Prognostic reclassification and survival outcomes in intermediate- and poor-risk nonseminomatous germ cell tumors (NSGCT).

5033 Background: Intermediate and poor-risk nonseminomatous germ cell tumors (NSGCT) form a large group of patients with varying prognosis. There is a need to revisit the existing International Germ Cell Cancer Collaborative Group (IGCCCG) classification to identify the ‘poorest subset’ of patients. Methods: We retrospectively analysed the outcomes of consecutive patients with intermediate and poor-risk NSGCT who were diagnosed between 2015 to 2021 at our centre and had received first-line chemotherapy. Overall survival (OS) and relapse-free survival (RFS) were calculated by Kaplan-Meier analysis. A multivariate risk factor model was constructed out of the prognostic factors for OS and RFS. Concordance statistic (c-statistic) was computed and compared between the proposed model and IGCCCG classification and boot-strap sampling was used to evaluate this comparison. Results: A total of 360 patients were included, having a median age of 29 (IQR: 24-34) years, with 149 in the intermediate-risk and 211 in the poor-risk according to the IGCCCG classification. Most of the patients had testis as the primary site of the tumor (336 patients, 93.3%). Median follow-up was 58.0 months (95%CI 50.7-65.3). The 5-year OS was 84.4% (95%CI 78.5-90.8) in the intermediate-risk versus 55.3% (95%CI 48.4-63.1) in the poor-risk (p < 0.001). RFS at 5 years was 76.8% (95%CI 70.0-84.3) in the intermediate-risk and 46.2% (95%CI 39.7-53.9) in the poor-risk (p < 0.001). Age above 35 years, baseline alpha-fetoprotein (AFP) >10,000 ng/mL, presence of liver and brain metastasis were identified to be poor prognostic risk factors for OS, while lung metastasis in addition to the above factors was prognostic for RFS by multivariate analysis. The multivariate model using the above four risk factors for OS had a higher c-statistic (0.668) than the existing IGCCCG classification (0.632) with an improvement of 0.036 (Table). This model also showed a similar improvement in c-statistic over IGCCCG classification for RFS (0.659 vs 0.625). Conclusions: The proposed model showed better discriminatory ability than the existing IGCCCG classification. This emphasizes the need to revise the risk stratification to identify the ‘very-high risk’ subset of patients for potential treatment intensification.[Table: see text]