Abstract
BACKGROUND: In the 12-month, randomized, double-blind, placebo-controlled Phase 2/3 NPC-002 study (NCT02612129), arimoclomol significantly reduced annual disease progression versus placebo, measured by the 5-domain NPC Clinical Severity Scale (5DNPCCSS). Arimoclomol has been approved in the US for treatment of Niemann-Pick disease type C (NPC) in combination with miglustat. This paper introduces the rescored 4-domain NPCCSS (R4DNPCCSS) as a post-hoc primary endpoint in NPC-002, discusses its validation, and presents the results of the post-hoc primary analysis. METHODS: To more accurately assess changes in disease course over a 12-month time period in a heterogeneous group of patients, the Cognition domain was removed from the 5DNPCCSS and the Swallow domain was rescored to reflect linearity in disease progression. Rescoring of the Swallow domain was based on input from clinical NPC and swallow experts from a qualitative interview-based study (N = 12), resulting in the R4DNPCCSS. To supplement prior validation analyses, data supporting the overall validity and reliability of the R4DNPCCSS was gathered through additional analyses of construct and convergent validity. The NPC-002 prespecified primary efficacy endpoint analysis based on the 5DNPCCSS score change from baseline to 12 months was repeated with R4DNPCCSS. RESULTS: Construct validity analysis demonstrated high agreement between the R4DNPCCSS domain scores and the Clinical Global Impression Scale of Severity (CGI-S) and NPC Clinical Database (NPC-cdb) scores. Convergent validity was confirmed by strong correlations between the R4DNPCCSS domains and corresponding items on the Scale for Assessment and Rating of Ataxia (SARA), 9-hole peg test (9-HPT), and Video Fluoroscopic Swallowing Study (VFSS) performance tests. The NPC-002 post-hoc primary analysis showed a mean standard error (SE) change in R4DNPCCSS score of 0.35 (0.40) with arimoclomol (N = 34) versus 2.05 (0.54) with placebo (N = 16), and a treatment effect in favor of arimoclomol over placebo of -1.70 (p = 0.0155). In the miglustat subgroup analysis, mean (SE) change in R4DNPCCSS score was -0.23 (1.02) with arimoclomol (N = 22) versus 1.92 (3.37) with placebo (N = 12), representing a treatment effect of -2.21 (p = 0.0077). CONCLUSION: The R4DNPCCSS is a valid and reliable measure of disease progression demonstrating consistent outcomes with the prespecified 5DNPCCSS endpoint. Arimoclomol significantly slowed disease progression through 12 months as measured by the R4DNPCCSS versus placebo.