A Computerized, Operator-Independent Morphometric Model for the Histological Assessment of Adrenal Neoplasia: Preliminary Data

Specific biomarkers for diagnosis and prognosis of adrenocortical carcinomas (ACCs) are still lacking.(1) The proliferation marker Ki-67 is under evaluation for validation as a reliable IHC marker of malignancy for ACCs.(2) Although a well defined diagnostic cutoff has not been demonstrated yet, Ki-67 staining between 5-7% is thought to be a feature of AT malignant behavior.(3) As for the morphological markers currently used in the determination of AT malignancy, also for Ki-67 an interobserver variability is frequently reported. The aim of this pilot study was to determine, with an operator independent and repeatable technique, the proliferative activity and cell morphology of AT histological samples; a preliminary comparison with visual perception was also attempted. METHODS IHC study of 14 ACCs and 7 adrenal adenomas (AAs) samples was done (Rabbit anti-human Ki-67 monoclonal Ab, Clone SP6, 1:400); two pathologists performed a blinded examination with a consensus determination and according to Weiss score. We generated a computerized morphometric model on AT slices to evaluate the volume fractions occupied by nuclei (nVF) and cytoplasm (cVF) of both Ki-67(+)/Ki-67(-) neoplastic cells, and by inflammatory infiltrate. All morphometric variables were obtained by a computerized image analyzer (Kontron-Zeiss KS400) with a color camera attached to a light microscope (40x objective) for microscopic fields examination. More than 200 fields were systematically selected and examined by an automatically controlled procedure to assure unbiased sampling. The method of point counting by using a counting frame was used to determine the relative volume proportion of the investigated structures. Statistics: t-test. RESULTS nVF (p<0.0001), cVF (p<0.0001), nuclear/cytoplasmic ratio (p<0.0001) were the most discriminatory features between ACCs/AAs. Moreover nVF of Ki-67(+) cells was more prominent in ACCs than in AAs (p<0.0001) but in some cases an overlapping was still evident. When compared with computerized analysis, Ki-67 pathologist's evaluation in ACCs showed a marked overestimation and a wide range (average of 9,9% vs 1,4%, range 1-25 % and 0,2-4,2% respectively; p=0,00036). CONCLUSIONS Based on these preliminary data our computerized morphometric method could be considered in the future as an helpful tool for pathologists in the histological assessment of ACCs/AAs, particularly minimizing subjectivity and possibly integrating the actual morphologic Weiss criteria.