The presence and prognostic impact of apoptotic and nonapoptotic disseminated tumor cells in the bone marrow of primary breast cancer patients after neoadjuvant chemotherapy

Introduction Neoadjuvant systemic therapy of primary breast cancer (PBC) patients offers the possibility to monitor treatment response. However, patients might have metastatic relapse despite achieving a pathologic complete response (pCR). This indicates that local response to therapy must not be representative for systemic treatment efficacy. Therefore, the aim of this study was to compare local response with systemic tumor cell dissemination by determining the presence of disseminated tumor cells (DTCs), including apoptotic tumor cells, in the bone marrow (BM) of PBC patients after neoadjuvant chemotherapy (NACT). Methods DTCs were detected by immunocytochemistry (pancytokeratin antibody A45-B/B3) and cytomorphology (DTC status). The presence of apoptotic tumor cells was determined by using the M30 antibody (M30 status). This antibody detects a neo-epitope that is expressed only during early apoptosis. Results BM aspirates from 400 PBC patients that had completed NACT were eligible for this study. Of these, 167 (42%) patients were DTC positive (DTC status). The M30 status was investigated in 308 patients. Apoptotic (M30-positive) tumor cells were detected in 89 (29%) of these. Whereas the DTC status was not correlated (P = 0.557) to local treatment response (that is, pCR or a clinical complete/partial response), the presence of M30-positive tumor cells was significantly higher in patients that responded to therapy (P = 0.026). Additionally, DTC-positive patients were at an increased risk for disease relapse (hazard ratio, 1.87; 95% CI, 1.11 to 3.15; P = 0.019). Conclusion The presence of DTC is independent of therapy response of the primary tumor. As patients that are DTC positive after NACT have an unfavorable outcome, they might benefit from additional systemic treatment.


Introduction
Neoadjuvant chemotherapy (NACT) aims to reduce tumor mass and has become a standard treatment in primary breast cancer (PBC) patients. Initially, it was used to treat locally advanced and nonoperable tumors. Currently, NACT is also offered to patients with resectable tumors, as no differences between NACT and adjuvant treatment in terms of overall survival and the probability of disease relapse are evident [1,2]. In these patients, NACT aims to facilitate breast-conserving surgery. Moreover, NACT offers the possibility of monitoring the primary tumor's response to treatment. This in vivo chemosensitivity testing helps clinicians to choose for appropriate therapy options in case of recurrent disease [3].
Breast cancer might relapse even years after successful treatment of the primary tumor. Accordingly, the disease must have the ability to persist in secondary sites of the body, a phenomenon that is called minimal residual disease (MRD). To prevent metastatic regrowth effectively, MRD has to be eradicated before it becomes clinically evident [4,5]. Monitoring of the treatment success against MRD would therefore improve therapy and the prognosis of PBC patients. During NACT, the success of systemic treatment is currently determined locally (that is, by reassessing the primary tumor size). It is, however, not clear whether therapy response of the primary tumor is representative for the entire tumor burden (that is, for systemic treatment efficacy). Although pathologic complete response (pCR) of the primary tumor is associated with a favorable prognosis [6][7][8][9], subgroups of patients seem to not have a prognostic benefit from achieving a pCR. Minckwitz et al. recently evaluated the clinical value of pCR as related to intrinsic breast cancer subtypes and found that the primary tumor's response to NACT has no impact on prognosis in luminal A as well as in luminal B/HER2-positive patients [10]. Other reports indicate that 15% to 25% of all patients receiving NACT will develop metastatic disease despite a pCR [11][12][13][14][15]. Therefore, the persistence of MRD after NACT may be independent of the success of primary tumor treatment.
The aim of this study was to compare local with systemic response to NACT in a large cohort of PBC patients. For this purpose, we evaluated the prevalence of DTCs and apoptotic tumor cells at the time of surgery and compared our findings with response of the primary tumor. Moreover, we determined the prognostic relevance of DTC persistence after NACT.

Study population
Patients with PBC (cT1-cT4, cN0-cN2) who were treated with NACT and underwent surgery at the Department of Gynecology and Obstetrics at Tuebingen University Hospital, Germany, between January 2001 and January 2012 were eligible for this retrospective analysis. Exclusion criteria were metastatic or recurrent disease, bilateral breast cancer, and a previous history of secondary malignancy. All patients provided written informed consent for BM aspiration, and the analysis was approved by the ethics committee of the University of Tuebingen (560/2012R). Treatment regimens are shown in detail in Table 1.

Determination of response to treatment
Pathologic complete response (pCR) was defined as the absence of residual invasive cancer on pathologic evaluation of resected breast specimens and lymph nodes [36]. For patients that did not achieve a pCR, response to treatment was determined by physical examination and imaging tests. The preferred imaging modality was ultrasound. However, if ultrasound appeared not to provide valid results, other imaging tests were used, with the following priority: MRI, mammography. The effect of NACT was graded according to the World Health Organization criteria [37]. Partial remission was defined as a reduction of the primary tumor area by 50% or more (multiplication of longest diameter by the greatest perpendicular diameter; in patients with multifocal or multicentric disease, the lesion with the largest diameters was used), and also includes patients with clinical complete remission (disappearance of all known disease) that did not achieve a pCR. Progressive disease was defined as the development of new, previously undetected lesions or an increase in the diameter product of a preexisting lesion by more than 25% after at least two treatment cycles. Stable disease was defined as anything between partial remission and progressive disease. To determine response to NACT, we defined patients with pCR or partial remission as responders, whereas those with stable or progressive disease were defined as nonresponders.

Bone marrow status and immunohistochemistry
Approximately 3 to 4 weeks after NACT, 10 to 20 ml of BM aspirates was collected during surgery. All samples were processed within 24 hours, as described elsewhere [38]. In brief, mononuclear cells were separated with density centrifugation (Ficoll, 1.077 g/ml; Biochrom, Germany), spun down onto a glass slide (Hettich cytocentrifuge; Germany) and fixed in 4% formalin. The presence of DTC (DTC status) was evaluated by immunostaining with the DAKO Autostainer (Dako, Denmark), the monoclonal mouse A45-B/B3 antibody directed against pancytokeratin (Micromet, Germany), and the DAKO-APAA detection kit (Dako). For each patient, two slides (2 × 10 6 cells) were automatically scanned by using the ACIS imaging system (ChromaVision; Medical Systems Inc., San Juan Capistrano, CA, USA) and evaluated based on consensus recommendations for standardized tumor-cell detection and the criteria of the European ISHAGE Working group [39,40]. An additional slide was stained by using an unspecific isotype-matched antibody. Moreover, with each batch of samples, leukocytes from healthy volunteers served as negative control, and the cell lines MCF-7 and SKBR-3 were used as positive control.
To evaluate the specificity of our method for DTCdetection, we analyzed BM samples of 100 individuals without evidence of malignant disease for the presence DTCs, earlier [28]. Of these patients, one harbored DTCs.
To detect apoptotic tumor cells (M30 status), additional slides were stained with the M30-antibody (Roche Applied Science, Mannheim, Germany) and analyzed by using the previously described detection method. The M30-antibody is directed against an epitope expressed only after caspase cleavage of CK18 during early apoptosis [41,42]. Identification of apoptotic tumor cells was based on positive M30-staining and cytomorphologic criteria, as described elsewhere [34,43,44].

Statistical analysis
Associations between categoric variables (DTC status/ M30 status and patient characteristics) were analyzed by using the χ 2 test. To determine survival, times from BM aspiration to any recurrence of disease (disease-free survival, DFS) and to death of any cause (overall survival, OS), respectively, were investigated separately. If no event occurred, data were censored at last follow-up. The influence of the DTC status/M30 status on survival was determined in a univariate analysis and described by hazard ratio (HR) and the corresponding 95% confidence interval (CI). Kaplan-Meier curves were plotted and compared with the log-rank test. For multivariate analysis, a Cox proportional regression model was used. Variables were entered stepwise backward, and a significance level of 0.1 was used to exclude a variable from the model. The initial model included menopausal status, histology, grading, nodal status before/after NACT, tumor size before/after NACT, ER/PR/ HER2 status, and the DTC status. The effect of each variable was evaluated by using the Wald test and described by HR and the corresponding 95% CI. All statistical tests were carried out with PASW Statistics 20 (SPSS Inc., Chicago, IL, USA) and reported two-sided with significance levels set to P < 0.05.

Prevalence of DTCs and apoptotic tumor cells
After NACT, DTCs were detected in the BM of 167 (42%) patients (Table 2). A positive DTC status was observed more frequently in PR-negative patients (49%) than in PR-positive patients (38%) (P = 0.036). Except for the PR status, no significant association was noted between the DTC status and any established prognostic factor. Importantly, the presence of DTC was not reflected by the primary tumor response to NACT (P = 0.912).
BM samples of 308 patients were additionally screened for M30-positive apoptotic tumor cells. M30-positive cells were detected in 89 (29%) of these patients ( Table 2). A significant association was found between the M30 status and the ER (P = 0.041) and PR status (P = 0.001), respectively. Also, patients that were nodal positive before NACT were more likely to be M30 positive (P = 0.020). Moreover, the prevalence of apoptotic tumor cells was significantly higher in patients that had received NACT including trastuzumab (P = 0.002). Importantly, there was a significant association between the presence of apoptotic tumor cells and response to NACT (P = 0.026).
As demonstrated in Table 3, 12 of 308 patients (4%) harbored apoptotic tumor cells only (M30-positive/ DTC-negative), whereas in 77 of 308 patients (25%) both, apoptotic tumor cells and DTC, were detected (M30  In multivariate analysis (Table 4), independent factors of DFS were DTC status, nodal status after NACT, and response to NACT. Independent factors of OS were tumor grade and response to NACT.

Discussion
In the past decade, NACT has become an increasingly important strategy to treat PBC patients. Many clinical trials on systemic breast cancer treatment are currently conducted in the neoadjuvant situation, by using response to therapy as an early and easy-to-perform study end point and a surrogate for prognosis. Although response to NACT is associated with disease-free and overall survival some subgroups of patients do not benefit from achieving pCR and approximately 15% to 25% of the patients may develop metastatic disease despite pCR [10][11][12][13][14][15]. This indicates that the disease must have the ability to persist in secondary sites of the body and that local response to therapy is not necessarily representative of systemic response. Because micrometastatic spread of DTCs into the BM is a promising marker of disease persistence, this large retrospective analysis aimed to compare local response to NACT, as reflected by the primary tumor, with systemic response, as reflected by the presence of DTC and apoptotic tumor cells. We moreover determined the impact of the DTC status after NACT on prognosis.
DTCs were detected in 42% of the patients. An explanation for this comparatively high prevalence is that most women had initially advanced stages with nodal involvement. In general, these patients tend to have higher positivity rates in BM aspirates [17,18,[20][21][22]34,45]. Importantly, the presence of DTCs after NACT was not reflected by response of the primary tumor, indicating that local response to treatment is independent of systemic treatment efficacy. This observation is in line with previous reports and most likely due to altered genomic characteristics between CTC/DTC and primary tumor tissue of the same patient [32,33,35,46]. For example, phenotypic changes like HER2-amplification or epithelialto-mesenchymal transition (EMT) can occur during tumor cell dissemination [47,48]. Moreover, a large proportion of DTCs in breast cancer patients display stem cell-like features, such as ALDH1 positivity or presence of CD44 and absence of CD24 [49]. The cancer stem-cell hypothesis postulates that these cells not only might initiate tumorigenesis and metastatic growth, but also contribute importantly to therapy resistance [50].  Apoptosis is the principal mechanism of chemotherapyinduced tumor regression, and several studies have reported on the detection of single apoptotic cells in either peripheral blood or BM of breast cancer patients [34,[51][52][53]. To detect apoptotic cells, we and others used immunocytochemical staining with the M30-antibody [34,51]. The neo-epitope M30 is expressed by caspase cleavage of CK18 only during early apoptosis. Interestingly, serum levels of the CK fragment were shown to correlate with tumor load and prognosis during breast cancer chemotherapy [54]. In line with our previously presented results, the presence of apoptotic tumor cells was associated with local therapy response [34].
To differentiate between DTCs and false-positive cells, morphologic evaluations were performed as recommended by the ISHAGE Working Group. Therefore, cells with typical morphologic signs of apoptosis (cellular shrinkage, membrane blebbing, nuclear condensation, and fragmentation) were not considered as DTCs. As the majority of DTC-positive patients also harbored M30-positive cells, we believe that tumor cells in the BM represent a heterogeneous population of apoptotic and nonapoptotic cells. However, we cannot exclude that some DTCs might be M30 positive (that is, apoptotic), as the M30 status was evaluated on additional slides, and a double-staining procedure was not performed for technical reasons. Moreover, A45-B/B3 detects CK8, CK18, and CK19, whereas M30 detects a neoepitope of CK18, only. Thus, differences in the results between these two antibodies might be biased by different CK-expression patterns.
Principally, two different mechanisms would explain the source of apoptotic tumor cells in BM aspirates. These could be the result of a passive cell shedding from the primary tumor during NACT [35] and therefore reflect local response to NACT. The association between local treatment response and the M30 status we observed supports this hypothesis. Also, advanced tumor  stage and proliferation are associated with an increased tumor-cell turnover, and a higher breast cancer grade was recently shown to result in increased shedding of apoptotic tumor cells into the circulation [51,55]. In accordance, we found that patients at advanced stages (nodal positive) before the beginning of NACT as well as hormone-receptor-negative patients were more likely to present with apoptotic tumor cells in BM.
Conversely, DTCs that are the result of an active premetastatic process might undergo apoptosis, caused by systemic efficacy of NACT. However, chemotherapy is often unable to eradicate nonproliferating DTCs [27]. Indeed, we could not find an association between the presence of apoptotic cells and survival. A subgroup analysis, regarding the impact of the M30 status on survival in DTC-positive patients, confirmed this observation. Some authors therefore suggested the use of bisphosphonates as an alternative for cell-cycle-independent treatment of DTC-positive patients [56][57][58]. Others indicate that targeted therapy might be more appropriate than conventional chemotherapeutic regimens to eradicate MRD successfully [27,59,60]. Interestingly, we found that the prevalence of apoptotic tumor cells in patients that received trastuzumab was increased. This indicates a specific action of HER2-directed therapy against a significant subpopulation of DTCs. We recently found, in a large cohort of PBC patients, that about half of the DTC-positive patients harbor HER2-positive DTCs, independent of the HER2 status of the primary tumor [48]. The clinical value of HER2-directed therapy to treat MRD will be evaluated within the forthcoming "TREAT CTC" study (NCT01548677). In this trial, HER2-negative PBC patients with one or more CTCs after (neo)adjuvant chemotherapy will be randomized to receive trastuzumab treatment or not.
In PBC patients that have not been treated systemically before BM aspiration, the detection of DTCs has clearly proven to be of prognostic relevance, whereas the role of DTC detection after NACT is less well described [17][18][19][20][21][22][23][24][25]32,33]. Hall et al. [33] recently conducted a study on 95 PBC patients that were treated with neoadjuvant chemotherapy and found that DTC persistence is associated with breast-cancer-specific survival. Mathiesen et al. [32] evaluated the presence of DTC in the BM before, directly after (at surgery), and 12 months after NACT [32]. Probably due to the considerably low number of patients (n = 69), the authors found no impact of the DTC status directly after NACT on survival. In contrast to the results presented by Mathiesen et al., our study demonstrates an independent prognostic value of DTC detection after NACT. Of note, it would be interesting to evaluate the prognostic relevance of DTC persistence in patients with luminal B/HER2-positive or luminal A tumors separately, as pCR is not predictive of survival in these subgroups [10]. However, the number of patients was too low to permit such an analysis.
Next to the herein investigated role of the DTC status after NACT, promising data exist on DTC persistence in PBC during or after adjuvant treatment [28,29,31,32]. As these studies could also demonstrate an independent prognostic impact in patients that have received systemic therapy, the DTC status might be a marker of treatment failure and is likely to indicate patients that are in need of additional treatment. In the Norwegian SATT study (NBCG9), 72 PBC patients that were DTC positive after anthracycline-containing adjuvant chemotherapy received secondary docetaxel treatment, and a majority (79%) experienced disappearance of DTCs [61]. However, the clinical relevance of these results is not clear, as follow-up is still ongoing.
We could not find a prognostic impact of DTC determination after NACT on OS. Interestingly, a subgroup analysis revealed that in pretherapeutic clinical tumor stage III patients, the DTC status was prognostic for DFS and also OS, whereas in stage I to II patients, no impact of the DTC status on survival was found (data not shown). This observation is in accordance with our recent analysis on 1,345 clinically nodal-negative PBC patients [62]. In that study, the DTC status had also no impact on survival.
As BM sampling is a comparatively invasive procedure, recent reports have focused on the detection of CTC in the peripheral blood of PBC patients [63][64][65]. Whereas CTC detection in metastatic breast cancer patients has proven to be of prognostic relevance [66][67][68], their role in PBC is less well described, and results on an association between CTC and response to NACT or prognosis are inconclusive [60,[69][70][71]. This is probably owing to methodologic differences and a lower sensitivity of CTC analysis [25]. Recently, Molloy et al. [72] found, in 733 PBC patients, that CTC detection by use of a PCR-based assay was highly associated with the presence of DTCs, and that both DTC and CTC detection were predictive of survival. Further trials should therefore evaluate the clinical value of DTC/CTC enumeration in the neoadjuvant situation, especially as CTC detection offers the possibility of serial blood sampling during the course of therapy.

Conclusions
The detection of DTCs in the bone marrow of PBC patients is considered as a marker of systemic disease. In line with recent smaller studies, we found that the presence of DTCs after NACT was independent of the primary tumor response to treatment. As the DTC status was moreover associated with an unfavorable outcome, even patients with pCR but DTC persistence after NACT might benefit from additional adjuvant therapy. Moreover,