Prognostic factors for relapse in stage I seminoma: a new nomogram derived from three consecutive, risk-adapted studies from the Spanish Germ Cell Cancer Group (SGCCG).

BACKGROUND

We aimed to analyze prognostic factors for relapse in stage I seminoma managed by either active surveillance or adjuvant chemotherapy, and to describe the long-term patterns of recurrence in both groups.

PATIENTS AND METHODS

From 1994 to 2008, 744 patients were included in three consecutive, prospective risk-adapted studies by the Spanish Germ Cell Cancer Group. Low-risk patients were managed by surveillance and high-risk patients were given two courses of adjuvant carboplatin. Relapses were treated mainly with chemotherapy. Patient age, tumor size, histological variant, pT staging, rete testis invasion, and preoperative serum BHCG levels were assessed for prediction of disease-free survival (DFS).

RESULTS

After a median follow-up of 80 months, 63 patients (11.1%) have relapsed: 51/396 (14.8%) on surveillance and 12/348 (3.2%) following adjuvant carboplatin. Actuarial overall 5-year DFS was 92.3% (88.3% for surveillance versus 96.8% for chemotherapy, P = 0.0001). Median time to relapse was 14 months. Most recurrences were located at retroperitoneum (86%), with a median tumor size of 26 mm. All patients were rendered disease-free with chemotherapy (92%), radiotherapy (5%), or surgery followed by chemotherapy (3%). A nomogram was developed from surveillance patients that includes two independent, predictive factors for relapse: rete testis invasion and tumor size (as a continuous variable).

CONCLUSION

Long-term follow-up confirms the risk-adapted approach as an effective option for patients with stage I seminoma. The pattern of relapses after adjuvant chemotherapy is similar to that observed following surveillance. A new nomogram for prediction of DFS among patients on surveillance is proposed. Rete testis invasion and tumor size should be taken into account when considering the administration of adjuvant carboplatin. Prospective validation is warranted.