Background:

A significant fraction of prostate cancer patients experience post–radical prostatectomy (RP) biochemical recurrence (BCR). New predictive markers are needed for optimizing postoperative prostate cancer management. STAT5 is an oncogene in prostate cancer that undergoes amplification in 30% of prostate cancers during progression.

Methods:

We evaluated the significance of a positive status for nuclear STAT5 protein expression versus STAT5 locus amplification versus combined positive status for both in predicting BCR after RP in 300 patients.

Results:

Combined positive STAT5 status was associated with a 45% disadvantage in BCR in Kaplan–Meier survival analysis in all Gleason grade patients. Patients with Gleason grade group (GG) 2 and 3 prostate cancers and combined positive status for STAT5 had a more pronounced disadvantage of 55% to 60% at 7 years after RP in univariate analysis. In multivariate analysis, including the Cancer of the Prostate Risk Assessment Postsurgical nomogram (CAPRA-S) variables, combined positive STAT5 status was independently associated with a shorter BCR-free survival in all Gleason GG patients (HR, 2.34; P = 0.014) and in intermediate Gleason GG 2 or 3 patients (HR, 3.62; P = 0.021). The combined positive STAT5 status improved the predictive value of the CAPRA-S nomogram in both ROC-AUC analysis and in decision curve analysis for BCR.

Conclusions:

Combined positive status for STAT5 was independently associated with shorter disease-free survival in univariate analysis and was an independent predictor for BCR in multivariate analysis using the CAPRA-S variables in prostate cancer.

Impact:

Our results highlight potential for a novel precision medicine concept based on a pivotal role of STAT5 status in improving selection of prostate cancer patients who are candidates for early adjuvant interventions to reduce the risk of recurrence.

Prostate cancer is among the three most common cancers in men worldwide and the second leading cause of cancer-related deaths of men in North America (1). The standard treatment for organ-confined prostate cancer is surgery or radiotherapy (1). After radical prostatectomy (RP), approximately 30% to 60% of patients with intermediate-risk prostate cancer experience biochemical disease recurrence (BCR) as evidenced by rising postoperative serum PSA levels (2). BCR is a surrogate marker of disease progression and triggers follow-up treatments aiming to prevent clinical prostate cancer recurrence. Although postprostatectomy PSA is a specific indicator of disease status, by the time it becomes detectable, it is likely that significant regrowth of prostate cancer has already occurred.

Risk stratification of prostate cancer patients at the time of prostatectomy aims to identify patients at high risk of prostate cancer mortality who are likely to experience BCR and would require early multimodal therapy versus those who are at a relatively low risk and might be spared from the potential impact of aggressive treatment on quality of life. A number of clinical variables are predictive for recurrence risk. The histologic grade of prostate cancer itself provides prognostic information at the extremes of the scale where Gleason Grade Group (GG) 1 prostate cancers are at a low risk and Gleason GG 4 and 5 prostate cancers are at a high risk (3). However, the prognostic information of Gleason GGs 2 and 3 is limited and, consequently, there is a great need for potent predictive biomarkers for identification of patients undergoing RP whose intermediate-risk organ-confined prostate cancer is likely to progress early. A number of models for predicting postprostatectomy BCR have been published including the D'Amico risk stratification scheme (4), the Waltz nomogram (5), Stephenson nomogram (6), Suardi nomogram (7), and the Cancer of the Prostate Risk Assessment Postsurgical nomogram (CAPRA-S; refs. 8–10). CAPRA-S score has been shown to outperform Stephenson nomogram and D'Amico scheme in terms of calibration and decision curve analysis (DCA; refs. 11, 12). Key predictive variables of the CAPRA-S nomogram for postprostatectomy BCR include preoperative PSA, pathologic Gleason score, extraprostatic extension, and seminal vesicle invasion (13). CAPRA-S score has been demonstrated to predict rapid PSA doubling time (14) as well as metastatic progression and disease-specific mortality in prostate cancer patients (15). However, improvement of the prediction models for better identification of patients at increased risk for BCR would allow initiation of adjuvant treatment strategies in association with minimal residual disease volume in order to achieve greater curative benefit to avoid development of metastatic disease and death due to prostate cancer.

We have shown that active STAT5 protein is a critical driver of prostate cancer progression by sustaining cancer cell viability in vitro and by inducing prostate cancer growth in vivo when grown as xenograft tumors in mice (16–19). In addition, active STAT5 triggers epithelial–mesenchymal transition in prostate cancer cells and induces metastatic dissemination of prostate cancer in preclinical models (20). STAT5 is comprised of two highly homologous isoforms STAT5a and STAT5b, which act both as signaling proteins and nuclear transcription factors and are encoded by separate juxtaposed genes on chromosome 17q21 (21, 22). Our previous studies demonstrated that nuclear STAT5 levels are high in prostate cancer tissues but not in normal prostate epithelium (21). We further showed that STAT5 gene locus undergoes amplification during prostate cancer progression to metastatic castrate-resistant prostate cancer, and that STAT5 locus amplification results in increased STAT5 protein levels in prostate cancer (21). Based on analyses of three independent cohorts, we demonstrated that elevated levels of active nuclear STAT5 protein in prostate cancer at intent-to-cure RP predicted disease recurrence and early prostate cancer–specific death (23, 24). These findings corroborate the role of STAT5 in growth and metastatic behavior of prostate cancer shown in preclinical models.

In the present study, we evaluated the predictive value of a positive status for STAT5 locus amplification versus nuclear STAT5 protein expression versus combined positive status for both STAT5 gene locus amplification and nuclear STAT5 protein expression for BCR after RP in univariate and multivariate analyses. Patients with combined positive status for STAT5 gene amplification and nuclear protein overexpression suffered a 45% disadvantage in BCR in Kaplan–Meier (KM) survival analysis compared with patients with negative status for STAT5. Importantly, patients with Gleason GG 2 and 3 prostate cancers and a combined positive status for STAT5 had even a more pronounce disadvantage of 55% to 60% at 7 years after RP in univariate analysis. We further evaluated the predictive value of combined positive STAT5 status as an additional variable in the CAPRA-S nomogram (8) which is currently used for risk assessment for BCR following RP. Our results show that when combined positive STAT5 status was added to the CAPRA-S variables in multivariate analysis, a positive status for STAT5 was a significant independent predictor for BCR. Moreover, combined positive status for STAT5 improved the ROC-AUC analysis and the DCA (25) in disease-free survival in all Gleason GG patients and in Gleason GG 2 and 3 patients. This work introduces a novel concept that patients with combined positive STAT5 status for both locus amplification and nuclear protein expression at RP may be at an increased risk of prostate cancer recurrence post-RP, which needs to be evaluated in larger cohorts. Incorporation of STAT5 status into the evaluation of surgically removed prostate cancer may significantly improve the selection of patients with the potential to benefit from early adjuvant therapy following RP.

Patient characteristics

Our patient cohort included 532 patients having undergone RP at the Turku University Hospital, Finland, during years 2000–2005 and analyzed as a subcohort in our previous study for nuclear STAT5 protein expression status (24). The study protocol was approved by the Ethics Committee of the University of Helsinki (Helsinki, Finland), and the National Data Protection Ombudsman (Helsinki, Finland) was notified about the collection of the information. After excluding patients who received neoadjuvant treatments (the exclusion criteria included any neoadjuvant treatments prior to RP), complete relapse-free survival–related clinical data and follow-up information, RP tissue materials for rereview and tissue microarray (TMA) construction were available for 457 patients (Table 1). The cohort for BCR did not include any patients who received salvage radiotherapy. The cohort was not controlled for other post-RP adjuvant therapies. Complete CAPRA-S variables were available for 396 patients, STAT5 protein expression status for 417 patients, and STAT5 gene locus amplification status for 376 patients. Complete information for CAPRA-S and STAT5 variables was available for 300 patients (Table 1). The 300 patients were comparable in clinicopathologic variables with the remaining 157 patients from the original study group as demonstrated in Table 1 and compared by Student t test and Fisher exact test. Approximately 24% of patients in both the entire cohort and in the subcohort of 300 patients with complete CAPRA-S and STAT5 information experienced BCR. Univariate Cox regression analyses of this subcohort of 300 patients demonstrate that PSA above 6, positive surgical margins, seminal vesicle involvement, Gleason GG above 1, extracapsular extension, and lymph node involvement, each predicted shorter BCR-free survival (Supplementary Table S1), as expected, thus validating the use of the subcohort.

Table 1.

Demographics of the study cohort

Total cohort (n = 457)Patients without complete data (n = 157)Patients with complete data (n = 300)P value
Age at RP, years (mean, SD; n = 457) 62.0 (5.8) 61.62 (6.0) 61.65 (5.6) 0.953a 
Preoperative PSA, ng/mL (n, %; n = 457) 
 ≤10.0 294 (69.2) 80 (70.4) 206 (68.7) 0.443b 
 10.1–20.0 96 (22.6) 30 (24.0) 66 (22.0)  
 >20.0 35 (8.2) 7 (5.6) 28 (9.3)  
GG at RP (n, %; n = 457) 
 1 168 (36.8) 65 (41.4) 103 (34.3) 0.294b 
 2 134 (29.3) 48 (30.6) 86 (28.7)  
 3 63 (13.8) 20 (12.7) 43 (14.3)  
 4 70 (15.3) 17 (10.8) 53 (17.7)  
 5 22 (4.8) 7 (4.5) 15 (5.0)  
pT (n, %; n = 440) 
 2 233 (53.0) 72 (51.4) 161 (53.7) 0.785b 
 3a 153 (34.8) 52 (37.1) 101 (33.7)  
 3b 54 (12.2) 16 (11.5) 38 (12.6)  
Lymph node status (n, %; n = 454) 
 Negative 434 (95.6) 150 (97.4) 284 (94.7) 0.230b 
 Positive 20 (4.4) 4 (2.6) 16 (5.3)  
Follow-up time after RP, yr (median, range; n = 457) 9.5 (0.2–14.0) 10.17 (0.6–13.99) 9.52 (0.73–13.97) 0.006a 
BCR (n, %; n = 457) 110 (24.1) 37 (23.7) 73 (24.3) 0.909b 
Death from any cause (n, %; n = 457) 73 (16.0) 21 (13.4) 52 (17.3) 0.286b 
Death from prostate cancer (n, %; n = 457) 19 (4.2) 8 (5.1) 11 (3.7) 0.468b 
Patients receiving secondary therapy after RP (n, %; n = 438) 136 (31.1) 42 (28.0) 94 (32.6) 0.330b 
STAT5 IHC (n, %; n = 417) 
 Negative 254 (60.9) 74 (63.2) 180 (60.0) 0.576b 
 Positive 163 (39.1) 43 (36.8) 120 (40.0)  
STAT5 FISH (n, %; n = 376) 
 Negative 282 (75.0) 61 (80.3) 221 (73.7) 0.300b 
 Positive 94 (25.0) 15 (19.7) 79 (26.3)  
Total cohort (n = 457)Patients without complete data (n = 157)Patients with complete data (n = 300)P value
Age at RP, years (mean, SD; n = 457) 62.0 (5.8) 61.62 (6.0) 61.65 (5.6) 0.953a 
Preoperative PSA, ng/mL (n, %; n = 457) 
 ≤10.0 294 (69.2) 80 (70.4) 206 (68.7) 0.443b 
 10.1–20.0 96 (22.6) 30 (24.0) 66 (22.0)  
 >20.0 35 (8.2) 7 (5.6) 28 (9.3)  
GG at RP (n, %; n = 457) 
 1 168 (36.8) 65 (41.4) 103 (34.3) 0.294b 
 2 134 (29.3) 48 (30.6) 86 (28.7)  
 3 63 (13.8) 20 (12.7) 43 (14.3)  
 4 70 (15.3) 17 (10.8) 53 (17.7)  
 5 22 (4.8) 7 (4.5) 15 (5.0)  
pT (n, %; n = 440) 
 2 233 (53.0) 72 (51.4) 161 (53.7) 0.785b 
 3a 153 (34.8) 52 (37.1) 101 (33.7)  
 3b 54 (12.2) 16 (11.5) 38 (12.6)  
Lymph node status (n, %; n = 454) 
 Negative 434 (95.6) 150 (97.4) 284 (94.7) 0.230b 
 Positive 20 (4.4) 4 (2.6) 16 (5.3)  
Follow-up time after RP, yr (median, range; n = 457) 9.5 (0.2–14.0) 10.17 (0.6–13.99) 9.52 (0.73–13.97) 0.006a 
BCR (n, %; n = 457) 110 (24.1) 37 (23.7) 73 (24.3) 0.909b 
Death from any cause (n, %; n = 457) 73 (16.0) 21 (13.4) 52 (17.3) 0.286b 
Death from prostate cancer (n, %; n = 457) 19 (4.2) 8 (5.1) 11 (3.7) 0.468b 
Patients receiving secondary therapy after RP (n, %; n = 438) 136 (31.1) 42 (28.0) 94 (32.6) 0.330b 
STAT5 IHC (n, %; n = 417) 
 Negative 254 (60.9) 74 (63.2) 180 (60.0) 0.576b 
 Positive 163 (39.1) 43 (36.8) 120 (40.0)  
STAT5 FISH (n, %; n = 376) 
 Negative 282 (75.0) 61 (80.3) 221 (73.7) 0.300b 
 Positive 94 (25.0) 15 (19.7) 79 (26.3)  

aStudent t test.

bFisher exact test.

TMA construction

Slides of the whole-mounted RP specimens were rereviewed by T. Mirtti and M.A. Kallajoki according to the current Gleason grading and WHO/ISUP Grade grouping criteria (26). TMAs were constructed as described previously (24). Briefly, each patient's RP hematoxylin and eosin (H&E) slides were evaluated, and a minimum of three cores (on average 9–10 cores), 1 mm in diameter each, from the dominant cancer focus were transferred to the recipient TMA block. In addition, at least one core representing benign prostate epithelium remote from the cancer site was also selected for each patient. The representativeness of each core was confirmed by H&E staining in a sequential section following the STAT5 IHC and STAT5 FISH analyses.

STAT5 IHC and scoring of the TMAs

IHC was conducted as described previously (16, 23, 24, 27). In brief, for evaluation of nuclear STAT5 protein levels, each TMA core was given a score based on both intensity and proportion of immunostained cells where 0 represented negative, 1 weak, 2 moderate, and 3 strong immunostaining for STAT5. The final score for each patient was the highest score of the individual scores. For example, if a subject had three core scores 0, 1, and 3, the final STAT5 score for that subject would be 3 (24). Representative pictures of IHC scores are shown in Fig. 1A. Scores 0 and 1 were considered negative, whereas scores 2 and 3 represented positive status for nuclear STAT5 protein. These cutoff points for nuclear STAT5 protein levels as a function of survival were defined in three previously analyzed cohorts (23, 24). In the current study, we evaluated the predictive value of the STAT5 locus amplification status alone or in combination with the STAT5 protein status.

Figure 1.

STAT5 protein levels and STAT5 locus amplification in prostate cancer TMAs. A, Individual prostate cancer TMA cores were scored for nuclear STAT5a/b levels, detected by IHC, on a scale from 0 to 3, where 0 represented negative, 1 weak, 2 moderate, and 3 strong immunostaining (top plot). B,STAT5A/B FISH analysis of prostate cancer TMA cores (bottom plot). The top row shows representative sections from 5 cases showing no STAT5 locus amplification. Most cells show 2 copies of STAT5A/B (red signal). A chromosome 17 centromeric probe was used (green signal) as a control. The bottom row shows representative sections from 4 cases with STAT5 locus amplification. Arrows point to cells with STAT5 locus amplification.

Figure 1.

STAT5 protein levels and STAT5 locus amplification in prostate cancer TMAs. A, Individual prostate cancer TMA cores were scored for nuclear STAT5a/b levels, detected by IHC, on a scale from 0 to 3, where 0 represented negative, 1 weak, 2 moderate, and 3 strong immunostaining (top plot). B,STAT5A/B FISH analysis of prostate cancer TMA cores (bottom plot). The top row shows representative sections from 5 cases showing no STAT5 locus amplification. Most cells show 2 copies of STAT5A/B (red signal). A chromosome 17 centromeric probe was used (green signal) as a control. The bottom row shows representative sections from 4 cases with STAT5 locus amplification. Arrows point to cells with STAT5 locus amplification.

Close modal

FISH

Evaluation of STAT5 locus amplification in TMA tissue cores using FISH analysis was performed as described earlier (21). Briefly, we designed a STAT5A/B probe consisting of a contig of four overlapping bacterial artificial chromosome (BAC) clones containing sequences of the STAT5A/B gene: RP11-60B4, RP11-1151C17, RP11-1151G10, and RP11-365D24 (BACPAC Resources). As a control probe, we used a chromosome 17 centromeric BAC clone: RP11-299G20 (BACPAC Resources). FISH analysis of the TMAs was performed using a standard protocol (21). The STAT5A/B probe was detected using red fluorescence and chromosome 17 centromeric control probe using green fluorescence. Scoring of cells and digital image acquisition were performed as described earlier (21). STAT5 locus amplification was defined as a signal ratio of gene probe to control probe ≥2 or five or more copies of the gene signal in ≥10% of the tumor nuclei (21). STAT5 locus amplification status was scored either positive or negative if any of the cores from a given patient was positive for STAT5 locus amplification. Thus, STAT5 locus amplification evaluation did not require a cutoff analysis in a training cohort.

Statistical analyses

BCR was defined as a minimum of two consecutive post-RP PSA measurements of 0.2 ng/mL or higher. Mortality data were registered as death due to prostate cancer or death due to any other cause for disease-specific survival (DSS) and overall survival (OS) analyses, respectively. KM, ROC-AUC, and DCA were run in R version 3.3.2 using the survival, survminer, pROC, and MASS packages.

Nuclear STAT5 levels were determined by IHC (Fig. 1A) and STAT5 gene copy-number status by FISH (Fig. 1B), as described previously (21, 23, 24). The analysis of a cohort of 457 prostate cancers, treated exclusively by RP, yielded complete information for all CAPRA-S and STAT5 variables for 300 patients (Table 1). Because the FISH and IHC were not conducted simultaneously, these two analyses were performed on serial sections of the TMAs and do not represent the FISH/IHC status of the same cells. The maximal patient-wise marker status (either STAT5 protein or STAT5 locus amplification) was used to assess the final patient marker status. Concordance of STAT5 protein status and STAT5 locus amplification was high at patient level (88.9%) and is shown in Supplementary Table S2 (all spots discordant).

KM analysis of disease-free survival of prostate cancer patients with combined positive STAT5 status

The levels of nuclear STAT5a/b expression in prostate cancers were examined by IHC, as described previously (23, 27, 28). Representative images of prostate cancers with no detectable nuclear STAT5a/b (score 0), weak (score 1), moderate (score 2), or strong (score 3) nuclear STAT5a/b expression are shown in Fig. 1A. To identify prostate cancers with the STAT5 gene locus amplification, we used FISH analysis of paraffin embedded tissue cores of localized prostate cancers removed by RP, as described previously (21). Figure 1B shows examples of prostate cancer cases with STAT5a/b diploid pattern (top plot) versus prostate cancers with STAT5a/b gene locus amplification (bottom plot). Optimal cutoff points for nuclear STAT5 protein levels as a function of disease-free survival were defined in three cohorts that we have described previously (23). Scores 0 and 1 represented negative status for STAT5 protein, whereas scores 2–3 were considered positive for nuclear STAT5. The purpose of the current study was to evaluate the predictive value of STAT5 gene locus amplification alone or in combination with positive nuclear STAT5 protein status for BCR of prostate cancer after RP.

The KM analysis showed, for the first time, that combined positive status for both nuclear STAT5 protein expression and STAT5 gene locus amplification (FISH+/IHC+) was associated with 55% to 60% disadvantage in disease-free survival at 7 years after RP in patients with Gleason GG 2 or 3 prostate cancers (Fig. 2). At the same time, combined positive status for STAT5 (FISH+/IHC+) was associated with approximately 45% disadvantage in disease-free survival in patients with prostate cancers of all histologic grades (Fig. 2A–C). Positive status for STAT5 locus amplification alone (STAT5 FISH+) was associated with less significant disadvantage of 20% in disease-free survival at 7 years after RP in patients with all Gleason GG prostate cancers and in patients with Gleason GG 2 or 3 prostate cancers (Fig. 2D–F). In addition, positive STAT5 locus amplification status predicted shorter DSS and OS in Gleason GG 2 or 3 patients (Supplementary Figs. S1 and S2). Consistent with our previous results (23, 24), nuclear STAT5 protein expression alone (STAT5 IHC+) in KM survival analysis predicted shorter BCR-free survival in patients with prostate cancers of all Gleason GG (Fig. 2G–I). Collectively, our results suggest that positive status for STAT5 gene locus amplification, when combined with positive status for nuclear STAT5 protein expression, provides a more potent predictor of BCR than positive status for either STAT5 locus amplification or nuclear protein expression alone for prostate cancer after RP and, especially, for patients with Gleason GGs 2 or 3 prostate cancers. These findings are important because only a limited number of predictive markers of prostate cancer recurrence after RP are currently available for patients with Gleason GG 2 or 3 prostate cancers.

Figure 2.

Univariate analysis of the combined status for both STAT5 gene locus amplification status and protein expression and progression-free survival with BCR as the endpoint in prostate cancers of all Gleason GG (A) versus prostate cancers of Gleason GG 2 and 3 (B) versus GG 4 and 5 (C). Univariate analysis of the STAT5 gene locus amplification status and progression-free survival in prostate cancers of all Gleason GG (D) versus prostate cancers of Gleason GG 2 and 3 (E) versus GG 4 and 5 (F). Univariate analysis of the STAT5 protein status and progression-free survival in prostate cancers of all Gleason GG (G) versus prostate cancers of Gleason GG 2 and 3 (H) versus GG 4 and 5 (I). KM curves, with global P values calculated by Mantel–Haenszel tests. The STAT5 protein status detected by IHC and STAT5 locus amplification by FISH. Global Mantel–Haenszel log-rank P values were calculated comparing the difference in survival between all patient groups.

Figure 2.

Univariate analysis of the combined status for both STAT5 gene locus amplification status and protein expression and progression-free survival with BCR as the endpoint in prostate cancers of all Gleason GG (A) versus prostate cancers of Gleason GG 2 and 3 (B) versus GG 4 and 5 (C). Univariate analysis of the STAT5 gene locus amplification status and progression-free survival in prostate cancers of all Gleason GG (D) versus prostate cancers of Gleason GG 2 and 3 (E) versus GG 4 and 5 (F). Univariate analysis of the STAT5 protein status and progression-free survival in prostate cancers of all Gleason GG (G) versus prostate cancers of Gleason GG 2 and 3 (H) versus GG 4 and 5 (I). KM curves, with global P values calculated by Mantel–Haenszel tests. The STAT5 protein status detected by IHC and STAT5 locus amplification by FISH. Global Mantel–Haenszel log-rank P values were calculated comparing the difference in survival between all patient groups.

Close modal

The predictive value of combined positive STAT5 status in prostate cancer in relation to CAPRA-S variables

To further investigate the independent predictive value of combined positive status for both nuclear STAT5 protein expression and gene locus amplification, we performed multivariate Cox regression analyses that included the CAPRA-S variables (ref. 8; Tables 2A2C). Table 2A demonstrates multivariate analysis of the predictive value of the CAPRA-S variables alone without the STAT5 status for BCR in this cohort of 300 patients with complete STAT5 and CAPRA-S information showing the expected risk stratification by the CAPRA-S variables in this cohort. When both the STAT5 status and the CAPRA-S variables were included in the multivariate analysis, combined positive status for STAT5 was associated with a 2.34-fold increased risk of BCR compared with patients with combined negative status for STAT5 (FISH-/IHC-) in prostate cancer [P, 0.014; 95% confidence interval (CI), 1.184–4.625; Table 2B]. At the same time, patients having positive status for nuclear STAT5 protein expression alone in prostate cancer were at a 1.88-fold increased risk of BCR compared with patients negative for STAT5 protein (P = 0.049; Table 2B). When patients with Gleason GG 2 or 3 prostate cancer were analyzed separately, combined positive status for STAT5 was associated with the highest risk of BCR (a 3.62-fold increased risk) in multivariate analysis (P, 0.021; 95% CI, 1.22–10.78; Table 2C). Importantly, in Gleason GG 2 and 3 prostate cancers, combined positive STAT5 status was a stronger predictor of BCR than any of the clinical parameters of the CAPRA-S nomogram alone (Table 2C). These data indicate that combined positive STAT5 status of prostate cancer, even when accounting for clinical variables in the CAPRA-S nomogram, was independently associated with increased risk of BCR after RP.

Table 2A.

Multivariate Cox regression analysis of the risk of BCR by CAPRA-S variables in all prostate cancer patients with complete data (n = 300)

VariableLevelHR95% CIP value
PSA 0–6 Reference — — 
 6.01–10 2.673 (1.161–6.153) 0.021 
 10.01–20 2.996 (1.261–7.12) 0.013 
 >20 8.059 (3.14–20.689) <0.001 
SM Negative Reference  — 
 Positive 0.563 (0.32–0.99) 0.046 
SVI No Reference  — 
 Yes 0.921 (0.487–1.743) 0.802 
GG Reference  — 
 2–3 1.697 (0.826–3.485) 0.15 
 4–5 3.283 (1.577–6.835) 0.001 
EPE No Reference  — 
 Yes 3.071 (1.714–5.5) <0.001 
LNI No Reference  — 
 Yes 1.495 (0.691–3.234) 0.307 
VariableLevelHR95% CIP value
PSA 0–6 Reference — — 
 6.01–10 2.673 (1.161–6.153) 0.021 
 10.01–20 2.996 (1.261–7.12) 0.013 
 >20 8.059 (3.14–20.689) <0.001 
SM Negative Reference  — 
 Positive 0.563 (0.32–0.99) 0.046 
SVI No Reference  — 
 Yes 0.921 (0.487–1.743) 0.802 
GG Reference  — 
 2–3 1.697 (0.826–3.485) 0.15 
 4–5 3.283 (1.577–6.835) 0.001 
EPE No Reference  — 
 Yes 3.071 (1.714–5.5) <0.001 
LNI No Reference  — 
 Yes 1.495 (0.691–3.234) 0.307 

Abbreviations: CI, confidence interval; EPE, extraprostatic extension; LNI, lymph node involvement; SM, surgical margin; SVI, seminal vesicle involvement.

Table 2B.

Multivariate Cox regression analysis of the risk of BCR by CAPRA-S variables and STAT5 status in all prostate cancer patients with complete data (n = 300)

VariableLevelHR95% CIP value
PSA 0–6 Reference — — 
 6.01–10 2.559 (1.099–5.960) 0.029 
 10.01–20 2.776 (1.165–6.614) 0.021 
 >20 7.256 (2.776–18.964) <0.001 
SM Negative Reference  — 
 Positive 0.544 (0.304–0.972) 0.04 
SVI No Reference  — 
 Yes 0.98 (0.522–1.840) 0.95 
GG Reference  — 
 2–3 1.812 (0.878–3.740) 0.108 
 4–5 2.905 (1.381–6.111) 0.005 
EPE No Reference  — 
 Yes 2.934 (1.624–5.301) <0.001 
LNI No Reference  — 
 Yes 1.471 (0.677–3.196) 0.33 
STAT5 STAT5 FISH−/IHC− Reference  — 
 STAT5 FISH+/IHC− 1.813 (0.875–3.754) 0.109 
 STAT5 FISH−/IHC+ 1.876 (1.003–3.508) 0.049 
 STAT5 FISH+/IHC+ 2.340 (1.184–4.625) 0.014 
VariableLevelHR95% CIP value
PSA 0–6 Reference — — 
 6.01–10 2.559 (1.099–5.960) 0.029 
 10.01–20 2.776 (1.165–6.614) 0.021 
 >20 7.256 (2.776–18.964) <0.001 
SM Negative Reference  — 
 Positive 0.544 (0.304–0.972) 0.04 
SVI No Reference  — 
 Yes 0.98 (0.522–1.840) 0.95 
GG Reference  — 
 2–3 1.812 (0.878–3.740) 0.108 
 4–5 2.905 (1.381–6.111) 0.005 
EPE No Reference  — 
 Yes 2.934 (1.624–5.301) <0.001 
LNI No Reference  — 
 Yes 1.471 (0.677–3.196) 0.33 
STAT5 STAT5 FISH−/IHC− Reference  — 
 STAT5 FISH+/IHC− 1.813 (0.875–3.754) 0.109 
 STAT5 FISH−/IHC+ 1.876 (1.003–3.508) 0.049 
 STAT5 FISH+/IHC+ 2.340 (1.184–4.625) 0.014 

Abbreviations: CI, confidence interval; EPE, extraprostatic extension; LNI, lymph node involvement; SM, surgical margin; SVI, seminal vesicle involvement.

Table 2C.

Multivariate Cox regression analysis of the risk of BCR by CAPRA-S variables and STAT5 status in patients with Gleason GG 2 or 3 prostate cancer (n = 129)

VariableLevelHR95% CIP value
PSA 0–6 Reference — — 
 6.01–10 1.461 (0.41–5.24) 0.560 
 10.01–20 2.442 (0.71–8.42) 0.157 
 >20 10.422 (2.69–40.43) 0.001 
SM Negative Reference — — 
 Positive 0.418 (0.15–1.17) 0.095 
SVI No Reference — — 
 Yes 0.800 (0.27–2.41) 0.692 
EPE No Reference — — 
 Yes 2.571 (0.97–6.82) 0.058 
LNI No Reference — — 
 Yes 1.189 (0.23–6.05) 0.834 
STAT5 STAT5 FISH−/IHC− Reference — — 
 STAT5 FISH+/IHC− 2.442 (0.79–7.52) 0.120 
 STAT5 FISH−/IHC+ 2.543 (0.91–7.07) 0.074 
 STAT5 FISH+/IHC+ 3.623 (1.22–10.78) 0.021 
VariableLevelHR95% CIP value
PSA 0–6 Reference — — 
 6.01–10 1.461 (0.41–5.24) 0.560 
 10.01–20 2.442 (0.71–8.42) 0.157 
 >20 10.422 (2.69–40.43) 0.001 
SM Negative Reference — — 
 Positive 0.418 (0.15–1.17) 0.095 
SVI No Reference — — 
 Yes 0.800 (0.27–2.41) 0.692 
EPE No Reference — — 
 Yes 2.571 (0.97–6.82) 0.058 
LNI No Reference — — 
 Yes 1.189 (0.23–6.05) 0.834 
STAT5 STAT5 FISH−/IHC− Reference — — 
 STAT5 FISH+/IHC− 2.442 (0.79–7.52) 0.120 
 STAT5 FISH−/IHC+ 2.543 (0.91–7.07) 0.074 
 STAT5 FISH+/IHC+ 3.623 (1.22–10.78) 0.021 

Abbreviations: CI, confidence interval; EPE, extraprostatic extension; LNI, lymph node involvement; SM, surgical margin; SVI, seminal vesicle involvement.

Given that data from both univariate and multivariate survival analyses supported a new concept that combined positive STAT5 status identifies a group of prostate cancer patients with the highest risk of recurrence, we next sought to assess whether the combined positive STAT5 status adds predictive value to the CAPRA-S nomogram in ROC-AUC analyses (Fig. 3A; Supplementary Fig. S3; Supplementary Table S3). In ROC-AUC analyses of all prostate cancer patients (Table 3A), positive status for either STAT5 IHC or STAT5 FISH, each, added predictive value alone (both by 0.013; 0.767 minus 0.754). Combined positive status for both STAT5 IHC and FISH added the most predictive value for BCR by 0.026 (0.780 minus 0.754), when compared with CAPRA-S alone (Table 3A; Fig. 3A). When exclusively Gleason GG 2 and 3 prostate cancer patients were evaluated, combined positive STAT5 status added predictive value to CAPRA-S by 0.032 (0.732 minus 0.700) in ROC-AUC analysis (Table 3B; Fig. 3A). To assess potential added value of STAT5 status to treatment decisions of patients, we conducted DCA (Fig. 3B; Supplementary Fig. S4) where models are compared against the percentage likelihood of an outcome event to occur (25). In DCA of both the whole cohort and subanalysis of Gleason GG 2 and 3 patients exclusively, combined STAT5 status with CAPRA-S (as compared with CAPRA-S alone) added clinical benefit across most probabilities of BCR occurring. Specifically, for the whole cohort (Fig. 3B), DCA added benefit in approximately the 10% to 25%, 30% to 40%, and 50% to 70% ranges. In Gleason GG 2 or 3 prostate cancers, DCA added benefit in the 10% to 25%, 30% to 35%, and 50% to 80% ranges (Fig. 3B). Although the decision curves intersect at a number of probabilities, this result suggests that for patients with a low-risk of BCR (<25% risk) assessing combined STAT5 status adds clinical benefit. Taken together, these results suggest that combined positive status for STAT5 protein and gene locus amplification may add value to the CAPRA-S nomogram to predict prostate cancer recurrence after RP, a finding that requires follow-up studies in other cohorts.

Figure 3.

ROC-AUC (A) and DCA of CAPRA-S (B) with and without the combined positive STAT5 status in the entire cohort, and a separate subanalysis of Gleason GG 2 and 3 prostate cancer patients exclusively.

Figure 3.

ROC-AUC (A) and DCA of CAPRA-S (B) with and without the combined positive STAT5 status in the entire cohort, and a separate subanalysis of Gleason GG 2 and 3 prostate cancer patients exclusively.

Close modal
Table 3A.

ROC-AUC values for CAPRA-S and STAT5 status in all prostate cancer patients (n = 300)

ModelBCR95% CIOS95% CIDSS95% CI
CAPRA-S 0.754 (0.694–0.814) 0.608 (0.522–0.694) 0.818 (0.685–0.952) 
CAPRA-S + STAT5 IHC 0.767 (0.710–0.824) 0.605 (0.519–0.692) 0.825 (0.786–0.948) 
CAPRA-S + STAT5 FISH 0.767 (0.709–0.826) 0.621 (0.536–0.705) 0.841 (0.724–0.957) 
CAPRA-S + STAT5 FISH+IHC 0.780 (0.725–0.835) 0.619 (0.534–0.704) 0.849 (0.743–0.956) 
ModelBCR95% CIOS95% CIDSS95% CI
CAPRA-S 0.754 (0.694–0.814) 0.608 (0.522–0.694) 0.818 (0.685–0.952) 
CAPRA-S + STAT5 IHC 0.767 (0.710–0.824) 0.605 (0.519–0.692) 0.825 (0.786–0.948) 
CAPRA-S + STAT5 FISH 0.767 (0.709–0.826) 0.621 (0.536–0.705) 0.841 (0.724–0.957) 
CAPRA-S + STAT5 FISH+IHC 0.780 (0.725–0.835) 0.619 (0.534–0.704) 0.849 (0.743–0.956) 
Table 3B.

ROC-AUC values for CAPRA-S and STAT5 status in patients with Gleason GG 2 or 3 prostate cancer (n = 129)

ModelBCR95% CIOS95% CIDSS95% CI
CAPRA-S 0.700 (0.596–0.805) 0.503 (0.375–0.631) 0.617 (0.365–0.869) 
CAPRA-S + STAT5 IHC 0.718 (0.622–0.814) 0.494 (0.366–0.622) 0.635 (0.635–0.861) 
CAPRA-S + STAT5 FISH 0.715 (0.614–0.816) 0.517 (0.391–0.643) 0.674 (0.449–0.900) 
CAPRA-S + STAT5 FISH+IHC 0.732 (0.639–0.825) 0.517 (0.390–0.645) 0.698 (0.498–0.898) 
ModelBCR95% CIOS95% CIDSS95% CI
CAPRA-S 0.700 (0.596–0.805) 0.503 (0.375–0.631) 0.617 (0.365–0.869) 
CAPRA-S + STAT5 IHC 0.718 (0.622–0.814) 0.494 (0.366–0.622) 0.635 (0.635–0.861) 
CAPRA-S + STAT5 FISH 0.715 (0.614–0.816) 0.517 (0.391–0.643) 0.674 (0.449–0.900) 
CAPRA-S + STAT5 FISH+IHC 0.732 (0.639–0.825) 0.517 (0.390–0.645) 0.698 (0.498–0.898) 

Radical prostatectomy is the mainstay of the treatment of localized prostate cancer. Numerous prediction models based on preoperative and/or postoperative status have been developed to identify patients who are likely to experience post-RP BCR (10) and eventually metastatic progression and prostate cancer–specific death (10, 13). Here, we evaluated the predictive value of combined positive status for both STAT5 locus amplification and nuclear protein expression in prostate cancers at the time of RP in relation to CAPRA-S nomogram for BCR in the cohort of 300 prostate cancers. Our data show that combined positive status for STAT5 in this cohort was independently associated with shorter disease-free survival in univariate analysis and was an independent predictor for BCR in multivariate analysis using the CAPRA-S variables in prostate cancer. In Gleason GG 2 and 3 prostate cancers, combined positive STAT5 status was a stronger predictor of BCR than any other clinical variables of the CAPRA-S nomogram. In addition, combined positive STAT5 status added predictive value to the CAPRA-S nomogram in ROC-AUC and DCA analyses in the cohort analyzed here.

A key finding of this study is that combined positive STAT5 status added predictive value to the CAPRA-S nomogram for BCR in this cohort of prostate cancers treated by RP. Future studies are needed in independent larger cohorts to validate the finding of the present study. Moreover, it will be crucial to evaluate whether combined positive status for STAT5 in prostate cancer at RP predicts development of metastatic prostate cancer and prostate cancer–specific mortality in patients experiencing BCR in relation to the CAPRA-S nomogram. This will require a larger patient cohort with enough events occurring within the follow-up period. Also, the risk of BCR in patient groups with conflicting CAPRA-S risk versus STAT5 status will need to be specifically interrogated.

Another clinical implication of the findings of the present study is that the patients with combined positive status for STAT5 in prostate cancer at RP may benefit from post-RP adjuvant therapy in the form of chemotherapy, radiotherapy, or antiandrogens. In addition, identification of the JAK2-STAT5 signaling pathway as an independent predictive variable for disease recurrence/progression provides a novel target for innovative adjuvant treatment strategies. Numerous Jak2 inhibitors are currently in phase II trials in myeloproliferative disorders (29) and therefore readily available for evaluation of their efficacy for prostate cancer. Evaluation of nuclear STAT5 protein status by IHC and gene locus amplification by FISH in paraffin-embedded tissue sections of surgically resected prostate cancer are both simple and highly feasible tests that can be easily standardized for clinical pathology laboratories.

Combined positive status for STAT5 gene locus amplification and nuclear STAT5 protein expression in the cohort evaluated here was a more powerful predictor of prostate cancer recurrence than either STAT5 locus amplification or nuclear protein expression alone. This finding suggests that somatic amplifications of the STAT5 locus may not always result in an intact reading frame and expression of a fully functional STAT5 protein in prostate cancer. However, it is important to note that the FISH and IHC were not conducted simultaneously on the same TMA section, but these two analyses were performed on serial sections of the TMAs and, therefore, the data do not represent the FISH/IHC status of the same cells. Furthermore, the STAT5 IHC was analyzed in a four-tier grouping system where scores 0 and 1 were considered negative, whereas scores 2 and 3 were considered positive for nuclear STAT5 protein status. Accordingly, the maximal patient-wise marker status (either STAT5 protein or STAT5 locus amplification) was used to assess the final patient marker status. STAT5 has been shown to be a potent inducer of prostate cancer proliferation, epithelial–mesenchymal transition, and castrate-resistant prostate cancer growth (16–20) and, therefore, the finding of combined positive status for STAT5 locus amplification and protein expression as a predictor of prostate cancer recurrence in patients is consistent with the data on the role of STAT5 in promoting prostate cancer growth and progression in preclinical prostate cancer model systems. Future studies need to evaluate the STAT5 protein and locus amplification status simultaneously in the same sections.

In conclusion, the results of this study introduce a new concept that combined positive status for both STAT5 protein expression and gene amplification in prostate cancer at the time of RP is a predictor of BCR in multivariate analysis that includes the variable of the CAPRA-S nomogram. These findings will need to be investigated further in larger cohorts in future studies. Evaluation of STAT5 protein status by IHC and gene locus amplification by FISH in paraffin-embedded tissue sections of surgically resected prostate cancer are both simple and highly feasible tests that can be easily standardized for clinical pathology laboratories. By identifying patients at increased risk for treatment failure, the incorporation of STAT5 status in the evaluation of surgically removed prostate cancer at RP would allow for improved patient counseling and identification of patients likely to benefit from early adjuvant therapy following surgery.

No potential conflicts of interest were disclosed.

Conception and design: B.R. Haddad, A. Erickson, M.A. Kallajoki, T. Mirtti, M.T. Nevalainen

Development of methodology: B.R. Haddad, T. Mirtti, M.T. Nevalainen

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): B.R. Haddad, V. Udhane, J.D. Rone, M.A. Kallajoki, A. Rannikko, T. Mirtti, M.T. Nevalainen

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): B.R. Haddad, A. Erickson, J.D. Rone, M.A. Kallajoki, T. Mirtti, M.T. Nevalainen

Writing, review, and/or revision of the manuscript: B.R. Haddad, A. Erickson, P.S. LaViolette, M.A. Kallajoki, W.A. See, A. Rannikko, T. Mirtti, M.T. Nevalainen

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): A. Erickson, V. Udhane, A. Rannikko, T. Mirtti

Study supervision: M.A. Kallajoki, T. Mirtti, M.T. Nevalainen

This work was supported in part by grants from the NIH/NCI to M.T. Nevalainen, B.R. Haddad, and J.D. Rone (7R01CA113580-10, 5R21CA178755-02, and AHW 5520368), from Finnish Medical Foundation (T. Mirtti), Academy of Finland (T. Mirtti and A. Erickson), and Finnish Cancer Society (T. Mirtti) and P.S. LaViolette (R01CA218144 and R21CA231892).

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

1.
Attard
G
,
Parker
C
,
Eeles
RA
,
Schroder
F
,
Tomlins
SA
,
Tannock
I
, et al
Prostate cancer
.
Lancet
2016
;
387
:
70
82
.
2.
Grossfeld
GD
,
Latini
DM
,
Lubeck
DP
,
Mehta
SS
,
Carroll
PR
. 
Predicting recurrence after radical prostatectomy for patients with high risk prostate cancer
.
J Urol
2003
;
169
:
157
63
.
3.
Epstein
JI
,
Zelefsky
MJ
,
Sjoberg
DD
,
Nelson
JB
,
Egevad
L
,
Magi-Galluzzi
C
, et al
A contemporary prostate cancer grading system: a validated alternative to the Gleason score
.
Eur Urol
2016
;
69
:
428
35
.
4.
D'Amico
AV
,
Whittington
R
,
Malkowicz
SB
,
Fondurulia
J
,
Chen
MH
,
Tomaszewski
JE
, et al
The combination of preoperative prostate specific antigen and postoperative pathological findings to predict prostate specific antigen outcome in clinically localized prostate cancer
.
J Urol
1998
;
160
(
6 Pt 1
):
2096
101
.
5.
Walz
J
,
Chun
FK
,
Klein
EA
,
Reuther
A
,
Saad
F
,
Graefen
M
, et al
Nomogram predicting the probability of early recurrence after radical prostatectomy for prostate cancer
.
J Urol
2009
;
181
:
601
7
;
discussion 7–8
.
6.
Stephenson
AJ
,
Scardino
PT
,
Eastham
JA
,
Bianco
FJ
 Jr
,
Dotan
ZA
,
DiBlasio
CJ
, et al
Postoperative nomogram predicting the 10-year probability of prostate cancer recurrence after radical prostatectomy
.
J Clin Oncol
2005
;
23
:
7005
12
.
7.
Suardi
N
,
Porter
CR
,
Reuther
AM
,
Walz
J
,
Kodama
K
,
Gibbons
RP
, et al
A nomogram predicting long-term biochemical recurrence after radical prostatectomy
.
Cancer
2008
;
112
:
1254
63
.
8.
Cooperberg
MR
,
Hilton
JF
,
Carroll
PR
. 
The CAPRA-S score: a straightforward tool for improved prediction of outcomes after radical prostatectomy
.
Cancer
2011
;
117
:
5039
46
.
9.
Cooperberg
MR
,
Pasta
DJ
,
Elkin
EP
,
Litwin
MS
,
Latini
DM
,
Du Chane
J
, et al
The University of California, San Francisco Cancer of the Prostate Risk Assessment score: a straightforward and reliable preoperative predictor of disease recurrence after radical prostatectomy
.
J Urol
2005
;
173
:
1938
42
.
10.
Lughezzani
G
,
Briganti
A
,
Karakiewicz
PI
,
Kattan
MW
,
Montorsi
F
,
Shariat
SF
, et al
Predictive and prognostic models in radical prostatectomy candidates: a critical analysis of the literature
.
Eur Urol
2010
;
58
:
687
700
.
11.
Vickers
A
. 
Prediction models in urology: are they any good, and how would we know anyway?
Eur Urol
2010
;
57
:
571
3
;
discussion 4
.
12.
Lughezzani
G
,
Budaus
L
,
Isbarn
H
,
Sun
M
,
Perrotte
P
,
Haese
A
, et al
Head-to-head comparison of the three most commonly used preoperative models for prediction of biochemical recurrence after radical prostatectomy
.
Eur Urol
2010
;
57
:
562
8
.
13.
Brockman
JA
,
Alanee
S
,
Vickers
AJ
,
Scardino
PT
,
Wood
DP
,
Kibel
AS
, et al
Nomogram predicting prostate cancer-specific mortality for men with biochemical recurrence after radical prostatectomy
.
Eur Urol
2015
;
67
:
1160
7
.
14.
Schroeck
FR
,
Aronson
WJ
,
Presti
JC
 Jr
,
Terris
MK
,
Kane
CJ
,
Amling
CL
, et al
Do nomograms predict aggressive recurrence after radical prostatectomy more accurately than biochemical recurrence alone?
BJU Int
2009
;
103
:
603
8
.
15.
Cooperberg
MR
,
Broering
JM
,
Carroll
PR
. 
Risk assessment for prostate cancer metastasis and mortality at the time of diagnosis
.
J Natl Cancer Inst
2009
;
101
:
878
87
.
16.
Ahonen
TJ
,
Xie
J
,
LeBaron
MJ
,
Zhu
J
,
Nurmi
M
,
Alanen
K
, et al
Inhibition of transcription factor Stat5 induces cell death of human prostate cancer cells
.
J Biol Chem
2003
;
278
:
27287
92
.
17.
Dagvadorj
A
,
Kirken
RA
,
Leiby
B
,
Karras
J
,
Nevalainen
MT
. 
Transcription factor signal transducer and activator of transcription 5 promotes growth of human prostate cancer cells in vivo
.
Clin Cancer Res
2008
;
14
:
1317
24
.
18.
Gu
L
,
Liao
Z
,
Hoang
DT
,
Dagvadorj
A
,
Gupta
S
,
Blackmon
S
, et al
Pharmacologic inhibition of Jak2-Stat5 signaling by Jak2 inhibitor AZD1480 potently suppresses growth of both primary and castrate-resistant prostate cancer
.
Clin Cancer Res
2013
;
19
:
5658
74
.
19.
Tan
SH
,
Dagvadorj
A
,
Shen
F
,
Gu
L
,
Liao
Z
,
Abdulghani
J
, et al
Transcription factor Stat5 synergizes with androgen receptor in prostate cancer cells
.
Cancer Res
2008
;
68
:
236
48
.
20.
Talati
PG
,
Gu
L
,
Ellsworth
EM
,
Girondo
MA
,
Trerotola
M
,
Hoang
DT
, et al
Jak2-Stat5a/b signaling induces epithelial-to-mesenchymal transition and stem-like cell properties in prostate cancer
.
Am J Pathol
2015
;
185
:
2505
22
.
21.
Haddad
BR
,
Gu
L
,
Mirtti
T
,
Dagvadorj
A
,
Vogiatzi
P
,
Hoang
DT
, et al
STAT5A/B gene locus undergoes amplification during human prostate cancer progression
.
Am J Pathol
2013
;
182
:
2264
75
.
22.
Levy
DE
,
Darnell
JE
 Jr
. 
Stats: transcriptional control and biological impact
.
Nat Rev Mol Cell Biol
2002
;
3
:
651
62
.
23.
Li
H
,
Zhang
Y
,
Glass
A
,
Zellweger
T
,
Gehan
E
,
Bubendorf
L
, et al
Activation of signal transducer and activator of transcription-5 in prostate cancer predicts early recurrence
.
Clin Cancer Res
2005
;
11
:
5863
8
.
24.
Mirtti
T
,
Leiby
BE
,
Abdulghani
J
,
Aaltonen
E
,
Pavela
M
,
Mamtani
A
, et al
Nuclear Stat5a/b predicts early recurrence and prostate cancer-specific death in patients treated by radical prostatectomy
.
Hum Pathol
2013
;
44
:
310
9
.
25.
Vickers
AJ
,
Elkin
EB
. 
Decision curve analysis: a novel method for evaluating prediction models
.
Med Decis Making
2006
;
26
:
565
74
.
26.
Epstein
JI
,
Egevad
L
,
Amin
MB
,
Delahunt
B
,
Srigley
JR
,
Humphrey
PA
, et al
The 2014 International Society of Urological Pathology (ISUP) Consensus Conference on Gleason Grading of Prostatic Carcinoma: definition of grading patterns and proposal for a new grading system
.
Am J Surg Pathol
2016
;
40
:
244
52
.
27.
Nevalainen
MT
,
Xie
J
,
Bubendorf
L
,
Wagner
KU
,
Rui
H
. 
Basal activation of transcription factor signal transducer and activator of transcription (Stat5) in nonpregnant mouse and human breast epithelium
.
Mol Endocrinol
2002
;
16
:
1108
24
.
28.
Li
H
,
Ahonen
TJ
,
Alanen
K
,
Xie
J
,
LeBaron
MJ
,
Pretlow
TG
, et al
Activation of signal transducer and activator of transcription 5 in human prostate cancer is associated with high histological grade
.
Cancer Res
2004
;
64
:
4774
82
.
29.
Komrokji
RS
,
Seymour
JF
,
Roberts
AW
,
Wadleigh
M
,
To
LB
,
Scherber
R
, et al
Results of a phase 2 study of pacritinib (SB1518), a JAK2/JAK2(V617F) inhibitor, in patients with myelofibrosis
.
Blood
2015
;
125
:
2649
55
.

Supplementary data