pharmacovigilance spontaneous reporting
DESCRIPTION
13TRANSCRIPT
Introduction to
Pharmacovigilance
Mondira Bhattacharya, MD
Senior Medical Director, Global Pharmacovigilance
Abbott Laboratories
Disclaimer
• The views and opinions expressed in this
presentation are solely those of the presenter and do not necessarily reflect
those of Abbott Laboratories or any of its affiliated organizations.
Course Overview: Focus on Postmarketing Safety
• History of U.S. Pharmacovigilance (the regulations!)
• Definitions: Pharmacovigilance, ICSR, MedDRA, Signal, Risk
• General PV Processes: Less data driven; greater emphasis on regulatory submissions
– Spontaneous reporting: ICSRs and Aggregate Reports
• Signal Detection and Evaluation: Systematic; Data Driven
– Qualitative – cases/cases series
– Quantitative – disproportionality algorithms
• Risk Management
• Benefit Risk Assessment and its Implications: ongoing evaluation of a drug’s profile
Discipline that employs scientific methodologies to detect, assess and understand adverse events associated with use of drugs/biologicals/vaccines
• Postmarketing activities• Utilizes data gathering from multiple sources: clinical
trials, spontaneous reports, literature, pharmacoepidemiologic studies, registries
• Major task is to identify new safety signals and to better understand the patterns of recognized signals (risk factors, severity, outcome)
PV activities contribute to the understanding of the risk/benefit profile of the drug and to the communication of changes identified to the existing profile
Definition of Pharmacovigilance
History of Drug Safety Regulations1906 Pure Food and Drug Act: Tetanus: contaminated diptheria antitoxin leading to deaths in children
• Required licensing of manufacturers and distributers
• Prevented mislabeling of drugs and adulteration of drugs
• NO requirement for proof of safety/efficacy.
Next major disaster: Elixir of sulfanilamide used to treat gonorrhea and pneumococcal infections was newly introduced in a more flavorful dosage
using a solvent which was the antifreeze diethylene glycol.
> 100 deaths
Only charge against the company was that the “elixir” contained no alcohol.
1938 Food, Drug and Cosmetic Act: requirement for collection of clinical
safety data on and submission of such data to the FDA prior to approval. No postmarketing safety reporting requirements.
Thalidomide Storm
• Thalidomide (sedative/antiemetic) disaster: malformed limbs along with other birth defects reported in Lancet in 1961
• Drug was never approved in the US for morning sickness because reports of fetal malformations were quickly picked up by the press and communicated worldwide
Thalidomide Storm: Outcome
Kefauver-Harris Amendment (1962):
A result of the thalidomide disaster (phocomelia in newborns). Extensive pre-clinical testing, submission of such data to FDA (IND) before clinical testing could begin. Three explicit phases of clinical testing were defined. Proof of efficacy was required.
Also mandated that pharmaceutical manufacturers having an NDA must report AEs to the FDA.
History of U.S. Regulations
• Historical Perspective
- 1967: computerization of AE reports
– 1985 NDA Rewrite : requirement for submission of spontaneous reports; prior to that these reports got filed away as part of IND submissions.
– 1993 MedWatch – devices, vaccines
– 1997 FDA Modernization Act: AERS database
– 2003 Concept Papers on Risk Management
– 2007 FDAAA
Adverse Event
• Any untoward medical occurrence in a
patient…administered a pharmaceutical
product and which does not necessarily
have a causal relationship with this
treatment.
• Can be any sign, symptom, or abnormal
lab result temporally associated with a
treatment.
• Unsolicited reports from health care professionals or consumers
• Includes adverse experiences, medication errors and product quality complaints sent directly to FDA
through the MedWatch program or via a manufacturer.
• Postmarketing surveillance: compilation of these spontaneous reports + data from formal clinical trials + literature reports + pharmacoepidemiological
studies = postmarketing surveillance data
Spontaneous Reporting: Sources of Data
• Identify new (previously unrecognized) ADRs
• Identify risk factors/at-risk populations for known ADRs
• Identify change in reporting of an AE/ADR over time
• Identify manufacturing problems
• Identify medication errors (wrong dose, wrong
patient, drug interactions, use in the face of clear contraindications) are common: estimated to contribute to 44,000-98,000 annual deaths in the
U.S. and are associated with 17-29 billion dollars of excess expenditure.
Uses of Spontaneous
Data
Why do we need spontaneous reporting after drug
approval? Why is Clinical Safety Data Inadequate
for Identifying all Risks?
• Very restricted patient population(inclusion/exclusion criteria; concomitant medications restricted)
• Too small (few thousand)
• Duration of exposure is usually less than 1 year
• Often not designed to identify specific change in incidence of outcomes of the underlying disease
• Errors associated with medical prescribing not captured
• Public failures of our system:
– 20% of new drugs get Black-box warnings
– 4% of new drugs removed from market due to safety reasons (industry inefficiency)
Greatest Challenge for Spontaneous
Reports: Under reporting
HIPPA Regulations have not helped!
Report Quality: Inclusion of details to establish causality: diagnostic test results?, progression of disease?, effect of concomitant medications?, etc.
Biases Affecting Spontaneous Reporting
• Nature of the event
– Severity
– Uniqueness
– Co-morbidity/outcome of disease being treated
• Length of time drug is on the market: first in class vs. later
– Weber effect: peak is 1-2 years after approval; not shown to be true for all classes of drugs
• Publicity/media attention
• Litigation
• Reporting regulations
Triggers for Increased Reporting
0
50
100
150
200
250
300
Jul-1
999
Sep
-199
9N
ov-
199
9Ja
n-2
000
Mar
-200
0M
ay-2
000
Jul-2
000
Sep
-200
0N
ov-
200
0Ja
n-2
001
Mar
-200
1M
ay-2
001
Jul-2
001
Sep
-200
1N
ov-
200
1Ja
n-2
002
Mar
-200
2M
ay-2
002
Jul-2
002
Sep
-200
2N
ov-
200
2Ja
n-2
003
Mar
-200
3M
ay-2
003
# o
f C
ase
s (In
itia
l o
r F
ollo
w-u
p)
0
100
200
300
400
500
600
700
800
900
Pre
scri
pti
on
s
(Th
ou
sa
nd
s)
Initial Versions Follow-up Versions Prescriptions
Another drug
in class
removed
from market
for safety
reasons
←
No denominator available!
Next step - Coding of Events: Medical Dictionary for Regulatory Activities (MedDRA)
• Each event within a report (ICSR – individual case safety report)
• To aggregate reported terms in medically meaningful groupings for the purpose of reviewing and/or analyzing safety data
• To facilitate consistent retrieval of specific cases or medical conditions from a database
• To identify frequency of medically similar ADR/AEs
• To improve consistency in comparing safety data from multiple sources (clinical trials and spontaneous reports)
Features that may Suggest a Causal Relationship:
• Absence of symptoms/condition prior to exposure to drug
• Timing of event in relationship to drug administration suggestive
• Evidence of a positive dechallenge and/or positive rechallenge
• No mitigating factors identified
• Consistency with the drug’s known pharmacological and/or toxicological effects
• Supporting evidence from other sources (clinical studies, nonclinical studies, reports for drugs of the same class)
• Event typically associated with a drug effect (i.e. Steven Johnson Syndrome)
Evaluation of Causality of a Single Case Report: Medical
Assessment and Query = Adverse Drug Reaction (ADR)
MEDICALLY SIGNIFICANT LIST – Examples of
events that require immediate follow-up
Acute respiratory failure/acute
Acute respiratory distress syndrome
Ventricular fibrillation
Torsade de pointes
Malignant hypertension
Convulsive seizures
Agranulocytosis
Aplastic anemia
Toxic epidermal necrolysis
Stevens-Johnson syndrome
Hepatic necrosis
Acute liver failure
Anaphylactic shock
Acute renal failure
Pulmonary hypertension Pulmonary fibrosis
Confirmed or suspected endotoxin shock
Confirmed or suspected transmission of infectious agent
by products
Aggregate Reports: Periodic vs. PSUR
Periodic Report (PADER) PSUR (a)
Covers FDA USA ICH International
Events No required events; analysis of expedited events; non-expedited and fatal reports
Required events (Overdose, Drug Interaction); selected events; analysis of Serious unlisted events
Foreign Reports Non-included / Optional Included
Period Quarterly for 3 yrs. Annual after 3 yrs. Every 6 mos (2yrs), annually (2yrs),
every 3yrs
Total Exposure Not included Included
Studies Targeted safety studies or studies yielding important safety data that were initiated during reporting period
Targeted safety studies or studies yielding significant safety data; tracked until final report
Regulatory Actions taken for safety reasons globally
Actions taken for safety reasons globally
Label Changes USPI Changes during the reporting period
CCDS changes during the reporting period
(a) Only medically confirmed cases are used
CLASSIC CLASSIC
PHARMACOVIGILANCEPHARMACOVIGILANCE
Adverse Drug ReactionAdverse Drug Reaction
No
No
Review Of Data From All SourceReview Of Data From All Source
Aggregate and Present In Aggregate and Present In
Periodic ReportsPeriodic Reports
ChangeChange
RegulatorsRegulators
YesYes
EXPED
ITED
Bulk of effort was to better characterize
what was known and heavily reliant on
company’s own database. Not a very
effective means for identifying signals!
What is a Signal?
Reported information on a possible causal
relationship between an adverse event
and a drug, of which the relationship is
unknown or incompletely documented
previously and which is of enough
interest to justify verification action.
Source: Trinks, Uwe. DIA Conference, Chicago, IL, 2011
How do you find Signals: many routine activities within PV
• Single case of concern (ie:SJS)
• Trial results revealing an imbalance in ADRs
between treatment arms
• Data mining:
– Methodology to identify hidden patterns of association
between use of a drug and an unexpected adverse reaction
– Computer algorithms to analyze records contained in huge drug safety databases.
– Estimated 1000 reports per day are added to FDA AERS database
Data Mining
• It is a prospective, timely methodology for analyzing these reports to ascertain whether there is an excess
of events of concern with use of a drug
• Is now considered an essential adjunct to other PV
activities and needs to be integrated into existing PV processes.
• Databases:
– US FDA-AERS:available through FOI Act (5 month lag)
– WHO Vigibase: licensing agreement
– EMEA Eudra Vigilance:products on EU market; only access to data on their own products
– In house corporate database
Data Mining
• Examination of reported AEs using statistical or mathematical tools to assess for an excess of specific AEsfor a product
• Output is generally a score that quantifies the disproportionality between the observed and expected values for a given product-event combination
• Comparison: assumption for these scores is that the proportion of specific events is the same across all drugs
– This comparison can be modified as appropriate (for example, for an anti-diabetic drug, the comparison may be made only to other anti-diabetic drugs rather than all drugs)
• A threshold is arbitrarily set, with scores above this thresholddeemed in excess of expected and meriting further investigation
Datamining Method
Disproportionality Analysis
Drug of Interest All Other Drugs Total
AE of interest A C A + C
All other AEs B D B + D
Total A + B C + D A + B + C + D
Comparison of relative reporting frequencies
Example:
Proportional Reporting Ratio (PRR) =
AA + B
CC + D
> 9,000 event codes (MedDRA Preferred Terms)> 7,000 drug/biological products by trade names > 60,000,000 drug-event combinations possible!
Data Mining
• Scores are determined using statistical
models
– Multi-item Gamma Poisson Shrinker (MGPS) algorithm
– Proportional Reporting Ratio (PRR) method
– Neural Network Approach
Results of Data Mining
• EBGM: Empirical Bayes Geometric Mean that can be understood as an average ratio of expected to observed events
• EB05: The lower limit of the 95% CI on the EBGM
– Example: if EB05=20, then the drug-event occurred AT LEAST 20 times more frequently in AERS than expected
• Data Mining Signal (“Threshold Interval”) EB05 >=2:
– The drug-event occurred AT LEAST twice as often as expected. This threshold gives assurance that potential signals are unlikely to be noise
Interpretation of Data Mining Results
• Advantages: large, reproducible, comparable – however only hypothesis generating
• Areas of uncertainty:
False associations:
- Drug constantly used with other drugs that do cause the event
- Outcome/event more related to natural history of disease than drug
Missed associations:
- Later in drug class → under reporting
- Event linked to the disease → under reporting
- Event occurs years after initiation of treatment (cancers, heart disease) → under reporting
• No confirmation of a causal relationship between a drug and an AE can be
established through this exercise.
•Case series
– Review of additional cases; if possible, use standardized case definitions (formal criteria for including or excluding a case in the series)
• Literature
• Epidemiologic studies: natural history; safety studies
• Clinical Trial Data
Signal Evaluation:
Confirm/Refute/Understand a Signal
CURRENT CURRENT
PHARMACOVIGILANCEPHARMACOVIGILANCE
SignalSignal
No
Assess SignalAssess Signal
Fully Characterize = Fully Characterize =
Risk ManagementRisk Management
Benefit/Benefit/
Risk AssessmentRisk Assessment
ClinicalClinical
DataData
RegulatoryRegulatory
ReportingReporting
Label & Other Label & Other
ChangesChanges
ADRsADRsDataData
MiningMining
Other Other
SourcesSources
Yes
PeriodicPeriodic
ReportsReports
Altered
Risk Management: The Patient’s
Perspective – Stop the Drug!
“Every substance is a poison; it is only a matter of dose” –
Paracelsus (1493 – 1541)
…and a matter of risk management!
Risk Benefit Decisions
Risk Management View
Maximum Maximum acceptableacceptable
risk for all risk for all
patientspatients
RiskRisk
AcceptableAcceptable
risk for risk for
some some
patients or patients or
physiciansphysicians
BenefitBenefit
Acceptable Acceptable benefit for benefit for
some patientssome patients
or physiciansor physicians
Minimum Minimum
acceptable acceptable
efficacy for efficacy for
all patientsall patients
Disapprove
Risk Management (RM) Zone
Approve
Source: Michael J. Klepper, MD (2006)
• Identify the significant risks and potential risks of the product
• Quantify the risks; identify populations for which risk is not known or quantified
• Identify the subpopulations who face the important risk
• Evaluate early detection and prevention potentials
• Strategic safety program designed to meet specific goals and objectives in minimizing known risks of a product while preserving its benefits.
• Proposed routine PV or active interventions (will include a PPI)but generally involves other activities (e.g.: change to the label, certification of prescribers, medication guide, restricted distribution, etc.)
Risk Management Plans: Quantify,
Characterize and Communicate Risks
FDA vs. EMEA• Convergence rather than divergence
• Both requiring estimation of risk (pharmacoepidemiology) with drug, with population, and lowering the thresholds for identifying risk in the label
• RMP in EU is a requirement, though a risk minimization
plan is not
• REMS is analogous to a risk minimization strategy and will
become more commonplace in the US. REMS must be
accompanied by a timetable for submission of assessments.
• Both utilizing these documents to shift the discussion to one of benefit/risk in a more formalized manner.
T.O.U.C.H. – Tysabri Risk Minimization
Plan• Monoclonal antibody launched in 2004 for treatment of MS
and Crohn’s disease
• Rare, fatal illness associated with use – PML – few cases in 2005
• Removed and reintroduced in 2007 (for relapsing MS only, neurologist only) with restrictive prescribing program that has allowed:- An efficacious drug to be available for a subgroup of patients with MS under
highly restrictive prescribing conditions
- Estimate the incidence rate and outcome of PML in patients using the drug
- Characterize early clinical presentations of PML
- Possibly reduce the number of fatal cases by working with clinicians to introduce new therapies
- Collect and characterize other opportunistic infections associated with use
Drug Safety Update March 2010, vol 3
Neurology Today May 2010, vol 10: 17-18
Summary: Changes in Emphasis within PV
Old Emphasis
• Collection, coding and timely
evaluation and reporting of individual cases
• Reliance on aggregate reports to identify signals (based primarily on
spontaneous reporting which lacks
quality, estimation of exposure)
• Concentration on analysis of
company’s own data
• Not analyzed in the context of reporting rates of similar events
with other drugs
• Less emphasis on the continued
evaluation of the benefit:risk profile of the drug
Current Practices• Collection, coding timely evaluation of ICSRs and
aggegate periodic reports continue.
• Implementation of signal detection methodologies to allow for identification of signals utilizing large
safety databases; standard processes for signal evaluation
• Comparison to epidemiologic data to understand incidence/prevalence in population
• Conduct more safety studies to answer
new/ongoing safety concerns
• Staffing to ensure that qualified persons are
reviewing these data outputs per company processes
• RMPs (including the specific risk minimization plans) provide opportunity to analyze the large volume of spontaneous reports in a more focused
manner and require utilization of data from other sources when risk/safety concerns exist and allows
the company to more formally place them in the context of benefit/risk
PV and Risk Management: Continuous
• Signal Detection: allows us identify what we
do NOT know (the needle in the haystack) –more data driven
• Risk Management/Benefit:Risk Assessments:
allows us to characterize better what we do know and attempt to mitigate the risk and
maintain a favorable profile
Benefit – Risk is a Familiar Concept
Questions?