A comparative survival evaluation and assessment of interclassification concordance in adult supratentorial astrocytic tumors

Abstract

Classification and grading of astrocytic tumors has been the subject of several controversies and no universally accepted classification system is yet available. Nevertheless, acceptance of a common system is important for assessing prognosis as well as easy comparative evaluation and interpretation of the results of multi-center therapeutic trials. We report the results of a single center study on comparative survival evaluation along with assessment of interclassification concordance in 102 cases of supratentorial astrocytic tumors in adults (≥16 years of age). Hematoxylin and eosin (H&E) stained slides of these 102 cases were reviewed independently by two pathologists and each case classified or graded according to four different classification systems viz. Kernohan, Daumas-Duport (SAM-A), TESTAST-268 and WHO. The histological grading was then correlated with the survival curves as estimated by the Kaplan-Meier method. The most important observation was that similar survival curves were obtained for any one grade of tumor by all the four classification systems. Fifty three of the 102 cases (51.9%) showed absolute grading concordance using all 4 classifications with maximum concordant cases belonging to grades 2 and 4. Intra-classification grade-wise survival analysis revealed a statistically significant difference between grade 2 and grades 3 or 4, but no difference between grades 3 and 4 in any of the classification systems. It is apparent from the results of this study that if specified criteria related to any of the classification systems is rigorously adhered to, it will produce comparable results. Hence, preferential adoption of any one classification system in practice will be guided by the relative ease of histologic feature value evaluation with maximum possible objectivity and reproducibility. We recommend the Daumas-Duport (SAM-A) system since it appears to be the simplest, most objectivized for practical application and highly reproducible with relative ease.