We conducted a clinical utility study of Genetic Risk Scores for cancer risk assessment in a primary care setting. Our goal was to provide a tool that could help further risk stratify patients to ultimately guide more personalized cancer screening plans.
Currently, though colorectal cancer screening recommendations are fairly standard and generally agreed upon by experts, risk is based primarily on family history, recommending earlier and more frequent screening (with colonoscopy) for those with a first- or second-degree family member diagnosed with colorectal cancer.[4, 28, 29] For breast cancer, guidelines regarding screening mammography initiation and frequency have been inconsistent for decades. Currently, the two major guidelines used in primary care are those of the U.S. Preventive Services Task Force (USPSTF)and American Cancer Society. While the American Cancer Society continues to recommend annual mammography screening for all healthy women beginning at age 45, with the option to begin at age 40, the USPSTF now recommends, as of 2009, biannual screening beginning at age 50, unless one has a family history of breast cancer in a first-degree family member. [30, 31] For prostate cancer, a major shift in screening was made in 2012, when the large, U.S., multi-institutional Prostate, Lung, Colorectal, Ovarian (PLCO) clinical trial found prostate specific antigen (PSA) screening to be ineffective in reducing prostate cancer mortality.[32, 33] Based on these results, the USPSTF recommended that PSA screening no longer be performed.
Divergent and conflicting recommendations regarding the initiation and frequency of colorectal, breast, and prostate cancer screening leave many individuals unsure about whether, when, and how often to undergo colonoscopy, screening mammography, or PSA testing. Individuals at increased risk are often advised to undergo screening earlier and more often, but these recommendations are largely based on family history, which, if known at all, may be incomplete or inaccurate. Using our GRS risk-prediction model in combination with family history, patients in a primary care setting can better understand their own individual risk for developing colorectal and either breast or prostate cancer. Afforded this new knowledge, we anticipated that patients at increased risk of developing colorectal, breast, and prostate cancer will be motivated to undergo appropriate cancer screening.
We found that most participants (79.4%) were at low or average genetic risk of breast, prostate and colorectal cancers based on their GRS (Fig. 1d). Approximately one-third of participants, however, reported first- or second-degree family histories of these cancers (Table 1). Though both are important to consider in assessing cancer risk and determining screening plans, we found that only 11.3%, 7.5%, and 26.3% of participants who had a positive FH for breast, prostate or colorectal cancers, respectively, also had a high GRS for the respective disease (Supplemental Fig. 1). Similarly, of participants who had a high GRS for breast, prostate or colorectal cancer, only 53.8, 41.7, and 38.5% of participants reported family history of the disease (Supplemental Fig. 1). This indicates that GRS may identify a new subset of the population at high risk for these cancers who would not have been identified as high-risk based on current risk-assessment criteria.
Though many participants were deemed low risk of developing one of the three tested cancers based on GRS, a significant concern with regards to the clinical utility of this test was whether participants at high, or even average, risk would suffer anxiety as a result of finding out their GRS. We found that those with a high-risk GRS did report significantly more anxiety, as well as worrying more about developing cancer. Of note, anxiety was measured based on how strongly participants agreed with a statement about having anxiety before and after receiving results rather than a verified anxiety screening questionnaire. To further address this concern, a systematic review was performed by Oliveri et al. to analyze the psychological implications of genetic testing. Regarding genetic testing cancer, they found that levels of anxiety and depression decreased significantly after receiving genetic testing results. However, these studies largely involved BRCA1 and BRCA2 testing for breast and ovarian cancers. Though potentially preventable with surgery, considering prophylactic surgery may lead to anxiety. The review also found that knowing results positively affected screening behaviors Alternatively, screening alone can lead to improved health outcomes.
Of note, this study was conducted in a group of particularly motivated patients as evidenced by their participation and follow up in a genetic study. Most had undergone breast (88.2%) or prostate (75.0%) cancer screening within 2 years of this study and/or colorectal cancer (53.5%) within 5 years (data not reported). Thus, it was difficult to assess for increased compliance with recommended cancer screening via EMR data in this population. An important next step in assessing the clinical utility of GRSs should include patient populations with lower cancer screening compliance to assess whether ‘high risk’ scores result in more motivation to undergo cancer screening. Another improvement over the current study design would be a follow up period greater than 3 years to assess the long-term compliance of participants with information about their genetic risk scores for various cancers.
Another limitation of this study was that, at present, most primary care providers do not regularly incorporate genetics into cancer risk assessment. This is, in part, due to the fact that genetics education targeted at primary care providers is lacking. A review of interventions that provide genetics education for primary care physicians found that receiving short-term genetics education did not necessarily lead to apparent changes in practice. The authors also concluded that there are insufficient studies available to be able to inform, and thus improve, current genetics education tailored to primary care physicians. Further, the lack of established guidelines for PCPs to use in advising patients with known GRS for the respective cancers is another hurdle for PCPs to incorporate GRSs into their daily practice.
It is encouraging, however, that larger prospective studies are being conducted to further assess the clinical utility of SNP-based risk scores in targeted cancer screening. Specifically, the ongoing WISDOM trial is comparing standard vs risk-based screening to determine onset and frequency of breast cancer screening via mammography for 100,000 women using a polygenic risk score based on 200 SNPs to stratify risk in the risk-based arm.  With knowledge provided by studies like this, it is promising that PCPs will be more comfortable utilizing genetics, and more specifically SNP-based risk scores, to guide their practice.
Through the present study, we were able to successfully incorporate genetic risk assessments for specific cancers, in the form of Genetic Risk Scores, into primary care practice. By educating 40 primary care physicians and reporting scores through existing EMR/patient portal workflow, GRSs were successfully reported to 280 patients. As data continues to become available regarding novel cancer-risk associated SNPs, associated with the cancers represented in this study as well as other cancers, we encourage further work to expand the use of Genetic Risk Scores in the primary care setting.