The predictive value of serum tumor markers for pathologic findings after chemotherapy for primary mediastinal nonseminomatous germ cell tumors

Abstract

5087 Background: While data exists on the predictive value of serum tumor markers (STM) alpha-fetoprotein and human chorionic gonadotropin for postchemotherapy pathology in testicular nonseminomatous germ cell tumors (NSGCT), similar data for primary mediastinal NSGCT (PMNSGCT) have not been reported. We undertook a study of PMNSGCT patients who received cisplatin-based chemotherapy followed by surgery to investigate the relationship between preoperative STM values and surgical pathology. Methods: From 1981–2007, 166 PMNSGCT patients who presented to our institution for surgical therapy following chemotherapy were retrospectively reviewed. Multiple variables were analyzed with respect to preoperative STM levels and the “worst” pathology identified within the residual mass {noncancer (necrosis/teratoma, 59%, n=97), persistent NSGCT with or without malignant transformation of teratoma to nongerm cell cancer (NGCCA) (31%, n=52), or pure NGCCA (10%, n=17). Results: Elevation of either STM was present in 40% (n=66) patients at the time of surgery however only 59% of these patients demonstrated pathologic evidence of malignancy with persistent NGSCT (n=34) or pure NGCCA (n=5). Thirty-seven patients had a documented rise in STM prior to surgery and demonstrated pathologic evidence of malignancy in 25 (67%) cases. All, but only 8, patients with either STM level > 1,000 uniformly demonstrated pathologic evidence of persistent NSGCT. The overall predictive value of either STM elevation for persistent NSGCT was 52% (CI 39–64%). In contrast, 100 patients presented to surgery with normal STMs and 30% (n=30) of these patients demonstrated pathologic evidence of viable malignancy. The predictive value of normal STMs for benign disease was 70% (CI 61–79%). Conclusion: STM levels following chemotherapy for PMNSGCT are imperfect predictors of residual malignant or nonmalignant pathology. These data, along with the poor response of PMNSGCT to “salvage” chemotherapy, further support our past and present policy to surgically remove residual disease after chemotherapy when deemed feasible regardless of STM status. No significant financial relationships to disclose.