Detection of circulating tumor cells and circulating tumor DNA before and after mammographic breast compression in a cohort of breast cancer patients scheduled for neoadjuvant treatment

Circulating tumor markers such as circulating tumor cells (CTCs) and circulating tumor DNA (ctDNA) can be found in the blood of cancer patients. As a liquid biopsy, these markers complement solid biopsies and have the advantage of being physically more accessible and patient-friendly than traditional tissue biopsies. This provides a possibility for prognosis prediction, closer monitoring of treatment response and disease progression, identification of drug targets, as well as an opportunity for early detection of recurrence. The presence of CTCs in the blood of patients with primary breast cancer has been shown to be an independent predictor of decreased disease-free and overall survival [1, 2], but the treatment predictive value of the cells is still under debate [3, 4]. The CTC methodology in primary breast cancer is also limited by the low number of detected cells, which makes enumeration and evaluation statistically challenging [5]. ctDNA, the cell-free DNA that originates from cancer cells, is a promising biomarker whose prognostic and treatment predictive power is emerging [6, 7]. Recent studies have shown that quantification of specific mutations in ctDNA can be associated with early detection of metastases and therapy resistance in breast cancer as well as in other diagnoses [8‐12].

The risk that tumor cells are released into the bloodstream from a primary tumor during surgical interventions has been addressed in a few studies [13‐17], although how and when tumor cells are shed as well as the clinical importance of this release is poorly understood [18]. Animal studies have also found that physical manipulation of a primary tumor by applying pressure to it causes tumor cell dissemination [19‐21]. We have previously investigated if mammographic breast compression in patients with an already present breast tumor could cause shedding of tumor cells to the peripheral circulation [22]. We found no indications that this would be the case in a pilot study of 24 patients with primary breast cancer.

However, the configuration of the human blood circulation can cause tumor cells released from the breast to pass through the capillary vasculature of the lungs before reaching the peripheral blood vessels. In our previous study [22], CTCs captured only in the peripheral blood might have resulted in an underestimation of CTC number. It has been shown that a higher number of CTCs can be found in central compared to peripheral venous blood in patients with metastatic breast cancer [23] as well as in other diagnoses [24‐26]. Animal studies of colon carcinoma cells have shown that the majority (80–100%) of tumor cells could be trapped in the capillary bed of the first organ they encounter [27]. In breast cancer, an autopsy study by Peeters et al. [28] showed that CTCs were trapped in the lung microvasculature in four of the nine patients who all had high CTC counts (> 100). Thus, it is likely that the number of CTCs found in the peripheral blood system is not representative of a possible release of tumor cells from the primary tumor during manipulation such as breast compression during mammography or surgery. The difference in CTC number between central and peripheral blood is possibly even more pronounced after specific interventions compared to a more steady-state-like condition of metastatic disease [24, 26].

To our knowledge, ctDNA levels have not been used to study a possible release of tumor cells or tumor cell debris after breast compression or any other mechanical intervention in breast cancer. Relatively few studies have so far compared the levels of both CTCs and ctDNA in the same clinical patient cohort at identical time points and our understanding of the relationship between the two liquid tumor markers is limited. However, both the level of ctDNA and the number of CTCs have been shown to have a prognostic value in mainly metastatic breast cancer cohorts [29, 30]. Mutation analysis of ctDNA and single CTCs suggests that ctDNA reflects the heterogeneity of mutations found in individual CTCs [30], but ctDNA levels have been found to have a higher correlation with tumor burden than CTCs [29].

The aim of this study was to investigate how the presence of CTCs and ctDNA are affected by breast compression during mammography in patients with primary breast cancer. Special emphasis was made on comparing circulating tumor marker burden between the central and peripheral blood circulation.

In total, 8/31 patients (26%) had ≥ 1 CTC in at least one of the blood samples taken before or after mammographic breast compression (Fig. 1). Correspondingly, 13/20 patients (65%) had ≥ 0.01% MAF and were defined as ctDNA positive. No agreement was found between CTC- and ctDNA-positive patients (κ = 0.02, p = 0.92) (A plot of CTC count versus % MAF can be found in supplementary Fig. S1).

Patient and tumor characteristics of the whole cohort, as well as of CTC/ctDNA-positive and CTC/ctDNA-negative patients separately, are shown in Table 1. No patient or pathologic characteristics were statistically associated with CTC positivity. Larger tumor size, non-ductal histological subtype, and older age were more predominant in the CTC-positive group but the difference was not statistically significant. ctDNA positivity was associated with higher age (p = 0.05). Higher Ki67, ductal histological type, and triple-negative breast cancer were more predominant in ctDNA-positive patients, without reaching statistical significance (Table 1). Notably, 4/4 triple-negative breast cancers (TNBC) and 4/4 T4 staged cancers were all ctDNA positive.

Thirty patients had CTC results from the central blood sample before and after breast compression and 22 of these patients had 0 CTCs at both time points. Five of eight patients with detectable CTCs had an increased number of CTCs after compression (p = 0.19) (Fig. 2a). The average CTC increase was 3.2 cells (median 1.0 cell). Only two evaluable patients had detectable CTCs in the peripheral blood sample (Fig. 2b). Both central and peripheral % MAF generally increased after compression with the latter reaching significance (p = 0.08 and p = 0.01) (Fig. 2c, d). The average increase of % MAF was relatively small, 0.77 and 0.35 (median 0.35 and 0.22% MAF) for central and peripheral, respectively. Of the 20 patients with assessable ctDNA samples before and after breast compression, eleven and eight patients had 0% MAF in central and peripheral plasma samples, respectively, at both time points.

The median fraction of Ki67-positive cells was 66% (range 30–90%) in the five patients that had an increase in CTC number after compression, compared to 45% (range 15–95%) in patients with no increase (p = 0.31) (Supplementary Table 1). Also, Ki67 fraction was significantly higher in the group with increasing ctDNA after compression, 45% versus 30% (p = 0.04) (Supplementary Table 2). No other factors were differentially expressed between patients with an increase in CTCs/ctDNA levels and patients with a stable or a decrease in CTCs/ctDNA levels after compression. However, the histological type of the primary tumor seemed to differ between patients with an increase in the number of CTCs and patients with an increase in the levels of ctDNA after compression (Supplementary Tables 1 and 2).

CTCs were more abundant in central compared to peripheral blood in 8/10 positive samples (p = 0.04) (Fig. 3a). Forty-nine comparisons between central and peripheral blood contained 0 CTCs in both samples. There was no significant difference in % MAF levels between central and peripheral sampling (8/20 favoring higher % MAF in the central blood sample, p = 0.50) (Fig. 3b). Twenty comparisons between central and peripheral blood contained 0% MAF in both samples.

CTCs were detected in 26% of the patients before start of neoadjuvant therapy for primary breast cancer. This is in line with a recent meta-analysis where 25.2% of breast cancer patients had CTCs before onset of neoadjuvant chemotherapy (deemed independent of blood sampling volume) [2]. ctDNA was positive in 65% of the patients. The concordance between CTCs and ctDNA has been shown to be higher in metastatic breast cancer patients [29] as compared to what was found in this study, which is most likely due to the lower rate of CTCs in primary breast cancer (Supplementary Fig. S1). ctDNA positivity, defined as ≥ 0.01% MAF in this study, was associated with higher age (p = 0.05) and a trend was noted that a more aggressive tumor phenotype, including high Ki67, TNBC, and T4 staged cancers, favors ctDNA positivity (not statistically significant).

For CTCs, no increase was seen in either central or peripheral blood after mammographic breast compression (p = 0.19 and p = 0.37, respectively). However, both central and peripheral ctDNA levels increased after breast compression (p = 0.08 and p = 0.01, respectively). Only one patient had a CTC count difference of > 5 cells/7.5 ml between samples taken before and after compression (Fig. 2a). This suggests a lack of a larger bolus release of cells during breast compression of women with primary breast cancer. The central blood samples were drawn on average 2 min after compression and, according to an animal study, the release of malignant cells starts at the manipulation procedure and stays elevated up to 60 min [19]. To the best of the authors’ knowledge, no published study has investigated how ctDNA levels vary with manipulation of a primary tumor with regard to applied pressure. The increase of ctDNA after breast compression found in this study can be considered relatively small, with only one case going from % MAF 0.95 to 2.36 which possibly could affect prognostication (Fig. 2d) [8]. High Ki67 were associated with increased ctDNA levels (p = 0.04) (Supplementary Table 2).

As hypothesized, CTCs were in general significantly more likely to be present in central than in peripheral blood samples (p = 0.04). Six patients presented CTCs only in the central samples (Fig. 3a) suggesting a differential CTC yield depending on sample location. The results are comparable to the work by Peeters et al. [23] in metastatic breast cancer but no data from studies involving differential blood sampling of primary breast cancer are hitherto available. This differential yield was not seen for ctDNA (p = 0.50). Since ctDNA is a much smaller moiety and soluble in the blood, we speculate that ctDNA is much less affected by physical hindrance in the capillaries as compared to CTCs. Hence, ctDNA blood sampling is independent of blood drawing location, a finding that could contribute to the definition of clinical sampling routines in primary breast cancer for ctDNA.

The major limitation of this study was the small sample size and low count of CTCs, despite that a total of up to 40 ml whole blood was drawn from each patient. The statistical nature of CTC sampling has been described by Tibbe et al. [5]. Due to the low sample size, a possible difference between CTCs detection before and after breast compression may have been underestimated. Similarly, the ctDNA analysis was limited by a relatively low plasma input volume, and therefore a limited number of genome equivalents being analyzed for the presence of mutations.

When CTCs and ctDNA markers are implemented into clinical routine, our understanding of how the concentrations fluctuate during different interventions should be better understood. The women in this cohort are continuously being monitored and follow-up data will be available and presented in future publications.

In summary, there was no significant release of CTCs after mammographic breast compression but more CTCs were present in central compared to peripheral blood. There was a small average increase in ctDNA levels after breast compression, unlikely to be clinically relevant, and no difference between central and peripheral levels was found.