Variant CJD risk associated with human plasma derivatives: Introduction and overview of risk model f

1

• Variant CJD risk associated with human plasma
  derivatives Introduction and overview of risk
  model for US manufactured
• Factor VIII
• Steven Anderson, PhD, MPP
• Associate Director for Risk Assessment
• Office of Biostatistics Epidemiology
• FDA-CBER
• TSE Advisory Committee Meeting
• October 31, 2005

2
Elements of Risk AssessmentNAS (1983)

• Hazard identification
• Establishes causality between hazard and adverse
  effects
• Dose response (Hazard characterization)
• Probability of response infection or illness
• Exposure assessment
• Frequency and level of exposure
• Estimates potential DOSE vCJD ID50
• Risk characterization
• Probability of occurrence, severity of adverse
  effects
• Uncertainty
• Sensitivity analysis

3
Risk AssessmentNAS (1983)

• Risk Assessment is conducted when information is
  limited and uncertainty high
• Value of Risk Assessment
• Provides estimate of risk
• Details the Uncertainties
• Determine effectiveness of Mitigations
• Identifies data gaps and research priorities

4
Risk Assessment

• Uncertainty
• Arises in risk assessment when
• Only limited information is available
• Data are lacking
• Use of assumptions / expert opinion
• Errors in measurement or data collection
• Incorrect specification of model

5
FDA Risk Assessment vCJD risks for US
manufactured Factor VIII

• Factor VIII assessment under development
• Models vCJD risk for Factor VIII made in 2002
• Assessments for additional years possible - 1999
• Beginning of process to assess risks for plasma
  derivatives in future may
• Assess vCJD risks additional product classes
• Possibly assess risk by specific product, per
  individual, etc.
• Process model analyzes
• Probability quantity vCJD agent in plasma pools
• Reduction in levels vCJD ID50 in manufacturing
• Quantity Factor VIII used exposure vCJD ID50

6
Overview of vCJD and US manufactured Factor VIII
Risk Assessment Model
7
(No Transcript)
8
Module 1 Prevalence of vCJD in United Kingdom

• Proposed Modeling Approaches
• Two major sources of UK vCJD prevalence data
• A. Predictive Modeling based on vCJD cases in UK
• B. Surveillance data examination
  tonsil/appendix samples
• Disparity of approximately 10 to 100 fold between
  the two approaches

9
Module 1Prevalence of vCJD in United Kingdom

• A. Predictive mathematical models
• Some of data we may use (select sample many
  others!)
• (1) Ghani et al 2003
• vCJD estimate median 100 cases (10 2,600 - 95
  CI)
• Median 1 in 500,000
• (2) Boelle, et al 2003
• vCJD estimated cases range 183 to 304 cases
• (3) Llewyn et al 2004
• vCJD infection 1 in 15,000 to 1 in 30,000
• Approximately 1,000 - 2,000 infections
• Include all genotypes in estimate - codon 129
  PrP MM, MV, VV

10
Module 1 Prevalence of vCJD in United Kingdom

• B. Surveillance data tissue samples UK
  patients
• Tonsil/appendix surveillance study (Hilton, et
  al. 2004)
• 3 prion positive samples in 12,674 samples tested
• Mean of 1 positive in 4,225 individuals
• Mostly in 20 30 yr old patients
• Approximately 13,000 vCJD infected UK individuals
• Data would be further age adjusted using
  reported UK vCJD cases age profile

11
Module 1 Prevalence of vCJD in United Kingdom

• Uncertainties of Proposed Modeling Approaches
• Predictive Modeling based on vCJD cases in UK
• Only estimate clinical cases of MM codon 129
  individuals
• Use of assumptions incubation period, time of
  infection, etc
• Surveillance data examination tonsil/appendix
  samples
• vCJD agent in appendix may not have agent in
  blood
• May not become symptomatic
• Overestimate (Sample size relatively small for
  rare disease)
• Underestimate vCJD prevalence - in one infected
  case vCJD agent not in appendix
• Generally neither approach adequately address
• Clinical or asymptomatic cases MV codon 129, or
  VV 129 PrP
• vCJD infections that dont progress to
  symptomatic disease

12
Module 1Prevalence of vCJD in United Kingdom

• Estimation of prevalence vCJD in UK population
• Critical parameter in model used to estimate
  vCJD prevalence for
• France
• Europe
• Plasma donors in United States

13
Module 2Prevalence vCJD in US plasma donors

• Proposed Modeling Approach
• Estimate size of US plasma donor population
  travel history to UK, France or Europe since 1980
• Determine travel characteristics from survey data
  
• Adjust travel data by several factors (duration
  of stay, year, etc.)
• Estimate probability of infection in individual
  donors
• Add up potential number cases for US plasma donor
  groups
• Apply effectiveness of donor deferral policy
• Model output predict number of
• Potential vCJD infected US plasma donors
• vCJD infected donors deferred from donation
• Potential donations with vCJD agent

14
Module 2Prevalence vCJD in US plasma donors

• vCJD in US donors possible from dietary exposure
  to BSE agent during travel
• Current policy Defers donors with travel
  history
• UK gt 3 months - 1980 1996
• France gt 5 years - 1980 present
• Europe gt 5 years - 1980 - present (blood
  donation only)
• Estimated 90 - 99 effective eliminating vCJD
  donors / risk
• Residual risk - Two donor groups of interest
• 1) Deferrable risk
• 1 - 10 with deferrable travel history risk
  but donate
• 2) Short duration travel
• UK lt 3 months - 1980 1996
• France lt 5 years - 1980 present
• Europe lt 5 years - 1980 - present (blood
  donation only)

15
Module 2Prevalence vCJD in US plasma donors

• Concept of Relative Risk
• Used to estimate vCJD prevalence France and
  Europe relative to UK prevalence
• Based on BSE exposure, number vCJD cases, etc.

16
Module 2 Potential vCJD Prevalence in US Plasma
Donors

• Relative risk for US plasma donor with travel
  history to UK, France or Europe since 1980
• FDA model adjusted relative risk for vCJD
  over 23 year period (1980 2002)
  based on
• Duration of Travel
• Relative risk is adjusted on a per month or per
  day basis
• Specific year(s) of travel
• Accounts for variation in BSE epidemic / exposure
• Age of donor
• To apply age specific rates for vCJD in UK
  (median age 28yrs)

17
Module 2 Potential vCJD Prevalence in US Plasma
Donors

• Propose to model two types of
  plasma donors
• 1. Source Plasma donors (gt 80 donations)
• Age specific donation rates
• 2. Recovered Plasma donors (lt 20 donations)
• Age specific donation rates

18
Module 2Estimation Potential vCJD Prevalence in
US Plasma Donors

• Plasma Donor travel estimated from
  survey data
• Survey random sample of blood donors by
  American Red Cross
• Conducted Dec 1998 Jan 1999
• Queried travel history and accumulated stay
  information for UK, Europe and/or (France) during
  period 1980 - 1996

19
Module 2Estimation Potential vCJD Prevalence in
US Plasma Donors

• Modeling effectiveness of geographic deferral
  policy for travel to UK, France and Europe
• Data be discussed by Dr. Alan Williams
• Potential values that could be used in FDA model
  for effectiveness vCJD deferral policies
• Reduces 90 to 95 of vCJD risk from first-time
  donations
• May reduce 99 vCJD risk from repeat donors

20
Module 2Estimation Potential vCJD Prevalence in
US Plasma Donors

• When is vCJD agent present in blood during
  incubation period?
• Detailed discussion of data by Dr. David Asher
• Two potential approaches could be modeled
• During entire incubation period
• Assumption used in FDA Factor XI risk assessment
• During last half of incubation period?
• Brown, et al. 1999 prions in blood later in
  incubation period
• Modeling would be complex
• Increased uncertainty
• Assumptions would be made about duration
  incubation periods

21
Module 2Estimation Potential vCJD Prevalence in
US Plasma Donors

• Uncertainties
• Survey conducted on whole blood (recovered plasma
  donors)
• No survey on travel characteristics Source Plasma
  donors
• Source plasma donors may travel less
• Blood donor travel information may overestimate
  risk for Source Plasma
• Estimation deferral effectiveness a challenge
  because self-deferral
• Estimation of vCJD agents presence in blood
  from animal data may not be accurate for humans

22
Module 3Factor VIII Processing

• Proposed Modeling Approach
• Estimation probability plasma pool contains vCJD
  donation
• Estimation quantity vCJD ID50 per ml plasma and
  per pool
• Efficiency of exposure through i.v. versus
  i.c. route
• Log10 reduction in quantity iv ID50
• Model output to predict
• Percentage plasma pools with vCJD agent
• Percentage vials with vCJD agent
• Quantity vCJD agent per vial

23
Module 3Factor VIII Processing

• Proposed modeling approaches for estimation
  quantity vCJD ID50 per ml plasma and in pool
• Data to be discussed by Dr. David Asher
• Data ic ID50 per ml blood
• Propose to use triangular distribution with
• Minimum 0.1
• Most likely 10
• Maximum 310

24
Module 3Factor VIII Processing

• Proposed approaches estimation efficiency of
  exposure route to vCJD ID50
• FDA Factor XI risk assessment - assumption with
  range 5 to 10 fold (Kimberlin, et al 1996)
• Recent unpublished data suggest adjustment 1 to 5
  fold for efficiency of intracerebral vs.
  intravenous route exposure
• FDA proposes to use estimate of 1 to 5 fold for
  adjustment of efficiency of intracerebral vs.
  intravenous route exposure

25
Module 3Factor VIII Processing

• Proposed approach estimation of
• Plasma pool size in FDA model
• Number of donations per plasma pool used in
  manufacturing ranges from 20,000 donations up to
  60,000 donations
• FDA needs more accurate information on size of
  pools used in manufacturing
• FDA proposes to use a bimodal distribution
  indicating plasma pools most likely contain
  20,000 donations or 60,000 donations

26
Module 3FVIII Processing

• Proposed modeling of reduction vCJD ID50 during
  FVIII processing
• Detailed information to be presented by Dr.
  Dorothy Scott
• Some reduction levels vCJD ID50 expected to occur
  during processing and manufacture of Factor VIII
• Designations for degree FVIII purity
  (intermediate or high purity) may have little
  relationship to level of vCJD ID50 clearance
• FDA proposes 3 values for range Log10 reduction
  ID50 during processing
• 2 Log10 , 5 Log10 , 8 Log10,
• Distributions about these values will capture
  uncertainty
• Possible distributions 1-3 Log10, , 4 6
  Log10, , 7 9 Log10, , etc.

27
Module 3FVIII Processing

• Uncertainties in the data
• Only limited amount of data available of TSE
  reduction for a small number processing steps and
  few products
• Levels of reduction achieved with spiked
  infectivity may not accurately reflect levels
  achieved during manufacturing
• Experimental data obtained for other TSE agents
  and not specifically for vCJD agent
• Does addition of orthogonal reduction steps
  reflect actual reduction ID50 during
  manufacturing?

28
Module 4 Utilization of Factor VIII

• Proposed Modeling Approach
• Inputs
• Percentage vials with vCJD agent
• Quantity vCJD agent per vial
• Annual utilization / dose Factor VIII per patient
• Model output to predict
• Annual potential dose vCJD ID50 per patient
• Prediction of risk of vCJD infection based on
• animal dose-response information

29
Module 4 Utilization of FVIII

• Utilization to be discussed by Dr. Mark Weinstein
• Utilization Factors being considered for model
• Severity of Hemophilia A
• Severe, Moderate and Mild
• Treatment regimens
• Prophylaxis and sporadic
• Utilization data from CDC Hemophilia Treatment
  Centers may be used in model
• 3000 patients followed 1993 1998 utilization
  based on review of medical chart
• If available may use additional data sources
  from medical databases

30
Module 4 Utilization of FVIII

• Uncertainties
• Utilization data not most current and may not
  accurately reflect current prescribing practices
• Patients on multiple products utilization not
  separately reported for each
• Patients may move among categories (prophylaxis
  to episodic, etc.) difficult accurately capture
  in estimate
• To reduce uncertainty FDA is seeking additional
  sources of FVIII utilization data

31
Exposure Assessment

• Should the FDA model address apparent cumulative
  nonlinear effects of
  repeated dosing?
• Data to be discussed by Dr. Mark Weinstein
• Single Dose
• Repeated / Cumulative Doses (Diringer H, et al.
  1998, Jacquemot, et al. 2005)
• Modeling the risk of repeated / cumulative doses
  would be a challenge and increase uncertainty in
  risk estimate
• Limited data suggest that in some cases there may
  be an added nonlinear increase in infection rates
  with repeated dosing
• FDA proposes to model cumulative vCJD exposure
  per year assuming a linear ID50 dose-response

32
Exposure Assessment

• Exposure assessment of FDA model will provide
  estimate of
• potential vCJD ID50 dose
• Estimated dose from model coupled with
  dose-response
  relationship for vCJD agent
• Model will assume vCJD ID50 is a linear
  dose-response

33
Exposure Assessment

• Exposure to Fractional ID50 or Indivisible ID50
• (1) Fractional ID50 infection
• 20 individuals exposed to 0.1 ID50 then
• likely that 1 individual exposed to 0.1 ID50
  would become infected
• (2) Indivisible ID50 infection
• 20 individuals exposed to 0.1 ID50 then
• Probability of 0.1 or 10 of receiving one ID50
• Implies that 19 individuals receive no ID50 and
  1 receives 1 ID50
• likely that 1 individual exposed to 1 ID50 and
  infected

34
Dose response (vCJD)

• Uncertainties
• Factor VIII risk assessment uses Animal ID50 as
  linear dose-response to estimate human risk !
• Animal data limited adding uncertainty to
  dose-response
• Human data not available
• Development of a human dose response model is not
  possible at this time

35
Risk CharacterizationvCJD risk for Factor VIII

• Conclusions
• Estimated risk of infection based on level of
  exposure (dose) can be predicted using model
• Risk Prediction based on animal data and animal
  dose response
• Therefore risk estimates will be highly uncertain
  
• Risk assessment will highlight data gaps and
  uncertainties
• Risk estimates do provide information on relative
  magnitude of risk for risk management purposes