Rpc. Singson et al., Estimation of tumor stage and lymph node status in patients with colorectal adenocarcinoma using probabilistic neural networks and logistic regression, MOD PATHOL, 12(5), 1999, pp. 479-484
Citations number
41
Categorie Soggetti
Research/Laboratory Medicine & Medical Tecnology","Medical Research Diagnosis & Treatment
Staging colorectal adenocarcinoma on the basis of biopsy specimens could id
entify patients who might benefit from neoadjuvant therapy without undergoi
ng resection first. In this study, we evaluated the ability of artificial n
eural networks with genetic algorithms and multivariate logistic regression
to predict the stage of 99 patients with primary colorectal adenocarcinoma
by analyzing age, tumor grade, and immunoreactivity to p53 and bcl-2 with
use of endoscopically obtained biopsy specimens. We correlated results with
regional lymph node status and tumor stage, identified in subsequent colec
tomy specimens. bcl-2 and p53 protein expression were demonstrated by immun
ohistochemical methods, using formalin-fixed, paraffin-embedded biopsy tiss
ues. Tumor grade was evaluated in hematoxylin- and eosin-stained sections.
Patients were divided into training (n = 75) and testing cases (n = 24). Se
veral probabilistic neural networks with genetic algorithm models were trai
ned, using the four prognostic features as input neurons and regional lymph
node status or stage as output neurons. Data were analyzed with univariate
statistics and multivariate logistic regression. The cases were divided in
to training (n = 40) and testing (n = 59). The best two models classified c
orrectly the lymph node status of 20 of 24 test patients (specificity, 80%;
sensitivity, 85%; positive predictive value, 86%) and the tumor stage of 2
1 of 24 test patients (specificity, 82%; sensitivity, 92%; positive predict
ive value, 85%), respectively. Tumor grade and p53 protein were statistical
ly significant (P <.05) by analysis of variance for lymph node status and t
umor stage. Logistic regression models with these two independent variables
correctly estimated the probability of lymph node metastases in 44 of 59 t
est cases and the tumor stage of 43 of 59 test cases, respectively. Results
indicated the usefulness of probabilistic neural networks in the populatio
n studied, but the findings should be validated with large groups of patien
ts.