An upgrade in prostate cancer prognosis predictions

From the Stanford Group, Public Health Sciences Division

Prostate cancer is the second leading cause of cancer deaths in men in the United States. The Gleason score is currently one of the standard tools used clinically to determine prostate cancer aggressiveness. This score is the sum of two grades, each of which is a value between one and five, based on how much the cancerous prostate tissue looks like normal tissue (low score) or abnormal tissue (high score). Although several factors are considered, the Gleason score is an important component in determining patients’ treatment options. However, the score is not always an accurate estimate of tumor aggressiveness potential and better predictive tools are needed to distinguish aggressive tumors from those that are slow growing. A recently published paper in the journal Prostate by researchers in the Division of Public Health Sciences reports that a panel of DNA methylation sites, when combined with the Gleason score, improves the accuracy of predicting prostate cancer outcomes in men initially diagnosed with local stage tumors.

DNA methylation is a type of epigenetic modification in which methyl groups are added to specific DNA locations referred to as CpG sites. These biochemical modifications are reversible and do not alter the DNA sequence itself but exert regulation over gene expression. Changes in the methylation status of DNA are characteristic of cancer cells and methylation profiles are increasingly being explored as cancer biomarkers. The foundation for the new study was set by previous discovery work by the Fred Hutch authors and collaborators when they identified and initially validated a panel of eight differentially methylated CpG sites predictive of metastatic-lethal prostate cancer in men treated for local stage disease (Zhao et al., 2017).

Dr. Janet Stanford, senior author of the new paper, summarized the efforts and design of the previous research, “This extensive work was made possible by two earlier studies that enrolled incident prostate cancer patients who have remained under long-term follow-up for outcomes (recurrence, metastasis, survival) since their initial diagnosis in either 1993-1996 or 2002-2005.” These studies included conducting detailed in-person interviews, collection of medical records and consent for collection of primary tumor tissue, and blood collection. Furthermore, “Several resources supported processing of the tumor tissues, and the long-term follow-up effort was partially supported by the PNW Prostate Cancer SPORE P50 grant. Collaboration with industry partners (Illumina, Inc.) provided support for generating the genome-wide tumor DNA methylation and gene expression data,” said Dr. Stanford.


Graphical representation of tumor DNA methylation scores by prostate cancer outcomes
Tumor DNA methylation scores by prostate cancer outcomes of no recurrence (blue) or metastatic-lethal (red). Image from the publication

In the new study, the authors first set out to technically validate the eight CpG sites identified in their 2017 study that successfully categorized metastatic-lethal patients from those who remained recurrence free. As explained by Dr. Stanford, “This recent publication reports confirmatory evidence for a prognostic DNA methylation score based on five aberrantly methylated CpGs plus Gleason sum that can distinguish patients who develop metastatic progression or die of prostate cancer compared to those who remain disease-free for five or more years after radical prostatectomy.” Pyrosequencing-based assays were utilized to assess methylation status of 36 tumor DNA samples from the original study. Five of the eight originally identified sites were strongly correlated with the assay used in the prior discovery study, providing technical validation for these CpGs.

Using a new independent study population that had not previously been included in the earlier development and validation study, the authors calculated a 5-CpG methylation score (based on the methylation level of the five CpGs plus Gleason score) and tested whether it could distinguish patients by cancer outcome. The mean DNA methylation score of metastatic-lethal patients was significantly higher than that for patients with no recurrence (see figure).

Next, the authors tested the predictive power of the validated 5-CpG panel when combined with the standard Gleason score to generate a DNA methylation score for each patient. The methylation score was better at predicting prostate cancer outcome as compared to the Gleason score on its own, with a sensitivity of 74% versus 53%, respectively. “In the independent testing dataset, the 5-CpG methylation score reduced the number of men incorrectly classified as false-negatives based on the Gleason sum alone. Thus, with high (>95%) specificity the methylation score could identify an additional ~25% of patients who are at truly at high-risk but who would be missed by Gleason sum alone,” said Dr. Stanford.

When considered in the context of public health, the use of the 5-CpG DNA methylation score in the clinic could help improve treatment planning and follow-up for a broad spectrum of prostate cancer patients. For example, patients with aggressive cancer could be identified sooner and benefit from early adjuvant therapy and more intense surveillance for relapse. In men with low-risk or slow growing tumors, the improved prediction model could reduce overtreatment.

A major goal of biomedical research is to improve precision medicine and results from this study are well suited to advance such efforts. The next step for this research is to conduct a larger trial and Dr. Stanford described the plan, “Fred Hutchinson recently signed a licensing agreement with a precision oncology company based in China (AnchorDx). The company is planning a large clinical study for further validation of these DNA methylation biomarkers, with plans for moving these promising prognostic biomarkers to the clinic for improving patient outcomes.”


This research was supported by the National Cancer Institute.

Research reported in the publication is a collaboration between Cancer Consortium members James Dai, Ziding Feng, and Janet Stanford (Fred Hutch) and Jonathan Wright and Colin Pritchard (UW).

Zhao S, Leonardson A, Geybels MS, McDaniel AS, Yu M, Kolb S, Zong H, Carter K, Siddiqui J, Cheng A, Wright JL, Pritchard CC, Lance R, Troyer D, Fan J-B, Ostrander EA, Dai JY, Tomlins SA, Feng Z, Stanford JL. 2018. A five-CpG DNA methylation score to predict metastatic-lethal outcomes in men treated with radical prostatectomy for localized prostate cancer. Prostate. doi: 10.1002/pros.23667

Additional Citation:

Zhao S, Geybels MS, Leonardson A, Rubicz R, Kolb S, Yan Q, Klotzle B, Bibikova M, Hurtado-Coll A, Troyer D, Lance R, Lin DW, Wright JL, Ostrander EA, Fan J-B, Feng Z, Stanford JL. 2018. Epigenome-wide tumor DNA methylation profiling identifies novel prognostic biomarkers of metastatic-lethal progression in men diagnosed with clinically localized prostate cancer. Clin Cancer Res. doi: 10.1158/1078-0432.CCR-16-0549.