We have attempted to establish a prognostic nutritional index prospectively in 633 patients with digestive
carcinoma making it possible to predict
tumor resectability and occurrence of postoperative complications. The following factors were studied prior to surgery: age, sex,
weight loss index, triceps skinfold thickness, midarm circumference, lymphocyte count,
plasma albumin,
transferrin and
prealbumin levels,
delayed hypersensitivity as investigated by 8 skin tests. The prognostic nutritional index was calculated using Mullen's formula. Unidimensional and multidimensional studies (linear discriminant and logistic discriminant analyses with partial correlations) were used for the statistical analysis. Mortality was 6,7 p. 100 while 16,2 p. 100 of patients presented a postoperative complication. The following variables showed a significant correlation with the resectability of the tumour:
weight loss (p less than 0.0001), triceps skinfold thickness (p less than 0.01), midarm circumference (p less than 0.01), total lymphocyte count (p less than 0.01), plasma
transferrin (p less than 0.01),
delayed hypersensitivity (p less than 0.05). In the occurrence of postoperative complications there was a statistically significant value only to
weight loss (p less than 0.001) and
delayed hypersensitivity (p less than 0.05). Logistic discriminant analysis gave a linear model predicting
tumor resectability in 78 p. 100 of patients and postoperative complications in 86 p. 100 of patients in which
weight loss and
delayed hypersensitivity were statistically significant. No prognostic value of Mullen's index was found. This study shows that assessment of the preoperative nutritional status of patients with
gastrointestinal cancer makes it possible to predict
tumor resectability and postoperative complications.