A nomogram predicting survival of patients (pts) with metastatic or unresectable urothelial cancer (UC) treated with cisplatin-based chemotherapy

Abstract

5055 Background: Cisplatin-based chemotherapy is the standard of care for pts with metastatic or unresectable UC with phase III studies reporting median survivals of 12–15 months. Even more survival variation exists in phase II studies and this disparity is most frequently due to prognostic factors and not individual regimens. Thus, better tools are needed to predict survival both for individual pts and to balance phase III trials. Nomograms have utility in predicting short- and long-term outcome in muscle-invasive UC treated by surgery but they have not been explored in more advanced UC. Methods: We identified 308 pts with metastatic and/or unresectable UC treated on prospective phase II MSKCC protocols of cisplatin-based therapy containing 3–5 total chemotherapy agents. 203 pts received methotrexate, vinblastine, doxorubicin and cisplatin (MVAC), 45 had ifosfamide, paclitaxel and cisplatin (ITP) and 60 pts received doxorubicin plus gemcitabine (AG) followed by ITP (AG-ITP). Survival distributions were compared across trials. Pre-treatment characteristics were then assessed for impact on survival and a nomogram from a fitted Cox model was created to predict 1-yr, 2-yr, 5-yr and median survival. Results: No difference in median survivals were seen among the 3 regimens; median survival was 14.8 months for MVAC, 18.0 months for ITP and 16.1 months for AG- ITP (p=NS). Median survival for all pts was 12.99 months; 268 pts died and 40 pts were censored. 288 pts had all pre-treatment data. Characteristics most associated with survival included visceral metastases (present versus absent, p=.00001) and Karnofsky poor performance status (≥ 80 versus < 80, p= .0005) followed by hemoglobin (normal versus < normal, p=.01) and albumin (actual values, p<.02). These characteristics were then used to construct a nomogram utilizing all 4 factors to predict probabilities of 1-yr, 2-yr, and 5-yr survival. Conclusions: The number and sequence of drugs utilized in cisplatin-based chemotherapy did not substantially impact survival of pts with advanced UC. A nomogram of pre-treatment clinical factors can predict probability of pt survival at 1 yr, 2yrs, and 5 yrs. This nomogram may also be useful to balance treatment arms in phase III trials. No significant financial relationships to disclose.