Preplanned Studies: Longitudinal Trajectories of Adiposity Indicators and Cancer Risk Over 14 Years: Evidence from Repeated Health Check-Ups of 10 Times or More — China, 2010–2023

Introduction: Obesity, particularly central adiposity, has been associated with elevated cancer risk. However, longitudinal data on adiposity trajectories and cancer incidence in Chinese populations remain limited.

Methods: We analyzed data from 25,653 adults with ≥10 health check-ups in the West China Hospital Alliance Longitudinal Epidemiology Wellness (WHALE) Study (2010–2023). Five adiposity indicators — body mass index (BMI), waist circumference (WC), waist-to-hip ratio (WHR), BMI-adjusted WC (WCadjBMI), and BMI-adjusted WHR (WHRadjBMI) — were evaluated using Poisson regression and generalized linear mixed-effects models. Latent class mixed modeling identified long-term adiposity trajectories. Analyses were stratified by sex and age (≥50 years).

Results: Over 14 years, 393 participants developed cancer. Higher BMI [risk ratio (RR)=0.873, P=0.019] was associated with lower cancer risk, whereas central adiposity indicators (e.g., WCadjBMI, RR=1.175, P=0.001) showed positive associations, particularly among men and those aged ≥50 years. WCadjBMI was significantly associated with lung cancer (RR=1.246, P=0.009), with similar trends for breast and liver cancers. Inverted U-shaped trajectories of BMI-adjusted waist measures were linked to elevated cancer risk, highlighting the relevance of long-term fat distribution.

Conclusions: Central adiposity and its trajectories are associated with cancer risk in Chinese adults, supporting dynamic obesity monitoring and targeted prevention in older adults and men.

Obesity is now recognized as a major modifiable cancer risk factor. Globally, an estimated 4%–8% of all cancer cases are attributable to elevated body mass index (BMI), with higher proportions in high-income countries. Prospective cohorts in Europe and North America have shown that obesity increases the risk of colorectal, breast, kidney, and liver cancers by 1.5 to 2 times (1). Beyond energy storage, adipose tissue acts as an endocrine organ that can promote carcinogenesis through chronic inflammation, insulin resistance, hormonal imbalances, and altered adipokine secretion (2). However, most studies rely on a single baseline adiposity measure, which does not reflect long-term exposure or changing fat distribution.

In China, both obesity and cancer burdens have increased sharply in recent decades. National survey data show that adult obesity has surpassed 16% (3), while cancer incidence reached 341.75 per 100,000 in 2022 (4). Rapid increases have been noted in cancers such as lung, colorectal, and thyroid, with accumulating evidence linking them to central obesity (5). Nevertheless, most research on adiposity and cancer is based on Western cohorts and may not fully represent the metabolic profiles or fat distribution patterns of Chinese adults. Furthermore, limited Chinese studies have leveraged repeated health checkup data to assess how long-term changes in adiposity influence cancer risk.

To address this evidence gap and support the “Healthy China 2030” strategy, which emphasizes proactive health management, this study used data from the West China Hospital Alliance Longitudinal Epidemiology Wellness (WHALE) Study. The study integrates routine health check-up records from multiple centers. By analyzing more than 25,000 adults with at least 10 check-ups, we investigated how general and central adiposity indicators and their long-term trajectories relate to cancer incidence. This study offers novel evidence on the predictive value of dynamic fat distribution, which may improve early risk detection and inform population-based cancer prevention strategies in China.

Data were obtained from the WHALE Study (2010–2023), which is based on periodic health check-ups conducted at the Health Management Center of West China Hospital and its four affiliated subcenters (6). After excluding individuals with fewer than 10 visits, missing BMI data, or insufficient pre-diagnosis records, 25,653 adults were included in the analysis. Cancer cases were identified through hospital-confirmed diagnoses during follow-up visits. Obesity-related indicators included BMI, waist circumference (WC), and waist-to-hip ratio (WHR), measured using standardized protocols. Residualized variables [BMI-adjusted waist circumference (WCadjBMI) and BMI-adjusted waist-to-hip ratio (WHRadjBMI)] were constructed to minimize collinearity. Repeated measures were handled using an Andersen-Gill structure, and Poisson regression was used to estimate risk ratio (RR) for incident cancer. Generalized linear mixed models with random intercepts accounted for within-individual correlation. Analyses were stratified by sex and baseline age. Latent class mixed modeling was applied to identify long-term obesity trajectories, which were then incorporated into regression models (Figure 1). All statistical analyses were conducted using R (version 4.2.3, R Foundation for Statistical Computing, Vienna, Austria). Detailed analytic procedures are provided in the Supplementary Materials.

Figure 1. 

Flowchart of study population selection from the WHALE Study, 2010–2023.

Note: A stepwise illustration of participant inclusion and exclusion criteria is used to derive the final analytic sample (n=25,653) from the original WHALE Study (n=685,163), based on repeated health check-up data and cancer diagnosis status.

Abbreviation: BMI=body mass index; WHALE=West China Hospital Alliance Longitudinal Epidemiology Wellness.

A total of 25,653 adults with at least 10 health check-up records were included in the final sample. The median age was 40 years, and 53.0% were male. During follow-up, 393 participants (1.5%) were diagnosed with malignant tumors. Compared with cancer-free participants, those who developed cancer were older (median age: 46 vs. 40 years, P<0.001) and had higher rates of hypertension (27.5% vs. 19.5%, P<0.001) and diabetes (10.9% vs. 6.6%, P=0.001) (Table 1). The distribution of cancer types is presented in Supplementary Table S1.

Poisson regression analysis showed distinct differences in how adiposity indicators were associated with cancer risk (Figure 2A and Supplementary Table S2). In the overall sample, higher BMI was consistently linked to lower cancer risk, even after full adjustment for demographic, behavioral, and metabolic factors [risk ratio (RR)=0.873; 95% confidence interval (CI): 0.778, 0.977; P=0.019]. By contrast, WCadjBMI was positively associated with cancer risk (RR=1.175; 95% CI: 1.071, 1.289; P=0.001), suggesting an independent contribution of central adiposity to cancer development.

Figure 2. 

Longitudinal trajectories of adiposity indicators and their associations with cancer risk in the WHALE Study (2010–2023). (A) Forest plot of RRs and 95% CIs for the associations between adiposity indicators and cancer risk across the total population and subgroups (by sex and age group); (B) Trajectories of adiposity indicators over time prior to cancer diagnosis, estimated by latent class mixed modeling: (B) WCadjBMI in participants aged ≥50 years; (C) WHRadjBMI in the overall population; (D) WHRadjBMI in males.

Note: (A), Model 1: adjusted for baseline age and sex (sex not adjusted in sex-stratified models); Model 2: additionally adjusted for smoking status and drinking habits; Model 3: additionally adjusted for hypertension and diabetes. (B), in participants aged ≥50 years, an inverted U-shaped WCadjBMI trajectory remained significantly associated with elevated cancer risk in the fully adjusted model (Model 3: RR=1.776, 95% CI: 0.973, 2.991, P=0.043). In the overall population, an inverted U-shaped WHRadjBMI trajectory was significantly associated with increased cancer risk in the unadjusted model (RR=1.448, 95% CI: 1.058, 1.937, P=0.016). Among males, the same trajectory showed a similar association (RR=1.862, 95% CI: 1.066, 3.066, P=0.020). No significant associations were observed for BMI trajectories in any subgroup. Detailed trajectory class distributions, model fit indices, and regression estimates are presented in Supplementary Tables S4–S5 and Supplementary Figure S1.

Abbreviation: CI=confidence interval; RRs=risk ratios; BMI=body mass index; WCadjBMI=Waist circumference adjusted for age, sex, and BMI; WHRadjBMI=Waist-to-hip ratio adjusted for age, sex, and BMI; WHALE=West China Hospital Alliance Longitudinal Epidemiology Wellness.

In sex-stratified analyses, both WCadjBMI and WC remained significantly associated with increased cancer risk in men (WCadjBMI RR=1.353; 95% CI: 1.194, 1.531; P<0.001; and WC RR=1.182, 95% CI: 1.031, 1.352, P=0.016), while BMI showed no significant association. Among women, WHR displayed a significant negative association with cancer risk (RR=0.830; 95% CI: 0.704, 0.975; P=0.025), while no other indicators were significant.

Age-stratified analyses showed that in participants aged ≥50 years, WHRadjBMI (RR=1.193; 95% CI: 1.016, 1.395; P=0.030) and WCadjBMI (RR=1.392; 95% CI: 1.189, 1.626; P<0.001) were both associated with higher cancer risk. For those aged <50 years, WHR remained the only significant indicator (RR=0.837; 95% CI: 0.705, 0.992; P=0.042), showing an inverse association. Findings were consistent across mixed-effects models, supporting the robustness of these results (Supplementary Table S2).

WCadjBMI was significantly associated with increased lung cancer risk and showed positive trends for other cancer types, in contrast to the inverse relationship observed for BMI (Supplementary Table S3).

Trajectory analysis revealed that an inverted U-shaped pattern of WHRadjBMI was significantly associated with greater cancer risk in the overall population (unadjusted RR=1.448; 95% CI: 1.058, 1.937; P=0.016) and in men (unadjusted RR=1.862, 95% CI: 1.066, 3.066; P=0.020). Among individuals aged ≥50 years, an inverted U-shaped WCadjBMI trajectory remained significantly associated with higher cancer risk after full adjustment (RR=1.776; 95% CI: 0.973, 2.991; P=0.043) (Figure 2B and Supplementary Table S4).

This longitudinal cohort study of more than 25,000 Chinese adults with at least 10 repeated check-ups over 14 years provides robust evidence that central adiposity indicators, particularly WC and WCadjBMI, are independently associated with higher cancer risk. Stratified analyses showed stronger predictive value of central adiposity measures among men and individuals aged ≥50 years. Most notably, trajectory modeling revealed that participants following an inverted U-shaped trajectory of WHRadjBMI or WCadjBMI had significantly greater cancer risk, suggesting that long-term adiposity dynamics may provide an early opportunity for cancer risk detection. The inverse BMI–cancer association may partly reflect reverse causation from pre-diagnostic weight loss (7). These findings have direct practical public health implications, supporting routine monitoring of central adiposity (thereby outperforming BMI in cancer prediction) as part of chronic disease prevention strategies in China.

Our results are consistent with both international and domestic studies which highlight the role of central adiposity in cancer development. A large UK Biobank analysis reported that waist-based indicators were more strongly associated with obesity-related cancers than BMI, with notable sex-specific differences (8). Similar findings have been observed in Chinese cohorts, including the MJ Health and Guangzhou Biobank studies, where waist-based measures outperformed BMI in predicting cancer risk, especially among women and for colorectal cancer (9-10). The present study reinforces these findings and emphasizes the relevance of fat distribution indicators, especially for men and older adults. The stronger link between central adiposity and cancer is likely mediated by visceral fat–driven inflammation and hormonal disruption (8,11). Additionally, the inverted U-shaped trajectories of WHRadjBMI and WCadjBMI may indicate early subclinical processes such as cachexia or systemic inflammation preceding a cancer diagnosis (12).

This study has several strengths, including its large and well-characterized sample, repeated anthropometric measurements over a long follow-up period, and the combined use of trajectory modeling and traditional regression approaches to better capture cancer risk dynamics. The use of residualized adiposity indicators allowed clearer differentiation between general and central adiposity effects.

The findings in this report are subject to four main limitations. First, the relatively small number of cancer cases may have limited statistical power, particularly in subgroup and trajectory analyses. Second, excluding individuals with fewer than 10 follow-up visits or pre-existing cancer may have introduced survivorship bias, potentially underestimating cancer incidence in higher-risk individuals who were lost to follow-up or died earlier. Third, residual confounding influences from unmeasured lifestyle factors, such as diet and physical activity, cannot be ruled out. Finally, cancer type-specific analyses were constrained by sample size and should be addressed in future large-scale studies.

Given China’s ongoing implementation of the “Healthy China 2030” initiative (6), which prioritizes personalized and proactive chronic disease prevention, these findings provide timely and actionable evidence to support targeted cancer screening and interventions addressing abdominal obesity. Future research should aim to incorporate more comprehensive lifestyle data, extend follow-up duration, and expand sample sizes to enhance subgroup analyses and clarify site-specific cancer associations, further strengthening the evidence base for targeted public health interventions in China.

All colleagues and staff members who contributed to the data collection for this study.