The (frequently long, circuitous, and bumpy) trip from clinical problems to clinical trials Bill Burman Denver Public Health © 2014 Denver Public Health One view of clinical trials © 2014 Denver Public Health Clinical trials: cautionary notes • Time for a Phase 3 trial – a decade (or more!) – Trial development – 1 to 5 years, trial conduct – 4 to 8 years, trial analysis – 1 to 2 years – Will the question be relevant in 10 years? • Cost of classic clinical trials (GCP, professional site monitoring): – $5-25,000 per patient/year – For a trial of 1000 patients, $5 million to $25 million per year of the trial • Opportunity cost – Limited funding for TB clinical trials – Is this a durable and critical question? © 2014 Denver Public Health Another view of clinical trials… © 2014 Denver Public Health Problems of not doing clinical trials: When to start ART • When to start antiretroviral therapy (ART) – classic case for a randomized trial – Important clinical and programmatic question – Randomization possible at the level of the patient • Challenges for “When to start” study – Shifting equipoise – Concerns about feasibility © 2014 Denver Public Health Shifting recommendations for “When to start ART” – IAS USA panel > 500 350-500 VL>30K VL>5K VL>10K VL>5K VL>5K 200-350 <200 CD4 © 2014 Denver Public Health 1996 1998 2000 2002 2004 2006 2008 2010 Shifting recommendations for “When to start ART” – IAS USA panel > 500 VL>30K VL>5K VL>10K VL>5K VL>5K 350-500 “It is questionable whether a randomized 200-350 trial to study the issue of When to Start will ever <200 be feasible.” 2002 guidelines CD4 © 2014 Denver Public Health 1996 1998 2000 2002 2004 2006 2008 2010 Shifting recommendations for “When to start ART” – IAS USA panel > 500 VL>30K VL>5K VL>10K VL>5K VL>5K 350-500 “It is questionable whether a randomized 200-350 trial to study the issue of When to Start will ever <200 be feasible.” 2002 guidelines CD4 © 2014 Denver Public Health 1996 1998 2000 2002 2004 2006 2008 2010 When to start ART: Potential of ART • Prevent classic AIDS outcomes • Prevent irreversible loss of immune function (clonal deletion) • Prevent other serious events: – cardiovascular, renal, and hepatic events – other cancers – neurocognitive dysfunction • Prevent HIV transmission © 2014 Denver Public Health When to start ART: Problems of ART • Short-term side effects – diarrhea, rash, etc. • Long-term metabolic effects – Hyperlipidemia, bone loss – Renal and hepatic toxicity • Behavioral disinhibition – leading to increased risk of transmission • Acquired drug resistance, transmitted resistance • Cost of antiretroviral therapy – drugs, monitoring, increased clinical care © 2014 Denver Public Health SMART study – ART-naïve or off ART for > 6 months • CD4 > 350 • Randomized – Immediate ART – Deferred ART: start when CD4 < 250 – ART: physician/patient choice • 477 enrolled (of 5472 in overall trial) – 249 ART-naïve – 225 off ART for > 6 months • Median follow-up – 18 months © 2014 Denver Public Health Rate per 100 person-years Outcomes of SMART sub-study of “When to Start” 8 7 6 5 P = 0.002 P = 0.02 P = 0.01 4 3 2 1 0 AIDS/death Serious non-AIDS < 250 © 2014 Denver Public Health Composite > 350 SMART Study Group. J Infect Dis 2008; 197: 1133-44 Lessons of the “When to start ART” controversy • > 15 years of controversy not resolved by many, increasingly large and increasingly sophisticated observational studies • Clarity provided by two relatively small randomized clinical trials (combined n = 1293, with median follow-up for < 2 years) • Recommendations for common, important clinical decisions should be based on randomized trials © 2014 Denver Public Health Examples of the problems of not doing clinical trials: TB © 2014 Denver Public Health Examples of the problems of not doing clinical trials: TB • Sub-optimal treatments – Dosing of 1st (and 2nd) line TB drug in children – MDR-TB regimens without clarity on critical questions (e.g., number of drugs, length of aminoglycoside use, selection of fluoroquinolone) • Uncertainty about managing toxicity – Drug-related liver injury – Dosing of PZA, use of PZA in higher-risk patients • Key details about 1st-line therapy – Optimal duration – Intermittency – when during treatment, degree © 2014 Denver Public Health Getting started • Identify a compelling clinical problem – important in terms of diagnosis, treatment effectiveness, toxicity, prevention, population control • Find relevant observational studies – Current practices – Key outcomes: definitions, rates, associated factors, estimation of the effect size of an intervention • Consider doing research on what to do research on – Survey clinicians: key questions, areas of equipoise – Survey patients: outcomes of TB treatment that are important to patients – Formal decision analysis – cost-effectiveness modeling of various forms of treatment of latent TB © 2014 Denver Public Health Getting started • Identify a compelling clinical problem – important in terms of diagnosis, treatment effectiveness, toxicity, prevention, population control • Find relevant observational studies – Current practices – Key outcomes: definitions, rates, associated factors, estimation of the effect size of an intervention • Consider doing research on what to do research on – Survey clinicians: key questions, areas of equipoise – Survey patients: outcomes of TB treatment that are important to patients – Formal decision analysis – cost-effectiveness modeling of various forms of treatment of latent TB © 2014 Denver Public Health Getting started • Identify a compelling clinical problem – important in terms of diagnosis, treatment effectiveness, toxicity, prevention, population control • Find relevant observational studies – Current practices (sometimes substantially different from guidelines) – Key outcomes: definitions, rates, associated factors, estimation of the effect size of an intervention • Consider doing research on what to do research on – Survey clinicians: key questions, areas of equipoise – Survey patients: outcomes of TB treatment that are important to patients – Formal decision analysis – cost-effectiveness modeling of various forms of treatment of latent TB © 2014 Denver Public Health Getting started • Identify a compelling clinical problem – important in terms of diagnosis, treatment effectiveness, toxicity, prevention, population control • Find relevant observational studies – Current practices (sometimes substantially different from guidelines) – Key outcomes: definitions, rates, associated factors, estimation of the effect size of an intervention • Consider doing research on what to do research on – Survey clinicians: key questions, areas of equipoise – Survey patients: outcomes of TB treatment that are important to patients – Formal decision analysis – for example, cost-effectiveness modeling of various forms of treatment of latent TB © 2014 Denver Public Health Trials of treatment-shortening with fluoroquinolones • Rationale – Mouse model studies – moxifloxacin allowed treatment shortening to 4 months, especially if substituted for INH – Improved treatment completion, leading to improved TB control – Fluoroquinolones are well-tolerated, known to be safe • Three Phase 3 trials undertaken © 2014 Denver Public Health Activity of moxifloxacin in combination therapy in a mouse model of TB Log CFU in entire lung 10 9 8 7 Untreated 2RHZ+4RH 2RHZM+4RHM 2RMZ+4RM 6 5 4 3 2 2.5 logs 1 0 0 1 2 3 4 Duration of treatment (mos.) Am J Respir Crit Care Med 2004; 164:421-6 5 6 Concerns about fluoroquinolones for treatment-shortening • Toxicity/tolerability – Uncertainty about safety in pregnancy, children, and the elderly – Concerns about possible cardiotoxicity from moxi – Selection for resistance in M. tuberculosis might limit 2nd-line regimen potency – Broad activity against common bacterial pathogens – selection for resistant strains, C. difficile • Lack of consistent activity in Phase 2 trials • Opportunity cost © 2014 Denver Public Health Effect of moxifloxacin on 2-month sputum culture conversion 100 90 80 70 60 50 40 30 20 10 0 82 71 80 71 64 Study 27 South Africa Moxi © 2014 Denver Public Health 63 Brazil Comparator 60 55 Study 28 EMB Effect of moxifloxacin on 2-month sputum culture conversion 100 90 80 70 60 50 40 30 20 10 0 N = 102 N = 336 71 82 71 64 Study 27 South Africa Moxi © 2014 Denver Public Health N = 150 80 63 Brazil Comparator N = 443 60 55 Study 28 EMB Treatment-shortening using moxifloxacin for drug-susceptible active disease Relapses • IRZE – 12 (2%) • IRZM – 46 (9%) • MRZE – 64 (12%) 2-month conversion • IRZE – 83% • IRZM – 85% • MRZE – 87% Gillespie S, et al. N Engl J Med 2014; 371: 1677-87 © 2014 Denver Public Health TB clinical trials: drug-susceptible TB • Lessons from the past decade – Regimens for drug-susceptible disease must be applicable to women of child-bearing age and children – Don’t start Phase 3 trials of treatment-shortening without compelling results from Phase 2 – Is treatment-shortening to 4 months the most important question in treatment of drug-susceptible TB? – We are far to quick to conclude that a new regimen is “safe and well-tolerated” © 2014 Denver Public Health Suppression of ventricular ectopy after acute MI • Sudden death is common after a heart attack • Ventricular ectopy is a risk factor for sudden death • Drugs are available that markedly suppress ventricular ectopy • In a pilot randomized trial, these drugs were welltolerated • Clinicians started to prescribe these drugs to suppress ventricular ectopy © 2014 Denver Public Health % of patients with > 70% suppression Efficacy in suppressing ventricular ectopy 90 80 70 60 50 40 30 20 10 0 flecainide encainide moricizine placebo Am J Cardiol 1988;61:501-9 © 2014 Denver Public Health Tolerability in Phase 2 Encainide (n = 99) Flecainide (n = 103) Moricizine (n = 98) Placebo (m = 100) Death 0 0 0 0 CHF 1 0 0 0 MI 0 0 1 2 Other SAE 0 1 0 0 Drug stopped 2 3 6 7 Am J Cardiol 1988;61:501-9 © 2014 Denver Public Health Mortality Effect of flecainide, encainide on mortality - data from Phase 3 (n = 1500) 9 8 7 6 5 4 3 2 1 0 total mortality cardiac mortality active drug placebo N Engl J Med 1989;321:406-12 © 2014 Denver Public Health Lessons of the CAST study • Fixing a surrogate marker may not fix the disease and may even exacerbate it • Other examples: – post-menopausal hormone replacement therapy and CVD risk – effect of antiretroviral therapy on CVD risk – effect of EPO on anemia in ESRD • Sample sizes in most clinical trials are inadequate to detect unusual but serious adverse effects © 2014 Denver Public Health TB clinical trials: drug-susceptible TB • Lessons from the past decade – Regimens for drug-susceptible disease must be applicable to women of child-bearing age and children – Don’t start Phase 3 trials of treatment-shortening without compelling results from Phase 2 – We are far to quick to conclude that a new regimen is “safe and well-tolerated” – Is treatment-shortening to 4 months the most important question in treatment of drug-susceptible TB? © 2014 Denver Public Health Prospective study of tolerability of treatment for pulmonary TB • Multicenter prospective study – 4 regions of China, sampling scheme to assure representative sample of patients with pulmonary TB • Baseline survey and labs • Treatment: IRZE for initial treatment, Strep added for re-treatment • Symptom diary during treatment • Repeat labs at 2 months • Adverse events and TB treatment outcomes evaluated using standardized criteria © 2014 Denver Public Health Demographic and clinical characteristics of the cohort Parameter Number (% of 4304) Median age (IQR) 42 (29 – 55) Male/female 3082 / 1227 TB treatment history © 2014 Denver Public Health Primary 3556 (83% Re-treatment 748 (17%) Hep B S Ag positive 469 (11%) History of drug reaction 118 (3%) Hepatobiliary disease 23 (0.5%) Gastroenteropathy 40 (1%) Diabetes 51 (1%) Other medical illnesses 103 (2%) Lv Z, et al. PLoS One 2013 Adverse drug reactions • Frequency – 766 (17%) had an adverse reaction, 1.4% had a serious adverse reaction, 1% hospitalized – Liver dysfunction – 6.3%, 0.6% had serious hepatotoxicity • Effect on TB treatment regimen – 43% of those with adverse reaction had regimen changed, 5% stopped all TB treatment • Effect on TB treatment outcomes – 2.8% with adverse reaction had unsuccessful TB treatment (vs. 1% of those without an adverse reaction) – 19% of all unsuccessful outcomes attributed to adverse reactions © 2014 Denver Public Health Adverse drug reactions • Frequency – 766 (17%) had an adverse reaction, 1.4% had a serious adverse reaction, 1% hospitalized – Liver dysfunction – 6.3%, 0.6% had serious hepatotoxicity • Effect on TB treatment regimen – 43% of those with adverse reaction had regimen changed, 5% stopped all TB treatment • Effect on TB treatment outcomes – 2.8% with adverse reaction had unsuccessful TB treatment (vs. 1% of those without an adverse reaction) – 19% of all unsuccessful outcomes attributed to adverse reactions © 2014 Denver Public Health Adverse drug reactions • Frequency – 766 (17%) had an adverse reaction, 1.4% had a serious adverse reaction, 1% hospitalized – Liver dysfunction – 6.3%, 0.6% had serious hepatotoxicity • Effect on TB treatment regimen – 43% of those with adverse reaction had regimen changed, 5% stopped all TB treatment • Effect on TB treatment outcomes – 2.8% with adverse reaction had unsuccessful TB treatment (vs. 1% of those without an adverse reaction) – 19% of all unsuccessful outcomes attributed to adverse reactions © 2014 Denver Public Health Comparison of hepatotoxicity risk among antimicrobial agents Drug Hepatotoxicity incidence per 100,000 courses Amox-clav * 1-17 Comments Generally benign Telithromycin * 17 Withdrawn from the market Levofloxacin * 0.02 Trovofloxacin * 6 Withdrawn from the market Rifampin 70 Annals Intern Med 2008; 149: 694 Isoniazid 380 Annals Intern Med 2008; 149: 694 Pyrazinamide 430 Ann Intern Med 2002; 137: 640-7 * Andrade R, Tulkens PM. J Antimicrob Chemother 2011; 66: 141-6 In population-based studies, TB drugs are among the most common causes of serious drug-related hepatotoxicity (Aliment Pharmacol Ther 2010;31:1200, Gastroenterol 2008;135:1924) © 2014 Denver Public Health Changes in age-distribution among cases of active TB (Hong Kong) As transmission decreases, active TB becomes increasingly a disease of the elderly Wu P, et al. PLoS One, 2010 © 2014 Denver Public Health Age-specific rates of active TB, by WHO region http://www.who.int/tb/publications/global_report/en/ © 2014 Denver Public Health Toxicity considerations in treatment of drug-susceptible disease • INH and PZA are much more hepatotoxic than antibiotics that have been removed from the market • The new norm of TB: – Elderly, increased prevalence of comorbid disease – Higher risk of adverse effects, including hepatitis and CVD • Encourage enrollment of elderly and “complicated patients” into TB clinical trials • Is it time for trials comparing interventions for toxicity avoidance? © 2014 Denver Public Health Endpoints – How to define “better” (or “not inferior”) • Clinical endpoints (TB, relapse, death) – Infrequent events in TB patients – Postulated differences in clinical event rates (treatment effect size) may be small – Infrequent events analyzed as dichotomous endpoints + small effect size = large sample sizes • Surrogate markers (EBA, effect on transaminases, cholesterol, etc.) – Continuous variables with large dynamic range = markedly smaller sample size © 2014 Denver Public Health Endpoints – the lessons of CAST and SMART • Surrogate markers – quite reasonable for initial studies, to generate hypotheses • Effects of interventions may be much more complex than their effect on one or more surrogate markers • Surrogate markers should not be relied on as the basis for major clinical decisions/policies © 2014 Denver Public Health Dealing with complex clinical situations • Heterogeneity of “MDR-TB” – Primary vs. acquired – Degree of resistance to 2nd-line drugs – Different definitions of “resistance” – Limitations posed by intolerance to 2nd-line drugs – Severity of pulmonary disease • Common conclusion – Impossible to do clinical trials for MDR-TB © 2014 Denver Public Health Example of “optimized background regimen” design for MDR-HIV • Study population – Prior therapy with NRTI, NNRTI, PI – Virological failure of current therapy (VL > 5000) • Randomization – Optimized Background Regimen (OBR) chosen by enrolling clinician – could include other investigational drugs – OBR + T-20 (enfuvirtide) © 2014 Denver Public Health T-20 trial: baseline characteristics Characteristic T-20 Control Previous ART > 5 PIs Lopinavr/R Tenofovir 49% 39% 3% 39% 28% 0 Viral load < 40,000 > 40,000 20% 80% 20% 80% N Engl J Med 2003;348:2175-85 © 2014 Denver Public Health T-20 trial: baseline characteristics Characteristic T-20 Control Genotypic susc. Score 0 1-2 3-4 >5 16% 52% 28% 4% 13% 56% 27% 2% Use of other investigational agents Lopinavir/R Tenofovir 62% 8% 62% 7% N Engl J Med 2003;348:2175-85 © 2014 Denver Public Health T-20 trial – efficacy results N = 501 N Engl J Med 2003;348:2175-85, N Engl J Med 2003;358: © 2014 Denver Public Health T-20 trial – efficacy results from paired trials N = 501 N Engl J Med 2003;348:2175-85, N Engl J Med 2003;358: © 2014 Denver Public Health Clinical trials for complex problems • Don’t try to over-simplify life – manage complexity, don’t try to eliminate it • We under-estimate the power of randomization to deal with clinical heterogeneity © 2014 Denver Public Health Alternate trial designs • Behavioral or system-level interventions (e.g., DOT vs. SAT, availability of rapid susceptibility testing, community-based testing to decrease TB transmission) – Cluster-randomized – Stepped-wedge trial © 2014 Denver Public Health Alternate trial designs • Behavioral or system-level interventions (e.g., DOT vs. SAT, availability of rapid susceptibility testing, community-based testing to decrease TB transmission) – Cluster-randomized – Stepped-wedge trial © 2014 Denver Public Health Andrew Nunn’s rules for clinical trials 1. Ask important questions 2. Use simple designs 3. Enroll enough patients to answer the question My corollaries • Don’t rush to trial design; do research on what to do research on • Seek out – don’t avoid – a wide range of feedback on the design are worthy © 2014 Denver Public Health Andrew Nunn’s rules for clinical trials 1. Ask important questions 2. Use simple designs 3. Enroll enough patients to answer the question My corollaries • Don’t rush to trial design; do research on what to do research on • Seek out – don’t avoid – a wide range of feedback on the design are worthy © 2014 Denver Public Health Closing thoughts • Clinical trials – Not for the faint of heart – Not for the impatient – Not for the loner collaborative exercise, political process – But the team can accomplish something big © 2014 Denver Public Health Closing thoughts • Clinical trials – Not for the faint of heart – Not for the impatient – Not for the loner collaborative exercise, political process – But the team can accomplish something big © 2014 Denver Public Health