S727 Machine Learning-Assisted Gene Risk Score for Calories to Satiation Predicts Weight Loss Outcomes After Sleeve Gastrectomy

Introduction: Sleeve gastrectomy (SG) is among the most performed bariatric procedures. Recently, we developed a genetic risk score to predict calories to satiation (CTSGRS) (PMID 38826309). We aimed to evaluate whether CTSGRS status is associated with weight loss after SG. Methods: The Mayo Clinic Biobank is a large-scale biorepository of adult patients with genetic whole-exome data. From the Biobank, we identified 102 patients with a history of SG and genetic information available. We estimated the gene risk score as a continuous value and dichotomized it at 0.5 into High CTSGRS and Low CTSGRS. Our primary endpoint was total body weight loss percentage (%TBWL) at 60 months. Secondary outcomes included %TBWL at 1, 3, 6, 9, 12, 18, 24, 36 and 48 months. We used a repeated measures mixed model (REML) to analyze longitudinal %TBWL. The model included random intercepts for each patient and fixed effects for CTSGRS group (high vs low), age, sex, and baseline weight. We estimated least‐squares means (LSM) and standard errors (SEM) and used Tukey Honest Significant Difference for post-hoc multiple comparisons. Results: Baseline age, sex, and body mass index were similar between High (n = 42) and Low (n = 60) CTSGRS groups; only baseline weight was higher in High CTSGRS (127.7 ± 30.5 vs 116.0 ± 18.6 kg; P = 0.030). In the mixed‐effects model, the CTSGRS × timepoint interaction was highly significant [F(10, 700) = 2.78; P < 0.001], indicating that weight‐loss trajectories differed by CTSGRS group over time. Patients with a High CTSGRS consistently lost more weight at each timepoint. At 60 months, %TBWL was -29.4 ± 2.1% in High CTSGRS vs -22.1 ± 1.3% in Low CTSGRS (Padj = 0.047). Conclusion: Patients with a High CTSGRS demonstrated significantly greater %TBWL up to 5 years postoperatively. A polygenic background, represented by the CTSGRS, might influence weight loss outcomes in sleeve gastrectomy. Incorporating genetic predictions may help tailor follow-up interventions and optimize postoperative outcomes.