Premium
Significant variations in nuclear structure occur between and within Gleason grading patterns 3, 4, and 5 determined by digital image analysis
Author(s) -
Veltri Robert W.,
Marlow Cameron,
Khan Masood A.,
Miller Michael C.,
Epstein Jonathan I.,
Partin Alan W.
Publication year - 2007
Publication title -
the prostate
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 1.295
H-Index - 123
eISSN - 1097-0045
pISSN - 0270-4137
DOI - 10.1002/pros.20614
Subject(s) - receiver operating characteristic , prostate cancer , grading (engineering) , digital image analysis , prostatectomy , logistic regression , digital pathology , pathology , prostate , nuclear medicine , medicine , cancer , biology , computer science , ecology , computer vision
BACKGROUND Alterations in nuclei structure and DNA content captured from Gleason grading patterns 3, 4 and 5 of radical prostatectomy (RP) cases were determined by a computer‐assisted microscope. Quantitative Nuclear Morphometry (QNM) profiles were created to evaluate variability in nuclear structure within each of these grades. METHODS A tissue microarray (TMA) was constructed using RP cases and the prostate cancer (PCa) TMA cores prepared from 20 GG‐3, 9 GG‐4, 10 GG‐5 patterns, and 20 benign cancer‐adjacent cases from RP archival paraffin blocks. Feulgen‐stained nuclei were captured from 0.6 mm spots using the AutoCyte™ system. Pools of 1100 nuclei captured from each test group were used to calculate Multivariate Logistic Regression (MLR) models that generated predictive indices and predictive probabilities (PP) to make comparisons between and within each set of pooled nuclei. RESULTS A single QNM profiles yielded areas of receiver operator characteristic curves (ROC) that distinguished differences among benign cancer‐adjacent nuclei and GG‐3 (ROC‐AUC = 0.78); GG‐4 (ROC‐AUC = 0.86) and GG‐5 (ROC‐AUC = 0.88) with accuracies of 73%, 78% and 80% respectively. Applying PP plots generated from MLR models of GG 3, 4, and 5 nuclei clearly demonstrated marked heterogeneity within each of these three GG patterns. CONCLUSIONS QNM signatures illustrate alterations in nuclei structure, based upon nuclear morphometry within each of these three GG patterns, and might signify potential variations in PCa disease risk of progression outcomes. In the future a modified system of Gleason grading that considers not only glandular architecture but also quantitative nuclear grade may ensure accuracy in prognosis. Prostate 67: 1202–1210, 2007. © 2007 Wiley‐Liss, Inc.