1. Microbiological Surrogate Endpoints in Clinical Trials of Infectious Diseases Microbiological Surrogate Endpoints in Clinical Trials of Infectious Diseases John H. Powers, MD Lead Medical Officer Antimicrobial Drug Development and Resistance Initiatives Office of Drug Evaluation IV Center for Drug Evaluation and Research U.S. Food and Drug Administration

2. 2Introduction  Differentiating clinical practice guidelines and clinical trials  Definitions of endpoints, clinical endpoints, surrogate endpoints, and biomarkers  Utility of surrogate and differentiating risk factors from surrogate endpoints  Strengths and limitations of surrogate endpoints  Surrogates in the setting of topical antiseptics  Conclusions

3. 3 Clinical Practice and Clinical Trials  Clinical practice  use of drug products already proven to be safe and effective  not concerned with causality of outcomes  treatment guidelines describe use of products based on best available evidence (optimally based on clinical trials)  Clinical trials  experiments in humans to determine if drugs products are safe and effective  determine causality of outcomes related to interventions  provide evidence for formulation of practice guidelines (not usually vice versa)  need “yardstick” for determining if products are safe and effective

4. 4Definitions  Endpoint - measure of the effect of an intervention on outcome (e.g. success or failure) in a clinical trial in treating or preventing a disease  Several implications:  WHAT to measure - should be clinically relevant to the disease in question  HOW to measure it - should be able to measure differences between therapies should they exist i.e. developing a “yardstick” to differentiate effective from ineffective drugs  WHEN to measure it (timing) should be clinically relevant to the disease  HOW MUCH to measure (magnitude) should correlate with clinical outcomes that are meaningful to patients  HOW TO ANALYZE results as method may affect conclusions

5. 5 GI symptoms bother me I’m worried and concerned I cannot bend over or exercise I cannot eat and drink whatever I like My whole life is affected Heartburn disturbs my sleep

6. 6Definitions  Disease  a constellation of signs and symptoms experienced by the patient  infectious diseases caused by pathogenic organisms with resultant host response causing symptoms  Koch’s postulates relate to proving cause of disease, not effect of an intervention  cause and effect are separate considerations  Patients seek the care of clinicians because of symptoms, not because of presence of organism  Verheij TJ et al. Fam Pract 1990;7:175-80.  In prevention trials, seek to prevent symptoms from occurring  Powers JH Clin Infect Dis 2004;39:S211-17.

7. 7Definitions  Clinical endpoint = direct measure of how a patient feels, functions or survives  mortality  resolution or prevention of symptoms of disease  Surrogate endpoint = laboratory measurement or physical sign used as a substitute for clinical endpoint  culture results, radiological testing, histology  other data like PK-PD predicts effect on organism  Surrogate endpoint by itself does not confer direct clinical benefit to the patient  NIH Biomarkers Definitions Working Group, Clin Pharmacol Ther.2001;69:89-95.  International Conference on Harmonization, ICH E9 document

8. 8Definitions  Biomarkers are analytical tools used to assess biological parameters  Biomarkers are useful for purposes other than surrogate endpoints in trials 1) diagnostic tool - use of test as an inclusion criteria to define the disease based on presence of organism (increases specificity of diagnosis) 2) describe mechanism of action of drug - effect on organisms is mechanism, not goal of therapy 3) risk factor for acquisition of disease - colonization with particular organisms may be risk factor 4) risk factor for outcome - indicator of disease prognosis (HIV viral load and CD4 in HIV)  Sande MA et al. NIH Consensus Conference on HIV, JAMA 1993;270:2583-89. 5) surrogate endpoints

9. 9Definitions  Surrogate - Etymology:  Surrogate - Etymology: Latin surrogatus, past participle of surrogare to choose in place of another, substitute, from sub- + rogare to ask : to put in the place of another:   Merriam-Webster Dictionary   Surrogate endpoint is substituting microbiological outcomes for clinical outcomes (both success AND failure) of disease in pivotal clinical trials to prove drug efficacy   Surrogate endpoints very useful in early drug development as proof of principle of drug efficacy and selecting candidate drugs

10. 10 Utility of Surrogates  Useful in phase 3 trials when surrogate endpoint can be measured sooner than clinical endpoint  When clinical endpoint events are more rare, can complete trial with smaller sample size  Examples of successful surrogates:  lowering cholesterol to prevent CV disease  lowering blood pressure to prevent CV disease  suppression of HIV viral load as surrogate endpoint in treatment of HIV/AIDS

11. 11 HIV Viral Load and CD4 as Risk Factors

12. 12 Risk Factor and Endpoints Differ  Level of HIV viral load and CD4 count both are risk factors for disease progression in HIV/AIDS  HIV viral load functions well as surrogate endpoint but CD4 count not as useful as endpoint in clinical trials  7 of 8 trials in with positive effect on CD4 showed positive effect on progression to AIDS/death  BUT effect on CD4 positive in 6 of 8 trials with negative effect on AIDS progression/death  CD4 count may not change fast enough over time course of trial  Fleming TR. Stat Med 1994;13:1423-35.

13. 13 Strengths and Limitations  The “logic string” for topical antiseptic products  colonization with organisms precede infection (risk factor)  organisms can cause infection and resulting host response  since organisms cause infection, eliminating or decreasing organisms should results in positive clinical outcomes for patients  So logical, so “objective”…..but is it correct?  DeGruttola V et al. Ann Intern Med 1997;175:237-46.  Deductive, not inductive reasoning  Need to test this logic to see if it is true  Prentice RL Stat Med 1989;8:431-440.

14. 14 Strengths and Limitations Intervention SurrogateEndpoint ClinicalEndpoint Unmeasured harm Unmeasured benefits Other pathways of disease Reasons why surrogate may not accurately predict clinical outcomes: unmeasured harms caused by intervention unmeasured harms caused by intervention unmeasured benefits caused by intervention unmeasured benefits caused by intervention other mechanisms of disease other than those affected by intervention other mechanisms of disease other than those affected by intervention issues with measuring surrogate issues with measuring surrogate issues with measuring clinical outcomes issues with measuring clinical outcomes

15. 15 Strengths and Limitations Surrogate may not take into account unmeasured Surrogate may not take into account unmeasured benefits and harms of treatment (not just “correlate”) Knowledge about how a drug achieves clinical results may be incomplete Knowledge about how a drug achieves clinical results may be incomplete Conservation of processes in human body Conservation of processes in human body Intervention SurrogateEndpoint ClinicalEndpoint Unmeasured harm Unmeasured benefits Other pathways of disease

16. 16 Strengths and Limitations  Unmeasured benefits  effects of drug other than “eradication” of organisms  sub-inhibitory effects of antimicrobials on organisms  bactericidal therapy not necessary in many infections  direct effects of antimicrobials on host immune system  Labro MT et al. Curr Opin Investig Drugs 2002:3:61-8.  Nau R et al. Clin Micro Rev 2002;15:95-110.  Unmeasured harms  deleterious effects on host that may promote infection  replacement of one organism with another that may cause infection  other sources of infection other than those affected by drug

17. 17 Strengths and Limitations  Unmeasured benefits  rifaximin approved for treatment of travelers diarrhea  drug have similar rate of positive cultures for pathogens in stool compared to placebo  drug still decreased time to resolution of diarrhea compared to placebo (for some, not all organisms)  Steffen R et al. Am J Gastroenterol 2003;98:1073-8.  Unmeasured harms  dose escalation trial of clarithromycin for disease due to Mycobacterium avium-intracellulare in patients with AIDS  higher doses had higher rates of negative blood cultures (microbiological outcomes) but also higher mortality (clinical outcomes)  Chaisson et al. Ann Intern Med 1994;121:905-911

18. 18 Strengths and Limitations Intervention SurrogateEndpoint ClinicalEndpoint Unmeasured harm Unmeasured benefits Other pathways of disease Intervention may affect one pathway of disease and not other pathways that are as important or more important in resulting clinical outcomes

19. 19 Strengths and Limitations  Other mechanisms of disease not affected by intervention  several trials show decreased rates of colonization with S. aureus (microbiological outcome) with intranasal mupirocin  prevention of infections (clinical outcome) in patients not lower than placebo  Kalmiejer MD et al. Clin Infect Dis 2002;35:353-8.  Perl TM et al. N Engl J Med 2002;346:1871-7.  Wertheimer HF et al. Ann Intern Med 2004;140:419-25.  May be due to infection caused by sources of S. aureus other than the nose

20. 20 Strengths and Limitations Issues with accuracy of how surrogate is measured Issues with accuracy of how surrogate is measured it may be reproducible (precision) but is it tellingit may be reproducible (precision) but is it telling us the correct inference (accuracy)? what, when, how and magnitude of what is measuredwhat, when, how and magnitude of what is measured Unmeasured benefits Unmeasured harm Intervention SurrogateEndpoint ClinicalEndpoint

21. 21 Strengths and Limitations  Culture techniques based on methodology from late 1800’s with inherent error (same organisms can give MIC one to tube dilutions different on serial testing)  Issues with microbiological outcomes  patient population sampled  sampling technique  culture methodology  when culture is performed (“on therapy”) and follow-up period  frequency of sampling  criteria applied for classification (“all or nothing” vs. quantitative)

22. 22 Strengths and Limitations Microbial load Time level of detection Patient 1 Patient 2 baselineend of studyon therapy Suppressionvs.“Eradication”

23. 23 Strengths and Limitations Unmeasured benefits Unmeasured harm Intervention SurrogateEndpoint ClinicalEndpoint Measurement of clinical endpoint may not Measurement of clinical endpoint may not be relevant based on natural history of disease inaccurate measurement of clinical endpoint does not inaccurate measurement of clinical endpoint does not justify use of unvalidated surrogate

24. 24 Strengths and Limitations  Example  Catheter tip decolonization claimed to be “validated” as surrogate endpoint for clinical trials in prevention of catheter-related bloodstream infections (BSI) based on “correlation” of two endpoints  Rijnders BJA et al. Clin Infect Dis 2002;35:1053-8.  Definition of “bloodstream infection” in some of these studies is positive blood culture AND a positive culture of a catheter tip  “  “residual antimicrobial activity in the removed catheter sufficient to prevent growth from the cultured catheter segments would substantially reduce the apparent rate of catheter-related bloodstream infections…. could it be that use of minocycline–rifampin impregnation prevents growth from catheters in the microbiology laboratory but does not eliminate the clinical syndrome of catheter- related bloodstream infection? ”  Paterson DL N Engl J Med 1999;340:1761.  Studies submitted to FDA show no effect on BSI relative to placebo despite effect on decolonization

25. 25 “Correlating” Surrogates Surrogate must measure effects similarly for Surrogate must measure effects similarly for all drugs studied % success with surrogate % clinical success Perfect correlation slope = 1 80% 80%

26. 26 “Correlating” Surrogates Surrogate must measure effects similarly for all drugs studied Surrogate must measure effects similarly for all drugs studied adapted from Baker SG et al. BMC Medical Research Methodology 2003;3:16. % clinical success Perfect correlation slope = 1 Drug A Drug B AB AB % success with surrogate

27. 27 Regulatory Issues  Traditional approval based on surrogate endpoints only in cases where endpoint already validated to predict clinical endpoint  Accelerated approval based on surrogate endpoints (Subpart H, 21 CFR 314.510)  serious and life threatening disease  surrogate endpoint reasonably likely to predict clinical outcome  requires confirmatory post-approval trial based on clinical endpoint (usually from a trial that is already ongoing)  TFM predates these regulations

28. 28 Surrogates and Topical Antiseptics  Unmeasured harms  unintended effects on microscopic breakage in skin may result in greater infection rate (example of shaving and peri-operative infections)  Seropian R et al. Am J Surg 1971;121:251-4.  effects on common pathogens may be less than that on marker organisms in skin  selection of resistance to systemic antimicrobials?  Unintended benefits  some products may have positive effects other than those on organisms  effects on common pathogens may be greater than that on marker organisms used to contaminate skin

29. 29 Surrogates and Topical Antiseptics  Other mechanisms not affected by intervention  data obtained on surrogate endpoint is from most superficial layer of skin (stratum corneum of epidermis)  source of organisms causing infection may not be the same as site measured by surrogate  “skin stripping” experiments using cellophane tape  12 different sites on body had variable colony counts of coagulase negative staphylococci  top 5 layers of stratum corneum had higher colony counts and next 20 layers with stable counts  return of same organisms 18 hours post-sterilization with alcohol indicates reservoir of normal skin flora may be below stratum corneum in hair follicles/glands in dermis  Brown E et al. J Infect Dis 1989;160:644-50.

30. 30 Skin Anatomy

31. 31 Skin Anatomy

32. 32 Surrogates and Topical Antiseptics  Issues with measurement of surrogate  Population in whom surrogate is measured may be different from actual use population in unmeasured ways  Organisms measured not necessarily those that cause infection  Is timing of measurements relevant to disease process?  Are conditions of testing same as those that would be encountered in real life situations?  Does variations in methodology (vigor of scrubbing, use of neutralizers) affect consistency of results?  What log reduction is clinically significant? How to analyze numbers obtained on log reductions? Changing method of analysis may change clinical conclusions

33. 33 Surrogates and Topical Antiseptics  Clinical endpoints  What is data showing correlation of reduction in bacteria with decrease in infection rates?  What does dose response curve look like for infection rates and numbers of bacteria? Is this a threshold effect (lower infection rates below some amount of bacteria) or a continuous variable?  Are correlations the same for all types of products?

34. 34 Surrogates and Topical Antiseptics Rate of infection Change in numbers of bacteria on skin “Dose Response” of Numbers of Skin Bacteria and Infection Rates

35. 35 Surrogates and Topical Antiseptics Rate of infection Change in numbers of bacteria on skin “Dose Response” of Numbers of Skin Bacteria and Infection Rates

36. 36 Surrogates and Topical Antiseptics What Kind of Endpoint? Clinical Endpoint Surrogate Endpoint What to Measure? Infections in Patients in Patients Numbers of Bacteria Design Study and Define “Success” Percent of Patients not infected not infected MeanLog Reduction Reduction MedianLogReduction Percent of Subjects Meeting Log ReductionThreshold Analyze Results Comparison with Control (soap and water? other product) superiority non-inferiority MeetSomeThreshold Comparison with Control (vehicle? other product?) superioritynon-inferiority

37. 37Conclusions  Surrogate endpoints must not only “correlate” with clinical outcomes but must also take into account :  unmeasured harms and benefits  methodology and uncertainties in measuring surrogate  appropriate measurement of clinical endpoint  Clinical endpoint for efficacy of topical antiseptic products would be prevention of infections  condition of testing today are not those of Semmelweis  Further discussion today on what is known about surrogates in setting of topical antiseptics

38. 38 “Far better an approximate answer to the right question, which is often vague, than an exact answer to the wrong question, which can always be made precise.” John W. Tukey (1962) Annals of Mathematical Statistics 1962;33:1-67.