E D I T O R I A L C O M M E N T A R Y

Nonsteroidal Antiinflammatory Drugs for AdjunctiveTuberculosis Treatment

Juraj Ivanyi1 and Alimuddin Zumla2

1Department of Clinical and Diagnostic Sciences, Kings College London at Guy’s Hospital Campus, and 2Centre for Clinical Microbiology, Department of Infection, Division ofInfection and Immunity, University College London, United Kingdom

(See the brief report by Vilaplana et al on pages 199–202.)

Keywords. Tuberculosis; adjunctive treatment; non-steroidal anti-inflammatory drugs; Ibuprofen; prostaglandin inhibitors.

Tuberculosis continues to cause 1.4 milliondeaths annually despite effective treat-ment being available since the 1970s. Thestandard 4-drug tuberculosis treatmentregimen involving isoniazid, rifampicin,pyrazinamide, and ethambutol achieves90%cure rates in programmatic settings [1].Mycobacterium tuberculosis bacilli aredifficult to eradicate since they exist in aspectrum of replication states, from met-abolically active, “rapid” replicators tonearly “dormant” nonreplicating persist-ers. Thus, treatment is required for a du-ration of at least 6 months in 2 phases: a2-month intensive phase involving all 4drugs and a 4-month continuation phaseinvolving rifampicin and isoniazid. Majorchallenges for tuberculosis control pro-grams include poor compliance, resultingin the emergence of multidrug-resistantM. tuberculosis strains, which require theuse of more toxic second-line drugs for atleast 24 months [1]; problems also arise

from failures in drug supply and fromdrug intolerance. Shortening of the tuber-culosis treatment period for both drug-susceptible and drug-resistant tuberculosiswould greatly improve tuberculosis man-agement and infection control.Immunotherapy has been considered

as a possible approach toward shorteningthe duration of chemotherapy, with awide range of cytokines or their inhibi-tors and chemical or biological immuno-modulatory compounds being explored[2]. However, the factors that determinethe replication rate ofM. tuberculosis andsynergize or antagonize the effectivenessof mycobactericidal drugs remain poorlyunderstood. Efforts at developing newdrugs that act on replicating populationsof M. tuberculosis, which thrive underthe influence of nitric oxide and hypoxia,identified rhodanines [3]. Postchemo-therapy relapse was shown to requirecyclosporine-sensitive and genetically in-fluenced host immunity, which developsearly after infection [4]. Such a relapsecan be alleviated by the immunomodula-tory agent muramyl dipeptide [5] or bypassive antibody, combined with interfer-on γ (IFN-γ) and anti-interleukin 2 treat-ment [6]. Recently, an antiinflammatorydrug, oxyphenbutazone, was found to becidal against both replicating and dor-mant forms ofM. tuberculosis [7].Adjunct treatment for tuberculosis in

experimentalmousemodels, using generic,

nonsteroidal antiinflammatory and anal-gesic drugs (NSAIDs), showed that diclo-fenac (2-[2,6-dichloranilino] phenylaceticacid) inhibited colony-forming units inspleen and liver [8]. Further synergisticactivity was observed in combinationwith streptomycin. Aspirin and ibupro-fen (iso-butyl-propanoic-phenolic acid)were reported to enhance the mycobac-tericidal effect of pyrazinamide [9]. Inhi-bition of the synthesis of PGE2, which iselevated in the lungs at a late phase oftuberculosis, has been shown to reduceinfection [10]. This finding is plausible,considering that PGE2 inhibits phagocy-tosis, bacterial killing, production ofnitrite, and T-helper 1 cytokines. Theseimmunomodulatory functions are exertedvia 4 EP receptors [11]. In this issue ofthe Journal, Vilaplana et al report thatibuprofen, in clinical use for severaldecades, can alleviate the lung pathologyin a special mouse model, in which thelesions are similar to those caused by tu-berculosis in humans. There are severalramifications of this finding relating tothe cellular targeting mechanisms andfor clinical translation into improving tu-berculosis treatment in humans.

NSAIDs are the most commonly usedanalgesic and antiinflammatory drugsworldwide. The therapeutic effects ofNSAIDs are primarily due to their abilityto inhibit COX-1 and COX-2 cyclooxy-genases, which convert arachidonic acid

Received and accepted 13 February 2013; electronicallypublished 5 April 2013.

Correspondence: Juraj Ivanyi, MD, PhD, Clinical and Diag-nostic Sciences Department, Hodgkin Bldg, 2nd Fl, Guy’sCampus of Kings College London, London SE1 1UL UnitedKingdom ( juraj.ivanyi@kcl.ac.uk).

The Journal of Infectious Diseases 2013;208:185–8© The Author 2013. Published by Oxford University Presson behalf of the Infectious Diseases Society of America.All rights reserved. For Permissions, please e-mail: journals.permissions@oup.com.DOI: 10.1093/infdis/jit153

EDITORIAL COMMENTARY • JID 2013:208 (15 July) • 185

at National D

ong Hw

a University on M

arch 26, 2014http://jid.oxfordjournals.org/

Dow

nloaded from

to prostaglandin H2 and prostacyclin(which lead to release of mediators ofpain, inflammation, and fever) and tothromboxane A2 (which stimulates plate-let aggregation and vasoconstriction).Ibuprofen is listed as a core medicine bythe World Health Organization [12] andis a widely used analgesic and anti-inflammatory drug that inhibits bothCOX-1 and COX-2 cyclooxygenases.COX-1 inhibition is responsible for un-wanted side effects, such as gastrointesti-nal ulceration and bleeding. However,even selective COX-2–inhibitory NSAIDscarry a cardiovascular risk, leading to hy-pertension and myocardial infarction dueto COX-2 inhibition in vascular endothe-lial and smooth muscle cells [13].

The results obtained by Vilaplana et alare highly relevant for further consider-ation of evaluation of the use of ibuprofenin humans for the adjunctive treatment ofdrug-susceptible and drug-resistant tu-berculosis. The rationale for use of ad-junctive antiinflammatory therapy fortuberculosis is to alleviate the excessiveand harmful host inflammatory respons-es that lead to pathological lung lesions.Induced by antigenic and immunomodu-latory glycolipid stimuli, these “decoys”[14] trigger host reactions, which are pro-tective in the majority of infected individ-uals but become pathogenic in a minorityof infected subjects. The lungs are theroute of entry of M. tuberculosis infectionand permit the pathogen to evade protec-tive systemic host defense, persist in alatent state, and form pathological lesions,which are essential for efficient trans-mission ofM. tuberculosis from the granu-loma lung lesion into the alveoli andaerosols. Antiinflammatory therapy is,thus, targeted toward lung granulomatouslesions generated by an influx of a rangeof immune cells such as monocytes, lym-phocytes, and neutrophils (PMNs) [15].All these cell types produce prostaglandinsand, therefore, can be targets for the actionof COX-1 and COX-2 cyclooxygenase in-hibitory drugs.

Considering the pilot nature of the re-sults involving C3HeB/FeJ mice reported

by Vilaplana et al, the authors restrainedfrom eluding on the question of the pos-sible target cell for ibuprofen therapy.PMNs need to be considered, bearing inmind that their function is probably de-fensive early after M. tuberculosis infec-tion but aggravating in advanced disease[16]. Therefore, the timing of ibuprofentherapy would need to aim at the latestage and avoid interfering with the earlystage of infection. The protective func-tion has been associated with PMNgranule–derived peptide and cationicprotein defensins, which can be cidal forboth intracellular and extracellularM. tu-berculosis [17], or with the activation ofmacrophages. On the other hand, thepathogenic function of PMNs has recent-ly been attributed to the IFN-inducibleprogrammed death 1 ligand, with thecaveat that either its excess or absencecould be pathogenic [16]. The perceivedneed for “balancing” indicates how diffi-cult it may be to identify a beneficialtherapeutic regimen. While the influx ofPMNs to lesions is due to attraction tointerleukin 17, produced by γδT cells orT-helper 17 cells, to interleukin 8, and tothe MIG chemokine, the mechanisms re-sulting in PMN depletion by ibuprofenmight involve their apoptotic death andclearance by macrophages. Thus, ibupro-fen may act by either reducing the influxof PMNs or by enhancing their clear-ance.Inhibition of PGE2 synthesis by ibu-

profen may be favorable for host resis-tance, considering that PGE2 is actingagainst the production of interleukin 1,TNF-α, and reactive oxygen and nitrogenintermediates by macrophages. It also in-hibits interleukin 12 expression by den-dritic cells and expression of IFN-γ andinterleukin 2, while promoting the pro-duction of interleukin 10 and interleukin4 by lymphocytes. On the other hand,COX-mediated inhibition of PGE2 syn-thesis in macrophages may be unfavor-able for the host in the early phase afterM. tuberculosis infection, since PGE2protects the mitochondrial membraneagainst damage from M. tuberculosis,

prevents necrosis, and promotes the apo-ptosis of infected macrophages, whichleads to both innate and adaptive immu-nity [18]. In fact, virulent but not the at-tenuated M. tuberculosis evades the hostresponse by inducing the production oflipoxin A4, which inhibits cyclooxygen-ase 2 production and PGE2 biosynthesis.However, the picture seems more complex,considering that the ESAT-6 secretoryantigen ofM. tuberculosis stimulates PGE2production, perhaps with the pathogen’sstrategy to permit initial but impede laterToll-like receptor–mediated signaling[19]. Finally, the possible action of ibu-profen on alveolar epithelial cells, whichalso make COX cyclooxygenases, alsoneeds to be considered.

Ibuprofen could also influence the pro-duction of leukotriene B4 (LTB4), whichis an icosanoid whose levels are elevatedby M. tuberculosis infection, but with op-posite effects than PGE2 [20]. Blockadeof COX-2 by administration of the cele-coxib NSAID enhanced metabolism viathe 5-LO pathway, indicating that theprotective effect against M. tuberculosismay be due to increased leukotriene pro-duction by alveolar macrophages andthat the immunostimulant effects of LTB4dominate over the immunosuppressiveactions of PGE2 [10].

The key open question remains ofwhether ibuprofen could be beneficial asan adjunctive treatment when combinedwith tuberculosis chemotherapy. Themouse experimental models offer thebest opportunity for further preclinicalresearch. However, suitable conditionswill need to be chosen for several aspects,such as the route and dose of infection,the time and schedule of therapy, and theend point criteria, such as lung pathol-ogy, bacterial load, weight loss, or sur-vival. It is commendable that mice withhuman-like lung pathology features, suchas the C3HeB/FeJ strain (used by Vilapla-na et al) or the I/St strain [21], developcaseating granulomas, which developinto liquefacted lesions, with a massivecellular invasion of alveolar spaces mainlyby neutrophils. In common for humans

186 • JID 2013:208 (15 July) • EDITORIAL COMMENTARY

at National D

ong Hw

a University on M

arch 26, 2014http://jid.oxfordjournals.org/

Dow

nloaded from

and mice, tubercle bacilli in such lesionstend to multiply extracellularly andupregulate the expression of hypoxia-asso-ciated genes. Significantly, mice with theselesions are prone to postchemotherapyrelapse to a greater extent than conven-tional inbred mice [21]. Use of low-doseaerosol infection that leads to late pulmo-nary pathology might be most suitablefor finding out whether ibuprofen treat-ment, as an adjunct to chemotherapy,could reduce the relapse rate.

Trials on the use of ibuprofen as anadjunct to tuberculosis chemotherapy inhumans are warranted in clinical situa-tions with pronounced inflammatorypathogenesis. This is strongly supportedby the already established routine use ofcorticosteroids with standard tuberculo-sis drug treatment for tuberculosis men-ingitis, brain tuberculomas, and severecases of military tuberculosis, resulting inreduced mortality from tuberculosismeningitis [22]. Aspirin has been provento be of use in preventing stroke and3-month mortality in patients with tu-berculosis meningitis [23]. A range ofpharmacological agents with immuno-suppressive and immunomodulatory ac-tivity have been used for tuberculosis-related paradoxical immune reconstitutioninflammatory syndrome (TB-IRIS) afterinitiation of antiretroviral therapy inhuman immunodeficiency virus (HIV)–infected patients with tuberculosis. Adjuncttherapy with steroids reduces morbidityassociated with moderately severe IRIS[24]. The beneficial effects of prednisonein TB-IRIS appear to be mediated viasuppression of predominantly proinflamma-tory cytokine responses of innate immuneorigin, rather than via a reduction of thenumbers of antigen-specific T cells in pe-ripheral blood. NSAIDs may be as effec-tive as corticosteroids for treatment ofnon–life-threatening TB-IRIS in HIV-infected patients.

Many patients who need to move onto second-line therapy for drug-resistanttuberculosis do not receive the treatmentthey need because the tuberculosis drugsare too expensive. Ibuprofen and other

NSAIDs are comparatively very cheap.Although ibuprofen does have some po-tential side effects, these are likely to beacceptable when compared with the tox-icity and side effects of tuberculosisdrugs. Because ibuprofen has compara-tively little risk and good safety profile,the temptation is to immediately evaluatethe use of ibuprofen as an adjunct to tu-berculosis treatment in humans in phase3 clinical trials, although other evalua-tions may be required. Other COX-2 in-hibitors with fewer side effects thanibuprofen could be more suitable for ad-junctive tuberculosis therapy. Thus, rofe-coxib and celecoxib, which preferentiallyinhibit COX-2, resulting in better tolera-bility [25], should be evaluated in parallelwith ibuprofen. Ibuprofen lysine saltwith increased water solubility allows in-travenous use with more rapid onset ofaction and is suitable for topical use.Could ibuprofen-lysine be of interest foraerosol application in tuberculosis? Ofthe other considerations, the timing ofibuprofen delivery could be critical toimproving treatment outcomes: if it isused to target PMN infiltration, than itshould be given late, when the inflamma-tory pathology is excessive, not early,when PMNs might be protective.Case reports documented reactivation

of pulmonary tuberculosis in 2 patientswho had used NSAIDs [26]. A case-control study of 38 patients designed totest the hypothesis that such an associa-tion exists found a statistically significantrelation between the reactivation of latentM. tuberculosis infection and the use ofNSAIDs [27]. Whether the association isdirect, indirect, or secondary remainsunknown, but it may involve mecha-nisms by which hydrocortisone has beenshown to reactivate dormant tuberculosisin Cornell model mice. The studieswould need to take into account the in-fluence of human population genetics,the influence of commensal environmen-tal microbiota, and the compound life-time history of immunological primingdue to vaccinations and exposure to in-fectious diseases. Since all patients with

active tuberculosis will be treated by stan-dard chemotherapy, any consideration ofclinical trials that evaluate adjunctivetreatment with ibuprofen will need to bejudged as to whether ibuprofen canshorten the period required to achieverelapse-free cure.

Note

Potential conflicts of interest. All authors:No reported conflicts.All authors have submitted the ICMJE Form

for Disclosure of Potential Conflicts of Interest.Conflicts that the editors consider relevant to thecontent of the manuscript have been disclosed.

References

1. Zumla A, Raviglione M, Hafner R, von ReynC. Current concepts—tuberculosis. NewEngl J Med 2013; 368:745–55.

2. Tomioka H. Adjunctive immunotherapy ofmycobacterial infections. Curr Pharm Des2004; 10:3297–312.

3. Bryk R, Gold B, Venugopal A, et al. Selectivekilling of nonreplicating mycobacteria. CellHost Microbe 2008; 3:137–45.

4. Cox JH, Ivanyi J. The role of host factors forthe chemotherapy of BCG infection in inbredstrains of mice. APMIS 1988; 96:927–32.

5. Venkataprasad N, Ledger P, Ivanyi J. Theeffect of glucosaminylmuramyl dipeptide in-jection to mice on the course of tuberculousinfection and in vitro superoxide anion pro-duction. Int Arch Allergy Immunol 1997;114:23–9.

6. Buccheri S, Reljic R, Caccamo N, et al. Pre-vention of the post-chemotherapy relapse oftuberculous infection by combined immuno-therapy. Tuberculosis (Edinb) 2009; 89:91–4.

7. Gold B, Pingle M, Brickner SJ, et al. Nonste-roidal anti-inflammatory drug sensitizesMy-cobacterium tuberculosis to endogenous andexogenous antimicrobials. Proc Natl AcadSci U S A 2012; 109:16004–11.

8. Dutta NK, Mazumdar K, Dastidar SG, ParkJH. Activity of diclofenac used alone and incombination with streptomycin against My-cobacterium tuberculosis in mice. Int J Anti-microb Agents 2007; 30:336–40.

9. Byrne ST, Denkin SM, Zhang Y. Aspirin andibuprofen enhance pyrazinamide treatmentof murine tuberculosis. J Antimicrob Che-mother 2007; 59:313–6.

10. Rangel Moreno J, Estrada Garcia I, et al.The role of prostaglandin E2 in the immuno-pathogenesis of experimental pulmonary tu-berculosis. Immunology 2002; 106:257–66.

11. Kaul V, Bhattacharya D, Singh Y, et al. Animportant role of prostanoid receptor EP2 inhost resistance to Mycobacterium tuberculo-sis infection in mice. J Infect Dis 2012;206:1816–25.

EDITORIAL COMMENTARY • JID 2013:208 (15 July) • 187

at National D

ong Hw

a University on M

arch 26, 2014http://jid.oxfordjournals.org/

Dow

nloaded from

12. World Health Organization. Model list of es-sential medicines. 3rd list. Geneva: WorldHealth Organization, 2011. http://www.who.int/medicines/publications/essentialmedicines/en/. Accessed 19 April 2013.

13. Yu Y, Ricciotti E, Scalia R, et al. VascularCOX-2 modulates blood pressure and throm-bosis in mice. Sci Transl Med 2012; 4:132ra54.

14. Ivanyi J. Pathogenic and protective interac-tions in mycobacterial infections. ClinImmunol and Allerg 1986; 6:127–57.

15. Dheda K, Booth H, Huggett JF, Johnson MA,Zumla A, Rook GA. Lung remodeling in pul-monary tuberculosis. J Infect Dis 2005;192:1201–9.

16. Lowe DM, Redford PS, Wilkinson RJ,O’Garra A, Martineau AR. Neutrophils intuberculosis: friend or foe? Trends Immunol2012; 33:14–25.

17. Tan BH, Meinken C, Bastian M, et al. Macro-phages acquire neutrophil granules for anti-microbial activity against intracellularpathogens. J Immunol 2006; 177:1864–71.

18. Divangahi M, Chen M, Gan H, et al. Myco-bacterium tuberculosis evades macrophagedefenses by inhibiting plasma membranerepair. Nat Immunol 2009; 10:899–906.

19. A SK, Bansal K, Holla S, Verma-Kumar S,Sharma P, Balaji KN. ESAT-6 induced COX-2 expression involves coordinated interplaybetween PI3K and MAPK signaling. MolImmunol 2012; 49:655–63.

20. Peres-Buzalaf C, de Paula L, Frantz FG, et al.Control of experimental pulmonary tubercu-losis depends more on immunostimulatoryleukotrienes than on the absence of immu-nosuppressive prostaglandins. Prostaglan-dins Leukot Essent Fatty Acids 2011; 85:75–81.

21. Radaeva TV, Nikonenko BV, Mischenko VV,Averbakh MM Jr, Apt AS. Direct comparisonof low-dose and Cornell-like models ofchronic and reactivation tuberculosis ingenetically susceptible I/St and resistantB6 mice. Tuberculosis (Edinb) 2005; 85:65–72.

22. Marx GE, Chan ED. Tuberculous meningitis:diagnosis and treatment overview. TubercRes Treat 2011; 2011:798764.

23. Misra UK, Kalita J, Nair PP. Role of aspirinin tuberculous meningitis: a randomizedopen label placebo controlled trial. J NeurolSci 2010; 293:12–7.

24. Lawn SD, Meintjes G. Pathogenesis and pre-vention of immune reconstitution diseaseduring antiretroviral therapy. Expert RevAnti Infect Ther 2011; 9:415–30.

25. Weberschock TB, Muller SM, Boehncke S,Boehncke WH. Tolerance to coxibs in pa-tients with intolerance to non-steroidal anti-inflammatory drugs (NSAIDs): a systematicstructured review of the literature. Arch Der-matol Res 2007; 299:169–75.

26. Brennan M. Activation of latent tuberculosisby nonsteroidal anti-inflammatory agents.Can Med Assoc J 1980; 122:400–2.

27. Tomasson HO, Brennan M, Bass MJ. Tu-berculosis and nonsteroidal anti-inflammato-ry drugs. Can Med Assoc J 1984; 130:275–8.

188 • JID 2013:208 (15 July) • EDITORIAL COMMENTARY

at National D

ong Hw

a University on M

arch 26, 2014http://jid.oxfordjournals.org/

Dow

nloaded from