Association of comorbidity burden with initial treatment allocation in older women with gynecologic cancer: a retrospective cohort study

Abstract

Background Comorbidity is a key determinant of treatment decisions in older cancer patients; however, its role in shaping initial treatment allocation among older women with gynecologic malignancies remains insufficiently characterized. We aimed to evaluate the association between comorbidity burden and treatment selection and to develop a clinically interpretable predictive model. Methods We retrospectively analyzed 972 women aged ≥65 years with newly diagnosed cervical, ovarian, or endometrial cancer treated at a tertiary hospital in Southwest China between 2019 and 2024. Disease-specific guideline-concordant standard treatment was defined as curative-intent initial treatment appropriate for tumor type and FIGO stage, including surgery, platinum-based chemotherapy, concurrent chemoradiotherapy, brachytherapy, or combined-modality treatment when indicated. Multivariable logistic regression was used to identify independent predictors of disease-specific guideline-concordant standard treatment, and a predictive model was developed and internally validated with assessments of discrimination, calibration, and clinical utility. Results Of the 972 patients, 63.7% received disease-specific guideline-concordant standard treatment. Higher comorbidity burden [Charlson Comorbidity Index (CCI) ≥ 4], ECOG performance status ≥2, age ≥75 years, and pulmonary disease were independently associated with lower odds of receiving disease-specific guideline-concordant standard treatment (all p < 0.05). In contrast, higher body mass index and serum albumin levels were associated with higher odds of receiving disease-specific guideline-concordant standard treatment. The final model demonstrated strong internally validated discriminatory performance (AUC = 0.933, 95% CI: 0.917–0.948), good calibration, and meaningful clinical utility across a wide range of decision thresholds. Conclusion In this pooled real-world observational cohort of older women with cervical, ovarian, or endometrial cancer, comorbidity burden was independently associated with receipt of disease-specific guideline-concordant standard treatment. The internally validated host-factor-oriented model may support individualized treatment discussions as an adjunct to multidisciplinary assessment, but it should not be interpreted as a tumor-specific treatment algorithm or as a replacement for disease-specific guideline-based decision-making. Further external validation and tumor-specific prospective studies are needed before broader clinical implementation.