HPV co-infection burden, vaginal microbiota restructuring, and host context are associated with Cervical precancer...

High-risk human papillomavirus (HPV) infection is necessary for cervical carcinogenesis, but HPV detection alone does not distinguish transient infection from lesions at greatest risk of progression. We evaluated whether HPV burden, vaginal microbiota structure, and host-context variables jointly characterize cervical intraepithelial neoplasia grade 3 (CIN3) in a Black/African American and White analytic cohort from the Vaginal Microbiome Health Project (VaMHP), integrating L1-based HPV typing, 16S rRNA vaginal microbiota profiling, and linked clinical metadata.

Among 1181 participants, 75 had CIN3. CIN3 was associated with HPV positivity (55/75, 73.3% vs 431/1106, 39.0%; odds ratio [OR] 4.31, 95% CI 2.55-7.29; Fisher exact p = 7.9 x 10^-9) and with multiple HPV infection among HPV-positive participants (35/55, 63.6% vs 176/431, 40.8%; OR 2.54, 95% CI 1.42-4.54; p = 0.0022). HPV communities in CIN3-positive samples showed higher Shannon diversity, greater observed strain richness, higher evenness, and significant beta-diversity separation. In vaginal microbiota analyses, alpha diversity did not differ by CIN3 status, but community composition did, and Lactobacillus crispatus was the only taxon depleted in CIN3 after multiple-testing correction. Race, age, and metronidazole exposure were central nodes in the host-factor network. In predictive modeling, a full integrated model combining metadata, HPV, and vaginal microbiota features (auROC = 0.745) outperformed both HPV + vaginal microbiota (auROC = 0.670) and HPV-only (auROC = 0.440) models.

CIN3 in this cohort was associated with coordinated shifts in virologic burden, vaginal community structure, and host social-clinical context. The results support a structure-function interpretation in which loss of Lactobacillus crispatus-dominant states and enrichment of dysbiosis-associated communities define a host-microbiome context that is more permissive to HPV persistence and precancer. These findings move beyond descriptive omics by showing that microbiome and host-context features add nonredundant discriminatory signal beyond HPV-only models.