METHODOLOGICAL CHARACTERISTICS OF PIVOTAL DATA SUPPORTING ORPHAN MEDICINAL PRODUCTS AUTHORIZED BY THE EMA IN THE FIRST 15 YEARS OF THE ORPHAN REGULATION

Objective:We reviewed European Public Assessment Reports(EPAR) for Orphan Medicinal Products (OMP) to describe the amountand type of information supporting positive opinions for orphan drugs(OD) issued by EMA.Methods:EPARs for first 100 orphan drugs (OD) in Europe (January2000–December 2014) have been reviewed to retrieve information onthe drug, therapeutic indication, prevalence of the indication, numberof trials supporting each indication and main trial characteristics. Addi-tional public information sources were used: OD designations, SMERegistry, clinicaltrials.gov, Pubmed, Integrity Database and Orphanet.This was part of the project“Advances in Small Trials dEsign forRegulatory Innovation and eXcellence, ASTERIX”(FP7 grant Health-F5-2013-603160).Results:125 new OMP marketing applications had positive opinionsissued by the EMA, related to 100 products (76% small molecules,23% biological drugs and 1% advanced therapies) and 86 differentmedical conditions (41.6% oncology, 28.8% endocrine and metabolicdisorders), of which only 16 were ultrarare (prevalence<0.1/10,000).Most regulatory decisions (88%) were based on clinical trial (CT)results, while 12% were approved using non-CT evidence. Overall 159CTs were included in the dossiers (Mean per dossier 1.44; range 1–5).Up to 65% were parallel CTs and 30% single arm trials. Only half ofthe trials were blinded, and 47.2% used some type of placebo. Themain endpoint was an intermediate variable in 75% of CTs, and onlyin 8% of cases included patient reported outcomes (PRO). The mainoutcome was not met in up to 12% of pivotal trials. Mean (interquar-tile range) sample size was 154 (62-312) for conditions with preva-lence over 0.1/10.000, and 35 (17–55) for ultrarare conditions.Conclusions:Most of orphan drug approvals are based on clinical tri-als with conventional designs using intermediate outcomes as primaryendpoints. Alternative designs proposed as more suitable for smallsample sizes are scarcely used.