Hiroyuki Inoue, Kazutaka Yoshihara
Daiichi Sankyo Co., Ltd., Japan
Objectives: Glycogen storage disease type 1 (GSD1) is a rare autosomal recessive disorder characterized by abnormal storage or release of glycogen, which causes severe hypoglycemia during periods of fasting [1]. GSD1 is an area of high unmet medical needs, where the standard of care (SoC) is daily dietary treatment, uncooked cornstarch (UCCS), and no drug therapy is available. In general, the number of patients of rare disease is limited, which makes difficult to conduct a clinical trials with control arm. The difficulty leads to the challenge of early go/no-go decision, and efficient use of external data is desired. Model-based meta-analysis (MBMA) allows to provide a quantitative information from literatures, and this will be helpful to confirm small proof of concept (PoC). Therefore, MBMA was performed so as to quantify the landscape of SoC (UCCS) and other developing drug therapies for GSD1. Fasting duration is time-to-event data represents the time to reach hypoglycemic threshold after taking UCCS, and it was used as an efficacy end point of Phase 1/2 study [2]. The feasibility to apply the MBMA framework to early clinical decision making was also evaluated.
Methods: A systematic literature search was conducted and extracted the UCCS or other drug therapy trial data including individual fasting duration data and/or Kaplan-meier (K-M) curves for fasting duration. If individual fasting duration was available, then a K-M curve was reproduced, else the K-M curve was digitized from the literature. The K-M curves were fitted by parametric proportional hazard models. Gnls and nlme functions provided in R were used to develop the models. Based on the estimates of a final model, the expected fasting duration of each treatment was simulated.
Results: Seven clinical trials, involving 158 patients, were eligible for this analysis. In addition to UCCS, Phase 1/2 study of DTX401, an adeno-associated virus based gene therapy, was included [2]. The extracted median of fasting duration of UCCS and gene therapy were 4.3 and 6.8, respectively. A Weibull model well described the K-M curves demonstrating the fasting duration. The included trials had various thresholds of hypoglycemia, and there was a trend that the higher threshold resulted in the shorter fasting duration, so the effect of threshold was incorporated. Other covariate effects were evaluated exploratory, but no significant covariates were detected. The simulated median fasting duration of UCCS and gene therapy were 5.3 hr (95% CI: 4.2-6.7 hr) and 8.0 hr (5.9-10.7 hr), respectively. Then distributions of fasting duration of 1,000 virtual patients were generated based on the parameter estimates to test the feasibility to assess small PoC. The distribution of UCCS and that of gene therapy were divided, suggesting the effectiveness of gene therapy. That meant, PoC of DTX401 could be evaluated based on the small phase 1/2 study result.
Conclusions: The developed model could well describe the K-M curves extracted from published literatures, and our MBMA framework revealed the landscape of GSD1 treatments. This framework also suggested that it might be utilized for confirming small PoC by comparing to UCCS efficacy derived from literatures. The results will be used in the planning of clinical program.
References:
[1] Ross KM et al, JIMD Rep. 26: 85-90 (2016)
[2] Press release by Ultragenyx (2019) Available from: http://ir.ultragenyx.com/news-releases/news-release-details/ultragenyx-announces-positive-data-phase-12-study-dtx401-gene
Reference: PAGE () Abstr 9313 [www.page-meeting.org/?abstract=9313]
Poster: Drug/Disease Modelling - Endocrine