Predictive value of vaginal lactic acid bacteria changes on occurrence of HR-HPV-infected cervical intraepithelial neoplasia and construction and validation of nomogram model

Abstract

Objective: To explore the predictive value of vaginal lactic acid bacteria changes on the occurrence of HR-HPV infected cervical intraepithelial neoplasia (CIN), and to construct and validate Nomogram model.

Methods: The community composition, quantitative distribution and other indicators of vaginal lactic acid bacteria in patients with and without CIN were detected and analyzed. The key lactic acid bacteria characteristics and other potential risk factors related to CIN were screened out using statistical methods, and then the Nomogram prediction model was constructed. The model discrimination was assessed by the receiver operating characteristic (ROC) and the calibration curves, and decision curves analysis (DCA) was performed to verify the reliability of the model.

Results: There was no significant difference in the incidence of CIN, baseline data and vaginal lactic acid bacteria parameters between the training set and the verification set (P > 0.05). Univariate analysis identified significant differences in clinical factors (e.g., age of first sexual activity, safety measures) and vaginal lactobacilli (e.g., L. acidophilus, L. jenseni, L. gasseri, L. rhamnosus) between groups (P < 0.05). Multivariate Logistic regression confirmed safety measures, serum progesterone levels, and relative abundances of L. acidophilus, L. crispatus, L. jenseni, and L. rhamnosus as independent risk factors for CIN (P < 0.05). Further, the nomogram prediction model was constructed, and the nomogram model had good calibration and fit between prediction and reality in the training set and the verification set, and the nomogram demonstrated excellent discrimination in the training set and validation set, with calibration curves showing minimal absolute error (0.048 vs. 0.046) and DCA confirming clinical utility across a threshold probability of 0.05-0.95.

Conclusion: Changes in vaginal lactobacilli (e.g., reduced L. acidophilus, L. jenseni, L. gasseri, and L. rhamnosus) are significantly associated with CIN risk, potentially reflecting dysbiosis-driven vulnerability to HR-HPV persistence The Nomogram model demonstrated high accuracy in internal validation, suggesting its potential utility as a clinician-friendly tool for individualized risk assessment. However, external validation in prospective cohorts is required before clinical implementation.

Keywords: Cervical intraepithelial neoplasia; HR-HPV infection; Nomogram model; Vaginal lactic acid bacteria.

Fig. 1

Alignment diagram of an alignment prediction model for CIN occurrence in patients with HR-HPV infection. Note: X1-X6 are respectively: safety measures, serum progesterone level, lactobacillus acidophilus relative abundance, lactobacillus crispatus relative abundance, lactobacillus jenseni relative abundance, lactobacillus rhamnosus relative abundance

Fig. 2

Calibration curve of predictive model for CIN occurrence in patients with HR-HPV infection (A is the training set calibration curve, B is the validation set calibration curve). Calibration curves show predicted vs. observed CIN probability. Dashed lines represent ideal fit; solid lines show model performance with shaded areas indicating 95% confidence intervals from bootstrapping (1000 replicates). Mean absolute error (MAE) after correction: 0.12 (training), 0.14 (validation)

Fig. 3

ROC curve (A is the ROC curve of the training set, and B is the ROC curve of the validation set). ROC curves with 95% confidence intervals (dotted lines). AUC values reflect bootstrap-corrected estimates

Fig. 4

Decision curve (A is the training set decision curve, B is the validation set decision curve)