There is increasing evidence supporting microRNA (miRNA) analysis for breast cancer diagnosis, prognosis and therapy [1-3]. miRNAs are ~22 nucleotide long RNA molecules encoded in the genome that can control gene expression. Their expression levels vary greatly among species and tissues [4]. Over 700 miRNAs have been identified from human tissue. Some of these short, endogenous non-protein coding miRNAs are linked to human malignancy and are differentially expressed in human cancer vs. matched normal tissue [5]. For example, a recent publication identified 29 miRNAs that were aberrantly expressed in primary breast cancer tissue. Of the 29, miR-10b, miR-125b, and miR-145 were down-regulated, while miR-21 and miR-155 were up-regulated [6]. Depending on the genes that the miRNAs act on, they may serve to stimulate or suppress tumor formation and growth. Another team reported 15 miRNAs that are expressed in a variety of normal human tissues. Among them, miR-16 is expressed at a higher level than the remaining 14 miRNAs [7]. These differences provide the molecular basis and rationale for the identification of miRNAs in breast cancer detection.

Cancer-related RNAs are present and detectable in serum [8-11], and it was recently reported that the levels of cyclin D1 and thymidylate synthase levels of free mRNA in plasma correlated with nonresponse to breast cancer therapy [12]. As RNase is present in the blood of healthy and cancer containing individuals [13], one would expect all circulating RNA (including miRNAs) in the blood to be rapidly destroyed. However, accumulated data suggest that circulating RNA may not be as fragile as was previously assumed. Tsui et al. reported that RNA in serum is more RNase resistant and stable than previously thought [14]. For example, tyrosinase mRNA has been found in the serum of melanoma patients but not in healthy individuals [9], and epithelial mRNA in the plasma of breast cancer patients was associated with poor prognosis [15]. Both observations demonstrate that cancer-related mRNA exists in the circulation and is detectable. Finally, a recent pilot study demonstrated that miRNAs can be detected in the serum of ovarian cancer patients [16]. Our hypothesis was that circulating miRNAs are also present and detectable in the serum of breast cancer patients.

Early detection is a major factor contributing to the 2.3% annual decline in breast cancer death rates over the past 10 years [17]. Nonetheless 40,480 women in the U.S. were projected to die from breast cancer in 2008 [18], in part because currently available breast cancer screening tools such as mammography and breast examination miss 10–40% of early breast cancers and are least effective in detecting cancer in young women, whose tumors are often more aggressive. An invasive needle or surgical biopsy must be performed when an area of suspicion is identified in order to confirm, by cytologic or histologic evaluation, the presence of malignancy, even though 66–85% of abnormalities are benign [19].

In principle, there are three different ways in which biomarkers can be used. First, they can be used to differentiate normal from diseased states. Second, they can be used to determine into which prognostic groups subjects should be placed. Finally, biomarkers can be used to monitor response to therapy. The identification of biologic markers such as miRNA species collected noninvasively in the circulation that would distinguish between women with or without breast cancer is therefore of crucial importance in early cancer detection, to improve disease free and overall survival from the disease. There is a recent report which outlines the usefulness of miRNA signatures to distinguish normal from diseased tissues and tissue origin, relying upon miRNA signatures from formalin fixed, paraffin embedded and frozen tissue specimens [20], demonstrating the stability of miRNAs. Our miRNA findings in archival serum support and extend these findings, with the potential that miRNA analysis may prove useful for noninvasive breast cancer detection.

We first tested if spiked-in cel-miR-39 and has-miR-145 were detectable in a 10 year old human serum and a goat serum specimen. The specimen was divided into five samples of equal volume. Decreasing inputs of single-strand cel-miR-39 and has-miR-145 (10⁵, 10³, 10, 10^-1 and 0 pg) added to each sample (Figures 1 and 2) led to the corresponding mean Ct values (19.2, 25.8, 32.4, 37.1 and 38.7), respectively, with similar values for cel-39 and has-145. The Ct value of 37.1 obtained using the most dilute sample (10^-1 pg) was slightly lower than background of 38.7, indicating that the level of endogenous miR-145 in the human serum to which the synthetic single stranded miR-145 mimic was spiked in was very low, and therefore unlikely to have influenced the amplification of the synthesized miR-145. The equidistant curves down to a concentration of 10^-1pg/400 μL serum suggest linear recovery (Figure 1) and efficient amplification (Figure 2, slope = -3.0) of the added miRNAs. The Ct values (Figure 3) for endogenous GAPDH mRNA in the serial serum samples to which has-miR-145 was spiked in were approximately 35. The results demonstrate that circulating RNA, including endogenous GAPDH mRNA and spiked-in cel-miR-39 and has-miR-145, can be efficiently extracted and amplified from serum.

We then performed a pilot study in serum specimens from 21 women with and without breast cancer (Table 2) to assess the expression of miR-16, -145 and -155 using 18s rRNA to normalize the relative expression of the miRNAs in archived serum specimens from women with and without breast cancer. Each miRNA and 18s rRNA was isolated and amplified efficiently in all subjects. The expression of all three miRNAs in serum was similar in samples from healthy women compared to those with breast cancer, unlike a prior report which found differences in the expression of mi-R-145 and -155 miRNAs in breast cancer tissue vs. benign tissue women [6]. On the other hand, progesterone receptor (PR, p = 0.016) positive tumors had higher miR-155 expression than tumors that were negative for this receptor. miR-155 has been classified as a multifunctional miRNA, having an important role in both normal and pathologic processes in immunity, inflammation, cancer and cardiovascular disease [21]. The significance of our finding regarding miR-155 and the PR will require a larger study.

In determining the proper endogenous control for qPCR, we initially tried three small non-coding nucleoar RNAs (Rnu44, Rnu48 and Rnu66). Unfortunately, none were useful because we could not detect them. GAPDH was also tested but relatively high Ct values were observed, so 18s rRNA was tried levels adequate to serve as a control in all human serum samples. We confirmed that our techniques yielded adequate RNA extraction efficiency by measuring levels of spiked-in cel-miR-39 and has-miR-145. The spiked-in miRNAs had the advantage of lacking homology to and thereby interference from endogenous human miRNAs.

The efficient extraction and accurate identification of miRNAs from serum represents a key first step toward the development of a noninvasive, blood-based detection test for breast cancer, and we believe represents an important advancement in achieving this goal.

ER: estrogen receptor; mi-, m- and rRNA: micro-, messenger and ribosomal ribonucleic acid; Rnu: nucleolar RNA; qPCR: quantitative polymerase chain reaction; PR: progesterone receptor; RT: reverse transcription.

The authors declare that they have no competing interests.

All authors read and approved the final manuscript. WZ and WQ helped conceive the project, designed and performed the experiments, UA helped design the project, provided scientific and technical insight and manuscript review, and ERS provided the samples, helped conceive the project, prepared the manuscript, provided scientific and clinical input, and submitted the manuscript.