Effective use of the Historical Control Trial

1
Effective use of the Historical Control Trial

• 7th Annual DIA Biotechnology Workshop, February
  1999
• Karen D. Weiss, M.D.

2
What is an historically controlled trial?

• Study participants are treated with the new
  (test) agent
• Outcome compared to outcomes from individuals who
  do not (or have not) received the test agent
• Unlike a randomized controlled trial RCT
• All study participants receive the test agent.
• Control group is external to the trial

3
Disadvantages of externally controlled trials

• Control arm does not arise via randomization
• Advantages of randomization
• Minimizes bias in allocation of treatments
• Helps ensure groups are balanced
• Basis for tests of significance

4
Disadvantages of trials that utilize an external
control

• Onus is on the applicant to support the
  assumptions
• comparability of the control e.g.,
• diagnostic and assessment methodologies
• supportive care
• natural history
• no bias in outcome assessments
• Basis for sample size calculations, other
  statistical tests not well established

5
Advantages of trials utilizing an external control

• In some cases, an alternative design (e.g,
  placebo, active control) is unethical
• Facilitates recruitment
• Belief that new better (active control), or
  that something is better than nothing (placebo or
  no-tx control)
• Inherent distaste for computer to decide

6
When to utilize an externally controlled trial?

• When preliminary data strongly suggest efficacy
• When course of disease predictable, consistently
  poor outcome
• When endpoints objective
• When impact of baseline and other variables on
  endpoint is well characterized

7
External controls - types

• Patient in consecutive series
• Patients from the literature
• Patients from previous studies (pool)
• Patient as own control - misnomer
• Concurrent, non-randomized

8
Consecutive series

• Specific institutions/specific diseases -
• Control patients selected from a sequence of
  studies
• often, preceding trial is the control
• may utilize some or all of patients
• Likely to be from the recent past
• Same institution(s), investigators
• Completeness of the control database

9
Literature control

• Only reports from the published literature
• usually do not contain sufficient detail
• entry criteria, analytical plan, etc.
• inability to verify
• Most problematic of of external controls

10
Study pool

• For each treated patient, identify a mate from
  available pool matched for key prognostic factors
• ideally 21 ratio control active
• compare the control groups to each other
• Is the control truly a random sampling?

11
Patient as own control

• Term patient as own control is a misnomer
• Generally implies comparison of patient status
  after intervention to patient status at baseline
• ex for cancer trials, comparison of tumor burden
  at baseline vs end of treatment
• In reality, the actual comparison is the change
  from baseline in the study patients vs. the
  change from baseline that would be expected
  without the study intervention

12
Concurrent control

• Control patients diagnosed, evaluated over same
  time period, but assigned to the control, rather
  than randomized
• only certain sites have the technology
• limited availability e.g., laminar flow
• Matching possible, desirable

13
Criteria for acceptability of historical control
in conjunction with RCT

• Received precise standard tx as in RCT
• Part of a recent clinical trial, same eligibility
• Important covariates comparable between studies
• Prior study part of same organization, same
  investigators
• No signal to suggest differences between RCT and
  historical
• S. Pocock J. Chron Dis 1976 Vol 29 pp 175-188

14
Examples of effective utilization of the
externally controlled trial

• Rituxan
• refractory NHL
• Leukine
• PBPC mobilization
• Engerix-B
• immunization against Hepatitis B

15
Rituxan

• Chimeric antibody to CD20
• Patients with relapsed, low grade or follicular
  NHL
• Targets
• ORR 35-40 range,
• TTP gt or 8 mos.,
• duration of response gt or 6 mos.

16
Rituxan - comparison to published literature

• To establish the targets, review single agent
  data from similar disease state
• For each agent - can look at overall response
  rate (ORR) of individual trials (each with an n
  of 20-40) and pooled
• Fludarabine
• ORR (pooled) 40, (95 CI 30-50)
• Cladribine
• ORR (pooled) 40, (95 CI 25-55)

17
Rituxan - results

• 166 patients in phase 3, 37 in phase 2
• 1-7 prior regimens, median of 2
• Regimens included AuBMT
• Responses - 6 CR, 42 PR
• pooled ORR 48, (95 CI 40-56)

18
Rituxan - patient as own control

• 78/161 (48) ORR to Rituxan
• 63/78 (81) responded to most recent chemotherapy
• 117/161 (73) response to last chemotherapy

19
Leukine (GM-CSF) for PBPC mobilization

• May 1994 discussions with BRMAC
• optimal control for a RCT for an indication for
  mobilization?
• bone marrow
• PBPC without mobilizing agent
• RCT difficult growth factors already in
  widespread use, standard of care, encouraged
  submission of existing data

20
Leukine for PBPC mobilization

• Series of studies at U of Neb. conducted over 5
  years
• 5 groups of 196 patients total received Leukine
  for PBPC transplantation
• 2 doses, 2 routes, and /- Leukine post PBPC
• 1 group of 100 patients received PBPC without
  mobilization

21
Leukine for PBPC mobilization
22
Leukine for PBPC mobilization
23
Leukine for PBPC mobilization

• Outcomes
• time to ANC gt 500/mm3
• time to last platelet transfusion
• duration of hospitalization
• number of phereses, MNC, CFU-GM, BFU-E

24
Leukine for PBPC mobilization
25
Engerix - B

• 87 trials to assess safety and efficacy, total of
  10,000 subjects
• Major efficacy shown in newborns of mothers who
  were HBsAg
• 58 neonates born to mothers who were carriers
• 2 (3) became chronic carriers vs historical
  situation, where rates ranged from 60-90

26
Summary

• Because control (comparison) groups do not arise
  via randomization, externally controlled trials
  are not optimal as principle source of safety and
  efficacy data.
• In considering the acceptability of a trial
  utilizing an external control, it is important to
  carefully consider alternative designs and to
  justify why they are not possible.
• If an external control design is to be utilized,
  the control must be identified a priori and
  justified.