Predictive models for lymph node metastases in patients with testicular germ cell tumors

摘要: Purpose: To develop predictive models for lymph node (LN) metastasis in testicular germ cell tumors. Materials and Methods: 291 patients with tumors were included, which divided into seminomatous nonseminomatous groups. For screening the risk factors LN metastasis, tumor-related characteristics (including histopathological information tumor markers) alpha fetoprotein node-related features on CT compared between metastatic cases nonmetastatic cases. Two logistic regression built each histological group, one depending all tumor- (Model 1) another only 2). Receivers operating characteristic curves used to evaluate abilities of these models. Results: 117 positive nodes/regions identified 68 patients, including 51 metastases 17 occult metastases. Based selected independent factors, sensitivity, specificity, accuracy, value, negative value Models 1 2 groups (95.5%, 95.3%, 77.8%, 99.2%), (63.6%, 83.6%, 80.7%, 40.0%, 93.0%), (93.5%, 94.7%, 94.3%, 89.6%, 96.8%), (89.1%, 44.2%, 58.9%, 43.6%, 89.4%), respectively. Conclusion: established based node- factors. In regular surveillance is likely sufficient predicting status, while without any a long interval-time follow-up should be considered. Additionally, right side tend involve contralateral LNs left ones. Positive inguinal more frequently occur history groin surgery.