Circulating MicroRNA-19 and cardiovascular risk reduction in response to weight-loss diets

Abstract

Objective:

MicroRNA-19 (miR-19) plays a critical role in cardiac development and cardiovascular disease (CVD). We examined whether change in circulating miR-19 was associated with change in CVD risk during weight loss.

Methods:

This study included 509 participants with overweight or obesity from the 24-month weight-loss diet intervention study (the POUNDS Lost trial) and with available data on circulating miR-19a-3p and miR-19b-3p at baseline and 6 months. The primary outcome for this analysis was the change in atherosclerotic CVD (ASCVD) risk at 6 and 24 months, which estimates the 10-year probability of hard ASCVD events. Secondary outcomes were the changes in ASCVD risk score components.

Results:

Circulating miR-19a-3p and miR-19b-3p levels significantly decreased during the initial 6-month dietary intervention period (P = 0.008, 0.0004, respectively). We found that a greater decrease in miR-19a-3p or miR-19b-3p was related to a greater reduction in ASCVD risk (β[SE] = 0.33 [0.13], P = 0.01 for miR-19a-3p; β[SE] = 0.3 [0.12], P = 0.017 for miR-19b-3p) over 6 months, independent of concurrent weight loss. Moreover, we found significant interactions between change in miR-19 and sleep disturbance on change in ASCVD risk over 24 months of intervention (P interaction = 0.01 and 0.008 for miR-19a-3p and miR-19b-3p, respectively). Participants with a greater decrease in miR-19 without sleep disturbance had a greater reduction of ASCVD risk than those with slight/moderate/great amounts of sleep disturbance. In addition, change in physical activity significantly modified the associations between change in miR-19 and change in ASCVD risk over 24 months (P interaction = 0.006 and 0.004 for miR-19a-3p and miR-19b-3p, respectively). A greater decrease in miR-19 was significantly associated with a greater reduction in ASCVD risk among participants with an increase in physical activity, while non-significant inverse associations were observed among those without an increase in physical activity.

Conclusions:

In conclusion, decreased circulating miR-19 levels during dietary weight-loss interventions were related to a significant reduction in ASCVD risk, and these associations were more evident in people with no sleep disturbance or increase in physical activity.

Trial registration:

ClinicalTrials.gov NCT00072995.

1. Introduction

Cardiovascular disease (CVD) remains a leading cause of mortality worldwide [1,2]. MicroRNAs (miRNAs) are small non-coding RNA molecules with a powerful ability to regulate gene expression post-transcriptionally. Accumulating evidence indicates that miRNAs play a crucial role in the pathogenesis of CVD, highlighting the potential of miRNAs as novel targets for preventing CVD [3,4].

The microRNA-19 (miR-19) family, consisting of miR-19a and miR-19b, is an ancient and conserved miRNA species with identical seed regions among its members [5]. While the derived mature sequences of miR-19a and miR-19b differ by one nucleotide outside of the seed sequence, they are known to share many targeted genes, but at times they may have distinct targets [5]. In recent years, growing studies have closely linked the unbalanced expression of miR-19 in blood, heart, and vessels with cardiac development and CVD risk in humans and animals [6–21]. As a result, miR-19 family members are considered to be diagnostic biomarkers and promising therapeutic targets for CVD. In addition, circulating miRNAs are subject to dietary and behavioral (such as sleep and physical activity) modulations and subsequently regulate important cardiovascular-related pathways involved in lipid metabolism, endothelial function, cardiac hypertrophy, and fibrosis [22–25]. However, no study has assessed the temporal changes of circulating miR-19 within the context of dietary interventions and whether the changes are related to the alteration of CVD risk.

Therefore, in this study, we investigated the association between changes in circulating miR-19 levels and CVD risk reduction in response to weight-loss diets in the Preventing Overweight Using Novel Dietary Strategies (POUNDS Lost) Trial. We further assessed the interactions of miR-19 with dietary interventions, sleep disturbance, and physical activity on CVD risk.

2. Methods

2.1. Study design and participants

The POUNDS Lost study was a randomized dietary intervention trial conducted at two clinical research sites: the Harvard T.H. Chan School of Public Health in Boston, MA and the Pennington Biomedical Research Center in Baton Rouge, LA. The study was approved by the human subjects committee at each site with provided written informed consent from participants [25]. Detailed descriptions of the study design, participants’ recruitment, inclusion, and exclusion criteria have been summarized [25]. We excluded participants with diabetes or unstable cardiovascular disease, using medications that affect body weight, or with insufficient motivation assessed by interview and questionnaire [25]. Briefly, a total of 811 individuals who were obese or overweight (body mass index [BMI] ≥25 kg/m² and ≤40 kg/m²) and aged 30e70 years were enrolled and randomly assigned to one of the four diets restricted by 750 kcal/d. The four diets consisted of the following macronutrient compositions: 1) low fat (20% of energy), average protein (15% of energy); 2) low fat (20%), high protein (25%); 3) high fat (40%), average protein (15%); and 4) high fat (40%), high protein (25%). Thus, we have two diets low in fat (20%), the other 2 diets high in fat (40%), 2 diets with average protein (15%), and the other 2 diets with high protein (25%), constituting a 2-by-2 factorial design. A graded difference is further established in carbohydrate intake ranging from 35% in the high-fat, high-protein diet to 65% in the low-fat, average-protein diet.

The present study included 509 participants based on the availability of data on circulating miRNAs (hsa-miR-19a-3p and hsa-miR-19b-3p) at both baseline and 6 months and complete data on changes in study outcomes across the 2-year intervention trial.

2.2. Assessment of circulating miRNAs

The measurements of miRNAs were performed at the QIAGEN laboratory. Fasting blood samples were collected into purple top EDTA vacutainer tubes at baseline and 6 months and stored at −80 °C. RNA was first isolated from 200 μl per plasma sample with an elution volume of 20 μl per sample using the miRNeasy Serum/Plasma Advanced (QIAGEN) based on the manufacturer’s instructions (http://www.qiagen.com/HB-2390). We applied the QIAseq^® FastSelect^™ blockers to deplete hemolysis-specific miRNAs after we performed a pilot study within our samples, confirming that hemolysis-specific miRNAs were successfully depleted and leaving more sequencing reads for miRNAs of interest. After the library preparation was done using the QIAseq miRNA Library Kit (QIAGEN), a total of 5 μl total RNA was converted into miRNA Next Generation Sequencing (NGS) libraries. Unique molecular identifiers (UMIs) were introduced in the reverse transcription step after adapter ligation. The cDNA was amplified using PCR (22 cycles) and then purified. Library preparation was quality-controlled using capillary electrophoresis (Agilent TapeStation D1000 ScreenTape). The libraries were pooled in equimolar ratios according to the quality of the inserts and the concentration measurements. The library pool(s) were quantified using qPCR and then sequenced on a NovaSeq (Illumina Inc.) sequencing instrument according to the manufacturer’s instructions (1 × 75, 1 × 6). Raw data was demultiplexed and FASTQ files for each sample are generated using the bcl2fastq2 software (Illumina Inc.). Nine samples failed the library preparation quality control, and two samples received a small number of reads and were excluded from the analysis. All the read mapping and quantification analyses were conducted using CLC Genomics Server 21.0.1. The workflow “QIAseq miRNA Quantification” of CLC Genomics Server with standard parameters was used to map the reads to miRBase version 22 for miRNAs. All reads that did not map to the smallRNA databases, without perfect matches or as isomiRs (maximum 2 mismatches and/or alternative start/end position of 2 nt), were mapped to the human genome GRCh38 with ENSEMBL GRCh38 version 97 annotation, using the “RNA-Seq Analysis” workflow of CLC Genomics Server with standard parameters. Finally, normalized data on miRNAs (normalized as TMM (trimmed mean of M values) -adjusted CPM (counts per million) values) were applied to our analyses.

In this study, 523 participants who took part in the initial examination with available blood samples were included. Following quality control procedures for miRNA library preparation and sequencing, 521 individuals had data available for circulating miRNAs at baseline. Furthermore, out of those 521 individuals, 509 had both baseline and 6-month miRNA data. The primary miRNA exposure of interest for our study was hsa-miR-19a-3p and hsa-miR-19b-3p. Circulating miR-19 data were log-transformed to improve normality. We then calculated the changes in standardized miR-19a-3p and miR-19b-3p from baseline to 6 months after the intervention began.

2.3. Assessment of anthropometrics, behavioral, and biochemical variables

Baseline information on age, sex, race, antihypertensive medication use, insulin use, and smoking status was collected using standardized questionnaires. Dietary intake was assessed using a 5-day diet record at baseline and a 24-h recall at 6 and 24 months to assess adherence to the diet intervention among a random sample of 50% of the total participants. Biomarkers of urinary nitrogen excretion for protein and respiratory quotient for fat versus carbohydrates were used to validate self-reported adherence to macronutrient targets [26]. Habitual physical activity (PA) levels at baseline and 6 months were assessed by the Baecke questionnaires. Change in total habitual PA at 6 months was calculated and classified into 2 groups of no change and decreases (≤0) or increases in PA (>0). Self-reported sleep disturbance information (“not at all,” “slight amount,” “moderate amount,” to “great amount) was collected by trained interviewers at baseline and every 6 months during the intervention period. We utilized the sleep disturbance data reported at 6 months, which coincided with the period for the change in miR-19. Due to the small sample sizes in the slight, moderate, and great categories, sleep disturbance during the initial 6 months was also classified into two groups, those without sleep disturbances (No) and those with slight, moderate, or significant disturbances (Yes, slight, moderate, and great amounts). Weight was measured using calibrated hospital scales before breakfast at baseline, and every 6 months thereafter. BMI was calculated as weight in kilograms divided by the square of height in meters. Waist circumference was also measured every 6 months using a non-stretchable tape measured 4 cm above the iliac crest at baseline. Systolic (SBP) and diastolic blood pressure (DBP) were measured using automatic blood pressure monitors. Blood samples were repeatedly collected in the fasting state at baseline, 6 months, and 24 months. The procedures for measuring serum lipids (total cholesterol, high-density lipoprotein (HDL) cholesterol, low-density lipoprotein (LDL) cholesterol, and triglycerides), fasting glucose, insulin, glycated hemoglobin (HbA1c), and calculating the homeostasis model assessment of insulin resistance (HOMA-IR) in POUNDS Lost have been described previously [25].

2.4. Assessment of ASCVD risk

The ASCVD risk score was used to predict both 10-year and lifetime ASCVD risk of an individual through the ACA/AHA pooled cohort equation [27]. Predicting variables included in the model were age, sex, race, total cholesterol, HDL cholesterol, SBP, antihypertensive therapy, history of diabetes, and current smoking status. The calculation of ASCVD risk was limited to individuals aged 40–79 years, as it was recommended to screen asymptomatic adults within this age range for traditional ASCVD risk factors and estimates their 10-year ASCVD risk based on the ACC/AHA guideline [27]. Additionally, the calculation of ASCVD risk was only applicable when total cholesterol falls between 130 and 320 mg/dL, HDL cholesterol is between 20 and 100 mg/dL, and LDL cholesterol is between 30 and 300 mg/dL [27]. In our study, we calculated the ASCVD risk among participants who met the aforementioned criteria and subsequently selected 496, 470, and 360 participants who had baseline, 6-month, and 24-month ASCVD data available, respectively.

2.5. Statistical analysis

Data distributions of continuous variables were checked for normality before analysis. Primary outcomes were changes in ASCVD risk at 6 months and 24 months. Secondary outcomes were changes in individual CVD risk factors, including total cholesterol, HDL cholesterol, glucose, HbA1c, and blood pressure.

General linear models were performed to calculate effect size per log-transformed change in circulating miR-19 with changes in study outcomes from baseline to 6 months and 24 months, adjusting for age, sex, race, blood collection site, diet group, baseline BMI, baseline levels of the miRNA, and the respective outcome at baseline (model 1). Specifically, to avoid possible over-adjustment, confounding variables that were used to calculate the ASCVD risk score (age, sex, and race) were removed from the analysis for ASCVD risk. In model 2, we further adjusted for concurrent weight change in the above models. We examined the modification effect of sleep disturbance or change in PA by evaluating multiplicative interactions between change in miR-19 and sleep disturbance and change in PA at 6 months on primary outcomes. We also analyzed potential differences in the main findings across different diet intervention groups (low vs high fat or average vs high protein) by testing miRNA-diet interactions. P < 0.05 was considered statistically significant. Statistical analyses were performed with SAS version 9.4.

3. Results

Table 1 shows the baseline characteristics of the included 509 study participants according to the tertiles of miR-19a-3p. Levels of miR-19a-3p and miR-19b-3p were very highly correlated at baseline and 6 months (Pearson’s r > 0.9 and P < 0.0001 for all). Across tertiles of miR-19a-3p at baseline, there was no significant difference in age, sex, race, diet groups, dietary intakes, smoking, physical activity, sleep disturbance, biomarkers of diet adherence (urinary nitrogen and respiratory quotient), blood pressure, lipids profiles, glucose, insulin, HbA1c, HOMA-IR, and adiposity measures (body weight, BMI, waist circumference) (all p-values >0.05; Table 1).

Table 1.

Baseline Characteristics of the study participants according to the tertiles of miR-19a-3p.

	N	Tertile 1 (n = 171)	Tertile 2 (n = 170)	Tertile 3 (n = 168)	P value
miR-19a-3pa	509	5.9 (0.3)	6.51 (0.14)	7.08 (0.31)	<0.0001
miR-19b-3pa	509	7.07 (0.31)	7.66 (0.2)	8.22 (0.33)	<0.0001
Age, year	509	51.98 (9.48)	51.54 (9.18)	51.53 (8.92)	0.88
Female	509	98 [57.31%]	109 [64.12%]	97 [57.74%]	0.36
White	509	142 [83.04%]	134 [78.82%]	138 [82.14%]	0.58
Body weight, kg	509	94.32 (16.11)	92.17 (15.35)	92.82 (14.84)	0.42
BMI, kg/m2	509	32.89 (4.2)	32.39 (3.61)	32.43 (3.59)	0.40
Waist circumference, cm	509	104.27 (12.98)	102.77 (13.13)	104.02 (13.08)	0.53
Diet groups
High fat diet	509	81 [47.37%]	75 [44.12%]	83 [49.4%]	0.62
High protein diet	509	75 [43.86%]	78 [45.88%]	78 [46.43%]	0.88
Dietary intake per dayb
Energy (kcal)	267	2033.44 (586.01)	2036.14 (611.26)	1945.36 (587)	0.52
Protein (%)	267	18.29 (3.6)	17.89 (2.79)	18.07 (3.03)	0.69
Fat (%)	267	37.6 (5.59)	36.53 (5.84)	35.66 (5.35)	0.07
Carbohydrate (%)	267	44.01 (7.82)	44.87 (7.4)	46.23 (6.64)	0.13
Biomarkers of adherence
Urinary nitrogen, g	508	12.07 (4.29)	12.29 (4.47)	12.3 (4.37)	0.86
Respiratory Quotient	508	0.84 (0.05)	0.84 (0.04)	0.84 (0.04)	0.95
SBP, mmHg	509	120.8 (13.94)	118.77 (13.42)	119.38 (12.1)	0.35
DBP, mmHg	509	75.77 (8.88)	74.81 (9.23)	75.39 (8.89)	0.62
Blood pressure treatment, yes	509	51 [29.82%]	56 [32.94%]	48 [28.57%]	0.67
Total cholesterol, mg/dl	509	202.99 (39.36)	201.86 (34.68)	198.23 (35.16)	0.46
HDL cholesterol, mg/dl	509	48 (14.31)	49.87 (14.18)	47.32 (12.99)	0.22
LDL cholesterol, mg/dl	509	126.4 (34.66)	124.79 (30.45)	122.45 (30.75)	0.52
Log-transformed triglyceride, mg/dl	509	4.86 (0.52)	4.78 (0.55)	4.85 (0.51)	0.35
Fasting glucose, mg/dl	509	91.79 (11.22)	93.66 (13.6)	91.94 (10.65)	0.27
HbA1C, %	508	5.45 (0.39)	5.42 (0.45)	5.36 (0.36)	0.10
Log-transformed -Insulin, μU/m	507	2.35 (0.54)	2.33 (0.58)	2.37 (0.6)	0.80
Log-transformed -HOMA-IR	507	0.86 (0.59)	0.86 (0.65)	0.89 (0.65)	0.89
Current smoker, yes	509	5 [2.92%]	7 [4.12%]	5 [2.98%]	0.79
ASCVD%c	496	4.57 (4.92)	3.9 (3.98)	4.29 (4.57)	0.40
Physical activity indices
Total index	509	7.15 (1.39)	7.04 (1.49)	6.94 (1.4)	0.40
Sport index	509	2.34 (0.69)	2.25 (0.67)	2.19 (0.65)	0.12
Leisure time index	509	2.74 (0.68)	2.71 (0.71)	2.69 (0.68)	0.78
Work index	509	2.07 (0.71)	2.08 (0.76)	2.06 (0.77)	0.97
Sleep	509				0.80
Not at all	292	95 [55.56%]	102 [60%]	95 [56.55%]
Slight amount	156	54 [31.58%]	49 [28.82%]	53 [31.55%]
Moderate amount	54	20 [11.7%]	18 [10.59%]	16 [9.52%]
Great amount	7	2 [1.17%]	1 [0.59%]	4 [2.38%]

Data are mean (SD) or N [% of the number of study participants].

BMI, body mass index; SBP/DBP, systolic/diastolic blood pressure; HDL, high-density lipoprotein; LDL, low-density lipoprotein; HbA1c, glycated hemoglobin; HOMA-IR, Homeostasis model assessment-of-insulin resistance; ASCVD, 10-year risk for atherosclerotic cardiovascular disease.

^alog-transformed miR-19a-3p and miR-19b-3p.

^bData on dietary intake per day were available for 267 individuals: n = 92/91/84 across the 3 tertiles, respectively.

^cData on ASCVD were available for 496 individuals: n = 165/165/166 across the 3 tertiles, respectively.

Circulating miR-19a-3p and miR-19b-3p levels significantly decreased from baseline to 6 months (difference [SD] −0.06 [0.51], P = 0.008 for miR-19a-3p; and difference [SD] −0.09 [0.54], P = 0.0004 for miR-19b-3p). We found that a greater reduction of miR-19a-3p or miR-19b-3p was related to a greater reduction in ASCVD risk (β[SE] 0.33 [0.13], P = 0.013; β[SE] 0.3 [0.13], P = 0.02 for miR-19a-3p and miR-19b-3p respectively) over 6 months (Table 2). Similar linear associations were observed for blood pressure; the decrease in miR-19a-3p or miR-19b-3p was associated with reductions in SBP (β [SE] 1.96 [0.86], P = 0.023; β [SE] 1.63 [0.84], P = 0.054) and DBP (β [SE] 1.67 [0.65], P = 0.011; β [SE]0.1.6 [0.64], p = 0.013) at 6 months (Table 2). Further adjustment for concurrent weight loss in model 2 did not change the significance of relationships (Table 2). However, no associations were observed between the 6-month change in miR-19a-3p or miR-19b-3p with changes in ASCVD risk and each component from baseline to 24 months.

Table 2.

Associations of changes in circulating miR-19a-3p and miR-19b-3p with change in ASCVD risk score and individual ASCVD risk score components over 6 months.

	N	Change in miR-19a-3p		Change in miR-19b-3p
Model 1		β (SE)	P	β (SE)	P
Δ ASCVD (%)a	465	0.33 (0.13)	0.013	0.3 (0.13)	0.020
Δ SBP (mmHg)	509	1.96 (0.86)	0.023	1.63 (0.84)	0.054
Δ DBP (mmHg)	509	1.67 (0.65)	0.011	1.6 (0.64)	0.013
Δ Total cholesterol (mg/dl)	509	1.52 (2.68)	0.572	2.32 (2.63)	0.378
Δ HDL cholesterol (mg/dl)	509	0.5 (0.68)	0.457	0.35 (0.66)	0.603
Δ Glucose (mg/dl)	509	0.29 (0.82)	0.728	0.09 (0.8)	0.909
Δ HbA1C (%)	508	0.05 (0.03)	0.115	0.06 (0.03)	0.071
Model 2
Δ ASCVD (%)a	465	0.33 (0.13)	0.010	0.3 (0.12)	0.017
Δ SBP (mmHg)	509	1.94 (0.84)	0.021	1.58 (0.82)	0.056
Δ DBP (mmHg)	509	1.64 (0.61)	0.007	1.51 (0.6)	0.011
Δ Total cholesterol (mg/dl)	509	1.48 (2.54)	0.560	2.12 (2.48)	0.393
Δ HDL cholesterol (mg/dl)	509	0.51 (0.67)	0.452	0.36 (0.66)	0.585
Δ Glucose (mg/dl)	509	0.27 (0.79)	0.729	0.04 (0.78)	0.961
Δ HbA1C (%)	508	0.05 (0.03)	0.113	0.05 (0.03)	0.075

β (SE) per log-transformed miR-19a-3p or miR-19b-3p change; N, Number of participants eligible for the analysis; ASCVD, 10-year risk for atherosclerotic cardiovascular disease; SBP/DBP, systolic/diastolic blood pressure; HDL, high-density lipoprotein; HbA1c, glycated hemoglobin.

Model 1: Data with GLM models adjusting for age, sex, ethnicity, diet group, blood collection site, baseline body mass index, respective baseline miR-19, and variable of interest at the baseline examination.

Model 2: Model 1+ concurrent weight loss.

^aage, sex, and ethnicity were removed from model 1 and model 2 for ASCVD.

We next assessed the modifying effects of sleep disturbance, physical activity, and dietary interventions varying in macronutrient intakes on the associations between changes in miR-19 and changes in ASCVD risk. During the initial 6 months, we observed significant interactions between change in miR-19a-3p or miR-19b-3p and sleep disturbance for the change in ASCVD risk (P interaction = 0.007 for miR-19a-3p and P interaction = 0.016 for miR-19b-3p in Fig. 1; P interaction = 0.019 for miR-19a-3p and P interaction = 0.116 for miR-19b-3p in Fig. 2). In addition, the data revealed that sleep disturbance modified the associations of the initial 6-month change in miR-19a-3p or miR-19b-3p with change in ASCVD risk from baseline to 24 months (P interaction = 0.01 for miR-19a-3p and P interaction = 0.008 for miR-19b-3p, Fig. 2). Among participants without sleep disturbance, the greater decrease in miR-19 was associated with a greater reduction in ASCVD risk at both 6 months and 24 months, although no significant associations were observed in the slight/moderate/great amounts of sleep disturbance group (Figs. 1 and 2). We also found a comparable significant modification effect of sleep disruption on the relationship between change in miR-19 and change in SBP (P interaction = 0.032 for miR-19a-3p and P interaction = 0.037 for miR-19b-3p in Figure S1).

Fig. 1.

Changes in ASCVD risk per log-transformed miR-19a-3p or miR-19b-3p change according to sleep disturbances over 6 months. Data are β ± SE after adjustment for diet group, blood collection site, baseline body mass index, baseline sleep disturbance, respective baseline miR-19 level, and baseline ASCVD score.

Fig. 2.

Changes in ASCVD risk from baseline to 6 months and 24 months according to 6-month change in miR-19a-3p or miR-19b-3p in response to different sleep disturbances groups. Data are means ± SE 6-month and 24-month change in ASCVD risk according to 6-month change in miR-19a-3p (A, C) and miR-19b-3p (B, D) for No or Yes (light/moderate/great amount) of sleep disturbance group after adjustment for diet group, blood collection site, baseline body mass index, baseline sleep disturbance, respective baseline miR-19 level, and baseline ASCVD score.

Modification effects by change in physical activity were significant for the association between the 6-month change in miR-19 and the change in ASCVD risk from baseline to 6 and 24 months. Participants with a greater decrease in miR-19 among the group with an increase in physical activity had a significantly greater reduction of ASCVD risk from baseline to 6 months (P interaction = 0.013 for miR-19a-3p and P interaction = 0.118 for miR-19b-3p) and 24 months (P interaction = 0.006 for miR-19a-3p and P interaction = 0.004 for miR-19b-3p) than those without an increase in physical activity (Fig. 3).

Fig. 3.

Changes in ASCVD risk from baseline to 6 months and 24 months according to 6-month change in miR-19a-3p or miR-19b-3p in response to different physical activity groups. Data are means ± SE 6-month and 24-month change in ASCVD risk according to 6-month change in miR-19a-3p (A, C) and miR-19b-3p (B, D) for no change/decrease or increase in physical activity group after adjustment for diet group, blood collection site, baseline body mass index, baseline physical activity, respective baseline miR-19 level, and baseline ASCVD score.

Moreover, the association between the change in miR-19 and the change in ASCVD risk was not modified by dietary interventions, such as fat and protein intake (P interaction>0.05 for all, Figures S2&S3). Even though there was no evidence of modifying effect by dietary interventions on the associations between change in miR-19 and change in ASCVD, we observed a positive tendency for greater reduction in ASCVD risk along with a greater decrease in miR-19 among participants with low fat or average protein diet intake.

4. Discussion

In this study, we found that a decrease in circulating miR-19 was associated with a reduction in ASCVD risk in response to weight-loss diets. Specifically, participants who experienced a greater reduction in miR-19a-3p or miR-19b-3p levels had a greater reduction in ASCVD risk from baseline to 6 months. In addition, the relations of change in miR-19 with change in ASCVD risk were significantly modified by the degree of sleep disturbance and change in physical activity. In participants without sleep disturbance or with increased physical activity, a greater decrease in miR-19 was associated with a significantly greater reduction in ASCVD risk, while no such associations were observed in those with sleep disturbance or without an increase in physical activity.

In alignment with the results demonstrated in our study, it has been reported that miR-19 may serve as a biomarker for predicting ASCVD risk. Patients with cardiovascular diseases [13], such as atrial fibrillation [14], myocardial infarction [12,15], acute coronary syndromes [16], heart failure [7,10], unstable angina pectoris [17], and coronary artery disease [11] displayed upregulated levels of miR-19 in the serum or platelet microparticles. In contrast, Yao et al. found that circulating miR-19b levels significantly decreased in patients with heart failure among those who had coronary heart disease [18]. Another study reported that patients with coronary heart disease had downregulated levels of miR-19a, but the decrease disappeared during the validation stage [19]. Additionally, Mayer et al. found that low expression of miR-19a contributed to a substantial additive mortality risk in patients with stable CVD, suggesting a cardioprotective role of miR-19a in the pathogenesis of CVD [8]. The discrepant results from previous experimental and epidemiological studies [28] may be due to differences in methodology for miRNA assessment, study design, study population, and sample size.

To our knowledge, this is the first study to investigate the associations of changes in circulating miR-19 levels and temporary changes in ASCVD risk in response to weight-loss diets. A greater decrease in circulating miR-19a-3p or miR-19b-3p was significantly associated with a greater reduction in ASCVD risk beyond concurrent weight loss. Recent studies on the pathophysiological roles of miR-19 involved in the development of CVD provide solid evidence to support our findings from a biological perspective. MiR-19 was documented to promote proliferation and migration and accelerate cardiac fibrosis, which is a crucial clinical alteration of heart failure, by targeting the transforming growth factor-beta pathway and phosphatase and tensin homolog [5,7]. Besides fibrosis, it has been found that miR-19 induced the apoptosis of endothelial cells and further led to the development of atherosclerosis [5]. Various immune responses including cell activation, proliferation survival, and inflammation were shown to be modulated by miR-19 [15,29]. In the present study, miR-19 was also proposed to have a positive correlation with blood pressure, suggesting that miR-19’s role in CVD might be mainly mediated by blood pressure-related cardiac mechanisms.

Intriguingly, we observed that sleep disturbance and physical activity significantly modified the association of change in miR-19 with change in ASCVD risk. A greater reduction in ASCVD risk was demonstrated in participants without sleep disturbance or with increased physical activity compared to those with sleep disturbance or without an increase in physical activity. Both sleep health and physical activity are recognized as preventive factors for CVD risk [30–33] and influence levels of miR-19 [23,24]. It was documented that decreased expression of miR-19b has been observed in patients with sleep behavior disorders [23]; and miR-19 has been identified as an aerobic exercise training-induced miRNA [24]. MiR-19 serves as a strong post-transcriptional regulator for circadian rhythm by modulating the circadian transcripts CLOCK and RORα [34,35]. Moreover, healthy sleep and physical activity help to maintain the human circadian rhythm by ameliorating the negative effects of disrupted circadian rhythms [36]. Based on the observed changes in circadian-associated miR-19 levels along with alterations in behaviors related to circadian rhythms, such as sleep and exercise, we postulated that miR-19 might contribute to its predictive effects on ASCVD risk, at least partially, by modulating the circadian rhythm; and the relations between the decrease in miR-19 with the reduction in ASCVD risk might be facilitated by physical activity and healthy sleep, which are known to promote the maintenance of circadian rhythm. These findings add to the existing literature by identifying modifiable factors that may impact the relationship between miR-19 and CVD risk.

Major strengths of our study included a large sample of participants in one of the largest and longest randomized weight-loss dietary interventions and repeated assessments of plasma miRNA level allowing for prospective investigation of the potential role of miR-19 in predicting CVD risk reduction in response to diet interventions. In addition, next-generation sequencing is a promising method for miRNA studies providing advantages in higher sensitivity for measuring differential miRNA expression, wider miRNA spectrum, and the ability of de novo sequencing of so far unknown miRNAs [37]. However, several limitations in this study should be considered. First, we were unable to directly assess incident CVD risk in the intervention trial. However, the estimated ASCVD risk was widely validated and used to estimate the 10-years risk of ASCVD. Second, our study mainly included white participants; whether our findings could be applied to other populations needs to be further examined. Third, our study, focusing on the associations between miR-19 and ASCVD risk, is limited by the lack of functional analyses to elucidate the mechanistic pathways. As a result, the biological implications of miR-19’s role in ASCVD remain speculative. We recommend future research to include functional studies, such as gene expression analysis, to validate and expand upon our findings. Fourth, our study’s scope is limited to the miR-19 family, despite previous evidence indicating the involvement of multiple miRNA families in ASCVD risk. This focus potentially overlooks the complex interplay of various miRNAs in ASCVD. Future studies should consider a broader spectrum of miRNAs to fully understand their collective impact on ASCVD risk. Finally, the absence of validation of our RNA-Seq findings with quantitative real-time PCR (qPCR) is a limitation, and we intend to further confirm the accuracy of our results by conducting qPCR analysis.

In summary, the current study adds novel evidence to the growing body of literature on the association between change in circulating levels of circadian rhythm-associated miR-19 and change in CVD risk and potential modification effects of sleep behaviors and physical activity on these relationships. Our study results offer valuable insights into the potential role of miR-19 as a biomarker in predicting the risk of CVD and forecasting the cardiovascular benefits of weight loss diets.

Appendix A. Supplementary data

Supplementary data to this article can be found online at https://doi.org/10.1016/j.clnu.2024.02.015.

Data availability

Some or all datasets generated during and/or analyzed during the current study are not publicly available but are available from the corresponding author on reasonable request.