Anti-inflammatory Diet Before Diagnosis and Survival After Urologic Cancer: Findings from Two Swedish Prospective Cohorts.

BACKGROUND

Dietary patterns with anti-inflammatory properties have been linked to lower risks of several chronic diseases and some cancers, but their impact on survival after cancer is less well understood.

OBJECTIVE

We assessed whether a higher prediagnostic Anti-Inflammatory Diet Index (AIDI) score was associated with all-cause mortality after urologic cancer diagnosis, with urologic cancer-specific mortality evaluated as a secondary end point.

DESIGN SETTING AND

PARTICIPANTS

In the Cohort of Swedish Men and the Swedish Mammography Cohort, incident urologic cancers and deaths were identified through Swedish registers. AIDI was derived from food-frequency questionnaires (1997; 2009 in a subset) and defined as the assessment at least 2 yr before diagnosis, preferentially 2009 when available. Follow-up ran from diagnosis to death, emigration, or December 31, 2020.

OUTCOME MEASUREMENTS AND STATISTICAL ANALYSIS: Cox models estimated hazard ratios (HRs) per 1 standard deviation (SD) higher AIDI, adjusting for demographic factors, lifestyle, comorbidities, and total energy intake. Fine-Gray models assessed urologic cancer-specific mortality while accounting for death from other causes as a competing event.

RESULTS AND LIMITATIONS: Among 7686 patients with valid AIDI, 6609 had complete covariate data, including 4990 patients with prostate cancer. During follow-up, 3281 deaths occurred, including 1435 urologic cancer deaths. Each 1-SD higher AIDI was associated with lower all-cause mortality (HR = 0.91, 95% confidence interval [CI] = 0.88-0.95). Evidence for an association with urologic cancer-specific mortality was weak and borderline in cause-specific Cox models (HR = 0.95, 95% CI = 0.90-1.00) and was attenuated in competing-risk analysis (subdistribution HR = 0.97, 95% CI = 0.92-1.02).

CONCLUSIONS

Higher prediagnostic AIDI was associated with lower all-cause mortality after urologic cancer diagnosis. Whether this reflects a causal effect of diet or residual confounding, healthy-lifestyle clustering, health care engagement bias, or differences in competing noncancer mortality requires further study.