Association of viral load values with cervical lesion severity and exploratory genotype analysis in HPV-positive Indian women
Abstract
Hybrid Capture 2 (HC2) provides a semi-quantitative estimate of high-risk human papillomavirus (HPV) burden as relative light units/cutoff (RLU/CO), and HPV genotyping may add value to triage assessment. We evaluated the association of RLU/CO with cervical lesion severity and explored whether genotype information improves diagnostic performance in HPV-positive Indian women. This cross-sectional study was conducted in two sequential phases. In the initial phase, 217 consecutive HPV-positive women underwent HC2 HPV DNA testing, colposcopy-guided biopsy, and histopathological assessment. In a subsequent phase, an additional group of 111 HPV-positive women underwent the same clinical and histopathological assessment and also provided specimens for HPV genotyping. The genotype-enriched group was analysed separately for genotype distribution, co-infection patterns, and exploratory genotype-integrated prediction models. Age, Histopathology, and viral load (HC2 RLU/CO) were analysed in the full cohort. The genotyped subset was additionally evaluated for HPV genotype distribution and genotype-integrated ROC models. Descriptive statistics, Kruskal–Wallis testing with Holm-adjusted Dunn's post-hoc comparisons, and ROC analyses with 95% confidence intervals and paired DeLong comparisons were performed. RLU/CO values differed across study-specific histopathological severity categories, with higher median values in LSIL, HSIL, and invasive cervical cancer than in normal histology and chronic cervicitis. In the cohort of 328 women, RLU/CO discriminated HSIL or invasive cervical cancer (severity score ≥3) with an AUC of 0.798 (95% CI, 0.726–0.868). A study-derived Youden-index cutoff of 12.08 RLU/CO yielded a sensitivity of 81.1% (95% CI, 69.5%–90.9%) and specificity of 73.1% (95% CI, 68.0%–78.0%). Among 111 genotyped participants, HPV16 was the most frequently detected genotype, and multiple-genotype infections occurred in 64 participants (57.7%). In the genotyped subgroup, comparative ROC analysis for HSIL or invasive cervical cancer showed AUCs of 0.659 for RLU/CO alone, 0.608 for RLU/CO plus HPV16/18, and 0.461 for RLU/CO plus other high-risk HPV. Neither genotype-integrated model improved discrimination compared with RLU/CO alone; these exploratory estimates require validation in larger independent cohorts. HC2 RLU/CO showed potential for identifying HSIL or invasive cervical cancer. The study-derived cutoff requires validation in independent populations. Genotype-integrated models showed variable performance in the small subgroup. The HPV16/18 model was comparable to RLU/CO alone, but findings were exploratory. Larger prospective studies are needed before clinical implementation.