Risk assessment for indeterminate pulmonary nodules using a novel, plasma-protein based biomarker assay

BACKGROUND

The increase in lung cancer screening is intensifying the need for a noninvasive test to characterize the many indeterminate pulmonary nodules (IPN) discovered. Correctly identifying non-cancerous nodules is needed to reduce overdiagnosis and overtreatment. Alternatively, early identification of malignant nodules may represent a potentially curable form of lung cancer.

OBJECTIVE

To develop and validate a plasma-based multiplexed protein assay for classifying IPN by discriminating between those with a lung cancer diagnosis established pathologically and those found to be clinically and radiographically stable for at least one year.

METHODS

Using a novel technology, we developed assays for plasma proteins associated with lung cancer into a panel for characterizing the risk that an IPN found on chest imaging is malignant. The assay panel was evaluated with a cohort of 277 samples, all from current smokers with an IPN 4-30 mm. Subjects were divided into training and test sets to identify a Support Vector Machine (SVM) model for risk classification containing those proteins and clinical factors that added discriminatory information to the Veteran's Affairs (VA) Clinical Factors Model. The algorithm was then evaluated in an independent validation cohort.

RESULTS

Among the 97 validation study subjects, 68 were grouped as having intermediate risk by the VA model of which the SVM model correctly identified 44 (65%) of these intermediate-risk samples as low (n=16) or high risk (n=28). The SVM model negative predictive value (NPV) was 94% and its sensitivity was 94%.

CONCLUSION

The performance of the novel plasma protein biomarker assay supports its use as a noninvasive risk assessment aid for characterizing IPN. The high NPV of the SVM model suggests its application as a rule-out test to increase the confidence of providers to avoid aggressive interventions for their patients for whom the VA model result is an inconclusive, intermediate risk.

Development and Validation of a Prostate Cancer Genomic Signature that Predicts Early ADT Treatment Response Following Radical Prostatectomy

Purpose: Currently, no genomic signature exists to distinguish men most likely to progress on adjuvant androgen deprivation therapy (ADT) after radical prostatectomy for high-risk prostate cancer. Here we develop and validate a gene expression signature to predict response to postoperative ADT.Experimental Design: A training set consisting of 284 radical prostatectomy patients was established after 1:1 propensity score matching metastasis between adjuvant-ADT (a-ADT)-treated and no ADT-treated groups. An ADT Response Signature (ADT-RS) was identified from neuroendocrine and AR signaling-related genes. Two independent cohorts were used to form three separate data sets for validation (set I, n = 232; set II, n = 435; set III, n = 612). The primary endpoint of the analysis was postoperative metastasis.Results: Increases in ADT-RS score were associated with a reduction in risk of metastasis only in a-ADT patients. On multivariable analysis, ADT-RS by ADT treatment interaction term remained associated with metastasis in both validation sets (set I: HR = 0.18, P_interaction = 0.009; set II: HR = 0.25, P_interaction = 0.019). In a matched validation set III, patients with Low ADT-RS scores had similar 10-year metastasis rates in the a-ADT and no-ADT groups (30.1% vs. 31.0%, P = 0.989). Among High ADT-RS patients, 10-year metastasis rates were significantly lower for a-ADT versus no-ADT patients (9.4% vs. 29.2%, P = 0.021). The marginal ADT-RS by ADT interaction remained significant in the matched dataset (P_interaction = 0.035).Conclusions: Patients with High ADT-RS benefited from a-ADT. In combination with prognostic risk factors, use of ADT-RS may thus allow for identification of ADT-responsive tumors that may benefit most from early androgen blockade after radical prostatectomy. We discovered a gene signature that when present in primary prostate tumors may be useful to predict patients who may respond to early ADT after surgery. Clin Cancer Res; 24(16); 3908-16. ©2018 AACR.

Validation of a genomic risk classifier to predict metastasis and prostate cancer-specific mortality in men with positive lymph nodes

125

Background: Men with lymph node involvement (LNI) at prostatectomy (RP) are at high risk of dying from prostate cancer. However, survival following RP is highly variable, with some men apparently cured. Nomograms developed for men with LNI have been based on series where all or the vast majority of men received adjuvant treatment. Because administration of adjuvant treatment is not universal, even for LNI, we evaluated whether the Decipher genomic classifier (GC) can improve upon clinical models to predict metastasis within 5 years (MET5) and prostate cancer specific mortality within 10 years (PCSM10) in a cohort of LNI patients, the majority of whom did not receive adjuvant treatment. Methods: 141 patients from 4 institutions (Johns Hopkins, Mayo Clinic, Leuven, MD Anderson) had LNI at RP, had adequate tissue and clinical data for analysis of MET5, and 86 were analyzed for PCSM10. 43% of men received adjuvant therapy. RP tumor tissue was analyzed by Affymetrix Human Exon 1.0 ST GeneChip; the GC was calculated based on 22 genes in the previously trained and validated algorithm. Logistic regression evaluated whether the GC, dichotomized as high risk (GC score > 0.6) vs low-intermediate risk (≤0.6), improved prediction of MET5 and PCSM10 beyond that achieved with established clinical prognostic factors. Results: 62 men (43%) developed MET5, and 35 (41%) developed PCSM10. For both MET5 and PCSM10 CAPRA-S, number of positive lymph nodes, and age were significant; adjuvant therapy was not significant. GC was a significant independent strong prognostic factor when added to the clinical model for prediction of MET5, odds ratio = 4.04 (95% CI: 1.48, 11.02), p = 0.006, and prediction of PCSM10, odds ratio = 6.71 (95% CI: 2.01, 22.38), p = 0.002. Addition of GC to the clinical model improved the AUC from 0.77 to 0.79 for MET5, and from 0.65 to 0.74 for PCSM10. Conclusions: The Decipher GC significantly improves upon clinical variables to predict metastasis and prostate cancer specific mortality in men at high risk due to LNI. This is the first study to show a genomic classifier predicts outcomes in men with LNI; validation is needed to determine if Decipher can improve treatment decisions in men with LNI.

Development and Validation of a Novel Integrated Clinical-Genomic Risk Group Classification for Localized Prostate Cancer

Purpose It is clinically challenging to integrate genomic-classifier results that report a numeric risk of recurrence into treatment recommendations for localized prostate cancer, which are founded in the framework of risk groups. We aimed to develop a novel clinical-genomic risk grouping system that can readily be incorporated into treatment guidelines for localized prostate cancer. Materials and Methods Two multicenter cohorts (n = 991) were used for training and validation of the clinical-genomic risk groups, and two additional cohorts (n = 5,937) were used for reclassification analyses. Competing risks analysis was used to estimate the risk of distant metastasis. Time-dependent c-indices were constructed to compare clinicopathologic risk models with the clinical-genomic risk groups. Results With a median follow-up of 8 years for patients in the training cohort, 10-year distant metastasis rates for National Comprehensive Cancer Network (NCCN) low, favorable-intermediate, unfavorable-intermediate, and high-risk were 7.3%, 9.2%, 38.0%, and 39.5%, respectively. In contrast, the three-tier clinical-genomic risk groups had 10-year distant metastasis rates of 3.5%, 29.4%, and 54.6%, for low-, intermediate-, and high-risk, respectively, which were consistent in the validation cohort (0%, 25.9%, and 55.2%, respectively). C-indices for the clinical-genomic risk grouping system (0.84; 95% CI, 0.61 to 0.93) were improved over NCCN (0.73; 95% CI, 0.60 to 0.86) and Cancer of the Prostate Risk Assessment (0.74; 95% CI, 0.65 to 0.84), and 30% of patients using NCCN low/intermediate/high would be reclassified by the new three-tier system and 67% of patients would be reclassified from NCCN six-tier (very-low- to very-high-risk) by the new six-tier system. Conclusion A commercially available genomic classifier in combination with standard clinicopathologic variables can generate a simple-to-use clinical-genomic risk grouping that more accurately identifies patients at low, intermediate, and high risk for metastasis and can be easily incorporated into current guidelines to better risk-stratify patients.

Gene Expression Correlates of Site-specific Metastasis Among Men With Lymph Node Positive Prostate Cancer Treated With Radical Prostatectomy: A Case Series

Predictors of site-specific metastasis after radical prostatectomy (RP) are unknown despite prognostic differences between metastatic sites. We performed RNA expression analysis for 19 genes known to be correlated with aggressive prostate cancer in primary tumors of 63 men pN+ at RP (N = 35 developing metastases after RP vs N = 28 without metastases after RP). Of the men developing metastases, 22 (62.9%) had bone metastases, 8 (22.9%) had nonregional nodal metastases, and 5(14.3%) had visceral metastases. Patients with nodal metastases had higher androgen receptor expression relative to other metastatic sites and nonmetastatic controls (P = .001). This may explain the favorable prognosis of nodal metastases as it may be more androgen dependent.

Development and validation of a novel clinical-genomic risk group classification for prostate cancer incorporating genomic and clinicopathologic risk

5000

Background: It is clinically challenging to integrate genomic classifier results that report a continuous numerical risk of recurrence into treatment decisions for prostate cancer (PCa). We aimed to develop a novel clinical-genomic risk system that can readily be incorporated into treatment guidelines for localized PCa. Methods: Four multi-center cohorts (n = 6928 men; 5937 prospective samples and 991 retrospective samples with long-term follow-up) were utilized to identify and validate our clinical-genomic risk system in radical prostatectomy (RP) samples and subsequently in pre-treatment biopsy samples. All patients’ FFPE tissue underwent microarray analysis, and the expression values for 22 prespecified biomarkers that constitute Decipher were extracted. Cumulative incidence curves were constructed to estimate metastasis risk. C-indices were calculated to compare NCCN and CAPRA score to our clinical-genomic system. Results: With a median follow-up of 8 years for men in our RP cohort, the 10-year distant metastasis rates for NCCN low, favorable-intermediate, unfavorable-intermediate, and high-risk were 7.8%, 9.4%, 40.1%, and 41.4%, respectively. Our 3-tier clinical-genomic risk groups had 10-year distant metastasis rates of 3.7%, 30.7%, and 57.7%, for low, intermediate, and high-risk, which were validated in our pre-treatment biopsy cohort with 10-year rate of distant metastasis of 0%, 30.3%, and 63.2%, respectively. C-indices for the clinical-genomic system (0.84, 95%CI 0.62-0.92) were significantly improved over NCCN (0.71, 95%CI 0.59-0.84) and CAPRA (0.71, 95%CI 0.60-0.81) score. A total of 33.4% of men would be reclassified by the clinical-genomic system, and specifically 17.1%, 41.3%, and 19.4% of men in NCCN low, intermediate and high risk groups would be reclassified by our new system. Conclusions: The use of a readily available genomic classifier in combination with clinicopathologic variables can generate a simple to use 3-tier clinical-genomic risk system that is highly prognostic for distant metastasis, is more accurate than clinical risk, and can be easily incorporated into NCCN guidelines to inform treatment decisions.

Individual Patient-Level Meta-Analysis of the Performance of the Decipher Genomic Classifier in High-Risk Men After Prostatectomy to Predict Development of Metastatic Disease

Purpose To perform the first meta-analysis of the performance of the genomic classifier test, Decipher, in men with prostate cancer postprostatectomy. Methods MEDLINE, EMBASE, and the Decipher genomic resource information database were searched for published reports between 2011 and 2016 of men treated by prostatectomy that assessed the benefit of the Decipher test. Multivariable Cox proportional hazards models fit to individual patient data were performed; meta-analyses were conducted by pooling the study-specific hazard ratios (HRs) using random-effects modeling. Extent of heterogeneity between studies was determined with the I² test. Results Five studies (975 total patients, and 855 patients with individual patient-level data) were eligible for analysis, with a median follow-up of 8 years. Of the total cohort, 60.9%, 22.6%, and 16.5% of patients were classified by Decipher as low, intermediate, and high risk, respectively. The 10-year cumulative incidence metastases rates were 5.5%, 15.0%, and 26.7% ( P < .001), respectively, for the three risk classifications. Pooling the study-specific Decipher HRs across the five studies resulted in an HR of 1.52 (95% CI, 1.39 to 1.67; I² = 0%) per 0.1 unit. In multivariable analysis of individual patient data, adjusting for clinicopathologic variables, Decipher remained a statistically significant predictor of metastasis (HR, 1.30; 95% CI, 1.14 to 1.47; P < .001) per 0.1 unit. The C-index for 10-year distant metastasis of the clinical model alone was 0.76; this increased to 0.81 with inclusion of Decipher. Conclusion The genomic classifier test, Decipher, can independently improve prognostication of patients postprostatectomy, as well as within nearly all clinicopathologic, demographic, and treatment subgroups. Future study of how to best incorporate genomic testing in clinical decision-making and subsequent treatment recommendations is warranted.

Individual patient level meta-analysis of the performance of the Decipher genomic classifier in high-risk men post-prostatectomy to predict development of metastatic disease

133

Background: The genomic classifier, Decipher, has been validated to predict risk of metastasis after radical prostatectomy (RP). However, the cohort size and event rate in the previous studies did not allow for a thorough investigation into performance within individual clinicopathologic or treatment subgroups. In this study, we present the first meta-analysis of the performance of the 22-marker genomic classifier in men with prostate cancer (PCa) post-RP. Methods: MEDLINE, EMBASE, and the Decipher genomic resource information database were searched for published reports of men with PCa treated by RP between 2010 and 2016 where the benefit of the Decipher genomic classifier test was assessed. The primary end point was the ability of Decipher to independently improve prognostication of regional or distant metastasis over routine clinicopathologic factors. Meta-analysis was performed with random-effects modeling, and extent of heterogeneity between studies was determined with the I2 test. Results: Five studies (975 total patients, and 855 with individual patient genomic and clinicopathologic data) were eligible for analysis. The median follow-up was 8 years. All patients had clinical high-risk disease, yet 60.9%, 22.6%, and 16.5% of patients were classified as low, intermediate, and high-risk, respectively by Decipher and had 10-year cumulative incidence rates of metastases of 5.5%, 15.0% and 26.7% (p < 0.001), respectively. Adjusting for standard clinicopathologic variables, on multivariable analysis Decipher remained a statistically significant predictor of metastasis (hazard ratio [HR] 1.30 per 0.1 unit, 95% confidence interval [CI] 1.14-1.47, p < 0.001), and the summary HR for metastasis of Decipher across the 5 studies was 1.52 (95% CI 1.39-1.67) per 0.1 unit. Conclusions: The genomic classifier test, Decipher, has the ability to independently improve prognostication of men post-RP, as well as within nearly all clinicopathologic and treatment subgroups. Strong consideration should be given to incorporating the use of genomic testing in clinical decision making and clinical trials to better individualize treatment.

Evaluation of the Decipher prostate cancer classifier to predict metastasis and disease-specific mortality from genomic analysis of diagnostic prostate needle biopsy specimens

4

Background: Accurate risk stratification after diagnosis of prostate cancer (PCa) is key to optimal treatment decision-making. Decipher RP is an extensively validated genomic classifier used to determine biological potential for metastasis. Here, in a multi-institutional cohort, we aimed to evaluate its ability to predict metastasis and prostate cancer-specific mortality from analysis of PCa needle biopsy tumor tissue specimens. Methods: We identified 175 patients treated with either first-line RP or first-line radiation therapy (RT) + androgen deprivation therapy (ADT) with available genomic expression profiles generated from diagnostic biopsy specimens obtained from three tertiary referral centers: Cleveland Clinic, Brigham and Women’s Hospital and Johns Hopkins. The core with the highest grade was sampled and Decipher was calculated based on a locked random forest model. Cox univariable and multivariable (MVA) proportional hazards model and survival c-index were used to evaluate the performance of Decipher. Results: Overall, 85% of patients had NCCN intermediate and high-risk disease. Of the 175 patients, 43% and 57% were treated with first-line RP and RT+ADT, respectively. With a median follow-up of 6 years, 32 patients developed metastases and 11 of these patients died of PCa. For prediction of metastasis 5 years post biopsy, Decipher had a c-index of 0.74 (95% confidence interval [CI] 0.63-0.84) compared to 0.66 (95% CI 0.53-0.77) for CAPRA and 0.66 (95% CI 0.55-0.77) for NCCN risk group. On MVA, when modeled with CAPRA, Bx Decipher remained a significant predictor of metastasis (Decipher Bx hazard ratio [HR] 1.33 per 10% increase in score, 95% CI 1.06–1.69, P = 0.01). Decipher Bx was also a significant predictor of PCSM (Decipher Bx HR 1.57 per 10%, 95% CI 1.07–2.40, P = 0.02) Conclusions: Decipher Bx was able to predict metastasis and PCSM from diagnostic biopsy specimens in a cohort of primarily intermediate and high-risk men regardless of first line treatment. This additional genomic information provides important risk stratification to help guide therapy for men with intermediate- and high-risk disease.

Molecular Analysis of Low Grade Prostate Cancer Using a Genomic Classifier of Metastatic Potential

PURPOSE

We determined how frequently histological Gleason 3 + 3 = 6 tumors have the molecular characteristics of disease with metastatic potential.

MATERIALS AND METHODS

We analyzed prostatectomy tissue from 337 patients with Gleason 3 + 3 disease. All tissue was re-reviewed in blinded fashion by genitourinary pathologists using 2005 ISUP (International Society of Urological Pathology) Gleason grading criteria. A previously validated Decipher® metastasis signature was calculated in each case based on a locked model. To compare patient characteristics across pathological Gleason score categories we used the Fisher exact test or the ANOVA F test. The distribution of Decipher scores among different clinicopathological groups was compared with the Wilcoxon rank sum test. The association of Decipher score with adverse pathology features was examined using logistic regression models. The significance level of all statistical tests was 0.05.

RESULTS

Of men with Gleason 3 + 3 = 6 disease only 269 (80%) had a low Decipher score with intermediate and high scores in 43 (13%) and 25 (7%), respectively. Decipher scores were significantly higher among pathological Gleason 3 + 3 = 6 specimens from cases with adverse pathological features such as extraprostatic extension, seminal vesicle involvement or positive margins (p <0.001). The median Decipher score in patients with margin negative pT2 disease was 0.23 (IQR 0.09-0.42) compared to 0.30 (IQR 0.17-0.42) in patients with pT3 disease or positive margins (p = 0.005).

CONCLUSIONS

Using a robust and validated prognostic signature we found that a small but not insignificant proportion of histological Gleason 6 tumors harbored molecular characteristics of aggressive cancer. Molecular profiling of such tumors at diagnosis may better select patients for active surveillance at diagnosis and trigger appropriate intervention during followup.

Evaluation of a genomic classifier in radical prostatectomy patients with lymph node metastasis

Objective

To evaluate the performance of the Decipher test in predicting lymph node invasion (LNI) on radical prostatectomy (RP) specimens.

Methods

We identified 1,987 consecutive patients with RP who received the Decipher test between February and August 2015 (contemporary cohort). In the contemporary cohort, only the Decipher score from RP specimens was available for analysis. In addition, we identified a consecutive cohort of patients treated with RP between 2006 and 2012 at the University of California, San Diego, with LNI upon pathologic examination (retrospective cohort). The retrospective cohort yielded seven, 22, and 18 tissue specimens from prostate biopsy, RP, and lymph nodes (LNs) for individual patients, respectively. Univariable and multivariable logistic regression analyses were used to evaluate the performance of Decipher in the contemporary cohort with LNI as the endpoint. In the retrospective cohort, concordance of risk groups was assessed using validated cut-points for low (<0.45), intermediate (0.45–0.60), and high (>0.60) Decipher scores.

Results

In the contemporary cohort, 51 (2.6%) patients had LNI. Decipher had an odds ratio of 1.73 (95% confidence interval, 1.46–2.05) and 1.42 (95% confidence interval, 1.19–1.7) per 10% increase in score on univariable and multivariable (adjusting for pathologic Gleason score, extraprostatic extension, and seminal vesicle invasion), respectively. No significant difference in the clinical and pathologic characteristics between the LN positive patients of contemporary and retrospective cohorts was observed (all P>0.05). Accordingly, among LN-positive patients in the contemporary cohort and retrospective cohort, 80% and 77% had Decipher high risk scores (P=1). In the retrospective cohort, prostate biopsy cores with the highest Gleason grade and percentage of tumor involvement had 86% Decipher risk concordance with both RP and LN specimens.

Conclusion

Decipher scores were highly concordant between pre- and post-surgical specimens. Further, Decipher scores from RP tissue were predictive of LNI at RP. If validated in a larger cohort of prostate biopsy specimens for prediction of adverse pathology at RP, Decipher may be useful for improved pre-operative staging.

Validation of the Decipher prostate cancer classifier for predicting 10-year postoperative metastasis from analysis of diagnostic needle biopsy specimens

59

Background: Accurate riskstratification after diagnosis of prostate cancer (PCa) is key to optimal treatment decision-making. Decipher is an extensively validated genomic classifier of metastasis after radical prostatectomy (RP). Here, we evaluate its ability to predict metastasis from analysis of prostate needle biopsy diagnostic tumor tissue specimens in a cohort of intermediate risk PCa patients treated with RP. Methods: Fifty-seven patients with available diagnostic biopsy specimens were identified from a previously reported post-RP validation study of Decipher in a cohort of 169 patients treated at Cleveland Clinic. The core with at least 1mm tumor of the highest Gleason grade was sampled and subjected to whole transcriptome analysis. Decipher was calculated based on a locked random forest model. Cox multivariable (MVA) proportional hazards model and survival c-index were used to evaluate the performance of Decipher. Results: 61% of patients had biopsy Gleason score 6 and 67% of patients had NCCN intermediate risk disease. With a median 8 years follow up, 8 patients metastasized and 3 of these patients died of PCa. Decipher had a c-index of 0.80 (95% confidence interval [CI], 0.58-0.95) compared to 0.58 (95% CI, 0.18-0.91) for biopsy Gleason score and 0.57 (0.57; 95% CI, 0.23-0.89) for preoperative PSA at 10 years post-RP for prediction of metastasis. A combined model consisting of Decipher, preoperative PSA, and Gleason score had a c-index of 0.84 (95% CI, 0.68-0.96). On MVA, Decipher was the only significant predictor of metastasis when adjusting for age, preoperative PSA and biopsy Gleason score (Decipher hazard ratio per 10% increase: 1.72; 95% CI, 1.04–2.83; P = 0.02). Conclusions: Decipher was able to predict metastatic outcome from diagnostic biopsy specimens in a cohort of primarily intermediate risk men treated with RP. This additional genomic information may help identify patients who may not be optimal candidates for active surveillance and better identify appropriate first line therapy for men with intermediate risk disease.

Efficacy of early and delayed radiation in a prostatectomy cohort adjusted for genomic and clinical risk

12

Background: In 3 published randomized trials, adjuvant radiation therapy (ART) for prostate cancer (PCa) resulted in improved progression free survival. However, the impact on metastases and overall survival is unclear. To date, there have been no published prospective trials examining the impact of salvage radiation therapy (SRT) in this disease state. Hence, we conducted a retrospective, nonrandomized comparative study of ART, SRT, or no radiation following radical prostatectomy (RP) for men with pT3 disease or positive margins (adverse pathologic features, APF). Methods: 422 PCa men treated at 4 institutions with RP and having APF were analyzed with a primary end point of clinical metastasis. ART (n = 111), early SRT (n = 70) and delayed SRT (n = 83) were defined by PSA levels of < 0.2, 0.2 to 0.5, and ≥ 0.5 ng/mL, respectively, prior to RT initiation. Remaining 157 men who did not receive additional therapy prior to metastatic onset formed the no RT arm. Clinical-genomic risk was assessed by CAPRA-S and Decipher. Cox multivariable (MVA) model was used to evaluate the impact of treatment on outcome. Results: During study follow-up, 37 men developed metastasis with a median follow-up of 8 years. Both CAPRA-S and Decipher had independent predictive value on MVA for metastatic outcome (both p < 0.05). On MVA adjusting for clinical-genomic risk, delayed SRT and no RT had an HR of 4.31 (95%CI, 1.20-15.47) and 5.42 (95% CI, 1.59-18.44) for metastasis compared to ART. No significance difference was observed between early SRT and ART (p = 0.28). Men with low to intermediate CAPRA-S and low Decipher have a low rate of metastatic events regardless of treatment selection. In contrast, men with high CAPRA-S and Decipher benefit from ART, however the cumulative incidence of metastasis remains high. Conclusions: The decision as to the timing and need for additional local therapy following RP is nuanced and requires providers and men to balance risks of morbidity with improved oncologic outcomes. This analysis provides the most robust and accurate quantification of risk for these men. Post-RP treatment can be safely avoided for men who are low risk by clinical-genomic risk, whereas those at high risk should favor enrollment in clinical trials.

Development and validation of an ADT resistance signature to predict adjuvant hormone treatment failure

106

Background: Androgen deprivation therapy (ADT) is one of the main treatment options for locally advanced and metastatic prostate cancer. Neuroendocrine prostate cancer (NEPC) is inherently less sensitive or even resistant to ADT. NEPC can be observed de novo (e.g., small cell prostate cancer) but more commonly arises after exposure to ADT. We hypothesized that a gene expression signature of NEPC when measured in primary tumor specimens (RP) of prostatic adenocarcinoma may be useful for predicting patients with innate resistance to ADT. Methods: Expression profiles of 1023 PCa patients treated with RP were obtained from the Decipher GRID database. These were split into training (n=529) and validation (n=494) sets and stratified by the receipt of adjuvant ADT (n=243) or no adjuvant ADT (n=780). A literature review of ADT resistance and neuroendocrine genes identified 1,557 genes as candidates. This set was further filtered, using logistic regression to select a 52-gene ADT resistance signature (ARS). ARS was trained using a generalized linear model with lasso regularization. Survival c-index and Kaplan Meier was used to compare survival differences between treated and untreated patients with high and low ARS scores (defined by median split). Results: In validation cohorts, the ARS was predictive of metastasis in cohorts receiving adjuvant ADT (10-year metastasis free survival c-index of 0.69 (95% CI 0.59-0.78) as compared to 0.45 (95% CI 0.29-0.61) in patients not treated with ADT). Similarly in a separate cohort of untreated patients that received no ADT until after metastatic onset, ARS was not prognostic (c-index 0.53). Among ADT treated patients, those with low ARS scores had a 10 year MFS of 87%, versus 70% in those with high ARS scores (p<0.001). In the subset of men who received ADT after metastatic onset and who developed castrate-resistant prostate cancer (CRPC, n = 41), median time to treatment failure was 1 year in patients with high ARS compared to 2 years for those with low ARS scores (p=0.07). Conclusions: A 52-gene ADT resistance signature was developed which showed significant differences in metastasis-free survival among adjuvant hormone treated but not untreated patients.

Deciphering the genomic fingerprint of small cell prostate cancer with potential clinical utility

303

Background: Small cell (SC) neuroendocrine carcinoma of the prostate has very poor prognosis and does not respond well to androgen receptor (AR)-targeted therapies. Neuroendocrine prostate carcinomas are increasingly recognized to show a spectrum of morphologic changes, and may not always be distinguishable from adenocarcinoma (adeno) based on morphology alone. Here we hypothesize that SC carcinoma harbor a unique gene expression signature that, when measured in primary prostatic adenocarcinoma, may help guide subsequent management. Methods: In total, 617 whole genome expression profiles were retrieved from the Decipher GRID providing gene expression data for 1.4 million markers. 17 morphologically-diagnosed SC carcinoma samples were compared to 32 Gleason 8-10 and 77 Gleason 6 RP adeno samples with no prior treatment and no evidence of metastasis, and this comparison was used to define SC Genomic Fingerprint (SCGFt) through adjusted median fold difference and Wilcoxon test. An additional 493 RP adeno from Mayo Clinic and John Hopkins Hospital were used to evaluate the utility of the SCGFt for predicting outcome. Results: Based on genomic prognostic tests calculated using the Decipher assay including Decipher, Penney and microarray-derived Cell Cycle Progression, SC carcinomas have very poor prognosis compared to high grade adeno. A total of 356 genes defined the SCGFt with 258 gene upregulated in SC and 98 down-regulated. PEG10, HELLS and RB1-loss program are the most overexpressed genes in SC and AR-related genes the most down-regulated. As expected, SC lacked ERG and alternative ETS expression and showed relatively low SPINK1 expression. A median summarization of the 356 genes in SCGFt was used to build a small cell genomic score (SCGS). Evaluating SCGS in RP adeno cohorts showed that patients with high SCGS are at higher risk of developing metastasis after RP in a natural history cohort or after adjuvant hormonal therapy based on Kaplan Meier analysis (both p< 0.001). When restricted to patients with high Decipher scores, SCGS provided independent prognostic information. Conclusions: SC carcinoma has a distinct genomic profile that may be useful for treatment management of patients with adenocarcinoma.

Validation of a Genomic Classifier for Predicting Post-Prostatectomy Recurrence in a Community Based Health Care Setting

PURPOSE

We determined the value of Decipher®, a genomic classifier, to predict prostate cancer outcomes among patients after prostatectomy in a community health care setting.

MATERIALS AND METHODS

We examined the experience of 224 men treated with radical prostatectomy from 1997 to 2009 at Kaiser Permanente Northwest, a large prepaid health plan in Portland, Oregon. Study subjects had aggressive prostate cancer with at least 1 of several criteria such as preoperative prostate specific antigen 20 ng/ml or greater, pathological Gleason score 8 or greater, stage pT3 disease or positive surgical margins at prostatectomy. The primary end point was clinical recurrence or metastasis after surgery evaluated using a time dependent c-index. Secondary end points were biochemical recurrence and salvage treatment failure. We compared the performance of Decipher alone to the widely used CAPRA-S (Cancer of the Prostate Risk Assessment Post-Surgical) score, and assessed the independent contributions of Decipher, CAPRA-S and their combination for the prediction of recurrence and treatment failure.

RESULTS

Of the 224 patients treated 12 experienced clinical recurrence, 68 had biochemical recurrence and 34 experienced salvage treatment failure. At 10 years after prostatectomy the recurrence rate was 2.6% among patients with low Decipher scores but 13.6% among those with high Decipher scores (p=0.02). When CAPRA-S and Decipher scores were considered together, the discrimination accuracy of the ROC curve was increased by 0.11 compared to the CAPRA-S score alone (combined c-index 0.84 at 10 years after radical prostatectomy) for clinical recurrence.

CONCLUSIONS

Decipher improves our ability to predict clinical recurrence in prostate cancer and adds precision to conventional pathological prognostic measures.

Characterization of 1577 primary prostate cancers reveals novel biological and clinicopathologic insights into molecular subtypes

BACKGROUND

Prostate cancer (PCa) molecular subtypes have been defined by essentially mutually exclusive events, including ETS gene fusions (most commonly involving ERG) and SPINK1 overexpression. Clinical assessment may aid in disease stratification, complementing available prognostic tests.

OBJECTIVE

To determine the analytical validity and clinicopatholgic associations of microarray-based molecular subtyping.

DESIGN, SETTING, AND PARTICIPANTS

We analyzed Affymetrix GeneChip expression profiles for 1577 patients from eight radical prostatectomy cohorts, including 1351 cases assessed using the Decipher prognostic assay (GenomeDx Biosciences, San Diego, CA, USA) performed in a laboratory with Clinical Laboratory Improvements Amendment certification. A microarray-based (m-) random forest ERG classification model was trained and validated. Outlier expression analysis was used to predict other mutually exclusive non-ERG ETS gene rearrangements (ETS(+)) or SPINK1 overexpression (SPINK1(+)).

OUTCOME MEASUREMENTS

Associations with clinical features and outcomes by multivariate logistic regression analysis and receiver operating curves.

RESULTS AND LIMITATIONS

The m-ERG classifier showed 95% accuracy in an independent validation subset (155 samples). Across cohorts, 45% of PCas were classified as m-ERG(+), 9% as m-ETS(+), 8% as m-SPINK1(+), and 38% as triple negative (m-ERG(-)/m-ETS(-)/m-SPINK1(-)). Gene expression profiling supports three underlying molecularly defined groups: m-ERG(+), m-ETS(+), and m-SPINK1(+)/triple negative. On multivariate analysis, m-ERG(+) tumors were associated with lower preoperative serum prostate-specific antigen and Gleason scores, but greater extraprostatic extension (p<0.001). m-ETS(+) tumors were associated with seminal vesicle invasion (p=0.01), while m-SPINK1(+)/triple negative tumors had higher Gleason scores and were more frequent in Black/African American patients (p<0.001). Clinical outcomes were not significantly different among subtypes.

CONCLUSIONS

A clinically available prognostic test (Decipher) can also assess PCa molecular subtypes, obviating the need for additional testing. Clinicopathologic differences were found among subtypes based on global expression patterns.

PATIENT SUMMARY

Molecular subtyping of prostate cancer can be achieved using extra data generated from a clinical-grade, genome-wide expression-profiling prognostic assay (Decipher). Transcriptomic and clinical analysis support three distinct molecular subtypes: (1) m-ERG(+), (2) m-ETS(+), and (3) m-SPINK1(+)/triple negative (m-ERG(-)/m-ETS(-)/m-SPINK1(-)). Incorporation of subtyping into a clinically available assay may facilitate additional applications beyond routine prognosis.

Novel Biomarker Signature That May Predict Aggressive Disease in African American Men With Prostate Cancer

PURPOSE

We studied the ethnicity-specific expression of prostate cancer (PC) -associated biomarkers to evaluate whether genetic/biologic factors affect ethnic disparities in PC pathogenesis and disease progression.

PATIENTS AND METHODS

A total of 154 African American (AA) and 243 European American (EA) patients from four medical centers were matched according to the Cancer of the Prostate Risk Assessment postsurgical score within each institution. The distribution of mRNA expression levels of 20 validated biomarkers reported to be associated with PC initiation and progression was compared with ethnicity using false discovery rate, adjusted Wilcoxon-Mann-Whitney, and logistic regression models. A conditional logistic regression model was used to evaluate the interaction between ethnicity and biomarkers for predicting clinicopathologic outcomes.

RESULTS

Of the 20 biomarkers examined, six showed statistically significant differential expression in AA compared with EA men in one or more statistical models. These include ERG (P < .001), AMACR (P < .001), SPINK1 (P = .001), NKX3-1 (P = .03), GOLM1 (P = .03), and androgen receptor (P = .04). Dysregulation of AMACR (P = .036), ERG (P = .036), FOXP1 (P = .041), and GSTP1 (P = .049) as well as loss-of-function mutations for tumor suppressors NKX3-1 (P = .025) and RB1 (P = .037) predicted risk of pathologic T3 disease in an ethnicity-dependent manner. Dysregulation of GOLM1 (P = .037), SRD5A2 (P = .023), and MKi67 (P = .023) predicted clinical outcomes, including 3-year biochemical recurrence and metastasis at 5 years. A greater proportion of AA men than EA men had triple-negative (ERG-negative/ETS-negative/SPINK1-negative) disease (51% v 35%; P = .002).

CONCLUSION

We have identified a subset of PC biomarkers that predict the risk of clinicopathologic outcomes in an ethnicity-dependent manner. These biomarkers may explain in part the biologic contribution to ethnic disparity in PC outcomes between EA and AA men.

Cyclin D1 Loss Distinguishes Prostatic Small-Cell Carcinoma from Most Prostatic Adenocarcinomas

PURPOSE

Small-cell neuroendocrine differentiation in prostatic carcinoma is an increasingly common resistance mechanism to potent androgen deprivation therapy (ADT), but can be difficult to identify morphologically. We investigated whether cyclin D1 and p16 expression can inform on Rb functional status and distinguish small-cell carcinoma from adenocarcinoma.

EXPERIMENTAL DESIGN

We used gene expression data and immunohistochemistry to examine cyclin D1 and p16 levels in patient-derived xenografts (PDX), and prostatic small-cell carcinoma and adenocarcinoma specimens.

RESULTS

Using PDX, we show proof-of-concept that a high ratio of p16 to cyclin D1 gene expression reflects underlying Rb functional loss and distinguishes morphologically identified small-cell carcinoma from prostatic adenocarcinoma in patient specimens (n = 13 and 9, respectively). At the protein level, cyclin D1, but not p16, was useful to distinguish small-cell carcinoma from adenocarcinoma. Overall, 88% (36/41) of small-cell carcinomas showed cyclin D1 loss by immunostaining compared with 2% (2/94) of Gleason score 7-10 primary adenocarcinomas at radical prostatectomy, 9% (4/44) of Gleason score 9-10 primary adenocarcinomas at needle biopsy, and 7% (8/115) of individual metastases from 39 patients at autopsy. Though rare adenocarcinomas showed cyclin D1 loss, many of these were associated with clinical features of small-cell carcinoma, and in a cohort of men treated with adjuvant ADT who developed metastasis, lower cyclin D1 gene expression was associated with more rapid onset of metastasis and death.

CONCLUSIONS

Cyclin D1 loss identifies prostate tumors with small-cell differentiation and may identify a small subset of adenocarcinomas with poor prognosis. Clin Cancer Res; 21(24); 5619-29. ©2015 AACR.

Clinical and genomic analysis of metastatic prostate cancer progression with a background of postoperative biochemical recurrence

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.

A genomic classifier to improve prediction of metastatic disease within 5 years after surgery in node-negative high-risk prostate cancer patients managed by radical prostatectomy without adjuvant therapy

154

Background: Surgery is a standard first line therapy for most intermediate-high risk men diagnosed with prostate cancer. While clinical factors such as tumor grade, stage and prostate specific antigen (PSA) are currently used to identify patients at risk of cancer recurrence, novel biomarkers that can improve risk stratification and distinguish local from systemic recurrence are needed. The Decipher Genomic Classifier (GC) is a validated model for predicting men at risk of metastasis. We evaluated its performance in predicting metastatic disease within 5 years after surgery (rapid metastasis, RM) in an independent cohort. Methods: Tumors and clinicopathologic data were obtained from a cohort of 2,641 RP patients treated between 1987-2008 at Cleveland Clinic. The final study cohort consisted of 15 RM patients and 154 patients as non-RM controls who met the following criteria: 1) preoperative PSA>20 ng/mL, stage pT3 or margin positive, or Gleason score ≥8; 2) pathologic node negative; 3) undetectable post-RP PSA; 4) no neoadjuvant or adjuvant therapy; and 5) minimum 5 year follow-up for the controls. Results: RM patients developed metastasis with a median of 2.3 (IQR: 0.8-5) years. In multivariable analysis, GC was a significant predictor of RM (OR=1.48, p=0.018) after adjusting for clinical risk factors. GC had the highest c-index, 0.77, compared to the Stephenson model (c-index 0.75) and CAPRA-S (c-index 0.72) as well as a panel of previously reported prostate cancer biomarkers unrelated to GC. Integration of GC into the Stephenson nomogram increased the c-index from 0.75 (95% CI: 0.65-0.85) to 0.79 (95% CI: 0.68-0.89). Conclusions: GC was independently validated as a genomic metastasis signature for predicting RM in a cohort of high-risk men treated with RP and managed conservatively without any adjuvant therapy. Integration of GC into clinical nomograms led to improvement in prediction of RM. GC may further allow identification of men most at risk for metastatic progression who should be considered for multimodal therapy or inclusion in clinical trials.

A novel biomarker signature to predict aggressive disease in African-American men with prostate cancer

24

Background: Numerous studies have reported a significantly higher incidence of PCa and/or adverse pathological features associated with African-American men compared to European-American men. Less however is known about the genomic disparities that exist between these two groups. In this report we compared the race-specific expression of biomarkers linked to PCa pathogenesis in a matched cohort of AA and EA men. Methods: PCa data from AA and EA patients were analyzed from four medical centers. Cases were matched based on CAPRA-S within each institution for a total sample size of 300 (121-AA; 179-EA). The distribution of mRNA expression levels of 20 validated biomarkers associated with PCa initiation and progression was compared by race using a false-discovery-rate adjusted Mann-Whitney U, and logistic regression models. Conditional logistic regression models were used to evaluate the interaction between race and biomarker expression for predicting pathologic T3 PCa. Results: Of 20 biomarkers interrogated, 6 showed statistically significant differential expression in AA compared with EA men in one or more statistical models. These include TMPRSS2-ERG (p<0.001), AMACR (p<0.001), SPINK1 (p=0.005), AR (p=0.018), SRD5A2 (p=0.005), and GSTP1 (p=0.021). Dysregulation of MYCBP (p=0.043) increases risk of pT3 disease in AA but decreases the risk in EA men, while the reverse is true for AMACR (p=0.013), TMPRSS2-ERG (p=0.026), FOXP1 (p=0.016), and GSTP1 (p=0.032). Loss-of-function mutation for tumor suppressors PTEN (p=0.046), TP53 (p=0.042), and RB1 (p=0.027), and dysregulation of AR (p=0.015), EZH2 (p=0.043), NKX3-1 (p=0.024), SRD5A2 (p=0.032), and SPOP (p=0.032) increased risk of pT3 disease for both AA and EA men. Conclusions: We have identified a subset of PCa biomarkers that predict risk of clinico-pathologic outcomes in a race-dependent manner. These biomarkers may in part explain the biological contribution to racial disparity in PCa outcomes between EA and AA men.

A genomic classifier improves prediction of metastatic disease within 5 years after surgery in node-negative high-risk prostate cancer patients managed by radical prostatectomy without adjuvant therapy

BACKGROUND

Surgery is a standard first-line therapy for men with intermediate- or high-risk prostate cancer. Clinical factors such as tumor grade, stage, and prostate-specific antigen (PSA) are currently used to identify those who are at risk of recurrence and who may benefit from adjuvant therapy, but novel biomarkers that improve risk stratification and that distinguish local from systemic recurrence are needed.

OBJECTIVE

To determine whether adding the Decipher genomic classifier, a validated metastasis risk-prediction model, to standard risk-stratification tools (CAPRA-S and Stephenson nomogram) improves accuracy in predicting metastatic disease within 5 yr after surgery (rapid metastasis [RM]) in an independent cohort of men with adverse pathologic features after radical prostatectomy (RP).

DESIGN, SETTING, AND PARTICIPANTS

The study population consisted of 169 patients selected from 2641 men who underwent RP at the Cleveland Clinic between 1987 and 2008 who met the following criteria: (1) preoperative PSA>20 ng/ml, stage pT3 or margin positive, or Gleason score≥8; (2) pathologic node negative; (3) undetectable post-RP PSA; (4) no neoadjuvant or adjuvant therapy; and (5) minimum of 5-yr follow-up for controls. The final study cohort consisted of 15 RM patients and 154 patients as non-RM controls.

OUTCOME MEASUREMENTS AND STATISTICAL ANALYSIS

The performance of Decipher was evaluated individually and in combination with clinical risk factors using concordance index (c-index), decision curve analysis, and logistic regression for prediction of RM.

RESULTS AND LIMITATIONS

RM patients developed metastasis at a median of 2.3 yr (interquartile range: 1.7-3.3). In multivariable analysis, Decipher was a significant predictor of RM (odds ratio: 1.48; p=0.018) after adjusting for clinical risk factors. Decipher had the highest c-index, 0.77, compared with the Stephenson model (c-index: 0.75) and CAPRA-S (c-index: 0.72) as well as with a panel of previously reported prostate cancer biomarkers unrelated to Decipher. Integration of Decipher into the Stephenson nomogram increased the c-index from 0.75 (95% confidence interval [CI], 0.65-0.85) to 0.79 (95% CI, 0.68-0.89).

CONCLUSIONS

Decipher was independently validated as a genomic metastasis signature for predicting metastatic disease within 5 yr after surgery in a cohort of high-risk men treated with RP and managed conservatively without any adjuvant therapy. Integration of Decipher into clinical nomograms increased prediction of RM. Decipher may allow identification of men most at risk for metastatic progression who should be considered for multimodal therapy or inclusion in clinical trials.

PATIENT SUMMARY

Use of Decipher in addition to standard clinical information more accurately identified men who developed metastatic disease within 5 yr after surgery. The results suggest that Decipher allows improved identification of the men who should consider secondary therapy from among the majority that may be managed conservatively after surgery.

Discovery and validation of novel expression signature for postcystectomy recurrence in high-risk bladder cancer

Background

Nearly half of muscle-invasive bladder cancer patients succumb to their disease following cystectomy. Selecting candidates for adjuvant therapy is currently based on clinical parameters with limited predictive power. This study aimed to develop and validate genomic-based signatures that can better identify patients at risk for recurrence than clinical models alone.

Methods

Transcriptome-wide expression profiles were generated using 1.4 million feature-arrays on archival tumors from 225 patients who underwent radical cystectomy and had muscle-invasive and/or node-positive bladder cancer. Genomic (GC) and clinical (CC) classifiers for predicting recurrence were developed on a discovery set (n = 133). Performances of GC, CC, an independent clinical nomogram (IBCNC), and genomic-clinicopathologic classifiers (G-CC, G-IBCNC) were assessed in the discovery and independent validation (n = 66) sets. GC was further validated on four external datasets (n = 341). Discrimination and prognostic abilities of classifiers were compared using area under receiver-operating characteristic curves (AUCs). All statistical tests were two-sided.

Results

A 15-feature GC was developed on the discovery set with area under curve (AUC) of 0.77 in the validation set. This was higher than individual clinical variables, IBCNC (AUC = 0.73), and comparable to CC (AUC = 0.78). Performance was improved upon combining GC with clinical nomograms (G-IBCNC, AUC = 0.82; G-CC, AUC = 0.86). G-CC high-risk patients had elevated recurrence probabilities (P < .001), with GC being the best predictor by multivariable analysis (P = .005). Genomic-clinicopathologic classifiers outperformed clinical nomograms by decision curve and reclassification analyses. GC performed the best in validation compared with seven prior signatures. GC markers remained prognostic across four independent datasets.

Conclusions

The validated genomic-based classifiers outperform clinical models for predicting postcystectomy bladder cancer recurrence. This may be used to better identify patients who need more aggressive management.

Validation of a genomic classifier that predicts metastasis following radical prostatectomy in an at risk patient population

PURPOSE

Patients with locally advanced prostate cancer after radical prostatectomy are candidates for secondary therapy. However, this higher risk population is heterogeneous. Many cases do not metastasize even when conservatively managed. Given the limited specificity of pathological features to predict metastasis, newer risk prediction models are needed. We report a validation study of a genomic classifier that predicts metastasis after radical prostatectomy in a high risk population.

MATERIALS AND METHODS

A case-cohort design was used to sample 1,010 patients after radical prostatectomy at high risk for recurrence who were treated from 2000 to 2006. Patients had preoperative prostate specific antigen greater than 20 ng/ml, Gleason 8 or greater, pT3b or a Mayo Clinic nomogram score of 10 or greater. Patients with metastasis at diagnosis or any prior treatment for prostate cancer were excluded from analysis. A 20% random sampling created a subcohort that included all patients with metastasis. We generated 22-marker genomic classifier scores for 219 patients with available genomic data. ROC and decision curves, competing risk and weighted regression models were used to assess genomic classifier performance.

RESULTS

The genomic classifier AUC was 0.79 for predicting 5-year metastasis after radical prostatectomy. Decision curves showed that the genomic classifier net benefit exceeded that of clinical only models. The genomic classifier was the predominant predictor of metastasis on multivariable analysis. The cumulative incidence of metastasis 5 years after radical prostatectomy was 2.4%, 6.0% and 22.5% in patients with low (60%), intermediate (21%) and high (19%) genomic classifier scores, respectively (p<0.001).

CONCLUSIONS

Results indicate that genomic information from the primary tumor can identify patients with adverse pathological features who are most at risk for metastasis and potentially lethal prostate cancer.

Whole-transcriptome profiling of thyroid nodules identifies expression-based signatures for accurate thyroid cancer diagnosis

PURPOSE

Due to the limitations of fine-needle aspiration biopsy (FNAB) cytopathology, many individuals who present with thyroid nodules eventually undergo thyroid surgery to diagnose thyroid cancer. The objective of this study was to use whole-transcriptome profiling to develop and validate a genomic classifier that significantly improves the accuracy of preoperative thyroid cancer diagnosis.

MATERIALS AND METHODS

Nucleic acids were extracted and amplified for microarray expression analysis on the Affymetrix Human Exon 1.0 ST GeneChips from 1-mm-diameter formalin-fixed and paraffin-embedded thyroid tumor tissue cores. A training group of 60 thyroidectomy specimens (30 cancers and 30 benign lesions) were used to assess differential expression and for subsequent generation of a genomic classifier. The classifier was validated in a blinded fashion on a group of 31 formalin-fixed and paraffin-embedded thyroid FNAB specimens.

RESULTS

Expression profiles of the 57 thyroidectomy training and 31 FNAB validation specimens that passed a series of quality control steps were analyzed. A genomic classifier composed of 249 markers that corresponded to 154 genes, had an overall validated accuracy of 90.0% in the 31 patient FNAB specimens and had positive and negative predictive values of 100% and 85.7%, respectively. The majority of the identified markers that made up the classifier represented non-protein-encoding RNAs.

CONCLUSIONS

Whole-transcriptome profiling of thyroid nodule surgical specimens allowed for the development of a genomic classifier that improved the accuracy of preoperative thyroid cancer FNAB diagnosis.

Development of a genomic-clinical classifier for predicting progression after radical cystectomy in patients with muscle invasive bladder cancer

4542

Background: The mainstay of muscle-invasive bladder cancer treatment is surgical resection with/without multi-agent chemotherapy. Management decisions are based on a small number of clinical and pathologic parameters with poor prognostic and predictive power. There is an urgent need for enhanced biomarkers to guide therapy of this lethal disease. Here we have developed a genomic signature of bladder cancer progression using whole transcriptome profiling technology. Methods: 251 FFPE bladder cancer specimens were obtained from patients undergoing radical cystectomy at the University of Southern California (1998-2004). All patients had pT2-T4a,N0 urothelial carcinoma in the absence of pre-operative chemotherapy. Median follow-up was 5 years. RNA expression levels were measured with 1.4 million feature oligonucleotide microarrays. Patients were divided into a training set (2/3 of cohort) to develop a genomic classifier for risk of progression (defined as any type of bladder cancer recurrence), and a validation set (1/3 of cohort). In parallel, multivariable analysis was used to develop a clinical classifier using typical clinical and pathologic variables. Finally, a genomic-clinical classifier was built combining the genomic classifier with clinical variables using logistic regression. The receiver-operator characteristic (ROC) area under the curve (AUC) metric was used to evaluate each classifier in the validation set. Results: The genomic classifier consisted of 89 features corresponding to 80 genes that were combined in a k-nearest neighbor model (KNN89). KNN89 showed an AUC of 0.77 in ROC analysis on the validation set. The best clinical classifier showed an AUC of 0.72. The genomic-clinical classifier demonstrated an AUC of 0.81. Multivariable analysis incorporating all clinical parameters and KNN89 further revealed that KNN89 was the only significant predictor of bladder cancer progression (p=0.0077). Conclusions: We have developed a combined genomic-clinical classifier that shows improved performance over clinical models alone for prediction of progression after radical cystectomy. External validation of this classifier is ongoing.

Validation of a genomic classifier that predicts metastatic disease progression in men with high-risk pathologic features postprostatectomy

36

Background: Following radical prostatectomy (RP), 30-40% of patients have adverse pathology and are deemed at high risk for metastatic progression. The objective of this study was to validate the ability of Decipher, a genomic classifier (GC), to improve prediction of metastatic disease progression compared with clinical variables in order to better identify candidates for therapy intensification. Methods: A previously developed 22-feature GC model was validated in a prospectively designed case-cohort study of a clinically high-risk population (i.e., with one or more adverse pathological features) of 1,010 RP patients treated at Mayo Clinic between 2000-2006. A random sample of 20% of the cohort was subjected to microarray analysis and GC scores were generated for 219 patients. The primary endpoint, the c-index for predicting metastatic disease progression (i.e., positive bone or CT scans) was evaluated in a blinded analysis. Cox modeling and decision curve analyses were used to compare the performance of GC to individual clinical variables and prediction models. Results: GC had a c-index 0.79 (95% CI 0.71-0.86) that was significantly better than any single clinical variable. Cumulative incidence curves in the cohort showed that 72% of patients had low GC scores with only 3% and 6% incidence of metastatic disease at 5 and 10 years post RP. In contrast, for the 28% of patients with high GC scores, the cumulative incidence was 17% and 25% at 5 and 10 years post RP. Decision curve analysis showed that the GC model had higher overall net benefit compared to clinical variables over a wide range of ‘decision-to-treat’ thresholds for risk of metastasis. In multivariable modeling with clinicopathologic variables, GC remained the only significant independent predictor of metastasis (HR=1.51, for each 0.1 unit increment, p<0.001). Conclusions: GC can better predict metastatic disease progression compared with clinical variables and may select among patients with adverse pathology a majority that is in fact at low risk for metastasis.

Development and validation of a digital Gleason score biomarker signature for risk stratification of patients with prostate cancer

40

Background: Gleason score (GS) is the most widely used grading system of cell differentiation in prostate cancer and one of the best pathological predictors of disease progression. Patients with high GS (> 7) are the most likely to experience metastasis, whereas patients with low GS (< 7) are expected to have favorable long-term outcomes. In addition to the subjective nature of GS assessment, patients with GS 7 represent a heterogeneous group in terms of patient outcomes. In this study a biomarker signature is developed and validated which could improve the prediction of high risk disease among GS 7 radical prostatectomy (RP) patients. Methods: Patient specimens from the Mayo Clinic RP Registry (n = 764) were used for feature selection and training of a 392-feature K-nearest neighbor (KNN, k = 11) classifier. These 392 genomic features were identified by assessing differential expression between patients with GS < 7 and those with GS > 7, using a Bonferroni adjusted T-Test with a p-value threshold of 0.05. The classifier was subsequently validated in two independent patient cohorts (Memorial Sloan Kettering [MSKCC] and German Cancer Research Center [DKFZ]) and compared to a state of the art biomarker signature (Penney et al. 2011). Results: In the MSKCC dataset (GSE21034) our model segregated GS < 7 from GS > 7 patients (n = 56) with an area under the receiver operating characteristic curve (AUC) of 0.97, comparable to the AUC of 0.94 obtained by Penney et al. in their independent validation set (n = 45). Strong performance was observed by our model when discriminating between primary Gleason grade (pGG) 3 and pGG 4 & 5 patients in the MSKCC (n = 130) and DKFZ prostate cancer (GSE29079, n = 47) datasets, achieving AUCs of 0.75 and 0.87, respectively. In the challenging GS 7 subset, our model segregated GS 3 + 4 and 4 + 3 patients in the DKFZ dataset (n = 64) with an AUC of 0.81 outperforming the Penney et al. signature (AUC = 0.60). Conclusions: A biomarker signature was developed which discriminates between low and high GS patients, outperforming a previously reported signature. Further validation of this biomarker signature in additional post RP patients, as well as, pretreatment biopsy specimens is warranted.

Clinical and genomic analysis of metastatic disease progression in a background of biochemical recurrence

90

Background: Biochemical recurrence (BCR) may be an indicator of metastatic prostate cancer; however, it isn't specific enough to differentiate patients who will experience rapid onset of metastases from those who do not. Such lack of specificity leads to overtreatment of men who will likely never experience metastatic disease in their lifetime. We hypothesize that genomic biomarkers, detectable in primary tumor tissue, may better identify rapidly progressing metastatic prostate cancer. Methods: FFPE prostatectomy specimens (n = 309) from the Mayo Clinic tumor registry were profiled using a high-density expression array. Patients were retrospectively classified into three outcome groups: NED (no evidence of disease recurrence); BCR-only (patients that experienced BCR but no metastatic progression); and MET (metastatic disease confirmed by positive CT or bone scans) and were matched on several criteria for analysis . Uni- and multi-variable analyses between outcome groups were used to analyze clinical and genomic variables (adjusting for adjuvant hormone therapy as a confounding factor). Results: Median follow-up was 16.8 years and MET patients were diagnosed a mean of 3.1 years following BCR. Univariable and multivariable analysis of clinical variables failed to show statistically significant differences between NED and BCR-only outcome groups. Differential expression analysis between NED and BCR-only groups showed no significant differences after false discovery correction. Conversely, 6,277 and 52,020 biomarkers were differentially expressed between MET and the BCR-only and NED groups, respectively. Conclusions: Considering the clinical and genomic results together, patients who experienced rapid metastatic disease were significantly different from both the NED and BCR-only groups. However, no significant differences were identified between the BCR-only and NED outcome groups. These results imply that BCR is not a good surrogate for men who are likely to experience rapid onset of metastatic disease and lethal prostate cancer.

Discovery of a biomarker signature that predicts upgrading or upstaging in patients with low-risk prostate cancer

37

Background: More than 90% of patients diagnosed with organ-confined prostate cancer (PCa) choose upfront definitive treatment (e.g., radical surgery) even though many are excellent candidates for delayed therapy (i.e., active surveillance [AS]). Therefore, patients may suffer from the adverse effects of treatment without gaining any benefit. Biomarker signatures that predict tumour aggressiveness are promising tools for identification of patients suited for AS. In this study, we use a transcriptome-wide assay to develop a biomarker signature for patients assessed as low risk at diagnosis who are upgraded or upstaged following radical prostatectomy (RP). Methods: Gene expression data of 56 RP samples from the Memorial Sloan Kettering Oncogenome Project (GSE21034) which met the low risk criteria (i.e., biopsy Gleason score (GS) ≤ 6, clinical stage T1 or T2A, and pre-operative PSA (pre-op PSA) ≤ 10 ng/ml) were used to develop the signature. Of these tumors, 31 underwent upgrading or upstaging (defined by pathological GS ≥ 7 or a pathological tumor stage > T3A). In the training set (n = 29) a median fold difference filter (MFD > 1.4) was applied to select features. The top 16 t-test ranked features were modelled with a K-nearest-neighbor (KNN) classifier (k = 3) which predicts upgrading/upstaging events. Results: The KNN was applied to the test set (n = 27) and achieved an area under the receiver operating characteristic curve (AUC) of 0.93, significantly better discrimination than pre-op PSA (AUC = 0.52) or tumor stage (AUC = 0.63). Compared to the null model’s accuracy of 56%, the KNN correctly predicts 81% (p-value < 0.005) of the upgrading/upstaging events. In multivariable analysis with pre-op PSA, tumor stage, and age at diagnosis, the KNN remained the only significant (p < 0.05) factor with an odds ratio of 2.7. Conclusions: A 16 marker signature was identified from RP specimens and shown to accurately segregate true low risk patients from those which transitioned to higher risk. Validation studies of this signature in prospectively designed cohorts of active surveillance candidates are underway to determine if the molecular signature can improve treatment and management decisions for low risk PCa patients.

Validation of a genomic-clinical classifier for predicting clinical progression in high-risk prostate cancer

4565

Background: The efficient delivery of adjuvant therapy after radical prostatectomy (RP) in patients with prostate cancer is limited by the lack of biomarkers, beyond clinicopathologic factors, that are able to assess the risk of clinically significant disease progression. Previously, routine FFPE patient specimens from the Mayo Clinic Radical Prostatectomy Registry with long term follow-up were selected to develop a genomic classifier (GC) to predict clinical progression. Here, we present the validation of a GC in a cohort of patients at high risk of disease progression. Methods: A case-cohort study of high-risk RP patients from the Mayo Clinic (N=219) was used to validate the genomic classifier (GC) for predicting clinical progression (defined by positive bone or CT scan post-RP). Its performance was compared to a multivariable clinical classifier (CC) and a genomic-clinical classifier (GCC) which combines GC with established clinicopathologic variables. Concordance index, Cox modeling and decision curve analysis were used to compare the different models. Results: GC and GCC were predictive of clinical progression in the high-risk cohort with c-indices of 0.79 and 0.82, respectively, compared to the clinical classifier (0.70). Multivariable survival analysis showed that the majority of prognostic information of GCC came from the GC with a minor contribution from Gleason score. Decision curve analysis showed that GCC had a higher overall net benefit compared to CC over a wide range of ‘decision-to-treat’ thresholds for the risk of progression. Conclusions: In this high-risk cohort, GC and GCC classifiers showed improved performance over CC in prediction of clinical progression. GC is an independent prognostic factor in this cohort and captures the majority of prognostic information. GC and GCC’s prognostic performance and their usefulness in guiding decision-making in the adjuvant setting after RP need further testing in studies of additional prostate cancer risk groups.