Liquid Biopsies in Breast Milk for the Early Detection of Breast Cancer

Breast cancer is one of the most common pregnancy-associated malignancies with an estimated incidence of 2.4 to 7.3 per 100,000 live births, predicted to rise, in part, due to increasing maternal age (1). The relationship between pregnancy and breast cancer risk is complex and dependent on several factors including age, family history, parity, and lactation (2). For 10 to 15 years postpartum, there is a transient increased risk of breast cancer, followed by a “cross-over” effect of reduced risk, whereby pregnancy overall becomes protective for women under 35 years (2). Increasing maternal age at the time of first live birth is associated with subsequent higher transient peak risk of breast cancer, such that over the age of 35 years, the longer-term protective effect of pregnancy does not lead to overall reduced risk (2).

Over the last decade, researchers have described an important distinction between breast cancer occurring in pregnancy (PrBC) and up to 10 years postpartum (PPBC) due to the differing clinicopathologic features and the impact on prognosis (3, 4). It is estimated that PrBC accounts for 4% of breast cancers in women <45 years old (4). Historically, PrBC was associated with worse prognosis; however, larger studies distinguishing between PrBC and PPBC have shown that PrBC has similar clinicopathologic features and prognosis when compared with young, nonpregnant patients with breast cancer (5). By comparison, PPBC accounts for approximately 35% to 53% of breast cancer in women <45 years old (4, 6). This phenotype is associated with an increased risk of metastasis (4, 6) and worse overall prognosis after adjusting for risk factors including subtype, stage, and year of diagnosis (4, 7). Breast tissue remodels during weaning, called involution, resulting in a breast microenvironment akin to wound healing with an influx of immunosuppressive cells (2). This altered microenvironment can persist for years after delivery, and it has been speculated that this may promote tumorigenesis and increase risk of metastasis in subclinical cancers (2), although further research is required. Identifying new ways to detect PPBC is a priority, especially as the high breast density of young age reduces mammographic sensitivity.

Liquid biopsy commonly refers to the analysis of circulating tumor DNA (ctDNA) in plasma, and extensive research has been directed toward the use of blood-based liquid biopsy for the early detection of cancer, which requires ultrasensitive assays to detect ctDNA present at very low levels. The current ctDNA detection/sensitivity rates of stage I and II breast cancer with blood-based ctDNA assay are low, suggesting new approaches are required. ctDNA can be detected in nonblood sources, including urine, saliva, stool, cerebrospinal fluid, and pleural fluid (8). Bodily fluids that have been in direct contact with malignant cells may have a higher proportion of ctDNA than plasma, thus increasing the chances of detecting cancer. For example, the ease and acceptability of urine collection make it an ideal biofluid to screen for the early detection of bladder, renal, and prostate cancers, with multiple studies showing high sensitivity (9, 10). Similarly, ctDNA detection in saliva has a sensitivity of 76% for oropharyngeal cancers, increasing to 100% for oral cavity cancers (11). It follows that the close relationship between breast milk (BM) and breast cancer cells may result in a higher concentration of ctDNA and offer a potential method to screen for breast cancer, and prior studies have begun to investigate the idea. Free DNA is detectable in the BM of healthy patients (12). Genome-wide DNA methylation patterns were compared between the BM of patients who later developed breast cancer (n = 23) and those who did not (n = 60; ref. 13). Of the 23 cases of breast cancer, 20 had BM collected prior to diagnosis. Distinct regions of CpG site hypomethylation were associated with breast cancer, compared with controls, in BM (13). Whether this analysis reflected the detection of ctDNA was not addressed in the study, although the study suggested that epigenetic changes in BM might provide a mechanism for identifying patients at risk of breast cancer (13).

Set against this background, Saura and colleagues sought to identify whether the analysis of ctDNA in BM was a fea­sible method for early diagnosis (14). In a case–cohort study, 19 patients diagnosed with PPBC or PrBC and 12 healthy volunteers were recruited, who provided a sample of BM from each breast. ctDNA was detected in the BM from the ipsilateral breast in 13 of 15 cases using digital droplet PCR to track one mutation known to be present in primary tumor sequencing. No ctDNA was detected in BM from the contralateral breast. Given a tumor-informed approach is unsuitable for screening purposes, they validated a custom hybrid capture next-generation sequencing–based panel targeting 54 genes frequently mutated in breast cancer occurring in women <45 years old (VHIO-YWBC). This assay was less sensitive, detecting ctDNA in BM of 10 of 14 cases, but without detection in all 12 healthy controls. Only one of 14 patients had detectable ctDNA in synchronous plasma, suggesting increased sensitivity of ctDNA detection in BM compared with plasma.

This study confirmed that ctDNA can be detected in the BM from the ipsilateral breast with good concordance between mutations in BM and the primary tumor, and as a proof-of-principle study it tantalizingly suggests a potential novel screening tool for the future (Fig. 1). However, there are a number of limitations of the study. Evidence for the clinical validity of this approach is limited by small numbers. PrBC and PPBC are overall relatively rare; therefore, a screening test requires very high specificity to give an acceptable positive predictive value. Even a relatively low false-positive rate could result in low positive predictive values for a test that could result in unnecessary investigations and patient distress, as well as conceivably drive patients to seek potentially unnecessary prophylactic mastectomy. Specificity was 100% in the report but based on only 12 healthy normal patients. Use of a methylation-based ctDNA assay, exploiting differences in DNA methylation between plasma ctDNA and normal cell-free DNA, might conceivably be an option for future larger longitudinal studies to confirm a clinically meaningful specificity.

Future studies will need to consider the populations in whom analysis of ctDNA BM would have the greatest clinical utility. The lead time between BM ctDNA detection and clinical/radiographic presentation with breast cancer has also not been established in the current report. Two cases in the study demonstrated that the detection of ctDNA in BM predated clinical and radiologic diagnosis of breast cancer by up to 18 months (14). One of these cases had paired BM and ultrasound imaging on three occasions prior to diagnosis, with ctDNA detected 6 months prior to imaging diagnosis. On a population level, women are at risk of PPBC up to 10 years after delivery; therefore, further studies are required to establish a lead time between ctDNA detected in BM and diagnosis of PPBC. Although it is estimated that up to 53% of breast cancers in women <45 years are PPBC, the number of pregnancies associated with breast cancer is low, and, therefore, establishing the benefit of screening in an unselected pregnant population will require large patient numbers and long follow-up. The cohort in which screening for ctDNA in BM is most likely to play a role is in patients with a high risk of cancer, such as those with germline BRCA1/2, PALB2, and TP53 mutations, breast irradiation as a consequence of prior hematologic malignancy treatment, or prior diagnosis of breast cancer. Routine screening for patients <45 years at high risk of cancer diagnosis involves imaging with a mammogram or MRI; therefore, a robust comparison between ctDNA detection in BM and standard imaging techniques will be required. Finally, an interesting set of practical considerations would lie ahead, such as requirement for screening BM from both breasts, with what frequency, and is there an optimal time of sampling from delivery/commencement of lactation for ctDNA detection?

In conclusion, BM provides a novel fluid for liquid biopsy, in which the analysis of ctDNA is feasible. Further research directed toward establishing the optimal assay, clinical validity, and utility is required, ideally before the initiation of screening trials. With the worse outcomes of PPBC, developing novel strategies for early detection is an important aim, as is research into potential treatments or ideally even preventative strategies.