miR-3113-5p, miR-223-3p, miR-133a-3p, and miR-499a-5p are sensitive biomarkers to diagnose sudden cardiac death

For the purpose of this study, we collected four categories of human heart samples that were autopsied at the Department of Forensic Medicine, School of Basic Medical Sciences, Gannan Medical University over the period 2012–2019. A total of 34 cases of subjects who died from SCD were selected. These cases included 18 of SCD-negative autopsy that presented none notable heart changes at autopsy and 16 of SCD with evidence of heart dysfunction (SCD-positive autopsy). The causes of SCD were made after systemic autopsy. SCD-negative autopsy was defined as sudden deaths without gross heart pathology, with or without histological evidence of early myocardial ischemia (i.e. wavy fiber, hyperesonophilia of cardiomyocytes, and contraction band necrosis). SCD-positive autopsy was defined as sudden deaths in clinic plus severe coronary artery stenosis at autopsy, often with gross or microscopic scarring within myocardium. In addition, 14 cases of fatal injury and 14 cases of carbon monoxide (CO) intoxication were selected as the year-matched control cases. In detail, the heart specimens from fatal injury served as negative control due to none myocardial ischemia in such conditions, and the heart specimens from acute CO intoxication were assigned as positive control due to the global, but not heart-restricted, deficiency of oxygen in such circumstances. Patients who had pre-operative chemo- or radiotherapy history were excluded. Any case with heart implantation was also excluded. All of these cases undergone autopsy examination and were also excluded from influence of other potential substances such as alcohol, illicit drugs, or psychoactive drugs. Detailed demographic and case information regarding the four categories of cases were documented in Table 1. Detailed heart pathology of the four groups of cases were recorded in Table 2. The use of human heart samples for the purpose of research was approved by the Ethical Review Board at the School of Basic Medical Sciences, Gannan Medical University.

All histological evaluation was performed using the human heart samples dissected from the apex of each heart. Due to difficulty in obtaining fresh surgical heart samples, we used the autoptic formalin-fixed paraffin-embedded (FFPE) samples for the purpose of this study. Unlike mRNAs or proteins that are prone to degradation and chemical modification, miRNAs are very small in size (~ 22 nt), less likely to be degraded, and easier to recover from FFPE samples, all of which confer to miRNAs high forensic relevance [27]. The extraction of miRNAs from autoptic FFPE samples was commonly documented in publications [28]. After histological staining, independent pathologists were recruited to review the Hematoxylin & Eosin (H&E) staining sections and PicroSirius Red (PSR) staining sections using a multi-head microscope to result in consensus.

For the H&E staining, human heart tissues were immediately fixed in 4% paraformaldehyde for 24 h, followed by embedding in paraffin. Tissues were then sliced into 5 μm-thick heart slices. Then the slices were dewaxed and dehydrated in a gradient alcohol. Slices were then stained with hematoxylin solution for 10 min at room temperature and incubated in 1% hydrochloric-alcohol solution. Then, slices were washed using tap water, stained with eosin solution for 5 min, and separated by 75, 85, 90, 95 and 100% alcohol solutions for 2 min, respectively. Next, the slices were dried and sealed with neutral gums. After H&E staining, the morphological changes were observed and photographed under an optical microscope (Olympus, Tokyo, Japan).

For the PSR staining, human heart slices were deparaffinized to water and then immersed in 0.1% Sirius red staining solution dissolved in picric acid for 1 h. Then slices were washed in acidified water containing 1.5% hydrogen chloride, dehydrated and mounted. Collagen and non-collagen components were red- and orange-stained, respectively.

Heart slices from the four categories were also subject to IHC analysis. Briefly, the slides were incubated in 3% H₂O₂ solution for 10 min and antigen retrieval was performed by steam heating in 10 mM citrate buffer (pH 6.0) for 30 min. After epitope recovery, the slides were then treated with 10% of normal goat serum for 60 min, followed by incubation with active caspase 3 antibody (1:500, Cell Signaling Technology, Catlog #9664, MA, USA), CD31 antibody (1:1000, Cell Signaling Technology, Catlog #3528), and CD68 antibody (1:500, Cell Signaling Technology, Catlog #76437) incubation overnight at 4 °C. The slides were washed and incubated with secondary horseradish peroxide (HRP)-linked secondary antibody (Vector Laboratories, MN, USA) at 1:500 dilutions for 1 h. The samples were treated with the chromogen DAB for antigen detection and the final counterstaining was performed with hematoxylin.

All paraffin was removed from the FFPE heart sections by treating with Deparaffinization Solution. The extraction of miRNAs from FFPE tissues was performed using a commercial miRNeasy FFPE kit (Qiagen, Catlog #217504, Hilden, Germany). Briefly, samples were incubated by heat treatment in an optimized lysis buffer, which contains proteinase K, to release RNAs from the sections. Supernatant was collected and treated with a DNase, followed by buffer RBC and ethanol treatments. The cleaned samples were then applied to an RNeasy MiniElute spin column, where the total RNAs including miRNAs, bound to the membrane and contaminants were efficiently washed away. Total RNAs including miRNAs were then eluted in a minimum of 14 μL of RNase-free water. Reverse transcription of the total RNA was performed using a Mir-X miRNA First-Strand Synthesis Kit (Takara, Tokyo, Japan) for RT-qPCR of miRNA. The synthesized cDNA was stored at − 80 °C for later use.

QuantiFast Multiplex RT-PCR Kit (Qiagen) was used for RT-qPCR analysis. According to the kit instructions, an aliquot of 25 μL reaction system was mixed and amplified using the Applied Biosystem 7500 fluorescence quantitative PCR instrument (Darmstadt, Germany). Primers used for this study were listed in Table 3. The 2^-ΔCt method was used to calculate the relative expression of the target genes where ΔCt_miRNA = Ct_miRNA − C_{thousekeeping.}

All data were expressed as mean ± standard error of the mean (SEM). Statistical analysis was performed using independent Student’s t test. The diagnostic powers of miR-3113-5p, miR-223-3p, miR-133a-3p, and miR-499a-5p were evaluated by receiver operating characteristic (ROC) analysis. Areas under the curve (AUCs) were calculated. All statistical analyses were performed using GraphPad Prism 8.0 software (La Giolla, CA, USA). P < 0.05 was considered as statistically significant.

As shown in Table 1, the four categories of cases showed similar survival intervals (within 2 h). The cases from the SCD-positive autopsy group had significant increases in the heart weight as compared with both control cases (p < 0.001). The heart weight of the SCD-negative autopsy group was, however, significantly lower than that of the SCD-positive autopsy group (p < 0.05), though it was still higher than both control cases (Table 1). Furthermore, SCD-negative autopsy cases were not identified of any gross cardiovascular change as similar with control cases, while 56.2% (n = 9) of SCD-positive autopsy cases showed coronary stenosis (50–75% stenosis), calcification and myocardial fibrosis, and the remaining 43.8% (n = 7) showed coronary stenosis (> 75%), severe myocardial fibrosis and disorder (Table 2 and Fig. 1). The myocardium from the SCD-positive autopsy group was disarrayed with a halo of fibrotic tissues within the degenerated myocardium (Fig. 1). In consistent with the short survival time, 88.9% of SCD-negative autopsy cases showed histological evidence of acute myocardial ischemia (within 2 h) such as interstitial edema, wavy fibers, and hypereosinophilia of cardiomyocytes (Table 2). No evident fibrotic tissue was observed for the SCD-negative autopsy group (Fig. 1).

To further examine molecular alterations in the heart samples of SCD, we assayed the common molecular biomarkers of myocardial injury, including cleaved-Caspase3 (cl-Casp3), CD31, and CD68. Cl-Casp3 is a marker of early myocardial injury and cell apoptosis. CD68 is a marker of macrophage infiltration which often occur one or two days after injury. CD31 is involved in angiogenesis, leukocyte transmigration and integrin activation, and the changes of CD31-positive cells are commonly observed during cardiac remodeling [29]. As expected, the heart samples from the SCD-positive autopsy group showed notable positivity of cl-Casp3 and CD68. Neovascularization was also evident in the SCD-positive autopsy group as by IHC staining of CD31. The SCD-negative autopsy group showed patchy positivity of cl-Casp3 and dim CD31 signal, and was absent of CD68 positivity. These changes were milder as compared with the SCD-positive autopsy group, though they were to some extent pathological as compared with the fatal injury and CO intoxication cases that were consistently absent of cl-Casp3, CD31, and CD68 positivity (Fig. 2). These data suggested that SCD-negative autopsy was hard to be discriminated using the histological and IHC assays.

In view that common histological and IHC techniques could not aid in diagnosing SCD-negative autopsy, we speculated that small-molecules might be helpful. We thus assessed whether four cardio-miRNAs (miR-3113-5p, miR-223-3p, miR-499a-5p, and miR-133a-3p) had any diagnostic values. The selection of these miRNAs was based on published literature [28, 30, 31] and our previous work [26] . Our aims were two-fold: one to assess whether these miRNAs could diagnose SCD, and the other one to assess whether specific miRNAs had higher efficacy as to discriminate the causes of SCD. To this end, we initially pooled the SCD-positive autopsy and SCD-negative autopsy groups as SCD cases. RT-qPCR was performed with U6 as a stable reference gene. Our results revealed that miR-3113-5p was significantly elevated by approximate five-fold in the SCD groups (Fig. 3A). miR-223-3p and miR-499a-5p were also increased by approximate five-fold (Fig. 3B and C). miR-133a-3p showed significant increases for approximate three-fold when compared to both controls (Fig. 3D).

We next assessed the diagnostic values of these miRNAs for distinguishing SCD from non-cardiac deaths. Receiver operating characteristic (ROC) curve analysis showed that the area under the curve (AUC) of miR-3113-5p were 0.7893 (SCD vs Fatal injury: 95% confidence interval (CI) = 0.6501 to 0.9176, sensitivity = 68.75% and specificity = 83.33%) and 0.8281 (SCD vs CO intoxication: 95% CI = 0.7071 to 0.9492, sensitivity = 68.75% and specificity = 91.67%) (Fig. 4A and B). The AUCs of miR-223-3p were 0.9043 (SCD vs Fatal injury: 95% CI = 0.8031 to 1.006, sensitivity = 83.33% and specificity = 88.89%) and 0.8917 (SCD vs CO intoxication: 95% CI = 0.7984 to 0.9849, sensitivity =75% and specificity = 100%) (Fig. 4C and D). The AUCs of miR-499a-5p were 0.8971 (SCD vs Fatal injury: 95% CI = 0.7972 to 0.9969, sensitivity = 82.35% and specificity = 90%) and 0.8853 (SCD vs CO intoxication: 95% CI = 0.7771 to 0.9935, sensitivity =97.06% and specificity = 70%) (Fig. 4E and F). The AUCs of miR-133a-3p were 0.7851 (SCD vs Fatal injury: 95% CI = 0.6466 to 0.9235, sensitivity = 67.65% and specificity =76.92%) and 0.8743 (SCD vs CO intoxication: 95% CI = 0.7693 to 0.9793, sensitivity =64.71% and specificity = 100%), respectively (Fig. 4G and H). Detailed ROC analysis results were listed in Table 4. These results demonstrated that miR-3113-5p, miR-223-3p, miR-499a-5p, and miR-133a-3p could be used as sensitive biomarkers for diagnosis of SCD.

To precisely evaluate the power of the four miRNAs in predicting SCD, SCD was divided into SCD-positive autopsy (with gross coronary artery stenosis) and SCD-negative autopsy (without gross heart changes). Meanwhile, the expression patterns of the four miRNAs were detected in SCD-positive autopsy and SCD-negative autopsy groups by RT-qPCR. Interestingly, miR-3113-5p (Fig. 5A), miR-223-3p (Fig. 5B), miR-499a-5p (Fig. 5C), and miR-133a-3p (Fig. 5D) were all significantly higher in the SCD-negative autopsy group than those in the SCD-positive autopsy group with the fold-change ranging from two to three (p < 0.05).

Next, ROC curve analyses were used to assess the diagnostic ability of the four miRNAs for discriminating SCD-negative autopsy from SCD-positive autopsy. It revealed that neither of these miRNAs alone efficiently discriminated SCD-negative autopsy (AUC was 0.7373 for miR-3113-5p, AUC was 0.7200 for miR-223-3p, AUC was 0.7569 for miR-133a-3p, and AUC was 0.7951 for miR-499a-5p), indicating that SCD-negative autopsy and SCD-positive autopsy both shared common pathologic basis, albeit SCD-negative autopsy occurring in a sudden and unidentified way. We then assessed the ROC by combing these miRNAs. Our results showed that the AUC of a combination of miR-3113-5p and miR-223-3p was 0.8667 (95% CI = 0.7388 to 0.9945, sensitivity = 82.35% and specificity =80%, Fig. 6A). The AUC of a combination of miR-3113-5p and miR-499a-5p was 0.8510 (95% CI = 0.7105 to 0.9915, sensitivity = 100% and specificity =66.67%, Fig. 6B). The AUC of a combination of miR-3113-5p and miR-133a-3p was 0.7686 (95% CI = 0.6032 to 0.9341, sensitivity =70.59% and specificity =80%, Fig. 6C). The AUC of a combination of miR-499a-5p and miR-133a-3p was 0.7569 (95% CI = 0.5938 to 0.9201, sensitivity = 66.67% and specificity =81.25%, Fig. 6D). The AUC of a combination of miR-223-3p and miR-499a-5p was 0.7741 (95% CI = 0.6062 to 0.9419, sensitivity = 88.89% and specificity =60%, Fig. 6E). The AUC of a combination of miR-223-3p and miR-133a-3p was 0.7407 (95% CI = 0.5700 to 0.9115, sensitivity = 66.67% and specificity =80%, Fig. 6F). The details were shown in Table 5. These results demonstrated that miR-3113-5p and miR-223-3p together yielded the highest diagnostic performance in discriminating SCD-negative autopsy from SCD-positive autopsy (maximum of AUC). miR-3113-5p and miR-499a-5p together also showed high AUC values and sensitivity (maximum of Youden’s index).