Plasma microRNA 210 levels correlate with sensitivity to trastuzumab and tumor presence in breast cancer patients

Authors

  • Eun-Jung Jung MD, PhD,

    1. Department of Experimental Therapeutics, The University of Texas MD Anderson Cancer Center, Houston, Texas
    2. Department of Surgery, School of Medicine, Gyeongsang National University, Jin-ju, South Korea
    Search for more papers by this author
  • Libero Santarpia MD, PhD,

    Corresponding author
    1. Translational Research Unit, Department of Oncology, Hospital of Prato and Tuscany Cancer Institute, Prato, Italy
    2. Department of Oncology, Sandro Pitigliani Unit, Hospital of Prato and Tuscany Cancer Institute, Prato, Italy
    • Libero Santarpia, Translational Research Unit, Department of Oncology, Hospital of Prato and Istituto Toscano Tumori, Prato, Italy

      George A. Calin, Department of Experimental Therapeutics, The University of Texas MD Anderson Cancer Center, 1881 East Road, Houston, TX 77054

    Search for more papers by this author
    • Fax: (011) 39-0574-29798

  • Juyeon Kim MD,

    1. Department of Surgery, School of Medicine, Gyeongsang National University, Jin-ju, South Korea
    Search for more papers by this author
  • Francisco J. Esteva MD, PhD,

    1. Department of Breast Medical Oncology, The University of Texas MD Anderson Cancer Center, Houston, Texas
    Search for more papers by this author
  • Erica Moretti PhD,

    1. Translational Research Unit, Department of Oncology, Hospital of Prato and Tuscany Cancer Institute, Prato, Italy
    2. Department of Oncology, Sandro Pitigliani Unit, Hospital of Prato and Tuscany Cancer Institute, Prato, Italy
    Search for more papers by this author
  • Aman U. Buzdar MD,

    1. Department of Breast Medical Oncology, The University of Texas MD Anderson Cancer Center, Houston, Texas
    Search for more papers by this author
  • Angelo Di Leo MD,

    1. Department of Oncology, Sandro Pitigliani Unit, Hospital of Prato and Tuscany Cancer Institute, Prato, Italy
    Search for more papers by this author
  • Xiao-Feng Le MD, PhD,

    1. Department of Experimental Therapeutics, The University of Texas MD Anderson Cancer Center, Houston, Texas
    Search for more papers by this author
  • Robert C. Bast Jr MD,

    1. Department of Experimental Therapeutics, The University of Texas MD Anderson Cancer Center, Houston, Texas
    Search for more papers by this author
  • Soon-Tae Park MD, PhD,

    1. Department of Surgery, School of Medicine, Gyeongsang National University, Jin-ju, South Korea
    Search for more papers by this author
  • Lajos Pusztai MD,

    1. Department of Breast Medical Oncology, The University of Texas MD Anderson Cancer Center, Houston, Texas
    Search for more papers by this author
  • George A. Calin MD, PhD

    Corresponding author
    1. Department of Experimental Therapeutics, The University of Texas MD Anderson Cancer Center, Houston, Texas
    2. Center for RNA Interference and Noncoding RNAs, The University of Texas MD Anderson Cancer Center, Houston, Texas
    • Libero Santarpia, Translational Research Unit, Department of Oncology, Hospital of Prato and Istituto Toscano Tumori, Prato, Italy

      George A. Calin, Department of Experimental Therapeutics, The University of Texas MD Anderson Cancer Center, 1881 East Road, Houston, TX 77054

    Search for more papers by this author
    • Fax: (713) 745-4526


  • Eun-Jung Jung, Libero Santarpia, Lajos Pusztai, and George A. Calin were responsible for study conception and design. Libero Santarpia, Lajos Pusztai, and George A. Calin were responsible for financial support, George A. Calin was responsible for administrative support. Libero Santarpia, Juyeon Kim, Francisco J. Esteva, Aman U. Buzdar, Angelo Di Leo, Xiao-Feng Le, Robert C. Bast, Jr., and Soon-Tae Park provided study materials and patients. Eun-Jung Jung, Libero Santarpia, Erica Moretti, Lajos Pusztai and George A. Calin were responsible for data collection and assembly. Eun-Jung Jung, Libero Santarpia, and George A. Calin were responsible for data analysis and interpretation. Eun-Jung Jung, Libero Santarpia, Lajos Pusztai, and George A. Calin wrote the article. All authors approved the final article.

  • We thank Stephanie Deming, Maude Veech, and Erica Goodoff (Department of Scientific Publications, The University of Texas MD Anderson Cancer Center) for help with the editing of this article.

Abstract

BACKGROUND:

Trastuzumab is part of the standard treatment for patients with human epidermal growth factor receptor 2 (HER-2)-positive breast cancer, but not all patients respond to trastuzumab. Altered microRNA (miR) expression levels in cancer cells have been correlated with prognosis and response to chemotherapy. The authors of this report hypothesized that altered miR expression levels in plasma are associated with sensitivity to trastuzumab in patients with HER-2 positive breast cancer.

METHODS:

Quantitative reverse transcriptase-polymerase chain reaction was used to analyze plasma samples, including samples from patients with breast cancer who were enrolled in a clinical trial of neoadjuvant trastuzumab-based chemotherapy. Expression levels of miR-210, miR-21, miR-29a, and miR-126 were analyzed according to the type of response (pathologic complete response [n = 18] vs residual disease [n = 11]). MicroRNA expression levels also were compared in trastuzumab-sensitive and trastuzumab-resistant breast cancer cells derived from BT474 cells and in an independent set of preoperative plasma samples (n = 39) and postoperative plasma samples (n = 30) from 43 breast cancer patients who did not receive any treatment.

RESULTS:

At baseline before patients received neoadjuvant chemotherapy combined with trastuzumab, circulating miR-210 levels were significantly higher in those who had residual disease than in those who achieved a pathologic complete response (P = .0359). The mean expression ratio for miR-210 was significantly higher in trastuzumab-resistant BT474 cells, and miR-210 expression was significantly higher before surgery than after surgery (P = .0297) and in patients whose cancer metastasized to the lymph nodes (P = .0030).

CONCLUSIONS:

Circulating miR-210 levels were associated with trastuzumab sensitivity, tumor presence, and lymph node metastases. These results suggest that plasma miR-210 may be used to predict and perhaps monitor response to therapies that contain trastuzumab. Cancer 2011;. © 2011 American Cancer Society.

INTRODUCTION

The human epidermal growth factor receptor 2 (HER2) is amplified in 20% to 30% of invasive breast cancers, and its amplification is associated with poor patient prognosis.1 Since the advent of treatment with the anti-HER2 monoclonal antibody trastuzumab, the survival of patients with HER2-positive breast cancer has improved significantly, and trastuzumab is now part of the standard treatment for HER2-positive breast cancer. Monotherapy or combination treatment with trastuzumab results in a 30% to 50% response rate; however, most patients develop resistance to trastuzumab within 1 year.2-5 Although 1 study revealed that combining trastuzumab with chemotherapy in a neoadjuvant setting resulted in a pathologic complete response (pCR) rate >65%, almost 35% of patients who received trastuzumab and chemotherapy still had residual disease.6

Although breast cancer can respond to chemotherapy, its sensitivity to a given drug regimen varies with each patient. The identification of patients who would benefit from specific chemotherapeutic agents before treatment could increase the proportion of cancers that respond to treatment and potentially may help patients avoid the toxicity of ineffective chemotherapy. Amplification and consequent overexpression of HER2 in breast cancer is necessary for the cancer to respond to trastuzumab, but it is apparently not the only factor involved: Approximately 70% of cancers with HER2 amplification fail to respond to trastuzumab.7 One interesting study reported that HER2 blockade (trastuzumab) improved tumor oxygenation in Her2/neu-positive tumors,7 and that study suggested that response to trastuzumab treatment correlates with tumor hypoxia. Several studies revealed that tumor hypoxia has been associated with a poor prognosis and resistance to chemotherapy and radiation therapy.8, 9 In particular, the robust induction of microRNA 210 (miR-210) in hypoxic MCF-7 cell lines in 1 study proved to be a marker for hypoxia levels in tumors.10

It is known that microRNAs (miRNAs), which are endogenous RNAs approximately 20 to 23 nucleotides in length, play important regulatory roles in animals and plants by targeting messenger RNA transcripts for cleavage or translational repression.11 The emergence of miRNAs as regulators of gene expression suggests that they may be able to serve as novel diagnostic and prognostic biomarkers for disease and as predictive biomarkers for treatment. Although studies of miRNA expression in breast cancer have suggested that some miRNAs are promising candidates for these roles, most of those studies used samples from tumor tissues,10, 12-14 which frequently are not available in the amounts necessary for detailed molecular investigation. Blood can be sampled much less invasively than tissue, and studies have reported that levels of miRNAs in blood from patients with cancer are detectable and remarkably stable.15 Many studies have sought to identify miRNA markers in plasma or serum from patients with cancer16-18; however, to our knowledge, none have examined miRNA levels in breast cancer as they relate to drug resistance.

We hypothesized that expression levels of miRNAs in plasma would be related to trastuzumab resistance in patients with HER-2-positive invasive breast cancer. To test this hypothesis, we measured expression levels of miRNAs in plasma from patients with HER2-positive breast cancer before and after neoadjuvant chemotherapy that included trastuzumab, and we analyzed the relation between mean relative miRNA expression levels and response to treatment. In a separate analysis, we also examined the relation between mean relative miRNA expression levels and the presence of breast cancer.

MATERIALS AND METHODS

Treatment Response Analysis

After obtaining Institutional Review Board approval from The University of Texas MD Anderson Cancer Center (MD Anderson) and written informed consent from participants, we collected whole blood samples from 29 consecutive patients with breast cancer who received neoadjuvant chemotherapy and trastuzumab at MD Anderson and from 28 healthy, age-matched volunteers women who were recruited from MD Anderson and served as controls. We collected the blood samples at baseline (before chemotherapy) and 24 weeks after the start of chemotherapy from patients in the treatment group, and we collected blood samples at the time of study enrollment from healthy women in the control group. All 29 patients with breast cancer in this study had histologically confirmed breast cancer and were positive for HER2 in either fluorescence in situ hybridization or immunohistochemical analysis. Relevant demographic and clinicopathologic patient data were obtained from a prospectively maintained breast cancer database (Table 1). The pCR rate for this group (defined as no evidence of residual cancer in either breast or in the axilla) was 65%, as reported previously.6 The median age of the 28 healthy volunteers was 53.5 years (range, 22-71 years), which was not significantly different from that of the treatment group (median age, 52.0 years; range, 21-70 years).

Table 1. Clinical Characteristics of Patients at the University of Texas MD Anderson Cancer Center who Received Chemotherapy (Paclitaxel Followed by Fluorouracil, Cyclophosphamide, and Epirubicin) Plus Trastuzumab
CharacteristicNo. of Patients, N = 29
  • Abbreviations: ER, estrogen receptor; FISH, fluorescence in situ hybridization; HER2 indicates human epidermal growth factor receptor 2; IHC, immunohistochemistry; PgR, progesterone receptor.

  • a

    FISH results were not available.

Median age (range), y52 (21-70)
Tumor classification 
 T13
 T220
 T36
 T40
Lymph node status 
 N010
 N119
 N20
 N30
Hormone receptor status 
 ER positive13
 PgR positive8
HER2 status 
 FISH positive28
 IHC positivea1
Response 
Pathologic complete response18
Residual disease11
Residual disease in breast, cm 
 <17
 1-31
 >32
No. of positive lymph nodes 
 1-33
 4-100
 >100

Each patient in the treatment group received 4 cycles of paclitaxel followed by 4 cycles of combined fluorouracil, epirubicin, and cyclophosphamide (FEC). Paclitaxel was administered at a dose of 225 mg/m2 as a 24-hour, continuous intravenous infusion, and cycles were repeated every 3 weeks for 4 cycles. FEC consisted of 500 mg/m2 of intravenous fluorouracil on days 1 and 4, 500 mg/m2 of intravenous cyclophosphamide on day 1 only, and 75 mg/m2 of intravenous epirubicin on day 1 only. Patients received trastuzumab at a dose of 4 mg/kg as a 90-minute intravenous infusion on day 1 of the first FEC cycle. Subsequent weekly treatments with trastuzumab were administered at a dose of 2 mg/kg as 30-minute intravenous infusions. Patients received weekly doses of trastuzumab for a total of 24 weeks.

Patients in Analysis of Tumor Presence

To analyze any potential relation between plasma miRNA expression levels and tumor presence, we prospectively collected 39 preoperative plasma samples and 30 postoperative plasma samples from a separate cohort of 43 Korean patients with breast cancer who did not receive neoadjuvant or adjuvant chemotherapy. This study was approved by the Institutional Review Board of Gyeongsang National University Hospital, and informed consent was obtained from all patients. All patients had histologically confirmed breast cancer. Plasma samples were collected preoperatively and in the second week postoperatively. Among the 43 patients, 13 tested positive for HER2 in immunohistochemical analysis (minimum score, 3+); fluorescence in situ hybridization analyses were not performed in patients whose tumors were negative for HER2 on immunostaining (Table 2).

Table 2. Clinical Characteristics of Perioperative Patients in the Korean Cohort who Did Not Receive Adjuvant or Neoadjuvant Treatment
CharacteristicNo. of Patients, N = 43
  1. Abbreviations: ER, estrogen receptor; FISH, fluorescence in situ hybridization; HER2 indicates human epidermal growth factor receptor 2; IHC, immunohistochemistry; ND, not done; PgR, progesterone receptor.

Sample group 
 Preoperative39
 Postoperative30
Median age (range), y53 (35-78)
Tumor classification 
 T130
 T211
 T32
 T40
Lymph node status 
 N035
 N11
 N25
 N32
Hormone receptor status 
 ER positive29
 PgR positive35
HER2 status 
 FISH positiveND
 IHC positive13

Plasma Collection

Up to 8 mL of whole blood were collected from each participant in an ethylene diamine tetracetic acid tube. Blood samples were centrifuged at ×1200 g for 10 minutes at 4°C to separate the blood cells, and the supernatant was transferred into microcentrifuge tubes and then centrifuged a second time at ×12,000 g for 10 minutes at 4°C to completely remove the cellular components. Plasma was aliquoted and stored at −80°C until use. Blood samples were processed and plasma was frozen within 4 hours of collection.

Establishment of a Trastuzumab-Resistant BT474 Breast Cancer Cell Clone

Wild-type BT474 cells were obtained from the American Type Culture Collection (Manassas, Va). These cells were seeded into 6-well cell-culture dishes and were treated continuously with trastuzumab (Genentech, South San Francisco, Calif) at a concentration of 10 μg/mL for 6 months. Cultures were replenished with fresh medium containing trastuzumab every week. After 6 months, the cells were tested for sensitivity to trastuzumab based on their levels of up-regulation of p27Kip1 protein and cell-cycle arrest. Individual colonies that were resistant to trastuzumab (ie, those without p27Kip1 induction and cell-cycle arrest at the G1 phase) were chosen microscopically, expanded, and rechecked for resistance to trastuzumab. Clone 65 (which we called BTR65) was the clone that exhibited the maximal resistance compared with wild-type BT474 cells. Both wild-type BT474 and BTR65 cells seeded onto 100-mm cell-culture dishes were treated once with trastuzumab 10 μg/mL for 48 hours. More details were published previously.19

RNA Extraction

Total RNA was isolated from plasma samples using the Norgen RNA Purification Kit (Norgen Biotek Corporation, Thorold, Ontario, Canada) according to the manufacturer's protocol. Briefly, lysis solution was added to 100 μL of plasma, and then ethanol was added. The lysates were then loaded onto the provided column, and most of the contaminating cellular proteins were removed as they flowed through it. The column was then washed 3 times with 400 μL of wash solution. The purified total RNA was eluted into as much as 50 μL or as little as 20 μL of elution buffer. Eluted RNA samples were quantified using a NanoDrop ND-1000 spectrophotometer (Thermo Scientific, Wilmington, Del). The mean amount of total eluted RNA in each sample was 182.39 ng (range, 111-378 ng).

Total RNA was extracted from BT474 cells and trastuzumab-resistant BTR65 cells using the TRIzol Reagent (Invitrogen, Carlsbad, Calif). The concentrations of all RNA samples were quantified using the NanoDrop ND-1000 spectrophotometer.

Quantitative Reverse Transcriptase-Polymerase Chain Reaction for Evaluation of MicroRNA Expression

We selected a group of 4 miRNAs (miR-210, miR-21, miR-29a, and miR-126) that were expressed abnormally in the initial study and in other studies that reported altered expression profiles of some miRNAs in breast cancer.12, 20-22 The miRNAs we chose also reportedly are influenced by hypoxia in breast cancer cells, and high expression levels of miR-21 have been correlated with trastuzumab resistance in breast cancer.23, 24

Expression levels of miR-210, miR-21, miR-29a, and miR-126 were detected by quantitative reverse transcriptase-polymerase chain reaction (PCR) using the TaqMan MicroRNA Assay kit (Applied Biosystems, Foster City, Calif) according to the manufacturer's instructions. Twenty nanograms of total RNA from each sample were reverse transcribed using the TaqMan MicroRNA Reverse Transcription kit (Applied Biosystems). PCR amplifications were carried out in final volumes of 10 μL using the CFX384 real-time PCR detection system (Bio-Rad, Hercules, Calif). Amplifications were initiated with 10-minute incubation at 95°C followed by 40 cycles at 95°C for 15 seconds and 60°C for 60 seconds. Each amplification reaction was performed in duplicate wells and was measured independently on 2 different days. To normalize the expression levels of miRNAs, we used U6 RNA as an internal control. The relative expression of each miRNA was calculated from the following equation: relative expression = 2− ΔCt, where Ct is the threshold cycle for a sample, and ΔCt = mean CtmiRNA − mean Ctcontrol U6. The mean relative expression levels for each miRNA were compared using 2-sided Student t tests (P < .05).

Statistical Analysis

All values are expressed as means ± standard deviations. Independent sample t tests were used to compare miRNA levels between patients with breast cancer at baseline and healthy controls, between patients who achieved a pCR and those with residual disease according to trastuzumab treatment response, and between the wild-type BT474 cell line and clone 65. Paired Korean samples were analyzed with t tests for paired data analysis. We dichotomized miR-210 levels (low and high) according to the median level for an analysis of the correlation between miR-210 expression and tumor size or lymph node status. Corresponding distribution plots were generated using GraphPad Prism 5.0 (GraphPad Software, San Diego, Calif). Statistical analyses were performed using the SPSS software package (version 16.0; SPSS Inc., Chicago, Ill). All tests were 2 sided, and differences in miRNAs expression were considered statistically significant at P < .05.

RESULTS

Expression of MicroRNAs in Plasma From Patients With Breast Cancer

First, we determined whether miR-210, miR-21, miR-29a, and miR-126 were expressed in plasma from patients with breast cancer in the treatment group (n = 29) and from healthy women in the control group (n = 28) in the MD Anderson cohort. In 2 baseline plasma samples, none of the tested miRNAs were expressed; therefore, we analyzed miRNA expressions in 27 samples at baseline and in 29 samples at 24 weeks after the start of chemotherapy. The mean Ct values for miR-210, miR-21, miR-29a, and miR-126 in the treatment group were 34.97 (range, 31.94-38.01), 26.26 (range, 23.77-28.95), 31.36 (range, 27.59-34.38), and 27.45 (range, 25.09-30.23), respectively. In the control group, the mean Ct values were 35.96, 27.89, 32.89, and 28.26 for miR-210, miR-21, miR-29a, and miR-126, respectively. These values indicate that the miRNAs were expressed in both groups. The mean Ct values for U6 (internal control) in the treatment and control groups ranged from 32.75 to 34.67, and differences in U6 Ct values between the treatment and control groups were not statistically significant (Fig. 1).

Figure 1.

Plasma microRNA (miR) levels are illustrated in patients with breast cancer. (A) Mean relative miR expression levels are illustrated in plasma samples from patients with breast cancer at baseline before neoadjuvant chemotherapy in the treatment group (Cancer) and from healthy women in the control group (Healthy controls). MiR expression levels were analyzed in 27 samples at baseline and in 29 samples at 24 weeks (2 plasma samples at baseline were excluded because they did not expressed any miR). Mean relative expression levels of miR-210, miR-21, miR-29a, and miR-126 were significantly higher in the treatment group than in the control group (P = .0119, P = .0430, P < .0001, and P < .0001, respectively). An asterisk indicates P < .05. (B) Mean threshold cycle (Ct) values for U6 RNA (internal control) are illustrated in plasma samples from patients with breast cancer at baseline before neoadjuvant chemotherapy in the treatment group and from healthy women in the control group. Mean Ct values for U6, which was used for normalization, are presented. No significant differences in mean U6 Ct values were observed between the treatment group at baseline and the control group.

Next, we determined whether mean relative miRNA expression levels differed between the treatment group and the control group. The mean relative expression levels for miR-210, miR-21, miR-29a, and miR-126 were significantly higher in baseline plasma samples (n = 27) in the treatment group than in the control group (P = .0119, P = .0430, P < .0001, and P < .0001, respectively) (Fig. 1), and the ratios of mean relative miRNA expression levels in the treatment group to mean relative miRNA expression levels in the control group (tumor:control expression ratios) were 2.42, 3.09, 4.28, and 2.82, respectively (Table 3). Thus, we detected measurably higher expression levels for the 4 miRNAs in the plasma samples from patients with breast cancer than in the plasma samples from healthy women of the same age.

Table 3. Ratios of Mean Relative MicroRNA Expression Levels in Plasma From Patients With Breast Cancera
 Ratio
miRNATumor: ControlResidual:pCR at BaselineResidual:pCR at 24 WeeksPreoperative: Postoperative
  • Abbreviations: miRNA, microRNA; pCR, pathologic complete response

  • a

    Tumor:Control indicates the ratio of mean relative miRNA expression levels in the treatment group (patients with breast cancer) at baseline to mean relative miRNA expression levels in the control group (healthy women in the MD Anderson Cancer Center cohort); Residual:pCR, mean relative miRNA expression levels in those with residual disease to mean relative miRNA expression levels in those who achieved a pCR in the MD Anderson Cancer Center cohort at baseline before neoadjuvant chemotherapy and at 24 weeks after treatment; Preoperative:Postoperative, mean relative miRNA expression levels in patients with breast cancer before surgery to mean relative miRNA expression levels in patients with breast cancer after surgery in the Korean cohort.

  • b

    P<.05.

miRNA-2102.42b2.18b0.971.53b
miRNA-213.09b1.300.960.73
miRNA-29a4.28b1.171.090.72
miRNA-1262.82b1.190.880.78

Expression Levels of MicroRNAs in Plasma From Patients With Breast Cancer Before and After Neoadjuvant Trastuzumab-Based Chemotherapy

To assess the relation between miRNA expression levels and response to trastuzumab treatment, we measured expression levels of the 4 miRNAs before and 24 weeks after the initiation of neoadjuvant chemotherapy that included trastuzumab. Among the treated patients, 18 achieved a pCR, and 11 still had residual disease after chemotherapy (Table 1). Baseline miRNA expression levels were not available in 2 samples for technical reasons; therefore, we analyzed 16 baseline samples from patients who achieved a pCR. Among 29 patients, 10 patients had residual disease in the breast, 3 patients had residual disease in lymph nodes, and 2 patients had residual disease in both sites. For miR-29a, the baseline mean relative expression level was not related to patients' ability to achieve a pCR. For miR-210, miR-21, and miR-126, the mean relative baseline expression levels were higher in samples from the residual disease group than in samples from the pCR group, but only the difference in mean relative expression levels for miR-210 reached statistical significance (P = .0359) (Fig. 2). Therefore, high baseline mean relative miR-210 expression levels were associated with resistance to treatment with trastuzumab in this set of patients.

Figure 2.

Expression levels of microRNA 210 (miR-210), miR-21, miR-29a, and miR-126 are illustrated in plasma samples from patients who had breast cancer with residual disease or who achieved pathologic complete remission (pCR) to trastuzumab combined with neoadjuvant chemotherapy. At baseline before patients received neoadjuvant chemotherapy combined with trastuzumab, expression of miR-210 was significantly higher in the residual disease group than in the pCR group (P = .0359). Baseline miR expression levels were not available for 2 samples (baseline pCR, n = 16). An asterisk indicates P < .05.

Expression of MicroRNAs in Trastuzumab-Sensitive and Trastuzumab-Resistant Cell Lines

To determine whether variations in miR-210 expression were related directly to trastuzumab resistance, we measured mean relative miR-210 expression levels in BTR65 cells (clone 65 derived from the HER2-overexpressing BT474 breast cancer cell line), which exhibited resistance to trastuzumab.19 Figure 3A indicates that the wild-type BT474 cells were sensitive to trastuzumab, as evidenced by up-regulation of the p27Kip1 protein. In contrast, trastuzumab did not induce p27Kip1 protein expression in the BTR65 clone cells (Fig. 3A).

Figure 3.

MicroRNA (miRNA) expression is illustrated in trastuzumab-sensitive and trastuzumab-resistant cells. (A) Clone 65 (BTR 65) that was used in this study exhibited resistance (no p27 up-regulation) to trastuzumab (Trast) treatment compared with wild-type BT474 cells. hIgG indicates human immunoglobulin G. (B) MiR-210, miR-21, miR-29a, and miR-126 expression ratios in BT474 cells and in trastuzumab-resistant BTR 65 cells are illustrated. The ratio of miR-210 expression (trastuzumab:control miR expression) was significantly lower in BT474s cell than in trastuzumab-resistant BTR 65 cells (P = .0075).

Next, we measured the ratio of mean relative expression levels after 48 hours of treatment with trastuzumab to mean relative expression levels with no treatment (trastuzumab:control expression ratio) for each of the 4 miRNAs in wild-type BT474 cells and BTR65 cells (Fig. 3B) For miR-21 and miR-29a, trastuzumab:control expression ratios were slightly higher in BTR65 cells than in wild-type BT474 cells, although the difference was not significant. For miR-126, the expression ratios did not differ between BT474 cells and BTR65 cells. For miR-210, the mean trastuzumab:control expression ratio was significantly higher in BTR65 cells than in BT474 cells (0.93 and 0.71, respectively; P = .0075), suggesting that high levels of miR-210, but not miR-21, miR-29a, or miR-126, indicate trastuzumab resistance.

Circulating MicroRNA 210 Levels and Tumor Presence

Finally, to understand the relation between plasma miR-210 and tumor burden in breast cancer patients, we measured the mean relative expression levels of miRNAs in preoperative samples (n = 39) and postoperative samples (n = 30) from 43 Korean patients who did not receive either neoadjuvant or adjuvant chemotherapy. Although we hypothesized that only miR-210 was related to tumor burden, we analyzed mean relative expression levels for all 4 miRNAs in the preoperative and postoperative plasma samples as well as in the 25 paired samples that were available.

First, we analyzed the mean relative expression levels of the 4 miRNAs in all samples and observed that the levels in miR-21, miR-29a, and miR-126 were slightly higher in postoperative samples than in preoperative samples (Fig. 4), although the difference was not statistically significant. Mean relative expression levels of miR-210 alone were significantly lower in postoperative samples than in preoperative samples (P = .0297).

Figure 4.

Plasma expression levels of microRNA 210 (miR-210), miR-21, miR-29a, and miR-126 are illustrated in patients with breast cancer patients surgery (Pre-op) (n = 39) and after surgery (Postop) (n = 30). None of the patients received any type of treatment for breast cancer. Only miR-210 expression was directly linked to the presence of tumor (P = .0297). U6 was used as an internal control. An asterisk indicates P < .05.

Similar findings were evident in the 25 paired samples. The mean relative expression levels of miR-210 were higher in preoperative samples than in postoperative samples in each pair (the mean relative expression levels were 0.92 and 0.58 for preoperative and postoperative samples, respectively; P = .0382). Mean relative expression levels for the other 3 miRNAs did not differ between preoperative and postoperative samples in each pair.

When we analyzed the relation between expression levels of miR-210 and clinical characteristics, the mean relative expression levels from the 39 preoperative samples were significantly higher in patients whose disease had spread to the lymph nodes than in patients whose disease had not spread to the lymph nodes (P = .0030), but there was no relation between mean relative expression levels of miR-210 and tumor size (Fig. 5). A similar analysis of the MD Anderson cohort did not reveal any relation between mean relative expression levels of miRNAs and tumor size, but the same positive trend was identified, although it was not statistically significant, for the association with lymph node status.

Figure 5.

Correlations between microRNA 210 (miR-210) expression levels are illustrated for (A) tumor size and (B) lymph node involvement. Samples from patients in the preoperative Korean cohort who had positive lymph nodes had significantly higher levels of miR-210 (P = .0030), but variations in tumor size were not significant. The expression of miR-210 at baseline in samples from the MD Anderson Cancer Center (MDACC) cohort also differed according to lymph node status, but the difference was not significant. LN(−), lymph node negative; LN(+), lymph node positive.

DISCUSSION

The current results suggest that high relative expression levels of miR-210 in plasma from patients with breast cancer are associated with trastuzumab resistance and the presence of tumor; furthermore, miR-210 expression levels were associated significantly with lymph node involvement in the preoperative group of Korean patients. We also produced evidence of this in a nonclinical setting, in which mean relative miR-210 expression levels were higher in cells from a trastuzumab-resistant cell line than in trastuzumab-sensitive cells. To our knowledge, this is the first time a link has been identified between trastuzumab sensitivity and relative expression levels of miR-210 in patient plasma.

The identification of patients who may benefit from chemotherapeutic agents is of great importance, because the individualized selection of treatment may maximize treatment benefit and minimize patient exposure to the adverse effects of ineffective therapy. Therefore, large-scale studies of the relation between relative miRNA expression levels and chemotherapy response are needed. In addition, as mentioned above, most studies of chemotherapy response-specific miRNAs have measured miRNA levels in tumor tissues or cell lines rather than in plasma.10, 12-14 However, frequently, tumor tissues are not available in the amounts necessary for detailed molecular investigation, and blood can be sampled much less invasively than tissue. Thus, investigations focusing on relative miRNA expression levels in plasma are particularly warranted.

Several studies have focused on the identification of miRNAs linked to the acquisition of a resistant phenotype in cancer cell lines. An investigation in cancer cell lines by Blower et al established the influence of 3 cancer-related miRNAs (let-7i, miR-16, and miR-21) on anticancer drug sensitivity. When testing 14 different anticancer compounds, those authors observed that increased levels of miR-21 reduced the efficacy of nearly half the tested anticancer compounds.25 Meng et al demonstrated that miR-21, miR-141, and miR-200b were highly overexpressed in malignant cholangiocytes and that inhibition of miR-21 and miR-200b increased the sensitivity of cholangiocarcinoma cells to gemcitabine.26

A recent article reported that high expression levels of miR-21 were associated with trastuzumab resistance in breast cancer.23 In our current study, instead, we observed that plasma mean relative expression levels of miR-210 only, and not miR-21, were related to trastuzumab sensitivity. However, we were able to confirm our findings by comparing the mean relative expression levels of each miRNA in a trastuzumab-resistant cell line with the expression levels in a trastuzumab-sensitive cell line.

Our finding of an association of miR-210 with tumor presence and drug resistance is not surprising given the results from previous studies on miR-210 expression. Recent clinical studies have reported higher expression levels of miR-210 in patients with breast cancer, head and neck cancer, and pancreatic cancer than in healthy controls and that high expression levels were associated with a poor prognosis.10, 27-29 A study of miRNAs as prognostic markers in pancreatic cancer revealed that the overexpression of miR-210 and of 3 other miRNAs (miR-155, miR-201, and miR-222) was associated with shorter survival.27 In fact, many reports on different types of cancer tissues and blood have revealed that the overexpression of miR-210 is associated with a poor prognosis.10, 28-30 In a study of plasma from cancer patients, Ho and colleagues reported that circulating miR-210 levels may serve as a diagnostic marker in pancreatic cancer.28

However, miR-210 is not always overexpressed in cancer. Mean expression levels of miR-210 are lower in esophageal squamous cell carcinoma tissues and cell lines than in healthy tissues, and this down-regulation is associated with poor differentiation.31 These results suggest that expression patterns of miR-210 are specific to tumor type. In our current study, we observed that mean expression levels of all miRNAs tested—miR-210, miR-21, miR-29a, and miR-126—were significantly higher in plasma from patients with breast cancer than in plasma from healthy women. However, when we compared plasma mean expression levels of these miRNAs in preoperative breast cancer patients with those in postoperative breast cancer patients, miR-210 expression alone was significantly lower after tumor resection, suggesting that expression levels of miR-210 may be related directly to the presence of breast tumor.

We also discovered that, in addition to predicting the presence of tumor, high expression levels of miR-210 were associated with positive lymph nodes in preoperative Korean patients and had the same trend (although it was not statistically significant) in the MD Anderson cohort at baseline. This can be explained by the smaller size of the MD Anderson cohort; and, although these results were obtained from a limited number of patients with different tumor characteristics, the results were consistent and substantial. However, studies with large-scale cohorts are needed to determine how well miR-210 may perform as a predictive marker. In our small cohort, the finding that mean expression levels of miR-210 were associated with tumor presence but not with tumor size also suggests that the expression levels potentially may indicate the early presence of a tumor, but this finding also needs further study.

The function of miR-210, which is directly regulated by hypoxia-inducible factor 1-alpha, also may depend on cancer type. MiR-210 inhibits apoptosis, bypasses cell-cycle arrest, and promotes cancer cell survival when it is overexpressed; however, but when it is under expressed, as it is in esophageal squamous cell carcinoma, it represses the initiation of tumor growth by inducing cell death and cell-cycle arrest.31-33 To date, known targets of miR-210 in cancer include the receptor tyrosine kinase ligand ephrin-A3, EF2 transcription factor 3 (E2F3), the DNA repair enzyme RAD52, and fibroblast growth factor receptor-like 1.34 MiR-210 also may interact with genes involved in the trastuzumab-resistance pathway. We searched miR-210 targets using RNA structure prediction software, including RNA22, miRanda, TargetScan, and PITA, and identified greater than 5300 targets; among these targets, we identified well known molecular targets for trastuzumab resistance, including MET, insulin-like growth factor 1 receptor, and membrane-associated mucin 4.35-37 Functional studies beyond the scope of this article should be performed to clarify the biologic role of miR-210 in the trastuzumab-resistance pathway and to determine whether this miRNA also may act after potential release from cancer cells.38

In summary, our results indicate that high mean expression levels of miR-210 in plasma are associated with the presence of tumor in patients with breast cancer and with trastuzumab resistance in patients with HER2-positive breast cancer. Although these results were obtained from small cohorts, they provide an important basis for larger, prospective, multi-institutional studies to investigate the potential role of plasma miRNAs as prognostic, diagnostic, and therapeutic markers of invasive breast cancer.

FUNDING SOURCES

E.-J.J. is supported in part by a National Research Foundation of Korea grant funded by the Korean government (NRF-2010-013-E00022). G.A.C. is supported as a Fellow at MD Anderson Research Trust, as a University of Texas System Regents Research Scholar, and by the CLL Global Research Foundation. Work in Dr. Calin's laboratory is supported in part by a Department of Defense Breast Cancer Idea Award, a Developmental Research Awards in Breast Cancer Specialized Program of Research Excellence (SPORE), a 2009 Seena Magowitz-Pancreatic Cancer Action Network American Association of Cancer Research pilot grant, MD Anderson Cancer Center's Support Grant CA016672, the Laura and John Arnold Foundation and the RGK Foundation. L.S. is supported by the Associazione Italiana per la Ricerca sul Cancro (AIRC grant 6251) and the Sandro Pitigliani Foundation. X.F.L. is supported by the Anne and Henry Zarrow Foundation.

CONFLICT OF INTEREST DISCLOSURES

Dr. Di Leo declares honoraria for participation in advisory boards and as speaker at satellite symposia organized by Roche and by GlaxoSmithKline. The other authors made no disclosures.

Ancillary