Concordância intra e interobservador da avaliação de microinvasão angiolinfática em carcinoma hepatocelular utilizando os métodos de histologia convencional e imunoistoquímica

Abstract

Introduction: Hepatocellular carcinoma (HCC) is the most common primary neoplasia of the liver (up to 80% of the cases), being the 4th in mortality worldwide. Its incidence and mortality are growing and the best treatment option is surgery, reserved mainly for inicial stage tumors. The angiolymphatic invasion is the most important independent factor for prognosis, recurrence prediction and disease-free survival of the patient. Because of its relevance, the AJCC/TNM pathological staging system includes vascular invasion as a criteria, playing importante role to change from pT1 stage to pT2 or even pT4, but it still needs better conceptual definition to increase its reproducibility and reliability. The aim of this study was to evaluate the inter and intraobserver agreement of microvascular invasion (MVI) and its association with morphological and laboratorial parameters in samples of HCC. Methods: Three pathologists performed the analysis of 76 HCC explants to characterize the presence of MVI using the hematoxylin eosin (HE) stain and immunohistochemistry (IHC) for CD34 and D2-40. The evaluations were made individually, in two moments, with a two months gap. Morphological data were collected by reviewing all cases and laboratorial data were collected by reviewing medical records and database analysis from HC-UFMG. Results were analyzed by the Kappas coeficiente (K) and logistic regression models for uni and multivariate analyzes. Results: By logistic regression models of the univariate analyzes of all morphological and laboratorial data, the independente variable mitotic index was the only to present threshold of statistical significance (p=0.05) with Odds Ratio of 1.07 (CI 95% of 0.99 to 1.15) when related to MVI. About the MVI, the interobserver agreement by HE was moderate in the first attempt (K=0.485) and fair (0.320) in the seccond (p<0.0001). The CD34 evaluation revealed fair agreement (0.234 and 0.218) in both attempts (p<0.001). The MVI search by D2-40 didnt present statistical significance, since there was no agreement (p>0,05). The intraobserver agreement by HE was almost perfect (K=0.837) for examiner A, moderate (K=0.546) for examiner B and substantial (K=0.682) for examiner C (p<0.0001). The CD34 evaluation revealed substantial agreement (0.636 and 0.629) for examiners A and B and almost perfect agreement (0.972) for examiner C (p<0.0001). There was no valid statistic correlation for D2-40. Using the ROC Curve to verify the acuracy of CD34 in relation to HE we note in intraobserver analysis for examiners A and B that CD34 tends to a high specificity, ranging from 82.1 to almost 100%, with sensitivity of 46.9 to 81.1%. In interobserver analysis, CD34 also has a high specificity (84.3 to 85.5%) while its sensitivity is a little shorter (81.2 to 84.3%). Discussion: Our study demonstrated similar agreement for MVI between HE and CD34 methods, regardless the comparison have been made inter or intraobserver. However, the intraobserver agreement values for both methods were higher than the interobserver ones, which allows us to infer that each pathologist employed own criteria to search for invasion, reinforcing the need of standardization. By the analysis of the ROC Curves, the CD34 method could be considered more specific than sensitive, pointing to the fact that this method shouldnt be used as a trial exam, but it could be useful in cases previously seen by HE to help confirming MVI. Conclusion: The mitotic index is an independent factor to predict MVI. The MVI evaluation criteria should be improved in HE. The CD34 can be used as auxiliary diagnostic method for MVI