[Pre-operative identification of patients at risk for septic complications after surgery for neoplasms of the digestive system. A multiphasic multicenter study].

ne hundred ninety-six patients undergoing surgical treatment for digestive tract cancer were recruited in a polycentric, multistep study to value significant differences of pre-operative nutritional and immunologic parameters between patients with or without post-operative infections. Such parameters might be useful for the identification of "septic risk" patients. Primarily, 63 patients were given a complete anthropometric evaluation, laboratory screening and immunologic assessment (surface marker analysis of peripheral blood mononuclear cells, polymorphonuclears in vitro chemotaxis and phagocytosis). Only a few serologic parameters were significantly reduced in patients with post-operative infections, namely: total serum proteins (p less than 0.02), albumin (p less than 0.02), beta-globulins (p less than 0.01), and C3c (p less than 0.05). These parameters were elaborated with Fisher's linear discriminant function to detect the optimal discriminant threshold able to identify "septic risk" patients. The predictivity of this discriminant function was validated in a second group of patients (n = 49) in which total serum protein with electrophoresis were pre-operatively assayed as well as skin test reactivity to intradermal injection of recall antigens (Multitest skin-testing); discriminant sensitivity was 80%, the specificity 50%, and positive predictivity 62%. This low discriminant power of the equation based only on serum proteins assessment coupled with the significant negative relationship between skin testing and post-operative infections (male, p less than 0.005; female, less than 0.025) suggested a re-elaboration of the discriminant function, including the skin-testing score. The definite assessment of this new discriminant equation was evaluated in the third sample of patients (n = 84) which showed 67% sensitivity, 88% specificity, and 76% positive predictivity, thus suggesting the high reliability of this test for the pre-operative selection of "septic risk" patients.