Search tips
Search criteria 


Logo of nihpaAbout Author manuscriptsSubmit a manuscriptHHS Public Access; Author Manuscript; Accepted for publication in peer reviewed journal;
Cancer Cell. Author manuscript; available in PMC 2010 January 7.
Published in final edited form as:
PMCID: PMC2749247

Latent bone metastasis in breast cancer tied to Src-dependent survival signals


Metastasis may arise years after removal of a primary tumor. The mechanisms allowing latent disseminated cancer cells to survive are unknown. We report that a gene-expression signature of c-Src activation is associated with late-onset bone metastasis in breast cancer. This link is independent of hormone receptor status or breast cancer subtype. In breast cancer cells, c-Src is dispensable for homing to the bones or lungs critical for the survival and outgrowth of these cells in the bone marrow. c-Src mediates AKT regulation and cancer cell survival responses to CXCL12 and TRAIL, factors which are distinctively expressed in the bone metastasis microenvironment. Breast cancer cells that lodge in the bone marrow succumb in this environment when deprived of c-Src activity.


We sought to identify signaling pathways that support the survival of disseminated breast cancer cells and thereby extend the period during which metastasis may emerge after the diagnosis and removal of a breast tumor. Based on bioinformatics analyses, we discovered a strong association between late-onset bone metastasis and c-Src activity in a cohort of over 600 breast cancer patients. We further demonstrated that c-Src supports cancer cell survival in the bone marrow microenvironment by facilitating CXCL12-CXCR4-AKT signaling and by conferring resistance to TRAIL. The link between c-Src-dependent signaling and metastatic cell survival provides mechanistic insights into metastasis latency, and suggests strategies to hasten the attrition of disseminated breast cancer cells.


Distant metastasis is the final and most threatening stage of malignant tumor progression, but its clinical course and biological features vary extensively depending on the type of cancer. One of the most strikingly variables in metastasis is its latency. In some types of cancer metastasis occurs soon after a primary tumor develops, whereas in others it emerges years or even decades after the primary tumor is removed (Karrison et al., 1999; Schmidt-Kittler et al., 2003; Weckermann et al., 2001). During such long latency periods the disseminated cancer cells survive maladapted to the host organ microenvironment, either in a dormant state or as indolent micrometastases, until they eventually meet all the requirements for metastatic outgrowth. The long-term permanence of latent metastatic disease implies the existence of mechanisms that keep the disseminated cancer cells alive. The identification of mechanisms that support the survival of disseminated cancer cells in their host microenvironments has clear implications for understanding the biology and improving the treatment of latent metastatic disease.

Latent metastasis leading to late-onset relapse occurs frequently in breast cancer. Nearly one third of cases of relapse emerge after three years of breast cancer diagnosis, with some cases emerging several decades later (Karrison et al., 1999; Schmidt-Kittler et al., 2003). A majority of these late-onset metastases arise in bone and may subsequently appear in other organs. Bone metastases from breast cancer are typically osteolytic, involving the mobilization of osteoclasts that cause pathological bone resorption, with intense pain, bone fractures, nerve compression, and hypercalcemia (Mundy, 2002). The development and osteolytic nature of these lesions are based on complex interactions between cancer cells and bone marrow stroma in a cycle of bone destruction and tumor expansion. The complexity of cellular interactions and molecular components implicated in bone metastasis has hindered a mechanistic elucidation of key biological features of this process, in particular the basis for long-term survival of metastatic cells in the bone marrow (Husemann et al., 2008; Stoecklein et al., 2008).

c-Src is the prototypic member of a non-receptor tyrosine protein kinase family. Its action on mammalian cells is broadly pleiotropic, including effects on cell morphology, adhesion, migration, invasiveness, proliferation, differentiation, and survival. c-Src participates in the activation of various downstream pathways through molecular interactions with growth factor receptors (EGFR, HER2), integrin cell adhesion receptors, steroid hormone receptors, G-protein coupled receptors, focal adhesion kinase, and cytoskeleton components (Bromann et al., 2004; Ishizawar and Parsons, 2004). Aberrant expression and activation of c-Src occurs in several tumor types and has been correlated with poor outcome, which has stimulated interest in Src kinase inhibitors as therapeutic agents in cancer, some of which have entered clinical trials (Finn, 2008; Rucci et al., 2008; Summy and Gallick, 2006). This knowledge notwithstanding, studies to date have not pinned specific roles of c-Src in particular aspects of tumor development in humans. Gene-expression profiling techniques that can register the activity of signaling pathways provide new tools to search for links between specific pathways and defined steps of tumor progression (Bild et al., 2006; Klapholz-Brown et al., 2007; Padua et al., 2008). Using this approach, we are reporting here on an association between Src pathway activity and late-onset bone metastasis.


Src activity in breast tumors is associated with late-onset bone metastasis

To investigate the association of pathway-specific gene expression signatures with breast cancer outcome we assembled a combined cohort of 615 breast tumors for which genome-wide gene expression data was available. Approximately two thirds of these cases were lymph node negative at diagnosis, and were not subject to adjuvant therapy after primary tumor resection (Wang et al., 2005; Yu et al., 2007). The rest of cases were lymph node positive or received adjuvant hormonal therapy and/or chemotherapy (Table S1). Among the 615 cases, 43% relapsed and the median metastasis-free survival time was 22.1 months. ER+ cases had a longer disease-free interval and a lower rate of recurrence during the first 20 months compared to the ER- cases (Figure 1A). The rate of late-onset metastasis (defined as relapse > 5 years after cancer diagnosis) was significantly higher in ER+ cases. The overall incidence of metastasis in ER+ and ER cases merged over time. These characteristics fit with the typical course of breast cancer progression (EBCTCG, 2005; Hess et al., 2003).

Figure 1
Src response signature (SRS) in breast tumors is associated with bone metastasis

ER+ breast tumors relapse most prominently to the bones over a protracted period (Hess et al., 2003). Indeed, the overall rate of bone metastasis and the rate of late-onset bone-metastasis in the 615 breast cancer patients were significantly higher in ER+ cases than in ER cases (Figure 1B). Reasoning that this metastatic latency may depend on a particular cell survival pathway, we employed gene-expression bioinformatics classifiers that register the activity of specific pathways, including the c-Src, H-Ras. E2F3, Myc, β-catenin (Bild et al., 2006), TCF/Wnt (van de Wetering et al., 2002), and TGFβ pathways (Padua et al., 2008). Notably, a gene expression signature that denotes c-Src activity (designated as Src responsive signature, SRS) was tightly associated with overall bone metastasis and late-onset of bone metastasis (Figure 1C). Bone relapse was more closely associated with SRS+ status than with ER+ status (p=7×10−9 versus p=0.001 for overall bone relapse; p=0.0002 versus p=0.003 for late-onset metastasis, log rank test). When we repeated the analysis after excluding patients who died from other metastases, SRS status was still strongly associated with overall bone metastasis (p=6.4×10−6) and late-onset bone metastasis (p=0.0037). The only other pathway that showed an association with bone relapse was the H-Ras pathway, but this association was marginal (p=0.04; Table 1). The TGFβ response signature in primary tumors was associated with lung but not bone relapse (Table 1), as reported (Padua et al., 2008). Breast tumors comprise a heterogeneous mix that includes several molecularly distinct subtypes (Sorlie et al., 2003). All these subtypes were represented among the SRS+ tumors (Figure 1D).

Table 1
Association of pathway gene-expression signatures with metastasis in breast cancer

A majority (88.4%) of ER+ tumors in this cohort scored as SRS+ (Figure 1C; Table S2). Indeed, c-Src and ERα engage in molecular interactions that lead to non-transcriptional effects (Collins and Webb, 1999; Ishizawar and Parsons, 2004). However, 23.0% of ER tumors also scored as SRS+, providing us with an opportunity to discern whether Src activity is specifically linked to bone metastasis independently of ER+ status. The ER status of these tumors was confirmed based on microarray ERα probe signal (Figure S1). Since the vast majority of bone relapses in ER- cases occur before 5 years, we decided to use a less stringent >3-year cutoff to define ER bone metastases that occur relatively late. Indeed, a multivariate analysis based on the Cox proportional hazard model indicated that the hazard ratio of SRS+ patient starts to dramatically increase 3 years after tumor resection, and remains at comparable levels at 4-year and 5-year cutoffs (Figure 2A). We therefore decided to use 3-year as the cutoff for late-onset bone metastasis in the particular case of ER- patients. Compared to ERSRS tumors, the ERSRS+ tumors showed a higher incidence of bone metastasis and late-onset bone metastasis (p=0.003, Figure 2B,C, log rank test). ERSRS+ cases developed bone metastasis with rate and kinetics that were similar to those of ER+SRS+ cases (Figure 2B,C). Conversely, when the analysis was restricted within the SRS+ or SRS subsets, ER status was no longer correlated with bone relapse (p=0.33 and p=0.69 for SRS and SRS+ groups, respectively; Figure 2B).

Figure 2
SRS defines a subset of ER- patients that develop delayed bone metastases

Multivariate analyses showed that the association of SRS+ status with bone relapse and/or late-onset bone relapse was independent of PR (progesterone receptor) status, HER2 status, molecular subtypes, Genomic Grade Index, and differentiation (Sotiriou et al., 2006) (Tables S3, S4 and S5). A subset of the patients was subject to adjuvant therapies (Table S6), but a multivariate analysis showed that SRS associates with bone metastasis independently of either hormonal or chemotherapy (Table S7). No correlation was observed between SRS status and lung or brain relapse (Table S8). An association between SRS status and liver metastasis was of marginal significance (p = 0.035) and limited to ER+ patients. For some patients, metastases occurred to multiple organs, which were diagnosed at the same time. These patients were scored for each of the corresponding sites in our analyses. This potentially confounding factor made little impact to the current study, because fewer that 20% of the patients with late onset bone metastases also developed metastases to other sites. Collectively, these observations suggest a specific link between SRS+ status and late-onset relapse to bone in breast cancer.

c-Src selectively promotes bone metastasis

Since ERSRS+ breast tumors indicated an ER-independent association between Src activity and bone metastasis we sought to identify a suitable ERSRS+ metastatic model system to further investigate this link. MDA-MB-231 (MDA231 for short) is a cell line derived from the pleural fluid of an ER breast cancer patient with disseminated disease. MDA231 is a basal subtype, “triple negative” (ER, PR, HER2) cell line (Neve et al., 2006). By in-vivo selection we previously isolated MDA231 cell populations with distinct metastatic organ tropisms in mice (Kang et al., 2003; Minn et al., 2005). Gene expression profiles from these derivatives were subjected to SRS analysis. The parental MDA231 population as well as various bone-metastatic and adrenal gland-metastatic derivatives showed a higher SRS score than did the lung metastatic and brain metastatic derivatives (Figure 3A; Figure S2). The level of activated (Tyr416-phosphorylated) c-Src (Thomas and Brugge, 1997) was higher in representative bone and adrenal metastatic MDA231 derivatives than in lung and brain metastatic derivatives (Figure 3B).

Figure 3
c-Src selectively promotes bone metastasis

To test the role of c-Src activity in bone metastasis, we stably reduced the level of endogenous c-Src using a shRNA vector in the MDA231 bone metastatic derivative BoM-1833 (Kang et al., 2003). shRNA-mediated knockdown reduced the level of endogenous c-Src and of activated c-Src by >90% (Figure 3C). We also generated a Src-rescued BoM-1833 line expressing this shRNA and the chicken c-Src gene, which is resistant to the shRNA. On injection of 3×104 cells into the left ventricle, all mice in the BoM-1833 control group were dead with bone metastasis within 70 days. Knockdown of c-Src extended survival by ~25 days, whereas Src rescue reversed this gain (Figure 3D).

We monitored the kinetics of emergence of hind-limb bone lesions by quantitative bioluminescence imaging (BLI) of luciferase activity from a stably integrated vector. The control, Src-knockdown, and Src-rescued BoM-1833 cells localized to the bones and initiated outgrowth at similar rates for up to 14 days (Figure 3E, F). However, from day 21 after injection, the rate of outgrowth of the bone lesions in the Src knockdown group decreased becoming 10-fold lower than that of the control groups (Figure 3F). This difference was accompanied with a sharp reduction in the extent of the hind-limb osteolytic lesions, as determined by X-ray imaging and histological examination (Figure 3E). Knockdown of two other Src family kinases, Fyn and Yes, did not inhibit the bone metastatic activity in BoM-1833 cells (Figure S3).

c-Src knockdown did not decrease the intrinsic proliferative activity of BoM-1833 cells in culture (Figure S4), or their growth as tumors in mammary glands (Figure 3G). c-Src knock down had no effect on the limited lung metastasis activity of these orthotopic BoM-1833 tumors (Figure 3G). Using the highly lung-metastatic MDA231 derivative line LM2-4175 (Minn et al., 2005) we additionally tested the effect of c-Src knockdown using tail-vein inoculation assays, which test for lung colonization activity (Figure 3H). The knockdown of c-Src had no effect on lung tumor burden. These observations pointed to a selective and specific role of c-Src in bone metastasis.

c-Src enhances metastatic outgrowth in the bone marrow microenvironment

The kinetics of BoM-1833 bone colonization suggested a role of c-Src in the sustained outgrowth of these colonies. Indeed, when directly inoculated into the bone marrow cavity of tibiae and tracked the tumor outgrowth, the c-Src–depleted cells were approximately 10-fold less active than control cells, and this inhibition was rescued by c-Src expression (Figure 4A-B). Dasatinib (BMS-354825) is a pharmacological inhibitor of Src family tyrosine kinases (Shah et al., 2004). Administration of dasatinib at 10mg/kg daily by oral gavage (Kamath et al., 2008) after inoculation of BoM-1833 cells in mice significantly prevented the formation of osteolytic metastases (Figure 4C). An inhibitory effect was observed when dasatinib treatment was started 7 days or 14 days after cancer cell inoculation. Dasatinib had no effect on the lung colonizing activity of LM2-4175 cells (Figure S5).

Figure 4
c-Src enhances metastasis survival and outgrowth in the bone marrow

Dasatinib inhibits various Src family kinases, and c-Src is functionally important in various cell types in the bone marrow stroma (Abram and Lowell, 2008). To establish the specificity of the dasatinib effect on bone metastasis, we stably expressed a dasatinib-resistant mutant form of c-Src (Du et al., 2009) in Src-knockdown BoM-1833 cells. The bone metastatic activity of BoM-1833 could be inhibited by dasatinib when the Src-knockdown was rescued with the wild type c-Src but not when it was rescued with the dasatinib-resistant c-Src mutant (Figure 4D). Thus, genetic and pharmacological evidence argued that c-Src activity is dispensable for bone marrow seeding by MDA231 cells but is rate-limiting for the sustained growth of these colonies.

Src has been implicated in cancer cell proliferation and survival, and in osteoclast differentiation (Horne et al., 2005; Lowell et al., 1996). Lesions formed by Src-knockdown BoM-1833 cells contained a high proportion of apoptotic (TUNEL+) cells compared to bone lesions formed by control BoM-1833 cells, whereas this proportion in Src-rescued BoM-1833 lesions was low and comparable to that in the controls (Figure 4E; Figure S6). Administration of dasatinib also induced a high level of TUNEL+ cells. By contrast, Src knockdown and dasatinib treatment had no effect on the proportion of proliferating (Ki67+) cancer cells (Figure 4F), or of tartrate resistant acid phosphatase (TRAP+) osteoclasts located at the tumor-bone matrix interface (Figure 4G,H). These results suggest that c-Src supports the aggressive outgrowth of BoM-1833 cells by promoting their survival in the bone marrow microenvironment.

Src supports survival of indolent breast cancer cells in the bone marrow

As BoM-1833 cells are a model of acute bone metastatic colonization, we also tested the role of c-Src in a more indolent model of bone metastatic disease. From the pleural fluid of a breast cancer patient we isolated a malignant cell population, termed CN34, which was ER (Figure S7A). CN34 cells transduced with a luciferase/GFP vector and a puromycin-resistant marker yielded one bone lesion after 9 weeks of intracardial inoculation in five mice. Extraction of these cells and expansion in culture, followed by two additional cycles of selection as bone lesions yielded CN34-BoM2 derivatives. Compared to a CN34-BrM3 derivative that is metastatic to brain but not bone (data not shown), CN34-BoM2 cells showed a higher level of activated Src (Figure 5A). However, compared to BoM-1833 cells, CN34-BoM2 grow in bone at a 100-fold lower rate (Figure 5B). CN34-BoM2 cells are also metastatic to lymph nodes (Figure 5C; Figure S7). Src-knockdown CN34-BoM2 cells (refer to Figure 7A) were almost devoid of bone metastatic activity, whereas the Src-rescued counterparts recovered the lost activity (Figure 5B). By contrast, c-Src knockdown did not inhibit lymph node metastasis (Figure 5C; Figure S7).

Figure 5
c-Src supports survival of indolent breast cancer cells in the bone marrow
Figure 7
c-Src mediates CXCL12-dependent survival and resistance to TRAIL death signals in breast cancer cells

To test the effect of Src on survival under even more indolent conditions, we inoculated mice with a low number (5×104) of control or Src-knockdown CN34-BoM2 cells. No BLI signal developed above background (<104 photons per second) in 35 days. The bone marrow was flushed from the femurs of these mice and placed in culture under puromycin selection (Figure 5D). The resulting colonies expressed GFP and stained positive for human vimentin, confirming that these were the human-tumor derived cells (Figure 5E). Compared to mice inoculated with control CN34-BoM2, mice inoculated with Src-knockdown CN34-BoM2 yielded 9-fold fewer bone marrow-derived human cell colonies (Figure 5F). Thus, c-Src was required for the survival of these indolent breast cancer cells in the bone marrow.

Cell survival factors in the bone metastasis microenvironment

In light of these results we postulated that c-Src may be required for the response of breast cancer cells to survival factors provided by the bone marrow microenvironment. To identify relevant factors we conducted microarray gene-expression analysis of a cohort of metastasis samples that were surgically removed from breast cancer patients at MSKCC, including 16 metastases from bone, 18 from lung, 19 from brain, and 5 from liver. Analysis of SRS status showed that 62.5% (10/16) of bone metastases are SRS+ (Table S9), in agreement with a report that the majority of breast cancer cells in bone metastases immunostain positive for activated c-Src (Planas-Silva et al., 2006). Only 28.6% (12/42) of metastases to other sites scored SRS+. Of 260 cytokines and chemokines (compiled by Gene Ontology database, ID: GO:0005125) in these samples, we focused on those that are expressed in bone metastases at a higher level (>2-fold) than in metastases from other sites. Seventeen genes met these criteria (Figure 6A). These genes include the chemokine CXCL12/SDF1, which is produced by bone marrow mesenchymal cells and acts as an attractant and survival factor for cells expressing the receptor, CXCR4 (Muller et al., 2001). Also included are insulin-like growth factors (IGFs), bone morphogenetic protein-2 (BMP2) and transforming growth factors β (TGFβs), which are stored in the bone matrix and released during osteolysis (Guise et al., 2006; Mundy, 2002). IGFs stimulate survival in various cell types (Stewart and Rotwein, 1996), whereas BMPs and TGFβs are implicated in the generation of osteoblasts and osteoclasts, respectively (Chen et al., 2004). Other factors of interest that were selectively expressed in bone metastasis tissues include platelet-derived growth factor-A (PDGF-A), placenta growth factor (PGF) and vascular-endothelial growth factor-C (VEGF-C), which are potential survival and microenvironment-modifying factors, and the pro-apoptotic member of the tumor necrosis factor family TRAIL (encoded by TNFSF10) (Ashkenazi, 2002).

Figure 6
Cytokines of the bone metastasis microenvironment

To determine whether a similar profile of cytokines is present in our experimental model, BoM-1833 cells were injected in high numbers (5×105 cells) to exploit their residual lung and brain metastatic activity of this cell line in addition to the bone metastatic activity. qRT-PCR assays of bone, brain and lung metastases formed by BoM-1833 xenografted in mice confirmed that CXCL12, IGF1, PDGF-A, VEGF-C and TGFβ1 were more abundantly expressed in bone metastases than in metastasis to brain or lungs (Figure 6B). TRAIL was expressed at higher level in bone and lung metastases than in brain metastases.

c-Src mediates CXCL12-dependent survival and resistance to TRAIL death signals

CXCL12 signaling through CXCR4 activates several signaling pathways including AKT (Epstein, 2004). c-Src has been independently implicated in the activation of the cell survival AKT kinase (Lu et al., 2003). CXCR4 is consistently expressed at a high level in our cell line systems and in a majority of human bone metastases (Kang et al., 2003). Indeed, the addition of CXCL12 to BoM-1833 and CN34-BoM cells under serum-free conditions caused AKT activation, and this activation was diminished by c-Src-knockdown cells and restored by rescuing c-Src expression (Figure 7A). When cultured in serum-free medium, BoM-1833 and CN34-BoM cells undergo apoptosis after two days (BoM-1833) or five days (CN34-BoM). CXCL12 significantly reduced apoptosis under these conditions, whereas c-Src depletion prevented this pro-survival effect (Figure 7B). Concordantly, CXCL12 enhanced cell viability in a c-Src-dependent manner (Figure 7C). Among the other cytokines enriched in the bone metastasis samples, IGF1 induced AKT activation (Figure S8) but not cell survival (data not shown), and TGFβ1, BMP2, PDGF-A and VEGF-C had little or no effect on either AKT or cell survival (Supplemental Figure 8; data not shown).

Because the pro-apoptotic cytokine TRAIL was abundantly expressed in bone metastasis tissue, we tested whether c-Src plays a role in protecting the bone metastatic cancer cells from TRAIL-induced cell death. TRAIL induces caspase activation and apoptosis by binding to cell death receptor DR4 and DR5 (Ashkenazi, 2002). DR5 is expressed in our cell line systems and in bone metastases (Affymetrix HG-U133A raw intensity range 300 to 600 after MAS5.0 normalization with global scaling factor set to 500). TRAIL addition caused a small increase in the level of activated caspase-3 (Figure 7D) and apoptosis in BoM-1833 cells (Figure 7B), and a small decrease in cell viability (Figure 7C). Notably, the knockdown of c-Src strongly increased the sensitivity of BoM-1833 cells to these pro-apoptotic effects of TRAIL, and this increase was reversed by the rescue of c-Src expression (Figure 7B-D). These results argue that in bone metastatic breast cancer cells c-Src is required for AKT activation and cell survival in response to CXCL12, and for resistance to the pro-apoptotic effect of TRAIL.


The present work provides clinical and experimental evidence for a role of c-Src as a critical mediator of survival signals in breast cancer cells that infiltrate the bone marrow. c-Src mediates survival responses of breast cancer cells to CXCL12 and TRAIL–factors that are prominently expressed in the bone metastasis microenvironment (Figure 7E). Strategies to suppress the survival of disseminated cancer cells can be envisioned based on these results.

c-Src activity linked late-onset bone metastasis in breast cancer

Using gene-expression signatures that denote activation of specific pathways we show that expression of a c-Src response signature (SRS) in primary breast tumors is associated with late-onset bone metastasis. Among 615 cases that we analyzed, nearly 15% of bone relapses from SRS+ tumors occurred five years after diagnosis, versus none in SRS tumors. This association occurs in breast tumors of different subtypes and independently of regional spread or adjuvant therapy. Although most ER+ breast tumors are SRS+, and estrogen receptors can interact with c-Src (Ishizawar and Parsons, 2004), the link of SRS status with bone metastasis is independent of ER status. Nearly one quarter of all ER tumors in this cohort are SRS+, and this subset also shows a high incidence of late bone metastasis. The association of SRS+ status with bone metastasis is more striking than the association of ER+ status with bone metastasis, and the same applies to the association with late-onset bone metastasis.

Our evidence suggests a more robust and sustained ability of SRS+ breast cancer cells to develop bone metastases. A majority of breast cancer bone metastases show activated c-Src by immunostaining (Planas-Silva et al., 2006), and we show that c-Src hyperactivity endows breast cancer cells with a superior ability to persist in the bone marrow microenvironment. Although c-Src has well-characterized effects on cell motility and invasion (Dehio et al., 1995), and the c-Src knockdown BoM-1833 cells show a decrease in these activities in vitro, no effect of c-Src on bone metastasis seeding was manifest in vivo. That most ER+ tumors are SRS+, and thus may derive a survival advantage in the bone marrow microenvironment, is consistent with previous evidence reports that interactions with estrogen receptor-α activate c-Src (Migliaccio et al., 2002). The mechanism for c-Src activation in ER- breast cancer cells remains unknown but could involve receptor tyrosine kinases, G-protein coupled receptors, integrins, and focal adhesion kinase (Bromann et al., 2004; Guo and Giancotti, 2004; Ishizawar and Parsons, 2004; Mitra and Schlaepfer, 2006).

c-Src-dependent CXCL12 survival signals in the bone marrow microenvironment

Our results suggest a cell autonomous, pro-survival role of c-Src in breast cancer cells in the bone marrow microenvironment. Analysis of human breast cancer metastasis tissues revealed a group of seventeen secreted factors that are selectively expressed in bone metastases but not in lung, liver, or brain metastases. We show that the pro-survival effect of one of these factors, CXCL12, on metastatic breast cancer cells requires c-Src. CXCL12 signals through the heterotrimeric G protein-coupled receptor CXCR4 (Epstein, 2004), which is clinically and experimentally implicated in breast cancer metastasis to bone (Kang et al., 2003; Liang et al., 2005; Muller et al., 2001). CXCR4 is highly expressed in BoM-1833 and other bone metastatic MDA231 derivatives (Kang et al., 2003) and in clinical bone metastasis samples. CXCR4 (Chinni et al., 2006) and c-Src (Lu et al., 2003) have been implicated in the activation of the AKT cell survival pathway. Our finding that c-Src is required for CXCL12 activation of AKT and cell survival in bone metastatic breast cancer cells provides a common thread for these various observations. We additionally show that c-Src activity is required for resistance of metastatic breast cancer cells to the pro-apoptotic effects of TRAIL. Through the death receptors and activation of the extrinsic pro-apoptosis pathway, TRAIL mediates immune surveillance and anti-tumor cytotoxicity by dendritic cells, monocytes, NK cells, and effector T cells (Ashkenazi and Herbst, 2008). By mediating CXCL12 survival signals and protecting against TRAIL cell death signals c-Src provides a cell-autonomous pro-survival function in cancer cells that infiltrate the bone marrow.

Bone marrow infiltration, cancer cell survival, and metastatic colonization

Breast tumors are competent for early dissemination and infiltration of distant organs (Husemann et al., 2008; Stoecklein et al., 2008), but unlike other types of tumors (e.g. lung adenocarcinoma, pancreatic carcinoma) the disseminated breast cancer cells often lack the competence for immediate metastatic outgrowth (Figure 7E). Years may pass before aggressive bone colonization occurs through the expression of osteoclastogenic factors and other mediators (Guise et al., 2006; Kang et al., 2003; Mundy, 2002; Yin et al., 1999). We suggest that by enhancing the responsiveness of breast cancer cells to bone marrow survival factors, c-Src extends the endurance of disseminated cancer cells, thereby increasing the probability of acquisition of full metastatic competence (Figure 7E). Other survival mechanisms may support latency in other organs (e.g. the lungs) or in other late-onset metastasis cancers (e.g. prostate carcinoma), although bone-derived secondary metastases to other organs are also possible.

Despite effective systemic treatments including chemotherapy, hormone therapy, and monoclonal antibody therapy, high-risk subsets of patients continue to have unacceptable rates of relapse in the bone and other organs (Hayes et al., 2007; Hudis, 2007). High-dose chemotherapy in combination with autologous hematopoietic stem-cell support has not shown a benefit in high-risk breast cancer patients compared to conventional standard-dose chemotherapy (Zander et al., 2004). At present, the major clinical benefits from post-operative adjuvant drug therapies are observed in the first few years after treatment, suggesting that latent cancer cells are at least partially resistant to conventional therapy (Berry et al., 2006). Disseminated cancer cell pools may act as reservoirs that are resistant to such treatments but potentially susceptible to treatments targeting mediators of latent cancer cell survival. Our data raise the possibility that we might be able to use such agents to attack reservoirs of disseminated latent cancer cells, including circulating tumor cells or mitotically-dormant tumor cells in various organs. Consistent with this is the fact that current chemotherapy and hormone-directed drugs do not specifically target c-Src. Recently, however, several agents that target c-Src pathways have become available for clinical testing (Finn, 2008; Rucci et al., 2008; Summy and Gallick, 2006). This approach could involve the use of c-Src inhibitors, TRAIL-receptor agonists (Ashkenazi and Herbst, 2008), or other latency-targeted agents either simultaneously with adjuvant drug regimens or afterwards in the form of long-term maintenance of disease-free status.

Experimental Procedures

Additional methods for cell culture assays, quantitative real-time PCR, immunoblotting, generation knockdown and overexpression cell lines, details of tumor xenografting, histological analyses, long-term survival assays and statistical analyses are provided in a supplementary section.

Primary tumor and metastasis tissue samples

We compiled a microarray dataset of 615 patients from MSKCC and EMC (EMC344, EMC189 and MSK82, GEO accession numbers: GEO2603, GSE5327, GSE2034 and GSE12276). These datasets were all normalized using MAS5.0 and each microarray was centered to the median of all probes. For each patient, metastasis free survival (MFS) is defined as the time interval between the surgery and the diagnosis of metastasis.

Archival human breast carcinoma metastasis specimens were obtained and processed in compliance with protocols approved by the MSKCC Institutional Review Board. Samples were snap frozen in liquid nitrogen and stored at -80°C. Each sample was examined histologically using hematoxylin and eosin stained cryostat sections. Regions were manually dissected from the frozen block to provide consistent tumor cell content of greater than 70% in tissues used for analysis. RNA was extracted from frozen tissues by homogenization in TRIzol reagent (GIBCO/BRL) and evaluated for integrity. Complementary DNA was synthesized from total RNA using a T7-promoter-tagged-dT primer. RNA target was synthesized by in vitro transcription and labeled with biotinylated nucleotides (Enzo Biochem, Farmingdale, NY). Labeled target was assessed by hybridization to Test3 arrays (Affymetrix, Santa Clara, CA). All gene expression analysis was carried out using HG-U133A (for 36 samples) or HG-U133plus2 (for 29 samples) GeneChips. Seven samples were profiled on both platforms and the data were averaged for the common probes. Gene expression was quantitated using GCOS.

Molecular pathway gene-expression signatures

To predict pathway activation from microarray gene-expression data, we used gene expression profiles derived from the overexpression of c-Src, H-Ras, β-catenin, E2F3, Myc in quiescent mammary epithelial cells (Bild et al., 2006). We derived a gene signature-expression classifier for each of these pathways using a false discovery rate of 0.05 and a fold-change of 1.5 as criteria. The 605 genes that met these thresholds in the c-Src responsive gene set were filtered by using the EMC-344 data set to eliminate non-informative genes. After eliminating genes that were either expressed at a low level (raw intensity < 64 or 26 in more than 25% tumors) or were non-variable across samples (standard deviation < 0.8 for log2 intensity, equivalent to the median of all genes), 159 genes remained that constitute the c-Src responsive signature (SRS) used here.

The SRS was applied to EMC-344 and MSK-82 datasets, which are based on HG-U133A and were combined, and to EMC-189 dataset, which is based on HG-U133plus2 and was processed separately. To search for breast cancers with SRS expression pattern similar to the c-Src-activated mammary epithelial cells (Bild et al., 2006) we performed unsupervised clustering (using heatmap.2 function in gplots package of R statistical software). Two clusters were consistently revealed by such procedure (R index = 0.85) (McShane et al., 2002). One cluster was identified as SRS+ based on the criteria that it exhibits gene-expression similarity to c-Src-activated mammary epithelial cells as gauged by positive Pearson’s correlation coefficients, “metagene” scores (Bild et al., 2006), and that it enriches ER+ tumors (Collins and Webb, 1999; Ishizawar and Parsons, 2004). The other cluster showed the opposite characteristics and was denoted as SRS.

The same approach was applied for β-catenin, E2F3, H-Ras, and c-Myc pathways. TGFβ pathway was gauged as previously described (Padua et al., 2008). For TCF/Wnt pathway, we performed unsupervised clustering using the dominant negative TCF4 signature (van de Wetering et al., 2002). One of the two clusters significantly overexpresses the vast majority of the dnTCF4 genes, and tumors in this cluster are therefore defined as TCF/Wnt+.

To examine the prognostic value of SRS in different subsets of breast cancers, we divided the breast cancer samples based on their ER status or molecular subtypes. For ER status, we used either published pathological annotations (for GSE2063, GSE5327 and GSE2034), or the intensity of probe “205225_at” (ESR1) on the Affymetrix chip when the pathological status was not available (for GSE12276). We used raw intensity of 1000 as the cutoff to define ER- vs. ER+. It has been established that this is an appropriate cutoff when the data is normalized with MAS5.0 and the global scaling is set to 600 (Foekens et al., 2006). Molecular subtype classification was done as previously described (Smid et al., 2008) according to published classifiers (Perou et al., 1999; Sorlie et al., 2003). Of note, a non-trivial proportion of Luminal tumors cannot be unambiguously determined between Luminal A and luminal B subtypes. We therefore merged the two subtypes in some analyses.

Survival analyses were carried out using the “survival” package of R. P values were calculated by the “survdiff” command in the package, which is based on log-rank tests. When sample size is small, we also performed Fisher’s Exact Test. Kaplan-Meier curves were drawn with the survfit command in the same package. The Cox proportion hazard regression analyses were performed using the “coxph” method in the same package.

Cytokine gene expression analysis

We compiled a list of 260 cytokine genes using GO database (cytokine activity entry: GO:0005125). These cytokines were mapped to 404 probes on the Affymetrix HG-U133A platform. We screened these probes in each tissue sample for those whose intensity value was greater than the median of all genes, and statistically overexpressed in bone metastasis samples compared to other metastasis (t test with Welch’s correction, p < 0.05 with correction for multiple tests).

Tumor xenografts and analysis

All procedures involving mice were approved by the MSKCC Institutional Animal Care and Use Committee. And the details are provided in the supplementary method section.

Statistical analysis

Bone metastasis assay in BoM-1833 line has been repeated for 2 times (n=7-10 for each cohort at each time). The results were pooled and shown as Figure 3D-F. The same result has been reproduced independently later in two additional experiments. For the bone metastasis assay of CN34-BoM2 (Figure 5), lung colonization assay (Figure 3H), orthotopic proliferation (Figure 3G), intratibial growth (Figure 4A), and Dasatinib treatment (Figure 4C-D), one experiment was performed with n=10-15 in each cohort. Results are reported as mean ± SEM (standard error of the mean), as indicated in the figure legends. Comparisons between Kaplan-Meier curves were performed using the log rank test. Other comparisons were performed using unpaired two-sided t test without equal variance assumption unless otherwise specified.

Accession numbers

The raw and normalized data of breast cancer metastases have been deposited to the Gene Expression Ominbus (GEO) database (GSE14020).

Supplementary Material



This work is dedicated to the memory of our esteemed colleague and coauthor William Gerald. We would like to thank M. Resh and H. Varmus, D. Nguyen, P. Bos, D. Padua, S. Tavazoie, Q. Chen, S. Gao and Q. Xi for helpful discussions and technical suggestions, T. Faye and M. Ladanyi for clinical annotations, and M. Donepudi and R. Somwar for reagents. This work was supported by grants from the National Institutes of Health (U54 CA126518), the Kleberg Foundation and the Hearst Foundation (JM), and the Netherlands Genomics Initiative (NGI)/Netherlands Organization for Scientific Research (MS, JAF). JM is an investigator of the Howard Hughes Medical Institute.


Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.


  • Abram CL, Lowell CA. The diverse functions of Src family kinases in macrophages. Front Biosci. 2008;13:4426–4450. [PMC free article] [PubMed]
  • Ashkenazi A. Targeting death and decoy receptors of the tumour-necrosis factor superfamily. Nat Rev Cancer. 2002;2:420–430. [PubMed]
  • Ashkenazi A, Herbst RS. To kill a tumor cell: the potential of proapoptotic receptor agonists. J Clin Invest. 2008;118:1979–1990. [PMC free article] [PubMed]
  • Berry DA, Cirrincione C, Henderson IC, Citron ML, Budman DR, Goldstein LJ, Martino S, Perez EA, Muss HB, Norton L, et al. Estrogen-receptor status and outcomes of modern chemotherapy for patients with node-positive breast cancer. Jama. 2006;295:1658–1667. [PMC free article] [PubMed]
  • Bild AH, Yao G, Chang JT, Wang Q, Potti A, Chasse D, Joshi MB, Harpole D, Lancaster JM, Berchuck A, et al. Oncogenic pathway signatures in human cancers as a guide to targeted therapies. Nature. 2006;439:353–357. [PubMed]
  • Bromann PA, Korkaya H, Courtneidge SA. The interplay between Src family kinases and receptor tyrosine kinases. Oncogene. 2004;23:7957–7968. [PubMed]
  • Chen D, Zhao M, Mundy GR. Bone morphogenetic proteins. Growth Factors. 2004;22:233–241. [PubMed]
  • Chinni SR, Sivalogan S, Dong Z, Filho JC, Deng X, Bonfil RD, Cher ML. CXCL12/CXCR4 signaling activates Akt-1 and MMP-9 expression in prostate cancer cells: the role of bone microenvironment-associated CXCL12. Prostate. 2006;66:32–48. [PubMed]
  • Collins P, Webb C. Estrogen hits the surface. Nat Med. 1999;5:1130–1131. [PubMed]
  • Dehio C, Prevost MC, Sansonetti PJ. Invasion of epithelial cells by Shigella flexneri induces tyrosine phosphorylation of cortactin by a pp60c-src-mediated signalling pathway. Embo J. 1995;14:2471–2482. [PubMed]
  • Du J, Bernasconi P, Clauser KR, Mani DR, Finn SP, Beroukhim R, Burns M, Julian B, Peng XP, Hieronymus H, et al. Bead-based profiling of tyrosine kinase phosphorylation identifies SRC as a potential target for glioblastoma therapy. Nat Biotechnol. 2009;27:77–83. [PMC free article] [PubMed]
  • EBCTCG. Effects of chemotherapy and hormonal therapy for early breast cancer on recurrence and 15-year survival: an overview of the randomised trials. Lancet. 2005;365:1687–1717. [PubMed]
  • Epstein RJ. The CXCL12-CXCR4 chemotactic pathway as a target of adjuvant breast cancer therapies. Nat Rev Cancer. 2004;4:901–909. [PubMed]
  • Finn RS. Targeting Src in breast cancer. Ann Oncol. 2008;19:1379–1386. [PubMed]
  • Foekens JA, Atkins D, Zhang Y, Sweep FCGJ, Harbeck N, Paradiso A, Cufer T, Sieuwerts AM, Talantov D, Span PN, et al. Multicenter Validation of a Gene Expression-Based Prognostic Signature in Lymph Node-Negative Primary Breast Cancer. J Clin Oncol. 2006;24:1665–1671. [PubMed]
  • Guise TA, Mohammad KS, Clines G, Stebbins EG, Wong DH, Higgins LS, Vessella R, Corey E, Padalecki S, Suva L, Chirgwin JM. Basic mechanisms responsible for osteolytic and osteoblastic bone metastases. Clin Cancer Res. 2006;12:6213s–6216s. [PubMed]
  • Guo W, Giancotti FG. Integrin signalling during tumour progression. Nat Rev Mol Cell Biol. 2004;5:816–826. [PubMed]
  • Hayes DF, Thor AD, Dressler LG, Weaver D, Edgerton S, Cowan D, Broadwater G, Goldstein LJ, Martino S, Ingle JN, et al. HER2 and response to paclitaxel in node-positive breast cancer. N Engl J Med. 2007;357:1496–1506. [PubMed]
  • Hess KR, Pusztai L, Buzdar AU, Hortobagyi GN. Estrogen receptors and distinct patterns of breast cancer relapse. Breast Cancer Res Treat. 2003;78:105–118. [PubMed]
  • Horne WC, Sanjay A, Bruzzaniti A, Baron R. The role(s) of Src kinase and Cbl proteins in the regulation of osteoclast differentiation and function. Immunol Rev. 2005;208:106–125. [PubMed]
  • Hudis CA. Trastuzumab--mechanism of action and use in clinical practice. N Engl J Med. 2007;357:39–51. [PubMed]
  • Husemann Y, Geigl JB, Schubert F, Musiani P, Meyer M, Burghart E, Forni G, Eils R, Fehm T, Riethmuller G, Klein CA. Systemic spread is an early step in breast cancer. Cancer Cell. 2008;13:58–68. [PubMed]
  • Ishizawar R, Parsons SJ. c-Src and cooperating partners in human cancer. Cancer Cell. 2004;6:209–214. [PubMed]
  • Kamath AV, Wang J, Lee FY, Marathe PH. Preclinical pharmacokinetics and in vitro metabolism of dasatinib (BMS-354825): a potent oral multi-targeted kinase inhibitor against SRC and BCR-ABL. Cancer Chemother Pharmacol. 2008;61:365–376. [PubMed]
  • Kang Y, Siegel PM, Shu W, Drobnjak M, Kakonen SM, Cordon-Cardo C, Guise TA, Massagué J. A multigenic program mediating breast cancer metastasis to bone. Cancer Cell. 2003;3:537–549. [PubMed]
  • Karrison TG, Ferguson DJ, Meier P. Dormancy of mammary carcinoma after mastectomy. J Natl Cancer Inst. 1999;91:80–85. [PubMed]
  • Klapholz-Brown Z, Walmsley GG, Nusse YM, Nusse R, Brown PO. Transcriptional program induced by Wnt protein in human fibroblasts suggests mechanisms for cell cooperativity in defining tissue microenvironments. PLoS ONE. 2007;2:e945. [PMC free article] [PubMed]
  • Liang Z, Yoon Y, Votaw J, Goodman MM, Williams L, Shim H. Silencing of CXCR4 blocks breast cancer metastasis. Cancer Res. 2005;65:967–971. [PMC free article] [PubMed]
  • Lowell CA, Niwa M, Soriano P, Varmus HE. Deficiency of the Hck and Src tyrosine kinases results in extreme levels of extramedullary hematopoiesis. Blood. 1996;87:1780–1792. [PubMed]
  • Lu Y, Yu Q, Liu JH, Zhang J, Wang H, Koul D, McMurray JS, Fang X, Yung WK, Siminovitch KA, Mills GB. Src family protein-tyrosine kinases alter the function of PTEN to regulate phosphatidylinositol 3-kinase/AKT cascades. J Biol Chem. 2003;278:40057–40066. [PubMed]
  • McShane LM, Radmacher MD, Freidlin B, Yu R, Li MC, Simon R. Methods for assessing reproducibility of clustering patterns observed in analyses of microarray data. Bioinformatics. 2002;18:1462–1469. [PubMed]
  • Migliaccio A, Castoria G, Di Domenico M, De Falco A, Bilancio A, Auricchio F. Src is an initial target of sex steroid hormone action. Ann N Y Acad Sci. 2002;963:185–190. [PubMed]
  • Minn AJ, Gupta GP, Siegel PM, Bos PD, Shu W, Giri DD, Viale A, Olshen AB, Gerald WL, Massagué J. Genes that mediate breast cancer metastasis to lung. Nature. 2005;436:518–524. [PMC free article] [PubMed]
  • Mitra SK, Schlaepfer DD. Integrin-regulated FAK-Src signaling in normal and cancer cells. Curr Opin Cell Biol. 2006;18:516–523. [PubMed]
  • Muller A, Homey B, Soto H, Ge N, Catron D, Buchanan ME, McClanahan T, Murphy E, Yuan W, Wagner SN, et al. Involvement of chemokine receptors in breast cancer metastasis. Nature. 2001;410:50–56. [PubMed]
  • Mundy GR. Metastasis to bone: causes, consequences and therapeutic opportunities. Nat Rev Cancer. 2002;2:584–593. [PubMed]
  • Neve RM, Chin K, Fridlyand J, Yeh J, Baehner FL, Fevr T, Clark L, Bayani N, Coppe JP, Tong F, et al. A collection of breast cancer cell lines for the study of functionally distinct cancer subtypes. Cancer Cell. 2006;10:515–527. [PMC free article] [PubMed]
  • Padua D, Zhang XH, Wang Q, Nadal C, Gerald WL, Gomis RR, Massagué J. TGFbeta primes breast tumors for lung metastasis seeding through angiopoietin-like 4. Cell. 2008;133:66–77. [PMC free article] [PubMed]
  • Perou CM, Jeffrey SS, van de Rijn M, Rees CA, Eisen MB, Ross DT, Pergamenschikov A, Williams CF, Zhu SX, Lee JC, et al. Distinctive gene expression patterns in human mammary epithelial cells and breast cancers. Proc Natl Acad Sci U S A. 1999;96:9212–9217. [PubMed]
  • Planas-Silva MD, Bruggeman RD, Grenko RT, Stanley Smith J. Role of c-Src and focal adhesion kinase in progression and metastasis of estrogen receptor-positive breast cancer. Biochem Biophys Res Commun. 2006;341:73–81. [PubMed]
  • Rucci N, Susa M, Teti A. Inhibition of protein kinase c-Src as a therapeutic approach for cancer and bone metastases. Anticancer Agents Med Chem. 2008;8:342–349. [PubMed]
  • Schmidt-Kittler O, Ragg T, Daskalakis A, Granzow M, Ahr A, Blankenstein TJ, Kaufmann M, Diebold J, Arnholdt H, Muller P, et al. From latent disseminated cells to overt metastasis: genetic analysis of systemic breast cancer progression. Proc Natl Acad Sci U S A. 2003;100:7737–7742. [PubMed]
  • Shah NP, Tran C, Lee FY, Chen P, Norris D, Sawyers CL. Overriding imatinib resistance with a novel ABL kinase inhibitor. Science. 2004;305:399–401. [PubMed]
  • Smid M, Wang Y, Zhang Y, Sieuwerts AM, Yu J, Klijn JG, Foekens JA, Martens JW. Subtypes of breast cancer show preferential site of relapse. Cancer Res. 2008;68:3108–3114. [PubMed]
  • Sorlie T, Tibshirani R, Parker J, Hastie T, Marron JS, Nobel A, Deng S, Johnsen H, Pesich R, Geisler S, et al. Repeated observation of breast tumor subtypes in independent gene expression data sets. Proc Natl Acad Sci U S A. 2003;100:8418–8423. [PubMed]
  • Sotiriou C, Wirapati P, Loi S, Harris A, Fox S, Smeds J, Nordgren H, Farmer P, Praz V, Haibe-Kains B, et al. Gene expression profiling in breast cancer: understanding the molecular basis of histologic grade to improve prognosis. J Natl Cancer Inst. 2006;98:262–272. [PubMed]
  • Stewart CE, Rotwein P. Growth, differentiation, and survival: multiple physiological functions for insulin-like growth factors. Physiol Rev. 1996;76:1005–1026. [PubMed]
  • Stoecklein NH, Hosch SB, Bezler M, Stern F, Hartmann CH, Vay C, Siegmund A, Scheunemann P, Schurr P, Knoefel WT, et al. Direct genetic analysis of single disseminated cancer cells for prediction of outcome and therapy selection in esophageal cancer. Cancer Cell. 2008;13:441–453. [PubMed]
  • Summy JM, Gallick GE. Treatment for advanced tumors: SRC reclaims center stage. Clin Cancer Res. 2006;12:1398–1401. [PubMed]
  • Thomas SM, Brugge JS. Cellular functions regulated by Src family kinases. Annu Rev Cell Dev Biol. 1997;13:513–609. [PubMed]
  • van de Wetering M, Sancho E, Verweij C, de Lau W, Oving I, Hurlstone A, van der Horn K, Batlle E, Coudreuse D, Haramis AP, et al. The beta-catenin/TCF-4 complex imposes a crypt progenitor phenotype on colorectal cancer cells. Cell. 2002;111:241–250. [PubMed]
  • Wang Y, Klijn JG, Zhang Y, Sieuwerts AM, Look MP, Yang F, Talantov D, Timmermans M, Meijer-van Gelder ME, Yu J, et al. Gene-expression profiles to predict distant metastasis of lymph-node-negative primary breast cancer. Lancet. 2005;365:671–679. [PubMed]
  • Weckermann D, Muller P, Wawroschek F, Harzmann R, Riethmuller G, Schlimok G. Disseminated cytokeratin positive tumor cells in the bone marrow of patients with prostate cancer: detection and prognostic value. J Urol. 2001;166:699–703. [PubMed]
  • Yin JJ, Selander K, Chirgwin JM, Dallas M, Grubbs BG, Wieser R, Massagué J, Mundy GR, Guise TA. TGF-beta signaling blockade inhibits PTHrP secretion by breast cancer cells and bone metastases development. J Clin Invest. 1999;103:197–206. [PMC free article] [PubMed]
  • Yu JX, Sieuwerts AM, Zhang Y, Martens JW, Smid M, Klijn JG, Wang Y, Foekens JA. Pathway analysis of gene signatures predicting metastasis of node-negative primary breast cancer. BMC Cancer. 2007;7:182. [PMC free article] [PubMed]
  • Zander AR, Kroger N, Schmoor C, Kruger W, Mobus V, Frickhofen N, Metzner B, Schultze W, Berdel WE, Koenigsmann M, et al. High-dose chemotherapy with autologous hematopoietic stem-cell support compared with standard-dose chemotherapy in breast cancer patients with 10 or more positive lymph nodes: first results of a randomized trial. J Clin Oncol. 2004;22:2273–2283. [PubMed]