Non-Prescription Mevacor

1
Non-Prescription MevacorMerck Co., Inc.New
Drug Application 21-213

• Nonprescription Drugs and Endocrinologic and
  Metabolic Drugs Advisory Committee Meeting
• Silver Spring, Maryland
• December 13, 2007
• Eric Colman, MD
• Division of Metabolism and Endocrinology Products

Center for Drug Evaluation and Research
2
Objective

• To discuss FDAs evaluation of data mining
  signals for amyotrophic lateral sclerosis (ALS)
  with statins
• Background
• ALS
• What is data mining?
• FDA data mining signal scores for ALS with
  statins
• FDAs evaluation of the data mining signals
• Next steps

3
FDA Observes Data Mining Signal for ALS with
Statins

• Earlier this year members from numerous offices
  and divisions within the Center for Drug
  Evaluation and Research met to discuss data
  mining signals for ALS and statins observed in
  FDAs Adverse Event Reporting System (AERS)
• Available data did not justify regulatory action
• Make evaluation publicly available
• Manuscript in progress

4
Edwards, et al.Drug Safety June, 2007
5
Edwards, et al.
.we hope that this signal for an ALS-like
syndrome with statins will be accepted not as
anything more than a hypothesis that needs to be
followed up to ensure the safer use of an
important group of medicines.
6
Wall Street Journal July, 2007
7
Amyotrophic Lateral Sclerosis

• Progressive destruction of motor neurons with
  retraction of axons from neuromuscular junction
• Often presents with muscle weakness
• Annual incidence 1.5 to 2 cases per 100,000
• Incidence increases with age
• Males gt females
• Etiology unknown

8
Drug Safety Data Mining

• Definition the use of computer algorithms to
  analyze adverse event data in a large, complex
  database
• FDAs Adverse Event Reporting System (AERS)
• Spontaneously-submitted adverse events from
  healthcare professionals, consumers, and drug
  companies
• Goal to identify reporting relationships that
  could signal possible adverse drug reactions
• Data mining CAN generate hypotheses regarding
  adverse drug reactions
• Data mining CANNOT prove or refute causal
  associations between drugs and adverse reactions

9
Proportional Reporting Ratios
Reports With Adverse Event Y Reports Without Adverse Event Y
Drug X a b ab
All Other Drugs c d cd
ac bd total

• Observed/Expected
• a/(ab)/c/(cd)

10
Example
Reports With Pancreatitis Reports Without Pancreatitis
Lipovent 10 200 10/10200
All Other Drugs In AERS 3 2000 3/32000

Observed ratio of pancreatitis reports for
Lipovent 10/10 200 0.048
Expected ratio of pancreatitis reports for
Lipovent 3/3 2000 0.001
Proportional reporting ratio 0.048/0.001 48
11
Data Mining Terminology

• EBGM is the Empirical Bayes Geometric Mean - an
  adjusted estimate of the mean proportional
  reporting ratio that addresses small cell counts
• EB05 lower bound of confidence interval
• EB95 upper bound of confidence interval

12
What Constitutes a Data Mining Signal?

• Criteria not written in stone
• EB05 gt 2
• EBGM gt 2
• Consider the whole range of scores for new drugs
  or very serious outcomes

13
FDA Data Mining Signal Scores for Statins and ALS
Ingredient EBGM EB05 EB95
Pravastatin 2.7 1.4 5.0
Fluvastatin 1.6 0.5 4.1
Atorvastatin 9.3 7.0 12.1
Cerivastatin 3.8 2.5 5.7
Lovastatin 2.5 1.0 5.7
Simvastatin 4.4 3.0 6.4
Rosuvastatin 2.4 1.2 4.4
EBGM is not an odds ratio or a measure of
relative or absolute risk EBGM is not a
causality score
14
What Did FDA Do In Response to the Data Mining
Signals for ALS and Statins?

• Reviewed
• AERS reports of ALS
• Statin clinical trial data
• Population-based ALS incidence data

15
Review of AERS Reports of ALS

• All reports reviewed by a safety evaluator and
  two neurologists
• 57 domestic reports
• Mean age 67 years
• 39 70 79 years old
• 53 male
• Clinical course after statin D/Cd
• 84 no improvement
• 2 improved
• 14 unknown
• Most reports received by FDA during or after 2000
• Initial Reporter Source
• 53 non-healthcare consumer
• 33 physician
• 11 non-physician healthcare provider

16
Retrospective Analyses of Statin Clinical Trials

• 42 placebo-controlled trials of all marketed
  statins 6 months to 5 years in duration
• Primary and secondary CAD prevention
• 200,000 person-years statin exposure
• 200,000 person-years placebo exposure
• 9 cases of ALS in statin groups
• 9 cases of ALS in placebo groups
• 4.3 cases per 100,000 vs. 4.6 cases per 100,000

17
Statin Use ALS Incidence

• Use of statins has increased steadily since early
  1990s
• Has the incidence of ALS increased?
• Data from Rochester, MN
• Before 1990 1.5 cases per 100,000 people per
  year (1.1 to 2.0)
• After 1990 1.9 cases per 100,000 people per year
  (1.0 to 2.8)
• Sorenson EJ, et al. Neurology.
• 200259(2)280-282.

Dispensed prescriptions by US retail pharmacies.
Verispan Vector One National, Years 1991-2006,
Extracted October 2007
18
What Should We Make of the Data Mining Signals?

• No imbalance in ALS from placebo-controlled
  statin trials
• No dramatic increase in the incidence of ALS
  despite dramatic increase in the use of statins
• Could statins unmask or exacerbate muscle
  symptoms of ALS?
• Could the data mining signals be the result of
  reporting bias?

19
Reporting Biases

• A major concern with all spontaneous reporting
  databases
• Both statins and ALS associated with muscle
  symptoms
• Detected a data mining signal for ALS with
  fenofibrate, another lipid-altering drug
  associated with myopathy
• Detected data mining signals for dermatomyositis
  and polymyositis with statins

20
Next Steps

• Ongoing case-control study
• Kaiser Permanente Northern California
• Lorene Nelson, PhD, Stanford University School of
  Medicine is the Principal Investigator
• Questions addressed
• Does the use of cholesterol-lowering drugs
  increase the risk of developing ALS?
• Is the use of cholesterol-lowering drugs
  associated with the length of survival after
  diagnosis among subjects with ALS?
• Is there a relationship between cholesterol
  levels prior to disease onset and the risk of
  developing ALS, independent of the use of
  cholesterol lowering agents?
• Study should be completed in mid-to-late 2008

21
Colleagues

• Ana Szarfman, MD, PhD
• Andy Mosholder, MD, MPH
• Devanand Jillapalli, MD
• Jay Levine, PhD
• Jo Wyeth, PharmD
• Mark Avigan, MD, CM
• Joe Tonning, MD, MPH
• Rita Ouellet-Hellstrom, PhD, MPH
• Allen Brinker, MD, MPH