Erho, Nicholas

Abstract Purpose: There is a clear need to improve risk stratification and to identify novel therapeutic targets in aggressive prostate cancer. The goal of this study was to investigate genes with outlier expression with prognostic association in high-risk prostate cancer patients as potential biomarkers and drug targets. Experimental Design: We interrogated microarray gene expression data from prostatectomy samples from 545 high-risk prostate cancer patients with long-term follow-up (mean 13.4 years). Three independent clinical datasets totaling an additional 545 patients were used for validation. Novel prognostic outlier genes were interrogated for impact on oncogenic phenotypes in vitro using siRNA-based knockdown. Association with clinical outcomes and comparison with existing prognostic instruments was assessed with multivariable models using a prognostic outlier score. Results: Analysis of the discovery cohort identified 20 prognostic outlier genes. Three top prognostic outlier genes were novel prostate cancer genes; NVL, SMC4, or SQLE knockdown reduced migration and/or invasion and outlier expression was significantly associated with poor prognosis. Increased prognostic outlier score was significantly associated with poor prognosis independent of standard clinicopathologic variables. Finally, the prognostic outlier score prognostic association is independent of, and adds to existing genomic and clinical tools for prognostication in prostate cancer (Decipher, the cell-cycle progression signature, and CAPRA-S). Conclusions: To our knowledge, this study represents the first unbiased high-throughput investigation of prognostic outlier genes in prostate cancer and demonstrates the potential biomarker and therapeutic importance of this previously unstudied class of cancer genes. Clin Cancer Res; 22(7); 1777–86. ©2015 AACR.

Abstract Background: Prostate cancer molecular subtypes based on ETS gene fusions and SPINK1 were originally identified through distinct gene expression profiles. Such molecular subtypes may have utility in disease stratification and clonality assessment, complementing available purely prognostic tests. Hence, we determined the analytical validity of molecular subtyping and explored clinical associations using global gene expression profiles in a large cohort of PCa. Methods: We analyzed 1,577 patient Affymetrix Human Exon 1.0ST GeneChip expression profiles from 8 radical prostatectomy (RP) cohorts; 5 of them generated as part of the Decipher® platform for the Decipher® discovery or validation. Multi-feature random forest classifiers and outlier analysis were used to define microarray-based molecular subtypes and characterize clinical associations. Results: A random forest (RF) classifier (m-ERG) was trained and validated to predict ERG fusion status using separate subsets of a single-institution RP cohort (total n=407) with known ERG rearrangement status defined by FISH, achieving >95% sensitivity and specificity in the validation subset. The model was then applied to 7 independent RP cohorts to predict ERG rearrangement status. Less frequent rearrangements involving other ETS genes (ETV1, ETV4, ETV5, FLI1) or SPINK1 over-expression were predicted based on gene expression outlier analysis. Across cohorts, 45%, 9% 8% and 38% of PCa were classified as ERG+, ERG—ETS+, ERG—SPINK+, and Triple Negative (ERG—/ETS—/SPINK1—), respectively. Global gene expression analysis shows that the four subtypes could be collapsed into three entities (ERG+, ERG—ETS+ and SPINK+/Triple Negative) based on expression patterns and clinical characteristics similarity. Based on multivariable analysis, ERG+ is significantly associated with lower pre-PSA (p<0.001), lower Gleason score (p<0.001), the presence of extraprostatic extension (p<0.001) and European Americans (p<0.001), but not associated with outcome. ERG—ETS+ is significantly associated with SVI (p=0.01) and ERG—SPINK+ is enriched in African Americans (p<0.001). Conclusions: The Decipher® platform can accurately determine ERG rearrangement status and PCa molecular subtypes. Inclusion of molecular subtyping, such as m-ERG status, may enable additional precision medicine opportunities in prognostic tests and provides insights into the development of novel therapeutic approaches. Citation Format: Mohammed Alshalalfa, Scott A. Tomlins, Nicholas Erho, Kasra Yousefi, Shuang Zhao, Robert B. Den, Adam P. Dicker, Bruce Trock, Angelo DeMarzo, Edward M. Schaeffer, Ashley Ross, Eric A. Klein, Cristina Magi-Galluzzi, Jeffery R. Karnes, Robert B Jenkins, Elai Davicioni, Felix Feng. Molecular and clinical characterization of 1,577 primary prostate cancer tumors reveals novel clinical and biological insights into its subtypes. [abstract]. In: Proceedings of the AACR Special Conference on Translation of the Cancer Genome; Feb 7-9, 2015; San Francisco, CA. Philadelphia (PA): AACR; Cancer Res 2015;75(22 Suppl 1):Abstract nr A1-64.

Abstract Purpose: Prostate cancers incite tremendous morbidity upon metastatic growth. We previously identified Asporin (ASPN) as a potential mediator of metastatic progression found within the tumor microenvironment. ASPN contains an aspartic acid (D)-repeat domain and germline polymorphisms in D-repeat-length have been associated with degenerative diseases. Associations of germline ASPN D polymorphisms with risk of prostate cancer progression to metastatic disease have not been assessed. Experimental Design: Germline ASPN D-repeat-length was retrospectively analyzed in 1,600 men who underwent radical prostatectomy for clinically localized prostate cancer and in 548 noncancer controls. Multivariable Cox proportional hazards models were used to test the associations of ASPN variations with risk of subsequent oncologic outcomes, including metastasis. Orthotopic xenografts were used to establish allele- and stroma-specific roles for ASPN D variants in metastatic prostate cancer. Results: Variation at the ASPN D locus was differentially associated with poorer oncologic outcomes. ASPN D14 [HR, 1.72; 95% confidence interval (CI), 1.05–2.81, P = 0.032] and heterozygosity for ASPN D13/14 (HR, 1.86; 95% CI, 1.03–3.35, P = 0.040) were significantly associated with metastatic recurrence, while homozygosity for the ASPN D13 variant was significantly associated with a reduced risk of metastatic recurrence (HR, 0.44; 95% CI, 0.21–0.94, P = 0.035) in multivariable analyses. Orthotopic xenografts established biologic roles for ASPN D14 and ASPN D13 variants in metastatic prostate cancer progression that were consistent with patient-based data. Conclusions: We observed associations between ASPN D variants and oncologic outcomes, including metastasis. Our data suggest that ASPN expressed in the tumor microenvironment is a heritable modulator of metastatic progression. Clin Cancer Res; 22(2); 448–58. ©2015 AACR.

Abstract Background ETS gene fusions and PTEN loss are the most common genomic alteration events in prostate cancer that stratify patients into molecularly distinct subtypes. Recently, a commercially available genomic classifier (GC, DecipherTM) has been validated to predict metastasis in patients with high-risk prostate cancer. In this study we evaluated the robustness of the GC model in the light of prostate cancer molecular heterogeneity. Methods: Analysis of PTEN expression loss, ERG overexpression and GC scores were performed on a previously reported genome-wide microarray dataset (Karnes et al., 2013). Microarray probesets mapping to exons of PTEN were benchmarked to identify a representative exon whose loss of expression was highly correlated to FISH analysis for PTEN deletion (PTEN-loss) performed on an independent dataset. ERG overexpression (ERG+) was determined similarly and the microarray probesets that best correlate with the FISH status (in independent dataset) were used to annotate the ERG status of the samples in this study. Receiver operating characteristic area under the curve (AUC), and multivariable regression analysis were used to assess GC performance for predicting metastasis in subsets of patient tumors with genomic alterations. Results Using the microarray method 42% of the tumors had ERG overexpression, 23% had PTEN-loss and 11% had both events. ERG+ and PTEN-loss were not on their own predictive of metastasis with AUCs of 0.52 (CI:0.44-0.6) and 0.47(CI 0.39-0.56), respectively. In ERG+ tumors GC had an AUC of 0.82 (p=4e-7) and high GC score ERG+ patients had a hazard ratio (HR) of 100 (p=6e-7) for metastasis. Similarly, in PTEN-loss tumors GC had an AUC of 0.83 (p=7.9e-5) and high GC score PTEN-loss patients had an HR of 62 (p=1.8e-4) for metastasis. When GC was evaluated in patient tumors with both ERG+ and PTEN-loss the AUC was 0.85 (p=2e-15) and high GC score patients had an HR of 95 (p=4e-4) for metastasis. In multivariable analysis, Decipher was the only significant risk factor with an HR of 55 (CI 12.2-249, p=1.8e-7) for metastasis after adjusting for ERG+, PTEN-loss, Gleason score, PSA, tumor stage, margins and adjuvant treatment. Conclusion The Decipher genomic classifier is a powerful prognostic marker that retains its significance for predicting metastasis in subsets of patients with tumors that harbor common genomic alterations. Patients with high GC scores and these alterations may represent an extremely high-risk subset of prostate cancer. Citation Format: Mohammed Alshalalfa, Ismael Vergara, Nicholas Erho, Elai Davicioni, Robert Jenkins, Kollmeyer Thomas. Evaluation of a genomic classifier (Decipher®) in subsets of primary tumors with common prostate cancer genomic alterations. [abstract]. In: Proceedings of the 105th Annual Meeting of the American Association for Cancer Research; 2014 Apr 5-9; San Diego, CA. Philadelphia (PA): AACR; Cancer Res 2014;74(19 Suppl):Abstract nr 4720. doi:10.1158/1538-7445.AM2014-4720

Objective To better characterize the genomics of patients with biochemical recurrence ( BCR ) who have metastatic disease progression in order to improve treatment decisions for prostate cancer. Methods The expression profiles of three clinical outcome groups after radical prostatectomy ( RP ) were compared: those with no evidence of disease ( NED ; n = 108); those with BCR (rise in prostate‐specific antigen [ PSA ] level without metastasis; n = 163); and those with metastasis ( n = 192). The patients were profiled using Human Exon 1.0 ST microarrays, and outcomes were supported by a median 18 years of follow‐up. A metastasis signature was defined and verified in an independent RP cohort to ensure the robustness of the signature. Furthermore, bioinformatics characterization of the signature was conducted to decipher its biology. Results Minimal gene expression differences were observed between adjuvant treatment‐naïve patients in the NED group and patients without metastasis in the BCR group. More than 95% of the differentially expressed genes (metastasis signature) were found in comparisons between primary tumours of metastasis patients and the two other outcome groups. The metastasis signature was validated in an independent cohort and was significantly associated with cell cycle genes, ubiquitin‐mediated proteolysis, DNA repair, androgen, G‐protein coupled and NOTCH signal transduction pathways. Conclusion This study shows that metastasis development after BCR is associated with a distinct transcriptional programme that can be detected in the primary tumour. Patients with NED and BCR have highly similar transcriptional profiles, suggesting that measurement of PSA on its own is a poor surrogate for lethal disease. Use of genomic testing in patients undergoing RP with an initial rise in PSA level may be useful to improve secondary therapy decision‐making.

Abstract Introduction: Multivariable models that predict biochemical recurrence (BCR) can indicate patients at higher risk for developing clinical progression after prostatectomy. However, most patients that experience BCR do not metastasize or die from prostate cancer, making BCR an unreliable surrogate endpoint. In this study we compared the clinical and genomic characteristics of prostatectomy patients in order to better delineate the relationship between BCR and aggressive prostate cancer. Methods: Postoperative patients were retrospectively classified into three outcome groups: NED (no evidence of disease recurrence); BCR-only (patients that experienced BCR but didn't have progressive disease as evidenced by bone or CT scans); and CP (clinical progression with BCR plus positive CT or bone scans). Patient specimens were from the Mayo Clinic Radical Prostatectomy Registry, selected using a nested case-control design with 14 years median follow-up. RNA expression levels from FFPE tumor specimens were measured with 1.4 million feature whole-transcriptome oligonucleotide microarrays on 106 matching patients from each outcome group (total 318 patients). We analyzed the clinical covariates of the outcome groups in a pairwise manner (NED vs. BCR-only; NED vs. CP; CP vs. BCR-only), using a conditional logistic model in univariable and multivariable analyses. We performed feature reduction for the genomic data set using Kruskal-Wallis tests and regularized logistic regression. Principal Component Analysis (PCA) was used to quantify the variance of the genomic data with respect to the patient outcome groups. Results: Our initial hypothesis was that all three outcome groups would have different clinical and genomic characteristics. Univariable analysis of clinical variables failed to show statistically significant differences between NED and BCR-only groups. Multivariable analysis of the clinical variables also failed to show statistically significant differences between NED and BCR-only groups. Among the clinical variables only pathological Gleason score was statistically significant when comparing the BCR-only vs. CP and NED vs. CP groups. Genomic analysis found 363 coding and non-coding features that passed thresholds for differential expression between the three outcome groups. PCA established that the primary source of variance in the genomic data stemmed from the separation of CP from NED and BCR-only patients; whereas only a small amount of variance separates NED and BCR-only groups. The 363 features were further compared in the three outcome groups using Wilcoxon tests, adjusting for false-discovery using Benjamini-Hochberg. At a cutoff of p < 0.001 there were no features differentially expressed between NED and BCR-only, while 52 and 34 statistically significant features were found in the NED vs. CP and CP vs. BCR-only comparisons, respectively. Conclusion: Considering the clinical and genomic results together, we find that: 1) Patients that developed clinical progression were significantly different from those that did not, irrespective of whether they experienced BCR or not; 2) the NED and BCR-only groups were statistically very similar to each other. We therefore conclude that BCR may not be the most appropriate endpoint for predictive models that attempt to stratify patients at risk for progression. This abstract is also presented as Poster B44. Citation Format: Anamaria Crisan, Elai Davicion, R. Jeffery Karnes, Robert B. Jenkins, Mercedeh H. Ghadessi, Christine Buerki, Eric J. Bergstrahl, Nicholas Erho, Ismael A. Vergara, Karla V. Ballman, Peter C. Black, Timothy J. Triche. Biochemical recurrence is not a definitive surrogate endpoint for development of clinically useful predictive models for post-prostatectomy patients [abstract]. In: Proceedings of the AACR Special Conference on Advances in Prostate Cancer Research; 2012 Feb 6-9; Orlando, FL. Philadelphia (PA): AACR; Cancer Res 2012;72(4 Suppl):Abstract nr PR4.