Association between obesity and pancreatic cancer in French women from the E3N cohort study
BOUTEILLE L. 1, LAOUALI N. 2, GELOT A. 1, ARTAUD F. 1, DESCOURVIÈRES C. 3,4, REBOURS V. 3,4, ELBAZ A. 1, SEVERI G. 1,5, TRUONG T. 1
1 Université Paris-Saclay, UVSQ, Inserm, Gustave Roussy, CESP, Villejuif, France; 2 Faculty of Medical Sciences, UM6P Hospitals, University Mohammed VI Polytechnic, Ben Guerir, Morocco; 3 Pancreatology and Digestive Oncology Department, Beaujon Hospital, Clichy, France; 4 Université Paris-Cité, Paris, France; 5 Department of Statistics, Computer Science, Applications “G. Parenti”, University of Florence, Florence, Italy
Background
Obesity has been identified as a modifiable risk factor for pancreatic cancer (PC). However, interpretation of this association is complicated by the long latency of PC and the occurrence of weight loss years before diagnosis, raising concerns about reverse causation. Evaluating time-varying adiposity measures and pre-diagnostic weight trajectories is therefore essential to address this issue. Objectives
To investigate the association between obesity and PC risk in French women, explicitly accounting for time-varying adiposity indicators and pre-diagnostic body weight trajectories. Methods
We analyzed data from 98,995 women enrolled in the French E3N cohort in 1990, among whom 291 incident PC cases were identified during follow-up until 2014. Cox proportional hazards models adjusted for potential confounders were used to assess associations between PC incidence and time-varying body mass index (BMI), waist circumference (WC), waist-to-height ratio (WHtR), and waist-to-hip ratio (WHR), applying different exposure lag periods. In addition, BMI trajectories were examined in a nested case–control study using multinomial generalized estimating equation (GEE) models. Results
Obesity (BMI ≥ 30 kg/m²) was associated with an increased risk of PC, with the risk estimate peaking at a five-year exposure lag (HR = 1.87, 95% CI: 1.18–2.96), whereas no clear association was observed with longer lag periods. Higher WC and WHtR were also significantly associated with increased PC risk (HR = 1.50, 95% CI: 1.09–2.06 and HR = 1.69, 95% CI: 1.08–2.63, respectively), while WHR was not associated with PC. Trajectory analyses showed similar obesity prevalence among cases and controls until approximately twelve years before diagnosis, after which trajectories diverged, with a higher prevalence of obesity among cases. This difference subsequently narrowed as diagnosis approached, consistent with pre-diagnostic weight loss. Conclusions / Implications for practice or policy
These findings underline the importance of considering the temporal dynamics of body weight when studying PC risk. They also support the use of alternative adiposity measures beyond BMI to better capture obesity-related risk and inform prevention strategies.