Jae Eun Ahn (1), Sima Ahadieh (1), Kevin Sweeney (1)
(1) Global Pharmacometrics, Pfizer, Inc.
Objectives:
Non-Alcoholic Fatty Liver Disease (NAFLD) is defined by excessive hepatic fat accumulation (steatosis, defined as a liver fat content greater than 5%) in the absence of significant alcohol consumption, autoimmune, viral, metabolic disorders, or drug induced liver disease [1]. Non-Alcoholic Steatohepatitis (NASH) is a clinical and histological subset of NAFLD that is associated with increased all-cause mortality, cirrhosis and end-stage liver disease, increased cardiovascular mortality, and increased incidence of both liver related and non-liver related cancers [2]. Currently, there are no medicines approved to treat NAFLD and NASH and clinical trials for several therapeutic agents in NASH patients are ongoing. Through model-based meta-analysis (MBMA), the following objectives were identified: 1) to understand the relationship between changes in plasma biomarkers and biopsy related endpoints, and 2) to quantify the effects of different therapeutic agents providing quantitative criteria to inform NASH clinical development strategies.
Methods:
A literature search was performed according to the following PICOS criteria [3]. Study populations (P) were NAFL and NASH patients with or without type 2 diabetes mellitus, fibrosis, and metabolic syndrome. Interventions (I) included but were not limited to pioglitazone, metformin, exercise, vitamin E, diet, lifestyle modification, ursodeoxycholic acid, liraglutide, pentoxifylline, bariatric surgery, and rosiglitazone as well as new investigational drugs. Placebo or standard of care (typically diet and exercise) served as comparators (C). The outcomes (O) of interest included circulating biomarkers such as alanine aminotransferase (ALT), aspartate aminotransferase, gamma glutamyl transferase, and alkaline phosphatase, and also liver fat, fibrosis grading, NAFLD Activity Score (NAS), and responders. Either placebo or active controlled studies (S) were included. The search databases included OVID MEDLINE® 1946-07Sep2018, OVID MEDLINE® In-process and Epub Ahead of Print, BIOSIS Previews 1969 to 2018 Week 40, Embase 1974-06Sep2018, Scopus®, and ClinicalTrials.gov (https://clinicaltrials.gov/). R (version 3.4.1) was used for data cleaning, modification, and plotting. NONMEM (version 7.4, ICON Development Solution, Gaithersburg, MD) was used for estimation. For biomarkers, the following structural model was applied for % change from baseline measures (Y), as reported or imputed using the reported baseline and change from baseline measures:
Yijk = -Pmax∙(1-exp(-0.693/ET50∙timek)-Trth+ηi+ηij+εijk
where Yijk denotes observation at ith study, jth arm, and kth time; Pmax, maximum placebo or control response; ET50, the time for reaching a half Pmax; Trth, treatment effects specific for hth drug class; ηi, inter-study random effect; ηij, inter-arm level random effect, nested within the study; εijk, residual variability, weighted according to the sample size of arm.
Results:
Following the literature search and critical review, 198 studies were included in the final NASH MBMA data set among which 173 publications reported ALT, and 22 studies reported liver fat measured by MRI. Biopsy endpoints such as NAS (either total or sub scores) and fibrosis were reported in 45 and 49 studies, respectively. A longitudinal model for ALT estimated a Pmax of 23.7 % (RSE 5.63 %) with ET50 of 12.4 weeks (11.8%) could be achieved by the control. Trth were characterized, and estimated to vary between an additional 3.85% (RSE 55.5 %) and 19.4% (RSE 9.24%) reduction relative to control. Liver fat data did not show any parametric trend in the control groups; however, overall changes varied between a 56% reduction to a 7% increase. The relationships between liver fat content and NAS or liver fat and fibrosis remain unclear, as the quantity of data informing this relationship is sparse. Responder definitions were heterogeneous among different studies and further characterization and grouping are ongoing.
Conclusions: Literature data were constructed for NAFLD/NASH disease area and the longitudinal MBMA was performed. While the preliminary modelling results provide some quantitative criteria to benchmark/compare the treatment effect of interest versus other agents, further discussion and additional data would be necessary to properly identify the relationship between the biomarkers and clinical endpoints from the literature data.
References:
[1] Vajro et al, JPGN 54:700-713 (2012)
[2] Sanyal et al, Hepatology 61(4): 1392-405 (2015)
[3] Liberati et al, PLoS Med 6: e1000100 (2009)
Reference: PAGE 28 (2019) Abstr 8818 [www.page-meeting.org/?abstract=8818]
Poster: Drug/Disease Modelling - Endocrine