SAN DIEGO – Information about prostate tumor biology obtained from a multigene assay helped investigators discriminate low- and intermediate-risk prostate cancers in the context of tumor heterogeneity and limited sampling with needle biopsies, a clinical validation study showed.
Furthermore, the information garnered from the assay might take the place of a second biopsy for identifying appropriate patients for active surveillance.
"I’d like to suggest that we are on the very front wave of a paradigm shift," Dr. Eric A. Klein said during a conference sponsored by the American Association for Cancer Research and the Prostate Cancer Foundation. "We have learned so much about the biology and genomics of prostate cancer. We’re at the point now where we can exploit that clinically and make precision decisions for our patients."
The test, known as the Oncotype DX Genomic Prostate Score (GPS), is a biopsy-based pretreatment tool of Genomic Health Inc. that can help clinicians predict which men are more likely to harbor an aggressive form of prostate cancer. It became available in May 2013.
Dr. Eric A. Klein
For patients with newly diagnosed, low- or intermediate-risk prostate cancer, limited accuracy of pretreatment risk assessment has led to underuse of active surveillance and overtreatment of nonlethal cancers, with all of the accompanying morbidity and cost, said Dr. Klein, who chairs the Glickman Urological and Kidney Institute at the Cleveland Clinic and who led the GPS development studies.
"The real problem for patients in active surveillance [is that] biopsy only samples potentially the low-grade tumor," he said. "So the question that exists is, how well does the Gleason grade predict biologic potential? And how accurately does the biopsy capture the biology of the whole prostate? The answer to the second question is unknown."
He discussed efforts to develop and validate a biopsy-based gene expression profile that predicts aggressive prostate cancer in the context of tumor heterogeneity, multifocality, and biopsy undersampling, which "limit the accuracy, precision, and confidence of current risk assessment. The goal was to explore genomics on biopsy to take decision-making from average risk based on grade, stage, and PSA [prostate-specific antigen], to more precisely define individual biological risk across the spectrum of each clinical risk group."
He and his associates conducted two development studies based on patients treated between 1987 and 2007: one to identify genes predictive of clinical recurrence, prostate cancer death, and adverse pathology at prostatectomy across multiple tumor regions sampled from each patient’s prostate, and a second to confirm the predictive value of these genes in prostate biopsies.
Next, independent researchers conducted a clinical validation study, which tested needle biopsies from patients with low to intermediate clinical risk who were treated during 1997-2011. This study was led by Dr. Peter Carroll, chair of the urology department at the University of California, San Francisco, and Dr. Matthew R. Cooperberg, also of UCSF. They used reverse-transcription polymerase chain reaction testing from prostate tumor tissue to quantitate gene expression, and used Cox proportional hazards or logistic regression to analyze associations with clinical recurrence and adverse pathology.
From a cohort sampling of 441 radical prostatectomy patients in the first development study, Dr. Klein and his associates identified 288 genes that predicted metastasis or death regardless of whether the gene expression was measured in the lowest Gleason pattern present or the highest Gleason pattern present. "This was a surprising finding," he said. "It challenges some of our notions about the biology of prostate cancer. What we’re suggesting is that grade, stage, and PSA only give us so much predictive power. We now have the tools available to unleash something that we can’t see under the microscope – the biology of the tumor – in a way that’s clinically exploitable."
For the second development study, the researchers evaluated 81 predictive genes from needle biopsy tissue in 167 patients, 58 of whom had adverse pathology at prostatectomy. Multivariate analysis of both development studies revealed 17 genes representing four biological pathways (stromal response, cellular organization, androgen signaling, and proliferation) that predicted outcome. Dr. Klein characterized the 17 genes as "a window" into the entire prostate.
"There is data in the literature that suggests the genomic signal that is present in patients with metastatic or advanced disease is almost always present in the primary tumor," he said. "It won’t be true in every case, but for the vast majority of prostate cancers I think this is true. The suggestion is that by measuring gene expression in the biopsy we can capture that signal."
The clinical validation study performed at UCSF included needle biopsy tissue from 395 patients, of whom 123 had adverse pathology at time of prostatectomy. The UCSF researchers reported that the GPS algorithm assessed in biopsies with tumor length as little as 1 mm from patients suitable for active surveillance predicted adverse pathology at prostatectomy, after adjustment for conventional pretreatment factors (P less than .005). In addition, the net reclassification improvement corresponding to at least a 5% change in predicted probability of favorable pathology with the addition of GPS to the CAPRA (Cancer of the Prostate Risk Assessment) score was 0.41 (P less than .001).