High microRNA-145 plasma levels are associated with decreased risk of future incident venous thromboembolism: the HUNT study

Venous thromboembolism (VTE), a collective term for deep vein thrombosis (DVT) and pulmonary embolism (PE), is a frequent and potentially fatal disease,^1,2 with severe long-term physical (ie, recurrence, postthrombotic syndrome, and post-PE syndrome)^3,4 and psychosocial complications (eg, depression and disability pension).^5,6 The incidence of VTE has slightly increased during the last 2 decades^7,8 and will likely continue to increase because of the rising prevalence of major VTE risk factors, such as an advancing age, obesity, and cancer.^9-11 To reduce the individual suffering and socioeconomic burden of VTE in society, there is a need to discover novel biomarkers and molecular mechanisms of VTE to improve risk prediction and pursue targeted prevention and treatment.

MicroRNAs (miRs) are small endogenous noncoding RNAs that downregulate gene expression at the posttranscriptional level by translational inhibition or messenger RNA degradation, with a single miR being able to regulate the expression of multiple genes.¹² MiRs have shown to play a critical role in several biological processes and in human diseases.^12-14 The level of various miRs in plasma is remarkably stable^15,16 and they have therefore emerged as attractive biomarkers of many diseases, including cancer and cardiovascular diseases (CVD).^13,17 Moreover, miRs could serve as potential therapeutic targets, because their expression levels can be either downregulated by synthetic miR inhibitors or upregulated by replacement therapy with miR mimics.^18,19 

Growing evidence from epidemiological studies and animal models suggests that the expression levels of miRs are dysregulated in VTE.²⁰ Several studies, most with a case-control design, have investigated the associations between various miRs and VTE, with inconsistent results between studies, probably because of differences in the study populations and limited statistical power of individual studies.^21-32 Furthermore, given the lack of temporality between exposure and outcome in a case-control design, the apparent associations between miRs and VTE may be a consequence of the VTE event itself rather than reflecting a role in disease pathogenesis, that is, reverse causation. In 1 of these studies, patients with VTE were found to have significantly lower plasma levels of miR-145 than controls.²⁵ Notably, in the same report, authors showed that administration of miR-145 mimics in a rat stasis model of venous thrombosis resulted in a dose-dependent reduction in thrombus formation, which was likely mediated by downregulation of tissue factor (TF).²⁵ In addition to TF,²⁵ coagulation factor XI (FXI)³³ and plasminogen activator inhibitor-1 (PAI-1)³⁴ were reported as gene targets of miR-145, and they are all recognized to be implicated in the pathogenesis of VTE.^35-38 Interestingly, miR-145 is also known to act as a tumor suppressor^39,40 and could be a possible shared risk factor in the mechanism of cancer-related VTE.

A prerequisite to clarify the apparent association between miR-145 and VTE previously reported in a case-control study²⁵ is to conduct a study with a well-established temporal sequence, in which the exposure (plasma miR-145) is assessed before the outcome (VTE). Here, we hypothesized that high plasma miR-145 levels are associated with reduced risk of future VTE. To examine this hypothesis, we used a case-cohort study derived from a general population-based cohort and investigated whether plasma miR-145 levels were associated with risk of future incident VTE.

The Trøndelag Health study (HUNT) is a population-based cohort study with repeated health surveys of inhabitants of the (former) Nord-Trøndelag County in Norway.⁴¹ To the third survey (HUNT3), conducted between 2006 and 2008, all inhabitants aged ≥20 years were invited, and 50 807 individuals participated (54% of those invited). The study was approved by the Regional Committee for Medical and Health Research Ethics, and all participants provided written informed consent for participation and use of data for medical research. Each participant was followed-up from the inclusion date onward, and all first-lifetime VTE events occurring among the participants up to 31 December 2019 were recorded.

Baseline information was collected at inclusion in the HUNT 3 survey by physical examination, blood samples, and validated self-administered questionnaires. The participants’ height and weight were objectively assessed by physical examination, and body mass index (BMI) was calculated as the weight divided by the square of height in meters (kg/m²). Information on cancer and history of arterial CVD (angina pectoris, myocardial infarction, and stroke) was collected through the self-administered questionnaires. Nonfasting blood samples were collected into vacutainer tubes containing EDTA as anticoagulant. Plasma was prepared by centrifugation (3000g for 10 minutes) and frozen at −80°C at the HUNT Biobank in Levanger, Norway.

We performed an extensive search in the hospital discharge diagnosis registries at the hospitals in Levanger and Namsos, as well as St Olavs hospital in Trondheim (the 3 hospitals covering the catchment area of HUNT participants), to identify all VTE cases occurring among the participants during follow-up. We performed a broad search with relevant International Classification of Diseases 10th Revision codes covering the years 2006 to 2019, and the medical record of each potential VTE case identified in the search was thoroughly reviewed for case validation. VTE events were adjudicated and recorded when signs and symptoms of lower extremity DVT or PE were objectively confirmed by radiological procedures, and treatment was initiated (unless contraindications were specified). Cases of concomitantly confirmed DVT and PE were classified as PE. Information on clinical risk factors and potential provoking factors in the 3 months preceding the event was extracted from the medical records using a standardized form. A VTE was categorized as provoked if ≥1 of the following provoking factors were present in the 3 months preceding the VTE event: surgery, trauma, acute medical condition (acute myocardial infarction, stroke, or severe infection), immobilization (bed rest of ≥3 days, confinement to wheelchair, or plaster cast), active cancer, or other factor specifically described as provoking in the medical record (eg, intravascular catheter).

For this study, we used a case-cohort design. We selected the first 600 VTE events occurring during follow-up, which yielded a maximum follow-up time of 7.3 years. A subcohort of 2000 participants was randomly sampled from the source cohort weighted by the age-distribution of the VTE cases in 5-year age groups. Participants who did not have plasma samples available (n = 99) and with a previous VTE (n = 56) were excluded, resulting in 527 incident VTE cases and 1918 subcohort members. We further excluded participants with missing data on miR-145 measurement (n = 30) and baseline BMI (n = 15), thus leaving 510 incident VTE cases and 1890 subcohort members in the final analytical sample of the case-cohort study (Figure 1). Because of the sampling design of the case-cohort, 42 of the VTE cases were also randomly sampled to the subcohort.

The samples were retrieved from the HUNT biobank, aliquoted, and sent to Qiagen Genomic Services (Hilden, Germany) for RNA isolation and miR quantification. Plasma samples were thawed on ice and centrifuged at 3000g for 5 minutes in a 4°C microcentrifuge. For RNA isolation, total RNA was extracted from the samples using miRNeasy Plasma Advanced Kit according to the high-throughput bead-based protocol version 1 (Qiagen, Hilden, Germany) in an automated 96-well format. The purified total RNA was eluted in a final volume of 50 μL. For measurement of miR-145, 2 μL RNA was reverse transcribed in 10-μL reactions using the miRCURY LNA RT Kit (Qiagen). Complementary DNA (cDNA) was diluted 50× and miR-145 was assayed by quantitative polymerase chain reactions (qPCR) on the miRNA Ready-to-Use PCR, custom panel using miRCURY LNA SYBR Green master mix. The qPCR analysis additionally included 2 reference genes, 2 synthetic RNAs (1 RNA purification control, and 1 cDNA synthesis control), 2 miRs to assess hemolysis, an interplate calibrator, and an empty negative control. The amplification was carried out in a LightCycler 480 Real-Time PCR System (Roche) in 384-well plates. The amplification curves were analyzed using the Roche LC software, both for determination of Cq (by the second derivative method) and for melting curve analysis.

The RNA isolation efficiency was analyzed by qPCR using primers targeting the synthetic UniSp2 added during RNA purification. cDNA synthesis efficiency was monitored by qPCR with primers targeting the synthetic UniSp6 added during the reverse transcription. To assess the technical variations between the panel plates, the interplate calibrator (UniSp3) was evaluated. Steady expression levels of these assays would indicate that extraction, reverse transcription, and qPCR were performed successfully. Expression levels of spike-in synthetic RNAs added during the RNA purification and cDNA synthesis did not vary between the samples, indicating comparable RNA isolation and cDNA synthesis (data not shown).

High-sensitivity C-reactive protein (CRP) was measured in nonfasting serum samples by latex immunoassay methodology using commercially available reagents (Abbott, Clinical Chemistry, IL) as previously described in the HUNT 3 survey.⁴² 

Statistical analyses were performed with STATA version 17.0 (Stata Corporation, College Station, TX) and R version 4.3.1 (The R Foundation for Statistical Computing, Vienna, Austria). The expression levels of miR-145 were normalized by the expression of miR-425-5p, as suggested by BestRef, and thereafter log transformed to make the skewed distribution close to normally distributed. MiR-425-5p is stably expressed in plasma and has been previously used as a reference miRNA in several studies.^43,44 The study population was divided into quartiles based on log-transformed expression levels of miR-145 in the subcohort participants. Baseline characteristics according to VTE and subcohort status and across miR-145 quartiles were presented using descriptive statistics and expressed as percentage for categorical variables and as mean (± standard deviation) or median (interquartile range) for continuous variables.

Follow-up time was calculated from the date of inclusion until the date of VTE, date of death or migration, or end of follow-up (10 August 2015), whichever came first. Weighted Cox-proportional hazards regression models were used to estimate hazard ratios (HRs) with 95% confidence intervals (CIs) for VTE according to quartiles of miR-145 using the lowest quartile as reference. The HRs were adjusted for age and sex in model 1; age, sex, and BMI in model 2; age, sex, BMI, and cancer and arterial CVD at baseline in model 3. In addition, subgroup analyses according to the presence of provoking factors (ie, provoked and unprovoked VTE) and anatomical location of VTE (ie, DVT and PE with or without DVT) were carried out. Because a sex difference in miR-145 levels was previously reported, with higher levels in women than in men,⁴⁵ we also conducted analysis for overall VTE using sex-specific cutoff values of miR-145 to conceive quartiles. Of note, low-grade inflammation, as reflected by high-sensitivity CRP levels, is reportedly associated with increased risk of VTE.^46,47 Furthermore, because the expression of miR-145 is often downregulated in conditions associated with increased inflammatory response, such as cancer^39,40 and metabolic disorders (eg, type 2 diabetes),^48,49 it is biologically plausible to assume that inflammation could potentially influence the association between miR-145 and VTE risk. Therefore, we added high-sensitivity CRP, a reliable downstream marker of inflammation, to model 3 of the Cox regression as a sensitivity analysis for overall VTE.

To assess potential nonlinearity between plasma miR-145 levels and risk of VTE, a generalized additive regression plot was generated to visualize the association by modeling miR-145 with a smoothing spline fit in a logistic model adjusted for the covariates of model 3. The miR-145 values were standardized to a mean value of 0 and a standard deviation of 1 before entering the analyses.

Given the relatively long follow-up time in the parent cohort (∼7 years for some participants), the results based on baseline measurements of miR-145 could be influenced by regression dilution bias.⁵⁰ To address this, we performed analyses that restricted the maximum follow-up time from blood sampling to the VTE events, while keeping all subcohort members in the analyses. The weighted Cox regression analyses on time restrictions were set to require at least 10 VTE events, and HRs adjusted for the covariates included in model 3 were generated at every follow-up time at which a new VTE event occurred and plotted as a function of this maximum time.

The distribution of baseline characteristics according to VTE and subcohort status is shown in Table 1. As expected, the mean age was virtually the same between VTE cases and subcohort members, whereas the proportion of men and self-reported history of cancer, the mean BMI, and the median CRP levels were slightly higher among cases. The distribution of baseline characteristics across miR-145 quartiles is described for the subcohort members in supplemental Table 1, available on the Blood website. It is worth noting that the proportion of men clearly decreased with increasing quartiles of miR-145 plasma levels. No substantial difference was observed in median CRP levels as well as in the other covariates across miR-145 quartiles.

The characteristics of patients with VTE assessed at the time of the VTE event are shown in Table 2. The mean age at the time of VTE was 70 years, and 50.8% were men. Of the total VTE events, 42.9% were DVTs and 57.1% were PEs, with 58.2% of the events being classified as provoked. Cancer was among the most common provoking factors, occurring in 23.5% of the VTE cases.

The HRs of VTE and the subgroups of provoked and unprovoked VTE across quartiles of plasma miR-145 levels are shown in Figure 2. Elevated expression levels of miR-145 in plasma were associated with decreased risk of VTE. Participants with miR-145 levels in the highest quartile had a 49% lower risk of VTE (HR, 0.51; 95% CI, 0.38-0.68) than those with miR-145 levels in the lowest quartile in age- and sex-adjusted analysis (model 1). Risk estimates were not affected after further adjustment for BMI in model 2 (HR, 0.52; 95% CI, 0.39-0.70), and cancer and arterial CVD at baseline in model 3 (HR, 0.52; 95% CI, 0.39-0.69). The inverse association was particularly pronounced for unprovoked VTE, with a 60% lower risk of VTE (HR, 0.40; 95% CI, 0.26-0.63) when comparing the highest vs lowest quartile in the fully adjusted model. For the other subgroups, the comparison of the highest vs lowest quartile yielded HRs of 0.60 (95% CI, 0.42-0.86) for provoked VTE (Figure 2), 0.62 (95% CI, 0.41-0.94) for DVT, and 0.45 (95% CI, 0.31-0.65) for PE (supplemental Figure 1). When sex-specific cutoff values were used to conceive quartiles (supplemental Figure 2) or when high-sensitivity CRP was added to model 3 (supplemental Figure 3), results remained essentially the same as those obtained in the main analysis.

The association between plasma miR-145, entered as a continuous variable, and risk of overall and unprovoked VTE is shown in Figure 3. There was an inverse relationship between plasma expression levels of miR-145 and thrombosis risk, with high levels of this miR being associated with a decreased thrombosis risk, particularly for unprovoked VTE (Figure 3B). It is important to note that risk estimates at more extreme levels are imprecise with wide 95% CIs because of limited number of individuals in the analysis.

To consider the possibility of underestimating the true association because of regression dilution bias, we estimated HRs of VTE (highest vs lowest quartile of miR-145) as a function of time between blood sampling and the events for overall (Figure 4A) and unprovoked (Figure 4B) VTE. The HRs of VTE by high plasma levels of miR-145 were lower with shortened time between blood sampling and VTE events, and the inverse association between miR-145 and thrombosis risk was especially strong for unprovoked VTE during the first 2 years after blood sampling (Figure 4B). Nonetheless, for both overall and unprovoked VTE, the inverse relationship between miR-145 levels and VTE risk remained significant during the entire study period (7 years).

In this population-based case-cohort study, elevated levels of miR-145 were associated with decreased risk of future incident VTE. Participants with plasma expression levels of miR-145 in the highest quartile had a 49% lower risk of VTE than those with miR-145 in the lowest quartile in age-and sex- adjusted analysis. The risk estimates remained essentially the same after further adjustment for BMI, and cancer and arterial CVD at baseline. Among the subgroups, the inverse relationship between miR-145 plasma levels and thrombosis risk was most pronounced for unprovoked VTE. Although the HRs for VTE were lower with shortened time between blood sampling and the thrombotic events, the association between high miR-145 levels and decreased VTE risk remained significant even many years after blood sampling. Our results indicate that plasma miR-145 displayed an inverse association with risk of future VTE in the general population and could serve as a reliable biomarker for assessing disease risk. The protective role of miR-145 against VTE found in our study is in line with previous experimental data and may form the basis for future studies designed to unravel the role of miR-145 in the pathogenesis of VTE and its potential to be used as therapeutic target for VTE management.

To the best of our knowledge, this is the first study that investigated the relationship between plasma levels of a miR and incident VTE using a prospective design at the population level. Our finding of an inverse association between plasma miR-145 levels and risk of future incident VTE confirms the results from a small case-control study comprising 40 male participants,²⁵ and another report involving only patients with chronic obstructive pulmonary disease, in which miR-145 expression level was downregulated in patients with PE compared with those without PE.³¹ Of note, our prospective study design prohibits the possibility for reverse causation, and supports the notion that miR-145 is involved in the development of VTE rather than being a marker or a consequence of the disease.

It is important to address that miR-145 has been shown to contribute to the pathogenesis of other human diseases, particularly cancer.⁵¹ MiR-145 is expressed in a variety of tissues and cell types and acts mainly as a tumor suppressor by targeting genes involved in different aspects of tumor growth and progression.^39,40 Indeed, miR-145 is found to be downregulated in a wide range of cancers, including colorectal, lung, breast, ovarian, prostate, gastric, and bladder cancers.^39,40 Given the established association between cancer and VTE,⁵² taking cancer into account when studying the effect of miR-145 on the risk of VTE is crucial. Accordingly, in this study, we demonstrated that the association between miR-145 and VTE did not seem to be driven by cancer, because this association was not affected by adjusting the regression models for a history of cancer at baseline. Moreover, the inverse association between expression levels of miR-145 in plasma and thrombosis risk was especially strong in analyses restricted to unprovoked VTE, in which patients with VTE related to active cancer were not included.

Although the HRs of VTE by elevated miR-145 levels were lower with shortened time between blood sampling and VTE events, the component of regression dilution because of intraindividual fluctuation of miR-145 plasma levels was modest at most. The stability of plasma miR-145 underscores its potential to serve as a reliable short- and long-term biomarker of VTE risk. Moreover, according to previous experimental data, miR-145 might be involved in the pathophysiological mechanisms that underlie venous thrombus formation.²⁵ In a stasis model of venous thrombosis in rats, Sahu et al demonstrated that systemic delivery of increasing dosages of miR-145 mimics before inferior vena cava ligation resulted in a dose-dependent decrease in thrombus weight.²⁵ The same study showed that the reduction in thrombus formation by miR-145 was likely mediated by downregulation of TF expression.²⁵ In vitro experiments further revealed that miR-145 had the potential to negatively regulate TF gene expression via direct interaction with TF 3' untranslated region.²⁵ Of note, miR-145 has been reported to regulate the expression levels of other hemostatic factors that are linked to VTE pathogenesis. Using an in silico consensus approach, Sennblad et al predicted the binding of miR-145 to F11 3' untranslated region, and further strengthened the finding using an in vitro luciferase report assay, thus suggesting that miR-145 indeed targeted F11 messenger RNA and regulated FXI levels.³³ Individuals with elevated plasma levels of FXI are at increased risk of VTE,³⁶ whereas those with severe deficiency of FXI appear to have a reduced thrombosis risk.³⁷ Additionally, FXI inhibitors have shown to reduce the risk of VTE in patients exposed to orthopedic surgery in clinical trials.⁵³ Finally, miR-145 was reported to negatively regulate PAI-1 expression in bladder cancer cells.³⁴ PAI-1 is the primary inhibitor of fibrinolysis and the VTE risk has been shown to increase with higher plasma levels of PAI-1 in a recent prospective nested case-control study.³⁸ Hence, our results on the protective role of miR-145 against VTE analyzed in light of the previous experimental data^25,33,34 may support the concept that miR-145 plays a critical role in the pathogenesis of VTE, and that replacement therapy with miR-145 mimics may emerge as a promising strategy for VTE treatment and prevention.

Strengths of our study include the recruitment of participants from a large population-based cohort with a wide age distribution, and the objective validation of VTE events. Because VTE cases and subcohort members were derived from the same unselected source population, the likelihood of selection bias was minimized. Our prospective study design established a clear temporal sequence between the exposure and the outcome, because blood samples for miR-145 assessment were collected at cohort inclusion. Some limitations also need attention. Plasma miR-145 levels were only measured at baseline, and changes in levels during follow-up (maximum ∼7 years) could have resulted in an underestimation of the true associations because of regression dilution bias.⁵⁰ Nonetheless, although we could identify some degree of regression dilution, the association between plasma levels of miR-145 and VTE remained substantial even several years after blood sampling. The blood samples used for miR-145 analysis were collected between 2006 and 2008 and stored at −80°C until measurement of plasma miR-145 levels in 2021. Although plasmas were stored for a long time (up to 15 years), it is unlikely that the storage time would affect the results, because miRs are reported to be stable in plasma samples.^15,16 Even if storage time would affect miR-145 levels, the storage effect is assumed to be similar in samples from cases and subcohort members because samples were stored under the same conditions and for the same amount of time. Thus, any potential misclassification would be nondifferential with regard to the VTE status, which could have led to an underestimation of the true associations.⁵⁰ Unfortunately, information on hemostatic factors that could potentially mediate the association between miR-145 and VTE risk, such as FXI and TF, was not available in the HUNT3 survey that originated the present study.

In conclusion, elevated expression levels of miR-145 in plasma were associated with decreased risk of future incident VTE. The protective role of miR-145 against VTE is consistent with previous experimental data and suggests that miR-145 may serve as a biomarker for assessing the risk of future VTE. Moreover, our findings may form the basis for future research aimed to unravel the contribution of miR-145 to VTE pathogenesis and its potential as therapeutic target for VTE.

The authors thank clinicians and other employees at Nord-Trøndelag Hospital Trust for their support and for contributing to data collection in this research project.

The Thrombosis Research Center has received an independent grant from Stiftelsen Kristian Gerhard Jebsen (2014-2020). The Trøndelag Health Study (HUNT) is a collaboration between the HUNT Research Centre (Faculty of Medicine and Health Sciences, Norwegian University of Science and Technology NTNU), Trøndelag County Council, Central Norway Regional Health Authority, and the Norwegian Institute of Public Health.

Contribution: V.M.M. designed the study, analyzed data, interpreted the results, and drafted the manuscript; K.D.H. performed the statistical analysis, interpreted the results, and revised the manuscript; O.S. and K.H. interpreted the results and revised the manuscript; S.K.B. and J.-B.H. designed the study, organized data collection, interpreted the results, contributed to the manuscript draft, and revised the manuscript; and all authors reviewed and approved the final version of the manuscript.

Conflict-of-interest disclosure: The authors declare no competing financial interests.

Correspondence: Vânia M. Morelli, Thrombosis Research Group (TREC), Department of Clinical Medicine, UiT The Arctic University of Norway, N-9037 Tromsø, Norway; email: vania.m.morelli@uit.no.