Table 1.
Clinicopathological characteristics of tissue microarray cases.
Figure 1.
A proliferation signature cluster in prostate cancer.
(A) Unsupervised cluster analysis of prostate cancers (data set from ref. [19]) reveals molecular subtypes of prostate cancer (1, 2 and 3, labeled), and gene-expression features reflecting underlying biological processes. Left, thumbnail heatmap of the cluster analysis. Right, enlarged view of the “proliferation cluster”, with selected genes shown (MKI67, TOP2A and E2F1 in red text, marked by arrow). Red and green expression levels reflect high and low values, respectively (see key). Red filled circles (below) identify lymph node metastases. (B) Overlap matrix of proliferation cluster genes (N = 94) with canonical pathway (CP) gene sets identifies top gene set matches (all significant, P<0.001) all relating to cell-cycle/proliferation. Solid blue fill indicates overlapping membership between proliferation cluster and queried gene sets.
Figure 2.
Immunostaining of Ki-67, TOP2A and E2F1 are prognostic.
(A) Immunostaining of proliferation markers Ki-67, TOP2A and E2F1; representative positive and negative cases shown. (B) Pairwise comparison of immunostain scores across cases. Pearson correlation (R) values shown. (C) Kaplan-Meier analysis of Ki-67 (<5% vs. ≥5% tumor nuclei), TOP2A (<5% vs. ≥5% tumor nuclei) and E2F1 (<50% vs. ≥50% tumor nuclei) immunostaining. P-values (log-rank test) shown. (D) Kaplan-Meier analysis of combined marker staining (see keys). P-values (log-rank test) shown.
Table 2.
Association of proliferation markers with clinicopathologic variables.
Table 3.
Receiver-operating characteristic (ROC) curve analysis.
Figure 3.
Incorporation of proliferation index improves prognostic value.
(A) Kaplan-Meier analysis comparing multivariate models based on clinicopathologic data (left) and clinicopathologic data plus the tri-marker proliferation index (right). Cases are grouped by tertile. Log-rank test P-values are indicated. (B) ROC curve analysis comparing multivariate models based on clinicopathologic data (left) and clinicopathologic data plus the tri-marker proliferation index (right). Analysis done at 8 years follow-up (the median follow-up time for the cohort). Areas under the curve (AUC) are indicated.
Table 4.
Analysis of PSA recurrence-free survival.