Back to 2025 Abstracts

---

Genome-Wide Association Study (GWAS) Using the NIH Precision Medicine Initiative All of Us to Predict 12-Month Weight Loss Following Vertical Sleeve Gastrectomy
*Yu Hou¹, *Daniel Leslie¹, *Jinhua Wang², *Rui Zhang¹, Sayeed Ikramuddin^1
1Surgery, University of Minnesota, Minneapolis, MN; ²Institute for Health Informatics, University of Minnesota, Minneapolis, MN

Background
Weight loss surgery (WLS) remains the most effective management strategy for individuals with a BMI ? 35. The sleeve gastrectomy (SG) is the most commonly performed WLS, achieving a mean of 28% total body weight loss (WL) that is normally distributed, within the first year. Little data exists to inform predictors of optimal weight loss. Here, we leverage the All of Us NIH Precision Medicine Initiative, which integrates comprehensive electronic health records (EHRs) and whole-genome sequencing (WGS) data to enhance our understanding of weight loss outcomes following sleeve gastrectomy (SG). By combining these datasets, we are able to apply a genome-wide association study (GWAS) to identify genetic markers associated with categorical weight loss success.
Methods
In addition to demographics, clinical features included pre-surgery maximum weight, pre-surgery BMI, and the status of diabetes, hypertension, and smoking. Genetic information was analyzed via whole-genome sequencing. Phenotypic classifications were based on postoperative WL outcomes, using 28% (mean WL from our clinical data) as the threshold to classify patients into low-responders and high-responders. Quality control included sex concordance, allele balance (AB > 0.2), relatedness checks, minor allele frequency (MAF > 0.2), Hardy-Weinberg Equilibrium (HWE p-value > 1e-11), and LD pruning (r² = 0.3). Logistic regression analyzed genetic associations, adjusting for sex and ancestry (using PCs). Significant Single Nucleotide Polymorphisms (SNPs) (?log(p-value) ? 3) were selected through p-value distribution analysis. To validate the significant SNPs identified through GWAS, predictive models were developed using clinical and genetic features. Genetic features were constructed for each individual based on their genotype information at the significant SNPs.
Results
We identified 1,127 individuals who underwent SG, of whom 483 had genetic data. Among these, 233 were classified as high-responders and 250 as low-responders, percent female was 86.34% (87.98% vs. 84.80%; p-value: 0.35), mean age 46.24 (42.82 vs. 49.43; < 0.001), mean BMI 47.56 (47.43 vs. 47.59; 0.79), diabetes 24.01% (19.74% vs. 28.00%; 0.04), hypertension 48.24% (40.34% vs. 55.60%; 0.001) and smoking 6.0% (5.6% vs. 6.4%; 0.85). GWAS identified genome-wide significant loci associated with WL outcomes, suggesting genetic markers that differentiate response. Predictive models combining clinical and genetic data outperformed models using clinical data alone (AUROC 0.80 vs. 0.68). Nineteen loci were identified, including those implicated in pathways related to lactose metabolism (LCT), inflammation (LAMA2), tissue remodeling (CDH13), and cellular signaling (RIPK2) (Table 1).
Conclusion
This study provides insights into the genomic factors associated with weight loss outcomes following SG. These findings may help identify candidates most likely to benefit from SG by incorporating a precision medicine approach to obesity management.
GWAS Results

Locus	p-value	SNPs	Gene	High Responder (AF)	Low Responder (AF)
chr2:135798089	1.47 × 10E-5	rs2322659	LCT	0.45	0.56
chr4:105438052	2.20 × 10E-4	rs2713862	PPA2	0.40	0.26
chr5:10255709	6.57 × 10E-4	rs3846599	CCT5	0.41	0.29
chr6:129143875	5.79 × 10E-4	rs7754560	LAMA2	0.30	0.24
chr7:41963902	3.81 × 10E-4	rs2051935	GLI3	0.41	0.30
chr8:89784141	5.28 × 10E-4	rs200269713	RIPK2	0.28	0.35
chr9:2622134	9.44 × 10E-4	rs34881325	VLDLR-AS1	0.24	0.32
chr9:129814386	9.44 × 10E-4	rs13294595	TOR1A	0.45	0.33
chr10:97611972	8.45 × 10E-4	rs933057	HOGA1	0.18	0.25
chr12:98532087	4.91 × 10E-4	rs3213900	TMPO	0.24	0.17
chr14:68513212	5.75 × 10E-4	rs1541390	RAD51B	0.28	0.20
chr15:41822821	7.59 × 10E-5	rs34055060	MAPKBP1	0.41	0.30
chr15:41855441	3.42 × 10E-4	rs2305655	SPTBN5	0.51	0.57
chr15:45253280	9.83 × 10E-4	rs11854484	SLC28A2	0.45	0.35
chr16:83031798	7.92 × 10E-4	rs9888896	CDH13	0.18	0.27
chr18:23533321	2.98 × 10E-5	rs2510344	NPC1	0.44	0.58
chr21:44528188	2.22 × 10E-4	rs2838579	TSPEAR	0.32	0.23
chrX:153930468	1.52 × 10E-4	rs2071129	NAA10	0.41	0.33

AF=allele frequency; Higher numerical value of AF indicates greater allele frequency between response groups.

---

Back to 2025 Abstracts