143: A Meta-Analysis to Understand the Correlation Between Weight Loss & Resolution of Nonalcoholic Steatohepatitis

Statement of Purpose: Nonalcoholic steatohepatitis (NASH) is a disease characterized by accumulation of fat in the liver accompanied by lobular inflammation and hepatocellular injury. NASH is believed to affect approximately 5% of the United States population. Currently, there is no pharmaceutical intervention approved for the treatment of NASH. Therefore, patients are prescribed lifestyle modifications to reduce body weight to reverse and/or prevent further liver damage. The current analysis was performed to describe the relationship between weight loss and resolution of NASH in the clinical setting. In the absence of direct evidence, published data from randomized control trials and observational studies was used to support the analysis.

Description of Methods & Materials: A search of published data was conducted to identify reports of resolution of NASH following changes in body weight resulting from intervention with diet or exercise, bariatric surgery or pharmacologic agents that promote weight loss. Articles were selected using current PRISMA guidelines. After selection of articles, reports were standardized to a common definition of weight loss (percent change from baseline) and resolution of NASH as assessed by liver histology. Resolution of NASH was expressed as the proportion of responders with complete NASH resolution and transformed to the logit scale. Data was weighted using the variance of the group and analyzed using the metafor and nlme packages in R v3.6.1. Covariates of interest included baseline body weight and intervention type, where interventions were categorized as “diet and exercise,” “bariatric surgery” or “pharmaceutical intervention.”

Data & Results: A total of 81 manuscripts, representing 79 studies were identified during the initial literature search. After standardizing endpoints, 11 papers with 17 representative study arms were included in the analysis. The most common reason for exclusion of records was inability to pool results due to reporting differences. The correlation between weight loss and NASH resolution is shown in Figure 1, where the size of circles represents study weight based on variance of the group and the blue line and grey shaded area represent the mean and 95% confidence intervals of the regression line. Although there was substantial heterogeneity across studies, accounting for weight loss was shown to statistically reduce variability around reports of NASH resolution (tau^2: 0.20 and 0.057, with and without accounting for weight loss, respectively; p value: 0.001). There was insufficient data to differentiate between interventions and NASH resolution. Baseline body weight was not shown to statistically influence resolution of NASH; however, baseline weight was maintained in the final model to address differences in baseline body weights across interventions.

Interpretation, Conclusion or Significance: There was a significant inverse relationship between percent change from baseline bodyweight and resolution of NASH. Neither intervention nor baseline weight were shown to significantly alter the correlation between weight loss and NASH resolution; however, caution should be used during interpretation of these data due to the small sample size for each separate intervention and the truncated range of baseline weight data.

Disclosures: All authors (Tina Checchio, Melissa O'Gorman, Sima Ahadieh, William Esler, Amanda Darekar, and Kevin Sweeney) are employees and shareholders of Pfizer Inc.