Circulating tumor cells in metastatic breast cancer

From prognostic stratification to modification of the staging system?


  • Presented in part at the San Antonio Breast Cancer Symposium, San Antonio, Texas, December 12-16, 2007.



The aim of the current study was to assess the prognostic value of baseline circulating tumor cells (CTCs) in a large cohort of patients with newly diagnosed metastatic breast cancer (MBC).


This retrospective study included 185 patients with newly diagnosed MBC evaluated between 2001 and 2007. CTCs were isolated and enumerated before patients started first-line treatment using the CellSearch system. Overall survival (OS) was calculated from the date of CTC measurement, estimated by the Kaplan-Meier product limit method, and compared between groups with the log-rank test. Cox proportional hazards models were fitted to determine the association between CTC levels and OS after controlling for other prognostic factors.


The median age of the patients at the time of MBC diagnosis was 49 years. Fifty-six (30.3%) patients presented with de novo metastatic disease, and 129 (69.7%) presented with newly recurrent breast cancer. A total of 114 patients (61.6%) had CTC <5, and 71 (38.4%) had CTC ≥5. The median OS was 28.3 months and 15 months (P < .0001) for patients with CTC <5 and CTC ≥5, respectively. Superior survival among patients with CTC <5 was observed regardless of hormone receptor and HER-2/neu status, site of first metastases, or whether the patient had recurrent or de novo metastatic disease. In the multivariate model, patients with CTC ≥5 had a hazards ratio of death of 3.64 (95% confidence interval, 2.11-6.30) compared with patients with CTC <5.


The results of this large retrospective study confirms that CTCs are a strong independent predictor of survival among women with either de novo or newly recurrent MBC. CTCs should be considered as a new stratification method for women with newly diagnosed MBC. Cancer 2008. © 2008 American Cancer Society.

The last decade has seen significant advances in the prevention, detection, and management of women with breast cancer that have resulted in a decrease in breast cancer mortality by approximately 2.3% annually.1 However, breast cancer remains 1 of the most common malignancies diagnosed among women, with the most recent cancer statistics estimating that 180,510 new cases of breast cancer would be diagnosed in 2007, with approximately 40,910 women expected to die of metastatic disease in the US.1 It is estimated of women diagnosed with early-stage breast cancer that approximately 24% to 30% of those with lymph-node-negative disease and 50% to 60% of those with lymph-node-positive disease will develop a recurrence.2 In addition, approximately 10% of women present with stage IV breast cancer at initial presentation.2, 3

Women with metastatic breast cancer (MBC) represent a heterogeneous group whose prognoses vary depending on factors such as age, tumor hormone receptor status and grade, anatomic site, and extent of disease, as well as the status of any existing comorbid conditions and associated performance status.4–6 Despite major advances seen in the management of MBC, the goal of treatment is palliation with a median survival of approximately 24 months.7 Only recent reports, based on single-institution experience, suggest a marginal improvement in outcome for this population, possibly related to improved diagnostics and therapeutics.8

Although attempts have been made to fit MBC into various prognostic strata using traditional prognostic indices,4 because of the known heterogeneity of breast tumors this method has had limited applicability in predicting survival accurately. In contrast, detection of circulating tumor cells (CTCs) in the blood of patients with metastatic breast cancer has repeatedly been shown to accurately predict survival, as well as to be an indicator of therapeutic efficacy in this cohort of patients.9–15 In a prospective trial performed in patients with measurable metastatic breast cancer, we found that CTCs were a strong prognostic factor in women about to start a new line of therapy.10 Furthermore, the prognostic value of CTCs was unrelated to hormonal receptor status, sites of recurrence, and previous treatments, and was demonstrated superior to tumor burden as measured by serum markers (CA 27-29) and Swenertone score, suggesting a further peculiar biological value.11 We were particularly intrigued by these results, which suggest the need for further independent validation studies of the prognostic value of CTCs in patients with newly diagnosed recurrent or de novo metastatic disease. The confirmation of the data in this population would support the value of a new staging stratification for MBC based on CTC measurement. Therefore, we designed a single-institution retrospective study that exclusively looked at patients who had their CTC levels measured at the time of staging of their newly diagnosed recurrent or de novo MBC.


Patient Population

We identified a total of 312 patients in the laboratory database of the University of Texas M. D. Anderson Cancer Center who had been diagnosed with CTCs between the period 2001 and 2007.15 Two hundred fifty-eight (82%) patients were evaluated beginning in 2004 after US Food and Drug Administration clearance of the CellSearch System (Veridex LLC, Raritan, NJ) for clinical use. This retrospective study was restricted to a cohort of 185 patients with newly diagnosed MBC. All patients had CTCs measured as part of their laboratory evaluation at the time of staging of their recurrent or newly diagnosed metastatic disease and before starting a new treatment. None of these patients was included in the previously reported pivotal trial or our previously reported retrospective study.10, 11 Patient eligibility criteria for this retrospective study included a newly diagnosed metastatic disease and no previous exposure to any therapy for recurrent disease. Moreover, patients were required to have clinical and radiologic evidence of metastatic breast cancer with either measurable or evaluable disease before initiating their first-line systemic therapy. All patients had imaging studies, laboratory evaluations, and treatment planning at our institution. The institutional review board at M. D. Anderson approved the study and granted a waiver of the informed consent, considering the retrospective nature of the study.

Isolation and Enumeration of CTCs

Operational details, precision, reproducibility, and accuracy of CTC measurement using the CellSearch System have been previously described.9 In summary, blood samples were drawn into 10-mL ethylenediamine tetraacetic acid Vacutainer tubes (Becton Dickinson, Franklin Lakes, NJ), supplemented with a cellular preservative, maintained at room temperature, and subsequently processed within 72 hours of collection using the CellSearch System. The system is composed of 1) a semiautomated sample preparation system; and 2) the CellSearch Epithelial Cell kit, which immunomagnetically enriches cells expressing the epithelial cell adhesion molecule (EPCAM) by incubating 7.5 mL of blood with a ferrofluid coated with anti-EPCAM antibodies.14 The cell suspension is then placed in a magnetic field that depletes it of hematopoietic cells that had ingested iron particles. The isolated cells are then incubated with the nucleic acid dye 4′,6-diamidino-2-phenylindole and labeled monoclonal antibodies that target leukocytes (CD45-allophycocyanin) and epithelial cell antigens (cytokeratin 8, 18,19-phycoerythrin). The resultant fluorescently labeled cells are then used to identify and enumerate CTCs using the CellSpotter Analyzer (Immunicon Corporation, Huntingdon Valley, Pa). The CellSpotter Analyzer is a semiautomated fluorescence microscopy system that allows for computer-generated reconstruction of cellular images.14 Samples were maintained at room temperature and processed within 72 hours after collection. All CTC evaluations were performed in 1 central laboratory (M. D. Anderson Cancer Center, Houston, Texas).

CTC Definition

The CellSpotter Analzer generates images of potential candidate CTCs in a sample. CTCs were defined as cells containing a nucleus (identified by positive 4′,6-diamidino-2-phenylindole staining) expressing cytokeratin and lacking CD45 (determined by lack of staining with CD45-allophycocyanin).9, 14 Results were expressed as number of cells per 7.5 mL of whole blood. Samples with <5 CTCs per 7.5 mL of whole blood and samples with ≥5 CTCs per 7.5 mL of whole blood were considered to be negative and positive for CTCs, respectively.16

Staging and Pathology Review

Clinical stage at diagnosis of primary disease was coded according to the criteria set forth by the American Joint Committee on Cancer (AJCC) (sixth edition, 2003).17 Histologic type and grade of invasive disease were coded according to the World Health Organization classification system18 and modified Black nuclear grading system, respectively.19 The method used to determine hormone receptor status depended on the year of primary diagnosis. For specimens obtained before 1993, estrogen and progesterone receptor status were determined using the dextran-coated charcoal ligand-binding method. For specimens obtained after 1993, immunohistochemistry (IHC) methods of staining using monoclonal antibodies were used on 4 μm paraffin-embedded tissue. Monoclonal antibodies 6F11 (Novacastra Laboratories Ltd., Burlingame, Calif) and 1A6 (Novacastra Laboratories Ltd.) were used for assessing estrogen and progesterone receptor status, respectively. HER-2/neu status was determined using IHC and/or fluorescent in situ hybridization (FISH) technique. Specimens exhibiting no evidence of staining on IHC and/or no gene amplification by FISH were considered to have a negative HER-2/neu status. Specimens staining 3+ on IHC and/or demonstrating gene amplification by FISH were considered to have positive HER-2/neu status. Triple negative status was assigned to those patients whose tumor specimens did not demonstrate positive staining for both estrogen and progesterone receptors and had a negative HER-2/neu status.

Statistical Analysis

Patient characteristics were tabulated and compared across CTC groups by chi-square test or Fisher exact test, as appropriate. Median follow-up was measured as the median observation time among all patients and among those still alive at the time of their last follow-up. Overall survival was calculated from the date of CTC measurement to the date of death or last follow-up and was estimated using the Kaplan-Meier product limit method and compared across groups using the log-rank test. Hormone receptor status, HER-2/neu status, triple receptor status, and site of first metastasis were identified a priori as important prognostic variables for this group of patients. The effect of CTCs was explored within each subgroup to determine whether the prognostic significance persisted within each group. Cox proportional hazards models were fit to estimate the association of CTC levels and overall survival after adjusting for hormone receptor status (positive/negative), HER-2/neu status (positive/negative), and site of first metastases (visceral vs other). We tested for all possible interactions between CTCs and the other variables in the model. Modeling was repeated (defined in post hoc analyses) considering triple receptor status (hormone receptor, and HER-2/neu negative vs any positive) in place of hormone receptor and HER-2/neu, as well as considering site of metastasis as bone versus other. All statistical tests were 2-sided, and P values <.05 were considered statistically significant. Analyses were performed using the SAS 9.1 and S-plus 7.0 statistical software (SAS Institute Inc., Cary, NC).


Patient Characteristics

One hundred eighty-five patients were identified and included in this analysis (Table 1); 114 (61.6%) patients and 71 (38.4%) patients had CTC levels <5 and ≥5, respectively. Fifty-six (30.3%) patients presented with de novo MBC, and 129 (69.7%) patients presented with newly diagnosed recurrent MBC. The median age at diagnosis for patients with CTC levels <5 and ≥5 was 49 years (range, 25 years-81 years) and 48.5 years (range, 23 years-84 years), respectively. The majority of patients had invasive ductal histology (80%) and histological grade III disease (55%). One hundred eighteen (63.8%) patients had hormone receptor-positive disease, and 67 (36.2%) patients had hormone receptor-negative disease. One hundred forty-four (77.8%), 33 (17.8%), and 8 (4.3%) patients had tumors that were HER-2/neu negative, positive, and unknown, respectively. Forty-eight (26.3%) patients had tumors that were triple receptor negative.

Table 1. Patient Characteristics by Circulating Tumor Cells Group
 CTC <5 (n=114)CTC ≥5 (n=71)P
  1. CTC indicates circulating tumor cells; ER, estrogen receptor; PR, progesterone receptor; HR, hormone receptor; LVI, lymphovascular invasion; BCS, breast-conserving surgery; ALND, axillary lymph node dissection.

Age at primary diagnosis, y     
Menopausal status     
Triple receptor     
Nuclear grade     
Site of first metastasis     
Bone as first site of metastasis     

Survival Estimates

At the time of last follow-up, 66 (35.7%) patients had died. The median follow-up among all patients and among patients still alive was 9.2 months (range, 0 months-41.7 months) and 9.4 months (range, 0 months-32.9 months), respectively. Table 2 summarizes the Kaplan-Meir overall survival estimates by CTC level and patient characteristics. Median and 1-year overall survival for the whole cohort was 21.4 months (95% confidence interval [CI], 17.3 months-28.3 months) and 78.1% (95% CI, 80.7%-83.8%), respectively. The median survival for patients with CTC <5 and ≥5 was 28.3 months (95% CI, 27.7 months-36.8 months) and 15 months (95% CI, 12.7 months-18.2 months) (Fig. 1), respectively, with an absolute difference in 1-year survival of 23.9% favoring patients with CTC <5 (P < .001).

Figure 1.

Kaplan-Meier plots are shown for overall survival among patients with newly diagnosed, untreated recurrent, and de novo breast cancer. The level of circulating tumor cells (CTCs) was measured at the time of initial presentation before treatment. The Kaplan-Meier plots illustrate our proposed staging system, with stage IVA representing patients with CTC <5 and stage IVB representing patients with CTC ≥5. The median survival for stage IVA is estimated to be 28.3 months. The median survival for patients with stage IVB disease is estimated to be 15 months. P values are 2-sided and unadjusted. HR indicates hazards ratio; CI, confidence interval.

Table 2. Kaplan-Meier Overall Survival Estimates Stratified by CTC Level and Patient/Tumor Characteristics
 nEvents, nMedian95% CI12 Months95% CIP
  1. CI indicates confidence interval; CTC, circulating tumor cells; HR, hormone receptor.

  2. P values are 2-sided and unadjusted.

CTC <51142828.3(27.7-36.8)87.5%(79%-92.7%) 
CTC ≥5713815.0(12.7-18.2)63.6%(50.2%-74.2%)<.0001
HR negative       
 CTC <5451139.9(17.3-39.9)84.0%(67.6%-92.5%) 
 CTC ≥5221213.458.6%(33.9%-76.8%).012
HR positive       
 CTC <5691728.3(27.7-36.8)89.7%(78.5%-95.3%) 
 CTC ≥5492616.5(13.2-19.1)65.5%(49.2%-77.8%)<.0001
Triple receptor positive       
 CTC <5801931.1(27.7-36.8)89.8%(79.6%-95%) 
 CTC ≥5552916.0(13.2-19.1)63.9%(48.7%-75.7%)<.0001
Triple receptor negative       
 CTC <532839.9(15-39.9)84.3%(62.8%-93.9%) 
 CTC ≥516913.4(6.1-18.2)61.9%(31.2%-82.1%).048
HER-2/neu negative       
 CTC <5872128.3(21.6-36.8)88.9%(78.9%-94.3%) 
 CTC ≥5573016.0(9-18.2)63.5%(48.2%-75.4%).0001
HER-2/neu positive       
 CTC <521431.194.4%(66.6%-99.2%) 
 CTC ≥512713.2(2.5-19.1)64.2%(30.2%-84.8%).002
Visceral metastases       
 CTC <5381628.3(21.6-36.8)82.1%(64.2%-91.6%) 
 CTC ≥5321712.7(4.7-16)51.5%(30.9%-68.7%).005
 CTC <5511228.3(19.5-41.7)90.8%(77.2%-96.5%) 
 CTC ≥5543015.0(9-17.3)60.7%(44.7%-73.3%).0003
Other metastases       
 CTC <5761235.3(20.2-35.3)90.3%(79.6%-95.6%) 
 CTC ≥5392116.8(13.5-21.4)72.5%(54.7%-84.2%).002
De novo stage IV       
 CTC <528731.1(28-39.9)87.9%(67%-96%) 
 CTC ≥5281516.8(13.6-21.4)79.5%(57.3%-91%).035
Recurrent stage IV       
 CTC low832028.3(20.2-36.8)87.0%(76.3%-93.1%) 
 CTC high412113.2(7.7-23.9)53.6%(35.8%-68.4%).0003

The prognostic significance of CTC was retained regardless of hormone receptor status (Fig. 2A) and HER-2/neu status. One hundred five patients presented with bone-only metastases as their first site of distant metastases, and 70 patients presented with visceral metastases as their first site of distant metastases. Patients who presented with either bone-only metastases or visceral metastases who had CTC ≥5 did worse than those with CTC <5, with a statistically significant absolute difference of 13 to 15 months in median survival favoring patients with CTC <5 (Fig. 2B, C). In addition, the prognostic impact of CTC was independent of whether patients had recurrent or de novo MBC (Fig. 2D). Specifically, among patients with de novo MBC, an absolute difference in median survival of 14.3 months was observed that favored patients with CTC <5 (P = .035).

Figure 2.

Kaplan-Meier plots are shown for overall survival by circulating tumor cells (CTCs) groups, further stratified by (A) hormone receptor–positive and hormone receptor–negative disease, (B) visceral metastases and other metastases, (C) bone metastases and other metastases, and (D) recurrent and de novo stage IV breast cancer. The plots illustrate that, regardless of stratification, patients with CTC <5 have significantly better overall survival compared with patients with CTC ≥5. P values are 2-sided and unadjusted. HR indicates hazards ratio; CI, confidence interval.

The effect of CTC on overall survival was further evaluated based on treatment groups received (eg, chemotherapy with anthracyclines, taxanes, or both, anthracyclines/taxanes, hormone therapy). Regardless of treatment received, a CTC level of <5 predicted for a significantly better overall survival in all subgroups (data not shown).

Multivariable Models and Interaction Terms

Table 3 summarizes the results for the Cox proportional hazards model. Adjusted for site of metastases (visceral vs other metastases), hormone receptor status, and HER-2/neu status, the hazard of death for patients with CTC level <5 compared with those with CTC ≥5 was 3.64 (95% CI, 2.11-6.30; P < .0001). The proportional hazards model was then run to look at all possible interaction terms. Although these interaction terms were not statistically significant, the estimated effects suggested that there might be some differences by subgroups. When we examined the interaction term between CTCs and bone versus other metastases, an additional risk of death was observed among patients with both bone metastases and CTC levels ≥5 (hazards ratio of interaction, 1.61; 95% CI, 0.52-5.04 [P = .410]).

Table 3. Cox Proportional Hazards Model
 Hazard RatioLower 95% CIUpper 95% CIP
  1. CI indicates confidence interval; CTC, circulating tumor cells; HR, hormone receptor.

CTC ≥5 vs <53.642.116.300<.0001
HR positive vs negative0.920.531.590.770
HER-2/neu positive vs negative0.870.441.700.680
Visceral metastases vs other1.520.892.610.130


The primary goal of this study was to validate the prognostic value of CTCs in an independent cohort of women with newly diagnosed recurrent and/or de novo MBC, to provide the basis for stage IV disease stratification. This retrospective study demonstrated significantly reduced overall survival for patients with a baseline value of ≥5 CTCs, independent of tissue expression of biomarkers (estrogen receptor, HER-2/neu, or triple negative disease) and site of recurrent or metastatic disease. To our knowledge, this is the first time the prognostic value of CTCs has been demonstrated in a cohort of patients with de novo metastatic disease who were treatment-naive. Furthermore, we observed that the prognostic power of CTC was independent of whether patients presented with recurrent or de novo MBC. The results of the current study suggest that the baseline detection of CTCs can discriminate disease with more aggressive features that may require a different therapeutic approach.

Despite improved systemic therapies, the management of MBC has remained palliative.20 CTCs have been the focus of a number of studies specifically examining it as an indicator of therapeutic efficacy and prognoses, with the objective of improving survival in this cohort. In a prospective multi-institutional pivotal trial, Cristofanilli et al10 evaluated the impact of CTCs detection on progression-free and overall survival in a cohort of 177 patients with MBC who were starting a new line of therapy on evidence of disease progression. In this seminal study, it was observed that patients with ≥5 CTCs/7.5 mL of blood at baseline had poorer median progression-free survival (2.7 months vs 7.0 months; P = .0001) and overall survival (10.9 months vs 21.9 months; P < .0001) compared with patients with <5 CTCs. In addition, it was also observed that patients who initially presented with ≥5 CTCs who responded to treatment reflected by a decrease in their CTC level to below 5 had better survival rates, essentially suggesting that CTC could be used jointly as an indicator of both prognoses and therapeutic efficacy. These results were then confirmed in a follow-up study of the same cohort by Hayes et al,12 who essentially confirmed that a CTC level ≥5 at any time point in the natural course of MBC was associated with a worse prognosis. The results of the current study strengthen the validity of these findings.

After the pivotal trial,10 our group recognized that before CTC measurement could be incorporated into the prognostic stratification of patients with MBC, it had to be compared with and outperform established prognostic markers. Prognostic markers that are used in MBC include the Swenerton score,5 CA 27.29, hormone receptor status,6 and subtype of disease (basal vs luminal disease).21, 22 By using information derived from a cohort of 151 patients with MBC and measured CTC levels, we demonstrated that the prognostic power of CTC was independent of and either equivalent or superior to these factors.11 Moreover, the results indicated that the prognostic power of CTC held irrespective of whether patients received hormonal treatment, chemotherapy, and/or trastuzumab treatment. These results essentially reflected the broad range of applicability of CTCs in determining prognosis within various subgroups of patients with MBC.

The results of the current study are unique in that we defined the prognostic power of CTCs within an independent cohort of patients with measurable and evaluable newly diagnosed recurrent and/or de novo metastatic disease. The evidence from the current study not only reaffirms the prognostic power of CTCs, but also provides evidence of the applicability of this new prognostic marker in the clinical setting. Moreover, the results of our subgroup analyses demonstrate interesting results. For example, it has been traditionally observed that patients with MBC who have bone-only metastases have better survival rates compared with other sites of disease such as the viscera.4 In our study, within the subgroup of patients who had bone-only metastases or visceral metastases at the time of MBC diagnosis, an absolute 1-year survival advantage of 30.1% (P = .0003) and 30.6% (P = .005), respectively, was observed favoring the group with CTC levels <5. Similar results were obtained among patients within various other subgroups. These results are important in that they demonstrate that whether we are dealing with a relatively good prognostic group of patients (eg, bone-only metastases subgroup, hormone receptor-positive subgroup, HER-2/neu-positive subgroup) or a group with features traditionally considered to be more aggressive (eg, visceral metastases, hormone receptor-negative subgroup, HER-2/neu-positive subgroup), CTCs can be used for further prognostic stratification.

The American Society of Clinical Oncology recently published their recommendations on the use of tumor markers in patients with both early-stage and advanced-stage breast cancer.23 The consensus was that although “Cellsearch assay appears to provide both a prognostic utility and a predictive use in metastatic breast cancer,” at the time of publication the panel could not recommend the use of the CellSearch System in patients with MBC “until further validation confirms the clinical value of this test.” Because of this publication, we argue that the results of this study and our previous study11 provide sufficient data to recommend use of this system for stratification of patients with newly diagnosed untreated MBC into 2 strata (Fig. 1).

The AJCC staging of disease17 quantifies only the physical extent of disease, without consistent incorporation of novel prognostic technology. In brief, the TNM system attempts to define the disease by incorporating all aspects of cancer distribution in terms of the primary tumor (T), lymph nodes (N), and distant metastasis (M). A “stage” group (0, I, II, III, IV) is then assigned based on the various possible TNM permutations, with 0 reflecting minimal involvement and IV either the greatest tumor involvement or distant metastasis. Our report argues that by using a technology that allows for detection of CTCs, we are able to further discriminate patients with metastatic disease into 2 distinct prognostic groups. Upfront stratification of patients with newly diagnosed MBC using this schema would have several advantages. First, it would provide a biologic (and not merely anatomic) distinction of prognoses upfront based on a reliable and reproducible test (eg, stage IVA vs IVB). Second, it would allow patients with more aggressive or stage IVB disease to be actively enrolled into clinical trials designed to address specific questions related to the selection of more selective and targeted therapies aimed at the elimination of CTCs, which could significantly impact survival in MBC.

In conclusion, to our knowledge the current study has 1 of the largest cohorts assessing the prognostic value of CTCs among women with newly diagnosed MBC. Although areas of uncertainty still exist regarding the use of CTCs as an indicator of therapeutic efficacy, which should be assessed prospectively in clinical trials, we believe our group has presented sufficient evidence to recommend the incorporation of CTCs as a prognostic indicator in patients with MBC. Furthermore, based on the data we presented, we believe that the AJCC should consider an in-depth discussion regarding these data and the possibility that such diagnostic technology could be considered for the stratification of newly diagnosed patients with stage IV disease into 2 groups: stage IVA (CTC <5) and stage IVB (CTC ≥5).