Perioperative physiological complexity trajectories and their association with molecular residual disease and long-term outcomes after curative colorectal cancer surgery: a prospective cohort study.

BACKGROUND: Postoperative recovery and long-term outcomes after curative colorectal cancer surgery remain heterogeneous. Although circulating tumor DNA-based minimal residual disease (ctDNA-MRD) provides tumor-specific prognostic information, dynamic host-recovery processes are not captured. We evaluated whether wearable-derived perioperative physiological complexity trajectories, operationalized as a multidomain host-recovery phenotype, provide prognostic information complementary to clinicopathologic variables and ctDNA-MRD.

METHODS: In this prospective cohort study, 1000 patients undergoing curative-intent colorectal cancer surgery were followed for 60 months. Continuous wearable-derived data from the preoperative period through postoperative day 30 were used to construct a composite physiological complexity score (PCS). PCS integrated cardiac autonomic complexity, rest-activity transition organization, and sleep-wake transition structure. Components were direction-aligned, standardized, and combined using prespecified equal weighting without outcome-derived optimization. Trajectory modeling identified perioperative recovery patterns. Associations with postoperative complications, ctDNA-MRD persistence, disease-free survival (DFS), and overall survival (OS) were evaluated using multivariable models, nonlinear analyses, calibration assessment, decision curve analysis, and sensitivity analyses.

RESULTS: Three PCS trajectories were identified: preserved complexity (n = 403), transient complexity loss (n = 372), and persistent complexity suppression (n = 225). Persistent complexity suppression was associated with higher major complication rates (18.2% vs 10.9%, P = 0.006) and increased ctDNA-MRD persistence (adjusted OR 3.02, 95% CI 2.10-4.35; P < 0.001). Over 60 months, persistent complexity suppression was associated with worse DFS (adjusted HR 1.89, 95% CI 1.45-2.47) and OS (adjusted HR 1.76, 95% CI 1.29-2.41). Findings were directionally consistent across sensitivity analyses. Adding PCS improved exploratory prognostic calibration and net clinical benefit.

CONCLUSION: Wearable-derived PCS trajectories were associated with postoperative morbidity, ctDNA-MRD persistence, and long-term outcomes after colorectal cancer surgery. PCS represents an investigational host-recovery phenotype, not a direct immune or tumor-biology measure. These observational findings are hypothesis-generating and do not establish causality. External validation, threshold calibration, device harmonization, and prospective clinical utility studies are required before clinical implementation.