Identification and Validation of Circulating MicroRNAs Distinguishing Metastatic from Local Breast Cancer
Peter McAnena1, K Tanriverdi2, C Curran1, J Freeman2, K Gilligan1, AJ Lowery1, J AL Brown1, MJ Kerin1
1 - Discipline of Surgery, National University of Ireland, Galway, Republic of Ireland
2 University of Massachusetts, MA, USA
Despite considerable recent advances in both diagnosis and treatment, 20-30% of breast cancer patients will develop distant metastatic disease. The risk of developing metastatic disease is determined by the initial stage as well tumour subtype. Luminal A is the most common subtype comprising up to 60% of all breast cancers . Bone is the most frequent site of metastasis among all subtypes, with 3.1% of those with luminal A breast cancer developed bone metastasis. The search for non-invasive biomarkers capable of augmenting conventional diagnostic and prognostic modalities in metastatic breast cancer is a priority in the era of individualised treatment regimens.
Mi(cro)RNAs are small, non-coding RNAs that regulate gene expression by targeting messenger RNA. Over 4,000 miRNAs have been described and it is estimated that they regulate up to 30% of all human genes . MiRNAs can operate as tumour-suppressors or as tumour-promoters and their dysregulation is intricately linked to cellular processes involved in the metastatic cascade. MiRNAs are stable in circulation and can be quantified relatively simply and inexpensively by qRT-PCR and show great promise in contributing to the diagnosis, prognosis, evaluation of response to therapy and treatment of breast cancer.
The aim of our study was to identify and validate circulating miRNAs capable of distinguishing metastatic breast cancer from locally confined breast cancer. This was a collaborative project in conjunction with UMASS Boston
Samples were prospectively collected from 2008-2015. The discovery cohort (n=8) included four patients with locally confined Luminal A breast cancer and four patients with metastatic disease to bone. RNAseq was performed at UMASS to identify miRNAs dysregulated between the two groups. 712 miRNAs were analysed with 12 showing significant dysregulation. 5 miRNAs were selected for validation on an independent validation cohort (n=73), including 21 healthy controls.
Discovery Cohort: RNAs was isolated from plasma samples using a Qiagen miRNeasy Serum/Plasma Kit . Small-RNA libraries were constructed using NEXTflex Small RNA Sequencing Kit. Small-RNAseq reads were processed and quantified using the exceRpt tool available on the Genboree Workbench. ExceRpt incorporates several modifications to existing analysis methods used to assess cytosolic microRNAs (miRNAs) that specifically address experimental issues pertinent to exRNA profiling.
Validation Cohort: Quantification of selected dysregulated miRNAs (mir-331, mir-329, mir181a, mir4433, mir6734 and mir-195) was determined by RQ-PCR using TaqMan miRNA assays. PCR reactions were performed in triplicate in final volumes of 10 µl on 96 well plates. Each plate included an inter assay control (IAC) to account for run-to-run variation. Raw fluorescence (cycle threshold, CT) data were subsequently calculated. High CT values indicated low miRNA expression and vice versa. CT values were scaled to highest expressing sample and normalized to mir-16 and mir-425.
Expression of mir-331 was significantly higher in the metastatic group (n=22) compared to both the local group (n=31) and the healthy control group (n=21), (p<0.001 and p<0.001, ANOVA and post-hoc Tukey analysis), corresponding to an average fold-change of 2.58 and 2.94 respectively. There was no significant difference in mir-331 expression between the local group and the control group (p=0.825).
Expression of mir-195 was significantly lower in the metastatic group compared to both the local and the healthy control groups (p<0.001 and p=0.043, average fold-change of 0.6 and 0.73 respectively). There was no significant difference in mir-195 expression between the local and healthy control groups (p=0.087).
A logistic regression was performed to ascertain the potential of circulating miRNAs in distinguishing metastatic from local disease. Analysing each individual miRNA and combination of miRNAs, we compared the area under the curve (AUC) produced from receiver operator characteristic (ROC) curve generation using binary logistic regression. The highest AUC of 0.902 was achieved combining mir-331 and mir-195, providing a sensitivity of 95% and a specificity of 76%. The logistic regression model was significant (x2 (4) =28.98, p<0.001).
Expression of mir-181a was significantly higher in the healthy control group in comparison to the metastatic group and the local group (p=0.001 and p=0.02 respectively), however it did not distinguish the metastatic from the local group (p=0.06).
Mir-329 could not be detected reliably across all 3 groups, with a persistent CT value >35 in over 50% of samples. Mir-4433 and mir-6734 demonstrated no difference in expression between the local and metastatic groups.
In this study we have identified 2 novel miRNAs (mir-331 and mir-195) dysregulated in the circulation of patients with distant metastatic Luminal A breast cancer compared to patients with locally confined Luminal A breast cancer. In addition, we have shown that levels of these miRNA are comparable in age-matched healthy controls to the patients with local breast cancer.
Mir-195 has previously been investigated as a suppressor of metastatic disease in breast cancer nut to our knowledge this is the first study to identify mir-331 as a potential promoter of breast cancer metastasis. Further research is required to elucidate the precise mechanism of mir-331 in breast cancer, and also to establish if dysregulation of this miRNA profile precedes the development of metastatic disease and can contribute to the evolving paradigm of breast cancer management.