Recognizing Drug-Induced Liver Injury (DILI) in Exposed Populations

1
Recognizing Drug-Induced Liver Injury (DILI) in
Exposed Populations

• John R. Senior, M.D.
• Associate Director for Science
• Office of Pharmacoepidemiology and Statistical
  Science
• Food and Drug Administration (FDA)

2

• Material presented here is based on the
  observations of the speaker for 20 years in
  academic hepatology and gastroenterology, 5 years
  as a senior executive in the pharmaceutical
  industry, 11 years in private consulting to
  industry, and upon 9.5 years at the FDA 4.5
  years as a medical reviewer for new
  gastrointestinal drugs, 3 as the Senior
  Scientific Advisor for hepatology in the Office
  of Drug Safety, 2 years as Associate Director for
  Science, Office of Pharmacoepidemiology and
  Statistical Science.
• Comments made here do not reflect official
  policies or positions of the Agency, but are
  personal opinions of the presenter based on his
  diverse experiences mentioned.

3

• First, Detect it !
• (if you dont ask or look, you wont get or
  find)
• Ask Is there liver injury or disease?
• serum transaminases, other enzymes, bilirubin,
  INR
• Is it progressive or serious?
• progressive getting worse or likely to do so
• serious disabling, life-threatening, fatal
• Drug-induced or some other cause?
• no pathognomonic test for DILI, including
  biopsy
• DILI may mimic any known liver disease

4

• some other causes
• What are they?
• mainly acute hepatitis viral A or B, seldom C,
  alcoholic, biliary stones, ischemic, or
  autoimmune other causes rarer
• How can they be ruled out?
• what information should be collected?
• What are the odds of other causes?
• Its really a problem of differential diagnosis.

5
Drug-Induced Liver Injury (DILI)

• Most people exposed to a new drug show no injury
• tolerators
• Some people show transient injury, but adapt
• adaptors
• A few fail to adapt and show serious toxicity !
• susceptibles

6
Drug-Induced Liver Injury (DILI)

• tolerators

7
Drug-Induced Liver Injury (DILI)

• adaptors

8
Drug-Induced Liver Injury (DILI)

• susceptibles

9
not DILI

10
Is It a Bad Drug or Is It aSusceptible Patient?

• Why do a few people not tolerate it?
• Whats different about them?
• Can we find out?
• A drug tolerated by nearly everybody except
  a very few cannot be considered toxic.
• And not give them this drug?

11
Given Some Drugs are More Toxic

• many or most eliminated in development
• molecular variations in class
• differing potencies of variants
• therapeutic index varies
• variable PK ADME among patients
• variable PD responses in patients
• one dose may (will) not suit all patients

12
Relative Hepatotoxicity in Class- glitazones
(thiazolidenediones)peroxisome
proliferator-activated receptor-? (PPAR?) agents
13
(No Transcript)
14
The Glitazone Family
15
Other Drug Families

• -pidems
• alpidem
• zolpidem
• -navirs
• ritonavir
• indinavir
• nefinavir
• saquinavir

• -profens
• benoxaprofen
• flurbiprofen
• ibuprofen
• -fenacs
• ibufenac
• bromfenac
• diclofenac

16
Need research on . . .

• 1) which cytochrome isoforms attack what part
  of the molecules?
• 2) which metabolites are more toxic?
• 3) proof that efficacy and toxicity are in
  different parts of same molecule ?
• 4) some rules to predict DILI
• 5) mechanisms causing the DILI
• 6) not just in animals, but in humans

17
Should We Be Looking More Carefully at the
Patients ?

• Humans are often very different than animal
  models
• They are incredibly diverse genetically
• They resist being standardized or controlled
• They take a lot of drugs, diets, supplements,
  alcohol
• They have lots of other diseases/disorders/activ
  ities
• The ones treated are not the same as the ones
  studied
• They dont always report troubles promptly

18
" Our study is man, as the subject of accidents
or diseases. Were he always, inside and outside,
cast in the same mould, instead of differing
from his fellow man as much in constitution and
in his reaction to stimulus as in feature, we
should ere this have reached some settled
principles in our art. The practice of
medicine is an art, based on science ... it has
not reached, perhaps never will, the dignity of a
complete science, with exact laws, like astronomy
or engineering.
William Osler, M.D. (1849-1919) Teacher and
Student,Minneapolis 1892
19
" he who aspires to treat correctly of human
regimen must first acquire knowledge and
discernment of the nature of man in general
knowledge of its primary constituents and
discernment of the components by which it is
controlled. though they are made from the same
materials, no two are alike
Hippocrates 460-377 B.C. The Father of
Medicine
20
Hippocrates Nature of Man elements
qualities constituents fire hot
blood water wet phlegm air
cold bile - yellow earth dry
bile - black Airs, Waters, Places
Epidemics Nutriment
21
Why Are They Susceptible ?

• idio-sug-krasia (Hippocrates, 400 B.C.)
• idios (?????) - ones own, self
• syn (???) - together
• crasis (??????) - mixing, mixture
• therefore,
• a persons own individual mixture of
  characteristics, factors nature and nurture,
  unique in the world
• it does NOT mean rare, unexpected, unexplained,
  although it may or may not be any or all of them

22
44_at_FL(6DVF4V (idiosugkrasia) s.f.,
(substantive, feminine) temperament
constitution
Pocket Oxford Greek Dictionary Oxford University
Press, Oxford UK, 2000
23
Factors in Idiosyncrasy

• genetic, bestowed at conception
• gender
• cytochromes, enzymes, transport systems
• acquired in life since conception
• age, activities, travels
• infections, immunities, diseases
• diet, obesity, dietary supplements
• other drugs, chemical exposures

24
Especially Susceptible Patients

• adverse effect, often not dose-related
• may not be duration-related
• may not depend on prior disease
• unexpected, unpredictable (up to now)
• risk factors not well known
• toxicity often uncommon or rare
• who are they? how can they be identified?

25
Drug Development Issues Idiosyncratic Drug
Toxicity

• Occurs despite careful preclinical testing in
  animals
• and despite careful and costly clinical
  trials
• Some problem (perhaps wrong) assumptions
• Assuming drug variability is the key to safety
• Obsolete concepts of safe and effective for
  all?
• Assuming one dose/regimen suits all patients
• Focus on efficacy, little on safety (especially
  if rare)

26
Needles in the Haystack Tyranny of the Numbers -
I

• if only 1 patient in 1000 reacts adversely to a
  drug, should we call the drug toxic?
• If 1 person in 1000 is idiosyncratically
  hyper-susceptible to show drug-induced liver
  injury, how many patients need to be observed to
  have 95 chance of finding at least 1?

27
A Binomial Answer . . .

• - if i incidence of a rare event, and u usual
  absence of the event, then (i u) 1, it
  happens or it doesnt. And (i u)N, where N
  number of persons
• - when uN, the chance of seeing no events in a
  group of N persons, is lt0.05, then the chance of
  seeing at least one is gt95, because 1 uN 1-
  lt0.05 gt95
• - for (0.001 0.999)N, need 2995 people to make
  (0.999)2995 0.04996, which is lt0.05
• - in general, N ? 3/i for rare events.

28
Detection of Rare Safety Problems

• Before approval and marketing
• NDA review seldom have enough patients, so rare
  but serious event may not be observed but if
  they are seen, investigate!
• After marketing and clinical use
• spontaneous, voluntary reporting useful, but
  severely underreported and incomplete data
• problems usually worse after marketing

29
Liver Function Tests

• what are they ?
• what do they mean ?
• Important distinguish between liver function
  and liver injury
• theyre the biomarkers we use

30
Commonly Used Tests
injury
transaminases ALT
AST
hepatocellular cholestatic
enzymes
alkaline phosphatase gamma-glutamyl transferase
function excretory synthetic
synthetic
bilirubin (total, direct) albumin prothrombin
(INR)
substances
31
Is Serum ALT a Liver Function Test ?

• serum enzymes not just from liver but also from
  skeletal and heart muscle, gut, etc.
• . . . so lets not assume liver
• it is not a function or job of the liver to
  regulate the level of serum enzyme activity
• . . . so lets not say function
• elevated serum ALT activity MAY indicate
  hepatocellular injury

32
Is Serum Transaminase Enough?
33
Need Very, Very Specific TestsTyranny of the
Numbers - II

• For relatively low prevalence or incidence
  events, need extremely high specificity of test
  to avoid finding many false positives
• Almost impossible to find a hyper-specific test
  compound tests may help ALTTBL
• Probably need to add clinical differential
  diagnostic data and methods to attain

34
Hy Zimmerman 1917-1999
35
ESTIMATED CASE FATALITY RATE IN DRUG-INDUCED
ACUTE HEPATOCELLULAR INJURYWITH NON-OBSTRUCTIVE
JAUNDICE
Zimmerman. HEPATOTOXICITY, 1978 1999
36
Intrinsic vs Idiosyncratic Toxicity

• intrinsic
• predictable
• dose-related
• similar in animals
• high incidence
• short interval
• types
• directly destructive
• indirect, metabolic
• cholestatic

• idiosyncratic
• unpredictable
• not dose-related
• not seen in animals
• low incidence, rare
• variably longer interval
• types
• hypersensitivity
• metabolic
• H. Zimmerman, 1978

37
HZbut thats an oversimplification!this
dichotomy into intrinsic toxicity or host
idiosyncracy is an overly simplified formulation.
The potential for injury is arranged along a
spectrum, modified by several different
mechanisms.

• Relative Importance for Hepatotoxicity
• phosphorus
• amanita phalloides
• carbon tetrachloride
• anabolic steroids
• tetracycline
• chlorpromazine
• phenytoin
• penicillin
• Intrinsic Toxicity -------------------------------
  -----some of both---------------------------------
  ------ Host Susceptibility
• __________________________________________________
  _________________________________________

38
Hys Law ...for drug-induced hepatotoxicity

• If both hepatocellular injury and jaundice occur,
  look out for at least 10 mortality!
• i.e., when both transaminase and bilirubin
  elevations occur together.
• Robert Temple, FDA, 1999

39
Searching for Clues, Predictors

• Can ways be found to search pre-approval clinical
  databases (NDA submissions) for clues or
  signals that predict liver failure in at least
  some who show the signal?
• Hy Zimmerman was on to a key concept - the
  combination of an indicator of injury and loss of
  overall function may be a valid predictor. It
  needs confirmation.

40
Temples Corollary ...for drug-induced
hepatotoxicityHys Law cases arise out of a
background of increased incidence of transaminase
elevations

• Hys Rule-type serious cases of DILI usually
  arise from a background of increased injury
  rates, so
• Look for more ALT elevations in those exposed to
  drug, compared to placebo or control drug.
• Robert Temple, FDA, 2004

41

42
What to Do When a Signal Detected

• signal laboratory abnormality, complaint,
  finding
• immediately (within day or two) confirm by
  repeat tests, examination, questions
• investigate possible causes AT ONCE ! r/o
  disease
• start close observation (3x/week) immediately
• obtain more information about possible causes
• obtain consultation as appropriate or needed
• follow closely (weekly, prn) to normalization or
  worsening and management decision

43
It may be DILI if its nothing else

• Diagnosis of exclusion no test FOR DILI
• Must gather data to rule out other causes
• Need to educate docs to do it better
• Develop model for quantitative likelihood
  estimate
• Prospective large safety studies needed
• for true incidence
• for risk factors and to design risk management
  plans
• for omic analyses (gen-, prote-, metabon-)
  specimens
• for elucidation of mechanisms

44
Tip of the Iceberg

• Death or Tx
• Acute Liver Failure
• Serious DILI Threatening
• Detectable DILI but Not Serious
• Patient Adaptation to New Agent Exposure
• Patients/People Tolerate Exposure Without Effects

45
Serious Adverse Events

• Often have relatively low incidence rate, i
• Difficult to establish correct value for i
• Voluntary reporting gives uncertain values
• May be missed in the usual efficacy trials
• Must balance modest clinical benefits for many
  against serious risks for a few
• Spontaneous reports (AERS) not enough

46
Retrospective Database Studies

• Epidemiologic, observational closer to reality
• A definite improvement on spontaneous reports
• Limited to who was included, data recorded
• No attribution of causality statistical
  differences
• Can improve estimates of incidence, risk factors
• But cannot investigate mechanisms or causes or
  do it right in real time, collect specimens

47
How many patients to study?

• Prospective, not retrospective
• Focus on safety, as drug actually prescribed
• Large and long enough to find rare events
• Full accounting, numbers of cases/exposed
• Use patient self-reporting, then investigate
  possible cases intensely get needed info
• Prospective case-control design
• Impartial study conduct NIH or AHRQ

48
Prospective Large Safety Studies

• Offer best chance to learn, and to protect
  patients
• Need to establish true incidence of DILI
• Need to identify risk factors for rational RM
• Need to investigate mechanisms
• In best interests of everybody, including
  patients, regulators, and sponsors
• In 2005, are we content with safety meaning not
  poison, and withdrawing toxic drugs ?