Prognostic Factors in Nonresectable Pancreatic Adenocarcinoma: A Rationale to Design Therapeutic Trials Joaquı ´n Cubiella, M.D., Antoni Castells, M.D., Constantino Fondevila, M.D., Miquel Sans, M.D., Luı ´s Sabater, M.D., Salvador Navarro, M.D., and Laureano Ferna ´ndez-Cruz, M.D. Departments of Gastroenterology and Surgery, Institut de Malalties Digestives, Department of Medicine, Institut d’Investigacions Biome `diques August Pi i Sunyer, Hospital Clı ´nic, University of Barcelona, Barcelona, Catalonia, Spain OBJECTIVE: Most patients with pancreatic cancer show an advanced tumor at the time of diagnosis. In recent years, new therapies have been developed in such patients, thus forcing the analysis of factors that influence their survival. The present study was aimed at determining the prognostic factors in a series of 134 consecutive patients with pancre- atic carcinoma not suitable for surgical resection, due to either locally advanced tumors or metastatic spread. METHODS: First, 34 epidemiological, clinical, analytical, and tumor-related parameters recruited at admission were included in a univariate analysis of survival by using the Kaplan-Meier method. Those significant variables in the previous step were thereafter introduced in a Cox regression procedure. A prognostic index for calculating the relative risk of death of each patient was also proposed. RESULTS: The median survival time in the whole series was 3.11 months, the 3, 6, and 12 months probability of survival being 51%, 28%, and 8%, respectively. Eight of 34 variables (jaundice and toxic syndrome at admission, serum choles- terol, iron and alanine aminotransferase concentrations, leu- kocyte count, baseline performance status, and the presence of distant metastases) were selected in the univariate anal- ysis. Of these eight parameters, the Cox regression analysis identified a preserved baseline performance status (OR = 2.14, 95% CI = 1.49 –3.04) and the absence of metastases (OR = 1.34, 95% CI = 1.03–1.73) as the only variables independently associated with a longer survival. Therefore, a prognostic index was constructed allowing the division of patients in three different groups according to their relative risk of death (RRD) = exp (performance status  0.7589 + presence of metastases  0.2891). CONCLUSIONS: The results of the present investigation sug- gest that baseline performance status and distant metastases should be considered in designing and evaluating any ther- apeutic trial in patients with nonresectable pancreatic carcinoma. (Am J Gastroenterol 1999;94:1271–1278. © 1999 by Am. Coll. of Gastroenterology) INTRODUCTION Pancreatic carcinoma is characterized by a very poor prog- nosis after the initial diagnosis of the disease, being the fifth leading cause of adult death from cancer and responsible for nearly 5% of all cancer-related deaths (1, 2). The only effective process for complete cure depends on early detec- tion and accurate surgical removal of the tumor. Unfortu- nately, figures concerning resectability are still very low, most of the patients showing an advanced pancreatic cancer at the time of diagnosis, due to either local progression or metastatic spread (3–7). The prognosis of patients with nonresectable pancreatic carcinoma is extremely poor, as no effective chemotherapy has been definitively established (8). However, in recent years, clinical trials using newly developed drugs achieved some encouraging results in these patients, both in terms of survival and disease-related symptoms (9, 10). The arrival of new therapies and the necessity of performing well- designed randomized control studies force us to analyze which parameters influence the survival of advanced pan- creatic cancer patients. Although several studies have de- fined the prognostic factors in pancreatic cancer, those pa- rameters specifically affecting the outcome of patients with nonresectable tumors are less well known. Most of the prognostic studies published up to now were done to define histological (11–14) or surgical factors (3, 4, 15–17), ap- plied with only univariate analysis (5, 16), or including patients with different tumor stages as a whole (18, 19). The only multivariate analysis performed in advanced pancreatic cancer included a short series of patients receiving systemic chemotherapy and some important variables, such as vas- cular invasion and lymph node involvement, were not ana- lyzed (20). The present study was designed to determine the prog- nostic factors in a large series of nonresectable pancreatic carcinoma patients recruited in one single center, staged according to a common diagnostic protocol and who re- ceived only palliative treatment. To translate results from THE AMERICAN JOURNAL OF GASTROENTEROLOGY Vol. 94, No. 5, 1999 © 1999 by Am. Coll. of Gastroenterology ISSN 0002-9270/99/$20.00 Published by Elsevier Science Inc. PII S0002-9270(99)00073-8