Predicting Survival following Surgery for Bronchogenic Carcinoma

¹ Roswell Park Memorial Institute, Buffalo

Data from 987 patients in a collaborative clinical trial were analyzed, by use of multivariate techniques to separate the effects of prognostic factors and to arrive at a mechanism useful to the clinician for prognosis. One-fifth of the total variation was accounted for by four factors. Nearly half of this (42 percent) was due to stage of disease, followed by tumor size (32 percent), cell type (10 percent) and postsurgical complications (7 percent). The absence of significant interactions simplified survival prediction. A mechanism is presented utilizing a table-look-up method for the first three variables to arrive at both expected mean survival and probabilities of surviving one, three, or five or more years. Two subgroups stood out: (a) patients having poor survival (14 percent of cases) averaging eight months, usually were stage III with medium to medium-large tumors of anaplastic cells, and (b) patients having superior survival (17 percent of cases) averaging 52 months, usually were stage I with small to small-medium tumors and all but anaplastic cell types.