Enhancing risk stratification in primary mediastinal nonseminomatous germ cell tumors (PMNSGCT): A 27-year experience at a tertiary cancer center.

Abstract

373 Background: Mediastinal GCTs and PMNSGCTs poorly benefit from CT and half of patients (pts) still die for disease. Enhancing the risk stratification may result in tailoring a personalized treatment strategy since diagnosis. Methods: Between 1985 and 2012, 87 pts with PM-GCT were treated at our center. Of them, pure seminomatous histology was excluded. Cox proportional hazards regression analysis was conducted to examine the prognostic impact of these candidate factors on overall survival (OS): type of 1st-line CT (high [HDCT] vs conventional dose [CDCT]), post-CT surgery, type of baseline elevated serum tumor marker (STM), presence of lung or liver-bone-brain metastases (LBB), STM response (still elevated vs normal or normalized), and histology (viable cancer [VC] vs necrosis/teratoma [NT]). OS curves were compared by Kaplan Meier method with the log-rank test. Results: The study included 68 cases with PMNSGCT. Median age was 28.5 yrs (IQR: 23-35). 12 pts (17.7%) presented with mediastinal syndrome, 23 (33.8%) had lung and 7 (10.3%) LBB metastases. 12 pts received upfront HDCT and 45 pts (66.2%) underwent post-CT surgery. The final model of poor prognostic factors included no surgery (HR: 8.74, 95%CI, 1.77-43.01), surgery + VC (HR: 6.97, 95%CI, 1.46-33.30), and lung metastases (HR: 2.92, 95%CI, 0.99-8.64). The model demonstrated moderate discriminatory ability for OS (c-statistic=0.68). A risk stratification model based on the combination of these factors and accounting for a 50% five-year survival cutoff identified 2 groups (poor prognosis, N=28 vs good prognosis, N=26) with distinct overall survival curves (p<0.001). Pre-operative STM and final histology were not associated (p=0.574 at Chi squared test). 5-yr OS after receiving 2nd line CT (n=25) was 18.7% (95%CI, 7.9-44.5). Results are limited by small numbers. Conclusions: Pts with PMNSGCT classified as having a good prognosis in this model had a fairly high survival estimate, while a strategy of consolidation CT for pts with poorest risk features warrants investigation, once the model is confirmed. The effect of surgery on survival was independent of post-CT STM, which also poorly predicted final histology.