Measuring the biological embedding of chronic stress: allostatic load algorithms and cancer risk in a prospective cohort

Autores

Daniel Redondo-Sánchez, José María Gálvez Navas, Encarnación González Flores, Maria del Señor López Vélez, Yahya Mahamat-Saleh, Javier Garcia-Pérez, Camino Trobajo-Sanmartín, Sandra M. Colorado-Yohar, Eunate Abilleira, Maria José Sánchez, Dafina Petrova

Abstract

Background: Epidemiological studies on chronic stress and cancer risk have produced inconsistent findings, likely due to challenges in measuring stress exposure. The allostatic load model offers an alternative biomarker-based framework by capturing the biological embedding of chronic stress.

Objectives: To examine the association between allostatic load and cancer risk, compare alternative operationalizations of the allostatic load construct, and assess whether biomarker patterns underlying cancer risk are consistent with the allostatic load model’s premise of multisystem dysregulation.

Methods: We conducted a nested case-control study within the European Prospective Investigation into Cancer and Nutrition (EPIC)-Granada cohort, including 955 incident cancer cases and 955 matched controls. Pre-diagnostic allostatic load was estimated using 16 biomarkers reflecting neuroendocrine, cardiometabolic, renal, hepatic, and immune system function, measured on average 15 years before diagnosis. Four alternative allostatic load indices were calculated using clinical vs. distributional cut-offs and biomarker vs. system-weighting. Conditional logistic regression estimated associations with cancer risk adjusted for known risk factors. Biomarker contributions were further investigated using machine learning and mixture modelling.

Results: Higher allostatic load was associated with increased cancer risk, with better model fit for indices based on clinical cut-offs (OR = 1.30, 95% CI: 1.18-1.42 per 1-unit increase for an equal-system-weighted clinical index), showing a dose-response relationship (OR = 2.24, 95% CI: 1.68-2.99 for Q4 vs. Q1; p-trend < 0.001). Machine learning and mixture analyses supported the role of multisystem physiological dysregulation, with contributions from most biomarkers and particularly strong influence of neuroendocrine and metabolic components.

Conclusions: Findings support allostatic load as a measure relevant to cancer risk and are consistent with the model’s premise that disease risk reflects coordinated physiological dysregulation. An allostatic load index based on pre-specified clinical cut-offs can detect cancer risk years before diagnosis, offering a reproducible option to increase comparability across studies and populations.