Early detection of breast cancer in high-risk women based on longitudinal changes in serum-based proteins: the TESTBREAST study

Background: Women who have a high risk of developing breast cancer due to a familial predisposition or mutations in susceptibility genes undergo adapted screening programs. The earlier onset of screening and higher number of screening moments leaves quite a burden on these women, next to that interval cancers can still occur between scheduled screening moments. Therefore, the prospective, multicenter Dutch TESTBREAST study aims to identify a novel panel of blood-based protein biomarkers. Hereby, the aim is to enable early breast cancer detection for high-risk women using a relatively simple blood test to monitor biomarker levels. Material and methods: The TESTBREAST study was designed to include 1250 women in several hospitals across The Netherlands, of which 40 women were expected to develop breast cancer during the study period. Using longitudinally acquired blood samples of high-risk women which were collected 1–4 times per year during screening appointments and at the time of breast cancer diagnosis, prediagnostic changes in protein levels were analyzed through targeted mass spectrometry. In a nested case-control analysis, serum samples of 3 women who developed a breast malignancy (cases) were longitudinally compared to 3 high-risk women who did not get the disease (controls). Cluster analyses were done to indicate differences between and within protein levels in serum samples of individuals. Statistical analyses were performed using ANOVA to select distinctive proteins for early breast cancer detection. Results: After 10 years, 1174 women have been enrolled in 9 Dutch hospitals. Over 3000 serum samples were acquired, of which some women surpassed 20 study visits. First analyses on 30 longitudinally acquired samples have resulted into unique, strong patterns of protein clustering for each patient, indicating a greater interpatient than intrapatient variability in protein levels of the longitudinally acquired samples. Moreover, a targeted panel of 6 distinctive proteins that are indicative of early breast cancer onset was selected (p < 0.05). These proteins differ between patients and controls using either personalized or population-based cut-off levels and levels already changed 1–2 years before clinical diagnosis. Conclusions: Using targeted mass spectrometry, strong clustering patterns within longitudinal intrapatient samples have demonstrated the importance of identifying small changes in protein levels for individuals over time. This underlines the significance of longitudinal serum measurements, that patients can serve as their own controls, and the relevance of the current study set-up for early detection. Additionally, a panel of 6 proteins for early breast cancer detection in high-risk women was established. Our promising protein panel will be validated in the complete TESTBREAST cohort, offering potential for more frequent (self) testing. No conflict of interest.