Disability Progression in MS: Why Bother? What Have We Learned? - PowerPoint PPT Presentation

2/15/2018 Clinical Trials and Long Term Follow-up: Disability Progression in MS: Why Bother? What Have We Learned? • Clinical trials offer meticulously curated data – Standardized data collection Bruce Cree, MD, PhD, MAS – Data undergoes quality control Weill Institute for Neurosciences • External study monitors review data • Data acquisition can include imaging, biometric assessments and University of California San Francisco even biological samples – Rarely if ever obtained in natural history studies – Treatment is standardized during the randomized controlled study period – Treatment is typically open-label during the LTFU Expectations from Long Term Some Long-term Follow up Study Interpretations Follow-up Studies • • Long term medication safety “Multiple sclerosis patients with mean disease duration of 22 years administering GA for up to 15 years had reduced relapse rates, and decreased disability progression and transition to SPMS.” Ford C. Mult Scler 2010;16:342-350 – Sample is limited to size of original cohort • “Disease activity despite treatment with IFNb is associated with unfavorable long-term • Longitudinal assessment of efficacy outcomes. Particular attention should be paid to gadolinium-enhancing lesions on IFNb therapy, as their presence strongly correlates with severe disability 15 years later. The – Lack of reference group results provide rationale for monitoring IFNb-treated patients with MRI, and for changing therapy in patients with active disease.” Bermel RA. Annal Neurol 2013;73:95– • Identification of predictors of long-term outcomes 103 • “This study provides class III evidence that long-term treatment with teriflunomide is – Assumes data on predictors are collected at baseline or well-tolerated and efficacy of teriflunomide is maintained long-term.” O’Connor P. Neurology 2016;86:920-930 during the study • “These findings suggest that higher cumulative exposure to sc IFN β-1a may be • Caveat: extent of attrition and baseline sample size associated with better clinical outcomes.” Kappos L. J Neurol Neurosurg Psychaitry 2015;86:1202-1207 limits interpretation of all of the above 1

2/15/2018 Attrition and Informative Censoring Long-term Follow-up Studies from Clinical Trials Medication Clinical Trial Duration of LTFU Ascertainment • Subject attrition in long-term studies occurs for multiple reasons Interferon β-1b Pivotal Trial 16 years 70% – Subject factors: relocation, loss of interest, life-changes Pivotal Trial- mortality 21 years 98% – Center factors: some centers chose not to or are unable to participate Secondary Progressive MS 10 years 50% – Disease factors: some subjects drop out or switch treatment because or BENEFIT (CIS) 11 year 59% worsening MS Interferon β-1a IM MSCRG 15 years 40% • This last feature known as informative censoring plagues interpretation of CHAMPS (CIS) 10 years 33% all studies with substantial drop-out making accurate assessment of Glatiramer Acetate Pivotal Trial 15 years 43% efficacy nearly impossible Interferon β-1a SC PRISMS 15 years 52% – For example in the glatiramer acetate (GA) long term follow up study subjects who stayed in the cohort experienced less progression than subjects who Teriflunomide TEMSO 9 years 63% discontinued GA Fingolimod FREEDOMS/TRANSFORMS 7 years 30% – A substantial proportion of subjects who discontinued GA did so due to Goodin DS. J Neurol Neuorsurg Psychiatry 2012;83:282-287. Goodin DS. Neurology. 2012;78:1315-22. worsening MS Kuhle J. Mult Scler 2016;22:533-43. Kappos L. Neurology 2016;87:978-87. Bermel RA. Mult Scler 2010;16:588-96. – Therefore, those retained in the study are positively selected to remain in the Kinkle RP. Arch Neurol 2012;69:183-90. Ford C. Mult Scler. 2010;16:342-50. Kappos L. J Neurol Neurosurg study as “GA responders” Psychiatry 2015;86:1202-1207. O’Connor P. Neurology 2016;86:920-930. Cohen JA. AAN2016 [P3.057] Interferon β-1b LTFU and Mortality: Interferon β-1b LTFU and Mortality Example of nearly complete LTFU ascertainment Survival from Randomization Survival from Clinical Onset Goodin DS. Neurology. 2012;78:1315-22. Goodin DS. Neurology. 2012;78:1315-22. 2

2/15/2018 Sources of Bias in LTFU Studies Interferon β-1b LTFU and Mortality • Nearly complete ascertainment: 366/372 (98.4%) Bias Impact Strategy • 21 year “hard” outcome of all cause mortality F/U must be as complete as possible Compare baseline and on-RCT characteristics • Modified therapeutic effect dependent on Comparison of placebo versus interferon beta-1b assignment without regard Ascertainment/Attrition of those patients in LTF to those not in LTF characteristics of participating patients. Sensitivity analysis of those lost to f/u assuming to later treatments worse outcomes • Follow-up analysis found death was could be attributed to MS in 54/69 Inflated estimate of therapeutic benefit because MPR: Use percent of total possible time on Informed Therapeutic subjects (78.3%), especially from pulmonary infections related to MS patients doing well continue therapy whereas therapy instead of absolute time to assess Decisions • failing patients switch or stop therapy. exposure. Study provided evidence that treatment with interferon β-1b has an impact on MS related mortality Propensity Scoring: Adjust for the likelihood that a particular treatment will be selected based • Provides evidence based rationale for early versus delayed initiation of Modified therapeutic effect dependent on patient Treatment Selection on available patient characteristics selection characteristics. treatment (assumes that such characteristics are relevant) • Results recently replicated in observational cohorts from Rennes France and Increased risk of Type 1 error from the use of Create a single model and apply adjustments to British Columbia, Canada (N=7009, 30% treated with IFN beta) Multiple Testing multiple predictor variables and weighting p-values according to the number of schemes predictors tested in the model. Goodin DS. Neurology. 2012;78:1315-22. Goodin DS. BMJ Open. 2012;2(6). Kingwell E. ECRTIMS 2017 abstract 2178 EPIC Subject Retention Long-term follow up in EPIC 1.0 • EPIC originally conceived of as a 5 year genotype-phenotype correlation study 0.9 – Emphasis on genetics and brain MRI 0.8 • Data available on 471/517 (91%) of subjects at • Collaborative effort with MS centers in Amsterdam Netherlands and Basel 0.7 ts the last follow up visit 10 years after baseline n tie Switzerland and Glaxo Smith Kline, formerly known as GENE-MSA a y P 0.6 • Rigorous analysis conducted on d tu • Long-term follow up of cohort proposed as a means of predicting disability S subjects lost to follow up with comparisons to f In 0.5 progression o n rtio subjects retained and sensitivity analyses 0.4 po • Anticipated observations: ro assuming worst outcomes P 0.3 – brain volume change, new lesions, EDSS change and active disease on 0.2 treatment would be predictive of long-term disability outcomes 0.1 – Escalation to high potency therapy based on clinical and MRI activity 0.0 would be associated with improved outcomes Baseline Y ear 1 Y ear 2 Y ear 3 Y ear 4 Y ear 5 Y ear 6 Y ear 7 Y ear 8−10 In Study Patients 517 505 489 484 479 472 461 460 456 Withdrawn Patients 0 9 24 28 31 36 47 48 49 Deceased (Not MS) 0 3 4 4 5 7 7 7 8 Deceased (MS) 0 0 0 1 2 2 2 2 4 Cree B. Ann Neurol 2016; 80:499-510 Cree B. Ann Neurol 2016; 80:499-510 3