Independent predictors of operative mortality and postoperative complications in surgically treated carcinomas of the oesophagus and cardia--is the aggressive surgical approach worthwhile?

During the 25 years 1960-84, 657 patients were operated on for carcinomas of the thoracic oesophagus (n = 347) or gastric cardia (n = 310). Resection was carried out in 514 (78%) and oesophagogastrostomy in 481 (73%). Overall operative mortality (defined as death within 30 days) was 19% (n = 122). Pulmonary complications developed in 167 patients (25%), cardiovascular complications in 100 (15%), and anastomotic leakage and mediastinitis in 36 (6%). After radical resection of a localised tumour (n = 144), or non-localised tumour (n = 224), pallative resection (n = 146), or exploration (n = 143), the operative mortality and five year cumulative survival were 10% and 26%, 15% and 8%, 27% and 0, and 24% and 0, respectively (p less than 0.01 and p less than 0.0001). Using logistic regression analysis several variables were found to be independent predictors of operative mortality, pulmonary complications, cardiovascular complications, and anastomotic leakage. The predictor variables reflected both general preoperative status of the patients, preexisting cardio-pulmonary diseases, stage of the cancer, and surgical procedures. Based on the final logistic regression models the patients were stratified into risk groups (12 for operative mortality, pulmonary complications, and cardiovascular complications, and eight for anastomotic leakage). Operative mortality varied from 0 to 80%, pulmonary complications from 3 to 100%, cardiovascular complications from 0 to 100%, and anastomotic leakage from 0 to 50% (p less than 0.0001 in each case). Given the high operative mortality and complication rates, and the low five year survival rate after palliative procedures or exploratory operations, a more selective surgical approach seems warranted. Patients likely to have a good response should be identified before operation.