Development of a Risk Assessment Model for the Recurrence of high-grade squamous intraepithelial lesions and differentiated vulvar intraepithelial neoplasia

Abstract Objective To construct a sophisticated risk prediction model for the recurrence of high-grade squamous intraepithelial lesions (HSIL) and differentiated vulvar intraepithelial neoplasia (dVIN) following treatment. This model seeks to facilitate the early detection and focused screening of individuals who are at increased risk of vulvar intraepithelial neoplasia. Methods Clinical data from 257 patients diagnosed with the differentiated vulvar intraepithelial neoplasia (dVIN) or the vulvar high-grade squamous intraepithelial lesions (HSIL), who underwent treatment and maintained regular follow-up at Anhui Maternal and Child Health Hospital between January 2010 and January 2025, were meticulously reviewed through a retrospective analysis. Patients were stratified into two distinct cohorts: the relapse group (n=60), comprising individuals who experienced post-treatment recurrence, and the non-recurrence group (n=197), consisting of those who remained disease-free following therapy. For robust model development, the dataset was methodically divided into a training subset, which included 70% of the total cases, and a validation subset, accounting for the remaining 30%. Leveraging logistic regression analysis, key predictors were identified and subsequently integrated to construct an intricate risk prediction model for post-treatment recurrence of HSIL and dVIN, thereby paving the way for enhanced clinical decision-making. Results Univariate logistic regression analysis unveiled that age, menopause, immunosuppression, HPV16 infection, histopathological characteristics, and positive surgical margins were positively correlated with recurrence risk, whereas low-risk HPV types exhibited a negative correlation with recurrence risk (all P < 0.05). Variables with a P-value less than 0.1 in the univariate analysis were subsequently included in the LASSO regression, where non-zero coefficient variables were meticulously selected based on the lambda value (0.008) corresponding to the minimum standard error deviation. The significant variables identified through the univariate analysis were subsequently integrated into the multivariable logistic regression analysis. The research findings revealed that age, smoking history, immunosuppression, HPV16 infection, and histopathology were robustly associated with an elevated recurrence risk (all P < 0.05). In the training dataset, the area under the receiver operating characteristic curve (AUC) was found to be 0.793 (95% CI: 0.77–0.880), accompanied by a median prediction success probability of 0.810. The maximum approximate Youden index was recorded at 0.521, with a sensitivity of 66.7% and a specificity of 85.4%. The positive predictive value was 0.583, while the negative predictive value stood at 0.893. Within the internal validation dataset, the AUC improved to 0.831 (95% CI: 0.726–0.937), featuring a median prediction success probability of 0.846, a maximum approximate entry index of 0.645, a sensitivity of 77.8%, a specificity of 86.7%. The positive predictive value was 0.636, and the negative predictive value was 0.929. The Hosmer-Lemeshow goodness-of-fit test indicated that the model calibration was acceptable for both the training set (P = 0.069) and the internal validation set (P = 0.086). Calibration curves displayed trends remarkably consistent with the ideal curve, signifying commendable calibration. Furthermore, the clinical decision curve substantiated the net benefit derived from the model. Conclusion A prediction model incorporating age, smoking history, immunosuppression, HPV16 infection, and histopathology demonstrates good predictive performance for post-treatment recurrence of HSIL and dVIN. This model can assist clinicians in evaluating recurrence risk, providing guidance for clinical consultations and enabling targeted follow-up and treatment plans for those at high risk.