Does primary neoadjuvant systemic therapy eradicate minimal residual disease? Analysis of disseminated and circulating tumor cells before and after therapy

The study was conducted in the Department of Gynecology and Obstetrics at the University Hospital of Essen. In total, 190 patients with diagnosed primary, nonmetastatic BC between January 2007 and June 2012 and treated with NACT were enrolled. Patient characteristics are documented in Table 1.

We conducted a retrospective, single-institution trial to determine the prognostic value of DTCs in the BM and CTCs in blood of the patients and proved the effectiveness of treatment on these cells. The median follow-up time was 54 months (range 2–93 months) for OS and 52 months (range 2–93 months) for PFS, with an OS rate of 89 % and 12 % for relapses.

The eligibility criteria were histologically proven BC, BM and blood samples obtained at the time of primary diagnosis and after neoadjuvant systemic therapy, no severe uncontrolled comorbidities or medical conditions, and no further malignancies at present or in the patient history.

Indications for NACT were: study participation for patients if comparable postoperative chemotherapy was indicated, patients with inflammatory BC, large operable BC primarily requiring mastectomy and adjuvant chemotherapy with the goal of breast conservation, such as patients with nondifferentiated or poorly differentiated tumors (G3) [47].

Neoadjuvant systemic therapy was performed according to guideline-based therapeutic regimens, including chemotherapy with anthracyclines, cyclophosphamides, 5-fluorouracil, and taxanes. In addition, patients with HER2-positive tumors were treated with HER2-targeted therapy (trastuzumab or lapatinib). Seven patients were treated with vascular endothelial growth factor targeted therapy with bevacizumab. Patients were included in clinical NACT trials and treated accordingly (e.g., the LAPADO, NeoALLTO, and GeparQuinto studies [48‐50]). After completing NACT and surgery, patients were treated according to guidelines, including radiation, antihormonal therapy in those with hormone-responsive tumors (tamoxifen or an aromatase inhibitor), and trastuzumab therapy was completed for at least 1 year in patients with HER2 positivity [51, 52]. Additional oral clodronate therapy (2 × 520 mg per day for at least 2 years) was recommended in case of DTC positivity after therapy.

Pathological response to therapy was defined according to the grading system of Sinn and colleagues [53] as pathological no response (regression according to Sinn 0 = no effect), pathological partial response [pPR; regression according to Sinn 1–3, where 1 = resorption and tumor sclerosis, 2 = minimal residual invasive tumor (<0.5 cm), and 3 = residual noninvasive tumor only; ductal carcinoma in situ (DCIS)], and pCR (defined as no evidence of residual invasive cancer and DCIS, both in breast and axilla; regression according to Sinn 4 = no tumor detectable).

Between 10 and 20 ml of BM was aspirated from the anterior iliac crests of 142 patients with primary BC before neoadjuvant systemic therapy during sentinel node biopsy or axillary lymph node dissection, as well as 165 patients during surgery of the tumor after NACT. Specimens were processed within 24 h. All specimens were obtained after written informed consent was provided, and they were collected using protocols approved by the clinical ethics committee of University Hospital Essen (05/2856). BM tumor cell isolation and detection were performed on the basis of recommendations for standardized tumor cell detection published by the German Consensus Group of Senology [54]. Details of the staining procedure (e.g., number of evaluated slides, controls, and cell detection) are described elsewhere [37, 55]. Briefly, BM cells were isolated from heparinized BM (5000 U/ml BM) by Ficoll-Hypaque density gradient centrifugation (density 1.077 g/mol; Pharmacia & Upjohn Diagnostics, Freiburg, Germany) at 400 × g for 30 minutes. Slides were analyzed for DTCs by immunocytochemistry using the pan-cytokeratin antibody A45-B/B3. Microscopic evaluation of the slides was carried out using the ARIOL system (Applied Imaging, Grand Rapids, MI, USA) according to the International Society of Hematotherapy and Graft Engineering evaluation criteria [56].

Two 5 ml ethylenediaminetetraacetic acid blood samples were collected for isolation of CTCs before the application of therapeutic substances with an S-Monovette (Sarstedt AG & Co., Nümbrecht, Germany) and stored at 4 °C until further examination. The samples were processed immediately or at latest 4 h after blood withdrawal.

Two 5 ml of blood before (n = 135 patients) and after (n = 133 patients) therapy were analyzed for CTCs with AdnaTest BreastCancer (QIAGEN Hannover GmbH) for the detection of transcripts of epithelial cell adhesion molecule (EpCAM); mucin 1, cell surface associated (MUC1); HER2; and β-actin. Expression of ER, PR, and ERCC1 was assessed in an additional reverse transcription polymerase chain reaction (RT-PCR) experiment. Establishment and validation of this assay are described in detail elsewhere [57, 58]. Briefly, all samples underwent immunomagnetic enrichment using the AdnaTest BreastCancerSelect (QIAGEN, Hannover GmbH). followed by RNA isolation and subsequent gene expression analysis by multiplex RT-PCR in separated tumor cells using the AdnaTest BreastCancerDetect (QIAGEN Hannover, GmbH). The primers generated fragments of the following sizes: GA 733-2, 395 bp; MUC1, 293 bp; HER2, 270 bp; PR, 270 bp; ER, 305 bp; ERCC1, 366 bp; and actin, 114 bp. Visualization of the PCR fragments was carried out with a 2100 Bioanalyzer using DNA 1000 LabChips (Agilent Technologies, Waldbronn, Germany) and the Expert software package (version B.02.03.SI307; Agilent Technologies).

Both the AdnaTest TumorStemCell and AdnaTest EMT require the enrichment of CTCs from 5 ml of blood using AdnaTest BreastCancerSelect before a single-plex PCR assay to analyze ALDH1 and a multiplex PCR assay to analyze EMT markers and actin as an internal control. In total, the analysis of 148 healthy controls resulted in a specificity of 97 % and a sensitivity of 96 % for this test procedure, which is comparable to our previously published data in smaller cohorts [33, 34]. The primers generate fragments of the following sizes: ALDH1, 165 bp; AKT2, 306 bp; Twist-related protein 1 (TWIST1), 203 bp; phosphoinositide 3-kinase alpha (PI3Kα), 595 bp; and β-actin, 119 bp.

The test result is considered positive if a PCR fragment of at least one tumor-associated transcript (MUC1, GA773-2, or HER2) is clearly detected. Using the software package for evaluation of the data on the Agilent 2100 Bioanalyzer, we found that peaks with a concentration >0.15 ng/μl were positive for the transcripts of GA733-2, MUC1, and HER2. Peaks with concentrations >0.60 ng/μl were positive for the ER transcript and >0.20 ng/μl were positive for the ERCC1 transcript. PR expression is considered positive when the transcript is detected without applying any cutoff. The cutoff values for the EMT markers and ALDH1 are 0.2 ng/μl for AKT2, 0.15 ng/μl for TWIST1, 0.25 ng/μl for PI3Kα, and 0.15 ng/μl for ALDH1.

For each of the 190 patients, the tumor type, TNM stage, and grade were assessed according to the World Health Organization classification of breast tumors [59] and the Sixth Edition of the TNM classification system [60]. ER and PR status were routinely determined by immunohistochemistry (IHC) in the pathology departments of each university hospital. The HercepTest score (Dako, Glostrup, Denmark) for the expression of HER2 was determined, and fluorescence in situ hybridization analysis was performed in cases of 2+ staining, as described elsewhere [61].

The statistical analysis was performed using SAS® 9.2 software (SAS Institute, Cary, NC, USA). Summary statistics are presented as counts and percentages in the case of categorically scaled measures and as mean, median, standard deviation, and range in the case of continuously scaled variables. All cases with any available information were analyzed, and no imputation of missing information was foreseen.

The statistical analysis of relationships among the variables was performed in an exploratory way, starting with description by contingency tables (χ² test or Fisher’s exact test for categorically scaled variables) or comparison of distributions (Mann-Whitney U test for continuously scaled variables). In parallel, univariable logistic regression models regarding binary outcome (yes or no) by factor were analyzed. Univariable Cox proportional hazards models were applied to investigate the influence of possible influencing variables on OS and PFS. In case of significant findings, Kaplan-Meier analyses were performed to create survival curves. The resulting odds ratio (OR) and hazard ratio (HR) are reported along with their 95 % confidence interval (CI) and p values. An α level of 0.05 was used, whereby an adjustment for multiple testing was not foreseen. Because of the limited sample size of patients with available cell count information and missing significant effects for cell counts on outcome measures in univariable analyses, a multivariable analysis was not performed.

Clinical data are shown in detail in Table 1. The exact numbers of patients who had the different tests before and after NACT are shown in Additional file 1: Fig. S1. A total of 190 patients were included in the study. The median age of the patients was 51 years (range 18–84 years), and most women were either premenopausal (n = 87) or postmenopausal (n = 79). The predominant histologic subtype was invasive ductal carcinoma (n = 140), and most patients had grade II (n = 85) and grade III (n = 88) tumors. Tumor size before therapy was cT1a–cT1c in 46 patients (24 %), cT2 in 111 patients (58 %), and above cT2 in 29 patients (15 %). After therapy, tumor size was available in 177 patients, resulting in ypT0 tumors in 26 %, ypT1a–ypT1c tumors in 40 %, and ypT2 tumors in 27 % of cases. Twelve patients had tumors above ypT2. Before therapy, 94 patients were classified as node-negative. All other patients were cN1 (n = 83) and cN2 or cN3 (n = 10). After therapy, nodal status was available in 180 patients, resulting in 116 (64 %) node-negative and 64 (37 %) node-positive patients. ER positivity was observed in in 69 % (n = 131) and PR positivity in 61 % (n = 116) of the tumors. In 29 % (n = 56) of the cases, HER2 was overexpressed. Classifying tumors in subtypes on the basis of their receptor status, 51 % (n = 97) of the tumors were ER- and/or PR-positive and HER2-negative, 19 % (n = 36) were triple-negative (ER−/PR−/ HER2−), and 8 % (n = 15) of the tumors expressed only HER2 (ER−/PR−/HER2+). Response to therapy could be evaluated in 176 patients, resulting in ratios of 93 % responders (21 % complete response, 72 % partial response) and 7 % nonresponders.

DTCs were found in 38 (27 %) of 142 patients before and in 33 (20 %) of 165 patients after therapy (Table 1). In 118 patients, DTC status could be evaluated before and after NACT, resulting in positivity rates of 26 % before (31 of 118 patients) and 19 % (22 of 118 patients) after therapy. In 73 patients, no DTCs were detected at any time point, while 8 patients had persisting DTCs. Twenty-three of the thirty-one DTC-positive patients before NACT turned negative after chemotherapy, and fourteen patients with a negative DTC status before therapy had a positive DTC status after chemotherapy (Table 2).

The detailed analysis for the evaluation of CTCs is illustrated in Fig. 1. Before therapy, CTCs were detected in 32 (24 %) of 135 of the patients expressing EpCAM (19 %), MUC-1 (41 %), HER2 (75 %), ER (19 %), PR (6 %), and ERCC1 (63 %). After therapy, 11 (8 %) of 133 of the patients were still positive for CTCs expressing EpCAM, MUC1, and HER2 (each 45 %); ER (18 %); PR (0 %); and ERCC1 (72 %). slCTCs were detected in 46 (51 %) of 91 of the patients before and in 18 (20 %) of 90 of the patients after therapy. Further, 47 % of the patients were positive for at least one of the EMT markers before and 14 % after therapy. For ALDH1, the values were 17 % and 12 %, respectively. Interestingly, after therapy, 50 % of the ERCC1-positive CTCs were also positive for ALDH1 and for at least one EMT marker.

Before therapy, 59 (43 %) of 136 patients were positive for DTCs and/or CTCs, 74 (69 %) of 107 were positive for DTCs and/or slCTCs, and 57 (62 %) of 92 were positive for CTCs and/or slCTCs. After therapy, the corresponding values were 34 % (44 of 140 patients), 47 % (48 of 103 patients), and 28 % (25 of 89 patients), respectively. The presence of CTCs before (OR 4.200, 95 % CI 1.549–11.389, p = 0.005) and after therapy (OR 0.688, 95 % CI 0.177–2.676, p = 0.05) was significantly associated with the presence of slCTCs (Tables 3 and 4).

In 92 patients, CTC status could be evaluated before and after NACT. We found positivity rates of 27 % before (25/92 patients) and 11 % (10/92 patients) after therapy. Interestingly, CTCs were eradicated in 24 of 25 CTC-positive patients before therapy, while only 1 patient had persisting CTCs and 9 (13 %) of 67 patients had a switch from negative to positive CTC status (Table 2). In addition, the switch from CTC-positive before to CTC-negative after NACT appeared more often in post- and perimenopausal women than in premenopausal women (p = 0.0499).

slCTC status before and after NACT was available in 48 patients, with positivity rates of 60 % (29 of 48 patients) before and 19 % (9 of 48 patients) after therapy. Eight patients had persisting slCTCs, while eighteen patients were negative/negative, 21 patients were positive/negative, and only one patient was negative/positive (Table 2). Furthermore, the eradication of slCTCs was rarer the bigger the tumor size (p = 0.044).

The correlation between the presence of tumor cells and clinical characteristics before and after therapy is shown in Tables 2 and 3. Whereas no significant associations with clinical parameters were found for CTCs and slCTCs before and after therapy, DTCs were significantly associated with nodal status (OR 0.335, 95 % CI 0.143–0.785, p = 0.03 for N0 vs N1) and histology (OR 2.926, 95 % CI 1.001–8.556, p = 0.046 for ductal vs lobular; OR 0.125, 95 % CI 0.021–0.732, p = 0.023 for lobular vs. other) before therapy and the IHC subtype (OR 3.954, 95 % CI 1.564–9.997, p = 0.02 for triple-negative vs. ER and/or PR-positive but HER2-negative). Although no significant correlations were observed for response and slCTCs before therapy, logistic regression identified a significant relationship between slCTCs and the group of complete responders vs. no remission (OR 0.091, 95 % CI 0.009–0.880, p = 0.04).

As shown in Table 1, the median follow-up time for PFS was 52 months (range 2–93 months), and for OS it was 54 months (range 2–93 months). The OS (not BC-specific) rate was 89 %. Relapses occurred in 12 % of cases, 3 patients had local recurrence, 16 patients had a distant recurrence, and 3 patients had both local and distant recurrence. Univariable Cox regression analysis revealed age (p = 0.0065), tumor size before (p = 0.0473), nodal status (p = 0.0137) after NACT, and response to NACT (p = 0.0136) were also significantly correlated with PFS, whereas age (p = 0.0162) and nodal status after NACT (p = 0.0243) were significantly associated with OS (Table 5). No significant correlations were found for DTCs or any CTCs before and after therapy with regard to PFS and OS (Figs. 2 and 3). In addition, all combinations tested between the presence of DTCs and/or any CTCs, as well as their change before and after therapy, and PFS and OS did not show any significant results (data not shown).

DTCs have been analyzed mainly after NACT, with a detection rate of 40–50 % but demonstrating that 30 % of the detected cells were apoptotic [21, 64]. In only one study have researchers analyzed these cells before and after NACT, showing positivity rates of 21 % and 16 %, respectively, which is quite in accord with our data showing 27 % positivity before and 19 % after NACT, respectively [11]. Interestingly, 14 of 87 patients with a negative DTC status before therapy switched to a positive DTC status after chemotherapy. We can only speculate and hypothesize that these patients probably were DTC-positive before therapy and that DTCs probably had undergone EMT and lost their epithelial character, which allowed them to travel to metastatic sites without being affected by conventional treatment [65].

Furthermore, DTCs in our patients were significantly associated with nodal status before and after NACT as well as with the histology- and IHC-determined subtype before NACT, while no significant correlations could be documented for PFS and OS, which is in contrast to the results obtained in the above-mentioned studies. The fact that DTCs persist after treatment and are associated with worse outcome has already been described and is explained by a clinically significant biological heterogeneity, most likely due to phenotypical changes, EMT and stem cell characteristics, and altered genomic characteristics between DTCs, CTCs, and the primary tumor [9‐11, 32, 35, 66‐70]. One treatment option for DTCs has been demonstrated by Naume et al., who administered six cycles of docetaxel as secondary treatment in DTC-positive patients after first-line therapy with fluorouracil, epirubicin, and cyclophosphamide, resulting in better survival of these patients [71]. We did not further characterize DTCs; thus, we can only speculate that residual cells might have stem cell characteristics in some patients. However, we did not see any negative prognostic effect of DTCs before and/or after NACT with regard to PFS and OS after a median follow-up of nearly 5 years, which is in accord with our other published adjuvant BC studies, where we demonstrated that the intake of clodronate in case of DTC positivity was able to eradicate DTCs even years after the first diagnosis [26, 72]. Thus, clodronate intake was also offered to DTC-positive patients in this study. However, it is still unknown whether bisphosphonates are also able to eradicate stem cell–like DTCs present among DTCs in the BM. The fact that stem cell–like and EMT-like cells, probably circulating from reservoirs other than the BM (e.g. liver, lung), were present in blood samples before and after therapy makes the hypothesis that DTCs with stem cell–like characteristics are eradicated by bisphosphonates more or less unlikely.

CTCs have been suggested to be potential surrogate markers for minimal residual disease, the precursor of metastatic disease. There is increasing evidence that CTCs could be a strong predictive biomarker of response to NACT because BM is too invasive and painful for monitoring purposes. Unfortunately, a recently published meta-analysis confirmed that, although CTCs before NACT significantly correlated with reduced PFS in some studies, the change (decrease or increase) in CTC numbers during NACT in patients with locally advanced BC was not associated with pCR, and a decrease in CTC counts after NACT did not indicate that patients had an improved response [44].

The main problem with using CTCs as a so-called liquid biopsy is the fact that, at present, there is no standard definition for the identification of CTCs [73]. Currently, the CELLSEARCH system, based on immunomagnetic EpCAM capturing, is the only system for CTC enumeration in BC approved by the U.S. Food and Drug Administration [7]. However, despite the prognostic impact of CTC counts in BC, it has been shown that this procedure is not able to detect the entire, highly heterogeneous population of CTCs, including slCTCs and CTCs in EMT [74, 75]. Despite the prognostic impact of CTC counts, it is indispensable to characterize these cells, which might complement these studies by improving the overall detection rate as well as sensitivity and thus permit the assessment of genomic markers in CTCs of patients with BC, as recently published [45].

Although we and others, by using molecular methods, have already characterized the heterogeneous CTC populations in primary BC before adjuvant therapy [7, 33, 37, 38, 76, 77], in few studies have researchers evaluated the presence, not the characteristics, of CTCs after adjuvant therapy, resulting in a positivity rate of 22–34 % [17, 20, 78]. Furthermore, in the neoadjuvant setting, the number of circulating epithelial cells was mostly used for monitoring the effect of chemotherapy after every cycle of treatment [23, 24]. To the best of our knowledge, our present study is the first in which CTCs have been characterized so comprehensively before and after NACT. In contrast to DTCs, most of the CTCs before therapy, present in about 24 % of the patients and reflecting the heterogeneous CTC population, were eliminated by the given therapy. Although we cannot definitively prove that all residual CTCs were stem cell–like and EMT-like, molecular marker expression might allow characterization of them as stem cell–like. In addition, 72 % of the residual cells were characterized as ERCC1-positive, indicating therapy-resistant tumor cell populations. Interestingly, seven of these eight patients had been treated with taxanes or anthracyclines, which were shown to eradicate DTCs in a Norwegian study [13]. The fact that ERCC1-positive CTCs were present after these therapies might indicate that these cells survived treatment. This knowledge might help clinicians to decide more precisely about further secondary treatment options in the future. However, CTCs and slCTCs before as well as after therapy did not significantly correlate with decreased PFS or OS, as described in the above-mentioned studies. On one hand, follow-up time might have been too short to see an effect on outcome; on the other hand, factors determining if single tumor cells form metastasis have not been identified yet. In this regard, it has been demonstrated that CTCs can be detected after more than 20 years in patients with BC without any sign of relapse [79].

Tumor stem cells are well known to be resistant to various chemotherapeutic agents and radiotherapy [80]. In this context, Creighton et al. reported supportive evidence that the residual breast tumor tissue cell populations surviving after letrozole or docetaxel were enriched for subpopulations of cells with both tumor-initiating and mesenchymal features [41]. Thus, additional therapeutic strategies are urgently needed to prevent relapse of the disease. In this regard, signaling pathways that maintain cancer stem cells are attractive targets for these therapies. One example is everolimus (RAD001), an oral inhibitor of mammalian target of rapamycin acting downstream of the PI3K/AKT pathway, which was shown to have effective inhibitory effects on cancer stem cells in vitro and in vivo, and combination treatment with RAD001 and docetaxel or trastuzumab has been reported to be effective in refractory metastatic BC [81]. Li et al. demonstrated that remaining tumorigenic cells after chemotherapy had unique properties of enhanced self-renewal as demonstrated by formation of mammospheres and increased propensity for tumor formation. In addition, lapatinib did not lead to an increase in these tumorigenic cells; thus, in combination with conventional therapy, specific pathway inhibitors may provide a therapeutic strategy for eliminating these cells to decrease recurrence and improve long-term survival [82]. Concerning HER2, a recently published study suggested that the clinical efficacy of adjuvant trastuzumab may relate to the ability of this agent to target the cancer stem cell population in a process that does not require HER2 gene amplification [83]. These results are partly comparable with ours because we observed clearance of slCTCs after NACT in some patients who received trastuzumab, lapatinib, or bevacizumab, whereas patients who did not receive these combinations had slCTCs left after therapy (data not shown). These observations have to be interpreted with caution, and more patients have to be followed to prove whether this observation holds true. Further promising agents that are thought to attack BC stem cells are salinomycin, where treatment resulted in the loss of expression of BC stem cells [84], and a new synthetic curcumin analogue against ALDH1 and glycogen synthase kinase-3β [85]. In the future, targeting the tumor microenvironment, such as by interrupting the immune cells (e.g., myeloid-derived suppressor cells) and cytokines [e.g., interleukin (IL)-6, IL-8] as well as the immune checkpoints [programmed cell death protein 1/programmed cell death ligand 1 (PD1/PDL1)], may provide additional new tools for immunological targeting of cancer stem cells [86].