Vaginal microbiome dynamics and clinical outcomes in cervical cancer patients undergoing radiotherapy

Advanced and recurrent cervical cancer (CC) is one of the major causes of death worldwide. Despite the advent of novel therapies and increasing utilization of immune checkpoint inhibitors, patients with advanced or recurrent CC have a poor prognosis, limited treatment options, and significant treatment-related toxicities. Limited evidence suggests that the vaginal microbiome (VM) may predict treatment outcomes and toxicity. Studies of the VM following radiotherapy (RT) using 16S amplicon sequencing observed a reduction in Lactobacillus and an increase in Prevotella . However, significant gaps remain in our understanding of the functional activities of these communities and their interactions with the host. To fill this gap, we initiated the “vaGinal hEalth in women ReceivinG pelvic radiation” (GEORGIA) trial to assess VM dynamics in CC patients undergoing RT and to investigate whether baseline microbiome composition predicts post-treatment outcomes. The GEORGIA trial (NCT04713618) enrolled 31 patients with CC who received pelvic RT. Vaginal swabs were collected at baseline and up to 24 months post-treatment. We assessed the metatranscriptome (metaT) using 2×150 bp sequencing libraries (>50 million reads per sample) to evaluate microbial abundance, activity, and human gene transcription. Longitudinal mixed-effects models (lme4 in R) and differential abundance analysis (ANCOM-BC2) were used to correlate microbial data with clinical variables, including recurrence status. A total of 27 patients had complete demographic information and longitudinal sequencing data. Over one-third were current smokers (35.5%), while 38.7% had never smoked, and 45.2% were post-menopausal at the time of treatment. Vaginal swab RNAseq reads aligned to human and microbe reference genomes in approximately equal numbers. Pelvic RT significantly altered the VM, driving a shift toward higher alpha diversity, as evidenced by increases in the Shannon (p = 0.001) and Simpson (p = 0.002) indices over time. When stratified by recurrence status, no significant differences in alpha diversity trajectories were observed. Differential abundance analysis revealed significant enrichment of taxa previously associated with HPV persistence and CC recurrence in VM studies, including Fusobacterium hominis (p = 0.02) and Fusobacterium gonidiaformans (p = 0.03), suggesting a potential role in disease progression. These findings underscore the possible influence of the VM on clinical outcomes and highlight its emerging role as a predictive factor in CC recurrence. Ongoing analysis will explore functional profiling and advanced longitudinal modeling to identify early microbial shifts preceding recurrence. We will use predictive modeling, integrating microbial and clinical features, to construct a recovery model and generate hypotheses for underlying biological mechanisms that can be tested experimentally.