Correlation of Breast Cancer Axillary Lymph Node Metastases With Stem Cell Mutations

Importance  Mutations in oncogenes AKT1, HRAS, and PIK3CA in breast cancers result in abnormal PI3K/Akt signaling and tumor proliferation. They occur in ductal carcinoma in situ, in breast cancers, and in breast cancer stem and progenitor cells (BCSCs).

Objectives  To determine if variability in clinical presentation at diagnosis correlates with PI3K/Akt mutations in BCSCs and provides an early prognostic indicator of increased progression and metastatic potential.

Design, Setting, and Participants  Malignant (BCSCs) and benign stem cells were collected from fresh surgical specimens via cell sorting and tested for oncogene mutations in a university hospital surgical oncology research laboratory from 30 invasive ductal breast cancers (stages IA through IIIB).

Main Outcomes and Measures  Presence of AKT1, HRAS, and PIK3CA mutations in BCSCs and their correlation with tumor mutations, pathologic tumor stage, tumor histologic grade, tumor hormone receptor status, lymph node metastases, and patient age and condition at the last follow-up contact.

Results  Ten tumors had mutations in their BCSCs. In total, 9 tumors with BCSC mutations and 4 tumors with BCSCs without mutations had associated tumor present in the lymph nodes (P = .001).

Conclusions and Relevance  Tumors in which BCSCs have defects in PI3K/Akt signaling are significantly more likely to manifest nodal metastases. These oncogenic defects may be missed by gross molecular testing of the tumor and are markers of more aggressive breast cancer. Molecular profiling of BCSCs may identify patients who would likely benefit from PI3K/Akt inhibitors, which are being tested in clinical trials.

In stem cell theory, breast cancer stem and progenitor cells (BCSCs) are central to cancer proliferation, metastatic potential, and outcome measures. Breast cancer stem and progenitor cells have the capacity to initiate tumors, produce multiple tumor cell lineages, and maintain a continuous population of malignant stem cells through self-renewal.^1,2 Although it is unknown exactly how malignant stem cells arise, the function of benign stem cells is to maintain normal tissues and organs. Following malignant transformation, they likely retain their stem cell qualities and acquire malignant abilities through the acquisition of cancer-associated genetic mutations.

Breast cancer stem and progenitor cells constitute a small percentage of the total number of cells within a tumor. However, experiments in murine models have shown that 200 cancer stem cells can form tumors when transplanted into the cleared mammary fat pad of transgenic mice.^3,4 In contrast, more than 10 000 nonspecific tumors cells are required to create the same result.^5-7 These qualities of self-renewal, tumor cell lineage propagation, and remodeling of surrounding tissues are required to sustain primary and metastatic disease.⁸

Findings from previous murine and human mammary tissue studies^4,7 suggest that benign breast stem and progenitor cells can be distinguished from malignant stems cells based on CD44, CD49f (or CD29), and CD24 cell surface markers. Neither benign nor malignant stem cells express endothelial or leukocyte markers CD31 and CD45. However, it has been shown that benign and malignant breast tissues contain positively and negatively expressing CD44, CD24, and CD49f cells.⁹ Another way to distinguish benign from malignant stem cells is genetic-based analysis to detect cancer-associated changes. Previous research found no oncologic abnormalities in stem and progenitor cells in benign breast tissue; however, oncogenes in the PI3K/Akt signaling pathway were identified in stem and progenitor cell populations in breast cancer.⁹

The PI3K/Akt pathway is implicated in cell proliferation, signaling, and metastatic potential and is a target for chemotherapeutic agents now in clinical trials.^10-12 PI3K is a transmembrane protein that when activated by cellular growth signaling molecules results in initiation of cellular proliferation through phosphatidylinositol 3,4,5-trisphosphate and Akt^13,14 (Figure 1). Mutations in proteins along the PI3K/Akt signaling pathway have been found in colon, pancreatic, and lung cancer, and it is the most commonly mutated pathway in breast cancer.^10,15,16 The PIK3CA gene encodes the catalytic subunit of the PI3K enzyme. Activating mutations result in increased initiation of the downstream cascade.^17-19 Previous findings indicate that PIK3CA mutations are present in 30% to 40% of breast tumors.¹⁵ Mutations have been associated with AKT1 (OMIM 164730) (4%-8% of tumors) and less frequently with HRAS (OMIM 190020) and PTEN (OMIM 601728).^10,20,21 Mutations of PIK3CA (OMIM 171834) cluster in “hot spots,” with the 3 most common mutations at E5542K, E545K, and H1047R, although additional oncogenic mutations have been identified.^10,17 Mutations in the pathway have been found in BCSCs, but the clinical implications have not been explored.⁹

We propose that differences in tumor behavior and clinical outcomes may be due to the genetic differences in BCSCs that are present in breast cancers.^22,23 We hypothesized that breast cancers with stem and progenitor cell mutations in genes of the PI3K/Akt signaling pathway would be associated with more aggressive breast cancers. To test this hypothesis, we correlated the mutation status of genes in the PI3K/Akt signaling pathway in malignant BCSCs with tumor pathologic features and early clinical outcomes among patients with breast cancer.

This study was approved by the Oregon Health & Science University institutional review board. Women with invasive ductal carcinoma of the breast exceeding 1.0 cm were identified at tumor board meetings and enrolled in the study. Solid-tissue breast specimens were collected at the time of mastectomy or lumpectomy before any adjuvant treatment. Specimens were obtained directly from the operating room and evaluated by a pathologist, and approximately 1 g of tumor tissue was immediately transferred to the laboratory for processing in mammary epithelial cell–specific medium (Epicult; StemCell Technologies). Samples were minced and placed in a solution containing 50% mammary epithelial cell–specific medium, 50% fetal bovine serum, and 6% dimethyl sulfoxide (ATCC) and cryopreserved at −80°C for a variable period ranging from several weeks to months. Samples were thawed, the cryopreserving fluid was removed by centrifuge, and minced tissue was digested in mammary epithelial cell–specific medium containing collagenase and hyaluronidase overnight. Cells were tested for viability with trypan blue, counted, and were labeled with fluorochrome-conjugated monoclonal antibodies against human CD45 and CD31 (fluorescein isothiocyanate conjugated), CD24 (phycoerythrin), CD49f (phycoerythrin–cyanine 5, and CD44 (phycoerythrin–cyanine 7). Isotype control testing indicated no nonspecific binding. Subpopulations were separated based on surface antibody labeling and collected by discriminatory gating. The CD31⁺ and CD45⁺ endothelial cells and leukocytes were removed leaving lineage negative cells. Cells were sorted into the following 4 lineage-negative populations: CD49f⁺CD24⁺, CD49f⁺CD24⁻, CD49f⁻CD24⁺, and CD49f⁻CD24⁻. The frequency of CD44 expression was evaluated for these cell populations in most patients. Sorted cell populations then underwent whole genomic amplification (REPLI-g Mini; Qiagen). Following amplification, samples of 10 ng of whole genomic DNA were screened for 410 mutations in 30 human oncogenes using an array system (MassARRAY; Sequenom Inc). As previously described,²⁴ this protocol involves polymerase chain reaction amplification of sequences of interest, followed by primer extension and mass spectrometry (matrix-assisted laser desorption ionization–time-of-flight mass spectrometry).

Following tissue analysis, medical records were reviewed, and additional data were compiled retrospectively for each patient. Data collected included patient age at diagnosis, race/ethnicity, tumor hormone receptor status, ERBB2 (formerly HER2/Neu) status, tumor histologic grade, pathologic tumor stage, and condition at the last follow-up contact. χ² Test and Fisher exact test were used for statistical analyses. t Test was used to compare parametric data between groups.

Thirty invasive ductal breast carcinomas were obtained. The characteristics of breast tumors, including patient age at diagnosis, tumor size, and tumor hormone receptor status, are summarized in Table 1. The mean follow-up time was 22 months after diagnosis. One patient was lost to follow-up contact. The most common tumor histologic grade was 2. Six patients (20%) had positive lymph nodes, and an additional 6 patients had micrometastatic deposits (<0.2-cm metastatic focus of tumor) identified in lymph nodes. The correlation between tumor size and macroscopic lymph node metastases did not achieve statistical significance (P > .05). All patients with macroscopic lymph node metastases underwent completion axillary lymph node dissection. Patients with only micrometastatic disease did not.

Among 30 tumors, 10 tumors (33%) had BCSCs with AKT1, HRAS, or PIK3CA mutations, 8 of which have been previously reported.⁹ Three different mutations (E545K, N345K, and H1047R) were detected in PIK3CA, a single mutation was detected in AKT1, and a single mutation was detected in HRAS (Table 1). PIK3CA G1049R (rs1219132) is considered a normal variant and was found in the lineage-negative CD49f⁺CD24⁺ and lineage-negative CD49f⁻CD24⁺ cells of one tumor. A subset of the tumors was specifically assessed for CD44 positivity, which varied based on BCSC population. In the CD49f⁺CD24⁺ sorted cell population of 20 of 26 patients, greater than 85% of the cells were also CD44⁺. In CD49f⁺CD24⁻, CD49f⁻CD24⁺, and CD49f⁻CD24⁻ populations, expression was variable among patients, with a mean of 60% (range, 20% [4 of 20 patients] to 90% [9 of 20 patients]) of cells being CD44⁺. Breast cancer stem and progenitor cells with and without mutations were assessed for CD44 positivity, and no significant difference was observed between the 2 groups.

When the presence of any BCSC mutation correlated with patient and breast cancer characteristics, no statistically significant correlations were found with patient age at diagnosis, tumor size, tumor histologic grade, estrogen receptor expression, progesterone receptor expression, or ERBB2 status (Table 2). However, a statistically significant correlation was observed between the presence of BCSC mutations and axillary lymph node metastases (P = .02). This significance was more pronounced when micrometastatic disease was included (P = .001) (Figure 2).

Three of 10 patients with BCSC mutations experienced disease progression after diagnosis following indicated chemotherapy, hormone therapy, and a trastuzumab regimen. Two patients died of disease, and 1 patient has brain metastases. No patients with BCSCs without mutations have evidence of disease. At the time of writing, patients with BCSC mutations had been followed up for a mean of 29 months, while patients without BCSC mutations had been followed up for a mean of 19 months. The difference in the mean follow-up periods was statistically significant (P = .001).

Patients with tumors in which BCSCs have a genetic abnormality of the PI3K/Akt signaling pathway are significantly more likely to have lymph node metastases. While axillary lymph node metastases are known to correlate with tumor size, BCSC mutation in this study was an independent predictor of lymph node metastasis. Because 4 of 20 patients (20%) without BCSC mutations had axillary lymph node metastases, a PI3K/Akt mutation in BCSCs may not be a requirement for axillary lymph node metastases. However, a significant correlation was found between the 2 factors, with 9 of 10 patients (90%) with BCSC mutations having nodal metastases. Five of 10 patients (50%) with BCSC mutations had axillary lymph node macrometastatic disease, and an additional 4 of 10 patients (40%) had micrometastatic disease. Micrometastatic disease in lymph nodes is of uncertain prognostic significance.^25,26 Given the link between PI3K and metastatic potential, it could be that micrometastases harboring PI3K/Akt mutations may carry a different risk for distant metastatic disease. Longer patient follow-up periods and a larger sample size will determine if this subset of patients demonstrates an increased risk and may benefit from specifically designed use of adjuvant chemotherapy.

In the present study, we showed that tumors in which BCSCs have PI3K/Akt mutations exhibit variable hormone receptor expression, tumor histologic grade, and other tumor characteristics, suggesting that these early BCSC mutations do not restrict the type of breast cancer that develops. This finding has also been observed in murine investigations in which PIK3CA mutations in luminal cells produced heterogeneous tumors.²⁷ In additional support of this theory, similar PIK3CA and AKT1 mutations are found in ductal carcinoma in situ.^20,21 Their detection in both ductal carcinoma in situ and BCSCs suggests that they are present in the early development of breast cancer and in some cases may be responsible for promoting early breast cancer development.^21,28-30 The retention of BCSCs along with PI3K/Akt mutations in tumors years after tumor initiation indicates that these cells continue to have an important role in tumor architecture and function. With regard to ERBB2, evidence suggests that PIK3CA mutations contribute to resistance to trastuzumab in ERBB2-positive breast cancers.³¹ In this study, an insufficient number of ERBB2-amplified tumors were examined to discern if there is a correlation between BCSC abnormalities and axillary lymph nodal metastases in this specific group of tumors.

PI3K/Akt signaling abnormalities also did not correlate with patient age at diagnosis. Unlike BRCA1 and BRCA2 mutations, which are associated with early-onset breast cancer, PI3K/Akt abnormalities are not germline mutations but rather somatic mutations. Accordingly, it is not unexpected that they have a later onset than the breast cancers associated with BRCA mutations.

The prognostic significance of specific PI3K/Akt signaling pathway mutations in breast cancers remains controversial.^32-34 It is likely that the variability of these mutations, the heterogeneity of the tumors, and the complexity of the pathway contribute to this conflicting evidence. Our findings in BCSCs are consistent with studies^10,15,31-34 showing that PIK3CA and AKT mutations in breast cancers are associated with factors that may indicate poor prognosis and decreased survival rates. However, other studies^32,35,36 have shown improved disease-free survival rates, lower tumor histologic grades, and increased rates of estrogen receptor positivity in patients with tumors bearing PIK3CA mutations. Our study findings indicate that the answer to this controversy may lie in identifying mutations in BCSCs, as well as mutations in the tumor as a whole.

The collection of BCSCs from fresh surgical specimens, performance of molecular analyses, and subsequent correlation with clinical outcomes support embarking on a new way of approaching breast cancer diagnosis and treatment planning. The results of this study support concomitant evaluation of BCSCs along with assessment of the breast cancer overall. The analysis of BCSCs can generate specific information about tumor growth and metastatic potential that may not be obtained from analysis of the tumor progeny cells alone. Simultaneous molecular analyses of both the tumor and BCSCs may better identify patients who are likely to benefit from specific therapeutic regimens. Similarly, simultaneous BCSC and tumor analysis may increase the number of patients who might benefit from treatment but be missed by tumor analysis alone. For example, PI3K/Akt signaling pathway inhibitors now being tested in clinical trials may prove beneficial to patients with BCSC mutations even if genetic analysis of the accompanying tumors demonstrates no PIK/Akt mutation. The use of BCSC-specific and tumor-targeted chemotherapeutic agents may prove to be synergistic with each other, providing a novel therapeutic approach. Cancer stem cell therapeutics is an area of rapidly expanding knowledge. Future studies with larger cohorts, more outcomes data, and longer follow-up periods will allow us to more critically evaluate the significant early results reported in this study.

Accepted for Publication: May 9, 2013.

Corresponding Author: Cory A. Donovan, MD, Division of Surgical Oncology, Department of Surgery, Oregon Health & Science University, Mail Code L619a, 3181 SW Sam Jackson Park Rd, Portland, OR 97239-3011 (donovan@ohsu.edu)

Published Online: July 24, 2013. doi:10.1001/jamasurg.2013.3028.

Author Contributions:Study concept and design: Donovan, R. Pommier, S. Pommier.

Acquisition of data: All authors.

Analysis and interpretation of data: Donovan, R. Pommier, O’Neill, Alabran, Vetto, S. Pommier.

Drafting of the manuscript: Donovan, R. Pommier, Vetto, S. Pommier.

Critical revision of the manuscript for important intellectual content: All authors.

Statistical analysis: Donovan, R. Pommier, Alabran, S. Pommier.

Obtained funding: S. Pommier.

Administrative, technical, and material support: Schillace, O’Neill, Muller, Alabran, Hansen.

Study supervision: R. Pommier, Schillace, Murphy, Vetto, S. Pommier.

Conflict of Interest Disclosures: None reported.

Funding/Support: This study was supported by the Janet E. Bowen Foundation (Dr S. Pommier).

Previous Presentation: This study was presented at the 84th Annual Meeting of the Pacific Coast Surgical Association; February 19, 2013; Kauai, Hawaii, and is published after peer review.