Gray level entropy matrix is a superior predictor than multiplex ELISA in the detection of reactive stroma and metastatic potential of high-grade prostatic adenocarcinoma.

Recent reports have indicated that not only the primary glandular tissue but also the surrounding stromal tissue plays an active role in the progression of carcinoma. Such is true for cancer tissues arising in the prostate. However, the precise role of stromal tissue in benign prostatic hyperplasia (BPH) and prostate adenocarcinoma is not well described. We undertook this current investigation to examine the changes in orientation of the extracellular matrix and correlate with prostatic cancer progression. We used a novel form of image analysis called gray level entropy matrix (GLEM) texture analysis to evaluate morphometric changes in stromal tissues. We used normal prostatic tissue obtained from cadaveric specimen and compared with BPH, prostatic intraepithelium neoplastic, hormone responsive prostatic adenocarcinoma and castration-resistant prostatic adenocarcinoma tissues. GLEM showed higher entropy in disease-resistant prostatic tissues, compared with benign forms of all spectra of pathologically diagnosed prostatic tissues (P < 0.05, ANOVA, between groups). Higher entropy is reflective of the disorganized morphological organization of the stroma, possibly reflecting the reactive matrix. In contrast, ELISA revealed that although individually correlated with the progressive stages of benign and carcinomatous prostatic tissues and trend correlation between groups, intergroup comparisons failed to arrive at statistical significance of comparisons between markers of neovasculogenesis, vascular endothelial growth factor, epithelial-mesenchymal transition (beta1-integrin, E-cadherin, MMP3) and osteogenic metastasis (RANKL and osteoprotegerin). The results of our study demonstrate the potential of GLEM entropy of gray level pixel in providing quasiquantitative estimate of a reactive stroma in advance stages of prostatic adenocarcinoma and thus can be routinely used in clinical decision making.