Associations of plasma extracellular microRNAs with new-onset breast cancer in the Framingham heart study

Abstract

Breast cancer is the second leading cause of cancer deaths among women. Multiple microRNAs (miRNAs) have been reported to be associated with breast cancer progression or metastasis. The purpose of the current study was to identify plasma extracellular miRNAs associated with incident breast cancer. Levels of 166 plasma miRNA were measured using qRT-PCR in 2140 Framingham Heart Study female participants with a median follow up of 15.7 years. Prospective analyses of the associations of miRNAs with the occurrence of 56 new-onset breast cancer events were conducted using proportional hazards regression. The expression levels miR-134-5p (P=0.002) and miR-505-3p (P=0.005) were found to be positively associated with incident breast cancer after adjusting for age, body mass index, and cigarette smoking. These results highlight plasma miRNAs as potential biomarkers of breast cancer risk. Validation of these findings in larger and more diverse cohorts is warranted.

Introduction

MicroRNAs (miRNAs) are small, non-coding RNAs comprised of 18-24 nucleotides that regulate gene expression by binding to complementary or near-complementary sequences in 3’-untranslated regions of messenger RNA (mRNA). In doing so, miRNAs can alter gene expression via mRNA transcript degradation or translational repression. It has been suggested that miRNAs regulate approximately two-thirds of human protein-coding genes [1,2]. Accordingly, miRNAs have been implicated as powerful regulators of a range of physiological and pathophysiological processes [1].

While miRNAs can be found inside the cell, miRNAs are also found extracellularly in circulating blood and other body fluids [1,3]. These miRNAs can be transported across cell membranes to mediate cell-to-cell communication [4]. Moreover, expression levels of extracellular miRNAs (ex-miRNAs) have been reported to be associated with a wide range of diseases, including cancer [5].

Ex-miRNAs play a role in numerous cancer-related processes, including cell proliferation, differentiation, and apoptosis [2]. Dysregulation of ex-miRNAs is often observed in various cancers due to altered miRNA biogenesis and perturbed transcription of miRNA-encoding genes. As a result, some ex-miRNAs become significantly overexpressed while the expression of others significantly decreases [6].

Given that ten percent of known human miRNAs can be detected in plasma, are stable, and can be measured by polymerase chain reaction (PCR), ex-miRNAs may be ideal candidates as cancer biomarkers [6]. Most studies of miRNAs and breast cancer have examined associations of miRNA expression in tumor tissue with prevalent breast cancer; there are few studies examining plasma miRNAs as predictors of new-onset breast cancer [7,8]. Therefore, in the present study, we investigated associations of plasma ex-miRNAs with incident breast cancer in a large population.

Sample collection

The Framingham Heart Study (FHS) is a population-based, observational cohort study established in 1948 to identify risk factors for the development of cardiovascular disease [9]. The study sample for the present investigation included female participants in the FHS Third Generation cohort (n=2,140 at Exam 1) in whom plasma miRNA levels were measured. A total of 24 participants with prevalent breast cancer were excluded, leaving 2,116 eligible participants. The FHS protocol was approved by the Boston Medical Center IRB; informed consent was provided by all participants.

Study participants attended the first examination in the years 2002-2005 and underwent a fasting blood draw. Plasma samples were immediately isolated (Applied Biosystems, catalog #A27828) and processed (ThermoFisher Flex Magnetic Particle Processor, 96DW) before being stored at -80°C until assay. The isolated miRNAs were converted to cDNA (Applied Biosystems, catalog #A28007) and synthesis and preamplification (Applied Biosystems, catalog #4484075) were performed. qRT-PCR (Applied Biosystems, catalog #A25576; BioMark HD, catalogue #BMKHD-BMKHD; Fluidigm, catalog #BMK-M-96.96) was used to quantitate miRNAs and data are reported as cycle threshold (Ct) values. We required that an ex-miRNA be measurable in at least 50% of participants, yielding 166 ex-miRNAs that were carried forward in the analysis. The methods for plasma miRNA isolation and measurement have been previously reported and are described in greater detail elsewhere [10]. Self-reported smoking history was obtained and cigarette smoking status (current/former/never) was defined according to previously published classifications [10].

Participants were prospectively followed through 2018 for the occurrence of incident breast cancer. Breast cancer events were identified by surveillance through routine examinations, health history updates, hospital admissions, or death records. Cancer cases were adjudicated and coded on the basis of tumor topology and morphology and graded by two independent physicians using International Classification of Diseases codes [11].

We used proportional hazards regression with plasma miRNA levels as the independent variable and incident breast cancer as the outcome. The covariates for the models were age, cigarette smoking status, and body mass index (BMI). All statistical analyses were conducted using SAS 9.4 (SAS Institute Inc. 2020. SAS/STAT^® 15.2 User’s Guide. Cary, NC: SAS Institute Inc.) procedure PHREG.

Results and discussion

Among eligible participants (n=2116; mean age: 40±9 years), 1160 (55%) were never smokers, 640 (30%) were former smokers, and 315 (15%) were current smokers. During a median follow-up time of 15.7 years, 56 women developed new-onset breast cancer. The full table of clinical characteristics of the study participants can be found in Table 1.

Table 1.

Clinical characteristics of the study participants

Clinical characteristic	Mean (standard deviation)
Age, years	40 (9)
Systolic blood pressure, mmHg	113 (14)
Diastolic blood pressure, mmHg	73 (9)
Triglycerides, mg/dL	97 (62)
Total cholesterol, mg/dL	185 (34)
High-density lipoprotein, mg/dL	61 (16)
Fasting blood glucose, mg/dL	92 (18)
BMI, kg/m2	26 (6)
Dichotomous outcome	Number of participants (%)
Never smoker	1160 (54.8)
Former smoker	640 (30.3)
Current smoker	315 (14.9)
Type 2 diabetes mellitus	46 (2.2)
Incident breast cancer	56 (2.7)

Among the 166 plasma ex-miRNAs included in analysis, two ex-miRNAs (miR-134-5p and miR-505-3p) were associated with incident breast cancer at P < 0.005 before adjusting for multiple testing (Table 2). The association results for all ex-miRNAs included in the prospective analyses are provided in Supplementary Table 1.

Table 2.

Ex-miRNAs associated with incident breast cancer in the FHS

Ex-miRNA	N analyzed	Event	Estimate*	Standard Error	Chi Squared	Probability Chi Squared	Hazard Ratio	HR Lower Confidence Level	HR Upper Confidence Level
134-5p	1594	51	-0.126	0.0413	9.296	2.30E-03	0.882	0.813	0.956
505-3p	1805	49	-0.163	0.0582	7.882	4.99E-03	0.849	0.758	0.952

^*Estimates are reported as Ct values such that a negative estimate indicates a higher Ct value (lower ex-miRNA expression).

The Kaplan-Meier breast cancer-free survival plot for miR-134-5p (Figure 1A) indicates that there was a higher proportion of event-free survival for participants with low miR-134-5p expression (miR-134-5p ≥ median Ct-values) compared to participants with high miR-134-5p expression (miR-134-5p < median Ct-values). The Kaplan-Meier survival plot for miR-505-3p depicts analogous behavior, i.e., lower miR-505-3p expression levels associated with higher event-free survival (Figure 1B).

Figure 1.

Kaplan-Meier plots for ex-miRNAs. The x-axis “Follow-up time” indicates years to new breast cancer event, and the y-axis “Prob(Free of Breast Cancer)” indicates the proportion of participants with event-free survival. The variable ex-miRNA was defined as: if ex-miRNA ≥ median Ct-value, then low ex-miRNA expression (red); if ex-miRNA < median Ct-value, then high ex-miRNA expression (blue). A. Among the 1622 female participants with analyzed miR-134-5p levels, there were 51 new breast cancer cases (among 811 participants with miR-134-5p levels above the median, there were 22 new breast cancer cases; among 811 participants with miR-134-5p levels below the median, there were 29 new breast cancer cases). B. Among the 1843 female participants with analyzed miR-505-3p levels, there were 49 new breast cancer cases (among 923 participants with miR-505-3p levels above median, there were 18 new breast cancer cases; among 920 participants with miR-505-3p levels below median, there were 32 new breast cancer cases).

The purpose of our study was to examine associations of plasma ex-miRNAs with new-onset breast cancer in middle-aged women. To this end, we identified two plasma ex-miRNAs (miR-134-5p and miR-505-3p) whose expression levels were positively associated with incident breast cancer during a median follow-up of 15.7 years. Previous studies have enumerated a negative association between the expression levels of miR-134-5p and miR-505-3p and breast cancer progression in tumor samples or cell lines [12-16]. However, such studies examining miRNAs isolated from tumor tissue or cell lines demonstrate an association between cellular miRNAs and prevalent breast cancer. To our knowledge, this is the first study to demonstrate the association of extracellular miR-134-5p and miR-505-3p expression with incident breast cancer. These findings highlight the potential of ex-miRNAs to serve as biomarkers of breast cancer risk.

Discordance in the direction of effect of miRNAs on breast cancer has been observed between breast cancer tissue samples or cell lines and miRNAs derived from cell-free blood samples for other miRNAs, including miR-122, miR-145-5p, and miR-125b-5p [17-23]. For example, expression levels of miR-122 isolated from human breast cancer cells have been reported to be negatively associated with prevalent breast cancer, whereas expression levels of serum miR-122 have been found to be positively associated with prevalent breast cancer [22,23]. One proposed explanation for this apparent paradox may be that breast cancer cells secrete these miRNAs at higher rates than normal cells, decreasing cellular miRNA levels and increasing ex-miRNA levels [22]. The current study, however, is unable to provide a mechanism of differential expression or evidence of ex-miRNA communication with specific tissues or cell lines. Functional studies investigating the molecular targets and pathways of miR-134-5p and miR-505-3p are warranted.

The present study has several strengths, including a large sample size of middle-aged women with available plasma ex-miRNA and the availability of new-onset breast cancer outcomes during long-term follow-up. Yet, there are several limitations. Of the 166 plasma ex-miRNAs included in the analysis, only two ex-miRNAs were significantly associated with incident breast cancer after adjusting for covariates; these associations were not significant, however, after correcting for multiple testing. Statistical power was limited by the relatively small number of new-onset breast cancer cases in the study sample. Additionally, the study population consisted largely of participants of European ancestry. Future studies that include larger sample sizes and greater racial and ethnic diversity are warranted.

This population-based study reveals two plasma ex-miRNAs associated with incident breast cancer. Moreover, these findings motivate further investigation into the molecular targets and mechanistic pathways of the identified ex-miRNAs. This effort may help advance the utility of plasma ex-miRNAs as clinically useful biomarkers.

Disclosure of conflict of interest

None.