Prediction of metastatic status in non-seminomatous testicular cancer

PURPOSE: To examine the significance of 90 biomarkers for predicting metastatic status in non-seminomatous germ cell tumors (NSGCT). By predicting metastatic status, it may be possible to eliminate unnecessary therapeutic or diagnostic efforts.

MATERIALS AND METHODS: We investigated 552 males who were diagnosed with non-metastatic (n = 273) and metastatic (n = 279) NSGCT between 2000 and 2011. The sample included cancers of different histologies: embryonal cell carcinoma (n = 131), teratoma (n = 55), and mixed histology (n = 366). We collected and analyzed more than 90 parameters via logistic regression: demographic characteristics, medical history, histopathological parameters, and levels of tumor markers and hormones.

RESULTS: Testis histology (p = 0.004), clinical symptoms (p = 0.0005), tumor length (p = 0.005), infiltration of the rete testis (p = 0.008), invasion of lymphatic (pL1) and blood vessels (pV1) (p < 0.0001), and levels of enzymes such as LDH, βHCG, AFP, and FSH (p values as small as <0.0001) were associated with metastatic status. With one model, we identified 14 out of 76 (18.4 %) metastatic NSGCT cases with 93-100 % certainty (positive predictive value) at 99 % specificity by the peripheral blood levels of LDH (day of operation) in combination with FSH measurements (1 day after operation). A second model included pV, tumor length, and FSH (1 day after operation). It identified 25 out of 90 (27.8 %) non-metastatic NSGCT with approximately 90 % certainty (negative predictive value) at 94-98 % sensitivity.

CONCLUSIONS: No single parameter was able to discriminate metastatic from non-metastatic NSGCT, but combinations of parameters in two predictive models accurately identified the metastatic status in 23 % of the cases in our sample.