August 2012, Vol 1, No 3

← Back to Issue

Prostate Cancer Roundup

Alice Goodman


Nearly 3000 abstracts were selected for presentation at the recent ASCO 2012 Annual Meeting, many of them related to some aspect of personalized medicine. Below are some highlights selected from the meeting that focus on potential genomic predictors of aggressive versus indolent disease and on potential biomarkers.

6-Gene Model Identifies Lower- Versus Higher-Risk CRPC Patients

A 6-gene model was found to discriminate between lower-risk patients and higher-risk patients with castration-resistant prostate cancer (CRPC) in both a training set and a validation study (Abstract 4516). Current models for risk assessment are based on clinical variables and only offer moderate predictive discrimination for men with CRPC who have a heterogeneous range of outcomes. Whole blood offers specific advantages as a biomarker – it is easy to collect, minimally invasive, can be standardized, and can be repeatedly collected over time.

“We demonstrated that the 6-gene model predicted survival,” stated presenting author William Oh, MD, professor at Mount Sinai School of Medicine in New York City.

Between August 2006 and June 2008, PAXgene Blood DNA Tubes were used to collect blood prospectively from 62 patients for a training set at Dana-Farber Cancer Institute, Oh told listeners. Subsequently, the researchers collaborated with the Memorial Sloan-Kettering Cancer Center (MSKCC), New York, for a validation set from 140 patients who had blood samples banked between August 2006 and February 2009. Two samples were eliminated because of poor-quality RNA. After an extensive review of studies in the literature, the researchers identified 6 candidate genes that would yield the best prediction of survival.

“When applied to the training set at Dana-Farber, we found that the lower-risk patients had a median survival of 34.9 months, while higher-risk patients had a median survival of 7.8 months (P=.0001),” Oh said. The gene model was superior to the Halabi nomogram variables based on data available for 6 of 7 of the variables, namely, alkaline phosphatase, ECOG performance status, hemoglobin, visceral metastases, prostate-specific antigen (PSA), and Gleason score. Area under the curve was 0.90 for the 6-gene model and 0.65 for the clinical model.

The MSKCC validation set had a median survival of 18.5 months for lower-risk patients and 9.2 months for the higher-risk group (P<.0001). As with the training set, the results were highly significant. The 6-gene model maintained its prognostic significance when clinical variables were added to it. The authors hope this study will provide models to help assist patient counseling and trial stratification.

Patient characteristics were typical for patients with CRPC. Metastatic disease was present in 87% and 90% of the training and validation cohorts, respectively.

The study was funded by Source MDx, which is no longer in business.

Gene Classifiers Predict Risk of Clinical Progression Following Prostatectomy

The genomic classifier (GC) and the genomic-clinical classifier (GCC) were validated as predictors of clinical progression after radical prostatectomy in prostate cancer patients at high risk for disease progression (Abstract 4565). Both GC and the GCC were superior to a multivariable clinical classifier (CC) in this regard, supporting the promise of applying GCs in guiding decision making following radical prostatectomy.

Christine Buerki, PhD, of GenomeDx Biosciences, Vancouver, Canada, reported these results, confirming that the GC is able to capture the majority of prognostic information.

The author believes that the lack of biomarkers, beyond clinical and pathologic factors, for predicting risk of clinically significant disease is a barrier to the efficient delivery of adjuvant therapy following prostatectomy.

The GC was developed from the Mayo Clinic radical prostatectomy registry of routine formalin-fixed, paraffin-embedded patient specimens.

In the case cohort study of 219 patients from the Mayo Clinic, clinical progression was defined as a positive bone or CT scan following prostatectomy. C-indices (measures of discrimination for model validation) of 0.79, 0.82, and 0.70 were found for GC, GCC, and CC, respectively.

Multivariable survival analysis revealed that most of the prognostic information of GCC was derived from the GC, with only a small contribution from Gleason score. GCC, which is a combination of GC and established clinical and pathologic variables, had an overall higher net benefit compared with CC over a wide range of decision-to-treat thresholds for the risk of progression. GC emerged as an independent prognostic factor in this study.

The utility of GC and GCC in informing decision making in the adjuvant setting following radical prostatectomy will depend on the results of additional studies in other prostate cancer risk groups.

FDHT and FDG Potential Imaging Biomarkers

Both 18F-16β-fluoro-5α-dihydrotestosterone (FDHT) and fludeoxyglucose (FDG) positive emission tomography (PET) are promising candidates for imaging biomarkers in men with metastatic castrate-resistant prostate cancer (mCRPC), as shown by a study designed to determine if FDHT and FDG PET scans are prognostic for survival (Abstract 4517). These findings suggest that more sophisticated imaging, such as FDHT and FDG, may be helpful in managing mCRPC. Current imaging modalities have limited ability to quantify disease burden and assess response to treatment.

Researchers at MSKCC in New York City prospectively scanned 170 patients in the FDG arm and 116 in the FDHT arm. All patients were diagnosed with mCRPC and had evidence of disease progression at time of the baseline scan.

Presenting author Karen A. Autio, MD, pointed out some important differences between the 2 imaging modalities used in the study. FDG images tumor metabolism but is not tumor specific and assumes that the lesions are glycolytic. FDHT, a structural analog of dihydrotestosterone, has a high affinity for the androgen receptor and captures its overexpression in bone, soft tissue, and viscera. FDHT measures androgen receptor expression and is prostate specific, but its utility requires a castrate state.

Each patient was assessed for standardized uptake values (SUV), specifically, SUVmax (ie, the hottest lesions) or SUVmaxavg (ie, average of the 5 hottest lesions).

“FDHTmaxavg and FDGmaxavg were significantly associated with survival (P=.049 and P=.0007, respectively),” Autio stated. “For FDHT, with a hazard ratio of 1.61, we can say that for every log 1 unit increase in SUV, the risk of death increased by 61%,” Autio said. In comparison, the hazard ratio for FDG was 2.54. In a multivariate model, neither FDHT SUV or FDG SUV was prognostic of survival, and neither tracer was strongly associated with SUVmax.

FDHT was superior to PSA and Gleason score as a prognostic marker of survival.

Preliminary data from this study indicate that both FDG and FDHT are linked to clinical outcome and have potential utility as imaging biomarkers in building an evidence database.

RT-PCR–Based Technique Discriminates Between Indolent and Aggressive Prostate Cancer

Reverse transcriptase-polymerase chain reaction (RT-PCR) provides a reliable measure of gene expression patterns and biological pathways associated with clinically aggressive prostate cancer in radical prostatec­tomy specimens obtained by needle biopsies, according to a study conducted at the Cleveland Clinic, which was confirmed by a study presented at a Poster Discussion Session (Abstract 4560). The technique also discriminated between indolent and aggressive prostate cancer.

The study supports the potential value of a biopsy-based genomic assay to guide the decision between immediate treatment and active surveillance for patients with biopsy-diagnosed prostate cancer. The study was presented by Eric A. Klein, MD, Glickman Urological and Kidney Institute, Cleveland Clinic, Cleveland, Ohio.

The study included 92 low-risk and 75 intermediate-risk patients who were biopsied and underwent radical prostatectomy between 1999 and 2010. The investigators used a novel design to assess gene expression in the context of tumor heterogeneity assessed by needle biopsy of tissue obtained from radical prostatectomy.

The researchers analyzed the expression of 81 prostate cancer–related genes, which were identified in a prior gene discovery study, and normalized to the average of 5 reference genes. Fifty-eight of the 81 discovery study genes (72%) also predicted adverse pathology and/or nonorgan-confined disease when assayed in biopsy tumor tissue. These included all stromal response and androgen genes and most (82%) cellular organization genes. Proportionately fewer proliferation (40%), stress response (29%), and basal epithelial (25%) genes were associated with an adverse path.

After covariate adjustment for clinical T stage, pretreatment PSA, and biopsy Gleason score, the researchers found that the predictive genes identified in biopsy specimens at diagnosis also predicted adverse pathology in biopsy tumor tissue.

An independent prospective study is currently under way to validate a clinical-grade multigene assay optimized for prostate needle core biopsy tissue. The assay is based on an algorithm incorporating the strongest genes and gene pathways.

Uncategorized - September 4, 2012

Facilitating the Next Generation of Precision Medicine in Oncology

Key Points Key stakeholders are increasingly considering new measures to protect and advance innovation and investment in diagnostics Efforts are under way to establish more appropriate and least burdensome regulatory policies to address stakeholder concerns specific to in vitro diagnostic tests FDA is concerned with the risk of false-positive and [ Read More ]

Interview with the Innovators - September 4, 2012

Incorporating Genomics Into Practice:

An Interview with Kimberly J. Popovits

Genomic Health, Inc, located in Redwood City, California, is a global cancer company founded in 2000. Genomic Health currently offers patients the Oncotype DX Breast Cancer Assay and the Oncotype DX Colon Cancer Assay. Genomic Health  maintains that the key to effectively using clinical genomics to improve cancer treatment and [ Read More ]