Simple prognostic model to predict survival in patients with undifferentiated carcinoma of unknown primary site

We performed this study to identify prognostic factors in a subgroup of patients with carcinoma of unknown primary site treated with cisplatin combination chemotherapy.

Seventy-nine patients with poorly differentiated adenocarcinoma or undifferentiated carcinoma of unknown primary site were treated on two consecutive phase II chemotherapy protocols. The first protocol consisted of treatment with 3-week courses of cisplatin, etoposide, and bleomycin (BEP). In the second protocol, cisplatin was administered weekly combined with oral administration of etoposide (DDP/VP). To identify prognostic factors, univariate and multivariate analyses were conducted.

In the univariate analysis, performance status, histology, liver or bone metastases, and serum levels of alkaline phosphatase and AST were significant variables to predict survival. In the multivariate analysis, performance status and alkaline phosphatase were the most important prognostic factors.

Good-prognosis patients had a performance score of 0 (World Health Organization [WHO]) and an alkaline phosphatase serum level less than 1.25 times the upper limit of normal (N). These patients had a median survival duration greater than 4 years. Intermediate-prognosis patients were characterized by either a WHO performance status < or = 1 or an alkaline phosphatase level > or = 1.25 N. These patients had a median survival duration of 10 months and a 4-year survival rate of only 15%. The poor-prognosis group had both a WHO performance status > or = 1 and an alkaline phosphatase level > or = 1.25 N. These patients had a median survival duration of only 4 months and none survived beyond 14 months. Treatment strategies for these three groups are discussed. It is suggested that this prognostic model be validated in other patients series.