nature outlook tuberculosis

A global disease under the microscope

outlookTUBERCULOSIS

Produced with support from:

Cover art: Neil Webb

S2 EPIDEMIOLOGY A mortal foe

Understanding history’s biggest killer

S4 DRUG DEVELOPMENT A combined effort

New types of drugs will help improve treatment outcomes

S7 PERSPECTIVE Graduation time

Universities need to join the effort, say David Russell & Carl Nathan

S8 VACCINES An age-old problem

Scientists search for better alternatives to the BCG vaccine, and ways to test it

S10 DIAGNOSIS Waiting for results

A clinic in Zambia serves as a litmus test for a new point-of-care diagnostic

S13 PERSPECTIVE Weigh all TB risks

Rethink the risks to lift burden of disease, say Christoper Dye & Mario Raviglione

S14 LATENCY A sleeping giant

Most infections don’t lead to illness, undermining our concept of disease

S16 TRANSMISSION Control issues

Knowing more about how tuberculosis spreads will lead to ways to stop it

10 October 2013 / Vol 502 / Issue No 7470

T U B E R C U L O S I SOUTLOOK

C O N T E N T S

Editorial Herb Brody, Michelle Grayson, Tony Scully, Afsaneh Gray, Vicki KitchenerArt & Design Wes Fernandes, Alisdair Macdonald, Yara Abdel RahmanProduction Karl Smart, Susan Gray, Ian Pope, Leonora Dawson-BowlingSponsorship David Bagshaw, Yvette Smith, Reya SilaoMarketing Elena Woodstock, Steven HurstProject Manager Christian MancoArt Director Kelly Buckheit KrausePublisher Richard HughesMagazine Editor Rosie MestelEditor-in-Chief Philip CampbellEditorial Advisor Andrew Jermy

Just when it seemed that humanity was ridding itself of its most lethal microbe, drug resistance and the HIV pandemic has kept Mycobacterium tuberculosis firmly on the map (S2).

To quell the rise of drug resistance, we need new types of drugs that act quickly and safely to increase the likelihood that a patient will finish their full course of treatment. In late 2012, the first new anti-TB drug in nearly half a century won market approval, and others should be soon to follow (S4).

Tuberculosis control also needs a practical point-of-care diagnostic – and one that also works in people with HIV. Our reporter visited clinics in Zambia to gauge the rapid roll out across Africa of the GeneXpert, and hears first hand the practical problems of using sophisticated technology in places with rudimentary facilities (S10).

These developments are the fruits of unprecedented public-private collaboration serving a market unable to afford market-priced medicines. To advance this enlightened effort, universities should learn to be less protective of intellectual property and promote affordable medicines for the world’s poorest (S7).

While new technologies are important, we also need a broader concept of risk when considering disease susceptibility to help identify ways to alleviate the suffering inflicted by TB using existing technology (S13). Researchers also need to clarify how tuberculosis takes hold in a population and spreads (S16).

Achieving the ultimate goal of eliminating tuberculosis will require an effective vaccine. Ten years ago, no vaccine candidates were undergoing clinical testing; today there are more than a dozen (S8). While the recent failure of a promising vaccine in a clinical trial was a setback, researchers are taking heart from the fact it was possible to conduct the trial in the first place (pages S8).

We are pleased to acknowledge the financial support of Janssen Research and Development LLC in producing this Outlook. As always, Nature retains sole responsibility for all editorial content.Tony ScullyScience Editor, Nature Outlook

1 0 O C T O B E R 2 0 1 3 | V O L 5 0 2 | N A T U R E | S 1

Nature Outlooks are sponsored supplements that aim to stimulate interest and debate around a subject of interest to the sponsor, while satisfying the editorial values of Nature and our readers’ expecta-tions. The boundaries of sponsor involvement are clearly delineated in the Nature Outlook Editorial guidelines available at http://www.nature.com/advertising/resources/pdf/outlook_guidelines.pdf

CITING THE OUTLOOKCite as a supplement to Nature, for example, Nature Vol XXX, No. XXXX Suppl, Sxx–Sxx (2013).

VISIT THE OUTLOOK ONLINEThe Nature Outlook Tuberculosis supplement can be found at http://www.nature.com/nature/outlook/tuberculosis It features all newly commissioned content as well as a selection of relevant previously published material.

All featured articles will be freely available for 6 months.

SUBSCRIPTIONS AND CUSTOMER SERVICESFor UK/Europe (excluding Japan): Nature Publishing Group, Subscriptions, Brunel Road, Basingstoke, Hants, RG21 6XS, UK. Tel: +44 (0) 1256 329242. Subscriptions and customer services for Americas – including Canada, Latin America and the Caribbean: Nature Publishing Group, 75 Varick St, 9th floor, New York, NY 10013-1917, USA. Tel: +1 866 363 7860 (US/Canada) or +1 212 726 9223 (outside US/Canada). Japan/China/Korea:Nature Publishing Group — Asia-Pacific, Chiyoda Building 5-6th Floor, 2-37 Ichigaya Tamachi, Shinjuku-ku, Tokyo, 162-0843, Japan. Tel: +81 3 3267 8751.

CUSTOMER [email protected] Copyright © 2013 Nature Publishing Group

COLLECTIONS18 TB’s revenge Leigh Phillips

S21 Global tuberculosis control: lessons learnt and future prospects

Christian Lienhardt et al.

S31 Out-of-Africa migration and Neolithic coexpansion of Mycobacterium tuberculosis with modern humans

Iñaki Comas et al.

S38 The Mycobacterium tuberculosis regulatory network and hypoxia

James E. Galagan et al.

Nature Outlooks are sponsored supplements that aim to stimulate interest and debate around a subject of interest to the sponsor, while satisfying the editorial values of Nature and our readers’ expecta-tions. The boundaries of sponsor involvement are clearly delineated in the Nature Outlook Editorial guidelines available at http://www.nature.com/advertising/resources/pdf/outlook_guidelines.pdf

CITING THE OUTLOOKCite as a supplement to Nature, for example, Nature Vol XXX, No. XXXX Suppl, Sxx–Sxx (2013). To cite previously published articles from the collection, please use the original citation, which can be found at the start of each article.

VISIT THE OUTLOOK ONLINEThe Nature Outlook Tuberculosis supplement can be found at http://www.nature.com/nature/outlook/tuberculosis

All featured articles will be freely available for 6 months.

SUBSCRIPTIONS AND CUSTOMER SERVICESFor UK/Europe (excluding Japan): Nature Publishing Group, Subscriptions, Brunel Road, Basingstoke, Hants, RG21 6XS, UK. Tel: +44 (0) 1256 329242. Subscriptions and customer services for Americas – including Canada, Latin America and the Caribbean: Nature Publishing Group, 75 Varick St, 9th floor, New York, NY 10013-1917, USA. Tel: +1 866 363 7860 (US/Canada) or +1 212 726 9223 (outside US/Canada). Japan/China/Korea:Nature Publishing Group — Asia-Pacific, Chiyoda Building 5-6th Floor, 2-37 Ichigaya Tamachi, Shinjuku-ku, Tokyo, 162-0843, Japan. Tel: +81 3 3267 8751.

CUSTOMER [email protected] Copyright © 2013 Nature Publishing Group

© 2013 Macmillan Publishers Limited. All rights reserved

“Quote black italic 10.5pt x 11pt over 5 or 6 lines with 2 clear line spaces above and below”

A MORTAL FOETuberculosis is one of the world’s most lethal infectious diseases. Further progress in consigning it to the past is a massive challenge. By Tom Paulson.

THETRANSMISS ION

CYCLE

Immune response

protection to adolescentsand young adults from the form of the disease that causes most

need to harness T cells, but scientists are still looking for correlates of protection (see ‘An age-old problem’, page S8).

Latency M. tuberculosis can be contained within granulomas for years. It is thought that latency may encompass a spectrum of states, from people who have completely controlled the disease, to those with undetected, subclinical disease (see ‘Latency: A sleeping giant’, page S14).

Activation If the immune system weakens as a person ages or contracts HIV infection, for example, bacterial replication can overcome the immune system and granuloma breaks down, releasing M. tuberculosis into the lungs and advancing disease (see ‘A

Transmission People with active disease develop symptoms such as a cough, which propels bacteria into the air where it can be inhaled by others. Tuberculosis must be diagnosed and treated as soon as possible to render the person non-infectious and prevent the spread of disease (see ‘Control issues’, page S16).

B-cells

T-cells

T cells can activate macrophages to kill M. tuberculosis as well as targeting the bacteria directly.

Infected macrophages signal other

immune cells to attack.

Dendretic cell

Macrophage

Granuloma

Necrotic macrophage

Epithelioid macrophage

In most people, their

immune system is able to control the infection in granulomas.

When infection overcomes the

immune system, it spreads

throughout the body.

Mycobacteria typically

spread through the air.

Lysosome

MacrophageOnce inside it takes over the cells internal machinery

internal phagocytic destruction and begins to replicate.

Once inhaled, M. tuberculosis invade the pulmonary alveoli and are engulfed by the body's front-line defense – the macrophages.

1

1

2

2

3

3

4

4

UncontainedM. tubercuosis in the lungs

prompts infectious diseases.

M. tuberculosis‘stough lipid-rich cell wall enables it to survivein air droplets.

In 2011, nearly 9 million people fell ill from TB and 1.4 million died, mostly in poor countries, with 60% of cases in Asia and 24% in Africa.


≥300

150–299

50–149

25–49

0–24

Prevalenceof TBper 100,000 people

Comparisonof TB incidenceand TB-related mortality

Best estimate ofincident cases* proportional to diameterby WHO regions

Number of people TB fatalities by WHO regions

In 2011, 8.7 million people

fell ill and 1.4 million died

from TB.

8.7M

Sub-Saharan AfricaTB kills more people living with HIV than anything else.

LondonCases rose by nearly 50% between 1999 and 2009.

Eastern EuropeThe fall of the Soviet Union led to the world’s worstoutbreak of drug-resistant TB.

IndiaThe worlds’s largest TB epidemic, with an estimated incidence of 2.2 million people.

The AmericasPopulation: 943 MIncidence: 260,000Mortality: 21,000HIV-positive(TB cases): 37,000

AfricaPopulation: 857 MIncidence: 2.3 MMortality: 220,000HIV positive: 870

EuropePopulation: 900 MIncidence: 380,000Mortality: 45,000HIV positive: 23,000

Eastern MediterraneanPopulation: 609 MIncidence: 660,000Mortality: 99,000HIV positive: 8,700

South East AsiaPopulation: 1. 8 B Incidence: 3.5 MMortality: 480,000HIV positive: 140,000

Western Paci�cPopulation: 1.8 BIncidence: 1.7 MMortality: 130,000HIV positive: 36,000

Tuberculosis has killed more than any other infectious disease in history. Over a billion lives in the past two hundred years.


T H E Rising living standards in industrialized nations, interupted by two World wars, and new antibiotics had tuberculosis in decline.


global burden of disease worsened by HIV.


Rat

e per

100,0

00 p

opula

tion

1990 1995 2000 2005 2010 2015

•Incidence

•Prevalence

•Incidence of tuberculosis-HIV coinfection

,000 ,000 ,000T U B E R C U L O S I S

D E AT H S

,000 ,000 ,000T U B E R C U L O S I S

D E AT H S

Tuberculosis

Smallpox

Malaria

PlagueIn�uenzaCholeraAIDS

0

50

100

150

200

250

300

Dea

ths

(mill

ions

year

–1)

in E

ngl

and a

nd W

ales

500

0

1,500

2,000

1,000

19601920 19401900

First World War

First TB drug

Second World War

KEY

New YorkHomelessness, overcrowding in shelters and HIV infections drove a TB epidemic in the early 1990s.

•Mortality

S 2 | N A T U R E | V O L 5 0 2 | 1 0 O C T O B E R 2 0 1 3

TUBERCULOSISOUTLOOK


A MORTAL FOETuberculosis is one of the world’s most lethal infectious diseases. Further progress in consigning it to the past is a massive challenge. By Tom Paulson.

THETRANSMISS ION

CYCLE

Immune response

protection to adolescentsand young adults from the form of the disease that causes most

need to harness T cells, but scientists are still looking for correlates of protection (see ‘An age-old problem’, page S8).

Latency M. tuberculosis can be contained within granulomas for years. It is thought that latency may encompass a spectrum of states, from people who have completely controlled the disease, to those with undetected, subclinical disease (see ‘Latency: A sleeping giant’, page S14).

Activation If the immune system weakens as a person ages or contracts HIV infection, for example, bacterial replication can overcome the immune system and granuloma breaks down, releasing M. tuberculosis into the lungs and advancing disease (see ‘A

Transmission People with active disease develop symptoms such as a cough, which propels bacteria into the air where it can be inhaled by others. Tuberculosis must be diagnosed and treated as soon as possible to render the person non-infectious and prevent the spread of disease (see ‘Control issues’, page S16).

B-cells

T-cells

T cells can activate macrophages to kill M. tuberculosis as well as targeting the bacteria directly.

Infected macrophages signal other

immune cells to attack.

Dendretic cell

Macrophage

Granuloma

Necrotic macrophage

Epithelioid macrophage

In most people, their

immune system is able to control the infection in granulomas.

When infection overcomes the

immune system, it spreads

throughout the body.

Mycobacteria typically

spread through the air.

Lysosome

MacrophageOnce inside it takes over the cells internal machinery

internal phagocytic destruction and begins to replicate.

Once inhaled, M. tuberculosis invade the pulmonary alveoli and are engulfed by the body's front-line defense – the macrophages.

1

1

2

2

3

3

4

4

UncontainedM. tubercuosis in the lungs

prompts infectious diseases.

M. tuberculosis‘stough lipid-rich cell wall enables it to survivein air droplets.



≥300

150–299

50–149

25–49

0–24

Prevalenceof TBper 100,000 people

Comparisonof TB incidenceand TB-related mortality

Best estimate ofincident cases* proportional to diameterby WHO regions

Number of people TB fatalities by WHO regions

In 2011, 8.7 million people

fell ill and 1.4 million died

from TB.

8.7M

Sub-Saharan AfricaTB kills more people living with HIV than anything else.

LondonCases rose by nearly 50% between 1999 and 2009.

Eastern EuropeThe fall of the Soviet Union led to the world’s worstoutbreak of drug-resistant TB.

IndiaThe worlds’s largest TB epidemic, with an estimated incidence of 2.2 million people.

The AmericasPopulation: 943 MIncidence: 260,000Mortality: 21,000HIV-positive(TB cases): 37,000

AfricaPopulation: 857 MIncidence: 2.3 MMortality: 220,000HIV positive: 870

EuropePopulation: 900 MIncidence: 380,000Mortality: 45,000HIV positive: 23,000

Eastern MediterraneanPopulation: 609 MIncidence: 660,000Mortality: 99,000HIV positive: 8,700

South East AsiaPopulation: 1. 8 B Incidence: 3.5 MMortality: 480,000HIV positive: 140,000

Western Paci�cPopulation: 1.8 BIncidence: 1.7 MMortality: 130,000HIV positive: 36,000







Rat

e per

100,0

00 p

opula

tion

1990 1995 2000 2005 2010 2015

•Incidence

•Prevalence

•Incidence of tuberculosis-HIV coinfection

,000 ,000 ,000T U B E R C U L O S I S

D E AT H S

,000 ,000 ,000T U B E R C U L O S I S

D E AT H S

Tuberculosis

Smallpox

Malaria

PlagueIn�uenzaCholeraAIDS

0

50

100

150

200

250

300

Dea

ths

(mill

ions

year

–1)

in E

ngl

and a

nd W

ales

500

0

1,500

2,000

1,000

19601920 19401900

First World War

First TB drug

Second World War

KEY

New YorkHomelessness, overcrowding in shelters and HIV infections drove a TB epidemic in the early 1990s.

•Mortality

WO

RLD

TU

BER

CU

LOS

IS R

EPO

RT

2012. L

IEN

HA

RD

T, C

. ET

AL.

NAT

UR

E R

EVIE

WS

MIC

RO

BIO

LOG

Y 10, 4

07-4

16 (2012)


OUTLOOKTUBERCULOSIS



The ringing in Dalene von Delft’s ears had become unbearable. The ines-capable noise was caused by the drug

amikacin, which von Delft — a doctor at Vergelegen Mediclinic hospital in Cape Town, South Africa — was taking to treat her drug-resistant tuberculosis (TB). But von Delft’s big-gest fear was that the ringing would go silent; in certain studies, up to half of those injected daily with amikacin for the two-year regimen were found to go deaf1. She wanted to cure her disease, but losing her hearing seemed like too high a price. As a doctor, “you need to be able to use a stethoscope”, she says. “It would have ended my career.” But stopping treatment alto-gether would have been an even more danger-ous option. Without proper treatment, up to two-thirds of people ill with TB will die2.

von Delft learned of a new anti-TB drug that was performing well in clinical trials. Although its safety had not been established, she pleaded with the drug’s maker, Janssen Pharmaceuticals, based in Titusville, New Jer-sey, and the South African Medicines Control Council, and managed to switch to the unap-proved drug, called bedaquiline. And although it made her heart palpitate temporarily, the remote danger of a heart attack seemed more bearable than losing her hearing. “You weigh the risks against the benefits,” she admits. After two years, the TB was gone and her hearing remained. “I was very lucky,” she says.

A new drug for TB has been a long time coming. Bedaquiline, approved by the US Food and Drug Administration (FDA) at the close of 2012, is the first novel anti-TB drug since rifampicin was introduced nearly half a century ago. With TB killing approximately 1.4 million people each year, industry, aca-demic institutions and public-health organiza-tions are having to collaborate like never before to stem the rise of drug-resistant TB.

NOT GOING QUIETLYRates of TB in the United States and Europe began to decline at the turn of the twentieth century because of sanatoriums built to isolate TB patients, along with improvements in overcrowded tenements that slowed TB transmission. Then, from the 1940s to the 1960s came antibiotics — streptomycin, isoniazid, rifampicin and 4-aminosalicylic acid (PAS) — that could effectively treat the disease. The arrival of these drugs led many to believe that TB would soon disappear altogether. “We were convinced in the 60s that we would end TB,” recalls Jacques Grosset, a professor at the Center for Tuberculosis Research at the Johns Hopkins School of Medicine in Baltimore, Maryland, who is himself a victim of TB (he had a portion of his lung removed during a bout with the disease in the 1950s).

But those expectations failed to materialize. Many patients, lacking careful supervision, would stop midway through the one- to two-year-long course of treatment. When this

D R U G D E V E L O P M E N T

A combined effortCombinations of anti-TB drugs are difficult to overcome because they attack Mycobacterium tuberculosis in different ways. B Y A M Y M A X M E N

NEI

L W

EBB

TUBERCULOSISOUTLOOK



happened, or when the drugs were given singly rather than in combination, naturally drug-resistant strains of the TB pathogen thrived and overtook the more sensitive population. People feel better after a few months and stop taking their pills, says Grosset. Once a patient acquired drug-resistant disease, they could spread their resistant pathogens directly to their neighbours.

Then came the HIV epidemic, which made people acutely vulnerable to TB. Developing countries, which had never managed to fully suppress the disease, experienced skyrocketing TB rates, and Western countries saw a resurgence. In 1993, the World Health Organization (WHO) in Geneva, Switzerland placed the disease back on the public-health agenda by declaring TB a global emergency. New drugs were needed with different mechanisms of action, which would suppress resistance and shorten treatment times. And yet pharmaceutical companies had little incentive to invest the massive amounts of money necessary for drug development — TB predominantly affects people from low-income countries, who cannot afford expensive treatments. Without a business case for TB drug development, progress stalled.

In the wake of the WHO declaration, the Bill & Melinda Gates Foundation, headquartered in Seattle, Washington, along with other non-governmental organizations and government agencies, including the National Institutes of Health in Bethesda, Maryland, got together to convince drug developers across the public and private sectors to work together. In response, the pharmaceutical giants GlaxoSmithKline (headquartered in Brentford, UK), Novartis (Basel, Switzerland), AstraZeneca (London) and Sanofi (Paris) agreed to collaborate with each other and with universities to pick up where TB drug developers had left off in the 1960s.

SHAKY START Developing a new drug for TB is a complex process. It involves screening thousands of compounds to find any that might kill the bac-terium, Mycobacterium tuberculosis; adapting these compounds into substances that work as a drug; testing for efficacy in animal models; and finally testing how effective and safe the drug is in human patients.

Various labs were able to identify com-pounds that inhibited metabolic pathways and other vital processes occurring within the bacterium; unfortunately, it became apparent that many compounds selected through these screens had trouble penetrating the membrane of whole mycobacteria and subsequently failed tests in mice. So researchers changed their screening protocols to look for the effects com-pounds had on whole mycobacteria — even so, it has not been smooth sailing.

In 2009, a team at the Novartis Institute for Tropical Diseases in Singapore landed on a new class of compound, pyrimidine–imidazoles,

which killed M. tuberculosis in vitro but had no effect on infected mice. In 2010, the researchers figured out why: the drug only blocked the abil-ity of the mycobacteria to survive in its glycerol suspension — leaving it with little relevance in the world beyond the test tube3. “That was not fun,” recalls Thomas Dick, who led the project at Novartis, and now directs the Antibacte-

rial Drug Discovery Laboratory at the National University of Singapore after Novartis dropped out of TB drug discovery. “It was the failure of a

two-year project that took a lot of investment.” Despite such disappointments, candidates

for new anti-TB drugs are continuously being identified. In May 2013, for example, researchers reported that high doses of vitamin C wipe out cultures of M. tuberculosis by triggering a DNA-damaging reaction4.

In July, a team at the Institute Pasteur Korea in Seongnam-si discovered a compound labelled Q203 that cuts off the energy supply to mycobacteria by blocking ATP synthesis both in culture and in mice 5.

Indeed, in the past decade, six types of com-pound that target M. tuberculosis in new ways have progressed to trials in humans. However, these trials face their own set of problems. Foremost among them is phase 2a testing, which checks whether a drug decreases the

mycobacterial load in patients’ sputum within two weeks of starting treatment. Because some drugs act slowly, the short timeframe of

this trial can be misleading. In fact, a num-ber of people at Janssen wanted to shut down the bedaquiline programme after it per-formed poorly in a phase 2a trial, says Myr-iam Haxaire-Theeuwes, who is developing bedaquiline at Janssen’s research and develop-ment branch in Beerse, Belgium. Spectacular results in cell culture and in mice convinced the team of bedaquiline’s worth. “You need to have strong product champions,” she says, “and senior management that will listen.”

Phase 2b trials are problematic for a different reason. For drug-resistant TB, such a trial needs to run for at least two years to determine whether the new treatment works: it can take that long to cure the disease. In the meantime, at least 310,000 patients with drug-resistant TB are taking potentially toxic medicines each year2.

To address the urgent need for new treatments for TB, the FDA announced in 2009 that in some cases they would grant ‘accelerated approval’ for promising drugs. For example, treatment success could be measured by no detectable mycobacteria in a patient’s sputum after six months of treatment, rather than ensuring that patients fully recover from drug-resistant TB at the end of a 2.5 year study.

It was under these new guidelines that bedaquiline was approved for multidrug-resistant TB in late 2012. Shorter clinical trials, however, leave more unknowns, including rates of cure and the incidence of side effects. “In situations where patients have few treatment options, healthcare providers will accept greater risk,” explains Edward Cox, the director of the FDA’s Office of Antimicrobial Products.

In June 2013, Janssen began to sell bedaqui-line in the United States, and approval is pending in several other countries. In the meantime, the drug is undergoing a phase 3 trial to assess with greater certainty its effects on drug-resistant TB. The Global Alliance for TB Drug Development (the TB Alliance), a non-profit organization based in New York, is also running a trial on bedaquiline — this time in combination with a leprosy drug, clofazimine, and another novel anti-TB agent, PA-824, in hopes of reducing the duration of the current six-month course of treatment for drug-susceptible TB.

JOINING FORCES Although they are often used as a marker of how well a drug is doing in trials, sputum bacterial counts are far from ideal. Counts can vary from day to day in a single patient, making the measure rather messy, says Clifton Barry, a TB researcher at the National Institute of Allergy and Infectious Diseases in Bethesda, Maryland, and at the International Tubercu-losis Research Center in Masan, South Korea. In his Korean laboratory, Barry scans patients’ lungs with computed tomography (CT) and positron emission tomography (PET) to see

NATURE.COMFor a review of tuberculosis drug discovery, visit: go.nature.com/ruw6so

TB survivor van Delft took a chance on a new drug.

“My fear is that what is really needed is orders of magnitude more funding.”

OUTLOOKTUBERCULOSIS



how drugs alter the state of M. tuberculosis inside the patient. Because the bacteria lurk in patients’ lungs long after they leave their spu-tum, he says this method is far more precise.

Scans may also help reveal complementary drug combinations. Thomas Dick in Singapore says that the ideal drug combo would include medicines that quickly kill M. tuberculosis rep-licating in the lung fluid, along with slower act-ing drugs that hit hard-to-reach mycobacteria hiding out in lesions in the lungs.

To find drugs that penetrate lesions, Dick proposes that researchers switch some of their animal studies to rabbits because, unlike in mice, M. tuberculosis forms lesions in their lungs. “It allows for a more rational selection of compounds that will work well in combina-tion,” Dick says.

One new drug with a novel mechanism is probably not enough to fight M. tuberculo-sis drug resistance. To tackle this, the Bill & Melinda Gates Foundation funds the TB Drug Accelerator programme, which partners seven drug companies with several laboratories at publically funded institutions. “We want at least one combination of three agents that

every TB patient is sensitive to so that the cur-rent notion of drug resistance just goes away,” says Ken Duncan, deputy director of drug dis-covery at the Gates Foundation.

The companies involved in the programme share their compound libraries and their results. Barry, who is a participant, says that sharing tools and information at each step of drug development eliminates the redundancy that can occur in more secretive and independ-ent drug development programmes where higher profits are at stake.

The FDA is also looking at ways to speed up the development of combination therapies. In 2010, they issued draft guidelines for testing

combinatorial treat-ments in urgent situa-tions. These guidelines will permit researchers to skimp on data about the effects of individual

components within new combinations in cases where the drugs are for the treatment of life-threatening diseases for which there are no satisfactory alternatives.

The coming decade looks bright for TB

drug development. By 2014, the European Medicines Agency in London is expected to approve delaminid, a novel drug from the Japa-nese drug company Otsuka, based in Tokyo. This compound, a nitroimidazole, poisons mycobacteria by releasing nitric oxide once it is metabolized. In July 2013, Pfizer, headquar-tered in New York, sold a novel drug candi-date — sutezolid — that was sitting on its shelf to Sequella, a small pharmaceutical company in Rockville, Maryland, for development and commercialization. Sutezolid, a drug that pre-vents mycobacteria from making proteins, looked promising in mouse studies, but Pfizer had frozen its development at that stage6.

MONEY MATTERSEarly results from the TB Alliance trial on bedaquiline, clofazimine, PA-824 and pyrazi-namide also look encouraging. Trials like this are exceptional in the drug development world because they involve compounds owned by multiple companies. Khisimuzi Mduli, a drug development project leader at the TB Alliance, suggests that the strength of the TB Alliance is that they have nothing to gain financially from drug development. This means that “our partners in pharma allow us to use their com-pounds in long clinical trials”, he says.

Even as such collaborations bring hope to the TB research community, uncertainty looms. As the founding director of the new KwaZulu-Natal Research Institute for Tuberculosis and HIV, Durban, South Africa, William Bishai is both thrilled with the recent infusion of money for TB research and filled with dread that it is not enough. “My fear is that what is really needed is orders of magnitude more funding rather than slight increases of 10 to 20 per cent per year,” says Bishai, who recently stepped down from the position to return to Johns Hopkins in Baltimore, Maryland.

Working in South Africa put Bishai at the heart of the TB epidemic. As well as giving his team access to a large pool of patients, he also made a happy discovery: there is no short-age of people willing to go after the disease. Like Dalene von Delft and Jacques Grosset, their personal experience of TB is motivation enough. “When I give a lecture here, two-thirds of the room has had TB or knows some-one who has died from TB,” he says. “So there’s a real fire in the belly of the young people here, and they want to be part of the next generation of scientists fighting TB.” ■

Amy Maxmen is a science writer based in Brooklyn, New York.

1. Frymark, T. et al. Evidence-Based Systematic Review: Drug-Induced Hearing Loss—Amikacin (American Speech-Language-Hearing Association, 2010).

2. World Health Organization. Tuberculosis Factsheet No. 104 (World Health Organization, 2013).

3. Pethe, K. et al. Nature Commun. 1, 57 (2010).4. Vilchèze, C. et al. Nature Commun. 4, 1881 (2013).5. Pethe, K. et al. Nature Med. 19, 1157–1160 (2013).6. Lessem, E. The 2013 Pipeline Report 22–261

(i-Base/TAG, 2013).

HIT LISTAnti-TB drugs attack Mycobacterium tuberculosis in di�erent ways, so combining di�erent drugs helps combat the bacterium developing ways to overcome a drug.

HIT LISTAnti-TB drugs attack Mycobacterium tuberculosis in di�erent ways, so combining di�erent drugs helps combat the bacterium developing ways to overcome a drug.

ATP synthase

RNA polymerase

mRNA

DNA

RibosomeProtein

Cel

l wal

l

Mycolicacid

Acyl lipids

Arabinogalactan

Peptidoglycan

Plasmamembrane

Cel

l cyt

opla

sm

Isoniazid1952Inhibits cell wall synthesis

Ethambutol1961Inhibits cell wall synthesis

Bedaquiline2012Inhibits ATPsynthase

Pyrazinamide1952Disrupts plasmamembrane

PA-824In phase IIGenerates toxic nitric oxide (NO)

Rifampicin1968Blocks mRNA synthesis

Cycloserine1956Inhibits proteinsynthesis

“You need to have strong product champions.”

TUBERCULOSISOUTLOOK



OUTLOOKTUBERCULOSIS

Tuberculosis (TB) is the single leading cause of death from bacterial infection. It is rapidly becoming untreatable, and untreated TB has a fatality rate of about 70% after three years. The challenges in

developing new drugs for TB are scientific, logistical, fiscal and societal. Over the past decade many pharmaceutical firms have abandoned

antibiotic research, having failed to discover effective candidates with new mechanisms of action. A further disincentive is the lower return on investment that rapidly curative drugs offer compared with palliative medications for prevalent conditions. The financial picture is particularly bleak for TB, which chiefly afflicts people in low- and middle-income countries. The prospect of even scantier profits makes it all the harder to entice drug companies to work on TB rather than infections that are common in wealthy markets.

The treatment of TB requires combination chemotherapy, because the use of a single agent virtually guarantees the rapid emergence of resist-ance. When new TB drugs do reach the market, rampant drug-cutting and counterfeiting in poorer countries mean that one or more of the drugs in the combination may be absent, or present at suboptimal concentra-tions, promoting the accelerated emergence of drug resistance, which is already prevalent to each of the widely used TB drugs. Consequently, one new TB drug is unlikely to do the trick: we need sets of drugs that work together. The difficulty of finding a new combination is more than additive, as each drug must not interfere with the others or with the antiretrovirals used to treat HIV infec-tion (a common co-infection in sub-Saharan Africa). Thus, even if a pharmaceutical company discovers one effective new TB drug (a big if), the chances are that such a drug would be rapidly lost to resistance unless it were used in combination with two or three other new drugs. Embark-ing on such a search takes an unprecedented level of social consciousness and cross-industry cooperation.

Despite these formidable challenges, a remarkable number of public–private partnerships (PPPs) have been launched for TB drug discovery and development. Among them are the Global Alliance for TB Drug Development, headquartered in New York, the Bill & Melinda Gates Foundation’s TB Drug Accelerator (Seattle, Washington), the Lilly TB Drug Discovery Initiative (Seattle), the Tres Cantos Open Lab Founda-tion (Guildford, UK), and the Innovative Medicines Initiative (Brus-sels). Many of these consortia pair academic researchers who have an up-to-date understanding of Mycobacterium tuberculosis biology and TB pathogenesis with pharmaceutical scientists who have access to chemi-cal compound collections that are larger than those in universities, more suited to drug development, and more extensively curated. Even more important, the pharmaceutical companies bring expertise in medicinal chemistry, chemoinformatics, pharmacology, pharmacokinetics and toxi-cology, along with the infrastructure to perform clinical trials. Such part-nerships provide outstanding opportunities for innovation and efficiency.

Although these PPPs commonly have more open approaches to intel-lectual property (IP) than are usual in drug development, several phar-maceutical companies have proved willing participants in the search for treatments for neglected diseases. Unfortunately, this enlightened

attempt to find a solution to the depleted TB drug pipeline has not been matched by all universities. Many academic institutions are strug-gling to fill a fiscal deficit. In the United States, as a consequence of the Bayh–Dole Act of 1980, universities in receipt of federal funding are free to license IP as a source of income. Under this profit-driven model, some institutions prefer to save precious patent-filing funds for IP with greater potential return. Some do not want to commit to the distribu-tion of drugs on a non-profit basis in the public markets of low-income countries, a condition required by many of the PPPs. There may also be concerns about sharing ownership of IP with a drug company, or universities may demand a share that is disproportionate with their contribution. Few institutions have heeded the call of the student Uni-versities Allied for Essential Medicines, a global group headquartered in Oakland, California, that challenges universities to adopt IP policies that promote affordable access to medicines and medical technologies for the world’s poor.

Although the technology transfer offices in universities may regard the potential income from the IP associated with the early stages of drug discovery as an attractive source of funds, this expectation is unrealistic for TB drug devel-opment because of the limited commercial return. Moreover, any such income pales in com-parison with the hundreds of millions of dollars that need to be invested to turn lead compounds into drugs. Who will pay to develop the best can-didates that emerge from the PPPs? To conduct

the clinical trials? To deliver the drugs to those in need? Regrettably, both a penchant for fiscal conservatism and a zeal for IP protectionism at some universities can obstruct the earliest stages of drug discovery.

The world cannot afford to wait long for answers to these questions, and universities need to play their part in finding solutions. For example, they will need to absorb some of the costs of IP protection and accept the potential for low fiscal returns for IP related to drug discovery for diseases like TB. More broadly, universities should lobby for larger, lasting fixes to the broken antibiotic pipeline, such as those called for by a recent panel of the Institute of Medicine of the US National Academies, based in Washington, DC. We need to examine ways of uncoupling the direct link between the rewards for antibiotic development and income from selling the drugs. As an alternative solution, a global fund — perhaps capitalized by financial transaction taxes — could compensate participating drug discoverers for their activities in proportion to their products’ reduction of the burden of disease. This would encourage industry and academia to become partners, promote cooperation, and render counterfeiting non-profitable. At this critical time it is imperative that universities re-evaluate their position and become activists for global health. ■

David G. Russell is the William Kaplan Professor of Infection Biology in the Department of Microbiology and Immunology at the College of Veterinary Medicine, Cornell University, Ithaca, New York 14853, USA. Carl F. Nathan is the R.A. Rees Pritchett Professor of Microbiology and Chair of the Department of Microbiology and Immunology at the Weill Cornell Medical College, New York, New York 10065, USA.

PERSPECTIVEGraduation time Universities should forego profits from tuberculosis, say David G. Russell and Carl F. Nathan.

THE FINANCIAL PICTURE IS PARTICULARLY

BLEAK FOR TB.



B Y S A R A H D E W E E R D T

It is one of the most widely administered vaccines in the world. In use for nearly a century, it has been given to over a billion

people. And yet, it is still not good enough.The bacillus Calmette–Guérin vaccine, or

BCG, is a live but weakened bacterium that is about 80% effective at protecting young healthy children from severe forms of tuberculosis (TB) — particularly tuberculous meningitis, which affects the brain, and miliary TB, a systemic infection of the body.

But BCG offers little protection to adolescents and young adults from the form of the disease that spreads widely and causes most deaths: pul-monary TB. Moreover, BCG is a replicating bac-terium; in children infected with HIV the vaccine itself can cause disease, known as BCG-osis.

Many experts say that eradicating TB will not be possible without an improved vaccine that protects against pulmonary TB. Because of drug resistance and the complexity of TB treatment regimens, “It is very clear that by far the most sustainable intervention would be if you have

a vaccine that prevents disease,” says Willem Hanekom, director of the South African Tuber-culosis Vaccine Initiative (SATVI) in Cape Town.

At first glance, the effort to develop an effec-tive vaccine against pulmonary TB seems poised to be richly rewarded. More than US$95 million went towards TB vaccine development in 2011 alone, and over a dozen candidates are in clinical trials. Just over ten years ago there were none.

The vaccine candidates are not only numer-ous but varied, using different mechanisms and dosing strategies. But this multitude of possibili-ties also reflects the field’s biggest problem: a lack of knowledge about Mycobacterium tuberculosis and how it interacts with the human immune system. “We don’t know enough to exclude any-thing at the moment,” says Helen McShane, a vaccinologist at the University of Oxford, UK.

These challenges were highlighted by the failure of one promising vaccine candidate, MVA85A, to show efficacy against TB in a clini-cal trial in early 2013.

MVA85A is designed to stimulate immunity to 85A, one of several thousand molecules, or antigens, produced by M. tuberculosis. The

vaccine delivers the antigen into cells — where it can trigger an immune response — via a geneti-cally engineered virus, so MVA85A is known as a viral vector vaccine.

The research team conducting the trial, includ-ing Hanekom and McShane, gave MVA85A injections to 1,399 South African infants who had been vaccinated with BCG earlier in infancy1. Several vaccine candidates are designed to work this way, known as a prime–boost strategy. It is thought that priming the immune system with BCG and then administering a booster dose of a new vaccine will continue to protect children against the most dangerous forms of TB while adding protection against pulmonary TB.

But, after about two years, children who had received the MVA85A vaccine had the same rate of TB infection and disease as children who had not1. “That it showed no efficacy is surprising, because this vaccine was vetted in many ways,” says Daniel Zak, a principal scientist at the Seattle Biomedical Research Institute in Washington.

Mouse and non-human primate studies had shown that MVA85A provides increased protec-tion against TB over BCG alone; it is not clear

JOS

HU

A M

ATTI

LLA

/UN

IVER

SIT

Y O

F P

ITTS

BU

RG

H

VA C C I N E S

An age-old problemResearchers are on the hunt for a better alternative to the BCG vaccine.

T cells (green) and nuclei (blue) in a tuberculous granuloma stained for regulatory T cells are important regulators of macrophage activation.

TUBERCULOSISOUTLOOK



why these results did not translate to humans. “Tuberculosis is a complex pathogen,” says

McShane. “It hides inside cells most of the time, which means the immune response required to clear it is predominantly a T-cell response.” T cells are involved in a process called cellular immunity, in which, among other things, they induce infected cells to ‘commit suicide’. By contrast, most of the successful vaccines in use today are based on humoral immunity, which mainly involves the production and secretion of antibodies into the extracellular fluid. Scientists have a better understanding of humoral immu-nity and the vaccines that stimulate it, but on its own it does not clear bacteria from cells.

TRIED AND TESTED Despite the lack of efficacy shown in the MVA85A trial, researchers say it was a success in other ways. It was the first efficacy trial of a TB vaccine in more than four decades. It demon-strated that a TB vaccine can be tested in infants according to modern-day ethical standards, and that researchers can develop the infrastructure and analytical methods needed to carry out a large clinical trial in a resource-poor country2. “In a way it’s become a template for how to do such a study,” says Hanekom.

With so many vaccine candidates waiting in the wings, it is a template that could get a lot of use. The next vaccine candidate likely to start efficacy trials is known as M72. It is an example of what scientists call a subunit or protein-adjuvant vaccine: a cocktail of two antigens, 32A and 39A, plus an adjuvant, a chemical that enhances immune responses. As with MVA85A, scientists envision using M72 as a booster vaccine in indi-viduals who have already received BCG.

Unlike the MVA85A trial, which was confined to infants, a team including SATVI researchers plan to test M72 efficacy in young adults, because they are more affected by pulmonary TB. So far, the team has shown that the vaccine is safe and induces a good T-cell response when adminis-tered to adults in South Africa either without TB or with latent infection3. An upcoming study will test whether the vaccine reduces the risk of developing active disease over the course of three years. It will involve around 7,000 participants in three African countries, Hanekom says, and is likely to begin by the end of 2013.

This is the first protein-adjuvant vaccine can-didate to be tested for efficacy, notes Jelle Thole, executive director of the Tuberculosis Vaccine Initiative in Lelystad, the Netherlands, which supports and coordinates TB vaccine research throughout Europe. It was chosen largely because it happens to be the vaccine in this cat-egory that was the most developed. But in the future, Thole adds, vaccine candidates within each category should be tested against each other, with the one that does best in animal stud-ies and at inducing immune responses in healthy volunteers moving forward first to efficacy trials.

A vaccine of a third type, based on a whole, live bacterium, may also be nearing efficacy

trials. “This is one of the few candidates that is trying to be better than BCG,” says Stefan Kaufmann, director of the department of immu-nology at the Max Planck Institute for Infection Biology in Berlin.

Kaufmann is one of the developers of the vaccine, called VPM 1002. It is based on BCG, which in turn is derived from M. bovis, which causes TB in cows. He notes that other types of vaccine contain only one or a few antigens to stimulate an immune response. His team chose to work with a whole live bacterium because “we thought that the whole antigen repertoire that is present in BCG might be a better choice,” he says.

If it works, Kaufmann hopes that VPM 1002 will one day replace BCG as the standard immunization given to infants. It could also be combined with one of the booster vaccines now in development.

To create VPM 1002, Kaufman’s team inserted a gene called listeriolysin into BCG and deleted another gene, encoding a subunit of urease4.

These genetic changes make BCG more visible to the immune system. For example, BCG mainly stimu-lates CD4 T cells whereas VPM 1002 is

better at also stimulating CD8 T cells. Kaufmann argues that a broader immune response is more likely to result in improved protection from the different forms of the disease.

VPM 1002 has bested BCG in mouse studies, and its safety has been assessed in newborns in South Africa. The results of that study have not yet been published, but “everything went very well,” Kaufmann says. The next step is to secure approval — and funding — to test the vaccine’s efficacy.

GOING ROGUE Meanwhile, some scientists are beginning to rethink what a successful TB vaccine might look like. For example, the Seattle-based Bill & Melinda Gates Foundation, which has con-tributed hundreds of millions of dollars to TB vaccine research, has issued a call to develop vac-cines that rely on mechanisms other than T cells, recruiting parts of the immune system such as natural killer cells or antibodies that are not big players in the body’s normal immune response to TB. That is likely to be a difficult exercise, but Zak thinks it might be worth trying, as “there’s something that’s missing” from the immune response induced by T-cell vaccines so far.

McShane, on the other hand, is not giving up on T-cell-inducing vaccines, or indeed on MVA85A. She is studying whether MVA85A will work if delivered directly to the airways as an inhaled vaccine. “We need to generate more potent vaccines. Delivering vaccines to the lung appears to be one way to do that,” she says. Her lab is also investigating whether combining MVA85A with another molecule designed to boost the immune response will improve efficacy.

Others are reconsidering not just how but when the vaccine should be delivered. For example, some investigators are asking whether a booster dose of BCG itself could improve pro-tection against pulmonary TB. And an effective vaccine may require even more frequent dosing. “I can imagine a scenario where you need to take a puff of your TB vaccine on your birthday,” says Kevin Urdahl, a principal investigator at the Seattle Biomedical Research Institute.

How to prioritize clinical trials and translate basic research more effectively into vaccine development is also generating interest.

“The big issue in vaccination right now is that we don’t have a good correlate of protection,” says Thole. In other words, there is no test or measure-ment that can define whether or not a person is protected from TB. “So we don’t know what kind of immunity a vaccine needs to induce.”

Researchers know that an effective TB vaccine must induce TB-specific T cells, but this does not seem to be the whole story. Take the MVA85A trial, in which researchers observed a T-cell response to the vaccine despite its lack of efficacy. Ideally, additional correlates of protection would be identified by comparing immune responses in individuals who are protected by a vaccine to those in individuals who are not protected.

The negative results of the MVA85A trial mean that researchers cannot make this com-parison, so this strategy remains elusive. Still, McShane hopes to mine the trial data to see if there are any differences in the immune responses between study participants who became infected with TB and those who did not.

Others are looking not for correlates of protec-tion but for what they call correlates of risk. By following the natural history of TB from the time of infection, “we can comprehensively define the immune responses in people that progress to [active] disease and those who do not progress”, says Zak, who is conducting such studies in col-laboration with SATVI.

Researchers could then conduct clinical tri-als of new vaccine candidates in people at high risk of progressing to active disease (see ‘Latency: A sleeping giant’ page S14), boosting the statisti-cal power of smaller, more rapid studies.

The need for a better clinical trial strategy is acute, some researchers say. “My fear is that if we push the clinical trials too quickly, and spend a lot more money getting results [from vaccines] that don’t provide much protection, then we’ll have lost our chance and the resources will run out,” says Urdahl. “There’s more basic research to do on the front end to make sure that we’re putting the right candidates into the pipeline.” ■

Sarah DeWeerdt is a science writer based in Seattle, Washington.

1. Tameris, M. D. et al. Lancet 381, 1021–1028 (2013).2. Tameris, M. et al. Tuberculosis 93, 143–149 (2013).3. Day, C. L. et al. Am. J. Respir. Crit. Care Med. 188,

492–502 (2013).4. Ottenhoff, T. H. & Kaufmann, S. H. PLoS Pathog. 8,

e1002607 (2012).

“We don’t know enough to exclude anything at the moment.”

OUTLOOKTUBERCULOSIS


S 1 0 | N A T U R E | V O L 5 0 2 | 1 0 O C T O B E R 2 0 1 3

B Y C A T H E R I N E D E L A N G E

Tobias Hamooya says he is feeling a bit better these days, but even to the casual observer it is clear that he is still not

doing well. Sitting on the porch outside the Macha Mission Hospital in rural Zambia, his slow, staggered speech is punctuated by a violent cough, and just talking seems to leave him drained.

A few weeks ago Hamooya’s cough got so bad he left his wife to look after their newborn baby and six other children and travelled the 150 kilometres to get here. Did he have any idea what was wrong with him? His answer needs no translation: “TB”. Since arriving at the clinic they also tested him for HIV, and the results came back positive for that too. Once his tuber-culosis (TB) medication kicks in, he will begin taking antiretrovirals (ARVs).

Pretty much everyone in the region is infected with TB, says John Spurrier, medical adviser at the hospital where Hamooya is being treated. “In most people your body handles

it. You may go all your life and never have a problem. Now with HIV, which destroys the immune system, it becomes a big issue.”

The highest proportion of new TB cases is in sub-Saharan Africa: more than 260 people per 100,000 in 2011. By comparison, in the same year France saw 4 cases per 100,000. And the region is in the grips of an HIV epidemic; TB kills more people living with HIV than any-thing else, and detection and treatment of TB is vastly complicated by HIV co-infection. Where once treatment programmes operated indepen-dently, many countries like Zambia now try to test and treat the two together. In 2004, just 3% of TB patients in the World Health Organiza-tion (WHO)’s African Region were tested for HIV; in 2011, it was 69%. Hamooya was one of the lucky ones: detecting TB is extremely diffi-cult in patients who also have an HIV infection.

HIV has an effective point-of-care test that can quickly and accurately detect infection and is practical and cheap enough for use in rural clinics. TB does not. In recent years, finding such a test for TB has been a major

public-health goal, and these efforts are begin-ning to bear fruit. Over the next few months, a new test called Xpert MTB/RIF that detects TB DNA will enter clinics in several high-burden countries in southern Africa; other tests are close behind. The question is, in the uncom-promising settings that are home to the great-est burden of disease, will these new tests fulfil their potential?

NEED FOR SPEEDA fast TB diagnosis is important. The sooner a patient is diagnosed, the sooner they can start treatment to mitigate the debilitating symp-toms of TB and limit the potential for trans-mission. But there is an added imperative for HIV-positive patients. “Starting someone on ARVs who has [untreated] TB means that, as their immune system recovers, they can get very sick,” says Spurrier.

Spurrier is referring to IRIS — immune recon-stitution inflammatory

D I A G N O S I S

Waiting for results There are several new tests for tuberculosis in the pipeline, but they must be shown to be effective in areas with limited resources and a heavy burden of HIV.

NATURE.COMFor more on the simple fluorescent dye test for TB, visit go.nature.com/wmtneq

CAT

HER

INE

DE

LAN

GE

The health clinic in Kanyama, a slum district of Lusaka, Zambia, is trying out a new device to test for tuberculosis.

TUBERCULOSISOUTLOOK


1 0 O C T O B E R 2 0 1 3 | V O L 5 0 2 | N A T U R E | S 1 1

syndrome. “Most of the symptoms you have from disease are [from] your body fighting the disease, not from the virus itself. So if your immune system is not working, you can have disease all through your body and have rela-tively few symptoms,” he says. Once the rebuilt immune system tries to fight TB, the immune response can be lethal.

The standard method for testing for TB in the clinic, smear microscopy, is not very fast. The technique has changed little since it was developed by German scientist Robert Koch in the 1880s; it involves staining and examin-ing a sample of sputum — the mucus patients cough up — under a microscope, counting the bacteria and grading the severity of the infec-tion. Although simple, this technique can take as long as a week to get a result and heavily depends on the experience of the technician performing the test. But even then it might be inconclusive: the test is estimated to detect TB in only around 40–50% of people with HIV1, rendering it impractical for parts of the world without the capacity for further tests. “Sputum is not very accurate: we miss people and then need to do additional work using X-rays and so on,” says Nzali Kancheya, TB programme director at the Center for Infectious Disease Research in Zambia (CIDRZ) in Lusaka.

HIGH HOPESXpert MTB/RIF aims to address these shortfalls, and its roll out is causing a big buzz. It is a fully automated machine and cartridge system that uses polymer chain reaction (PCR) to amplify and detect DNA sequences specific to Mycobacterium tuberculosis in sputum. The technology uses the GeneXpert system, an existing molecular testing device developed by molecular diagnostics company Cepheid, based in Sunnyvale, California. Cepheid, together with the Foundation for Innovative New Diagnostics in Geneva, Switzerland, and researchers from the University of Medicine and Dentistry of New Jersey, with additional funding from the National Institutes of Health in Bethesda, Maryland, developed the TB-specific system in 2008.

“It’s made to be used by people with minimal training, even though it’s a high-tech piece of equipment,” says Monde Muyoyeta, one of the lead researchers on the Zambia AIDS Related Tuberculosis (ZAMBART) Project, a non-gov-ernmental organization formed through a col-laboration between the University of Zambia’s School of Medicine in Lusaka and the London School of Hygiene and Tropical Medicine. Spu-tum samples are loaded onto cartridges and inserted into the machine, which is connected to a computer to read off the results. In theory, this set-up should make it easy to diagnose patients, as the machine does all the analysis.

In December 2010, WHO recommended the use of Xpert MTB/RIF to diagnose TB and is monitoring a massive global roll out. Part of this comes in the form of the TB Xpert Project,

a US$25.9 million programme funded by UNITAID, which is hosted by WHO in Geneva, and will provide over 220 machines and 1.4 million cartridges to 21 countries, mainly in East Africa and South-East Asia. Many countries are able to buy the system at a discount; as of March 2013, more than 6,100 machines and 2.3 million cartridges had been purchased2 (see ‘Xpert sales’, page S12).

In Zambia, several Xpert machines have been bought for use in clinics, in anticipation of government guidelines due later this year. In the meantime, other machines are already in use for research, for instance by ZAMBART, to

look at how effective it will be in resource-poor settings. A lot rests on this technol-ogy, says Kancheya. “In Western coun-tries they have a lot of other tech-

nologies, but in places like Zambia, GeneXpert will really help us,” she says. But first, solid evidence about its utility needs to be collected from the field, with the ZAMBART Project taking the lead.

PRACTICAL EXAMThe health clinic in Kanyama is a bright spot in this densely populated slum in the west of Zambia’s capital, Lusaka. It is also one of two sites chosen by ZAMBART to test GeneXpert in a primary healthcare setting. In addition to the GeneXpert units, the clinic is home to

a make-shift chest X-ray clinic, housed away from the dust in a shipping container painted with scenes of TB care in Africa. Kanyama sees the highest rates of TB in the city, and the clinic is the ideal arena to study how the GeneXpert technology performs far from the sterile lab.

For example, GeneXpert depends on a con-tinuous supply of electricity. “If you don’t have that, how does it affect the performance of the tests?” says Muyoyeta. Regular power cuts are a reality in Kanyama. “With smear microscopy, if you run out of electricity you can just come back and finish your test later,” she says. “But with GeneXpert, if you have a power outage and you don’t have backup, you have to start all over again.” If you are lucky, you have some leftover sample; if not, you will have to call the patient back to provide another. These issues are crucial in determining whether the system can perform as hoped. “If you’re wasting sam-ples and not getting the result the same day then it’s not a point-of-care test,” says Barry Kosloff, lab manager on the ZAMBART Project.

Then there is the cost. Under conces-sional pricing, the GeneXpert machine plus a computer to display the results costs around US$17,000. The cartridges cost just under $10 each. You need one cartridge per test, and the simplest version of the machine can process four cartridges at a time.

The cost is significant, Muyoyeta says. By comparison, “a microscopy slide is less than one dollar.” And, when you add in the capital cost and additional costs like electricity, Gen-eXpert will be far more expensive. Technology like this might leave low- or middle-income countries more dependent on outside help. “I think it can make a difference,” says Muyoyeta, “but it’s so expensive there’s no way that the minister of health can afford to roll out Gen-eXpert without support.”

GeneXpert delivers results in as little as two hours, which means patients should be able to wait to receive them. In a busy month, Kanyama clinic will see as many as 500 patients for TB screening. At present, it takes staff between 24 and 48 hours to get results, which means a return trip to the clinic for patients. “Transportation is very difficult,” says Kosloff. “If they are told they need to come back tomor-row or next week, they may say, ‘Well, I need to decide — is my family going to eat tomorrow, or do I make another trip to the clinic?’ If you can’t get the results to people the same day, why put this device in such a difficult location?”

There are alternatives to Xpert that are less high-tech. One innovation is an improved form of smear microscopy that uses light-emitting diode (LED) bulbs and fluorescent dyes to show up the bacteria more clearly under the microscope. Although this technique picks up more M. tuberculosis than standard micros-copy, it also mis-identifies contaminants, lead-ing to false-positive results.

A different approach is to try to rule out, rather than rule in, TB in HIV-infected people

CAT

HER

INE

DE

LAN

GE

Tuberculosis patient Tobias Hamooya.

“Sputum is not very accurate: we miss people and then need to do additional work.”

OUTLOOKTUBERCULOSIS



with suspected infection. Medical diagnostic firm Alere, based in Stockport, UK, has devel-oped a dipstick that tests urine for lipoara-binomannan (LAM), a component of the cell wall of M. tuberculosis. Alere’s test costs US$3.50 per dipstick and provides results in 25 minutes. It works best in people with a heavily compromised immune system, such as HIV patients with advanced infection. Overall, sensitivity — the proportion of TB it finds — in people also living with HIV is moderate at 40–60%, but its high specificity means that it can rule out TB in almost all uninfected people3.

Keertan Dheda, a pulmonologist at the University of Cape Town in South Africa, has conducted numerous studies into TB diagnostics. Dheda thinks that, despite its limitations, the dipstick is a valuable tool — particularly given the difficulties of detecting TB in HIV-infected people. “In this category of patients, a urine-orientated approach may be very useful.” Chris Smit, director of Alere’s infectious disease unit, says he suspects a number of countries are waiting for WHO prequalification of the device before they begin to use it more widely. Alere is working with WHO to get approval by the end of 2014.

BIGGER ISSUESThe GeneXpert system has an additional ben-efit that these other tests do not: it also checks for resistance to rifampicin, one of the cocktail of four first-line drugs used to treat TB. And if a patient’s TB is resistant to rifampicin, there is a good chance it may also be resistant to some

of the other drugs. In other words: multidrug-resistant (MDR) TB.

In countries with widespread MDR TB, such as South Africa or throughout Eastern Europe, this ability is valuable. However, in countries where drug resistance remains limited, this resistance test has a low predictive value, meaning that any samples that look like they might be resistant to rifampicin need to be sent to a reference lab to be confirmed. In Zambia, there are only three such labs — and few clinics carry alternative drug regimens.

This is the catch with the use of sophisticated technology in such low-resource settings, says

Muyoyeta. “It’s a problem because you’re identify-ing these patients and then you can’t do anything.” This would be especially problemat ic in more remote parts of the countr y,

which are many days travel from a reference laboratory.

Take the case of Tobias Hamooya, back at the Macha Mission Hospital in southern Zambia, where it is more than a day’s walk to the nearest small town. This is the second time Hamooya has been diagnosed with TB; the first time he stopped taking the drugs when he started to feel better. Consequently, it is likely that his infection has developed some resistance to first-line drugs, but being so far from a reference laboratory, this would take

many weeks to confirm, even if it was picked up by GeneXpert.

The jury then is still out on whether Gen-eXpert will live up to its expectations in Zam-bia. Muyoyeta and her team will publish the findings of the ZAMBART studies in the next few months. She acknowledges the amount of work required before GeneExpert can fulfil its potential. “We refer to GeneXpert as a new-born baby — it needs a lot of attention and is very demanding.”

And as Carol Nyirenda, a Zambian activ-ist who sits on the Stop TB Partnership’s New Diagnostics Working Group, points out, fail-ing to tackle TB undermines the tremendous progress that has been made to keep people living with HIV alive for longer. “In the earlier days if you were told you have HIV it was like a death sentence,” says Nyirenda, who is living with HIV and is a TB survivor. “I’ve been on medication [ARVs] for close to ten years. TB could come and take me away within a week or two. If they don’t invest in TB, then all the money they have put into investing in HIV treatment will just be wiped away.” ■

Catherine de Lange is Naturejobs editor and travelled to Zambia as a new media journalism fellow with the International Reporting Project.

1. Padmapriyadarsini, C., Narendran, G. & Swaminathan, S. Indian J. Med. Res. 134, 850–865 (2011).

2. World Health Organization. WHO Monitoring of Xpert MTB/RIF Roll-out (World Health Organization, 2013).

3. Lawn, S. D. et al. Lancet 12, 201–209 (2012).

“If they don’t invest in TB, then all the money they have put into investing in HIV treatment will just be wiped away.”

X P E R T S A L E S Sales of GeneXpert devices, which can help to diagnose tuberculosis in HIV-infected people, have shot up since it was recommended by WHO in 2010.

X P E R T S A L E S Sales of GeneXpert devices, which can help to diagnose tuberculosis in HIV-infected people, have shot up since it was recommended by WHO in 2010.

Cumulative number of GeneXpert instrument models and Xpert MTB/RIFcartridges procured.

Mod

ule

s (t

hou

sands)

Car

trid

ges

(mill

ions)

7

6

5

4

3

2

1

0

Q4

2010

Q1 Q2 Q3 Q4

2011

Q1 Q2 Q3 Q4

2012

Q1 Q2

2013

3.5

3.0

2.5

2.0

1.5

1.0

0.5

0

Sales begin to rise sharply after WHO recommends the use of Gene Xpert MTB/RIF for TB diagnosis in HIV infected people in December 2010.

As sales continues to rise, concerns raised about whether manufacturer Cepheid can meet demand.

0–45–1920–49≥50

HIV prevalence in people living with tuberculosis (%)

The African region has the highest burden of HIV/TB coinfection and accounts for approximately 79% of HIV-infected tuberculosis cases.

1–45–1617–100101–500≥501

Numbers of orders

South AfricaThe biggest buyer of GeneXpert machinesto date, having purchased almost half of the total.

ZambiaStudies are underway in to test whether GeneXpertMTB/RIF works in resource poor settings.

WH

O D

ATA

PR

OVID

ED B

Y FI

ND

TUBERCULOSISOUTLOOK



OUTLOOKTUBERCULOSIS

What factors put people at risk of illness, disability and death? The answers to this question have far-reaching implica-tions: identifying a risk factor suggests interventions that

could avoid or alleviate sickness and suffering. Unfortunately, for tuberculosis (TB) and other diseases, current risk assessments are not up to the job.

The Global Burden of Disease Study (GBD) includes an ambitious attempt to pinpoint the major causes of illness worldwide, and to use them to set a global agenda for preventive health care. But, despite listing 67 risk factors in 10 categories1, the GBD is selective and has little relevance to some important diseases, including TB. A broad set of both disease determinants and factors that limit disease control and treatment should be included in future studies, with this combined set used as the basis for developing a wider range of options for disease prevention and care. This proposal goes beyond TB: for many causes of ill health, an unidentified risk is a missed opportunity.

BROADER CONCEPTS Most of the risk factors selected by the GBD are environmental exposures, harmful behaviours, such as alcohol abuse, or physiological abnormalities, such as hypertension or high choles-terol. Only 3 of the 67 factors listed are linked to TB — tobacco smoke, alcohol abuse and diabetes. And yet there are clearly many other factors that deter-mine who becomes ill or infectious. For example, migration, urbanization and the way people interact through contact networks are important for the transmission of infection. Genetic factors are also excluded, even though these risks might one day be man-aged or treated. But perhaps the most significant drawback of the GBD and similar risk assessments is that they do not consider the limitations of current interventions as avoidable risks.

In TB control, these limitations include poor awareness of symptoms, lack of access to diagnostic and treatment facilities, the prohibitive cost of drugs to treat multidrug-resistant strains, medical malpractice, poor quality of care from health workers, broken drug supply chains, and patients not completing their treatment. Although these shortcomings are not conventionally thought of as risk factors, they account for a large proportion of the avertible burden of disease. To choose the best options for disease control, their importance must be considered next to conven-tional TB risk factors such as overcrowded housing, diabetes, tobacco smoking, HIV co-infection and under-nutrition. We need, in short, to adopt a more comprehensive view of risk.

ADVERSITY INTO OPPORTUNITY Finding new ways to reduce TB is a global health priority, and expand-ing the concept of risk will generate more options for control. By com-paring the costs and benefits of possible interventions, we can prioritize the best among them. For example, in India we recently found that the

increases in TB risk from diabetes, malnutrition and urbanization are modest compared with the expected positive impact of early detection and treatment2. In this setting, the next step is to compare the costs and potential benefits of better case detection and treatment strategies with those of interventions to mitigate other risk factors.

To be comprehensive, this work should go beyond evaluating measures targeted specifically at TB to look at those that have wider benefits for public health, such as health insurance schemes. In this way, the healthcare profession will be encouraged to evaluate interventions that could benefit TB but that lie beyond the reach of current disease control programmes.

CHANGING THE AGENDA The United Nations (UN) Millennium Development Goals (MDGs) will expire in 2015. During the MDG era, the rise in the TB inci-dence rate has been halted and reversed, but the decline is still only a disappointing 2% per year globally3. Effective TB control programmes should be able to reduce incidence by at least 5–10% each year4 The new UN agenda for international development will probably focus on

poverty reduction and sustain-able development5. Given limited resources, the challenge for TB control is to take a broader view of risk, set-ting priorities that overcome a diverse array of obstacles and exploit all pos-sible opportunities. These priorities should include better ways to use existing technologies while promot-ing the most effective new technolo-gies; working closely with the control of non-communicable diseases; and participating in initiatives to improve health that come not only from the

health sector, but also from agriculture, education, finance, indus-try and housing. This demands a big but potentially rewarding pro-gramme of data collection, quantitative analysis and modelling — one that enlarges the idea of risk to unify TB treatment and prevention, and places both in the wider context of health and development. ■

The author(s) alone are responsible for the views expressed in this article. The article does not necessarily represent the decisions, policy or views of the WHO. The WHO retains copyright. Christopher Dye is Director of Health Information and Mario Raviglione is Director of the Global Tuberculosis Programme in the HIV/AIDS, Tuberculosis, Malaria & Neglected Tropical Diseases Cluster at the World Health Organization, Geneva, Switzerland. [email protected], [email protected]

1. Lim, S. S. et al. Lancet 380, 2224–2260 (2012).2. Dye, C. et al. PLoS ONE 6, e21161 (2011).3. World Health Organization. Global Tuberculosis Control: WHO Report 2012

(World Health Organization, 2012).4. Dye, C. et al. Annu. Rev Public Health 34, 271–286 (2013).5. United Nations. A New Global Partnership: Eradicate Poverty And Transform

Economies Through Sustainable Development (United Nations, 2013).

PERSPECTIVEWeigh all TB risksA narrow definition of risk is hampering the search for new methods of tuberculosis control, say Christopher Dye and Mario Raviglione.

THIS PROPOSAL GOES BEYOND TB: FOR MANY CAUSES OF ILL

HEALTH, AN UNIDENTIFIED RISK IS A MISSED OPPORTUNITY.


B Y C O U R T N E Y H U M P H R I E S

Among the elderly population in Hong Kong, a spate of tuberculosis (TB) is worrying public-health officials. But

these cases are not the result of recent infec-tion; they signal the emergence of disease from an infection picked up decades ago. “Almost all recent cases in the elderly have arisen through reactivation of long-term latent infections rather than recent new infections,” says Benjamin Cowling, an infectious disease epidemiologist at the University of Hong Kong’s School of Public Health in China.

Improved standards of living and public-health campaigns in most of the industrialized world dramatically reduced rates of TB during the first half of the twentieth century. But, unlike in the West, Hong Kong, Japan and Korea only began to get to grips with TB after the Second World War. So people who were first infected as children back in the 1940s and 50s — and who have shown no sign of TB for over 50 years — are succumbing to the disease as their immune systems weaken. “Given a high prevalence of latent TB in the elderly, and currently no strate-gies to stop reactivation,” Cowling says, “it is dif-ficult to foresee major changes in TB incidence in the coming years.”

By current estimates, two billion people are infected with Mycobacterium tuberculosis world-wide, but only 10% will develop active disease in their lifetimes. The rest have what are called latent infections: they are not sick and will not spread the infection to others1. The risk of developing disease is roughly 5% in the first 18 months, and 5% over the rest of one’s life2.

This lurking threat is motivating researchers to study TB latency. Are there differences in the bacteria or in the host immune response that determine who gets the disease? Should every-one with latent infection be treated, or are there biomarkers or molecular signatures that could distinguish those who will develop disease?

TRICKY CUSTOMERLatency is not a well-defined biological condi-tion; it describes anyone who tests positive for TB and does not have clinical symptoms, says David Sherman, a biochemist at the Seattle Biomedical Research Institute in Washington. The assump-tion has been that M. tuberculosis remains viable in people with latent infections, not causing dis-ease but maintaining its potential to do so.

This ability to lie dormant may be an evo-lutionary strategy of the bacterium. Douglas Young, a microbiologist at the MRC National Institute for Medical Research in London, points

out that TB first appeared when early modern humans lived in small communities. “If TB was as virulent then as it is now, you would expect those populations to be wiped out,” he says. “There’s a big advantage to being latent, letting the population reproduce and then infecting the next generation.” The microbe may have only acquired more virulent traits as human settlements grew and it was possible to flourish despite proving fatal to some of its hosts.

But the distinction between latent and active disease may not be as clear-cut as once thought. Ongoing studies of patients with latent TB sug-gest that although some may have completely controlled the infection, others might have undetected subclinical disease. “Latency encom-passes a grab bag of individuals at different points in the spectrum of disease,” says Clifton Barry, chief of the Tuberculosis Research Section at the National Institute of Allergy and Infectious Diseases in Bethesda, Maryland.

This emerging view of latent infection is supported by research on the pathology of M. tuberculosis infection in animals. JoAnne Flynn, a microbiologist at the University of Pittsburgh in Pennsylvania, worked for years to create a

L AT E N C Y

A sleeping giant Most people infected with Mycobacterium tuberculosis never get the disease, but predicting who will is turning out to be a complex problem.

NATURE.COMFor more on transcriptional profiling, see: go.nature.com/9ivteq

WH

O/N

. OTT

O

Hong Kong and other regions in South-East Asia are trying to get to grips with a resurgence in tuberculosis.

TUBERCULOSISOUTLOOK



model of latency in mice. But her breakthrough came about a decade ago when she was work-ing with cynomolgus macaque monkeys, and noticed that about half of the TB-infected ani-mals did not go on to develop disease.

Flynn saw a surprising heterogeneity in lung pathology among individual monkeys — and even within the same individual. She has focused her sights on pockets of inflammation called granulomas: collections of mycobacteria, mac-rophages and other immune cells, sometimes surrounding dead tissue. Granulomas can vary in size, composition and degree of organization, and are thought to represent an attempt to con-tain the infection3. “It turns out latent monkeys can have a number of granulomas, even one or two that look active,” she says. The pathology of the lungs is less severe that those with active infections, but the distinction is not absolute.

SUMMING THE PARTSUsing positron-emission tomography (PET) and computed tomography (CT) imaging, Flynn is able to observe individual lesions in the lung over time, and then perform molecular analyses in more depth. “Each granuloma is its own little world,” she says, and whereas some of them are able to control the infection, some are not. Far from hiding or going dormant, the bacterium seems to be involved in an active, dynamic bat-tle in which the immune system fights to keep the disease in check across various individual sites. If that is true, says Sherman, “the distinc-tion between latent TB and active TB becomes an arbitrary one”.

The notion that latency represents the sum of its parts would make it difficult to find sys-temic biomarkers of TB progression. “It suggests that looking at the global immune response is unlikely to be able to predict who is going to do well and who is not,” Flynn says. However, studying individual lesions could illuminate the factors that allow host and microbe to coexist for long periods of time, and why the balance sometimes fails.

Barry has been using the same imaging method to study lung lesions in people with active TB, and now in two groups in South Africa and South Korea with latent infections. He has already observed lesions in some people with latent infections and the study “definitely sup-ports the idea that latency in people is incredibly heterogeneous”. But it is too early to say whether the extent of the lesions predicts whether people get the disease, as it seems to in monkeys.

To figure out why TB lesions can vary so much, researchers are looking at how the het-erogeneity within populations of mycobacteria might influence the microenvironment of the granulomas, and their interaction with the host’s immune system. Sarah Fortune, a microbiolo-gist at the Harvard School of Public Health in Boston, Massachusetts, has been working with Flynn to develop molecular tools to analyse the variability of bacteria in monkeys with TB.

Fortune and her team have created bacteria

engineered with genetic sequence tags that act as molecular barcodes: one sequence that identifies each strain, and a randomized sequence unique to each bacterium. It allows them to track the fate of different strains and of individual bacteria and their progeny. They have also taken advantage of the fact that when TB bacteria die, they remain in granulomas. Researchers can count the genomes of dead and living bacteria in these bac-terial graveyards, helping them to understand the course of the infection. Fortune believes that even bacteria that are genetically identical may differ in gene expression. “There’s probably a lot of individuality in the bacterial population that hasn’t been appreciated,” she says.

For public-health efforts to tackle the reacti-vation of latent infections, the challenge is first to identify, among the vast numbers of people worldwide with latent infections, those most in need of treatment. “The key issue in managing latent TB is diagnosing people who are at risk of progressing to disease,” says Barry.

In 2010, Anne O’Garra, an immunolo-gist at the MRC National Institute for Medi-cal Research, reported progress on this front. Her research showed that the transcriptional profiles of people with active versus latent TB differed. Such profiling could lead to bet-

ter diagnostic tests to distinguish latent from active infection, or a prognostic tool to determine which people with latent infections will go on to get active disease

or to track treatment success in diseased peo-ple. Robert Wilkinson, an infectious disease specialist at Imperial College London and the University of Cape Town who collaborates with O’Garra’s team, says that they are now testing the approach in larger and more complex groups of patients, including people with HIV and other diseases. “The crucial clinical question is to tell [the profile] apart from diseases that might mimic TB,” he says, pointing out that it is not yet clear whether this information can be divined from a transcriptional profile.

TREAT IT RIGHTLatent infections can be successfully treated, but if doctors treated everyone then ten people would be getting drugs for every one who would have developed disease, says Richard Menzies, director of the Respiratory Division at the Montreal Chest Institute in Canada.

Consequently, Western governments have made recommendations to screen for and treat latent TB based on risk factors. For example, updated guidelines in 2011 for England and Wales recommend screening all recent migrants from countries where TB is widespread, and suggest that treatment should be considered in people with latent infection who are immuno-compromised (such as through HIV infection).

In poor countries where TB infection is

endemic and resources are limited, that level of treatment is not feasible, nor is it the prior-ity when so many people with active infections are transmitting disease. But even in developed countries, both doctors and patients are reluc-tant to initiate treatment in the absence of dis-ease. “There’s widespread acceptance of drug treatment for conditions that predispose people to illness, such as hypertension or high choles-terol,” says Menzies, “but somehow the notion is you don’t have to treat latent TB infection,” even though the odds of preventing later disease and of experiencing side effects are comparable.

Ensuring patients complete their treatments, a challenge in patients with active TB, is even more problematic for patients who are not even sick: a long medication regimen for a disease they might never get is a tough sell. Menzies believes that until treatment regimens are avail-able that have a negligible risk of side effects, doctors will continue to be reluctant to prescribe treatment and patients unwilling to accept it.

Developing drugs specifically for latent infection is difficult, as at present drug mak-ers are required to show the efficacy of a drug against replicating bacteria or active disease in mice, but Sherman says that there is a lot of interest among researchers in targeting latency. “Many people now believe that the key to shortening therapy for active disease is to find agents that work on latent disease,” says Sherman, because bacteria may be living in a variety of conditions and metabolic states that overlap between the two conditions, and these states may make them more or less susceptible to different drugs.

Isoniazid targets bacteria as they replicate, but drugs developed to exploit other bacterial functions could be more effective at treating latent infections. For instance, researchers at the Institute Pasteur Korea in Seongnam-si have found a compound that targets ATP synthesis in M. tuberculosis, a process that is slow in bacteria that are not replicating, but which they still need to survive. Such a drug could benefit both active and latent disease4.

The situation in Hong Kong illustrates how leaving latent infections untreated is a stum-bling block to achieving very low incidence rates. “It’s guaranteed that we can never break the cycle if we only treat the active cases,” says Barry. Cowling agrees. “Latent TB is going to be a problem for many years to come,” he says. And he believes that what is happening now in more developed areas such as East Asia is a harbinger of future problems in less developed countries that are still grappling with active disease. ■

Courtney Humphries is a freelance writer based in Boston, Massachusetts.

1. Esmail, H., Barry, C. E. III & Wilkinson R. J. Drug Discov. Today 17, 514–521 (2012).

2. Zumla, A. et al. N. Eng. J. Med. 368, 745–755 (2013).

3. Ramakrishnan, L. Nature Rev. Immunol. 12, 352–366 (2012).

4. Pethe, K. et al. Nature Med. 19, 1157–1160 (2013).

“It’s guaranteed that we can never break the cycle if we only treat the active cases.”

OUTLOOKTUBERCULOSIS



B Y E W E N C A L L A W A Y

The USS Richard E. Byrd left its Norfolk, Virginia port in January 1965, bound for the Mediterranean. The 4,000-

tonne destroyer carried more than 300 sailors — one of whom was infected with Mycobac-terium tuberculosis, the bacteria that cause

tuberculosis (TB). By the time he left the ship more than a year later, coughing violently, almost half the crew had acquired silent infec-tions and seven of his cabin mates and other close contacts had full-blown TB.

“It was just a beautiful case of being able to pin down when transmission happened and when disease happened,” says Clifton Barry,

head of the Tuberculosis Research Section at the National Institute of Allergy and Infectious Disease in Bethesda, Maryland.

The USS Byrd and other locally limited out-breaks offer an opportunity to monitor the spread of the bacterial disease. They provide critical epidemiological data for scientists studying TB transmission in less cramped quarters, and highlight the challenges to com-batting a disease that can spread wildly before even being noticed. “We never find the cases before they’ve infected 10 or 15 other people,” Barry says.

Many experts see reducing the spread of M. tuberculosis infection as the only way to manage the disease. “Transmission is really the dominant problem we have in TB control,” says Chris Dye, an epidemiologist at the World Health Organization (WHO) in Geneva, Swit-zerland. “We know how to treat patients in clinics and cure them and save their lives and reduce illness, but what we’ve been far less suc-cessful at doing is cutting out transmission.”

STEMMING THE TIDETB infections come in two general forms: active and latent. People with latent TB, like most of the infected sailors aboard the USS Byrd, are not symptomatic and are unlikely to develop the disease (see ‘Latency: A sleeping giant’, page S14). A long course of antibiotics can rid the body of the low levels of bacteria they carry, but even without treatment, they will probably never pass on their infection to others.

Most TB control efforts target the active form of the disease. Active infections can spread the mycobacterium like wildfire, even before patients develop symptoms such as a cough — and start expelling droplets filled with the infectious bacteria into the air for other people to inhale. Mark Perkins, chief scientific officer at the Foundation for Innovative New Diag-nostics, in Geneva, uses the analogy of mop-ping up “the spill” of highly infectious people with active TB “before you turn off a tap that is barely dribbling”.

Poor quality medical care can exacerbate transmission. In countries such as India it is not uncommon for patients with active TB to visit four or more healthcare providers over a period of up to several months before getting a correct diagnosis1, and unregulated medical practices are rife. “Patients are spending far too long in the community transmitting TB to other people, before they end up in a clinic that can diagnose their TB and put them on the right treatment,” says Dye.

In most parts of the world, a diagnosis of active TB is confirmed by observing the tiny bacilli in a sample of sputum put under the microscope (see ‘Diagnosis: Waiting for results’, page S10). But by the time M. tubercu-losis reaches high enough numbers to be seen in this way, patients are already infectious. “You’ve missed the boat on stopping transmis-sion,” says Perkins.

T R A N S M I S S I O N

Control issuesOnce tuberculosis takes hold in a population it can be hard to control, but scientists are finding new ways to understand and stop its spread.

NEI

L W

EBB

TUBERCULOSISOUTLOOK



Efforts to control TB should focus on lower-ing the costs of diagnostics (and drugs) for gov-ernments, healthcare providers and, ultimately, patients, says Dye. “The TB control community are not by themselves going to remedy the prob-lem of poor health systems, but they’re going to have to make a contribution.”

BEYOND BORDERS Among the countries with a high burden of TB in Eastern Europe, Africa and Asia, one stands out as having had particular success at getting to grips with TB transmission. Estonia, a nation of 1.3 million people perched on the Baltic Sea, experienced a surge in TB in the early 1990s, after the collapse of the Soviet Union — and with it the healthcare infrastructure it provided.

With support from nearby Scandinavian countries, the Estonian government imple-mented a TB control programme, involving medical training, public-health screening and treatment programmes, says Manfred Danilo-vits, a physician at Tartu University Hospital who heads the programme. As a result, inci-dence dropped dramatically — from over 50 cases per 100,000 in 2002 to about 25 cases in 2011 (ref. 2). The TB control programme is now experimenting with ways to tackle drug-resistant TB and co-infection with HIV, says Danilovits. For instance, some people living with HIV, many of whom will have acquired the infection through intravenous drug use, who are also infected with TB can get all their medicines — antibiotics, antiretrovirals and, if necessary, methadone — at one clinic instead of three, improving their adherence to a strict treatment programme.

Understanding transmission is critical to stemming the rise of TB in Western countries as well. London earned a reputation as the ‘TB capital of Europe’ after cases rose by nearly 50% between 1999 and 2009, from 2,309 to 3,450 ( ref. 3). More than half of TB cases in regions such as the United Kingdom are in immigrants from high-burden countries, says Iacopo Baussano, an epidemiologist at WHO’s International Agency for Research on Cancer in Lyon, France, who has studied the effectiveness of TB screening among new immigrants.

In a 2012 survey4 of 29 high-income countries, Baussano and colleagues found that 25 of them

screen immigrants for active TB infections — typically with a skin test or chest X-ray, or by checking for symp-toms such as a cough. Only a third of those

countries screen people seeking entry before arrival; nearly all of them screen migrants soon after they arrive. Just 16 of the countries screened any immigrants for latent TB infection.

Undocumented immigrants and refugees are at the greatest risk of TB infection, Baus-sano says, because they often emigrate from countries with a high incidence of TB and live in

high-density housing that fosters transmission. Existing TB screening programmes do a good

job of identifying cases in new immigrants, with infection rates typically matching those of the immigrants’ country of origin. But it has proven more difficult to prevent new TB cases in peo-ple who have recently migrated to cities such as London. His work suggests that marginalized immigrant populations should be screened more closely for TB in the years after they arrive.

TB risk tends to decrease the longer indi-viduals live in a country, presumably after people gain access to health care and more stable housing. And Baussano says that there is no evidence to suggest that the TB circulat-ing among some groups of immigrants ever spreads into the general community.

Unfortunately, in the European Union, there is a political impediment to coordinated TB

screening. Procedures vary widely between countries and even between cities, and many cases slip through the region’s open borders. “It’s not easy to integrate the systems,” says Baussano.

TAKING THE STRAINOne major challenge to understanding the transmission of different forms of TB, such as those that are resistant to drugs or spread more quickly, is identifying them, says Ruth McNerney, a molecular microbiologist at the London School of Hygiene and Tropical Medi-cine. Related TB strains can differ very little at the genetic level, and most laboratory tests can-not tell them apart. McNerney’s team is turn-ing to sequencing the entire genome (made up of approximately 4 million building blocks) of clinical isolates of TB, to track their spread more closely.

McNerney and colleague Taane Clarke are

building a reference library of TB varieties so that researchers can easily trace a strain they have identified back to those circulating elsewhere in the world. Even with the cost of genome sequencing falling and states such as the United Kingdom integrating the technology into routine health care, genome sequencing is unlikely to influence care in poor, high-burden countries any time soon, McNerney concedes. But she believes that data gained from sequenc-ing will be used to develop simpler, cheaper tests to discern different strains of TB and help inform treatment.

TACKLING RESISTANCESequencing is also helping researchers to ana-lyse outbreaks of drug-resistant TB, McNerney says. Her team recently looked at strains of drug-resistant TB from people in Uganda who had been hospitalized twice with the disease, comparing the M. tuberculosis genomes of the two infections. In some patients, the strains dif-fered very little, suggesting that drug resistance had resulted from poor compliance with antibi-otics. But in other patients they differed mark-edly, says McNerney, and the second infection often matched the genome of a strain circulating in the same clinic. “There seems to be problem of multidrug resistance being shared amongst re-treatment cases,” she says.

Whole-genome sequencing can also help reveal the underlying social causes of a TB out-break. A team led by Jennifer Gardy, a molecular epidemiologist at the British Columbia Centre for Disease Control in Vancouver, investi-gated an outbreak in Vancouver that resulted in 41 cases of active TB between 2006 and 2008. Genome sequencing revealed that the epidemic was really two outbreaks, caused by distinct strains of TB that had emerged at the same time5. Gardy’s findings suggested that the rise of crack-cocaine use in the city sparked the outbreaks, with crack houses becoming centres of TB transmission.

Her team also found that a handful of ‘super-spreaders’ were responsible for transmitting most of the TB cases. Biological factors, such as the levels of TB bacilli in a person’s sputum or cough, or a suppressed immune system that allows rampant bacterial replication, may explain this phenomenon, she says. But Gardy’s work indicates that social contact is often the most important element of an outbreak. “TB fundamentally is a social disease,” she says. “Social factors are what makes an outbreak, versus one sick person who doesn’t transmit to anybody else.” ■

Ewen Callaway is a Senior News Reporter for Nature.

1. Dye, C. Indian J. Med. Res. 135, 737–744 (2012).2. The World Bank. Incidence of Tuberculosis (Per

100,000 People) (The World Bank, 2013).3. Zumla, A. Lancet 377, 10–11 (2011).4. Pareek, M. et al. Emerg. Infect. Dis. 18, 1422–1429 (2012).5. Gardy, J. L. et al. N. Engl. J. Med. 364, 730–739 (2011).

MA

RIA

NN

E LO

OR

ENTS

/OH

TULE

HT

TB patients await treatments in Estonia.

“Transmission is really the dominant problem we have in TB control.”

OUTLOOKTUBERCULOSIS


If there was any doubt that tuberculosis (TB) was fighting back, it was dispelled in 2005, at the Church of Scotland Hos-pital in the village of Tugela Ferry, South

Africa. Doctors at the hospital, in a rough, remote corner of KwaZulu-Natal province, were hardened to people dying from gunshots and AIDS. But even they were puzzled and frightened when patients with HIV who were responding well to antiretroviral drugs began dying — rapidly — from TB.

With ordinary TB, patients start to feel better after a few weeks or months on a sel-ection of four mainstay antibiotics. But of the 542 people with TB at the hospital in 2005 and early 2006, 221 (41%) had a multi-drug-resistant (MDR) form, against which these therapies are mostly powerless. Worse, 53 of them did not even respond to the few

antibiotics that form a second line of defence. Eventually, doctors had nothing left to try: all but one of the 53 died, half of them within 16 days of diagnosis. It was the first major out-break of what became known as extensively drug-resistant (XDR) TB — and a wake-up call to the world that TB had taken a turn for the worse1.

In the early 1980s, TB cases had dropped to such low rates that Western policy-makers frequently talked of eradication of the disease. Then came the HIV epidemic, which triggered a resurgence of TB in the late 1990s. But the latest report on TB from the World Health Organization (WHO), published in October, revealed signs of progress against normal — or drug-sensitive — cases of the bacterial disease. New infections have fallen and the mortality rate has dropped by 41% since 1990. But, the

TB’S REVENGEJ.

MAT

THEW

S/P

AN

OS

B Y L E I G H P H I L L I P S

A chest X-ray from a patient with tuberculosis (TB) in Lira, Uganda. Uganda is one of 22 countries accounting for roughly 80% of new TB cases each year.

The world is starting to win the war against tuberculosis, but drug-resistant forms pose a new threat.

1 4 | N A T U R E | V O L 4 9 3 | 3 J A N U A R Y 2 0 1 3

FEATURENEWS


report warned, “drug-resistant TB threatens global TB control”. Some 3.7% of new cases and 20% of previously treated cases are MDR-TB. And whereas in 2000 the highest incidence of MDR-TB was 14%, in Estonia; in 2010 that figure had jumped to 35%, in Russia’s Arkhan-gelsk province. An estimated 9% of drug-resistant cases are XDR-TB, which has now been reported in 84 countries.

It is a tale of two TBs. Once detected, drug-sensitive TB is almost always treatable, as long as the appropriate drugs are provided and taken. Simple practices — such as check-ing that patients take their medicine — can be transformative. But in some countries, par-ticularly in eastern Europe, Asia and Africa, the weakening or collapse of health-care systems over the past two decades has meant that patients do not always finish their drugs, or they take the wrong ones, allowing highly transmissible, drug-resistant strains to emerge and spread.

Drug-resistant TB is harder, more expensive and more time-consuming to treat. New tools are needed — but there have been no new anti-TB drugs in more than 50 years, and the current vaccine is largely ineffective. The most common diagnostic technique — analysing sputum samples under a microscope — can determine that Mycobacterium tuberculosis bacteria are present but not whether they are drug resistant. Meanwhile, researchers have lacked interest in developing drugs and tests, and drug companies have lacked market incentives to do so.

The growth of multi-drug resistance is an “escalating public-health emergency”, says Grania Brigden, TB adviser for Médecins Sans Frontières (Doctors Without Borders) in Geneva, Switzerland: “With barely 1 in 20 TB patients being tested for drug resistance, we’re just seeing the tip of the iceberg.”

But scientists are careful to temper their alarm. In the past decade, researchers and policy-makers have fought for and won a rever-sal in funding and attention for TB. Several new drugs are in development, and progress is being made towards an effective vaccine.

“I do worry when people stand up at conferences and talk about MDR-TB and say it’s a big disaster and the whole world is going to collapse. It’s not that severe yet,” says Tim McHugh, head of the Centre for Clinical Microbiology at University College London, who leads a team that is trialling one of the two most advanced candidates for new TB drugs. “The big anxiety is that if we don’t act now, it will easily run away from us.”

RISE AND FALLTB is one of the world’s leading killers, stealing 1.4 million lives and causing 8.7 million new and relapse infections in 2011. One-third of the world’s population carries the bacterium, but most will never develop the active form of the disease.

The first modern TB epidemic took off

in the late 1700s, during the Industrial Revolution. Rural workers in Europe and North America moved in droves to cities, where poverty — and related malnutrition and overcrowding — created an ideal environ-ment for the disease’s spread. But as hygiene, nutrition and medicine improved, what was known as the Great White Plague began to ebb.

“By the 1940 and 50s, things looked quite bright,” says McHugh, who seems almost as interested in the history of TB as in its micro-biology. The Bacillus Calmette–Guérin (BCG) vaccine, first used in the 1920s, helped. But BCG is now effective mainly against child-hood TB, which is not infectious, rather than the adult form. What really broke TB’s back was the introduction of isoniazid, in 1952, and then rifampicin, in the 1970s. “If you look at a graph of TB from the 1950s onward, [infection rates] just collapsed,” McHugh says.

Then, in the 1980s and 1990s, HIV hit. “You can’t underestimate the importance of HIV,” McHugh says. A co-infection of TB and HIV produces a powerful biological synergy, accel-erating the breakdown of the body’s immune defences; latent TB infection is 20–30 times more likely to become active in people who have HIV. In 1993, the WHO declared TB a global emergency. Worldwide, TB is now the leading cause of death among people with HIV.

The resurgence of ordinary TB set the stage for drug-resistant forms. Resistance develops when people do not stick to their drug regimens — which typically last six months for drug-sensitive TB and 20 months for MDR-TB — allowing naturally occurring resistant mutants to grow and evolve. MDR-TB, which grew more threatening during the 1990s, is resistant to iso-niazid and rifampicin. People with this form require second-line drugs — broad-spectrum antibiotics called fluoroquinolones or injectable

agents (amikacin, capreomycin and kanamy-cin). These treatments are less effective, more toxic and take many more months to work than first-line therapies. An infection is classified as XDR-TB if it is also resistant to fluoroquin-olones and at least one of the injectables. The XDR-TB outbreak at Tugela Ferry, when it was reported in 2006, rattled the TB research and policy community.

Experts agree that the biggest driver for the growth in drug-resistant TB has been the deterioration in some countries’ health-care infrastructures, including TB programmes, since the 1990s — particularly in the former Soviet bloc. This decline has meant that patients are not diagnosed and treated; and in some countries, over-the-counter availabil-ity of anti-TB drugs also encourages people to take inappropriate second-line therapies,

accelerating the growth of drug resistance. In a stroke of bad luck, the virulent and often drug-resistant ‘Beijing’ strain of TB, identi-fied in 1995 in China, swept through Russia and eastern Europe just as the region’s public-health provision was being dismantled. “There was a confluence of the biology of the organ-ism and its progress into Russia, where lots of people had a health-care system that was col-lapsing around their ears,” says McHugh. (In 2010, the Beijing strain was found in around 13% of active TB infections worldwide.) All of this helps to explain why the recent WHO report shows the highest burden of MDR TB in Russia’s Arkhangelsk province and in Belarus, Estonia, Kazakhstan, Kirghizia and Moldova (see ‘Two faces of TB’).

FIGHTING BACKOver the past decade or so, the paths of drug-sensitive and drug-resistant TBs have diverged. The solution for drug-sensitive TB is simply to deliver the drugs and diagnos-tics to patients — and the drive to do so has grown. One of the United Nations Millennium Development Goals set in 2000 was to halt and begin to reverse the incidence of TB by 2015; in 2001, the international Stop TB Partnership was established, bringing together government programmes, researchers, charitable foun-dations, non-governmental organizations (NGOs) and the private sector.

One major result of these and other efforts was a global expansion of ‘directly observed treatment, short course’ (DOTS), a strategy promoted by the WHO to combat drug-sensi-tive TB. Once diagnosed, the disease is treated with a supply of first-line drugs, taken under the close observation of health-care workers to ensure that people finish the course. Thanks in large part to such efforts, the WHO says that

the world is on track to halve TB mortality from 1990 levels by 2015.

Tackling drug-resistant TB, however, will require not just the rebuilding of health-care infrastructure, but also new weapons, such as diagnostics, drugs and vaccines. The private sector has had little incentive to invest in basic research whose eventual products, if any emerge, would largely be sold at low cost in poor countries. “They have to look at the bottom line,” says Anthony Fauci, head of the US National Institute of Allergy and Infectious Diseases in Bethesda, Maryland.

In the past decade or so, global TB pro-grammes have pumped money into research. One major development came in 1998, when researchers at the Wellcome Trust Sanger Institute in Hinxton, UK, published the genome sequence of M. tuberculosis,

“WITH BARELY 1 IN 20 PATIENTS WITH TB BEING TESTED FOR DRUG RESISTANCE, WE’RE JUST SEEING THE TIP OF THE ICEBERG.”

3 J A N U A R Y 2 0 1 3 | V O L 4 9 3 | N A T U R E | 1 5

FEATURE NEWS


allowing researchers to identify and study genes that underlie the bacterium’s virulence and ability to evade the immune system2. In 2012, the US National Institutes of Health in Bethesda started a bigger genome-sequencing project that aims to uncover the genetic roots of drug resistance. “We’ll use next-generation sequencing technologies to sequence 1,000 TB clinical isolates from around the world — South Africa, Korea, Russia, Uganda — anywhere drug-resistant TB is heavily present,” says Fauci.

WAR CHESTThere are now ten TB drugs in clinical trials. The aim is to find compounds that are effect ive against resistant strains and that work faster and have fewer side effects, so that patients will be more likely to finish the course. McHugh and his team, for example, are running a clinical trial at sites across Africa and Asia to test the antibiotic moxifloxacin, which is commonly used for pneumonia and skin infections. (They expect to release prelim-

inary results in 2013.) The researchers are also working to speed up the screening process for potential drugs by using mycobacterial species that are less pathogenic and more fecund than M. tuberculosis, which is slow-growing, finicky and poses a biosecurity risk. “Previously, what you had was a chemist who says ‘I’ve got this molecule that will kill Escherichia coli. I’m fairly sure it should kill TB. But there’s nowhere I can see if it does,’” McHugh says.

Accurate and fast diagnostic tests for drug-resistant strains are also a key part of the fight, and a number of tests have come online

in the past five years. One, called GeneXpert, takes 90 minutes to complete and is based on a gene-amplification technique that detects DNA sequences specific to M. tuberculosis and to rifampicin resistance. The system has been endorsed by the WHO and subsidized by a coa-lition of organizations, but researchers are still seeking simpler, cheaper options.

Only better vaccines will solve the problem for good. “We must invest in vaccine research if our ultimate goal is to be able to prevent the disease rather than forever chase growing drug resistance,” says Helen McShane, a vaccine researcher at the University of Oxford, UK.

In 2008, the European Commission pushed for the creation of the TB Vaccine Initiative, which draws funding from Euro-pean countries, NGOs and private funders. These and other efforts have helped to boost the number of vaccine candidates from 0 to 12 since 2000.

McShane and her team are on the cusp of the first efficacy results for MVA85A, one

of the most clinically advanced TB vaccines in the pipeline at present. The shot, which McShane helped to develop as a PhD stu-dent 15 years ago, contains a virus designed to ramp up the activity of T cells that have already been primed by BCG. In 2009, in part-nership with the South African Tuberculosis Vaccine Initiative, McShane launched a major phase II clinical trial on nearly 3,000 BCG-vaccinated babies in South Africa; early results are expected in the first quarter of 2013. In parallel, she and her colleagues are also testing the vaccine’s efficacy in HIV-infected adults in

South Africa and Senegal. But are these efforts enough? “Unfortunately

not,” concludes Karin Weyer, coordinator of laboratories, diagnostics and drug resistance at the WHO Stop TB Department in Geneva. Annual funding for TB diagnosis and treat-ment is expected to reach some US$4.8 billion in 2013 — but TB care and control are expected to demand up to $8 billion a year by 2015. The $600 million contributed to TB research in 2010 also falls well short of the $2 billion the WHO estimates will be needed annually — and the economic crisis has slowed financing across the board. “I need to be and want to be optimistic,” says Weyer. “But we’re still working with shoestring budgets compared to HIV.”

Meanwhile, the bacterium is not resting. In December last year, clinicians in Mumbai, India, reported3 the identification of 12 patients with what they termed totally drug-resistant TB, or TDR-TB. Similar claims had been made a few years earlier in Italy and Iran, but this time the WHO took it seriously enough to investigate. In March 2012, 40 experts convened by the WHO concluded that there was not enough evidence to say that TDR-TB was substantially different from XDR-TB.

McHugh agrees. But he does not need fur-ther evidence to act. In the face of marching drug resistance, it is the responsibility of researchers to speak out, he says. “I think we can no longer be scientists in our labs doing fascinating stuff and think we’re doing good work. We have to evangelize a little bit too.” ■

Leigh Phillips was an International Development Research Council fellow with Nature until October 2012.

1. Gandhi, N. R. et al. Lancet 368, 1575–1580 (2006).2. Cole, S. T. et al. Nature 393, 537–544 (1998).3. Udwadia, Z. F., Amale, R. A., Ajbani, K. K. &

Rodrigues, C. Clin. Infect. Dis. 54, 579–581 (2012).

PERCENTAGE OF NEWTB CASES THAT AREMULTI-DRUG RESISTANT

No dataSubnational data onlyNot applicable

0–2.9

6–11.912–17.9≥18

3–5.9

TWO FACES OF TBE�orts to tackle drug-susceptible strains have started to cut the global incidence of TB (left), but collapsing health-care systems in the former Soviet bloc have helped drug-resistant strains to emerge and spread.

Co-infection with HIV

200

150

100

50

0

Rat

e per

100,0

00 p

eople

1990 1995 2000 2005 2010

ESTIMATED INCIDENCE OF TB

Total globalincidence

SO

UR

CE:

WH

O

“WE MUST INVEST IN VACCINE RESEARCH IF OUR ULTIMATE GOAL IS TO BE ABLE TO PREVENT THE DISEASE.”

1 6 | N A T U R E | V O L 4 9 3 | 3 J A N U A R Y 2 0 1 3

FEATURENEWS


ARTICLEdoi:10.1038/nature12337

The Mycobacterium tuberculosisregulatory network and hypoxiaJames E. Galagan1,2,3,4, Kyle Minch5*, Matthew Peterson1*, Anna Lyubetskaya3*, Elham Azizi3*, Linsday Sweet6*,Antonio Gomes3*, Tige Rustad5, Gregory Dolganov7, Irina Glotova3, Thomas Abeel4,8, Chris Mahwinney1, Adam D. Kennedy9,Rene Allard10, William Brabant5, Andrew Krueger1, Suma Jaini1, Brent Honda1, Wen-Han Yu1, Mark J. Hickey5, Jeremy Zucker4,Christopher Garay1, Brian Weiner4, Peter Sisk4, Christian Stolte4, Jessica K. Winkler5, Yves Van de Peer8, Paul Iazzetti1,Diogo Camacho1, Jonathan Dreyfuss1, Yang Liu7, Anca Dorhoi11, Hans-Joachim Mollenkopf12, Paul Drogaris10, Julie Lamontagne10,Yiyong Zhou10, Julie Piquenot10, Sang Tae Park2, Sahadevan Raman2, Stefan H. E. Kaufmann11, Robert P. Mohney9,Daniel Chelsky10, D. Branch Moody6, David R. Sherman5,13 & Gary K. Schoolnik7,14

We have taken the first steps towards a complete reconstruction of the Mycobacterium tuberculosis regulatory networkbased on ChIP-Seq and combined this reconstruction with system-wide profiling of messenger RNAs, proteins, metabolitesand lipids during hypoxia and re-aeration. Adaptations to hypoxia are thought to have a prominent role in M. tuberculosispathogenesis. Using ChIP-Seq combined with expression data from the induction of the same factors, we have reconstructeda draft regulatory network based on 50 transcription factors. This network model revealed a direct interconnection betweenthe hypoxic response, lipid catabolism, lipid anabolism and the production of cell wall lipids. As a validation of this model, inresponse to oxygen availability we observe substantial alterations in lipid content and changes in gene expression andmetabolites in corresponding metabolic pathways. The regulatory network reveals transcription factors underlying thesechanges, allows us to computationally predict expression changes, and indicates that Rv0081 is a regulatory hub.

Mycobacterium tuberculosis (MTB) has been associated with humandisease for thousands of years and its success is due in part to theability to survive within the host for months to decades in an asymp-tomatic state. The mechanisms underlying this persistence in the hostare poorly understood, although adaptations to hypoxia are thoughtto have a prominent role1,2. Hypoxia produces widespread changes inthe bacterium and induces a non-replicating state characterized byphenotypic drug tolerance. Within the host, MTB also shifts to lipids,including cholesterol, as a primary nutrient3–6. Lipid catabolism is, inturn, linked to the biosynthesis of lipids that serve as energy stores,factors associated with virulence and immunomodulation, and com-ponents of the unique and complex cell wall of MTB7–9.

The regulatory mechanisms underlying these and other adaptationsare largely unknown, as functions for only a small fraction of the 1801MTB transcription factors (TFs) are known, direct DNA binding dataexist for only a handful of sites, and the interactions between TFs nece-ssary for complex behaviour have not been studied. We also lack a compre-hensive understanding of the cellular changes underlying pathogenesis,with existing studies typically focused on specific molecular compo-nents that can be difficult to integrate with results from other studies.To address these challenges, we have performed a systems analysis ofthe MTB regulatory and metabolic networks, with an emphasis on hyp-oxic conditions thought to contribute to MTB persistence in the host.

Mapping and functional validation of TF binding sitesTo systematically map TF binding sites, we performed chromatinimmunoprecipitation followed by sequencing (ChIP-Seq)10–12 using

Flag-tagged transcription factors episomally expressed under controlof a mycobacterial tetracycline-inducible promoter13–15 (Supplemen-tary Fig. 1). The inducible promoter system allows us to study all MTBTFs in a standard and reproducible reference state without a prioriknowledge of the conditions that normally induce their expression.Using a custom pipeline (Supplementary Fig. 2 and SupplementaryTable 1) we identified binding sites in regions of enrichment with highspatial resolution. Using this method, we mapped 50 TFs. We com-pared the results with previous reports for two well-studied regulatorsfor which strong evidence for direct binding exists: the activator DosR(Rv3133c) and the repressor KstR (Rv3574).

Our method shows high sensitivity and reproducibility. We identi-fied all known direct binding regions for DosR (Supplementary Fig. 3)and KstR (Fig. 1a) and recovered the known motifs for these fac-tors (Supplementary Material). Coverage for enriched sites is highlycorrelated between replicates (Fig. 1b and Supplementary Fig. 4).There is also high reproducibility in binding location, with distancesbetween replicate binding sites less than the length of predicted bind-ing site motifs for the vast majority of sites (Fig. 1b). Moreover, for 11different TFs we also see substantial concordance between bindingobserved in normoxia and binding observed in hypoxia (Supplemen-tary Fig. 5).

ChIP enrichment is a function of the number of cells in which a siteis bound16 which in turn is governed by the affinity of the site and theconcentration of the factor. Thus, increasing TF induction was pre-dicted to increase the occupancy of strong sites up to a saturation limitwhile occupying weaker affinity sites. This is confirmed by comparing

*These authors contributed equally to this work.

1Department of Biomedical Engineering, Boston University, Boston, Massachusetts 02215, USA. 2Department of Microbiology, Boston University, Boston, Massachusetts 02215, USA. 3BioinformaticsProgram, Boston University, Boston, Massachusetts 02215, USA. 4The Eli and Edythe L. Broad Institute of Harvard and MIT, Cambridge, Massachusetts 02142, USA. 5Seattle Biomedical Research Institute,Seattle, Washington 98109, USA. 6Division of Rheumatology, Immunology and Allergy, Brigham and Women’s Hospital and Harvard Medical School, Boston, Massachusetts 02115, USA. 7Departments ofMedicine and of Microbiology and Immunology, Stanford Medical School, Stanford, California 94305, USA. 8Department of Plant Biotechnology and Bioinformatics, Ghent University, 9052 Gent, Belgium.9Metabolon Inc., Durham, North Carolina 27713, USA. 10Caprion Proteomics, Inc., Montreal, Quebec H4S 2C8, Canada. 11Department of Immunology, Max Planck Institute for Infection Biology, 10117Berlin, Germany. 12Microarray Core Facility, Max Planck Institute for Infection Biology, 10117 Berlin, Germany. 13Interdisciplinary Program of Pathobiology, Department of Global Health, University ofWashington, Seattle, Washington 98195, USA. 14Division of Infectious Diseases and Geographic Medicine, Department of Medicine, Stanford Medical School, Stanford, California 94305, USA.

1 7 8 | N A T U R E | V O L 4 9 9 | 1 1 J U L Y 2 0 1 3

Macmillan Publishers Limited. All rights reserved©2013

www.nature.com/doifinder/10.1038/nature12337

ChIP-Seq experiments after inducing three different factors to differ-ent expression abundances (Fig. 1c, Supplementary Fig. 6 and Sup-plementary Fig. 7).

Consistent with this observation, at the highest levels of TF induc-tion we identify more binding sites than previously reported for DosRand KstR (Fig. 1a); most, but not all, of these newly-identified siteshave lower ChIP-Seq coverage than the majority of previously iden-tified sites. Abundant binding of transcription factors, particularly tolow affinity sites, has been reported in yeast, worm, fly and mamma-lian cells16–18 but, to our knowledge, these data represent the first large-scale observation in a prokaryote. We have confirmed that many novelsites can be bound at physiological levels of these TFs, and that sitesshow sequence specificity for each TF. In addition, for DosR, nearly allnovel sites are also found when performing ChIP using anti-DosR anti-bodies in a wild-type background (Supplementary Material Section 2.4).

To assess the degree to which binding is associated with transcrip-tional regulation, we performed transcriptomic analysis from the samecultures in which regulators were induced for ChIP-Seq. Using thesedata we developed a procedure for determining the possible regulatoryroles of identified binding sites (Supplementary Fig. 11). This methodidentified a regulatory effect for 92% and 80% of previously identifiedDosR and KstR sites, respectively, and associated regulation with 43%and 36% of new DosR and KstR binding sites revealed using ChIP-Seq(false discovery rate (FDR) 5 0.15). Many, but not all, newly identifiedsites show weaker ChIP-Seq enrichment, indicating evidence for reg-ulatory effects of weak binding even for well-studied regulators19–21.This was corroborated by knockout expression data for these TFs(Supplementary Fig. 12).

Applying our method to all peaks from all 50 TFs, we could assign apotential regulatory role to 25% of peaks within 1,000 base pairs (bp)

on either side of the site (FDR 5 0.15; 18% of sites were significantwith q value 5 0) (Fig. 1e). Stronger binding sites are more often asso-ciated with regulation than weaker sites, independent of window size,suggesting a possible correlation between binding strength and regu-latory impact (Supplementary Fig. 13). Such a correlation could explainwhy the stronger sites have been reported, as they would be more easilydetected. The use of a 1-kilobase (kb) window ensures that predictionsare not a priori biased to proximal promoter regions. However, evenwith 4-kb windows, the distance between binding sites and associatedtarget genes is consistent with expectation: binding sites are typicallylocated within 500 bp of the start codon of the predicted regulatedgene (Fig. 1f), with 24% located in the upstream intergenic region. Bycontrast, 76% of sites fall into annotated coding regions and a significantproportion are associated with regulation. Extensive genic binding hasbeen reported17,18 and there remains no consensus on its functionalsignificance. Prokaryotic binding sites have been largely mapped withlower resolution ChIP-Chip that frequently show broad binding over-lapping both genic and intergenic regions22. Our method detects bind-ing at high spatial resolution and indicates that some genic binding mayreflect the extension of promoter regions into upstream genes, alterna-tive promoter regions within genes, or errors in the current annotationof genic regions. As with previous reports17, we cannot assign regulatoryroles to all detected binding sites (Supplementary Fig. 13). We discusspotential issues with false positives and negatives in SupplementaryMaterial.

We also tested the degree to which observed binding could beused to develop models predictive of gene expression. We developedcomputational models relating the expression of target genes to theexpression of TFs predicted to bind the target (Supplementary Fig. 14).The relationship between TFs and target genes was parameterized

f

a b

c d

0 10 20 30 40 50 60 70 80 90 1000

100

150

200

Distance between binding sites (bp)

Num

ber

of

bin

din

g s

ites

Motif length

0

30

40

50

60

70

1 51 101 151 201 251

Lo

g(p

eak h

eig

ht/

med

ian c

overa

ge)

0

10

20

30

40

50

60

70

80

90

Co

vera

ge/m

ed

ian, exp

t 2

0 50 100 150 200

Coverage/median, expt 1

Binding peaks, ordered by height

Motif comparison scoreChIP-Seq region height

Sco

re k

no

wn v

s n

ovel m

otif

R2 = 0.9025

100

101

102

103

10

20

0 50 100 150 200 250100

101

102

103

Lo

g(p

eak h

eig

ht/

med

ian c

overa

ge)


100 ng10 ng1 ng0.1 ng

Lo

g(p

eak h

eig

ht/

med

ian c

overa

ge)

0 50 100 150 200 250


KnownNewly identified

100

101

102

103

Peak detection

threshold

Ind

uctio

n

020406080

Perc

ent

of

sites a

ssig

ned

reg

ula

tio

n

Percent of highest peak for TF

All peaks

Intergenic peaks

Observed binding sites Random binding sites

Genic peaks

0 20 40 60 80 10010 30 50 70 90

100

020406080

100

020406080

100

e

0

20

40

60

80

100

120

140

160

180

Distance from binding to ATG

No

. in

tera

ctio

ns a

ssig

ned

reg

ula

tio

n

–2,000 2,000–1,000 1,0000

Target gene

Figure 1 | ChIP-Seq binding shows high sensitivity, reproducibility andsequence specificity. a, We identify all known binding sites (red bars) for KstRand DosR (Supplementary Fig. 3). Binding site heights plotted as bars andordered by peak height. b, Binding site identification is highly reproducible. Barplot shows the distance between corresponding sites in two KstR replicates. Themajority of replicates fall within the motif (cyan line). Inset shows correlation ofheights of corresponding peaks in two replicates (R2 . 0.83 for all TFs).c, Increasing TF expression increases peak height. Shown are plots of peaks

identified at different levels of KstR induction. Corresponding peaks are plottedat the same position on the horizontal axis. d, KstR binding peak heightcorrelated with motif structure. The canonical palindromic motif is identifiedin all strong binding sites. At weaker sites, however, we detect degraded motifs.e, Fraction of peaks assigned regulation as a function of relative peak height.f, Stacked histogram of the number of peaks assigned regulation as a function ofthe distance to the start codon of the predicted target gene and coloured bygenomic location relative to the target gene and genic or intergenic context.

ARTICLE RESEARCH

1 1 J U L Y 2 0 1 3 | V O L 4 9 9 | N A T U R E | 1 7 9


based on subsets of the overexpression data and tested on the remain-ing using cross-validation. We could generate models that predictmore accurately than random TF assignments for 28% of genes withbinding (positive false discovery rate (pFDR) , 0.15; SupplementaryTable 4). More importantly, as described below, we confirmed theability of these models to predict expression for genes in an indepen-dent data set.

An MTB regulatory network modelUsing the combination of binding site mapping and functional valida-tion via expression profiling, we analysed the regulatory interactionsof 50 TFs (26% of predicted MTB TFs). Our TF selection was weightedtowards those that respond to hypoxia or are associated with lipidmetabolism. By linking TFs with genes based on binding proximity(Supplementary Text) and potential regulation, we constructed theregulatory network model shown in Supplementary Fig. 15 (also Sup-plementary Fig. 16). The TB regulatory network model has topologi-cal features seen for other organisms (Supplementary Text), includingthe presence of ‘hubs’ or TFs that interact with many genes. Surpri-singly, Rv0081 forms the largest hub identified among the TFs reported,and interacts with another hub, Lsr2, an MTB analogue of the H-NSnucleoid binding protein23,24 (Supplementary Text).

The network also begins to reveal interactions between transcrip-tion factors mediating responses of MTB to its environment (Sup-plementary Material). Of particular interest is a subnetwork involvingresponses to altered oxygen status and lipid availability (Fig. 2). Theseresponses, among the most extensively studied in MTB, have beenviewed largely as separate phenomena. DosR and Rv0081 mediate theinitial response to hypoxia, whereas a larger stimulon termed the endu-ring hypoxic response (EHR) is induced later in hypoxia25. KstR con-trols a large regulon mediating cholesterol degradation and lipid andenergy metabolism26,27. KstR was identified as part of the EHR, but thebiology linking these responses was unclear.

We identified two potential regulators for KstR. Rv0081 is predic-ted to repress both Rv0324 and KstR, whereas Rv0324 is predicted toactivate KstR. Rv0081 is the only regulator in the initial hypoxic res-ponse apart from DosR, and our network identifies an interactionunderlying the known induction of Rv0081 by DosR. Rv0324 is aregulator associated with the EHR25.

We also identify several potential regulators of DosR: Rv2034,Rv0767c and PhoP (Rv0757). Rv2034 is an EHR regulator predictedto activate DosR, thus providing possible positive feedback from theenduring to the initial hypoxic response (during revision, this linkbetween Rv2034 and DosR was confirmed28). PhoP mediates a rangeof responses, including upregulating DosR29–31, although direct regu-lation of DosR by PhoP had not been previously demonstrated. PhoPbinding to DosR is the strongest among 50 TFs, providing a mecha-nism for this regulatory link and supporting the conclusion that regu-lation of hypoxia adaptation by PhoP is indirect through this connectionwith DosR29. PhoP also mediates pH adaptation and our data confirmdirect binding between PhoP and the aprABC locus required for this32.PhoP is known to modulate the production of virulence lipids and wepredict PhoP to bind upstream of and directly regulate WhiB3 (Rv3416),which codes for a redox-sensitive protein that directly regulates theproduction of these lipids33. In addition to PhoP, both Rv0081 and Lsr2also display binding to whiB3, with activation predicted by Rv0081.Taken together, the data reveal an interconnected subnetwork linkinghypoxic adaptation, lipid and cholesterol degradation, and lipid bio-synthesis (Supplementary Text).

Profiling and prediction during hypoxia and re-aerationTo broadly assess the changes associated with altered O2 availability,and assess the explanatory power of the regulatory network in theseresponses, we performed systems level lipidomic, proteomic, meta-bolomic and transcriptomics profiling of MTB during a time courseof hypoxia and subsequent re-aeration (Supplementary Fig. 17 and

Methods). We cultured MTB in a medium without detergent or exo-genous lipids. All measurements were normalized to baseline levelsbefore hypoxia, and integrated with a manually curated model of MTBmetabolism (Supplementary Fig. 18). We summarize key results hereand provide additional details and results in Supplementary Text.

Changes in oxygen availability result in expression changes to nearlyone-third of all MTB genes (Supplementary Fig. 19A). To identifytemporal trends and associate them with possible regulators, we clus-tered expression data into paths using DREM34 (Supplementary Text).We identified Rv0081 as a candidate high-level regulator broadly pre-dictive of the overall expression of sets of genes during hypoxia and re-aeration (Supplementary Fig. 19b). A broad regulatory role for Rv0081is thus supported by three independent sources of evidence: Rv0081overexpression in normoxia alters the expression of numerous genes,Rv0081 ChIP-Seq reveals a large number of binding sites which arealso detected during hypoxia (Supplementary Fig. 20), and the expres-sion and predicted regulatory role of Rv0081 correlates with the expres-sion of the genes it binds during hypoxia.

We next sought to assess the degree to which the regulatory net-work could be used to predict changes in the expression of individualgenes during hypoxia and re-aeration. We used the regression modelsdescribed above—parameterized by independent ChIP-Seq and TFoverexpression transcriptomics data (Supplementary Material)—andgenerated predictions that are significantly better than random for 66%of genes with significant changes. Examples are shown in Fig. 3 andSupplementary Fig. 21. In particular, we correctly predict the pattern ofexpression of KstR, confirming an implication of the network topology.

Rv0081

DosR

PhoP

ClgR

Lsr2

SigF

Redox andlipid biosynthesis Proteases

Hypoxia/lipids?

Hypoxia and redox adaptation

Enduring hypoxic response

Sigma factors

pH and lipidbiosynthesis

WhiB3

Chromatin

Cholesterol and fatty acid catabolism

Rv2034 Rv0494

Rv2021c

Rv0324

Rv3249c

Rv2324

KstR

Rv1990c

Rv2989 SigA

SigL

SigB SigE

1.0 0.5Line thickness : peak height/TF maximum

Days

Gene expression log(fold change)

1 785 923

Hypoxia Re-aeration

–2 –1 0 1 21214

(relative to day 0 / normoxia)

Figure 2 | TF regulatory interaction subnetwork linking hypoxia, lipidmetabolism and protein degradation. The figure shows a subset of theregulatory network model for selected transcription factors. Edges are colouredby z-score (see text) with red edges indicating positive z-scores and activation,and blue indicating negative z-scores and repression. Grey edges indicate linkswithout significant z-scores, TFs without induction expression data, orautobinding. The width of edges indicates the height of the correspondingbinding site relative to the maximum binding site for the corresponding TF.Selected TFs are colour-coded by functional association and heat maps showexpression data during hypoxia and re-aeration as shown in legend.

RESEARCH ARTICLE

1 8 0 | N A T U R E | V O L 4 9 9 | 1 1 J U L Y 2 0 1 3


Importantly, these data also indicate that the regulatory network, builtfrom a normoxic baseline, can generalize to hypoxia.

Alterations in lipid metabolismConsistent with predictions of the regulatory network during hypoxia,we found strong induction of genes associated with lipid catabolism andcholesterol degradation, including the regulator kstR (Fig. 3, Supplemen-tary Fig. 18 and Supplementary Fig. 22). KstR induction by hypoxia ispredicted by the core regulatory network. However, kstR is a repressor26

and kstR-repressed cholesterol degradation genes are among those indu-ced. KstR de-repression occurs during growth on cholesterol27. However,no cholesterol or other exogenous lipids are present in our medium.Follow-up studies suggest that de-repression of kstR may be due to fattyacids endogenous to MTB or their metabolites (Supplementary Text).

The accumulation of triacylglycerides (TAGs) during hypoxia andin TB patient sputum samples, and their utilization upon re-aeration,has been reported7,8,35. We also observe TAG accumulation during hyp-oxia and rapid depletion during re-aeration (Fig. 4). A detailed systemsview associated with these changes (Supplementary Text) suggests ascenario in which metabolites upstream of DAG decrease in produc-tion, and TAG accumulation results from conversion of existing DAGsto TAGs via triacylglyceride synthase. We also observe changes poten-tially related to TAG utilization. The regulatory network identifiesseveral regulatory links potentially relevant to these changes (Supplemen-tary Fig. 18). Induction of tgs1 by DosR is well established7,36,37, and weidentify this link. The network also identifies oxygen-responsive reg-ulators of tgs2 (Rv0081, Rv0324) and tgs4 (DosR, Rv0324) and ourmodels predict positive regulation of these genes in hypoxia by theseTFs (Fig. 3). Further, three of four lipase genes (Rv3176, Rv1169c andRv3097c) induced during hypoxia are influenced by regulators in thecore network, and in these three cases we are able to predict theirexpression profiles using our gene expression models (Fig. 3).

MTB uses methylmalonyl-CoA as a precursor to synthesize a com-plex set of surface-exposed methyl-branched lipids including acylatedtrehaloses (PAT/DAT), sulphoglycolipids (SGL) and phthiocerol dimy-cocerosates (PDIM), the latter two associated with virulence in murinemodels38–42. During hypoxia, the expression of biosynthetic genes forSGL, PAT/DAT, PDIMs and methylmalonyl are generally downregu-lated (Supplementary Fig. 18). Correspondingly, during hypoxia mass

spectral signals corresponding to diacylated sulphoglycolipid (AC2SGL)(a precursor to SL-1, the major SGL in MTB) and DATs seemed unal-tered, whereas ions corresponding to PDIMs showed a modest decline(Fig. 4, DATs not shown). Conversely, during re-aeration, we obser-ved induction of genes encoding enzymes in the methylmalonyl path-way. The activation of the methylcitrate cycle and accumulation ofmethylcitrate suggests the availability of precursors for methylmalo-nate. Consistent with this hypothesis, we see statistically significantincreases in AC2SGL (Fig. 4).

The regulation of the methylmalonyl pathway is partially explainedby the regulatory network. All three subunits of the propionyl-CoAcarboxylase (PCC) complex (AccA3, AccD5 and AccE5) are regula-ted by hypoxia regulators (Fig. 3). Both MutA and MutB also displayregulation by KstR and Lsr2. Regulation associated with methyl-branched lipid biosynthesis, in contrast, is complex. Whib3 is regu-lated by PhoP in the model, and both are known to modulate theproduction of PAT/DAT (via pks3) and SL (via pks2)29,33. Our net-work predicts a PhoP/WhiB3 FFL underlying this phenomenon, withPhoP regulating whiB3 and both regulating pks2/pks3 (Supplemen-tary Fig. 25). Similar regulatory complexity is seen for DIM, althoughregulation of key steps in DIM synthesis by Rv0081, PhoP, DosR andKstR is predicted.

Mycolyl glycolipids are important immunomodulatory componentsof the mycobacterial cell wall. As seen in other systems43–45, we observeincreases in free mycolic acids during hypoxia that are reversed duringre-aeration (Fig. 4). Conversely, we observe the opposite effects ontrehalose monomycolates (TMMs) (Fig. 4) and trehalose dimycolates(TDMs) (not shown). Similar effects have recently been reported forTDMs in Mycobacterium smegmatis during biofilm formation45 andTMMs in MTB during the transition into a dormant ‘‘non-culturable’’state induced by a potassium-free medium43.The rapid, reversible andnearly complete mobilization of glycosylated to free mycolates duringhypoxic dormancy is also compatible with decreased need to delivermycolic acids to non-dividing cells.

Concluding remarksThis report presents an initial step in the reconstruction of the MTBregulatory network, based on 50 TFs, and its integration with system-wide profiling of MTB during a time-course of hypoxia and re-aeration.

Lipid metabolism

Actual Predicted Hypoxia Re-aeration

No

rmaliz

ed

exp

ressio

n

Iron/sulphur oxidation

atpH (Rv1307)

narI (Rv1164)

narX (Rv1736c)

cyp123 (Rv0766c)

2 4 6 8 10 12 14

Central metabolism

sucA (Rv1248c)

acn (Rv1475c)

gpm1 (Rv0489)

Rv2419c

2 4 6 8 10 12 14

Time (days)

tgs1 (Rv3130c)

fas1 (Rv2524c)

inhA (Rv1484)

lipY (Rv3097c)

2 4 6 8 10 12 14

otsB1 (Rv2006)

accE5 (Rv3281)

accD4 (Rv3799c)

fabG1 (Rv1483)

2 4 6 8 10 12 14

TFs

0

0.5

1.0kstR (Rv3574)

ideR (Rv2711)

Rv3855

Rv0767

0

0.5

1.0

0

0.5

1.0

2 4 6 8 10 12 140

0.5

1.0

Figure 3 | Predicting gene expression during hypoxia and re-aeration.Using the models described in text, we predict the expression pattern of 66% ofgenes (533) whose expression changes during hypoxia and re-aeration. Selected

examples shown. Green lines, actual scaled expression with error bars fromreplicates; dashed black lines, model-predicted expression.

ARTICLE RESEARCH

1 1 J U L Y 2 0 1 3 | V O L 4 9 9 | N A T U R E | 1 8 1


Although necessarily incomplete, the regulatory network confirmspreviously known physical interactions, provides possible mechan-isms for known regulatory interactions, provides a framework for re-interpreting existing data, and identifies network motifs thought tounderlie dynamic behaviour. The predictive models take a first steptowards systems modelling, and integration of the network modelwith profiling data provides new insight about the physiological con-sequences of regulatory programs induced by changes in oxygenavailability—a perturbation relevant to host adaptation. The resultsprovide a foundation for ongoing efforts to map the complete tran-scriptional regulatory network, and to extend it to include signallingand non-coding RNAs46. The results presented here identify compel-ling questions for further investigation (Supplementary Text). Studiesnow focus on determining how the in vitro network connections andphysiological changes identified here relate to adaptations of the mic-robe in the intracellular environment of the macrophage.

METHODS SUMMARYMTB H37Rv was used for all experiments with the single exception of one experi-ment performed in M. smegmatis (Supplementary Fig. 21). This MTB strain wasfully sequenced by the Broad Institute (GI:397671778). For Chip-Seq, cells werecultured in Middlebrook 7H9 with ADC (Difco), 0.05% Tween80, and 50 mg ml21

hygromycin B at 37 uC with constant agitation and induced with 100 ng ml21

anhydrotetracycline (ATc) during mid-log-phase growth, and ChIP was perfor-med using a protocol optimized for mycobacteria and related Actinomycetes. Forthe hypoxia and re-aeration time-course, bacilli were cultured in bacteriostaticoxygen-limited conditions (1% aerobic O2 tension) for seven days, followed by re-aeration. Bacteria were cultured in Sauton’s medium without detergent or exoge-nous lipid source. Profiling samples were collected as described in the Supplementary

Text. All data available at http://TBDB.org. Expression data also available atGEO (accession number GSE43466).

Received 30 December 2011; accepted 23 May 2013.

Published online 3 July 2013.

1. Manabe, Y. C. & Bishai, W. R. Latent Mycobacterium tuberculosis-persistence,patience, and winning by waiting. Nature Med. 6, 1327–1329 (2000).

2. Flynn, J. L. & Chan, J. Tuberculosis: latency and reactivation. Infect. Immun. 69,4195–4201 (2001).

3. Schnappinger, D. et al. Transcriptional adaptation of Mycobacterium tuberculosiswithin macrophages: insights into the phagosomal environment. J. Exp. Med. 198,693–704 (2003).

4. Yang, X., Nesbitt, N. M., Dubnau, E., Smith, I. & Sampson, N. S. Cholesterolmetabolism increases the metabolic pool of propionate in Mycobacteriumtuberculosis. Biochemistry 48, 3819–3821 (2009).

5. Miner, M. D., Chang, J. C., Pandey, A. K., Sassetti, C. M. & Sherman, D. R. Role ofcholesterol inMycobacterium tuberculosis infection. Indian J. Exp. Biol. 47, 407–411(2009).

6. Chang, J. C. et al. igr genes and Mycobacterium tuberculosis cholesterolmetabolism. J. Bacteriol. 191, 5232–5239 (2009).

7. Daniel, J., Maamar, H., Deb, C., Sirakova, T. D. & Kolattukudy, P. E. Mycobacteriumtuberculosis uses host triacylglycerol to accumulate lipid droplets and acquires adormancy-like phenotype in lipid-loaded macrophages. PLoS Pathog. 7,e1002093 (2011).

8. Low, K. L. et al. Triacylglycerol utilization is required for regrowth of in vitro hypoxicnonreplicating Mycobacterium bovis bacillus Calmette-Guerin. J. Bacteriol. 191,5037–5043 (2009).

9. Russell, D. G., Mwandumba, H. C. & Rhoades, E. E. Mycobacterium and the coat ofmany lipids. J. Cell Biol. 158, 421–426 (2002).

10. Robertson, G. et al. Genome-wide profiles of STAT1 DNA association usingchromatin immunoprecipitation and massively parallel sequencing. NatureMethods 4, 651–657 (2007).

11. Mikkelsen, T. S. et al. Genome-wide maps of chromatin state in pluripotent andlineage-committed cells. Nature 448, 553–560 (2007).

0

500,000

1,000,000

1,500,000 C47:0C48:1C48:0C49:0

C51:1C51:0C52:1

C50:1

C52:0C54:1C56:0C57:0C58:0C60:0C61:0

0

2,000,000

4,000,000

6,000,000

8,000,000C33:0C34:2C34:1C35:1

C34:0C35:0C36:1

0

10,000,000

20,000,000

30,000,000

40,000,000

50,000,000C33:0C34:0C35:0C36:0

0

500,000

1,000,000

1,500,000

2,000,000C87C89C90

C91C92

C94C95C96C97

C98C100

0

50,000

100,000

C76:0 alphaC78:0 alphaC80:0 alphaC82:0 alphaC82:0 ketoC84:0 ketoC86:0 ketoC88:0 ketoC83:0 methoxyC85:0 methoxyC87:0 methoxyC89:0 methoxyC91:0 methoxy

0 1 2 3 4 5 6 7 8 9 10 11 12 13 140

50,000

100,000C70:0 +H+

C70:0 +NH4+

C70:0 +Na+

C72:0 +NH4+

C72:0 +Na+

C73:0 +NH4+

0 1 2 3 4 5 6 7 8 9 10 11 12 13 140

50,000

100,000

150,000C78:0 alphaC80:0 alpha

Triacylglycerides

Diacylglycerides

Phosphatidylinositol

0

10,000,000

20,000,000

30,000,000C35:0+H+

C35:0+NH4+

Phosphatidylethanolamine

Phthiocerol dimycocerosates

Diacylated sulphoglycolipids

Trehalose monomycolates

Free mycolates

Ab

so

lute

peak a

rea

Time (days)Hypoxia Re-aeration

Figure 4 | Lipid changes during hypoxia and re-aeration. HPLC-MS of totallipids from M. tuberculosis analysed in the positive-ion mode as ammoniatedadducts unless otherwise indicated. Among more than 5,000 ions detected ateach time point, m/z values for unnamed lipids were converted to named lipidswhen they matched the masses (,10 p.p.m.) retention time (,1 min) andcollisional mass spectrometry patterns in MycoMass and MycoMap databases.Within each lipid class individual molecular species are reported by intensity

and tracked by mass, converted to deduced empiric formulas and reportedseparately corresponding to the R group variants of mycolic acids (alpha, keto,methoxy) and as CX:Y, where X is the alkane chain length and Y is theunsaturation in the combined fatty acyl, mycolyl, phthioceranyl, pthiocerol,mycocerosyl units of one molecule. Error bars are standard deviations fromfour replicates.

RESEARCH ARTICLE

1 8 2 | N A T U R E | V O L 4 9 9 | 1 1 J U L Y 2 0 1 3


http://TBDB.org

http://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE43466

12. Johnson, D. S., Mortazavi, A., Myers, R. M. & Wold, B. Genome-wide mapping of invivo protein-DNA interactions. Science 316, 1497–1502 (2007).

13. Ehrt, S. et al. Controlling gene expression in mycobacteria withanhydrotetracycline and Tet repressor. Nucleic Acids Res. 33, e21 (2005).

14. Ehrt, S. & Schnappinger, D. Controlling gene expression in mycobacteria. FutureMicrobiol. 1, 177–184 (2006).

15. Klotzsche, M., Ehrt, S. & Schnappinger, D. Improved tetracycline repressors forgene silencing in mycobacteria. Nucleic Acids Res. 37, 1778–1788 (2009).

16. Farnham, P. J. Insights from genomic profiling of transcription factors. Nature Rev.Genet. 10, 605–616 (2009).

17. MacQuarrie, K. L., Fong,A.P.,Morse,R. H.&Tapscott, S. J.Genome-wide transcriptionfactor binding: beyond direct target regulation. Trends Genet. 27, 141–148 (2011).

18. Galagan, J., Lyubetskaya, A. & Gomes, A. ChIP-Seq and the complexity of bacterialtranscriptional regulation. Curr. Top. Microbiol. Immunol. 363, 43–68 (2013).

19. Chauhan, S., Sharma, D., Singh, A., Surolia, A. & Tyagi, J. S. Comprehensive insightsinto Mycobacterium tuberculosis DevR (DosR) regulon activation switch. NucleicAcids Res. 39, 7400–7414 (2011).

20. Gautam, U. S., Chauhan,S. & Tyagi, J. S. Determinants outside the DevRC-terminaldomain are essential for cooperativity and robust activation of dormancy genes inMycobacterium tuberculosis. PLoS ONE 6, e16500 (2011).

21. Vasudeva-Rao, H. M. & McDonough, K. A. Expression of the Mycobacteriumtuberculosis acr-coregulated genes from the DevR (DosR) regulon is controlled bymultiple levels of regulation. Infect. Immun. 76, 2478–2489 (2008).

22. Cho, B. K., Federowicz, S., Park, Y. S., Zengler, K. & Palsson, B. O. Deciphering thetranscriptional regulatory logic of amino acid metabolism. Nature Chem. Biol. 8,65–71 (2012).

23. Colangeli, R. et al. The multifunctional histone-like protein Lsr2 protectsmycobacteria against reactiveoxygen intermediates.Proc.NatlAcad. Sci.USA106,4414–4418 (2009).

24. Gordon, B. R. et al. Lsr2 is a nucleoid-associated protein that targets AT-richsequences and virulence genes in Mycobacterium tuberculosis. Proc. Natl Acad. Sci.USA 107, 5154–5159 (2010).

25. Rustad, T. R., Harrell, M. I., Liao, R. & Sherman, D. R. Theenduring hypoxic responseof Mycobacterium tuberculosis. PLoS ONE 3, e1502 (2008).

26. Kendall, S. L. et al. A highly conserved transcriptional repressor controls a largeregulon involved in lipid degradation in Mycobacterium smegmatis andMycobacterium tuberculosis. Mol. Microbiol. 65, 684–699 (2007).

27. Nesbitt, N. M. et al. A thiolase of Mycobacterium tuberculosis is required forvirulence and production of androstenedione and androstadienedione fromcholesterol. Infect. Immun. 78, 275–282 (2010).

28. Gao, C. H., Yang, M. & He, Z. G. Characterization of a novel ArsR-like regulatorencoded by Rv2034 in Mycobacterium tuberculosis. PLoS ONE 7, e36255 (2012).

29. Gonzalo-Asensio, J. et al. PhoP: a missing piece in the intricate puzzle ofMycobacterium tuberculosis virulence. PLoS ONE 3, e3496 (2008).

30. Gonzalo Asensio, J. et al. The virulence-associated two-component PhoP-PhoRsystem controls the biosynthesis of polyketide-derived lipids in Mycobacteriumtuberculosis. J. Biol. Chem. 281, 1313–1316 (2006).

31. Ryndak, M., Wang, S. & Smith, I. PhoP, a key player in Mycobacterium tuberculosisvirulence. Trends Microbiol. 16, 528–534 (2008).

32. Abramovitch, R. B., Rohde, K. H.,Hsu, F. F. & Russell, D. G. aprABC: a Mycobacteriumtuberculosis complex-specific locus that modulates pH-driven adaptation to themacrophage phagosome. Mol. Microbiol. 80, 678–694 (2011).

33. Singh, A. et al. Mycobacterium tuberculosis WhiB3 maintains redox homeostasis byregulating virulence lipid anabolism to modulate macrophage response. PLoSPathog. 5, e1000545 (2009).

34. Ernst, J., Vainas, O., Harbison, C. T., Simon, I. & Bar-Joseph, Z. Reconstructingdynamic regulatory maps. Mol. Syst. Biol. 3, 74 (2007).

35. Garton, N. J. et al. Cytological and transcript analyses reveal fat and lazy persister-like bacilli in tuberculous sputum. PLoS Med. 5, e75 (2008).

36. Park, H. D. et al. Rv3133c/dosR is a transcription factor that mediates the hypoxicresponse of Mycobacterium tuberculosis. Mol. Microbiol. 48, 833–843 (2003).

37. Baek, S. H., Li, A. H. & Sassetti, C. M. Metabolic regulation of mycobacterial growthand antibiotic sensitivity. PLoS Biol. 9, e1001065 (2011).

38. Cox, J. S., Chen, B., McNeil, M. & Jacobs, W. R. Jr. Complex lipid determines tissue-specific replicationofMycobacteriumtuberculosis inmice.Nature402,79–83(1999).

39. Camacho, L. R., Ensergueix, D., Perez, E., Gicquel, B. & Guilhot, C. Identification of avirulence gene cluster of Mycobacterium tuberculosis by signature-taggedtransposon mutagenesis. Mol. Microbiol. 34, 257–267 (1999).

40. Converse, S. E. et al. MmpL8 is required for sulfolipid-1 biosynthesis and Mycobac-terium tuberculosis virulence. Proc. Natl Acad. Sci. USA 100, 6121–6126 (2003).

41. Domenech, P. et al. The role of MmpL8 in sulfatide biogenesis and virulence ofMycobacterium tuberculosis. J. Biol. Chem. 279, 21257–21265 (2004).

42. Rousseau, C. et al. Production of phthiocerol dimycocerosates protectsMycobacterium tuberculosis from the cidal activity of reactive nitrogenintermediates produced by macrophages and modulates the early immuneresponse to infection. Cell. Microbiol. 6, 277–287 (2004).

43. Nazarova, E. V. et al. Role of lipid components in formation and reactivation ofMycobacterium smegmatis ‘‘nonculturable’’ cells. Biochemistry 76, 636–644(2011).

44. Ojha,A.K.etal.GrowthofMycobacteriumtuberculosisbiofilmscontaining freemycolicacids and harbouring drug-tolerant bacteria. Mol. Microbiol. 69, 164–174 (2008).

45. Ojha, A. K., Trivelli, X., Guerardel, Y., Kremer, L. & Hatfull, G. F. Enzymatic hydrolysisof trehalose dimycolate releases free mycolic acids during mycobacterial growthin biofilms. J. Biol. Chem. 285, 17380–17389 (2010).

46. Arnvig, K. & Young, D. Non-coding RNA and its potential role in Mycobacteriumtuberculosis pathogenesis. RNA Biol. 9, 427–436 (2012).

Supplementary Information is available in the online version of the paper.

Acknowledgements This project has been funded in whole or in part with Federalfunds from the National Institute of Allergy and Infectious Diseases National Institute ofHealth, Department of Health and Human Services, under contract no.HHSN272200800059C and U19 AI 076217, R01 AI 071155, the Paul G. Allen FamilyFoundation (to DRS), the National Science Foundation Pre-doctoral FellowshipProgram (to K.M.), and the Burroughs Wellcome Fund Award for TranslationalResearch. We acknowledge D. C. Young for lipidomics mass spectrometry services andadvice. We wouldalso like to thankL. Carvalho for his adviceon the statistical analysis ofthe gene expression modelling. We are grateful for the administrative assistance ofS. Shiviah and S. Tucker and for the support and advice of V. Di Francesco,K. Lacourciere, P. Dudley and M. Polanski.

Author Contributions J.E.G. led the project with G.K.S., oversaw ChIP-Seq, wrote thepaper and produced figures, discussed results and implications, oversaw dataintegration, and performed analyses. K.M. co-designed and performed ChIP andtranscriptomic experiments, discussed results and implications, and commented onthe manuscript. M.P. developed the analysis pipeline for ChIP-Seq data, performed allChIP-Seq data analysis, and contributed multiple figures and text. A.L. performed allanalysis of the integration of TF induction transcriptomics with ChIP-Seq data,contributed to analysis of ChIP-Seq binding data, and contributed multiple figures andtext. E.A. developed the predictive models of gene expression, and contributed allcorresponding figures and text. L.S. performed lipidomics experiments and dataanalysis, discussed the results and implications, and contributed figure and text to thepaper. A.G. developed the improved blind deconvolution algorithm for ChIP-Seq,contributed to analysis of all ChIP-Seq data, and contributed corresponding figures.T.R. designed and performed hypoxic time course and transcriptomic experiments,discussed results and implications and commented on the manuscript. G.D.performed all RT–PCR transcriptomics experiments and contributed analyses to thepaper. I.G. performed the DREM analysis and provided corresponding the figure. T.A.analysed ChIP-Seq data, developed the interfaces for data sharing and public release,and provided text. C.M. performed all library preparation and sequencing for ChIP-Seq.A.D.K. performed the metabolomics measurements, data analysis and theirinterpretation, discussed the results and implications and commented on themanuscript. R.A. was responsible for overview of bioinformatics and statistical dataanalysis. W.B. performed hypoxic time course, ChIP and transcriptomic experiments,and discussed results and implications. A.K. performed the experimental analysis ofKstR de-repression and provided the corresponding figure. S.J. performed theexperimental analysis of KstR de-repression, and provided the corresponding figure.M.J.H. produced individual MTB strains for ChIP-Seq experiments, and discussedresults and implications. J.Z. developed and curated the MTB metabolic model. C.G.contributed to analysis of profiling data. J.K.W. performed ChIP and transcriptomicexperiments, and discussed results and implications. Y.V.P. provided support andadvice. P.I. contributed to the analysis of KstR expression and the validation of KstRbinding sites. B.W. contributed to the ChIP-Seq analysis pipeline. P.S. and C.S.developed the interfaces for data sharing and public release. D.C. contributed to initialnetwork analysis. J.D. contributed to analysis of profiling data. Y.L. contributedexpression data for TB under different lipids. P.D. was responsible for experimentaldesign and mass spectrometry analysis. J.L. was responsible for coordinating sampleanalysis, data generation, annotation and results reporting Y.Z. was responsible forproteomics statistical data analysis. J.P. was responsible for analysis of LC-MS andLC-MS/MS data analysis, protein identification and maintenance of annotationdatabases. A.D. and H.-J.M. discussed the results and implications and commented onthe manuscript. B.H. and W.-H.Y. developed the ChIP protocol; S.T.P. developed theChIP protocol, performed the KstR RT–PCR experiments, and performed the MTB KstRnative promoter ChIP-Seq experiments. S.R. developed the ChIP protocol, oversawexperimental work on KstR and commented on the manuscript. S.H.E.K. discussed theresults and implications and commented on the manuscript. R.P.M. performed themetabolomics measurements, data analysis, and their interpretation; discussed theresults and implications and commented on the manuscript. D.C. was responsible foroverall scientific direction of the proteomic core. D.B.M. oversaw lipidomicsexperiments, contributed to integration of methods across mass spectral platforms,discussed the results and implications and commented on the manuscript. D.R.S.oversaw the hypoxic culture, ChIP and transcriptomic experiments, discussed resultsand implications, provided text and commented extensively on the manuscript. G.K.S.led the project with J.E.G., oversaw RT–PCR experiments, discussed results andimplications, provided text and commented extensively on the manuscript. G.K.S. andD.R.S. are co-last authors.

Author Information Expression data were deposited at GEO (accession numberGSE43466). The proteomics data have been deposited in the ProteomeXchange withthe identifier PXD000045. Reprints and permissions information is available atwww.nature.com/reprints. The authors declare no competing financial interests.Readers are welcome to comment on the online version of the paper. Correspondenceand requests for materials should be addressed to J.E.G. ([email protected]).

ARTICLE RESEARCH

1 1 J U L Y 2 0 1 3 | V O L 4 9 9 | N A T U R E | 1 8 3



http://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE43466

www.nature.com/reprints


mailto:[email protected]

1176 VOLUME 45 | NUMBER 10 | OCTOBER 2013 Nature GeNetics

A rt i c l e s

Tuberculosis killed one in five adults in Europe and North America between the seventeenth and nineteenth centuries1 and today remains a cause of high morbidity and mortality in much of the developing world2. Infectious diseases of humans can be divided into two broad categories3. Crowd diseases are generally highly virulent and depend on high host population densities to maximize pathogen transmission and reduce the risk of pathogen extinction through the exhaustion of susceptible hosts4. Many crowd diseases emerged during the Neolithic Demographic Transition (NDT) starting around 10,000 years ago, as the development of animal domestication increased the likelihood of zoonotic transfer of novel pathogens to humans and agricultural innovations supported increased population densities that helped sustain the infectious cycle3. In contrast, older human infections are often characterized by slow pro-gression to disease, sometimes involving reactivation after many years of latent or asymptomatic infection; these characteristics have been pro-posed to reflect adaptation to low host population densities by allowing repletion of the reservoir of susceptible individuals5. Tuberculosis is

reminiscent of a typical crowd disease in killing up to 50% of individuals when left untreated3,6 and having evolved a mode of aerosol transmis-sion that is promoted by high host densities. However, tuberculosis also displays a pattern of chronic progression, latency and reactivation that is characteristic of a pre-NDT disease7. Human tuberculosis was tradition-ally believed to have originated from animals4, but more recent phyloge-netic analyses of MTBC have suggested that the strains adapted to cause tuberculosis in animals diverged from the major human strains before NDT8–13. Moreover, human-associated MTBC is an obligate human pathogen with no known animal or environmental reservoir, suggesting that changes in human demography are likely to affect the evolution of MTBC. Here we used a population genomics approach to explore the evolutionary history of human MTBC, with a particular focus on the impact of changing host population sizes over time. Our results suggest a model that allows reconciliation of the apparent discrepancy between MTBC features characteristic of crowd diseases and those indicative of adaptation to low host densities.

Out-of-Africa migration and Neolithic coexpansion of Mycobacterium tuberculosis with modern humansIñaki Comas1,2, Mireia Coscolla3,4,23, Tao Luo5,23, Sonia Borrell3,4, Kathryn E Holt6,7, Midori Kato-Maeda8, Julian Parkhill9, Bijaya Malla3,4, Stefan Berg10, Guy Thwaites11,12, Dorothy Yeboah-Manu13, Graham Bothamley14, Jian Mei15, Lanhai Wei16, Stephen Bentley9, Simon R Harris9, Stefan Niemann17, Roland Diel18, Abraham Aseffa19, Qian Gao5, Douglas Young20–22,24 & Sebastien Gagneux3,4,24

Tuberculosiscaused20%ofallhumandeathsintheWesternworldbetweentheseventeenthandnineteenthcenturiesandremainsacauseofhighmortalityindevelopingcountries.Inanalogytoothercrowddiseases,theoriginofhumantuberculosishasbeenassociatedwiththeNeolithicDemographicTransition,butrecentstudiespointtoamuchearlierorigin.Weanalyzedthewholegenomesof259M. tuberculosiscomplex(MTBC)strainsandusedthisdatasettocharacterizeglobaldiversityandtoreconstructtheevolutionaryhistoryofthispathogen.CoalescentanalysesindicatethatMTBCemergedabout70,000yearsago,accompaniedmigrationsofanatomicallymodernhumansoutofAfricaandexpandedasaconsequenceofincreasesinhumanpopulationdensityduringtheNeolithicperiod.ThislongcoevolutionaryhistoryisconsistentwithMTBCdisplayingcharacteristicsindicativeofadaptationtobothlowandhighhostdensities.

1Genomics and Health Unit, Centre for Public Health Research (CSISP-FISABIO), Valencia, Spain. 2CIBER (Centros de Investigación Biomédica en Red) in Epidemiology and Public Health, Barcelona, Spain. 3Department of Medical Parasitology and Infection Biology, Swiss Tropical and Public Health Institute, Basel, Switzerland. 4University of Basel, Basel, Switzerland. 5Key Laboratory of Medical Molecular Virology, Institutes of Biomedical Sciences and Institute of Medical Microbiology, Shanghai Medical College, Fudan University, Shanghai, China. 6Department of Biochemistry and Molecular Biology, The University of Melbourne, Melbourne, Victoria, Australia. 7Bio21 Molecular Science and Biotechnology Institute, The University of Melbourne, Melbourne, Victoria, Australia. 8Division of Pulmonary and Critical Care Medicine, University of California, San Francisco, San Francisco, California, USA. 9Pathogen Genomics, The Wellcome Trust Sanger Institute, Wellcome Trust Genome Campus, Hinxton, Cambridge, UK. 10TB Research Group, Veterinary Laboratories Agency, Weybridge, New Haw and Addlestone, UK. 11Department of Infectious Disease, King’s College London, London, UK. 12Centre for Clinical Infection and Diagnostics Research, King’s College London, London, UK. 13Noguchi Memorial Institute for Medical Research, University of Ghana, Legon, Ghana. 14Department of Respiratory Medicine, Homerton University Hospital, London, UK. 15Department of Tuberculosis Control, Shanghai Municipal Center for Disease Control and Prevention, Shanghai, China. 16Ministry of Education Key Laboratory of Contemporary Anthropology, School of Life Sciences and Institutes of Biomedical Sciences, Fudan University, Shanghai, China. 17Molecular Mycobacteriology, Research Center Borstel, Borstel, Germany. 18Institute for Epidemiology, Schleswig-Holstein University Hospital, Kiel, Germany. 19Armauer Hansen Research Institute, Addis Ababa, Ethiopia. 20Medical Research Council (MRC) National Institute for Medical Research, Mill Hill, London, UK. 21Department of Medicine, Imperial College London, London, UK. 22Centre for Molecular Bacteriology and Infection, Imperial College London, London, UK. 23These authors contributed equally to this work. 24These authors jointly directed this work. Correspondence should be addressed to I.C. ([email protected]), Q.G. ([email protected]) or S.G. ([email protected]).

Received 17 December 2012; accepted 1 August 2013; published online 1 September 2013; doi:10.1038/ng.2744

npg

© 2

013

Nat

ure

Am

eric

a, In

c. A

ll rig

hts

rese

rved

.

http://www.nature.com/doifinder/10.1038/ng.2744

http://www.nature.com/naturegenetics/

http://www.nature.com/naturegenetics/

Nature GeNetics VOLUME 45 | NUMBER 10 | OCTOBER 2013 1177

A rt i c l e s

RESULTSTheglobaldiversityofhuman-adaptedMTBCWe sequenced the whole genomes of 186 strains representative of the global diversity of MTBC, combining these sequences with data from 34 already published strains and 39 additional newly sequenced strains corresponding to the lineage 2 ‘Beijing’ family (Supplementary Table 1). In the global data set, after excluding repetitive and mobile elements, we identified 34,167 polymorphic sites (SNPs) (Supplementary Table 2), which we used to reconstruct phylogenetic relationships between these strains (Fig. 1a). This genome-based phylogeny was congruent with previous phylogenies based on other markers and resolved seven major lineages, with animal-adapted strains clustering together with the strains from line-age 6 (ref. 8). The phylogeny included the recently described lineage 7, which so far has only been observed in Ethiopia or in recent Ethiopian

emigrants14. Principal-component analysis confirmed all main MTBC lineages and highlighted the close phylogenetic relationship between Eurasian lineages 2–4. These three lineages have collectively in the past been referred to as evolutionarily ‘modern’ (Fig. 1b) because of their comparably more derived position on the MTBC phylogeny and because they are thought to have spread more recently8,11. The maximum genetic distance between any 2 strains was 2,188 SNPs and involved a human and an animal strain and was 1,856 SNPs when only human clinical isolates were considered. Only 387 of the SNPs (1.1%) were homoplastic. Homoplasy can arise as a consequence of false- positive SNP calls because of positive selection or recurrent muta-tions, or because of recombination, as recently suggested15. However, the fact that only 1.1% of the sites were homoplastic supports the view that the population structure of MTBC is largely clonal, with little ongoing recombination occurring between strains16,17.

a

Southeast Asia, Oceania:Macrohaplogroup M

Eurasia:Macrohaplogroup N

Africa:Macro

haplogroup L

HumanmtDNA

0.0020

dSoutheast Asia, Oceania:Lineage 1

Eurasia:Lineages 2–4

Africa:Lineages 5 and 6

MTBC

0.0040

c

Lineage 4Europe, America,

Africa

Lineage 1East Africa, the

Philippines,Indian Ocean rim

Lineage 7Ethiopia

Lineage 5West African 1

Lineage 6West African 2

Lineage 2East Asia

Lineage 3East Africa,Central Asia

ModernMTBC

Animalstrains

100/100

100/100100/100

100/100

100/100

0.0050

100/100100/100

100/100100/100

100/10096/95

M. canettii

99/100ModernMTBC

0.20

0.15

0.10

0.05

PC

3

PC 2PC 1

0

–0.05

–0.10

0.100.05

–0.05–0.10 0.064

0.0650.066

0.0670.068

0.069

0

b

Figure 1 The genome-based phylogeny of MTBC mirrors that of human mitochondrial genomes. (a) Whole-genome phylogeny of 220 strains of MTBC. Support values for the main branches after inference with neighbor-joining (left) and maximum-likelihood (right) analyses are shown. (b) Principal-component analysis of the 34,167 SNPs. The first three principal-component axes (PC 1–PC 3) are shown; these discriminate between evolutionarily modern (gray circle) and ancient (all other) strains. Individual lineages are shown with the same colors as in a. (c,d) Comparison of the MTBC phylogeny (c) and a phylogeny derived from 4,955 mitochondrial genomes (mtDNA) representative of the main human haplogroups (d). Color coding highlights the similarities in tree topology and geographic distribution between MTBC strains and the main human mitochondrial macrohaplogroups (black, African clades: MTBC lineages 5 and 6, human mitochondrial macrohaplogroups L0–L3; pink, Southeast Asian and Oceanian clades: MTBC lineage 1, human mitochondrial macrohaplogroup M; blue, Eurasian clades: MTBC lineage 2–4, human mitochondrial macrohaplogroup N). MTBC lineage 7 has only been found in Ethiopia, and its correlation with any of the three main human haplogroups remains unclear. Scale bars indicate substitutions per site.

npg

© 2

013

Nat

ure

Am

eric

a, In

c. A

ll rig

hts

rese

rved

.


A rt i c l e s

AfricanoriginandcodivergenceofMTBCwithmodernhumansSeveral studies have proposed an African origin for MTBC8,10,12. We decided to formally test this hypothesis using our new whole-genome data. We used three independent phylogeographic analyses to determine the likely geographic origin of the most recent com-mon ancestor (MRCA) of MTBC. Two different Bayesian analyses identified Africa as the most likely origin of MTBC, with East and West Africa showing combined posterior probabilities of 90% and 67%, respectively (Supplementary Figs. 1–3). Similarly, a maximum parsimony approach predicted 100% probability of an African origin. Taken together, these data support the hypothesis that MTBC originated in Africa.

Next, we sought to determine the putative age of the association between MTBC and its human host. Given that human-adapted MTBC is limited to humans and that both anatomically modern humans and MTBC originated in Africa, we tested whether MTBC and humans might have diverged in parallel; this would be partic-ularly likely if the association between the two predates the NDT, as previously postulated8,10,12. To explore this possibility, we first compared our new MTBC phylogeny to a corresponding tree con-structed from 4,955 mitochondrial genomes representative of the main human haplogroups (Supplementary Table 3)18. We observed striking similarities (Fig. 1c,d). In both cases, the early branching clades were found exclusively in Africa. Moreover, the trichotomy formed by the branching of the Out-of-Africa M and N mitochon-drial macrohaplogroups from the L3 African source population was mirrored in the MTBC phylogeny by a similar relationship between lineage 1, Eurasian lineages 2–4 and African lineages 5 and 6. In addi-tion to this qualitative similarity, comparison of the most common mitochondrial haplogroups with the most frequent MTBC lineages in the same country identified a strong quantitative association (by parsimony score and association index tests; P < 0.01 in all cases) (Supplementary Fig. 4, Supplementary Table 4 and Supplementary Note). Taken together, these data are consistent with MTBC evolving in parallel with its human host.

AgeoftheassociationofMTBCandhumansSimilarities in tree topology and phylogeographic distribution suggest that MTBC infected the early human populations of Africa. To further explore the association between MTBC and its human host, we tested for possible imprints of ancient human divergence times on the main phylogenetic lineages of MTBC using a Bayesian approach19. Several approaches have been used to date bacterial phylogenies (see refs. 20–22 for some examples). Unfortunately, none of these were applicable here

because of the following reasons. First, although ancient DNA has been used to study the evolutionary history of other bacteria20 and similar studies have been performed in tuberculosis in the past23, no relevant whole-genome data are currently available for ancient DNA from MTBC strains. Second, although a mutation rate for MTBC has recently been estimated on the basis of a macaque infection model and molecular epidemiological data24,25, it is well known that such short-term mutation rates cannot easily be extrapolated to the long-term substitution rates relevant for the time scale discussed here26,27. Third, and related to the previous point, although the isolation dates of some of the strains included in our analysis are known, at best they would allow the calculation only of a short-term mutation rate. Moreover, when performing a tip-to-date analysis of those strains (N = 49), we found that, in contrast to several other bacterial species21,28–30, MTBC had no significant correlation between isolation time and phyloge-netic divergence (correlation coefficient = 0.047).

Because of these limitations, we used an alternative approach to date our MTBC phylogeny. Specifically, we used as initial calibration points several key dates in human evolution. We tested three alterna-tive models in which the coalescent time for the most basal MTBC lineages 5 and 6 was calibrated against (i) the emergence of anatomi-cally modern humans 185,000 ± 20,000 years ago (MTBC-185)31, (ii) the coalescent time of the L3 mitochondrial haplogroup 70,000 ± 10,000 years ago (MTBC-70)32 and (iii) the beginning of the NDT 10,000 ± 2,000 years ago (MTBC-10)3 (Table 1). We compared the timing of the branching points predicted by each of the models with estimated dates of known events in human history. A recent model based on the analysis of human whole-genome variation data sets sug-gests that the global dispersal of modern humans occurred through two major waves: an initial eastern dispersal around the Indian Ocean starting 62,000–75,000 years ago and a later dispersal into Eurasia 25,000–38,000 years ago33. Our MTBC-70 model showed a striking correlation with these human migration events by dating a first split of lineage 1 at 67,000 years ago (95% highest probability density (HPD) = 48,000–88,000 years ago), coinciding with the first wave of human migration31, and a second split at 46,000 years ago (95% HPD = 31,000–61,000 years ago), matching the later dispersal throughout Eurasia (Fig. 2a and Supplementary Fig. 5)34,35. Coalescent dates for the branch leading to lineages 2 and 4 in the MTBC-70 model (30,000–46,000 years ago and 32,000–42,000 years ago, respectively) showed a good correlation with archaeological evidence of the pres-ence of modern humans in Europe35 and East Asia36. In contrast, our alternate model MTBC-185 postulated initial branching of Out-of-Africa lineages as early as 126,000–174,000 years ago when focusing

table 1 comparison of different dating scenarios for MtBc evolutionDating scenario MTBC-70 MTBC-185 MTBC-10 MTBC-65

Rationale Emergence of MTBC with human mitochondrial DNA

haplogroup L3

Emergence of MTBC with anatomically modern humans

Emergence of MTBC during NDT

Emergence of Out-of-Africa MTBC with human mitochon-

drial DNA haplogroup M

Dates inferred from models (in thousands of years ago)a

MRCA of MTBC 73 (50–96) 198 (170–229) 11 (9–14) 67 (44–91)

Coalescent time for lineages 5 + 6 70 (48–88)b 184 (164–203)b 10 (8–12)b 61 (40–81)

Coalescent time for lineage 1 67 (46–88) 183 (160–207) 10 (8–12) 62 (42–82)b

Coalescent time for lineages 2–4 46 (31–61) 126 (104–148) 7 (6–10) 41 (26–55)

Period of maximum logistic growth 4–7 31–34 1 4–7

Substitution rate (SNPs per polymorphic site per thousand years)c

3.37 × 10−4 (2.38 × 10−4 to 4.65 × 10−4)

1.23 × 10−4 (1.04 × 10−4 to 1.46 × 10−4)

2.17 × 10−3 (1.71 × 10−3 to 2.68 × 10−3)

3.78 × 10−4 (2.62 × 10−4 to 5.36 × 10−4)

aDates are shown as the median value and 95% HPD interval predicted in the corresponding Bayesian analysis. bValue provided as prior input in Bayesian analysis. cBEAST-predicted rate of SNP accumulation (per polymorphic position and thousand years). In the main text we use the estimated genomic substitution rate (per position per year) for comparative purposes with published estimations from other bacterial species.

npg

© 2

013

Nat

ure

Am

eric

a, In

c. A

ll rig

hts

rese

rved

.


A rt i c l e s

on the branch leading to modern strains (Supplementary Fig. 6), which would suggest that the global dispersal of MTBC preceded that of anatomically modern humans. The MTBC-10 model, by definition, implies global dispersal within the last 10,000 years (Supplementary Fig. 7). Although MTBC has been spread by trade and conquest in recent centuries8, the pattern of this dispersal does not match the phy-logeographic distribution discussed above. Finally, a fourth model, MTBC-65, using the coalescent time of mitochondrial haplogroup M as a calibration time point for MTBC lineage 1, generated very similar results to the MTBC-70 model (Table 1). In summary, our phyloge-netic analysis based on a 70,000-year time frame shows that MTBC has been infecting humans for at least the last 70,000 years.

NeolithiccoexpansionofMTBCandhumansAll the data presented so far strongly support the notion that human tuberculosis indeed predated NDT. How then could the features of tuberculosis typical of crowd diseases have arisen? To address this question, we used Bayesian skyline plots to estimate the changes in effective population size over time in the pathogen and human popu-lations19. Analysis of our full MTBC sequence data set identified a main signal of population size increase starting about 10,000 years ago (Fig. 2b), suggesting that the expansion of MTBC occurred as a consequence of the increase in population densities that followed the establishment of the first human settlements during NDT37 and not only because of a general increase in the total number of humans peopling the planet at the time. To test whether human population dynamics around that period coincided with those for MTBC, we used a data set previously described to maximize the information on human demographics during the Neolithic (Supplementary Table 5)38. The resulting skyline plot showed a Neolithic expansion

of humans around 4,000–8,000 years ago (Supplementary Fig. 8), coinciding with the expansion of MTBC (Spearman’s R = 0.99; P < 0.00001; Fig. 2b and Supplementary Fig. 8). Taken together, these findings indicate that the Neolithic period contributed to the success of MTBC, not by enhancing the likelihood of zoonotic transfer to humans as previously proposed, but because of combined increases in host population size and density.

TheevolutionaryhistoryofMTBConaregionalscaleTo analyze MTBC evolution at a regional level, we focused on lin-eage 2, which includes the Beijing family of strains. These strains have received particular attention because of their hypervirulence in laboratory models, their recent dissemination in human populations and their association with drug resistance39. Supplementing our glo-bal diversity set by sequencing the whole genomes of an additional 39 lineage 2 strains from China, we observed a strong correlation between skyline plots derived from lineage 2 genomes and a set of human mitochondrial genomes enriched for haplogroups from East Asia that likely originated just before, during or after the Neolithic

Lineage 4Europe

Lineages 5and 6

West Africa

46

64

67

73

42

Lineage 7Ethiopia

Lineage 3Central Asia

Lineage 1Indian Ocean

Lineage 2East Asia

a

100,000

10,000

b

1,000

100

Ne per generation

1060 50 40 30

Time before present (thousand years)20 10 0

Figure 2 Out-of-Africa and Neolithic expansion of MTBC. (a) Map summarizing the results of the phylogeographic and dating analyses for MTBC. Color coding of lineages is the same as in Figure 1a. Major splits are annotated with the median value (in thousands of years) of the dating of the relevant node. Lineage 7 has so far been isolated exclusively in individuals with known country of origin in the Horn of Africa14. Lineage 7 diverged subsequent to the proposed Out-of-Africa migration of MTBC; it may have arisen among a human population that remained in Africa or a population that returned to Africa. (b) Bayesian skyline plots showing changes in population diversity of MTBC (red) and humans based on mitochondrial DNA (blue) over the last 60,000 years. Dashed lines represent the 95% HPD intervals for the estimated population sizes. Ne, effective population size.

32,000–42,000 years agoEmergence of lineage 2

linked to arrival of modernhumans in East Asia

6,000–11,000 years agoExpansion of lineage 2linked to humans NDT

3,000–5,000 years agoExpansion of Beijing family

linked to spread of agriculture

a

Ne per generation

c

100,000

10,000

1,000

100

10

125 20 15

Time before present(thousand years)

10 5 0

b

MRCA of Beijing family8,000 years ago

(95% HPD =6,000–11,000 years ago)

30 25 20 15Time before present

(thousand years)

10 5 0

Figure 3 Neolithic expansion and spread of MTBC lineage 2 Beijing strains in East Asia. (a) Bayesian skyline plots indicating changes in lineage 2 diversity over time (red) compared with human mitochondrial DNA haplogroups from East Asia (blue). Dashed lines show 95% HPD intervals for the population size estimations. (b) Dated Bayesian phylogeny of MTBC lineage 2 based on coalescent analysis. (c) Map of the parallel origin and migration of MTBC and humans in East Asia, indicating the first archaeological evidence of modern humans in the region 32,000–42,000 years ago, coinciding with the migration of MTBC from central to East Asia, the start of the Neolithic period in the region indicated by the first evidence of domesticated crops in China coinciding with the origin of the MTBC Beijing family 8,000 years ago (HPD = 6,000–11,000 years ago) and the coexpansion of agriculture and the MTBC Beijing family into neighboring countries 3,000–5,000 years ago.

npg

© 2

013

Nat

ure

Am

eric

a, In

c. A

ll rig

hts

rese

rved

.


A rt i c l e s

period (Spearman’s R = 0.97; P < 0.001; Fig. 3a and Supplementary Fig. 9). MTBC-70 dating for lineage 2 is consistent with an initial arrival coincident with archaeological evidence of anatomically mod-ern humans in East Asia36 (32,000–42,000 years ago; Supplementary Fig. 5), a first expansion (6,000–11,000 years ago; Fig. 3b,c) alongside the emergence of agriculture in China 8,000 years ago40 and a subse-quent main expansion of the Beijing strains (3,000–5,000 years ago; Supplementary Fig. 9) coinciding with the spread of agriculture to neighboring regions (Fig. 3b,c)37.

In summary, our data on the global and regional expansion of MTBC during NDT support the view that, although NDT was not the only period leading to large increases in human population sizes, it was the period where, in addition to human population growth, the densities of human populations increased following the first establish-ment of permanent human settlements. Hence, in addition to provid-ing a springboard for global domination by modern humans, NDT was also central to the success of MTBC by generating growing numbers of susceptible hosts living under increasingly crowded conditions.

DISCUSSIONThe common origin in Africa of MTBC and humans, the congruence in their phylogeographies and the dating of major branching events lead us to conclude that MTBC has been coevolving with anatomically modern humans for tens of thousands of years. The marked expan-sion of MTBC during NDT but not during earlier human expansion events41,42 suggests that the success of this pathogen was primarily driven by increases in human host population density, which is typi-cal of crowd diseases. However, the striking match between MTBC and human mitochondrial phylogenies supports a much older asso-ciation between MTBC and its host and suggests that carriage of MTBC was ubiquitous in hunter-gatherer populations migrating out of Africa well before NDT. The fidelity of this match is unexpected. Considering the vulnerability and small numbers of human groups (some of today’s hunter-gatherers live in groups of 20 or less43), it might have been anticipated that tuberculosis would have substan-tial detrimental impact on these groups and might therefore have precipitated its own extinction. In fact, the correspondence of the MTBC phylogeny with early human migration is strikingly similar to that observed with low-virulence Helicobacter pylori44. Perhaps latent infection with MTBC imparted some degree of immunity against more lethal pathogens encountered in the new environment or in contact with archaic human populations. Ongoing analyses of human microbiota highlight the fuzzy boundaries between com-mensalism and pathogenicity in health and disease45. A recent study has suggested that coinfection with H. pylori might protect against active tuberculosis disease46. Conversely, whether latent tuberculosis infection protects against gastric ulcers or stomach cancer caused by H. pylori in individuals infected with both bacteria is unknown but represents an intriguing possibility. In such a case, positive feedback between both infections would result in an asymptomatic individual benefiting from being infected by both bacterial species.

Alternatively, one could think of a model in which early popula-tions carried the infection in a less virulent form, with transmission

sustained by reactivation of disease in elderly individuals beyond the reproductive age. The possibility that disease characteristics might have changed over time as different MTBC populations were selected in different human societies may help to explain current epidemio-logical trends associated with increased dissemination of the Beijing family of MTBC39 and decreased rates of disease caused by evolution-arily ‘ancient’ lineages of MTBC47. In addition to changes in popula-tion density, it can be anticipated that the pathology of tuberculosis during NDT would have been influenced by coinfections with novel crowd diseases and by variations in key nutrients such as vitamin D48. Similarly, it is important to consider the possibility of reciprocal adaptive changes to the human genome as a result of prolonged coevolution with MTBC49.

In this study, we have compared MTBC phylogenic diversity to human diversity inferred from mitochondrial genome data. One advantage of using mitochondrial data is that these data have been used extensively to study recent human evolution. Furthermore, such data are available from almost any region of the world, and there is a large body of work studying human migrations that is based on the distribution of mitochondrial haplogroups. However, mitochondrial DNA is also limited in that it contains little phylogenetic informa-tion, and the existing data sets suffer from potential sampling bias. Increasingly, new DNA sequencing technologies are paving the way for studies of human diversity based on whole genomes33. Hence, in the context of a pathogen such as MTBC, future studies should be based on paired human and bacterial whole-genome informa-tion that is collected prospectively. Such an integrated approach will allow investigation of the molecular determinants of host-pathogen coevolution in human tuberculosis and other diseases.

The accumulation of more than 30,000 SNPs by human MTBC strains over the proposed time frame of 70,000 years corresponds to a long-term genome-wide substitution rate of 2.58 × 10−9 substitutions per site per year (95% HPD = 1.66 × 10−9 to 2.89 × 10−9; Table 1). This rate is much lower than recent estimates of short-term sub-stitution rates for experimental models and human outbreaks24,25. A decrease in substitution rates measured over increasing time inter-vals is a common feature of phylogenetic analyses27, and an exponen-tial decrease is observed in the substitution rate with time when we pool our data with those from other similar genome-based studies published recently (correlation coefficient = −0.9614; P < 0.0001; Fig. 4). Fixation or removal of single-nucleotide changes by natu-ral selection can contribute to this phenomenon, although retention of a high proportion of nonsynonymous mutations suggests that

1.00 × 10–4

1.00 × 10–5

H. pylori

S. aureus ST239

S. sonnei

C. difficile

Y. pestis

MTBC

V. cholerae

S. pneumoniae1.00 × 10–6

1.00 × 10–7

log 10

(su

bstit

utio

ns p

er b

ase

pair

per

year

)

1.00 × 10–8

1.00 × 10–9

1 10log10 (time since MRCA)

100 1,000 10,000 100,000

Figure 4 Time-dependent decay of substitution rates in bacteria based on whole-genome data sets. The scatter plot graph shows the relationship between substitution rate and time span between the MRCA and the last sampling date for each studied pathogen. Values were extracted from relevant publications that used whole-genome representative data sets and coalescent analysis of substitution rates (for a complete list of references, see the supplementary Note).

npg

© 2

013

Nat

ure

Am

eric

a, In

c. A

ll rig

hts

rese

rved

.


A rt i c l e s

natural selection has had a low impact on MTBC8. Alternative mecha-nisms to account for the reduction in genetic diversity over long time scales include serial founder effects linked to sequential expansions of human subpopulations and their associated pathogenic and com-mensal microbial flora50.

In conclusion, we propose that MTBC has been a constant com-panion of anatomically modern humans during our evolution and global dissemination over the last 70,000 years. Furthermore, MTBC has been able to adapt to changing human populations. Exploration of changes that have occurred in this interaction over time may help predict future patterns of disease and to design rational strategies to bring an end to this historic partnership.

METhODSMethods and any associated references are available in the online version of the paper.

Accession codes. Sequencing reads for the previously unpublished genomes of strains that were used in this study (225 strains) have been deposited in the European Nucleotide Archive (ENA) under study number ERP001731. Additionally, we have analyzed the genomes from 34 previously published strains that are available at the Sequence Read Archive (SRA) under accessions SRS002426, SRS003212, SRS003328, SRS004666, SRS004753, SRS004754, SRS004756, SRS004757, SRS004758, SRS004759, SRS004760, SRS004761, SRS004762, SRS004763, SRS004764, SRS004831, SRS004841, SRS005175, SRS005448, SRS005450, SRS006765, SRS074557, SRS084142, SRX002001, SRX002002, SRX002003, SRX002004, SRX002005, SRX002429, ERS001592, ERS003236, ERS003237 and ERS003250). A complete list of the strains analyzed in this study together with sequenc-ing and origin information is given in Supplementary Table 1.

Note: Any Supplementary Information and Source Data files are available in the online version of the paper.

ACKNoWLEDGMENTSWe thank D. Behar and S. Rosset for providing the mitochondrial genome sequences and C. Gignoux for advice on the mitochondrial Neolithic data set, N. Mistry (The Foundation for Medical Research) for providing bacterial strains and C. Dye, F. Balloux and L. Weinert for comments on the manuscript. This work was supported by the MRC UK (grants U.1175.02.002.00015.01 to S.G. and U117581288 to D.Y.), the Swiss National Science Foundation (PP0033-119205 to S.G.), the US National Institutes of Health (AI090928 and HHSN266200700022C to S.G.), the Leverhulme-Royal Society Africa Award (AA080019 to S.G.) and the Natural Science Foundation of China (grant 91231115 to Q.G.). DNA sequencing was partially supported by core funding of the Wellcome Trust (grant 098051) and by a Framework Programme 7 project of the European Community (SysteMTb HEALTH-F4-2010-241587 to D.Y.). I.C. is supported by European Union funding from the Marie Curie Framework Programme 7 actions (project 272086) and project BFU2011-24112 from the Ministerio de Economía y Competitividad (Spain).

AUTHoR CoNTRIBUTIoNSI.C., Q.G., D.Y. and S.G. designed and supervised the study. M.C., S. Borrell, K.E.H., M.K.-M., J.P., B.M., S. Berg, G.T., D.Y.-M., G.B., J.M., L.W., S.R.H., S.N., R.D., A.A., Q.G. and S.G. provided MTBC strains and/or reagents. J.P., S. Bentley and S.R.H. contributed to the genome sequencing. I.C., M.C. and T.L. analyzed the data. I.C., M.C., T.L., S. Borrell, K.E.H., J.P., S. Berg, G.T., D.Y.-M., S. Bentley, S.R.H., S.N., A.A., Q.G., D.Y. and S.G. contributed to the manuscript writing. All authors read and approved the manuscript.

CoMPETING FINANCIAL INTERESTSThe authors declare no competing financial interests.

Reprints and permissions information is available online at http://www.nature.com/reprints/index.html.

1. Wilson, L.G. Commentary: medicine, population, and tuberculosis. Int. J. Epidemiol. 34, 521–524 (2005).

2. World Health Organization. The Global Plan to STOP TB 2011–2015 (World Health Organization, Geneva, 2011).

3. Wolfe, N.D., Dunavan, C.P. & Diamond, J. Origins of major human infectious diseases. Nature 447, 279–283 (2007).

4. Diamond, J. Guns, Germs, and Steel: The Fates of Human Societies 496 (W.W. Norton & Company, New York, 1999).

5. Blaser, M.J. & Kirschner, D. The equilibria that allow bacterial persistence in human hosts. Nature 449, 843–849 (2007).

6. Berg, G. The prognosis of open pulmonary tuberculosis: a clinical-statistical analysis. J. Am. Med. Assoc. 114, 1954–1955 (1940).

7. Barry, C.E. III et al. The spectrum of latent tuberculosis: rethinking the biology and intervention strategies. Nat. Rev. Microbiol. 7, 845–855 (2009).

8. Hershberg, R. et al. High functional diversity in Mycobacterium tuberculosis driven by genetic drift and human demography. PLoS Biol. 6, e311 (2008).

9. Mostowy, S., Cousins, D., Brinkman, J., Aranaz, A. & Behr, M.A. Genomic deletions suggest a phylogeny for the Mycobacterium tuberculosis complex. J. Infect. Dis. 186, 74–80 (2002).

10. Wirth, T. et al. Origin, spread and demography of the Mycobacterium tuberculosis complex. PLoS Pathog. 4, e1000160 (2008).

11. Brosch, R. et al. A new evolutionary scenario for the Mycobacterium tuberculosis complex. Proc. Natl. Acad. Sci. USA 99, 3684–3689 (2002).

12. Gutierrez, M.C. et al. Ancient origin and gene mosaicism of the progenitor of Mycobacterium tuberculosis. PLoS Pathog. 1, e5 (2005).

13. Gagneux, S. et al. Variable host-pathogen compatibility in Mycobacterium tuberculosis. Proc. Natl. Acad. Sci. USA 103, 2869–2873 (2006).

14. Firdessa, R. et al. Mycobacterial lineages causing pulmonary and extrapulmonary tuberculosis in Ethiopia. Emerg. Infect. Dis. 19, 460–463 (2013).

15. Namouchi, A., Didelot, X., Schöck, U., Gicquel, B. & Rocha, E.P.C. After the bottleneck: genome-wide diversification of the Mycobacterium tuberculosis complex by mutation, recombination, and natural selection. Genome Res. 22, 721–734 (2012).

16. Hirsh, A.E., Tsolaki, A.G., DeRiemer, K., Feldman, M.W. & Small, P.M. Stable association between strains of Mycobacterium tuberculosis and their human host populations. Proc. Natl. Acad. Sci. USA 101, 4871–4876 (2004).

17. Comas, I. & Gagneux, S. The past and future of tuberculosis research. PLoS Pathog. 5, e1000600 (2009).

18. Behar, D.M. et al. A “Copernican” reassessment of the human mitochondrial DNA tree from its root. Am. J. Hum. Genet. 90, 675–684 (2012).

19. Drummond, A.J., Ho, S.Y.W., Phillips, M.J. & Rambaut, A. Relaxed phylogenetics and dating with confidence. PLoS Biol. 4, e88 (2006).

20. Bos, K.I. et al. A draft genome of Yersinia pestis from victims of the Black Death. Nature 478, 506–510 (2011).

21. Mutreja, A. et al. Evidence for several waves of global transmission in the seventh cholera pandemic. Nature 477, 462–465 (2011).

22. Morelli, G. et al. Yersinia pestis genome sequencing identifies patterns of global phylogenetic diversity. Nat. Genet. 42, 1140–1143 (2010).

23. Djelouadji, Z., Raoult, D. & Drancourt, M. Palaeogenomics of Mycobacterium tuberculosis: epidemic bursts with a degrading genome. Lancet Infect. Dis. 11, 641–650 (2011).

24. Ford, C.B. et al. Use of whole genome sequencing to estimate the mutation rate of Mycobacterium tuberculosis during latent infection. Nat. Genet. 43, 482–486 (2011).

25. Walker, T.M. et al. Whole-genome sequencing to delineate Mycobacterium tuberculosis outbreaks: a retrospective observational study. Lancet Infect. Dis. 13, 137–146 (2013).

26. Morelli, G. et al. Microevolution of Helicobacter pylori during prolonged infection of single hosts and within families. PLoS Genet. 6, e1001036 (2010).

27. Ho, S.Y.W. et al. Time-dependent rates of molecular evolution. Mol. Ecol. 20, 3087–3101 (2011).

28. Holt, K.E. et al. Shigella sonnei genome sequencing and phylogenetic analysis indicate recent global dissemination from Europe. Nat. Genet. 44, 1056–1059 (2012).

29. Croucher, N.J. et al. Rapid pneumococcal evolution in response to clinical interventions. Science 331, 430–434 (2011).

30. Harris, S.R. et al. Evolution of MRSA during hospital transmission and intercontinental spread. Science 327, 469–474 (2010).

31. Soares, P. et al. Correcting for purifying selection: an improved human mitochondrial molecular clock. Am. J. Hum. Genet. 84, 740–759 (2009).

32. Soares, P. et al. The expansion of mtDNA haplogroup L3 within and out of Africa. Mol. Biol. Evol. 29, 915–927 (2012).

33. Rasmussen, M. et al. An Aboriginal Australian genome reveals separate human dispersals into Asia. Science 334, 94–98 (2011).

34. Henn, B.M., Cavalli-Sforza, L.L. & Feldman, M.W. The great human expansion. Proc. Natl. Acad. Sci. USA 109, 17758–17764 (2012).

35. Stewart, J.R. & Stringer, C.B. Human evolution out of Africa: the role of refugia and climate change. Science 335, 1317–1321 (2012).

36. Jin, L. & Su, B. Natives or immigrant: modern human origins in East Asia. Nat. Rev. Genet. 1, 126–133 (2000).

37. Bellwood, P. & Oxenham, M. The Neolithic Demographic Transition and its Consequences 13–34 (Springer, New York, 2008).

npg

© 2

013

Nat

ure

Am

eric

a, In

c. A

ll rig

hts

rese

rved

.



http://www.ebi.ac.uk/ena/data/view/ERP001731

http://trace.ncbi.nlm.nih.gov/Traces/sra/sra.cgi?sample=SRS002426























http://www.ncbi.nlm.nih.gov/sra/SRX002001






http://trace.ncbi.nlm.nih.gov/Traces/sra/sra.cgi?sample=ERS001592





http://www.nature.com/reprints/index.html

http://www.nature.com/reprints/index.html


38. Gignoux, C.R., Henn, B.M. & Mountain, J.L. Rapid, global demographic expansions after the origins of agriculture. Proc. Natl. Acad. Sci. USA 108, 6044–6049 (2011).

39. Parwati, I., van Crevel, R. & van Soolingen, D. Possible underlying mechanisms for successful emergence of the Mycobacterium tuberculosis Beijing genotype strains. Lancet Infect. Dis. 10, 103–111 (2010).

40. Barton, L. et al. Agricultural origins and the isotopic identity of domestication in northern China. Proc. Natl. Acad. Sci. USA 106, 5523–5528 (2009).

41. Atkinson, Q.D., Gray, R.D. & Drummond, A.J. mtDNA variation predicts population size in humans and reveals a major Southern Asian chapter in human prehistory. Mol. Biol. Evol. 25, 468–474 (2008).

42. Wei, W. et al. A calibrated human Y-chromosomal phylogeny based on resequencing. Genome Res. 23, 388–395 (2013).

43. Hamilton, M.J., Milne, B.T., Walker, R.S., Burger, O. & Brown, J.H. The complex structure of hunter-gatherer social networks. Proc. Biol. Sci. 274, 2195–2202 (2007).

44. Linz, B. et al. An African origin for the intimate association between humans and Helicobacter pylori. Nature 445, 915–918 (2007).

45. Littman, D.R. & Pamer, E.G. Role of the commensal microbiota in normal and pathogenic host immune responses. Cell Host Microbe 10, 311–323 (2011).

46. Perry, S. et al. Infection with Helicobacter pylori is associated with protection against tuberculosis. PLoS ONE 5, e8804 (2010).

47. de Jong, B.C. et al. Progression to active tuberculosis, but not transmission, varies by Mycobacterium tuberculosis lineage in The Gambia. J. Infect. Dis. 198, 1037–1043 (2008).

48. Martineau, A.R. et al. Reciprocal seasonal variation in vitamin D status and tuberculosis notifications in Cape Town, South Africa. Proc. Natl. Acad. Sci. USA 108, 19013–19017 (2011).

49. Barnes, I., Duda, A., Pybus, O.G. & Thomas, M.G. Ancient urbanization predicts genetic resistance to tuberculosis. Evolution 65, 842–848 (2011).

50. Ramachandran, S. et al. Support from the relationship of genetic and geographic distance in human populations for a serial founder effect originating in Africa. Proc. Natl. Acad. Sci. USA 102, 15942–15947 (2005).

A rt i c l e s

npg

© 2

013

Nat

ure

Am

eric

a, In

c. A

ll rig

hts

rese

rved

.

Nature GeNeticsdoi:10.1038/ng.2744

ONLINEMEThODSData sets. MTBC data sets. We have analyzed a total of 259 MTBC strains (including 1 Mycobacterium canettii strain used as the outgroup). We used two different strain sets for different aspects of the analyses.

(1) Global MTBC data set (n = 220). This data set represents a global col-lection of MTBC clinical strains covering all the known phylogenetic lineages of MTBC and including representatives from 46 countries. In addition, three strains from the animal-adapted lineage (including one strain of the Mycobacterium bovis BCG vaccine) were included as reference, and one strain of M. canettii was used as the outgroup. More detailed information can be found in Supplementary Table 1.

(2) MTBC lineage 2–enriched data set (n = 75). To explore the evolution of MTBC in a regional setting, we extended our collection of 36 MTBC strains from lineage 2 with an additional 39 strains that represent the population diversity of lineage 2 in China based on standard genotyp-ing (Supplementary Table 1).

Illumina reads for the genomes of the new MTBC strains sequenced and described in this study have been deposited under project number ERP001731.

Human mitochondrial data set. For comparisons with human genetic diver-sity, we analyzed large data sets of complete mitochondrial genomes. There are limitations inherent to mitochondrial DNA. First, estimating the most frequent mitochondrial DNA haplogroup in a particular country is always difficult and is dependent on sampling. Second, mitochondrial DNA con-tains limited phylogenetic information. However, the reasons to focus on a mitochondrial marker rather than on a chromosomal marker include (i) the availability of information for most regions and countries in terms of mito-chondrial DNA haplogroup frequencies and (ii) the possibility of comparison with previously published studies dealing with human mitochondrial DNA haplogroups, human migrations and population dynamics. We used three different sets of human mitochondrial genomes that were available in public repositories. These are listed in Supplementary Tables 3, 5 and 6.

(1) Global reference data set of human mitochondrial DNA (n = 4,955). This data set is a compilation of most of the publicly available human mitochondrial genomes for which the haplogroup has been deter-mined18. This data set includes representatives of most known human mitochondrial macrohaplogroups and derived haplogroups.

(2) Neolithic population expansion data set of human mitochondrial DNA (n = 423). This data set is derived from the data set reported by Gignoux et al.38 and includes selected representative haplogroups known to have their origin either before, during or shortly after the Neolithic period. This data set is therefore maximized to detect signatures of popula-tion expansion around this period that could be obscured by earlier expansion events.

(3) East Asia–enriched Neolithic data set of human mitochondrial DNA (n = 72). For MTBC lineage 2, we complemented the data set for East Asia by adding any newly published human mitochondrial genome from the mitochondrial DNA haplogroups of interest (B4a1, F1a1, E1a and E1b).

Sequencing of MTBC strains. The majority of MTBC strains were sequenced during the present project at different sequencing centers (GATC (Germany), Wellcome Trust Sanger Institute (UK) and Southern Genome Center (China)); a few additional sequences were retrieved from publicly available databases. MTBC DNA was extracted using standard procedures. Single- or paired-end multiplexed Illumina sequencing was performed as described previously51. Briefly, sequencing was performed on a HiScanSQ instrument with TruSeq SBS kit HS chemistry (Illumina) to generate sequencing reads of between 51 and 100 bases in length, depending on the strain. Average genome coverage was 146.5 of the reference genome (strain-specific genome coverage is shown in Supplementary Table 1).

Mapping Illumina sequencing reads and SNP calling. Sequencing reads for each MTBC strain were mapped to the inferred MRCA of MTBC as previously

determined52 (the sequence of the MTBC MRCA is available upon request). We used two mapping approaches, the ungapped Mapping and Assembly with Quality (MAQ)53 algorithm and the Burrows-Wheeler algorithm described in BWA54, along with the MAQ SNP caller and SAMtools55, respectively, to generate two different lists of SNPs. We kept those polymorphic positions called by both approaches (Supplementary Table 7). For a complete descrip-tion of the SNP-calling procedure and annotation of the positions, see the Supplementary Note, as well as Supplementary Figure 10 for a workflow of the SNP-calling procedure.

Phylogenetic and principal-component analyses. Human mitochondrial DNA data sets were obtained from the database of variant positions used by Behar et al.18. For the population expansion data set of human mitochondrial DNA during the Neolithic period, sequences from the relevant accessions described in Gignoux et al.38 were downloaded, and genomes were aligned using the ClustalW56 implementation in the BioEdit package57 followed by manual curation. We removed the poorly aligned region known as the D-loop and kept polymorphic sites for subsequent phylogenetic and coalescent analy-ses. For the MTBC data sets, we used variable positions for all downstream analyses. In both cases, we applied phylogenetic distance as well as maximum-likelihood methods. For a complete description of the phylogenetic analyses, the identification of homoplastic sites and the principal-component analysis of the SNPs used, see the Supplementary Note.

Phylogeographic analyses. For the phylogeographic analyses, we used the BSSVS model implemented in BEAST 1.6 (ref. 58). We also used RASP59, which implements both Bayesian and parsimony approaches to analyze the ancestral geographic ranges of MTBC lineages. We subdivided the world map into seven broad geographic areas and used them as a proxy for the most likely origin of each strain (see Supplementary Fig. 1 for subdivisions and Supplementary Table 1 for the origins of infected individuals). We used broad geographic areas instead of exact locations because the large number of locations to consider and, hence, the exchange rates to estimate would be unmanageable if using all individual countries. Predefined geographic areas were introduced for each MTBC strain according to the country of origin of the infected individuals. See the Supplementary Note for a complete descrip-tion of the settings for the different phylogeographic analyses.

MTBC–mitochondrial DNA association test. We tested the hypothesis that modern lineages 2– 4, lineage 1 and the African lineages 5 and 6 are associated with the N, M and L human mitochondrial DNA lineages, respectively. To this end, we assigned for each MTBC strain from a given country a mitochondrial DNA haplogroup according to the frequency of the haplogroup in that country on the basis of a review of the published literature (Supplementary Fig. 4). Only the two most frequent MTBC lineages of a country and the two most frequent mitochondrial DNA haplogroups were considered, unless only one MTBC lineage occurred in the country, in which case it was assigned to the most frequent mitochondrial DNA of the country (Supplementary Table 4). We used BaTs (Bayesian Tip-association significance testing)60 to test whether the main lineages of MTBC for each country tended to be associated with a particular human mitochondrial DNA macrohaplogroup (L, M or N) or haplogroup (A, B, D, E, F, G, H, K, L, M, R or U) (Supplementary Fig. 4, Supplementary Table 4 and Supplementary Note). For the tests, we assumed that there was no MTBC lineage that corresponded with the L0, L1, L2 and L4 human mitochondrial lineages on the basis of the fact that no lineage 5 or 6 strains are found outside of West Africa where the human mitochondrial L3 haplogroup has the highest frequency31. However, even when we introduce L0, L1, L2 and L4, the test results did not change. BaTs implements two association indexes, a parsimony score that quantifies the number of state changes in the phylogeny (a low number indi-cates high clustering of states) and an association index that examines internal nodes and records the most frequent state in the taxa downstream of the node. A statistical test was carried out by reshuffling the various states across the phylo-geny. Given the constrained phylogeographic distribution of lineages 5 and 6 (Mycobacterium africanum) to West Africa and their basal but close position to all the Out-of-Africa lineages, these M. africanum lineages correlate best with the human mitochondrial L3 haplogroup, which shows remarkable similarities.

npg

© 2

013

Nat

ure

Am

eric

a, In

c. A

ll rig

hts

rese

rved

.

http://www.ebi.ac.uk/ena/data/view/ERP001731

Nature GeNetics doi:10.1038/ng.2744

BEAST analyses. We used BEAST v. 1.6 (ref. 19) to date the evolutionary events and population dynamics of MTBC and the human mitochondrial DNA haplogroups. BEAST implements the joint sampling of the posterior distribution of different evolutionary parameters, such as the substitution rate or the population size, under a coalescent framework. In all cases, we used a skyline plot before looking for changes in population size over time. For MTBC, we used two data sets. To explore different dating hypotheses, we used the complete MTBC data set, a total of 216 strains excluding the outgroup (M. canettii) and the animal-based strains. We used an uncorrelated log- normal distribution for the substitution rate in all cases. We imposed different prior values on the coalescent times of lineages 5 and 6 according to plausible time estimates. Because no fossil records or good substitution rate estimates are available for MTBC, we used this approach as a way to narrow down the origin and age of the extant strains of MTBC. We imposed normal distribu-tions in the coalescent time of lineages 5 and 6, as time estimates for mito-chondrial haplogroups are usually given in coalescent times and not in times of splitting events between groups: 185,000 (± 20,000), 70,000 (± 10,000) and 10,000 (± 2,000) years ago. We also added as a second anchor point the split of MTBC lineage 1 with a normal prior of 65,000 ± 10,000 years ago, based on the coincident geographic distribution of lineage 1 with human mitochondrial macrohaplogroup M. Similar approaches were followed to analyze mitochon-drial DNA data sets, where we used both a molecular clock approach (by specifying a published substitution rate31) and a dating approach (by assuming that the height of the phylogeny was distributed normally around 185,000 years ago as a mean ± 20,000 years ago). Both approaches yielded similar results,

and we report the results for the dating analyses. Similarly, for the East Asian clade, we specified priors for the age of the whole data set (60,000 ± 10,000 years ago) and for the individual haplogroups as described in the literature (B4a1, 11,000 ± 3,000 years ago; E1a, 9,000 ± 3,000 years ago; E1b, 6,000 ± 3,000 years ago)18. For a detailed description of the models and the statistical comparison of skyline plots, see the Supplementary Note.

51. Quail, M.A. et al. A large genome center ‘s improvements to the Illumina sequencing system. Nature Methods 5, 1005–1010 (2008).

52. Comas, I. et al. Human T cell epitopes of Mycobacterium tuberculosis are evolutionarily hyperconserved. Nat. Genet. 42, 498–503 (2010).

53. Li, H., Ruan, J. & Durbin, R. Mapping short DNA sequencing reads and calling variants using mapping quality scores. Genome Res. 18, 1851–1858 (2008).

54. Li, H. & Durbin, R. Fast and accurate long-read alignment with Burrows-Wheeler transform. Bioinformatics 26, 589–595 (2010).

55. Li, H. et al. The sequence alignment/map format and SAMtools. Bioinformatics 25, 2078–2079 (2009).

56. Larkin, M.A. et al. Clustal W and Clustal X version 2.0. Bioinformatics 23, 2947–2948 (2007).

57. Hall, T.A. BioEdit: a user-friendly biological sequence alignment editor and analysis program for Windows 95/98/NT. Nucleic Acids Symp. Ser. 41, 95–98 (1999).

58. Lemey, P., Rambaut, A., Drummond, A.J. & Suchard, M.A. Bayesian phylogeography finds its roots. PLoS Comput. Biol. 5, e1000520 (2009).

59. Yu, Y., Harris, A.J. & He, X. S-DIVA (Statistical Dispersal-Vicariance Analysis): a tool for inferring biogeographic histories. Mol. Phylogenet. Evol. 56, 848–850 (2010).

60. Parker, J., Rambaut, A. & Pybus, O.G. Correlating viral phenotypes with phylogeny: accounting for phylogenetic uncertainty. Infect. Genet. Evol. 8, 239–246 (2008).

npg

© 2

013

Nat

ure

Am

eric

a, In

c. A

ll rig

hts

rese

rved

.

“The struggle [against tuberculosis] has caught hold along the whole line and enthusiasm for the lofty aim runs so high that a slackening is no longer to be feared. If the work goes on in this powerful way, then the victory must be won.”

It is with these powerful and optimistic words that Robert Koch, discoverer of the tubercle bacillus, concluded his Nobel Lecture on December 12, 1905. Armed only with the conviction of the infectious nature of the disease and the importance of isolating infected individuals to decrease transmission, and convinced of the need to improve the social environment of patients with tuber‑culosis (TB), Koch was confident that society could win the battle against this disease. More than 100 years on (TIMELINE), how has TB control changed? Are we any closer to seeing this scourge eliminated in the foresee‑able future? In this Review, we discuss the challenges and obstacles associated with global TB control, particularly those associated with the development of DOTS (directly observed therapy, short course) and the WHO Stop TB Strategy, and assess how the lessons learnt can aid future action towards elimination of the disease.

Early TB control policiesThroughout the 20th century, TB mortality declined stead‑ily in most industrialized countries, with interruptions during the two World Wars1. In England and Wales, the decline in reported mortality from TB that began to be observed in the mid‑nineteenth century continued stead‑ily until the 1960s1, most probably as a result of improved

socioeconomic conditions, better nutrition and living standards, and the isolation of infectious patients2 (FIG. 1a). In the Netherlands, mortality due to TB fell by ~3% annu‑ally from 1900–1920 and by ~5.5% in the 1920s and 1930s3 (FIG. 1b). The data for developing countries, especially in the pre‑chemotherapy era, are scarce, but information can be obtained from tuberculin surveys that were carried out by the WHO between the 1950s and 1970s4. In Uganda, the annual risk of infection decreased by 1.4% from 1950 to 1970, which equated to halving the incidence of TB in more than 50 years, whereas in the Netherlands the annual risk of TB infection decreased by 14% over the same period, leading to a halving of the incidence of TB in 5 years3.

The development of the Mycobacterium bovis bacil‑lus Calmette–Guérin (BCG) vaccine generated high hopes for TB prevention. First administered in 1921, it was used increasingly in Europe following early evi‑dence of its efficacy from studies of student nurses in Norway5. After the Second World War, the re‑emergence of TB became a major public health concern, and mass BCG vaccination campaigns were encouraged in many countries within and outside of Europe, stimulated by UNICEF, the Scandinavian Red Cross Societies and the WHO5. Routine BCG vaccinations were established in many countries worldwide using various schedules, and the vaccine was incorporated into UNICEF’s expanded programme on immunization of infants in 1974.

The discovery of streptomycin by Waksman in 1944, and the nearly simultaneous development of para‑ aminosalicylic acid (PAS) by Lehmann led to an effective

Stop TB Department, World Health Organization, Desk D4 6017, 20 Avenue Appia, CH-1211 Geneva 27, Switzerland.Correspondence to C.L. e-mail: [email protected]:10.1038/nrmicro2797Published online 14 May 2012

Global tuberculosis control: lessons learnt and future prospectsChristian Lienhardt, Philippe Glaziou, Mukund Uplekar, Knut Lönnroth, Haileyesus Getahun and Mario Raviglione

Abstract | Tuberculosis (TB) is an ancient disease, but not a disease of the past. After disappearing from the world public health agenda in the 1960s and 1970s, TB returned in the early 1990s for several reasons, including the emergence of the HIV/AIDS pandemic and increases in drug resistance. More than 100 years after the discovery of the tubercle bacillus by Robert Koch, what is the status of TB control worldwide? Here, we review the evolution of global TB control policies, including DOTS (directly observed therapy, short course) and the Stop TB Strategy, and assess whether the challenges and obstacles faced by the public health community worldwide in developing and implementing this strategy can aid future action towards the elimination of TB.

R E V I E W S

NATURE REVIEWS | MICROBIOLOGY VOLUME 10 | JUNE 2012 | 407


mailto:[email protected]

treatment that had a marked effect on TB mortality, end‑ing the sanatorium era. The establishment of the first multi‑therapy approach combining streptomycin and PAS was based on the results of a landmark clinical trial carried out by the UK Medical Research Council that showed the superiority of the combined treatment over either agent alone6. The discovery of isoniazid in 1951 completed the armamentarium of what became the first triple therapy for an infectious disease7, curing patients with TB in 18–24 months. PAS was replaced by etham‑butol in the early 1960s. The addition of rifampicin in the 1970s and the replacement of streptomycin by pyrazina‑mide in the 1980s were key to the development of the short‑course chemotherapy of 6–8 month duration that became the pivotal element of modern TB control8.

In 1947, based on the expected high impact of mass BCG vaccination campaigns, and the discovery of the first anti‑TB drugs, the WHO declared TB a priority and established a TB section to assist governments in developing effective control programmes9. This inau‑gurated the era of large vertical control programmes (1948–1963), based on mass vaccination, mass radio‑logical screening and TB case management10 (vertical programmes are single‑purpose programmes that are independent of both the general health infrastructure and the structure of other vertical programmes, and are staffed with specialized personnel from the central level to the local level at which the technical control activities are delivered). This approach, which took place in an era of sustained economic development, was successful in many industrialized countries, thereby accelerating the decline in the incidence of TB3,11.

However, it was not easily transferred to developing countries1; large‑scale mass BCG vaccinations did not have a long‑lasting effect on TB epidemiology5,12, and anti‑TB campaigns could not be sustained outside of the general health services in resource‑poor settings. This led to a radical move towards integration of TB pro‑grammes into general health services in the mid‑1960s,

based on pilot studies carried out in India13,14. This new policy was formalized in the eighth report of the WHO Expert Committee in 1964 (REF. 15), which empha‑sized the integration of service delivery, using simple standardized diagnostics and treatment approaches. Unfortunately, in most resource‑poor settings, the transfer of responsibilities from vertical to general health services was not accompanied by an increase in resources, leading to weakened TB control with almost no effect on TB indicators. To remedy this, when the WHO embraced primary healthcare (PHC) in the late 1970s, a second wave of integration occurred, aimed at further integration of service delivery and full manage‑rial functions, dismantling the remains of the vertical approaches10. This resulted in severe weakening of case‑finding and treatment activities in many developing countries, mainly owing to the deterioration of public health infrastructure, compounded by a series of eco‑nomic crises. The neglect of TB control worldwide was further accelerated in the late‑1980s with the promotion of health sector reforms that involved decentralization of authority and managerial integration of programmes16. This shift in the overall concept of health management had a staggering effect on disease control programmes, which temporarily disappeared as specific entities10,17.

The resurgence of TB in the 1990sThe 1990s saw major disruptions in the world, with the end of the Cold War, the dissolution of the former Soviet Union and the inexorable expansion of the HIV/AIDS pandemic. New global TB estimates in 1989–1990 revealed a huge burden of disease in developing coun‑tries, accounting for an estimated 7.1 million out of 8 million new cases globally, with TB becoming a leading cause of mortality in sub‑Saharan Africa18. The driving forces were the HIV/AIDS pandemic, which produced sharp increases in notifications of TB cases, particularly in sub‑Saharan Africa19, associated with insufficient case detection and low cure rates within disorganized and

Timeline | Landmarks in TB control

1882 1921 1944 1948 1952 1959 1974 1978 1980 1981 1982 1989 1991 1993 1994 1997 1998 1999 2000 2001 2002 2003 2004 2006 2008 2010

•WHO recognizes TB as a major public health problem

•CDC reports MDR-TB epidemics in the United States

Discovery of the tubercle bacillus

First use of BCG vaccine

Discovery of streptomycin and PAS

Efficacy of streptomycin demonstrated in pioneering RCT

First combination treatment regimen: S–PAS–H for 24 months

Short-course chemotherapy: H–R–Z–E for 6–8 months

Recognition of the impact of the HIV/AIDS epidemic on TB

First WHO meeting on TB–HIV co-infection

Emergence of the HIV/AIDS epidemic

Self-administered ambulatory treatment shown to be as safe and efficient as hospital treatment of TB

Emergency declaration by the WHO

BCG part of UNICEF’s EPI Programme

“Health for all by the year 2000”

BCG, Mycobacterium bovis bacillus Calmette–Guérin; DOTS, directly observed therapy, short course; EPI, expanded programme of immunization; FIND, Foundation for Innovative New Diagnostics; H–R–Z–E, isoniazid–rifampicin–pyrazinamide–ethambutol; MDR-TB, multidrug-resistant TB; PAS, para-aminosalicylic acid; RCT, randomized controlled trial; S–PAS–H, streptomycin–PAS–isoniazid; TB, tuberculosis; XDR-TB, extensively drug-resistant TB.

R E V I E W S

408 | JUNE 2012 | VOLUME 10 www.nature.com/reviews/micro


insufficiently resourced TB control programmes1,20. In parallel, TB started re‑emerging in several industrial‑ized countries. In the USA, after years of decline, the number of newly reported cases began to increase in 1986, peaking in 1992 (REF. 21). A similar situation was observed in most western European countries, and in Eastern Europe the crisis resulting from the collapse of the former Soviet Union led to a major increase in TB incidence and mortality in practically all the newly established states20.

The development of DOTS. This alarming situation led to a re‑appraisal and reinvigoration of global TB con‑trol activities. After decades of neglect, a new strategy was promoted by the WHO, based on the approach developed by Karel Styblo and the International Union Against Tuberculosis and Lung Disease in Tanzania and Malawi in the 1980s3, which involved moving away from strict integration management theories and emphasized the need for specialized managerial func‑tions at all levels of health care9,10,22,23. In 1991, the 44th World Health Assembly (WHA) adopted a resolution that recognized TB as a major public health problem24. Two global targets were set for the year 2000: detect‑ing 70% of infectious cases and curing 85% of them. As poor adherence to, and premature interruption of, treat‑ment contribute to prolonged infectiousness and drug resistance, the newly developed strategy, which centred around direct supervision of drug intake by patients, was labelled DOTS, and was characterized by a package of five elements (BOX 1) that constituted a framework for effective TB control25. A key element of this strategy was the supervision of drug intake. This approach relied on early work by Wallace Fox, who investigated the possibil‑ity of ambulatory (that is, outpatient) self‑administration of drugs in the 1950s26. The results of a controlled trial in Chennai, India, showed that “the merits of domiciliary therapy [were] comparable to those of sanatorium treat‑ment” (REF. 27). However, to be succesful, domiciliary

therapy required an adequate supply of drugs, proper staffing, adequate transport, availability of hospital beds for referrals, supervision of drug intake by members of the patient’s family or neighbours, and surprise visit checks. This notion was further embraced in the United States as a fundamental component of proper TB care and named directly observed therapy (DOT)28.

In 1995, after the establishment of a TB global sur‑veillance and monitoring system, in principle 23% of the world’s population had access to DOTS29. This figure reached 56% by 1998 (REF. 30), and by 1999 127 coun‑tries had adopted this approach, including the 22 high‑est burden countries, which collectively were responsible for 80% of global TB mortality31–33. By 2005, 89% of the world’s population was living in areas where DOTS services were available34. However, in 2000 it was esti‑mated that only 27% of smear‑positive cases worldwide were reported and managed by DOTS programmes35. Furthermore, the effectiveness of DOTS was questioned, particularly in areas of high HIV prevalence and in resource‑poor settings36–39. Indeed, DOTS expansion in the early years was limited by many factors, including the lack of uniform and complete coverage within countries, the high variability of interventions incorporated in TB control practices, the perceived lack of flexibility and adaptability of the strategy, and the lack of explicit refer‑ence to social support to facilitate access and adherence to treatment38–41. This called for a revision that would allow wider access for all infected individuals, particu‑larly for the underprivileged in the poorest countries10, and became the foundation for the development of the new Stop TB Strategy in the mid‑2000s (see below).

The new challenges: HIV infection and MDR‑TB. The link between TB and HIV was reported soon after the first descriptions of AIDS in 1981 (REFs 42,43). HIV infection rapidly emerged as the strongest risk fac‑tor for the development of TB in individuals infected with Mycobacterium tuberculosis19,44, with the risk up to

Timeline | Landmarks in TB control

1882 1921 1944 1948 1952 1959 1974 1978 1980 1981 1982 1989 1991 1993 1994 1997 1998 1999 2000 2001 2002 2003 2004 2006 2008 2010

Genome sequencing of Mycobacterium tuberculosis

First global survey of anti-TB drug resistance

ProTEST TB–HIV projects started

UNAIDS and WHO policy on isoniazid preventive therapy

Creation of the Global Alliance for TB Drug Development

Creation of AERAS TB Vaccine Foundation

•Creation of FIND •WHO TB–HIV

policy launchedCreation of the TB Vaccine Initiative

•Establishment of the Green Light Committee

•Ministerial meeting on TB, Amsterdam, The Netherlands

•Global Drug Facility launched

•Formalization of the Stop TB Partnership governance

•Global Plan to Stop TB 2001–2005

Establishment of the Global Fund to fight AIDS, Tuberculosis and Malaria

•Global Plan to Stop TB 2006–2015

•Establishment of UNITAID

•Launch of the WHO Stop TB Strategy

•Emergence of XDR-TB

Updated Global Plan to Stop TB 2011–2015

First WHO ad hoc committee on the TB epidemic in London, United Kingdom

Launch of the new framework for effective TB control, leading to DOTS strategy launch in 1995

R E V I E W S



Nature Reviews | Microbiology

Dea

ths

(mill

ions

yea

r-1)

Infe

ctio

n ra

te (1

0,00

0 ye

ar-1

)

500

0

1,500

2,000

1,000

19601920 1940

Year

Year

First World War

1900

10

1

100

2,000

1,000

1960 19801920 1940 1950 19701910 19301900

SecondWorld War

First anti-TB drug available

Incidence (reactivated cases excluded), since 1951Reactivated cases, since 1951Mortality, since 1901Risk of tuberculosis infection, since 1910

a

b

37 times higher than in HIV‑negative patients infected with the bacterium45. This resulted in an increase in TB case notifications in HIV‑prevalent countries, particularly in sub‑Saharan Africa, which mirrored the increase in the prevalence of HIV infection with a 4–7‑year delay45. TB control faced increasing challenges, with rising case notifications, difficulties in diagnosis, drug‑related side effects, high case‑fatality rates, increased recurrence rates and increased transmission of M. tuberculosis within crowded settings44,46. In response to this, in 1997 the WHO launched the ProTEST initiative, an opera‑tional research project to address the unprecedented scale of the epidemic of HIV‑related TB and develop a district‑based strategy for joint TB–HIV control efforts. This initiative promoted voluntary HIV testing as the point of entry for access to HIV and TB interventions,

including intensified TB case finding and isoniazid pre‑ventive therapy47. Evaluation of pilot projects conducted in Malawi, South Africa and Zambia48 and elsewhere in Africa and Asia by organizations such as Family Health International and the CDC49 showed that HIV/AIDS and TB control programmes can work together effec‑tively and that such collaborative activities are necessary to improve the care available for people co‑infected with TB and HIV. The rapid expansion of access to antiretro‑viral therapy (ART) through the 3 by 5 initiative (which aimed to provide 3 million people living with HIV/AIDS in low‑ and middle‑income countries with ART by the end of 2005) has accelerated the implementation of the “policy on collaborative TB/HIV activities” (REF. 50).

At the same time, in the early 1990s, outbreaks of multidrug‑resistant (MDR) TB (that is, strains resistant to isoniazid and rifampicin) were reported in the United States, mainly among HIV‑infected individuals51–54. By the end of the 1990s, virtually all countries participating in a global survey of anti‑TB drug resistance reported MDR‑TB cases55. This emerging problem was linked to improper prescribing practices, lack of patient adher‑ence to treatment and irregular supply and low quality of drugs, all of which reflected poor TB control practices56. In addition, ongoing primary transmission was found to contribute to rising rates of MDR‑TB55,56.

Building on the DOTS strategy, a new programme‑based approach was devised to address the rise in MDR‑TB around the world57. In June 2000, the WHO, together with other partners (Harvard Medical School, the CDC, Médecins Sans Frontières, the Royal Dutch Tuberculosis Foundation and the National TB Programme of Peru) established a pooled procure‑ment mechanism to increase the availability of quality second‑line drugs at low cost, named the Green Light Committee (GLC)58. The objective was to allow coun‑tries in which TB is endemic to access concessionally priced and quality‑assured second‑line anti‑TB drugs59, while ensuring safe and rational drug use.

How TB control reappeared on the global public health agenda. By 1998, only a handful of countries had achieved the 70% case detection and 85% treatment success targets that had been set by the 44th WHA, so the WHO convened an ad‑hoc committee in London, United Kingdom, to address the delays in implementa‑tion of the DOTS strategy. The committee made sev‑eral key recommendations that modified the approach to worldwide TB control60. First, it established that the response to the TB epidemic had to be based on a new coalition with common aims of all partners and govern‑ments worldwide. Subsequently, the Stop TB Initiative, a coalition of stakeholders in the global fight against TB, was launched at the World TB Conference in Bangkok, Thailand, in November 1998. It called for the develop‑ment of a global action plan for TB control focusing on high‑burden countries, and addressing drug‑resistant and HIV‑associated TB61. Second, the committee called for a central mechanism of drug procurement and supply that would address the serious issue of first‑line drugs being out of stock in numerous developing countries.

Figure 1 | Early declines in TB mortality in Europe. a | Decline in tuberculosis (TB) mortality in England and Wales for the period 1900–1960. Data from REF. 101.b | TB incidence, reactivation and death rates (per 100,000) and the average annual risk of TB infection (per 10,000) in the Netherlands for the period 1901–1983. Data from REF. 3.

R E V I E W S



In accordance with this, the Global TB Drug Facility (GDF) was established in 2001 and delivered its first drugs to Moldova in May of that year62. Third, the com‑mittee called on the governments of the 22 high‑burden countries to commit to TB control. Following a ministe‑rial meeting in Amsterdam, the Netherlands, on World TB Day in 2000, a declaration for action was issued, stat‑ing the strategic directions and targets that were to be achieved by a new deadline of 2005 (REF. 63). This was endorsed by the WHA 2 months later64.

The original Stop TB Initiative did not function immediately. However, by the year 2000, a new environ‑ment was created by a change of leadership, the good‑will of key partners and the agreement reached around the GDF. In February 2001, in Bellagio, Italy, under the auspices of the Rockefeller Foundation, a new govern‑ance was approved based on a secretariat located at the WHO, a permanent coordinating board and several technical working groups. The Stop TB Partnership was established61, and the first coordinating board meeting was held in 2001 in Annapolis, Maryland, United States.

A business plan was needed to describe the strate‑gic directions that the Stop TB Partnership would take to make a difference in TB control. The ‘Global Plan to Stop TB 2000–2005’ defined strategies and activities to expand DOTS coverage, address HIV‑associated TB and MDR‑TB, and pursue innovative research for new TB diagnostics, drugs and vaccines65. This was followed by a second plan that set the global goals to halve TB prevalence and mortality compared with 1990 levels by 2015, the deadline year for the Millennium Development Goals (MDG), and to achieve TB elimination, defined as less than one case per million population per year, by 2050 (REF. 66).

In parallel, talks were held at the G8 Summit in Okinawa, Japan, in 2000, about a “new global part‑nership to address the infectious diseases of poverty, including TB”. Moroever, at the UN General Assembly in June 2001, the then Secretary‑General of the UN, Kofi Annan, called for the creation of a global fund to fight HIV/AIDS. The merging of the two concepts led to the establishment of a Global Fund to fight AIDS, Tuberculosis and Malaria, which was announced at the G8 Summit in Genoa, Italy, in 2001 (REF. 67) and began operating formally in early 2002 (REF. 68). The first dis‑bursements were made in 2004, and TB control budgets

began to increase dramatically. Later in the decade, other international financing mechanisms were created, including UNITAID, a funding initiative proposed by France, Brazil and Chile based on a tax levy on air flights to support essential commodities for the fight against HIV/AIDS, TB and malaria (see UNITAID website). Support for TB control greatly increased in the 2000s, and by 2011, US$3.1 billion was invested in TB con‑trol, approximately three times more than was invested in 2002.

The development of the Stop TB StrategyThe increased political commitment from high‑burden countries and their national and international partners led to progress in global TB control. By 2004, more than 20 million patients had been treated in DOTS programmes around the world, and more than 16 mil‑lion of them had been cured69. However, although the implementation of DOTS helped achieve good progress, it was insufficient to accomplish the international tar‑gets of halving TB mortality and prevalence by 2015 (REF. 70). Urgent action was needed, particularly in parts of the world where the epidemic was becoming worse, notably in Africa, Eastern Europe and Asia. In these areas, identifying and reaching those in need of care required concomitant progress of efforts to control TB and efforts to strengthen health services as a whole. Thus, in 2005, based on crucial reports71,72, the WHA passed a resolution advocating for “sustainable financ‑ing for TB control and prevention”, with countries mak‑ing a commitment to strengthen efforts to achieve the TB‑related international targets73.

Since the development and initial promotion of DOTS, the WHO and other international partners had been exploring complementary approaches to address the major constraints in TB control, particularly the new challenges of TB–HIV co‑infection and MDR‑TB. Together with the rapid expansion of access to anti‑retrovirals, collaborative TB–HIV activities were defined and implemented50. Strategies for programme‑based management of MDR‑TB were developed and tested74, and effective ways of undertaking community care to support patients and expand access were identified75. In addition, there was increased recognition of the crucial part played by patients and their community members in the design and implementation of policies and pro‑grammes, both globally and nationally76,77. The WHO and its partners developed evidence‑based strategies to engage diverse public, voluntary, corporate and private providers to widen the network of TB services offered78, and formulated and published the International Standards for TB Care to ensure quality of care across all providers79. Moreover, they piloted initiatives that strengthened primary respiratory care while expanding quality services for TB80.

In addition to mechanisms such as the GDF and GLC, which aimed to improve access to quality drugs for TB and MDR‑TB, respectively59,62, options for tackling poverty in TB care and control were inves‑tigated81. New resources were becoming available for health systems and disease control from domestic and

Box 1 | The five components of DOTS

•Generatinggovernmentcommitmenttomobilizesufficientresourcesfortuberculosis(TB)control.

•Casedetectionthroughpassivecasefindingusingsputumsmearmicroscopyinpatientswithrespiratorysymptoms.

•Treatmentusingstandardshort-coursechemotherapyregimenscontainingrifampicin,administeredunderdirectobservationforatleastthefirst2 monthsoftreatment.

•Securingaregularsupplyofessentialanti-TBdrugs.

•Establishingareliablemonitoring,recordingandreportingsystemforprogrammesupervisionandevaluation.

DOTS,directlyobservedtherapy,shortcourse.

R E V I E W S



http://www.unitaid.eu

international sources, including: the Global Fund to fight AIDS, Tuberculosis and Malaria; bilateral agencies; and philanthropic organizations. This created a favourable environment to build a new strategy, expanding beyond DOTS and drawing attention to the need for compre‑hensive action on several fronts (BOX 2). The new Stop TB Strategy was formally launched on World TB Day in 2006. Building on the expansion of high‑quality DOTS, this strategy integrated the need to address the new chal‑lenges posed by TB–HIV co‑infection and MDR‑TB, the need to reach out to poor and vulnerable populations and foster private sector and community involvement in TB care and control, and the need to address the health system82. It provided the basis and the context for the Global Plan to Stop TB 2006–2015.

The role of research and developmentTB research and development stagnated for several decades, as the prevailing concept was that the essen‑tial tools were available and that better application of these tools would suffice. However, most developing countries in the 1990s and 2000s were using old tools

that had found their limits with TB–HIV co‑infection and the emergence of anti‑TB drug‑resistance: sputum microscopy and traditional solid culture when and where available as diagnostics; the four‑drug treatment regimen that was developed in the 1970s; and the same, largely ineffective BCG vaccine that has been in use since the 1920s.

The need for research to develop new tools for opti‑mal diagnosis, prevention and treatment of all forms of TB in people of all ages (including children and those living with HIV) was evident. The establishment of the Bill & Melinda Gates Foundation coincided with the establishment of other public–private partnerships that have played a major part in resurrecting TB research and development. The Global Alliance for TB Drug Development was created in New York, United States, in 2000, with support from the Rockefeller Foundation, with the aim of facilitating research and development of new TB drugs. A decade later, and with salient invest‑ment from various pharmaceutical companies, there are several promising compounds in the pipeline, and a few new drugs belonging to new classes of anti‑TB mol‑ecules might be released for clinical use in 2012–2013 (REFs 83,84). For TB diagnostics, FIND (Foundation for Innovative New Diagnostics) was established in 2004 in Geneva, Switzerland, with grants from the Bill & Melinda Gates Foundation and other donors to facilitate the development of new diagnostics. Only a few years after its creation, FIND has facilitated the development of new, effective diagnostics, such as innovative nucleic acid amplification tests85, and a series of new strate‑gies for diagnostics have been endorsed by the WHO86. Finally, in the field of vaccines, with the involvement of several academic groups, research institutions and public–private partnerships such as AERAS and the TB Vaccine Initiative (TBVI), a pipeline exists that contains at least 10 vaccine candidates, including modified and strengthened BCG vaccines, as well as novel vaccines designed to be used in prime‑boost vaccination strate‑gies87. This reinvigorated role of research was reflected in the revised Global Plan to Stop TB 2011–2015 (REF. 88). Overall, TB research and development increased sub‑stantially over the past decade, thanks to the involve‑ment of key donors, and today’s annual investment is more than $0.6 billion89. However, according to the esti‑mates in the Global Plan 2011–2015, expenditure on TB research and development should be ~$2 billion per year to deliver according to expectations88.

Impact of TB control since 1995The reductions in the burden of disease achieved to date are the result of more than 15 years of intensive efforts to improve TB care and control. Between 1995 and 2010, 46 million patients with TB were successfully treated in DOTS programmes, and up to 7 million lives were saved, including 2 million women and children90. Overall, the treatment success rate has increased from 57% in 1995 to 87% in 2009 among sputum smear‑positive cases.

In 2010, there were 5.7 million official case notifi‑cations of TB cases worldwide, up from 3.4 million in 1995, and there were an estimated 8.8 million incidents

Box 2 | Components of the Stop TB Strategy and implementation approaches

1. Pursue high-quality DOTS expansion and enhancement•Securepoliticalcommitment,withadequateandsustainedfinancing.

•Ensureearlycasedetectionanddiagnosisthroughquality-assuredbacteriology.

•Providestandardizedtreatment,withsupervision,andpatientsupport.

•Ensureeffectivedrugsupplyandmanagement.

•Monitorandassessperformanceandeffect.

2. Address TB–HIV, MDR-TB and the needs of poor and vulnerable populations•Scaleupcollaborativetuberculosis(TB)–HIVactivities.

•Scaleuppreventionandmanagementofmultidrug-resistant(MDR)TB.

•AddresstheneedsofcontactsofpatientswithTBandofpoorandvulnerablepopulations.

3. Contribute to strengthening the health system based on primary health care•Helptoimprovehealthpolicies,humanresourcedevelopment,financing,supplies,servicedeliveryandinformation.

•Strengtheninfectioncontrolinhealthservices,othercongregatesettingsandhouseholds.

•Upgradelaboratorynetworksandimplementpracticalapproachestolunghealth.

•Adaptsuccessfulapproachesfromotherareasandsectors,andfosteractiononthesocialdeterminantsofhealth.

4. Engage all care providers•Involveallpublic,voluntary,corporateandprivateprovidersthroughpublic–privatemixapproaches.

•PromotetheuseoftheinternationalstandardsforTBcare.

5. Empower people with TB and communities through partnership•Pursueadvocacy,communicationandsocialmobilization.

•FostercommunityparticipationinTBcare,preventionandhealthpromotion.

•PromoteuseofthePatients’CharterforTBCare.

6. Enable and promote research•Undertakeprogramme-basedoperationalresearch.

•Advocateforandparticipateinresearchtodevelopnewdiagnostics,drugsandvaccines.

DOTS,directlyobservedtherapy,shortcourse.

R E V I E W S



2000 20101995 20051990 2000 20101995 20051990 2000 20101995 20051990

Rat

e pe

r 100

,000

pop

ulat

ion

year

-1

350 AFR AMR EMR

EUR SEA WPR

300

250

200

150

100

50

0

Year

Incident TB cases, all forms HIV-positive TBTB case notifications

Year Year

2000 20101995 20051990 2000 20101995 20051990 2000 20101995 20051990

60

50

40

30

20

10

0

120

140

100

80

60

40

20

0

60

50

40

30

20

10

0

200

150

100

50

0

150

100

50

0


Year Year Year

(8.5 million−9.2 million) of TB globally (equivalent to 128 cases per 100,000 population)91. Global TB inci‑dence rates have been falling since 2002, and the number of case notifications has been falling since 2006 (REF. 91) (FIG. 2). If these trends are sustained, the MDG target will be achieved. However, global incidence is falling too slowly for TB elimination to be reached by 2050. Also in 2010, there were an estimated 0.9 million−1.2 mil‑lion TB‑related deaths among HIV‑negative individuals globally and 0.32 million−0.39 mil lion TB‑related deaths among HIV‑infected individuals. Global TB mortality rates fell by around 40% between 1990 and 2010 (REF. 91), and the target of a 50% reduction by 2015 could be achieved if the current rate of decline is sustained (FIG. 3); this target could be achieved in five out of the six WHO regions, the exception being the African Region.

HIV infection has markedly increased the number of TB cases reported annually from sub‑Saharan Africa, and in 2010 the African Region accounted for 74% of the world’s HIV‑positive TB cases. Overall, 34% of patients with TB in 2010 knew their HIV status (up from 28% in 2009)91, including 59% of patients in the African Region. Globally, a total of 300,000 HIV‑positive patients with TB were enrolled on co‑trimoxazole preventive therapy, and almost 200,000 on ART (77% and 46%, respectively, of patients with TB who were HIV positive). However,

the use of isoniazid to prevent the development of TB remains an inconsistently implemented global policy, and only 180,000 people living with HIV received prophy‑lactic treatment in 2010 (REF. 91). In 2010, there were an estimated 460,000–870,000 prevalent cases of MDR‑TB, mostly in China, India, the Russian Federation and South Africa. By November 2011, 58 countries and ter‑ritories had reported at least one case of extensively drug‑resistant TB (XDR‑TB). Among TB case notifications in 2010, an estimated 290,000 patients had MDR‑TB. Of these, slightly more than 46,000 patients (16%) were diagnosed with MDR‑TB and put on treatment91.

Future prospectsAs the international community continues to make strides to reach the TB control targets set for 2015, atten‑tion is gradually shifting towards TB elimination. Based on the progress made to date, the experiences arising from more than 50 years of global TB control and the lessons learnt, we suggest that actions will be required on four distinct but synergistic fronts to achieve this ambitious goal92.

Securing the core functions. Ensuring full and contin‑ued funding and implementation of all components of the Stop TB Strategy is the highest priority for all

Figure 2 | Recent trends in TB incidence and case notification rates. The graphs show trends in case notification rates (new and relapse cases, all forms), estimated incidence rates, including HIV-positive tuberculosis (TB) and estimated incidence rates of HIV-positive TB cases in the WHO regions for the period 1990–2010. The shaded areas represent uncertainty bands. AFR, Africa Region; AMR, Region of the Americas; EMR, Eastern Mediterranean Region; EUR, European Region; SEA, South East Asia Region; WPR, Western Pacific Region. Data from REF. 91.

R E V I E W S



Rat

e pe

r 100

,000

pop

ulat

ion

year

-1

2000 201520101995 200519902000 201520101995 200519902000 201520101995 20051990


AFR AMR EMR

EUR SEA WPR

80

60

40

20

0

Year Year Year

2000 201520101995 200519902000 201520101995 200519902000 201520101995 20051990Year Year Year

8

10

6

4

2

0

40

30

20

10

0

15

10

5

0

50

40

30

20

10

0

15

20

10

5

0

countries. When and where the fundamentals of TB care and control are not yet in place, efforts should first focus on establishing these principles, and when they are in place, should focus on sustaining them. In par‑ticular, it will be crucial to improve access and intensify action to reach out to those who are not diagnosed early enough through passive case‑finding approaches. This includes intensified screening of TB among people seek‑ing healthcare using sensitive tools and algorithms, as well as active case finding among TB contacts and other high‑risk groups92.

Health system support. The lack of progress in control efforts in recent years in some settings is largely due to the limitations of the health systems and services within which TB control programmes operate93. Inspiring statements about bold policy changes for health system strengthening must be urgently translated into action, including: increased public spending on health care; improved human resources for health care; universal coverage with free diagnosis and treatment; and full pro‑tection against the catastrophic direct and indirect cost of illness. Better intelligence on the TB burden requires improved health information systems. Alignment of care to best evidence across the health system requires much stronger collaboration with private practitioners, including increased education and technical assistance78.

Prevention and management of drug‑resistant TB requires enforced drug regulation, as well as rational prescribing and dispensing of drugs. Addressing all these barriers is a tall order and will require political will on the highest level, stimulated by forceful grassroots demands.

Investing in research. Any realistic prospect of elimi‑nating TB relies both on better and wider use of exist‑ing technologies and the development of new tools for TB diagnosis, treatment and prevention. Research must be accelerated across the continuum, from discovery (the basic science that underpins the development of new diagnostics, drugs and vaccines) to implementa‑tion research (ensuring that newly developed tools are rapidly and effectively taken up in areas where they are most needed). With this view, the TB Research Movement has been launched, with the aim of boost‑ing TB research and accelerating progress in TB control towards international targets94. This is, however, insuffi‑cient. As suggested by a recent mathematical model, the elimination of TB by 2050 relies on the combined and synergistic implementation of several new strategies, including improved diagnosis of drug‑susceptible and drug‑resistant TB, shorter treatment of overt TB cases (≤2 months), scaled‑up treatment of latently infected individuals (especially in high‑risk populations) and

Figure 3 | Recent trends in TB mortality. The graphs show tuberculosis (TB)-related mortality, including deaths in HIV-infected individuals, for the period 1990–2010 and projected mortality for the period 2011–2015, for the WHO regions. The shaded areas represent uncertainty bands. Uncertainty is comparatively lower when the data come from vital registration records. AFR, Africa Region; AMR, Region of the Americas; EMR, Eastern Mediterranean Region; EUR, European Region; SEA, South East Asia Region; WPR, Western Pacific Region. The dashed line represents the Stop TB Partnership target of 50% reduction in the mortality rates by 2015 compared with 1990 levels. Data from REF. 91.

R E V I E W S



mass vaccination campaigns using a more effective vac‑cine95. The goal of eliminating TB will remain elusive without the much needed transformation in research to develop new technologies for diagnosis, treatment and prevention94. To catalyse this transformation and pave the way for future research, in 2011 the TB Research Movement produced the International Roadmap for TB Research96, which outlines crucial priority areas for future scientific investment, with the aim of synergiz‑ing research efforts globally and catalysing the develop‑ment of new research collaborations to address difficult and unanswered questions in TB research97.

Addressing the social determinants of TB. The social and economic forces behind TB infection and disease have remained the same for centuries. Poverty, undernour‑ishment, and poor living and working conditions, main‑tained by social injustice, political instability and war, continue to create fertile ground for TB to spread and thrive. Rapid urbanization concentrates these conditions into large pockets of extreme poverty. Increased migra‑tion allows TB to be transmitted faster and over a wider area. Shifting lifestyles have brought many unhealthy choices to the developing world, with increasing preva‑lence of smoking98 and diabetes99, as well as alcohol and drug abuse100, all of which are risk factors for TB. These risk factors are associated with economic progress, but disproportionally affect the lower social classes in all soci‑eties, making those already vulnerable to TB infection even more vulnerable, completing the vicious circle that links the disease and poverty101.

Actions on social determinants can have a positive effect on TB control on at least three levels92. First, to improve TB treatment outcomes, treatment must be free of charge and delivered in a way that optimizes chances

of full treatment adherence and minimizes direct and indirect costs. Second, to improve early case detection, access barriers need to be addressed within a broader health system strengthening agenda, and high TB risk populations must be targeted with intensified and pro‑active case detection efforts. Finally, socioeconomic con‑ditions must be improved through social and economic reforms, and the prevalence of TB risk factors should be tackled through broad public health actions.

The fight against TB is becoming ever more multi‑faceted and has to be carried out on many fronts. The stakeholders involved in TB control need to become more diverse and grow in number. The health sector is but one of many sectors that need to be fully engaged. The research community needs better stimuli to take on new developments. All government sectors, non‑ governmental organizations, the private sector, civil society organizations and TB‑affected communities have key parts to play in creating demand for, and ensuring effective delivery of, both medical and social interventions102.

ConclusionsTo achieve maximum TB care and control, countries must move rapidly towards universal access to prevention, diag‑nostic and treatment services for all forms of TB. From a broader perspective, realizing the Stop TB Strategy’s vision of a TB‑free world will require policies and actions beyond the remit of TB control programmes and non‑state health stakeholders, and will need contributions from other sec‑tors to eradicate the key determinants of the epidemic. In the context of health and human development, research to accelerate progress in TB control is inextricably associ‑ated with efforts to alleviate poverty and promote social and economic development.

1. Styblo, K. in Selected Papers, Vol. 24 (KNCV, 1991).2. Lienhardt, C. From exposure to disease: the role of

environmental factors in susceptibility to TB. Epidemiol. Rev. 23, 288–301 (2001).

3. Styblo, K. Overview and epidemiological assessment of the current global tuberculosis situation with an emphasis on control in developing countries. Rev. Infect. Dis. 1, S339–S346 (1989).

4. Roelsgaard, E., Iversein, E. & Blocher, C. Tuberculosis in tropical Africa. Bull. World Health Organ. 30, 459–518 (1964).

5. Fine, P. E. M., Carneiro, I. A. M., Milstien, J. B. & Clements, C. J. Issues Relating to the Use of BCG in Immunisation Programmes. (WHO, 1999).

6. Medical Research Council. Treatment of pulmonary tuberculosis with para-aminosalicylic acid and streptomycin: a preliminary report. BMJ 2, 1521 (1949).

7. Crofton, J. Chemotherapy of pulmonary tuberculosis. BMJ 1, 1610–1664 (1959).

8. Fox, W., Ellard, G. A. & Mitchison, D. A. Studies on the treatment of tubeculosis undertaken by the British Medical Research Council Tuberculosis Units, 1946–1986, with relevant consequent publications. Int. J. Tuberc. Lung Dis. 10, S231–S279 (1999).

9. World Health Organization. Official Records of the World Health Organization, No 6. (WHO, 1947).

10. Raviglione, M. D. & Pio, A. Evolution of WHO policies for tuberculosis control, 1948–2001. Lancet 359, 775–780 (2002).

11. Styblo, K. Recent advances in epidemiological research in tuberculosis. Adv. Tuberc.Res. 20, 1–63 (1980).

12. Styblo, K. & Meijer, J. Impact of vaccination programmes in children and young adults on the tuberculosis problem. Tubercle 57, 17–43 (1976).

13. Mahler, H. The tuberculosis programme in the developing countries. Bull. Int. Union Tuberc. 37, 77–82 (1966).

14. Banerji, D. Tuberculosis: a problem of social planning in developing countries. Med. Care 3, 151–161 (1965).

15. WHO Expert Committee on Tuberculosis. Eighth Report. World Health Organization Technical Report Series No. 290. (WHO, 1964).

16. Berman, P. A. Health Sector Reform. Making health development sustainable. Health Policy 32, 13–29 (1995).

17. Miller, B. Health sector reform: scourge or salvation for TB control in developing countries? Int. J. Tuberc. Lung Dis. 4, 593–594 (2000).

18. Murray, C. J. L., Styblo, K. & Rouillon, A. Tuberculosis in developing countries: burden, intervention and cost. Bull. Int. Union Tuberc. Lung Dis. 65, 6–24 (1990).

19. Narain, J. P., Raviglione, M. C. & Kochi, A. HIV-associated tuberculosis in developing countries: epidemiology and strategies for prevention. Tubercle Lung Dis. 73, 311–321 (1992).

20. Raviglione, M. C., Snider, D. E. Jr & Kochi, A. Global epidemiology of tuberculosis. Morbidity and mortality of a worldwide epidemic. JAMA 273, 220–226 (1995).

21. Bloom, B. & Murray, C. J. L. Tuberculosis: commentary on a re-emergent killer. Science 257, 1055–1064 (1992).

22. Enarson, D. A. Principles of IUATLD collaborative tuberculosis progammes. Bull. Int. Union Tuberc. Lung Dis. 66, 195–200 (1991).

23. Kochi, A. The global tuberculosis situation and the new control strategy of the World Health Organization. Tubercle 72, 1–6 (1991).

24. World Health Organization. Forty-fourth World Health Assembly: Resolutions and Decisions. Resolution WHA 44.8. WHA44/1991/REC/1. (WHO, 1991).

25. World Health Organisation. Framework for Effective Tuberculosis Control. WHO Tuberculosis Programme.(WHO, 1994).

26. Fox, W. The problem of self-administration of drugs, with particular reference to pulmonary tuberculosis. Tubercle 39, 269–274 (1958).

27. Tuberculosis Chemotherapy Center, Madras. A concurrent comparison of home and sanatorium treatment of pulmonary tuberculosis in South India. Bull. World Health Organ. 21, 51–144 (1959).

28. Iseman, M. D., Cohn, D. L. & Sbarbaro, J. A. Directly oberved therapy: we can’t afford not to do it. N. Engl. J. Med. 328, 576–578 (1993).

29. Raviglione, M. C., Dye, C., Schmidt, S. & Kochi. A. Assessment of worldwide tuberculosis control. Lancet 350, 624–629 (1997).

30. World Health Organisation. Global Tuberculosis Control. WHO Report 2001, Document WHO/CDS/TB/2001.287. (WHO, 2001).

31. China Tuberculosis Control Collaboration. The effect of tuberculosis control in China. Lancet 364, 417–422 (2004).

32. Netto, E. M., Dye, C. & Raviglione, M.C. Progress in global tuberculosis control 1995–1996, with emphasis on 22 high-incidence countries. Int. J. Tuberc. Lung Dis. 3, 310–320 (1999).

33. Suarez, P. G. et al. The dynamics of tuberculosis in response to 10 years of intensive control effort in Peru. J. Infect. Dis. 184, 473–478 (2001).

34. World Health Organization. WHO Report 2007: Global Tuberculosis Control (Surveillance, Planning, Financing). (WHO, 2008).

R E V I E W S



FURTHER INFORMATIONUNITAID website: http://www.unitaid.eu

ALL LINKS ARE ACTIVE IN THE ONLINE PDF

35. Dye, C., Watt, C. J. & Bleed, D. Low access to a highly effective therapy: a challenge for international tuberculosis control. Bull. World Health Organ. 80, 437–444 (2002).

36. Zwarenstein, M., Schoeman, J. H., Vundule, C., Lombard, C. J. & Tatley, M. Randomised controlled trial of self-supervised and directly observed treatment of tuberculosis. Lancet 352, 1340–1343 (1998).

37. De Cock, K. M. & Chaisson, R. E. Will DOTS do it? A reappraisal of tuberculosis control in countries with high rates of HIV infection. Int. J. Tuberc. Lung Dis. 3, 457–465 (1999).

38. Volminck, J., Matchaba, P. & Garner, P. Directly observed therapy and treatment adherence. Lancet 355, 1345–1350 (2000).

39. Walley, J. D., Khan, A. N., Newell, J. N. & Khan, M. H. Effectiveness of the direct observation component of DOTS for tuberculosis: a randomised controlled trial in Pakistan. Lancet 357, 664–669 (2001).

40. Khan, M. A., Walley, J. D., Witter, S. N., Imran, A. & Safdar, N. Costs and cost-effectiveness of different DOT strategies for the treatment of tuberculosis in Pakistan. Directly observed treatment. Health Policy Plan 17, 178–186 (2002).

41. Lienhardt, C. & Ogden, J. A. Tuberculosis control in resource-poor countries: have we reached the limits of the universal paradigm? Trop. Med. Int. Hlth 9, 833– 841 (2004).

42. Pitchenik, A. E. & Fischl, M. A. Disseminated tuberculosis and the acquired immunodeficiency syndrome. Ann. Intern. Med. 98, 112 (1983).

43. Mann, J. et al. Association between HTLV-III LAV infection and tuberculosis in Zaire. JAMA 256, 346 (1986).

44. De Cock, K. M., Coulibaly, S. I. M. & Lucas, S. B. Tuberculosis and HIV infection in sub-Saharan Africa. JAMA 268, 1581–1587 (1992).

45. Getahun, H., Gunneberg, C., Granich, R. & Nunn, P. HIV infection-associated tuberculosis: the epidemiology and the response. Clin. Infect. Dis. 50, S201–S207 (2010).

46. Harries, A. et al. Tuberculosis 3. Lancet 375, 1906– 1919 (2010).

47. Godfrey-Faussett, P. et al. How human immunodeficiency virus voluntary testing can contribute to tuberculosis control. Bull. World Health Organ. 80, 939–945 (2002).

48. World Health Organization. Report of a ‘Lesson Learnt’ Workshop on the Six ProTEST Pilot Projects in Malawi, South Africa and Zambia. (WHO, 2004).

49. Getahun, H., Scano, F. & Nunn, P. in Tuberculosis, a Comprehensive Clinical Reference. (eds Schaaf, H. S. & Zumla, A.) (Saunders Elsevier, 2009).

50. World Health Organization. Interim Policy on Collaborative TB/HIV Activities. (WHO, 2004).

51. Iseman, M. D. & Madsen, L. A. TB isolates resistant to the two principal drugs, isoniazid and rifampicin, or to those and additional drugs are referred to as multidrug-resistant strains. Clin. Chest Med. 10, 341 (1989).

52. Centers for Disease Control. Nosocomial transmission of multidrug-resistant tuberculosis among HIV-infected persons— Florida and New York, 1988–1991. JAMA 266, 1483–1485 (1991).

53. Edlin, B. R. et al. An outbreak of multidrug-resistant tuberculosis among hospitalized patients with the acquired immunodeficiency syndrome. N. Engl. J. Med. 326, 1514–1521 (1992).

54. Frieden, T. et al. The emergence of drug-resistant tuberculosis in New York City. N. Engl. J. Med. 328, 521–526 (1993).

55. Pablos-Mendez, A. et al. Global surveillance for antituberculosis-drug resistance, 1994–1997. N. Eng. J. Med. 338, 1641–1649 (1998).

56. Cohn, D., Bustreo, F. & Raviglione, M. Drug-resistant tuberculosis: review of the worldwide situation and the WHO/IUATLD global surveillance project. Clin. Inf. Dis. 24, S121–S130 (1997).

57. World Health Organization. Guidelines for Establishing DOTS-Plus Pilot Projects for the Management of Multidrug-Resistant Tuberculosis (MDR-TB) (WHO, 2001).

58. Gupta, R. et al. Responding to market failure in tuberculosis control. Science 293, 1049–1051 (2001).

59. Gupta, R. et al. Increasing transparency in partnerships for health-introducing the Green Light Committee. Trop. Med. Intern. Health 7, 970–976 (2002).

60. World Health Organization. Global Tuberculosis Programme. Report of the Ad Hoc Committee on the Tuberculosis Epidemic. London, March 17–19, 1998. (WHO, 1998).

61. Kumaresan, J., Heitkamp, P., Smith, I & Billo, N. Global Partnership to Stop TB: A model of an effective public health partnership. Int. J. Tuberc. Lung Dis. 8, 120–129 (2004).

62. Kumaresan, J., Smith, I., Arnold, V. & Evans, P. The global TB drug facility. Innovative global procurement. Int. J. Tuberc. Lung Dis. 8, 130–138 (2004).

63. World Health Organization. Stop TB Initiative. Amsterdam. Declaration to Stop TB: a Call for Accelerated Action against Tuberculosis. (WHO, 2000).

64. World Health Organization. Fifty-third World Health Assembly. Resolutions and Decisions. Resolution WHA 53.1. WHA53/2000/REC/. (WHO, 2000).

65. World Health Organization. The Global Plan to Stop Tuberculosis – Stop TB Partnership. WHO/CDS/STB/2001.16. (WHO, 2001).

66. Stop TB Partnership and WHO. Global Plan to Stop TB 2006–2015. Actions for Life—Towards a World Free of Tuberculosis. WHO/HTM/STB/2006.35. (WHO, 2006).

67. Brugha, R. & Walt, G. A global health fund: a leap of faith? BMJ 323, 152–154 (2001).

68. Brugha, R. et al. The Global Fund: managing great expectations. Lancet 364, 95–100 (2004).

69. Uplekar, M., Weil, D., Raviglione, M. C. in Reichman and Hershfield’s Tuberculosis: A Comprehensive, International Approach (Informa Healthcare, 2006)

70. Dye, C. et al. Evolution of tuberculosis control and prospects for reducing tuberculosis incidence, prevalence, and deaths globally. JAMA 293, 2767– 2775 (2005).

71. World Health Organization. Macroeconomics and Health: Investing in Health for Economic Development. Report of the Commission on Macroeconomics and Health. (WHO, 2001).

72. World Health Organization. Report on the Meeting of the Second Ad-Hoc Committee on the TB Epidemic—Recommendations to Stop TB partners—Montreux, Switzerland, Sept. 2003. WHO/CDS/STB/2004.28. (WHO, 2004).

73. World Health Organization. Fifty-eighth World Health Assembly. WHA58/2005/REC/1. (WHO, 2005).

74. World Health Organization. Guidelines for the Programmatic Management of Drug-Resistant Tuberculosis. WHO/HTM/TB/2006.361. (WHO, 2006).

75. World Health Organization. Community Contribution to TB care: Practice and Policy. WHO/CDS/TB/2002.318. (WHO, 2002).

76. Achmat, Z. Science and social justice: the lessons of HIV/AIDS activism in the struggle to eradicate tuberculosis. Int. J. Tuberc. Lung Dis. 10, 1312–1317 (2006).

77. Harrington M. From HIV to tuberculosis and back again: a tale of activism in 2 pandemics. Clin. Infect. Dis. 50, S260–S266 (2010).

78. Lönnroth, K. et al. Public-private mix for implementation: what makes it work? Bull. World Health Organ. 82, 580–586 (2004).

79. Hopewell, P. C., Pai, M., Maher, D., Uplekar, M. & Raviglione, M. C. International standards for tuberculosis care. Lancet Infect. Dis. 6, 710–725 (2006).

80. World Health Organization. Practical Approach to Lung Health (PAL). A Primary Health Care Strategy for the Integrated Management of Respiratory Conditions in People Five Years of Age or Over. WHO/HTM/TB/2005.351. (WHO, 2005).

81. [No authors listed]. Tackling poverty in tuberculosis control. Lancet 366, 5 (2005).

82. Raviglione, M. C. & Uplekar, M. WHO’s new stop TB strategy. Lancet 367, 952–955 (2006).

83. Lienhardt, C., Vernon, A. & Raviglione, M. C. New drugs and new regimens for the treatment of tuberculosis: review of the drug development pipeline and implications for national programmes. Curr. Opin. Pulm. Med. 16, 186–193 (2010).

84. Ma, Z. et al. Global tuberculosis drug development pipeline: the need and the reality. Lancet 375, 2100– 2109 (2010).

85. Boehme, C. C. et al. Rapid molecular detection of tuberculosis and rifampicin resistance. N. Engl. J. Med. 363, 1005–1015 (2010).

86. Wallis, R. S. et al. Biomarkers and diagnostics for tuberculosis: progress, needs, and translation into practice. Lancet 375, 1920–1937 (2010).

87. Kaufmann, S. H. E., Hussey, G. & Lambert, P.-H. New vaccines for tuberculosis. Lancet 375, 2110–2119 (2010).

88. Stop TB Partnership and WHO. Global Plan to Stop TB, 2011-2015: Transforming the Fight Towards Elimination of Tuberculosis. (WHO, 2010).

89. Treatment Action Group. 2011 Report on Tuberculosis Research Funding Trends, 2005–2010. (Treatment Action Group, 2012).

90. Glaziou, P. et al. Lives saved by tuberculosis control and prospects for achieving the 2015 global target for reducing tuberculosis mortality. Bull. World Health Organ. 89, 573–582 (2011).

91. World Health Organization. Global tuberculosis control: WHO Report 2011. (WHO, 2011).

92. Lönnroth, K. et al. Tuberculosis control and elimination 2010-50: cure, care, and social development. Lancet 375, 1814–1829 (2010).

93. Atun, R., Weil, D. E., Eang, M. T. & Mwakyusa, D. Health-system strengthening and tuberculosis control. Lancet 375, 2169–2178 (2010).

94. Lienhardt, C., Espinal, M., Maher, D., Pai, M. & Raviglione, M. What research is needed to stop TB? Introducing the TB Research Movement. PLoS Med. 8, e1001135 (2011).

95. Abu-Raddad, L. J. et al. Epidemiological benefits of more effective tuberculosis vaccines, drugs and diagnostics. Proc. Natl Acad. Sci. USA 106, 13980– 13985 (2009).

96. Stop TB Partnership and WHO. The International Roadmap for Tuberculosis Research - Towards a World Free of TB. (WHO, 2011).

97. Lienhardt, C. Why is research critical to stop tuberculosis? PLoS Med. community blog [online], http://blogs.plos.org/speakingofmedicine/2012/01/26/why-is-research-critical-to-stop-tuberculosis (2012).

98. Lin, H., Ezzati, M. & Murray, M. Tobacco smoke, indoor air pollution and tuberculosis: a systematic review and meta-analysis. PLoS Med. 4, e142 (2007).

99. Jeon, C. Y. & Murray, M. B. Diabetes mellitus increases the risk of active tuberculosis: a systematic review of 13 observational studies. PLoS Med. 5, e152 (2008).

100. Lönnroth, K., Williams, B. G., Stadlin, S., Jaramillo, E. & Dye, C. Alcohol use as a risk factor for tuberculosis—a systematic review. BMC Public Health 8, 289 (2008).

101. Lönnroth, K., Jaramillo, E., Williams, B. G., Dye, C. & Raviglione, M. Drivers of tuberculosis epidemics: the role of risk factors and social determinants. Soc. Sci. Med. 68, 2240–2246 (2009).

102. Getahun, H. & Raviglione, M. Transforming the global tuberculosis response through effective engagement of civil society organizations: the role of the World Health Organization. Bull. World Health Organ. 89, 616–618 (2011).

AcknowledgementsThe authors thank Katherine Floyd, Diana Weil and Paul Nunn for their contribution to many of the elements described in this Review.

R E V I E W S



http://www.unitaid.eu

nature outlook tuberculosis

Documents

tuberculosis control

tuberculosis spreads

new antitb drug

new point

mycobacterium tuberculosis

new technologies

promising vaccine

bcg vaccine