Immune remodeling during cervical cancer chemoradiotherapy: temporal biomarkers and the timing of checkpoint blockade

Concurrent chemoradiotherapy (CRT) is the backbone of curative treatment for locally advanced cervical cancer, and PD-(L)1 blockade has entered first-line practice. Yet the pivotal trials diverge: pembrolizumab added to CRT improved survival in KEYNOTE-A18, whereas durvalumab did not in CALLA. Most prior syntheses organize the cervical tumor immune microenvironment by cell type and rely on cross-sectional, pre-treatment sampling, and several predate KEYNOTE-A18 and recent longitudinal multi-omic data. Here we reframe the question along the treatment time axis. Section 1.1 reports that evidence base in full, under tiered eligibility criteria. Drawing on the still-small body of studies with paired or serial pre-, on-, and post-treatment sampling of tumor and blood, we reconstruct four sequential states: an HPV-shaped baseline ecology; an early immunogenic priming window driven by DNA-damage sensing and type I interferon; a mid-treatment window in which intratumoral activation coexists with radiation-induced lymphodepletion and compensatory checkpoint upregulation; and a post-treatment state of viral-antigen persistence, minimal residual disease, and adaptive resistance. We use this lens to interpret CRT–immune checkpoint inhibitor (ICI) trial heterogeneity without over-attributing it, and derive a falsifiable prediction, which neither pivotal trial can adjudicate and which the best available randomized-evidence synthesis currently opposes: blockade delivered within the priming window may outperform equal exposure given as post-CRT maintenance, in patients who mount on-treatment activation without catastrophic lymphodepletion. This is a mechanistically motivated hypothesis, not an inference from existing results. Testing needs a factorial or biomarker-stratified trial. We also propose a temporal biomarker framework spanning baseline-predictive, on-treatment early-response, and post-treatment residual-disease markers, graded by evidence and offered as a research agenda. For each stage we name the design that would qualify or refute it. Stage 1 needs registered analysis of archival trial tissue. Stage 2 needs a nested correlative sub-study with prespecified effect modifiers. Stage 3 needs a blinded, prespecified prognostic cohort. We propose, as a hypothesis for prospective testing, that response to CRT–ICI may depend less on any single baseline feature than on whether a patient completes the dynamic transition from viral-antigen exposure through antigen presentation and T-cell recruitment to durable surveillance.