Pot otentia ial l trial d des esigns s and suitable e study populati tions EMA MA stakeho eholde der i interaction o n on the d dev evelopm pmen ent o of medicinal p produc ducts f for ch chronic n c non-infectious l liver er di dise sease ses ( s (PBC, P , PSC, N NAS ASH) 3 3 De December 2018 2018 Bettina E Hansen IHPME, University of Toronto Toronto Center for Liver Disease, UHN Gastro & Hepatology, Erasmus MC, The Netherlands
Selecti ection on o of study population PBC Symptomatic/Asymptomatic UDCA Responders Non-Responders Reduced liver Progressive disease related deaths or OLT Cirrhosis, portal hypertension and/or HCC Premature death or OLT
New Treatment (Rx) if insufficient response to UDCA PBC Symptomatic/Asymptomatic UDCA Responders Non-Responders New Rx Reduced liver Progressive disease related deaths or OLT Cirrhosis, portal hypertension and/or HCC Premature death or OLT
Inclu clusio ion cr criteria ia of often r rela elated t to o res esponse cr criteria ia Duration: 1 year POISE 1 – trial Inclusion: ALP>1.67 OR abnormal bilirubin, but bilirubin < 3xULN Response: ALP<=1.67 AND min. 15 % reduction compared to baseline AND normal bilirubin BEZURSO 2 - trial Inclusion: Non-responder according to Paris I Response: normal bilirubin, normal ALP, AST, ALT, albumin and PT 1 Nevens et al ; NEJM 2016; 2 Corpechot at al; NEJM 2018
Study y population: h high r risk EMA advocates a study population: • at highest risk for progression in urgent need of new treatment 8.9% 17.5% • risk population after min 1 year of UDCA: • ALP >2 xULN ? AND ? • abnormal bilirubin • additional selection may depend on 65.8% 7.9% • AST, albumin, GGT, Mayo risk Hansen, Global PBC dec 2018 Reflection paper on regulatory requirements for the development of medicinal products for chronic non-infectious liver diseases (PBC, PSC, NASH); 2018; Prentice, Stat in Med; 1989
Su Suitable s study dy popul ulatin Selection of an appropriate study population is critical to: • Ethical acceptability • Minimize bias confounders • Numbers of subjects • Speed of enrollment • Interpretation and extrapolation of data • Acceptance by physicians and regulatory authorities
St Study p population: at r risk ALP < 1.67 Normal ALP ≥ 1.67 bilirubin Abnormal ALP < 1.67 bilirubin ALP ≥ 1.67 Lammers et al, Gastroenterology 2014
Zoomin ing i in on A ALP LP belo elow 2 2 and n nor ormal b l bilir lirubin in Alkaline phosphatase (ALP) Bilirubin tra n s p la n ta tio n o r d e a th (9 5 % C I) T im e 0 c o h o rt tra n s p la n ta tio n o r d e a th (9 5 % C I) 6 4 Abnormal 5 H a z a rd ra tio fo r H a z a rd ra tio fo r 3 4 3 2 2 1 1 0 0 0 1 1.67 2 3 0 .0 0 .5 0 .7 1 1 .0 1 .5 A lk a lin e p h o s p h a ta s e ( × U L N ) B iliru b in ( × U L N ) ALP: lower is better Bilirubin: > 0.6 - 0.7 at higher risk Murillo et al., AASLD 2017; Murillo et al., AASLD 2018
Rotterdam am Di Disease se S Stage: B Biliru rubin & & albumin Advanced Early MODERATE Both Both normal One Abnormal abnormal N=2039 N=1084 N=238 N=133 N=57 N=106 CLINICAL EVENT N=296 1 Kuiper et al; Gastroenterology 2009; Hansen 2017 APASL
Use e of Globe s scor ore or or oth other r risk s scores t to s o sel elect ct study po population These patients could potentially benefit of additional therapies HR globe score > threshold = 4.5 C-stat = 0.82 50 th percentile 0 100 Lammers et al., Gastroenterology 2015 http://globalpbc.com/globe
Discu cussio ion: s : sel elect ctio ion of of h high r risk p pop opula latio ion PROS CONS • May be to late • In urgent need • Treatment not efficient in high risk group • Balance of cost benefit? • Other population need to wait • …. • Extrapolation of results questionable • Ethical aspects • ….
Recycle an and R Reuse se d data an and knowled edge Hu Huge d databanks a are e part t of t the soluti tion - Especially these are powerful: • For rare diseases and events are distant in time • To gain knowledge of the natural history / standard of care • To understand differences in disease stage, patient characteristics, geographical differences • To study outcomes • To study biomarkers • To support the search for potential surrogate endpoints • To use for design of new studies (power analysis, selection of patients) • To use as potential historical controls
Des esign of of phase 3 3 and 4 4 stu tudie ies • Phase 3 • Two/3 arm study (active arm (add on ) versus control arm (UDCA)) • intermediate endpoint • Phase 4 confirmatory study • two arm study: active versus control • true endpoint = liver transplantation or death, decompensation, MELD>14 • Power calculation – min 8-15 years follow-up n>500 patients – event driven
Des esign p phase 4 4 confir firmatory s stu tudy u using a a (historical) m matched con ontrol a l arm phase 2 phase 2 long-term follow-up phase 3 phase 3 long-term follow-up long-term follow-up study of Standard of Care (SOC)
Des esign P Phase 4 4 con onfir firmatory s stu tudy u using a a match ched con ontrol a l arm Pros • reuse of gained knowledge • reuse of data = recycling data and knowledge • reduction of study-time New • to clinical endpoint SOC • to assess benefit or harm • to approval for the patients Cons • selection bias • heterogeneity • quality bias
Des esign p phase 4 4 confir firmatory s stu tudy u using a a match ched con ontrol a l arm Consider if disease is rare and/or chronic = clinical endpoint is far away How to solve Cons • selection bias Use incl/excl criteria • heterogeneity New SOC use weights (IPTW) to stabilize differences • quality bias minimize bias, install quality control
An e example Phase 3 SOC
Selecti ection on In the control cohort apply Selection criteria phase 3 • ALP>1.67xULN or bilirubin>1xULN • Bilirubin<threshold xULN • UDCA min 12 months or untreated Phase 3 • Of all visits fulfilling above first visit selected SOC • diagnosed after 1990 to control for • population differences • UDCA dosage differences • Changes in treatment of decompensation • Listing for liver transplantation
Comparison on P Phase e 3 an and Sele elect ctio ion Phase 3 Selection p n=137 n=361 Sex %Female 83.9% 92.5% 0.007 Age, yr (mean,SD) 58.8 (11.9) 54.9 (12.2) 0.002 UDCA % 97.8% 94.2% 0.10 Duration UDCA(yr) 3.6 (3.4) 3.9 (3.7) 0.001 (mean, SD)
Comparis ison P Phase 3 3 and S Sele elect ctio ion p=0.18 bilirubin 10 p=0.44 ALP albumin p=0.001 9 AST p=0.001 ALT 8 plat/100 p=0.02 GGT p=0.39 7 GLOBE SCORE p=0.001 6 5 4 3 2 1 0 -1 Phase 3 Control
Comparis ison P Phase 3 3 and G Global S l Sele elect ctio ion IPTW weig eighted a analy lysis is Phase 2 Global p n=135 n=361 sum of weights sum of weights Sex %Female 90.4% 88.7% 0.75 Age, yr (mean,SD) 55.8 (11.8) 56.4 (12.7) 0.63 UDCA 94.1% 95.2% 0.65 Duration UDCA(yr) 3.8 (3.7) 3.8 (3.6) 0.98 (mean, SD) New SOC
Comparis ison P Phase 2 2 and G Global S l Sele elect ctio ion IPTW w weighte ted a analys ysis p=0.63 bilirubin p=0.31 10 ALP p=0.70 albumin 9 p=0.25 AST 8 ALT p=0.75 plat/100 p=0.73 7 GGT p=0.51 GLOBE SCORE 6 p=0.66 5 4 3 2 1 0 -1 control with IPTW phase 3 with IPTW
Des esign p phase 4 4 confir firmatory s stu tudy u using a a match ched con ontrol a l arm Consider if disease is rare and/or chronic = clinical endpoint is far away How to solve Cons selection bias use incl/excl criteria New SOC heterogeneity use weights to stabilize differences • quality bias minimize bias, install quality control
Qu Quality ity c contr trol • SOP which includes: • Site visits: at site data inspection/capture • REDCAP data collection – safe tracking and storing • Queries automatically generated • Lab-test provided with units and Upper/Lower Limit of Normal • All clinical endpoints (decompensation, HCC, liver transplantation, death and cause of death) reassessed by board of experts Inclusion of other SOC-databases: • Prospective data collection in parallel with phase 3
Comments a and d discu cussio ion • In case of rare/chronic disease reuse/recycle of historical database is feasible • Selection bias can be avoided • Heterogeneity can be avoided with use of IPTW weights to mimic a RCT • Quality control rules must be applied and standardized • Consider prospective SOC/registry cohort to run in parallel
Recommend
More recommend