Same Food, Different Outcome: How Biology Shapes the Effects of Food on Health (Oral Session 3)

(OR03-04-25) Prediction for Weight Loss and Regain Based on Multi-Omics and Phenotypic Features: Results From a Caloric-Restricted Feeding Trial

Saturday, May 31, 2025

10:36 AM – 10:48 AM ET

Location: S220 B-D

Presenting Author(s)

Lin Li, PhD candidate

PhD Candidate
Tongji Medical College, Huazhong University of Science and Technology, School of Public Health
Wuhan, Hubei, China (People's Republic)

Disclosure(s): No relevant financial relationship(s) with ineligible companies to disclose.

Co-Author(s)

Rui Li, MPH (he/him/his)

PhD candidate
Huazhong University of Science and Technology
Wuhan, Hubei, China (People's Republic)
RL

Ruyi Li, PhD candidate

None
Tongji Medical College, Huazhong University of Science and Technology
wuhan, Hubei, China (People's Republic)
KZ

Kai Zhu, PhD candidate

Tongji Medical College, Huazhong University of Science and Technology
Wuhan, Hubei, China (People's Republic)
ZQ

Zixin Qiu, n/a

School of Public Health, Tongji Medical College, Huazhong University of Science and Technology, Wuhan, China
Wuhan, Hubei, China (People's Republic)
HY

Hancheng Yu

Huazhong University of Science and Technology
AP

An Pan, PhD

Professor
Huazhong University of Science and Technology
Wuhan, Hubei, China (People's Republic)
GL

Gang Liu

Huazhong University of Science and Technology, Tongji Medical College, School of Public Health

Objectives: Predicting individual responses to weight loss and regain during lifestyle interventions may help personalize weight management. We aimed to identify baseline multi-omics and phenotypic predictors and develop prediction models for changes in body weight and composition during weight loss and regain phases of the low-carbohydrate diet and time-restricted eating (LEAN-TIME) trial (ChiCTR2200056363).

Methods: In the post hoc study of the LEAN-TIME trial, which included a 12-week caloric-restricted feeding intervention and a 28-week follow-up, data from 88 overweight or obese adults completing weight loss phase and 79 completing the weight regain phase were used. Baseline diet features, metabolic parameters, fecal metabolites, and gut microbial species were candidate predictors for changes in body weight, fat mass (BFM), and soft lean mass (SLM) during both phases. Multivariable regression and the least absolute shrinkage and selection operator (LASSO) model were used to identify predictors, and prediction models were developed using the weighted sum of LASSO-selected variables.

Results: Compared to prediction models for changes in body weight and composition based on baseline phenotypic features (R² < 0.27, root mean square errors (RMSE): 1.28-2.03 kg), the multi-omics and phenotypic models showed significantly improved performance (P < 0.05). Among prediction models for changes during the weight loss phase, body composition models outperformed the weight model. The R² values were 0.61, 0.54, and 0.48 for BFM, SLM, and weight loss, with RMSE of 1.41, 0.98, and 1.60 kg, respectively. Models for changes during weight regain phase outperformed those during weight loss, with R² of 0.72, 0.77, and 0.67 for weight, BFM, and SLM regain, and RMSE of 1.42, 1.48, and 0.72 kg. Among the shared predictors, higher baseline levels of Ruminococcus callidus were associated with greater SLM but less BFM loss, while Romboutsia timonensis and galactonic acid were linked to less SLM but more BFM regain, independent of weight change (P < 0.05).

Conclusions: Baseline multi-omics and phenotypic features effectively predict changes in body weight and composition, particularly for body composition and during weight regain phase, offering insights into personalized weight management.

Funding Sources: This study was supported by grants from the National Key R&D Program of China.