Cervical cancer causes 330,000 deaths and 600,000 new cases annually worldwide1–3. Concurrent cisplatin-based chemoradiotherapy (CRT) is standard of care (SoC) for locally advanced cervical cancer (LACC)4–6. However, 40% of LACC patients experience primary CRT resistance or recurrence and do not benefit from current SOC1,7. Despite candidate biomarkers being proposed, none have achieved reliable clinical validation to stratify patients based on CRT response. Recent studies demonstrate immunotherapy as an adjuvant to CRT improves progression-free survival and overall survival in LACC (stage III–IVA)8. GCIG INTERLACE trial shows induction chemotherapy administered before CRT improves survival outcomes in stages IB2–IVA9. Given these therapeutic advances, a biomarker-informed refinement of CRT intensity is critical, ensuring patients poised to benefit from innovative strategies avoid unnecessary radiation-induced toxicity. Our group with access to longitudinal samples from 130 patients (pre- and post-CRT at week five) and a repository of over 3,000 cervical tumor and dysplasia samples, is well-positioned to address this need.

The goal of this proposal is to understand CRT resistance in cervical cancer, and in long term, develop effective methods to translate findings into improved clinical outcomes through better patient stratification and targeted therapeutics. Historically, biomarker research on cervical cancer has focused on early-stage or metastatic disease only. Because most clinical cases receive CRT, tumor sample collection is restricted to stage I cancers where surgery is performed, or metastatic cases allowing core biopsies. However, Colbert Lab pioneered a non-invasive swab-based sample collection protocol —using cytobrush and isohelix swabs— that overcomes these barriers, adequately assesses tumor genomic and immune landscape, and enables generation of patient-derived primary cell lines and organoids10,11. Karacosta Lab team also has expertise in building reference maps of phenotypic states using single-cell proteomics (CyTOF) 12,13. With mentorship and expertise of Dr. Colbert and Dr. Karacosta, I will address the knowledge gap regarding predictive biomarkers for CRT resistance in cervical cancer by completing following specific aims:

#Specific Aim 1: Functionally validate somatic mutations associated with RT resistance in cervical tumors.

Rationale: Cervical tumors under therapeutic pressure accumulate resistant subclones via somatic evolution16. Whole-exome sequencing (WES) of 76 longitudinal patients (pre-treatment and at week five of CRT) identified 1,145 genes bearing clonal mutations present at week five with loss-of-function (LOF) and/or gain-of-function (GOF) effects. Many of these genes are linked to CRT resistance in cervical cancer: DDR genes (PRKDC, TP53, RAD51B, etc.), oncogenic drivers (KRAS), apoptosis regulators (BIRC3), Fanconi Anemia factors (FANCF, FANCM), and P glycoproteins (ABCC2/3/4, etc.)14,15. For genes with LOF mutations observed in resistant tumors, we hypothesize that inactivating these genes in sensitive cells will recapitulate resistance. Conversely, for genes with GOF mutations, we hypothesize that introducing these specific mutations into sensitive cells will confer resistance to CRT. Sub aim 1a: We will conduct a CRISPR knockout screen to validate LOF mutation effects using a customized library (2 sgRNAs/gene for 1145 genes plus 50 non-targeting controls) in cervical cell lines (SiHa, HeLa, CaSki, C33A) and 3-5 patient-derived primary cells (PDCs) under fractionated radiation (IR: 2–4Gy over 14 days). sgRNA frequencies will be measured by NGS, and BAGEL2 will compute gene fitness scores. Genes whose knockouts yield high-survival phenotypes (enrichment) likely restrain resistance under normal conditions, whereas lethal knockouts (depletion) indicate essential genes under IR16. We will identify top-ranked genes using FDR <0.1, log2FoldChange ≥2, prioritizing relevance to resistance pathway. Sub aim 1b: We will validate GOF mutation effects by introducing clinically observed missense or truncating alleles from WES into CRT-sensitive cells via CRISPR/CAS9 to see whether these variants confer heightened resistance, using clonogenic assays under 2–4Gy of IR after five weeks against no IR controls.

#Specific Aim 2:Identify CRT resistance heterogeneity states in cervical cancer and develop a phenotypic reference map

Rationale: Beyond genetic drivers, plasticity and phenotypic heterogeneity often arise from differential protein expression, epigenetic changes, adaptive signaling, and tumor-stromal interactions, producing subpopulations that persist under genotoxic stress17,18. We hypothesize a high-dimensional single-cell proteomics-based reference map can define distinct “resistant states”, each with unique biomarker signatures. Sub aim 2a: We will design a 40–50-antibody CyTOF panel that profiles tumor cells and microenvironment features. This panel will target tumor identification markers (p63, CK5/6; MUC5AC) and pathways linked to DDR(DNA-PKcs, RAD51, BRCA1, etc), hypoxia adaptation (CAIX), cell-cycle checkpoints & apoptosis evasion(ATM, CHK1, WEE, etc. ), oncogenic signaling(pAKT, ps6, PTEN, p65, etc) , drug efflux (MRP1, MDR1, etc.), EMT(E-cadherin, Slug, Vimentin, etc) and immune evasion PD-L1, TIM3, CD8, CD45 etc). Subjecting PDCs to fractionated radiation (2–4 Gy) plus cisplatin will reveal changes reflecting partial resistance. By systematically titrating metal-labeled antibodies for specificity and robust detection, we will ensure panel captures critical phenotypic. Sub aim 2b: We will apply optimized panel to 50 longitudinal patient samples collected before and during CRT and build a CRT resistant states map. Using dimensionality reduction (UMAP) and clustering (CCAST), we will identify discrete resistant subpopulations of cells, and delineate boundaries to create a two-dimensional map with defined resistant states19. The resulting phenotypic map will show where each clinical sample falls and a supervised learning model will assign new samples to states, enabling rapid identification of clinically significant resistant phenotypes using CyTOF.

By successfully accomplishing this aim, we will generate a valuable resource for understanding and predicting CRT resistance in cervical cancer, potentially leading to improved personalized treatment strategies.