Studies in prognostic factors relating to chemotherapy for advanced gastric cancer

The prognostic value of pretreatment information relating to prior treatment, demography, physical status, symptoms, disease involvement, pathologic, immunologic, and clinical chemistries were analyzed for a series of 322 patients with advanced gastric cancer. All patients received chemotherapy upon entry into Gastrointestinal Tumor Study Group protocols which were active between 1975 and 1978. Multivariate models were used to study relationships between prognostic factors and survival for all patients and objective tumor resonse for a subset of 137 patients with measurable disease. The initial performance status was a leading determinant of survival (P less than 0.0001). In addition, new summary measures relating to blood chemistries (P less than 0.01) and differential counts (P less than 0.001) were shown to influence patient survival. Blood chemistry parameters included SGOT, total serum protein, and total direct bilirubin while differential counts included absolute granulocytes, lymphocytes, and monocytes. Thus, the initial performance status, measurable disease status, blood chemistries, and differential counts are recommended as stratification factors in the design and analysis of trials involving patients with advanced gastric cancer. The initial performance status was examined in relation to other pretreatment data. The performance status at study entry correlated independently with the degree of weight loss (P less than 0.001), blood chemistries (P less than 0.01), differential counts (P less than 0.05), and peritoneal metastases (P less than 0.05). The measurable and nonmeasurable subgroups were compared with respect to baseline characteristics. Patients with measurable disease had more liver metastases (56 versus 35%) and less peritoneal metastases (76 versus 49%) than patients with nonmeasurable disease. Controlling for the imbalance in liver and peritoneal metastases, the presence of measurable disease was less favorable than nonmeasurable disease with respect to survival. Regarding the pathways of disease spread, there was a strong correlation (P less than 0.001) between primary tumor site within the stomach and location of metastases. Diffuse lesions were associated with the lowest frequency (25%) of liver metastases. Diffuse lesions (58%) and tumors of the pyloris (54%) were associated with the highest percentage of peritoneal metastases. Tumors of the cardia or fundus were more likely to metastasize to the liver while diffuse tumors were more likely to spread to the peritoneum. Pretreatment factors under study did not appear to be the dominant factors responsible for prolongation of survival in patients with an objective tumor response. Pretreatment factors predicted a three week advantage; however, a 22 week advantage was observed for responders over nonresponders.