Discriminatory capacity of allostatic load score in colorectal cancer: A pilot case-control study

Colorectal cancer (CRC) represents the leading cause of cancer-related deaths in Puerto Rico. Allostatic load (AL) reflects a cumulative dysregulation across inflammatory, neuroendocrine, and metabolic systems. It may serve as a composite index for distinguishing CRC cases from controls. Unlike individual markers, AL captures multi-system physiological dysregulation observed in CRC patients. This study aimed to assess the discriminatory capacity of a composite AL score derived from inflammatory, neuroendocrine, and metabolic biomarkers for distinguishing CRC patients from healthy controls in a sample of Hispanics individuals living in Puerto Rico (HPR).This case-control study used plasma samples from the Puerto Rico Familial Colorectal Cancer Registry (PURIFICAR) including 18 CRC cases and 18 healthy controls. AL score was constructed using control derived percentile scoring (75th percentile cutoffs, 25th for IGF-1) across 8 markers: IL-1beta, TNF-alpha, IL-8, MCP-1, CRP (inflammatory), cortisol (neuroendocrine), and IGF-1 and BMI (metabolic). Logistic regression models evaluated discriminatory capacity with progressive adjustment for demographic and sociodemographic factors. Area under the curve (AUC) comparisons were performed using DeLong's test. Analyses were performed in Stata 18. Participants had a median age of 48 years (range: 34-65) and 44% were male. CRC cases had significantly lower BMI (median 23.8 vs 28.2, p=0.016), higher educational attainment (50% vs 33%, p=0.31), and lower private insurance coverage (33% vs 56%, p=0.18) than controls. The AL score showed higher median values in cases versus controls. Higher AL scores were associated with increased odds of CRC in unadjusted analysis (OR=1.26, p=0.295), which strengthened slightly after adjusting for age and sex (OR=1.30, p=0.247). When including education and insurance type for the full model, the association remained similar (OR=1.25, p=0.362). Private insurance demonstrated a marginally significant protective association (OR=0.20, p=0.075). Discriminatory capacity was modest for AL alone (AUC=0.619) and remained similar after demographic adjustment (AUC=0.590) but improved with full adjustment (AUC=0.713). DeLong's test indicated non-significant differences between models (Model 1 vs 2: p=0.626; Model 1 vs 3: p=0.329; Model 2 vs 3: p=0.171). This pilot study demonstrated non-significant associations between AL score and CRC, though discriminatory capacity improved with sociodemographic adjustment. The marginally significant protective effect of private insurance suggests healthcare access may influence CRC risk independent of physiological stress burden. These findings support the feasibility of AL measurement in HPR populations and highlight the importance of considering structural determinants of health in CRC risk assessment. Validation in larger samples is warranted.