Abstract
OBJECTIVE: To develop and internally validate a nomogram to predict the probability that a clinically suspected ocular surface squamous epithelial tumor is histopathologically malignant. METHODS: This retrospective study included 92 patients with ocular surface squamous epithelial tumors who underwent surgical excision and histopathologic confirmation between 2015 and 2020. Lesions were classified as benign (squamous papilloma) or ocular surface squamous neoplasia (OSSN) according to the latest AJCC criteria. Clinical and pathologic parameters were analyzed using univariate and multivariate logistic regression to identify independent predictors of malignancy. These predictors were incorporated into a nomogram model. Model performance was assessed by calibration and receiver operating characteristic (ROC) curve analysis with internal validation using 1,000 bootstrap resamples. RESULTS: Among the 92 cases, 50 (54.3%) were squamous papilloma, 24 (26.0%) were conjunctival intraepithelial neoplasia (CIN), and 15 (16.3%) were squamous cell carcinoma (SCC). After multiple factor logistic regression analysis, we selected preoperative prediction Models 1 (sex + age + corneal invasion + tumor diameter) and 2 (papillary hyperplasia + cytoplasmic changes + degree of differentiation). We established nomogram models for Models 1, 2, and 3 (Model 1+ Model 2). The results showed that all models had a good fit, which reflected a higher diagnostic value. All models could reliably discriminate malignant from benign lesions. The model predicts histopathologic malignancy (CIN/SCC) at the time of excision. CONCLUSION: The nomogram model based on clinicopathologic parameters provides a reliable tool for differentiating benign from OSSN. This model may assist clinicians in preoperative risk assessment, guide biopsy or excision decisions, and improve diagnostic accuracy in ocular surface tumor management.