Treatment Pathways and Candidate Correlates of Surgical Intervention in Locally Advanced Cervical and Other Gynecological Cancers: A Hypothesis-Generating Single-Center Cohort Study

Abstract

Background/Objectives: Management of locally advanced cervical cancer (LACC) and other gynecological malignancies aims to achieve locoregional control while avoiding morbid surgical salvage. This exploratory, hypothesis-generating study described the treatment pathways of three protocol groups and examined whether clinical, sociodemographic, and treatment-deviation variables are associated with any oncological surgery. Methods: Sixty-nine adult patients with gynecological or urogenital cancers (89.9% cervical carcinoma) treated with platinum-based chemotherapy at a Romanian tertiary center between April 2024 and April 2025 were stratified into INTERLACE-type dose-dense induction chemotherapy followed by chemoradiotherapy (Group A, n = 27), classic three-weekly neoadjuvant chemotherapy followed by chemoradiotherapy (Group B, n = 7), and no induction chemotherapy (Group C, n = 35). Associations with any oncological surgery (upfront, completion, or salvage) were explored by multivariable logistic regression with bootstrap internal validation, Firth penalized regression, and restricted sensitivity analyses. Results: Twenty-five patients (36.2%) underwent surgery (14 upfront, 8 completion, 3 salvage), and none of them were from Group A, whose median follow-up was only 8.6 months and whose allocation was confounded by diagnosis and stage. The model showed an apparent area under the curve of 0.717 (optimism-corrected 0.672), accuracy of 73.9% against a no-information rate of 63.8% (p = 0.049), sensitivity of 0.480, and specificity of 0.886. Urban provenience (OR 0.343, 95% CI 0.113–1.039, p = 0.059), Delta Target Dose (OR 1.109 per Gy, p = 0.076), and number of chemotherapy cycles (OR 1.298 per cycle, p = 0.076) showed non-significant trends whose direction was unchanged in all sensitivity analyses. Conclusions: The absence of surgery after induction chemotherapy is hypothesis-generating rather than confirmatory, and the exploratory model is not a clinical prediction tool. Geographic provenience and planning-to-delivery treatment deviation are candidate variables for prospective evaluation in a homogeneous LACC population with standardized radiotherapy parameters and mature oncological endpoints.