Developing a predictive nomogram evaluate missed diagnosis risk of High-grade squamous intraepithelial lesion+(HSIL+) in patients with Low-grade squamous intraepithelial lesion (LSIL)

Abstract

Abstract Objective Patients with abnormal cervical cancer screening and pathologically diagnosed as LSIL by colposcopic biopsy had a risk of missed diagnosis of HSIL+. To explore the possible risk factors of missed diagnosis of HSIL + in patients with LSIL diagnosed by colposcopic biopsy. Having a predictive model may help to reduce the rate of missed diagnosis. Methods The retrospective single-center study involved 505 patients with LSIL diagnosed by colposcopic biopsy and treated with surgery. The enrolled patients were divided into a training set and a validation set. Logistic regression was used to establish a predictive model which generated risk scores for missed diagnosis HSIL+. The model was a nomogram and was assessed with goodness of fit test,calibration and decision curves. Results HPV16/18 infection (OR 2.071; 95%CI 1.039–4.127;p = 0.039), TCT ≥ ASC-H(OR 4.147; 95%CI 1.392–12.355༛p = 0.011), TZ3 (OR 1.966; 95%CI 1.003–3.853༛p = 0.049) and Colposcopic impression G2 (OR 3.627; 95%CI 1.350–9.743; p = 0.011) were the independent risk factors for missed diagnosis of HSIL+. The nomogram model based on these factors had a high predictive value, which was internal validated(AUC:0.747(95% CI: 0.672–0.822)) and external validation(AUC:0.83 (95%CI༚0.716–0.944)). Calibration suggested good coherence between training set and validation set. Decision curve analysis also suggested the model had high clinical practical value. Conclusion HPV16/18 infection, TCT ≥ ASC-H, TZ3 and colposcopic impression G2 were independent risk factors for patients preoperative with LSIL missed diagnosis of HSIL+, and the nomogram model based on these factors had good guiding value in helping clinicians to avoid missed diagnosis of HSIL + and overtreatment of LSIL.