Ultra-processed foods and disease-free survival after colorectal cancer diagnosis: Findings from the Colocare Study

Introduction: Ultra-processed foods (UPFs), industrially formulated products high in fat, sugar, and additives, account for ~60% of daily calories in the United States. UPF intake has been linked to obesity, diabetes, and risk of some cancers, yet its impact on colorectal cancer (CRC) survivorship remains unclear. CRC survivors often face treatment-related dietary challenges and may rely on UPFs for convenience or to meet energy needs. Methods: We analyzed 418 stage I-III CRC patients in the ColoCare Study across four U.S. sites. Baseline demographic, lifestyle, and clinical data were collected. Dietary intake was assessed 6 months post-diagnosis using a food frequency questionnaire (FFQ) reflecting the prior six months. Foods were classified into one of four NOVA groups: unprocessed, minimally processed, processed, and ultra-processed foods (UPFs). We focused on UPF intake (NOVA Group 4), categorized into quartiles (Quartile 1 = lowest; Quartile 4 = highest) using two metrics: absolute intake (grams/day) and relative intake (percent of total grams/day). Cox proportional models estimated hazard ratios (HRs) and 95% confidence intervals (CIs) for disease-free-survival (DFS), adjusting for study site, age, sex, tumor site adjuvant treatment, BMI, and total caloric intake (kcal). Sensitivity analyses without BMI/kcal adjustment were also done. Results: Participants had a mean age of 58 years (standard deviation [SD]=13y) and a mean BMI of 29 kg/m² (SD=6.8). 50% of participants were female, 49% were diagnosed with stage III CRC, and 59% with colon cancer. The median follow-up was 4 years (SD=2y). UPFs comprised ~16% of the total daily food intake by weight. The mean Healthy Eating Index (HEI) score was 63 (SD=10), but scores decreased across quartiles of UPF consumption (p < 0.001). When evaluated by g/day, BMI differed by UPF quartiles (p=0.04), and a similar pattern was observed when quartiles of % g/day were used (p=0.12). UPF consumption, whether in g/day or percent of total intake, was not significantly associated with DFS. For g/day quartiles, HRs (95% CI) for quartiles 2-4 vs 1 were 0.92 (0.52-1.65), 0.88 (0.47-1.62), and 0.88 (0.45-1.74). For percentage-based quartiles, HRs were 1.12 (0.61- 2.09), 1.02 (0.54-1.93), and 1.27 (0.70-2.30). Results were consistent in both unadjusted models and those without BMI/kcal adjustment. Discussion: In this large, multi-center cohort, UPF intake was common among CRC survivors and associated with lower diet quality. However, UPF consumption was not associated with DFS, regardless of whether intake was measured in g/day or percentage of total intake. Modest differences in HR direction likely reflect how absolute versus relative intake captures dietary patterns. These findings offer important insights into post-diagnosis diet and cancer outcomes.