Breast cancer risk assessment: current challenges and the emerging role of blood-based protein biomarkers

Breast cancer risk is influenced by a combination of factors, including genetics, lifestyle, and reproductive history.1-3 Having inherited certain gene mutations as with BRCA1 and BRCA2 or having a family history of breast cancer in close relatives increases breast cancer risk.1,4 However, most women with breast cancer do not have a family history.5 Obesity, lack of physical activity, and alcohol consumption can influence breast cancer risk.6,7 Dense breast tissue can make it harder to detect tumors on mammograms and may also increase risk.8 There are inconsistencies and a lack of consensus in breast cancer screening recommendations. Although risk increases with age, there is concern about breast cancer incidence rising among younger women.9 The US Preventive Services Task Force updated breast cancer screening recommendations with a lowered starting age at 40 years and continuing through age 74 years, with the recommended screening method being biennial screening mammography. This recommendation elicited concern that the evidence is insufficient to support routine screening for women in their 40s given the potential harms from false positives.10 Thus, there is a compelling need to tailor breast cancer screening based on personalized risk. A person’s risk for breast cancer or other cancers may evolve over time.11 A search for biomarkers indicative of breast cancer risk, which may be integrated with a patient’s characteristics, may provide a real-time assessment of risk allowing tailoring screening according to risk and allowing a trajectory of risk biomarker changes over time to be established. A variety of circulating biomarker types may be indicative of a cancer risk including DNA, proteins, and metabolites, among others. Discovery of circulating risk biomarkers may be pursued using platforms appropriate for each type of biomarker. Discovery platforms may be highly elaborate and specialized, but for the resulting findings to be implemented at scale requires a cost-effective way to do the testing as part of regular checkups. In a recent publication in this Journal, Syleouni et al.12 applied a high throughput proteomic technology based on aptamers to identify circulating proteins associated with breast cancer risk among postmenopausal women. The technology they used was intended to quantify levels of 4712 proteins in plasma samples from the Atherosclerosis Risk in Communities study cohort consisting of 4403 women among whom 340 had an incident breast cancer. Of the total aptamers tested, 174 were found to be associated with breast cancer risk prior to false discovery rate (nominal P < .05). Of these, 2 plasma proteins LEG1 homolog and adenosine diphosphatase (ADP)–dependent glucokinase passed false discovery rate and were associated with increased breast cancer risk over the longer-term in postmenopausal women (hazard ratio = 1.45 and 2.52 per doubling, respectively for LEG1 homolog and ADP-dependent glucokinase). Some 79.4% of patients were diagnosed more than 5 years after blood collection. The findings for these 2 proteins were validated in a case-cohort study within the European Prospective Investigation into Cancer and Nutrition cohort. The authors found for both proteins the associations were increasing but not linearly (with a weak correlation of 0.06). To differentiate proteins associated with underlying breast cancer risk from those that are tumor-derived or elevated in response to undiagnosed tumors, a 3-year lagged analysis was performed. The associations for LEG1 homolog and ADP-dependent glucokinase remained robust even after excluding patients diagnosed shortly after blood draw, suggesting that these proteins are more likely indicative of true risk markers rather than markers of subclinical disease. Furthermore, both proteins demonstrated strong performance in predicting medium-term risk (within a 5-year follow-up window) and long-term risk (within a full follow-up time of 23 years), indicating that a single measurement was associated with risk over the medium- to long-term follow-up. Other prior studies have identified proteins associated with breast cancer risk among postmenopausal women (Table 1). Insulin growth factor–related proteins have been associated with breast cancer risk in some studies.13,14 However, the study by Syleouni et al.12 did not find a statistically significant association between insulin-like growth factor and breast cancer risk, although the hazard ratio indicated a modest, positive trend (Table 1). Of note, prior studies have shown that the association between insulin-like growth factor and breast cancer is largely limited to estrogen receptor–positive tumors,15 information that was not available for most participants in this study, potentially contributing to the observed weak association. Comparison of previously established breast cancer risk markers with findings from Syleouni et al.12 Abbreviation: P = P-value. Comparison of previously established breast cancer risk markers with findings from Syleouni et al.12 Abbreviation: P = P-value. In a study using UK Biobank samples, several proteins were associated with risk of breast cancer.16 Three key proteins were identified (amphiregulin, stanniocalcin 2, and cytokine receptor-like factor 1) of which only 1 was ascertained in Syleouni and colleagues’12 study with a reported hazard ratio of 1.21; however, the adjusted P value was greater than .05. Although the UK Biobank cohort included female participants across pre- and postmenopausal stages, this diversity may have contributed to the attenuated performance observed. Other studies based on the Atherosclerosis Risk in Communities cohort have identified genetically predicted and measured plasma proteins in relation to breast cancer risk.17,18 A few of the candidates were ascertained in Syleouni and colleagues’12 study with no statistically significant findings. Another study led to the identification of 22 blood proteins associated with breast cancer risk at a false discovery rate less than 0.05.19 Similarly, a study of women enrolled in the Karolinska Mammography Project using Olink’s Proximity Extension Assay identified 5 proteins to play a potentially etiological role in breast cancer risk.20 None of these markers emerged as statistically significant in the study by Syleouni et al.12 To this end, a persistent challenge in the field of breast cancer risk biomarkers is the lack of consistent and reproducible findings across studies and cohorts. This study makes an important contribution by identifying 2 key biomarkers for breast cancer risk using rigorous statistical methods and independently validating the findings in an external cohort. However, several previously reported markers did not validate in this cohort of postmenopausal women using the current platform. This may be because of differences in breast cancer subtypes, underlying population characteristics, hormonal status, or platform characteristics, which underscore the need for comprehensive and rigorous efforts to discover robust and generalizable biomarkers for breast cancer risk prediction. Importantly, given that breast cancer is a heterogeneous disease, biomarkers may need to be tailored to specific subgroups of women, such as postmenopausal women or women with a family history of breast cancer, and to encompass risk across molecular subtypes, to achieve clinical utility. Whereas the study by Syleouni et al.12 presents a robust and rigorous approach for biomarker discovery for breast cancer risk assessment, additional biomarkers may emerge from other discovery platforms. Moreover, the value of risk biomarkers needs to be considered in the context of established breast cancer risk models such as Tyrer-Cuzick et al.21 and Gail et al.22 to assess their contribution to a comprehensive framework for breast cancer risk prediction. None. Ehsan Irajizad (Conceptualization, Writing—original draft) and Samir Hanash (Conceptualization, Writing—original draft) No funding was used for this editorial. E. Irajizad is an inventor on multiple intellectual property filings related to cancer biomarker discovery. S. Hanash is an inventor on CancerSeek and other single and multicancer biomarkers intellectual property filings. No new data were generated or analyzed for this editorial.