data mining approaches to signal detection
TRANSCRIPT
Dr. Bhaswat S. ChakrabortySenior VP, R&D, Cadila
PharmaceuticalsJnauary 22, 2009
Presented at Pharmacovigilance 2010, January 21-22, 2010
Risk Management and Pharmacovigilance Increased focus on safety and risk management is a
global issue with diminishing boundaries
Knowledge and experience of the drug and its life-cycle further develops as we understand its use and hazards
Development of global risk management plan rather than individual region or country Risk
Management plans would be an important step forward to more effectively and accurately assess the safety of pharmaceutical drug products
An early risk management planning between company, regulator and healthcare professionals is key for successful product life cycle management
Risk Communication Plan between Company, Regulator, Healthcare Professionals and Patients is getting more transparent
And Yet…Despite this increased importance, stakeholders
still use traditional methods for PV
Conservative methods do not capture many SAEs and USAEs that are possible to capture from huge databases without further experimentation
Data mining is one such approach
Data mining can help find unrecognized toxic signals
Two categories of approved drugCategory 1
Those who are unequivocally superior to existing drugs of that class in efficacy
May or may not be superior in safetyCategory 2
Those who are not superior to existing drugs of that class in efficacy
Non-inferiorSuperior to placebo but inferior to existing
standard careMay or may not be superior in safety
Premature Approval?Many Category 2 drugs whose complete safety
profile is still unknown were approvedIn some cases, drugs are approved despite
identification of SAEs in premarketing trialsAlosetron hydrochloride – ischemic colitisGrepafloxacin hydrochloride – QT prolongation and deathsRofecoxib – heart attack and stroke (long-term, high-
dosage use)
They were all subsequently withdrawn from the market because of these SAEs
Market Uptake and Sales Volume Many drugs whose complete safety profile is still
unknown actually have/had a rapid market and very high sales volume through increased Rx.
Promotion of early use of new drugs by sponsorsPhysicians' adoption of such drugsDirect-to-consumer drug advertisingPharma companies concern for patent life, a desire to
mold prescribing habits prior to the market entry of competitors
“Ramp-up" in sales encourages investors and increase stock prices.
New drug safety may be further compromised by the failure to conduct postmarketing studies
J. Herson. In Data and Safety Monitoring Committees in Clinical Trials
Having an Adverse Events Database
Is not a bad ideaAll good pharma companies have AE databaseAlmost all developed country regulatory
agencies have AE databaseThe WHO Uppsala Monitoring Centre (UMC)
now receive >1,000,000 reports per year
Such databses can really help in bringing down drug induced morbidity
Desirable Attributes of AE Database SoftwareShould be well integrated with Clinical data
management softwareUser friendlyIndividual reports management featuresEasy for queryLine listing of the entire database or part is
possible and easyData extraction is easy, with desirable filtersMay also keep track of postmarketing Rx
utility and complaints data
Collection of ICSRs from CADRMP
Conversion of free text to structured information
Data cleaning and duplicate detection
Applying quantitative or statistical methods
Computing an accurate measure for SD
Gavali, Kulkarni, Kumar and Chakraborty (2009), Ind J Pharmacol, 41, 162-166
or any comprehensive database
Targeted Event Y All other events Total
Targeted Drug X
A B A+B
All other drugs
C D C+D
Total A+C B+D A+B+C+D
Criteria for a Toxic Disproportional ADR
ROR =
χ2 =
)(
)(
DCC
BAA
PRR
DC
BA
Expected
ExpectedObserved 2)(
Significant disproportional Signal is detected when 2 is ≥ 4.0 and the rest ≥ 2.0
Casestudy Example: Propranolol-Bradycardia
PRR = 2.51
ROR = 2.58
χ2 = 3.26
Therefore, bradycardia is not a significant disproportional signal (Serious Adverse Event) associated with Propranolol
BradycardiaNot
Bradycardia
Propranolol HCL
4 82
Not Propranolol HCL
52 2749
Gavali, Kulkarni, Kumar and Chakraborty (2009), Ind J Pharmacol, 41, 162-166
Casestudy Examples – Significant Signals
AssociationBupropion – seizuresOlanzapine – thrombosisPergolide – increased libidoRisperidon – diabetes mellitusTerbinafine – stomatistisRosiglitazone – liver function abnormalities
Dis-associationIsotretinoine– suicide
Source: LAREB
Thank You Very Much
Acknowledgment:Sharwan Singhal