Related Experiment Video
Updated: Aug 19, 2026
Microarray-based Identification of Individual HERV Loci Expression: Application to Biomarker Discovery in Prostate Cancer
Published on: November 3, 2013
Quantitative alterations in nuclear structure predict prostate carcinoma distant metastasis and death in men with
Masood A Khan1, Patrick C Walsh, M Craig Miller
1The James Buchanan Brady Urological Institute, The Johns Hopkins University School of Medicine, Baltimore, Maryland 21287, USA.
Background:
Microscopic histologic grade has been the best predictor of prostate carcinoma (PCa) progression in men after surgical therapy. The ability to predict accurately, at the time of surgery, which patients are likely to develop metastatic PCa would enable optimization of disease management with adjuvant therapy. The authors assessed the ability of pathologic, nuclear morphometric, and chromatin parameters to predict metastatic PCa progression and/or death in 227 men with biochemical recurrence and long-term follow-up after undergoing radical prostatectomy.
Methods:
Multivariate logistic regression (LR) was used to calculate quantitative nuclear grade (QNG) solutions using the variances of 60 nuclear morphometric descriptors (NMDs) of nuclear size, shape, DNA content, and chromatin organization that predicted distant metastasis and/or PCa-specific death. An LR model also was generated to predict this outcome using a combination of pathologic variables and the best QNG solution. Cox proportional hazards models were generated, and Kaplan-Meier plots were used to display three risk groups based on pathology, QNG, and a combination of these variables.
Results:
A multivariate LR model using pathology retained lymph node (LN) status, seminal vesicle status, and prostatectomy Gleason score, yielding an area under the curve-receiver operator characteristic (AUC-ROC) of 75% with an accuracy of 59% at 90% sensitivity. The best QNG solution used the variance of 25 NMDs, yielding an AUC-ROC of 84% and an accuracy of 70% at 90% sensitivity. The combined pathology-QNG model retained LN status, prostatectomy Gleason score, and QNG, yielding an AUC-ROC of 86% with an accuracy of 76% at 90% sensitivity. The Cox proportional hazards models produced the following significant univariate and multivariate hazard ratios: QNG, 3.5 and 2.9, respectively; LN, 2.7 and 1.8, respectively; and prostatectomy Gleason score, 2.8 and 2.1, respectively.
Conclusions:
Alterations in the structure of tumor nuclei measured by computer-assisted image analysis were strong predictors of PCa progression and death in men with long-term follow-up who had biochemical recurrence after undergoing radical prostatectomy. QNG solutions can serve as a new supplemental biomarker for accurate prediction of PCa progression at the time of surgery.
