1 Study Design Issues and Considerations in HUS Trials Yan Wang, Ph.D. Statistical Reviewer Division...

14
1 Study Design Issues and Considerations in HUS Trials Yan Wang, Ph.D. Yan Wang, Ph.D. Statistical Reviewer Statistical Reviewer Division of Biometrics IV Division of Biometrics IV OB/OTS/CDER/FDA OB/OTS/CDER/FDA April 12, 2007 April 12, 2007

Transcript of 1 Study Design Issues and Considerations in HUS Trials Yan Wang, Ph.D. Statistical Reviewer Division...

Page 1: 1 Study Design Issues and Considerations in HUS Trials Yan Wang, Ph.D. Statistical Reviewer Division of Biometrics IV OB/OTS/CDER/FDA April 12, 2007.

1

Study Design Issues and Considerations in HUS Trials

Yan Wang, Ph.D.Yan Wang, Ph.D.

Statistical ReviewerStatistical ReviewerDivision of Biometrics IVDivision of Biometrics IVOB/OTS/CDER/FDAOB/OTS/CDER/FDA

April 12, 2007April 12, 2007

Page 2: 1 Study Design Issues and Considerations in HUS Trials Yan Wang, Ph.D. Statistical Reviewer Division of Biometrics IV OB/OTS/CDER/FDA April 12, 2007.

2

Outline

• Adequate and well-controlled studies• Study design issues and considerations in HUS

trials– Choice of primary efficacy endpoint

– Efficacy evaluation and sample size

– Safety evaluation and sample size

• Conclusion

Page 3: 1 Study Design Issues and Considerations in HUS Trials Yan Wang, Ph.D. Statistical Reviewer Division of Biometrics IV OB/OTS/CDER/FDA April 12, 2007.

3

Adequate and Well Controlled Studies 21 CFR 314.126 (b)

1. Clear statement of objectives

2. Study design permits valid comparison with appropriate control to provide quantitative assessment of drug effect

3. Select patients with disease (treatment) or at risk of disease (prevention)

4. Baseline comparability (randomization)

5. Minimize bias (blinding, randomization, etc.)

6. Appropriate methods of assessment of outcome

7. Appropriate methods of analysis

Page 4: 1 Study Design Issues and Considerations in HUS Trials Yan Wang, Ph.D. Statistical Reviewer Division of Biometrics IV OB/OTS/CDER/FDA April 12, 2007.

4

Ideal/Preferred Study Design in HUS Prevention Trials

• Randomized, double blind, placebo-controlled

• Target population may include– Patients with STEC infection and at risk of developing HUS

• Primary efficacy endpoint– Incidence of HUS

• Adequate power to detect treatment effect

• Adequate data to demonstrate safety

Page 5: 1 Study Design Issues and Considerations in HUS Trials Yan Wang, Ph.D. Statistical Reviewer Division of Biometrics IV OB/OTS/CDER/FDA April 12, 2007.

5

Challenges for Designing HUS Prevention Trials

• Rare and sporadic nature of STEC infection

• Unpredictable nature of HUS development from STEC infection

• Therapeutic window may be narrow (possibly within 48 hours after infection)

• Low incidence rate of HUS in patients with STEC infection (5%-15%)

Page 6: 1 Study Design Issues and Considerations in HUS Trials Yan Wang, Ph.D. Statistical Reviewer Division of Biometrics IV OB/OTS/CDER/FDA April 12, 2007.

6

Sample Sizes Required for Efficacy in HUS Prevention Trials Under Various Scenarios

Number of Patients per Arm (80% power, alpha=5%, 2-sided)

906

337

1102

435163278

105

2306

701

33% 50% 75%

Assumed Treatment Effect (% Reduction of Placebo Rate)

5%

10%

15%

Placebo Placebo IncidenceIncidence

Rate of HUSRate of HUS(STEC Infection)(STEC Infection)

Page 7: 1 Study Design Issues and Considerations in HUS Trials Yan Wang, Ph.D. Statistical Reviewer Division of Biometrics IV OB/OTS/CDER/FDA April 12, 2007.

7

Use of a composite endpoint can reduce sample size because the number of events increases

Can we use a composite endpoint in a HUS trial (e.g. “HUS + other clinically relevant events) ?

Number of Patients per Arm (80% power, alpha=5%, 2-sided)

2306

906

337163

1102

435701

278105

300121 47

33% 50% 75%

Assumed Treatment Effect (% Reduction of Placebo Rate)

HUS: 5%

HUS: 10%

HUS: 15%

HypotheticalComposite: 30%

Placebo Placebo IncidenceIncidence

RateRate

Page 8: 1 Study Design Issues and Considerations in HUS Trials Yan Wang, Ph.D. Statistical Reviewer Division of Biometrics IV OB/OTS/CDER/FDA April 12, 2007.

8

Difficulty in interpretation when treatment effects on components are not homogenous

Hypothetical example of a HUS trial using a composite endpointHypothetical example of a HUS trial using a composite endpoint

Composite Endpoint

Component #1: HUS

Treatment Effect

Component #2

Component #3

Page 9: 1 Study Design Issues and Considerations in HUS Trials Yan Wang, Ph.D. Statistical Reviewer Division of Biometrics IV OB/OTS/CDER/FDA April 12, 2007.

9

Considerations for Using a Composite Endpoint

• Are the individual components clinically relevant and of similar importance to patients?

• Do the more and less important endpoints occur with similar frequency?

• Is the underlying pathophysiology of the components similar?

• Are the components likely to have similar relative risk reductions?

Page 10: 1 Study Design Issues and Considerations in HUS Trials Yan Wang, Ph.D. Statistical Reviewer Division of Biometrics IV OB/OTS/CDER/FDA April 12, 2007.

10

• Issues/considerations for efficacy evaluation in HUS trials– Sample size can be prohibitive if “prevention of HUS” is

used for efficacy measurement

– Use of a composite endpoint (if possible) can reduce sample size for testing treatment effect

– Potential difficulty in interpreting results of the composite endpoint when the treatment effect on composite endpoint cannot be translated to an effect on HUS prevention

• Safety evaluation– Need to have an adequate number of treated patients

for safety evaluation

Page 11: 1 Study Design Issues and Considerations in HUS Trials Yan Wang, Ph.D. Statistical Reviewer Division of Biometrics IV OB/OTS/CDER/FDA April 12, 2007.

11

Chance of Observing No Serious Adverse Events

• What does it mean when no serious event (SAE) was observed or no safety issues were identified in a clinical trial of a new product?

• It doesn’t necessarily mean that the new product is safe because the chance of observing no events can be high when trial size is small.

Chance of Observing No Events = (1-p)N

Incidence of SAE (p)

Number of Patients Treated in a Trial (N)

50 100 200 300 600 1000 2000

0.5% 78% 61% 37% 17% 4.9% 0.67% 0.00%

1% 61% 37% 13% 4.9% 0.24% 0.00% 0.00%

Page 12: 1 Study Design Issues and Considerations in HUS Trials Yan Wang, Ph.D. Statistical Reviewer Division of Biometrics IV OB/OTS/CDER/FDA April 12, 2007.

12

Rule of Three (3/N)

6.00%

3.00%

1.50%1.00%

0.50%0.30% 0.15%

50 100 200 300 600 1000 2000

Number of Patients Treated in a Trial

Upper Bound of 95% CI for Incidence of an Adverse Event

Provide an answer to “What is the worst possible scenario for the risk of a serious adverse event when no events occur in a trial?”

If no events occur in N treated patients, the upper bound of the 95% confidence interval for the risk (p) can be estimated as

3/N.

Page 13: 1 Study Design Issues and Considerations in HUS Trials Yan Wang, Ph.D. Statistical Reviewer Division of Biometrics IV OB/OTS/CDER/FDA April 12, 2007.

13

Majority of Patients With STEC Infection May Not Benefit From Prophylactic Therapy for HUS

# of Patients Needed to be Treated to Prevent one HUS Case

Assumed Treatment Effect

Placebo HUS Incidence Rate

5% 10% 15%

33% Reduction 60 30 20

50% Reduction 40 20 14

75% Reduction 27 14 9

98% 98%

95% 95%

89%89%

• 85% to 95% of patients with STEC infection will not develop HUS

• Number of patients needed to be treated to prevent one HUS case could be large

Page 14: 1 Study Design Issues and Considerations in HUS Trials Yan Wang, Ph.D. Statistical Reviewer Division of Biometrics IV OB/OTS/CDER/FDA April 12, 2007.

14

Conclusion

• Need to have adequate and well controlled clinical trials to evaluate efficacy and safety of new products for HUS prevention

• Challenges for designing HUS prevention trials with– Adequate statistical power to evaluate a clinically

meaningful efficacy endpoint – Sufficient data to demonstrate safety

Return to Main Menu.