Three presentations during the 2018 American Urological Association Annual Meeting in San Francisco, California, together demonstrated the potential and utility of different assays to identify prostate cancer and guide treatment decisions for patients with prostate cancer. Each assay suggested a simpler and more cost-effective tool for guiding decision making in prostate cancer than immediate tissue biopsy.
James M. McKiernan, MD
Three presentations during the 2018 American Urological Association (AUA) Annual Meeting in San Francisco, California, together demonstrated the potential and utility of different assays to identify prostate cancer and guide treatment decisions for patients with prostate cancer. Each assay suggested a simpler and more cost-effective tool for guiding decision making in prostate cancer than immediate tissue biopsy.
One such study of a urine-based gene-expression assay showed greater than 90% accuracy for distinguishing between high- and low-grade prostate cancers, results from more than 1000 patients showed.1
In 2 separate patient cohorts, the exosome-expression assay identified 92% to 93% of cancers that proved to have Gleason score ≥7/ International Society of Urological Pathology grade ≥2 at pathology. Use of the test to guide biopsy decisions eliminated 20.1% of biopsies in one cohort and 26.6% in the second.
“The EPI [ExoDx Prostate (IntelliScore)] is a noninvasive, easy-to-use, 3-gene expression urine assay,” James M. McKiernan, MD, the John K. Lattimer Professor of Urology at Columbia University and the urologist-in-chief at NewYork-Presbyterian/ Columbia, said during a presscast at the AUA annual meeting. “In 2 independent prospective validation studies, we showed that the EPI discriminates high-grade from low-grade prostate cancer and benign disease.”
“The EPI improves stratification of patients with higher-grade disease and improves the number of unnecessary biopsies,” McKiernan said.
The assay takes advantage of what McKiernan described as the “rich source of molecular information” contained in exosomes, which are excreted by virtually all cells into all biofluids. Structurally, exosomes are lipid-bilayer protected vesicles that maintain stability under varying conditions, protecting their contents from degradation. Exosome contents include multiple types of RNA as well as DNA and proteins.
The EPI requires a urine sample (not from a digital rectal exam), from which exosomal RNA is extracted. After extraction, reverse transcription-quantitative polymerase chain reaction is used to determine the gene signature associated with exosomal RNA for 3 genes: ERG, PCA3, and SPDEF. The signature undergoes analysis by a multivariate algorithm, resulting in a risk score (range, 0-100). A score >15.6 is associated with an increased likelihood of high-grade prostate cancer.
An initial validation study involved 519 men scheduled for prostate biopsy as a result of elevated prostate-specific antigen (PSA) levels (2-10 ng/mL), enrolled at 22 sites. The median age in the study was 63 years, and 22.5% of patients had a family history of prostate cancer. The EPI results identified high-grade prostate cancer in 28.5% of the patients, and 26.6% of biopsies would have been avoided. Performance metrics showed that the assay had a sensitivity of 91.9%, a specificity of 34.0%, and a negative predictive value of 91.3%.2
McKiernan reported findings from a second validation assessment of the assay, involving 503 men with baseline characteristics similar to those of the men in the first assessment. The cohort had a median age of 64 years and a median PSA of 5.4 ng/mL. African Americans accounted for 14.1% of the cohort, and 14.3% of the study participants had a family history of prostate cancer. The study involved investigators at 14 sites.
Use of the EPI cut point of 15.6 led to detection of high-grade cancer in 31.4% of the patients, 20.1% of whom could have avoided biopsies. The test had a sensitivity of 93.0%, a specificity of 26.1%, and a negative predictive value of 89.1%.
ROC analysis yielded an area under the curve (AUC) of 0.70 for discrimination of high-grade prostate cancer. That compared with an AUC of 0.63 for standard of care (PSA plus risk factors) and 0.58 for PSA alone. The initial validation study yielded similar results, with an AUC of 0.71 for EPI, 0.62 for standard of care, and 0.54 for PSA alone.
“I have to emphasize that this is a secondary test,” said McKiernan. “The studies involved men who already had a test [PSA] that singled them out as being at high risk. Performance characteristics of tests like this [EPI] can shift massively, depending on the type of patient population.”
DISTINGUISHING BETWEEN RISK SUBGROUPS
A follow-up validation study of an assay for abnormal PSA isoforms also showed potential for distinguishing high-risk (Gleason ≥7) from low-risk (Gleason = 6) prostate cancer and benign disease.
Development of the IsoPSA test had its genesis in recognition that abnormal PSA proteins emerge during prostate cancer evolution. Rather than focus on specific PSA isoforms, the assay detects any malignancy-associated protein, Mark Stovsky, MD, MBA, explained during a presentation at the AUA meeting.3
In a preliminary validation study involving 261 men scheduled for PSA biopsy, the assay demonstrated potential as a decision-making aid. Pathology results showed an overall prevalence of 53% for any prostate cancer and 34% for high-grade disease. ROC analysis for the assay yielded an AUC of 0.79 for any cancer and 0.81 for high-grade cancer. The results demonstrated a 45% to 48% reduction in false-positive biopsies, depending on the chosen risk threshold. The recommended cutoff for biopsy is a 17% likelihood of high-grade disease.4
Stovsky, a consultant urologic oncologist at the Cleveland Clinic, reported findings during the meeting from a multicenter follow-up study involving 271 men scheduled for prostate biopsy, prompted by abnormal results from PSA tests. Results obtained with the assay were compared with findings from a standard 12-core biopsy. A subgroup of patients underwent MRI assessment of the prostate followed by MR-ultrasound fusion biopsy, and the analysis of the assay’s performance was repeated for that subgroup.
ROC analysis produced an AUC of 0.79. The assay’s performance improved in the analysis of patients who underwent MRI-ultrasound fusion biopsy up to an AUC of 0.83 when patients with Gleason 6 pathology were included and 0.87 when the analysis included only patients with Gleason ≥7.
“MR-ultrasound fusion is superior at finding high-grade disease,” said Stovsky. “Using the MR fusion biopsy results to evaluate IsoPSA showed both high exclusion powera low chance for high-grade prostate cancer—and a very high specificity to cancer with acceptable sensitivity.”
A statistical model that included IsoPSA results and age slightly improved the performance metrics, he added. A model that incorporated prostate volume, race, and PSA level did not improve performance metrics, however.
Several additional evaluations of IsoPSA have been planned to determine the test’s performance in different clinical scenarios: larger multicenter studies with greater reliance on the emerging standard of MR-ultrasound fusion biopsies, evaluation of the assay in conjunction with MRI-based parameters as a means of monitoring grade migration in patients on active surveillance, and studies enriched with high-risk populations.
COMPARING GENOMIC TESTS
A third presentation provided data from a small retrospective comparison of the 3 genomic assays currently available for evaluating prostate cancer risk: Prolaris, Decipher, and Oncotype DX.
Genomic testing may help risk-stratify patients beyond traditional criteria and aid in treatment decision making, Joseph R. Wagner, MD, a urologic oncologist at Hartford HealthCare Cancer Institute in Connecticut and the director of robotic surgery at Hartford Hospital, said when presenting at the AUA presscast. According to the 2018 National Comprehensive Cancer Network (NCCN) guidelines, genomic testing “can be considered” for evaluation of low- and favorable intermediate-risk prostate cancers in men who have a life expectancy of at least 10 years.5
By means of chart review, Wagner and colleagues identified 22 patients whose prostate cancer had been evaluated by at least 2 of the available genomic tests. By NCCN criteria, 17 of the men had very low- or low-risk prostate cancer, 3 had favorable intermediate-risk disease, and 2 had unfavorable intermediate-risk disease.
Records showed that 20 patients had undergone evaluation with the Prolaris assay, 15 with Decipher, and 10 with the Oncotype DX assay. The Prolaris results agreed with the NCCN criteria, supporting active surveillance in 15 of 20 cases (75%). The Decipher results agreed with NCCN in 9 of 15 cases (60%), and the Oncotype DX in 5 of 10 cases (50%).
None of the comparisons produced results that achieved statistical significance. Moreover, kappa analysis showed less than moderate agreement (≥0.60) with NCCN criteria for the Prolaris test (k = 0.21) or Decipher (k = 0.15). Too few patients had been tested with the Oncotype DX to perform a kappa analysis, Wagner said.
Medical records showed that 12 patients had been evaluated by both the Decipher and Prolaris tests, 8 by Prolaris and Oncotype DX, and 2 by the Decipher and Oncotype DX assays. The Decipher and Prolaris results were concordant in 8 of 12 cases (67%, k = 0.31, P = .276), Prolaris and Oncotype DX in 6 of 8 (75%, k = 0.39, P = .168), and the Decipher and Oncotype DX agreed on one case and disagreed on the other.
Comparison of genomic test results and surgical pathology in 8 patients showed that Oncotype DX results were concordant in 1 of 2 cases, Prolaris in 5 of 6, and Decipher in 5 of 8.
Wagner suggested that the 3 tests might be improved by incorporating additional information, if feasible. “Can mutation patterns be used to determine [whether] a patient is better suited for a certain type of local therapy?” he asked. “Could Oncotype DX and Decipher add outcomes for radiation and conservative therapy [in addition to surgery]? Could Prolaris separate outcomes for radiation and surgery and provide surgical pathology predictions?