Yue Yu, Xiaoyan Sun and Xianghai Zhou*

Department of Endocrinology and Metabolism, Peking University People's Hospital, Xicheng, Beijing 100044, China

*Corresponding Author:
Xianghai Zhou
Department of Endocrinology and Metabolism, Peking University People's Hospital, Xicheng, Beijing 100044, China
E-mail: xianghaizhou1@163.com
This article was originally published in a special issue, “Innovations in Biomedical Research and Drug Development”
Indian J Pharm Sci 2023:85(3) Spl Issue “211-222”

This is an open access article distributed under the terms of the Creative Commons Attribution-NonCommercial-ShareAlike 3.0 License, which allows others to remix, tweak, and build upon the work non-commercially, as long as the author is credited and the new creations are licensed under the identical terms

DOI: 10.36468/pharmaceutical-sciences.spl.694

Abstract

Non-alcoholic fatty liver disease is a high-risk factor for the morbidity and mortality of type 2 diabetes mellitus. Thus, we conducted a meta-analysis to investigate whether the microRNAs associated with non-alcoholic fatty liver disease are associated with diabetes and their potential to predict diabetes in non-alcoholic fatty liver disease individuals. We conducted a systematic literature search using PubMed, Cochrane library and Embase databases. The literature about circulating microRNA levels in nonalcoholic fatty liver disease individuals was screened according to the inclusion and exclusion criteria. The sensitivity, specificity and area under curve of different microRNAs were summarized. These microRNAs with higher diagnostic efficacy and stability were selected for further analysis in people with diabetes. In all, 10 of 1950 studies determining microRNA expression profiles in non-alcoholic fatty liver disease were included in the final analysis. Using area under curve ≥0.7 as the criterion, we screened out 10 microRNAs that have moderate diagnostic efficacy in non-alcoholic fatty liver disease, including microRNA-122, microRNN-34a, microRNA-99a, microRNA-16, microRNA-379, microRNA-197, microRNA-181d, microRNA-146b, microRNA-375 and microRNA-12. The pooled area under curve of microRNA-122 and microRNA-34a in the diagnosis of non-alcoholic fatty liver disease was 0.88 (95 % confidence interval: 0.81–0.96) and 0.86 (95 % confidence interval: 0.82–0.91), respectively. Further, 30 of 2674 studies related to diabetes and the above microRNA were included in the final analysis. The results showed that microRNA-122 and microRNA-34a showed higher stability in non-alcoholic fatty liver disease and diabetes. In conclusion, microRNA-122 and microRNA-34a are common biomarkers of both non-alcoholic fatty liver disease and type 2 diabetes mellitus. However, whether they can be used as effective biomarkers for the early prediction of diabetes in patients with non-alcoholic fatty liver disease lacks valid evidence.

Keywords

Meta-analysis, microRNA, non-alcoholic fatty liver disease, type 2 diabetes mellitus

Pathologically, Nonalcoholic Fatty Liver Disease (NAFLD) is characterized by >5 % hepatic fat accumulation without other recognized causes of fatty liver, such as alcohol, viruses, drugs and autoimmunity[1]. Currently, it is the most common liver disorder worldwide and the leading cause of chronic liver diseases[2]. By 2030, it is predicted to be the predominant cause of orthotropic liver transplantation[3]. Diabetes is a metabolic disorder with rapidly rising prevalence worldwide and is characterized by chronically elevated blood glucose levels due to pancreatic beta (β)-cell dysfunction or insulin resistance[4,5]. Although the underlying relationship between NAFLD and Type 2 Diabetes Mellitus (T2DM) is complicated and controversial, there is a consensus that NAFLD increases the risk of T2DM[1,6-8]. Retrospective analysis of 2920 participants revealed that the non-obese NAFLD individuals had a significantly higher risk of approximately twofold for DM than the non-obese controls[9]. In a cohort study of 129 adults with biopsy-proven NAFLD, researchers found that the prevalence rate of T2DM increased from 8.5% at baseline to 78 % in 13 y, including patients who developed either T2DM (58 %) or impaired glucose tolerance (20 %)[10]. A meta-analysis of a pooled population of 117 020 patients summarized that NAFLD was associated with an increased risk of incidence of T2DM with a pooled relative risk of approximately twofold[11]. Furthermore, diabetic patients with NAFLD frequently have poorer glycemic control compared with patients without NAFLD and NAFLD was also associated with worse all-cause and cardiovascular mortality[12-14]. An investigation of 167 621 diabetic individuals revealed that they have an approximately threefold higher risk of dying due to chronic liver diseases, which is largely attributed to NAFLD[15]. In a cohort study of 2103 patients with T2DM, the prevalence of Chronic Kidney Disease (CKD) was almost twice in those with ultrasonography-diagnosed NAFLD than in those without NAFLD[16]. Considering the high risk and poor clinical outcomes associated with NAFLD in diabetic patients, it is imperative to identify biomarkers that can predict the risk of diabetes before the onset of glycemic abnormalities. Few studies have examined the common markers of diabetes and NAFLD and even fewer have investigated the appropriate markers for predicting diabetes in patients with NAFLD.

MicroRNAs (miRNAs) are small endogenous non- coding Ribonucleic Acid (RNA) with a length of about 22–25 nucleotides that regulate target gene expression at the post-transcriptional level[17,18]. Mostly, a single miRNA targets various messenger RNA (mRNAs) and influences the expression of several genes that are often involved in an interconnected pathway[19]. Owing to the remarkable stability that protects them from degradation by endogenous ribonucleases, circulating miRNAs have greater application compared with other molecular markers[20,21]. Researchers found that parts of specific miRNAs can influence various pathophysiological processes, such as the metabolism of lipids and glucose, chronic inflammation and liver function[22,23], and have used them as biomarkers to diagnose diseases, including NAFLD and other metabolic diseases[24-27]. Therefore, we conducted a meta- analysis to investigate whether miRNAs associated with NAFLD are associated with diabetes. This will eventually aid in identifying valid biomarkers to predict diabetes in patients with NAFLD.

Materials and Methods

Search strategies:

We comprehensively retrieved data from the PUBMED, EMBASE and Cochrane databases. Firstly, we explored studies concerning miRNA expression profiling in NAFLD using the following MeSH terms and keywords; “miRNA”, “NAFLD” and a combination of these terms. All studies published from inception to 7th November 2021 were included. The first round of literature screening and analysis identified target miRNAs for the second round of research. In the second step, studies relevant to specific miRNA expression profiling in patients with T2DM were searched using the following MeSH terms; “miRNA-12”, “miRNA-122”, “miRNA-146”, “miRNA-16”, “miRNA-181”, “miRNA-197”, “miRNA-34a”, “miRNA-375”, “miRNA-379” and “miRNA-99a” in combination with “T2DM” and “noninsulin-dependent diabetes mellitus” in December 2021. The language of searching records was not limited and all retrieved items were managed with Endnote X9.

Study selection:

Two independent investigators, namely Yu and Sun, manually screened the studies based on the following inclusion and exclusion criteria. In case of a discrepancy, it was resolved by the corresponding author Zhou. We included studies that reported miRNA expression profiles in adult patients with NAFLD or T2DM, age range 18-65 y. The exclusion criteria were as follows; duplicate reports; studies conducted on animals or cell lines; conference abstracts, case reports, comments, letters to the editors, systematic reviews and meta- analyses; studies that used tissue samples other than serum, plasma or blood; studies without data proving diagnostic efficacy in retrieval results of NAFLD and studies without relevant data about target miRNAs in retrieval results of T2DM.

Data extraction and literature quality assessment:

We extracted the following information and data from the full text and supplementary materials of each study. First author, journal, year of publication, country of the study, sample type, miRNA expression assay type, sample size, the direction of expression difference, cut-off criteria of deregulated miRNAs and fold-change from miRNA expression profile studies. Moreover, we obtained relevant data proving diagnostic efficacy, including Sensitivity (SEN), Specificity (SPE), Positive Likelihood (PLR), Negative Likelihood (NLR), Diagnostic Odds Ratio (DOR) and the area under the Receiver Operating Characteristic (ROC) and Area Under the Curve (AUC). In addition, for articles without clear data reports, we extracted data from the graphical plots to calculate the mean fold changes using Web Plot Digitizer (version 4.5)[24]. Different miRNA names used in various studies were standardized according to miRBase version 22.1[28]. We evaluated the methodological quality of the included studies using the Quality Assessment of Diagnostic Accuracy Studies-2 (QUADAS-2) criteria[29]. It assesses the research bias based on four domains; patient selection, index test, reference standard and flow timing. Furthermore, the first three domains were also used to evaluate the applicability of the articles[29].

Statistical analysis:

This study was mainly conducted in two stages. First, we selected all efficient miRNAs in the diagnosis of NAFLD followed by identifying their roles in T2DM. For selecting miRNAs in the first step, we ranked them according to the vote-counting strategy[30,31]. According to this strategy, all miRNAs were ranked in order of importance based on the following criteria; number of studies reporting a miRNA in the same expression direction, total sample size of the case and control in the studies, and mean fold changes. We used AUC as the dominant parameter of diagnostic power. For miRNAs with more than one study reporting AUC, we built the diagnostic 2×2 contingency table based on the extracted data and calculated the pooled AUC. In addition, we assessed the heterogeneity among studies using I2 statistics. If I2>50 %, heterogeneity was considered statistically significant and a random effects model was used. After identifying all useful miRNAs in diagnosing NAFLD, the second round of retrieving, screening and extracting was performed to confirm their role in T2DM. All the statistical analyses were displayed using Review manager version 5.4 and Meta-DiSc version 1.4. p<0.05 was considered statistically significant.

Results and Discussion

In all, we obtained 1950 studies that explored miRNA expression profiles in NAFLD. Of these, 557 duplicate records were recognized and removed using EndnoteX9. Further, during the primary screening of the titles and abstracts, 1322 records were excluded for unfit literary forms (n=774), animal or cellular studies (n=273) and irrelevant studies (n=275). Next, we conducted a full-text review of the remaining 71 records, among which 61 were excluded for irrelevant diseases (n=24), non-circulating samples except serum, plasma, and blood (n=16), and irrelevant data (n=21). Ultimately, 10 studies were included in the analysis[32-42]. Fig. 1 shows the flow diagram of the process of literature screening. The primary characteristics of the studies included are summarized in Table 1, and they are arranged by publication year, from 2011-2020. The included studies provided information on 700 patients with NAFLD and 417 healthy individuals. Two studies were conducted in China, two in the United States of America (USA), and others were conducted in Turkey, Philippines, Spain, Thailand, Egypt and Japan. Serum was used as the circulating sample and miRNA expression was measured by Real-Time quantitative Polymerase Chine Reaction (RT-qPCR) in all studies. Researchers typically detected more than one miRNA in their studies. We included only miRNAs with clear reports of diagnostic performance (SEN, SPE or AUC) (Table 1).

Study Country Sample type Measurement methods Sample size No. of miRNAs miRNAs
NAFLD Control
Cermelli[32] USA Serum RT-qPCR 34 19 4 miR-122, miR-16, miR-34a and miR-21
Tan[34] China Serum RT-qPCR 152 90 6 miR-122-5p, miR-1290, miR-27b-3p, miR-192-5p, miR-99a-5p and miR-148a-3p
Celikbilek[41] Turkey Serum RT-qPCR 20 20 4 miR-197, miR-146b, miR-181d, miR-99a
Salvoza[35] Philippines Serum RT-qPCR 28 36 2 miR-34a and miR-122
Liu[38] China Serum RT-qPCR 48 37 2 miR-34a
Auguet[42] Spain Serum RT-qPCR 61 61 1 miR-122
Jampoka[39] Thailand Serum RT-qPCR 58 34 3 miR-29a, miR-29c and miR-122
Hendy[40] Egypt Serum RT-qPCR 210 90 3 miR-122, miR-34a and miR-99a
Pillai[36] USA Serum RT-qPCR 10 20 5 miR-122, miR-34a, miR-375, miR-16 and miR-12
Okamoto[37] Japan Serum RT-qPCR 79 10 1 miR-379

Table 1: Characteristics of the Studies Included In the miRNA Profiles of NAFLD

IJPS-selection

Fig. 1: Flow diagram of the selection of studies exploring miRNA profiles in NAFLD

The 10 articles studied in this meta-analysis included 18 circulating miRNAs, including miR-122, miR- 34a, miR-99a, miR-16, miR-99a-5p, miR-27b-3p, miR-192-5p, miR-148a-3p, miR-1290, miR-122-5p, miR-29c, miR-29a, miR-379, miR-197, miR-181d, miR-146b, miR-375 and miR-12, ranked according to the vote-counting strategy (Table 2)[30,31]. Of these circulating miRNAs, six studies reported miR-122 to be significantly up regulated with a total sample size of 661 and a mean fold-change ranging from 3.76-7.20 in patients with NAFLD compared with healthy controls. The pooled AUC of miR-122 in diagnosing NAFLD was 0.88 (95 % Confidence Interval (CI): 0.81–0.96) and the pooled DOR was 18.05 (95 % CI: 8.68–37.54) (fig. 2A). In addition, four studies reported up regulation of miR-34a with a total sample size of 479 and a mean fold-change of 1.93–2.8. The pooled AUC of miR-34a was 0.86 (95 % CI: 0.82– 0.91) and DOR was 14.83 (95 % CI: 9.09–24.17) (fig. 2B). Down regulation of miR-99a expression was reported in two studies with a combined sample size of 340. The other two studies with a combined sample size of 83 revealed significantly higher levels of miR-16 expression and both AUC was higher than 0.85. Other miRNAs were reported only once, which indicates low credibility. Further, taking AUC as a reference for diagnostic power, we selected 10 miRNAs with an AUC ≥0.7, which is regarded as a moderate diagnostic value for distinguishing patients with NAFLD. These miRNAs were, namely, miR- 122, miR-34a, miR-99a, miR-16, miR-379, miR- 197, miR-181d, miR-146b, miR-375 and miR-12.

miRNA Study Sample size Expression direction Mean p Diagnostic power
NAFLD/Control Sen (%) Spe (%) AUC
miR-122 Cermelli 34/19 Up regulated 7.2 <0.0001 - - 0.927
Salvoza 28/36 Up regulated - <0.0001 78.6 77.8 0.858
Auguet 61/61 Up regulated - - 83.1 69.8 0.82
Jampoka 58/34 Up regulated 3.76 <0.001 75 82.4 0.831
Hendy 210/90 Up regulated 5.06 <0.001 92 85 0.92
Pillai 10/20 Up regulated 30.3 <0.05 - - 0.85
miR-34a Salvoza 28/36 Up regulated - <0.0001 75 75 0.781
Liu 48/37 Up regulated 2.8 <0.05 70.4 87.5 0.811
Hendy 210/90 Up regulated 1.93 <0.05 82 79 0.77
Pillai 10/20 Up regulated 8.3 <0.01 - - 0.86
miR-99a Celikbilek 20/20 Down regulated 0.26 0.0262 65 95 0.76
Hendy 210/90 Down regulated 0.82 <0.05 78 76 0.73
miR-16 Cermelli 34/19 Up regulated 5.5 <0.0001 - - 0.962
Pillai 10/20 Up regulated 23 <0.01 - - 0.86
miR-99a-5p Tan 152/90 Up regulated 2.38 <0.01 69.1 41.1 0.559
miR-27b-3p Tan 152/90 Up regulated 2.71 <0.01 59.9 72.7 0,693
miR-192-5p Tan 152/90 Up regulated 2.61 <0.01 34.9 93.3 0.652
miR-148a-3p Tan 152/90 Up regulated 2.04 <0.01 25.7 90 0.54
miR-1290 Tan 152/90 Up regulated 4.05 <0.01 0.586 0.656 0.629
miR-122-5p Tan 152/90 Up regulated 9.27 <0.01 0.934 0.467 0.729
miR-29c Jampoka 58/34 Down regulated 0.37 0.3 0.434 0.6129 0.529
miR-29a Jampoka 58/34 Down regulated 0.08 0.006 0.6087 0.8235 0.679
miR-379 Okamoto 79/10 Up regulated - 0.026 - - 0.72
miR-197 Celikbilek 20/20 Down regulated 0.33 0.0109 0.6 0.95 0.77
miR-181d Celikbilek 20/20 Down regulated 0.18 <0.0001 0.7 0.85 0.86
miR-146b Celikbilek 20/20 Down regulated 0.35 0.0133 0.55 1 0.75
miR-375 Pillai 10/20 Up regulated 34.2 <0.01 - - 0.88
miR-12 Pillai 10/20 Up regulated 9.9 <0.01 - - 0.83

Table 2: A Systematic Review of the Diagnostic Performance of miRNAs in Patients with NAFLD

IJPS-Diagnostic

Fig. 2: Diagnostic accuracy of (A): miR-122 and (B): miR-34a in NAFLD

Overall, we retrieved 2674 studies that explored miRNA expression profiles in T2DM. Of these, 500 duplicate records were removed. Next, we excluded 1943 reports in the initial screening for the following reasons, unfit literary forms (n=820), animal or cellular studies (n=535) and irrelevant studies (n=588). Further, we conducted a full-text review of the remaining 231 records, among which 201 were excluded for the following reasons; non-circulating samples (n=63), no information about target miRNAs (n=118) and lack of valid data verified by RT-qPCR (n=20). Finally, 30 studies were included in the analysis[43-72]. Fig. 3 exhibits the flow chart of the process of literature screening. Table 3 summarizes the main characteristics of the included studies, in order of the publication year, ranging from 2010-2022. In all, 1229 patients with T2DM, 409 pre-diabetic patients and 1659 healthy controls were included. Of the 30 studies, ten were conducted in China, five in the USA and other studies were conducted in different countries. Most studies used serum as the test sample, while plasma and whole blood were also used. The miRNA expression levels were determined by RT-qPCR in all studies. Notably, the majority of studies detected a greater number of miRNAs than those listed here. Considering the purpose of this study, we extracted and listed data of only target miRNAs and only one or two target miRNAs were identified per research.

Study Country Sample type Measurement methods Sample size No. of miRNAs miRNAs
DM Pre-DM Control
Zampetaki[71] USA Plasma RT-qPCR 80   80 1 miR-197
Kong[57] China Serum RT-qPCR 18 18 19 2 miR-34a and miR-375
Sun[65] China Serum RT-qPCR 100   100 1 miR-375
Wang (1)[66] China Plasma RT-qPCR 54 44 53 1 miR-375
Wang (2)[67] China Plasma RT-qPCR 33   119 1 miR-197
Flowers[49] USA Plasma RT-qPCR 72   55 2 miR-122-5p and miR-197-3p
Rojas[45] Ecuador Serum RT-qPCR 56   40 1 miR-122
Lopez[62] USA Serum RT-qPCR 17   20 1 miR-34a
Seyhan[64] USA Serum RT-qPCR 31 12 27 2 miR-375 and miR-34a
Candia[47] Italy Serum RT-qPCR 9 9 9 2 miR-99a-5p and miR-122-5p
Dias[48] South Africa Serum RT-qPCR 4   4 1 miR-379
Jones[55] New Zealand Plasma RT-qPCR 15   12 1 miR-34a
Krauskopf[58] USA Serum RT-qPCR 7   22 1 miR-375
Yin[70] China Plasma RT-qPCR 100   100 1 miR-375
Al-Muhtaresh[43] Bahrain Serum RT-qPCR 30 30 30 1 miR-375
Fomison-Nurse[50] New Zealand Plasma RT-qPCR 39   32 1 miR-34a
Jaeger[54] Switzerland Serum RT-qPCR 43 43 43 1 miR-122
Ashoori[44] Iran Serum RT-qPCR 50   50 1 miR-375
García-Jacobo[51] Mexico Serum RT-qPCR 54 16 35 2 miR-34a and miR-375
Gok[52] Turkey Serum RT-qPCR 25   25 1 miR-34a
Kokkinopoulou[56] Greece Blood RT-qPCR 40   37 1 miR-34a
Meerson[59] Israel Serum RT-qPCR 29   30 1 miR-122-5p
Mononen[60] Finland Blood RT-qPCR 19 207 427 1 miR-122-5p
Regmi[63] China Serum RT-qPCR 50   25 1 miR-122-5p
Banerjee[46] India Plasma RT-qPCR 30   30 1 miR-34a
Hu[53] China Serum RT-qPCR 50   50 1 miR-197-3p
Wu[68] China Serum RT-qPCR 10   21 1 miR-375
Ye[69] China Plasma RT-qPCR 40   40 1 miR-16-5p
Zeinali[72] Iran Serum RT-qPCR 30 30 30 1 miR-122
Nie[61] China Plasma RT-qPCR 94   94 1 miR-122-5p

Table 3: Characteristics of the Studies Included in the Target miRNA Expression Profiles of T2DM

IJPS-selection

Fig. 3: Flow diagram of the selection of studies exploring miRNA profiles in T2DM

From 30 included articles, we acquired data on seven target circulating miRNAs, including miR-375, miR-34a, miR-122, miR-197, miR-16-5p, miR-99a-5P and miR-379. We divided the data into two groups based on the type of disease as follows; T2DM vs. healthy controls (Table 4) and prediabetes vs. healthy controls (Table 5). The miRNAs in both groups were ranked according to the vote-counting strategy[30,31]. miR- 375, which was reported only once in the diagnosis of NAFLD, had the greatest number of 10 studies in T2DM. It was consistently up regulated in seven of the ten studies with a total sample size of 735 and a mean fold-change ranging from 1.72–5.9. miR-34a also showed a consistent up regulation in six studies with a total sample size of 290 and a mean fold-change ranging from 2.04–11.3. For miR-122, five of the nine studies reported a consistent up regulation with a total sample size of 787 and a mean fold-change ranging from 1.53–8.61, while the other four studies revealed no significant change in its expression. miR-197 displayed no significant dysregulation in most studies. The remaining three miRNAs, miR-16- 5p, miR-99a-5P, miR-379, have only been reported by a single study, with lower accuracy and reliability. In the prediabetic group, miR-122 showed consistent up regulation with a sample size of 712, consistent with the T2DM group. However, the direction of expression change was not always the same in the diabetic and prediabetic patients. miR-34a showed no significant dysregulation in all three prediabetic studies and miR-375 was reported with inconsistent directions, implying low credibility and reference value.

miRNA Study Sample size Expression direction Mean p
DM Control
miR-375 Kong 18 19 Up regulated - 0.004
Sun 100 100 Up regulated - 0.0313
Wang 54 53 Up regulated 1.72 <0.05
Seyhan 31 27 n.c. 1.77 n.s.
Krauskopf 7 22 Down regulated - <0.05
Yin 100 100 Up regulated 3 0.026
Al-Muhtaresh 30 30 Up regulated 5.9 <0.001
Ashoori 50 50 Up regulated 2.1 0.02
García-Jacobo 54 35 n.c. - n.s.
Wu 10 21 Up regulated - <0.001
miR-34a Kong 18 19 Up regulated - <0.0001
Lopez 17 20 Up regulated 2.04 0.011
Seyhan 31 27 Up regulated 3.3 0.0184
Jones 15 12 Up regulated 11.3 0.0049
Fomison-Nurse 39 32 Up regulated - <0.01
García-Jacobo 54 35 n.c. - n.s.
Gok 25 25 n.c. - n.s.
Kokkinopoulou 40 37 Down regulated - < 0.0001
Banerjee 30 30 Up regulated >2 0.045
miR-122 Flowers 72 55 n.c. 1.12 n.s.
Rojas 56 40 n.c. 0.86 n.s.
Jaeger 43 43 n.c. - n.s.
Zeinali 30 30 Up regulated 8.61 <0.001
Candia 9 9 Up regulated 1.75 >0.05
Meerson 29 30 n.c. - n.s.
Mononen 19 427 Up regulated 1.53 <0.0001
Nie 94 94 Up regulated 2.82 0.001
Regmi 50 25 Up regulated - <0.05
miR-197 Flowers 72 55 n.c. 1.09 n.s.
Zampetaki 80 80 Down regulated 0.456 0.0013
Wang 33 119 n.c. - n.s.
Hu 50 50 n.c. - n.s.
miR-16-5p Ye 40 40 n.c. 1.711 n.s.
miR-99a-5P Candia 9 9 Up regulated 1.35 >0.05
miR-379 Dias 4 4 n.c. 1.64 n.s.

Table 4: Summary of the Target miRNA Expression Profiles in T2DM NAFLD

miRNA Study Sample size Expression direction Mean p
Pre-DM Control
miR-122 Jaeger 43 43 n.c. - n.s.
Zeinali 30 30 Up regulated 3.73 <0.001
Candia 9 9 Up regulated 2.58 <0.01
Mononen 207 427 Up regulated 1.14 0.006
miR-34a Kong 18 19 n.c. - n.s.
Seyhan 12 27 n.c. 0.86 n.s.
García-Jacobo 16 35 n.c. - n.s.
miR-375 Kong 18 19 n.c. - n.s.
Wang 44 53 Down regulated 0.88 <0.05
Seyhan 12 27 Down regulated 0.65 0.4004
Al-Muhtaresh 30 30 Up regulated >3 <0.05
García-Jacobo 16 35 n.c. - n.s.
miR-99a-5P Candia 9 9 Up regulated 1.74 <0.01

Table 5: Summary of the Target miRNA Expression Profiles in Pre-Diabetes

According to previous data and reports, miRNAs can be used as diagnostic markers of NAFLD. This study conducted a meta-analysis of the diagnostic miRNAs of NAFLD to determine whether they are associated with diabetes. Since different miRNAs have varying diagnostic power, we selected 10 miRNAs with a pooled AUC >0.7, which suggests moderate diagnostic accuracy. Of these, miR-122 and miR-34a were the most frequently reported miRNAs with consistent up regulation and both had a superior diagnostic power with a pooled AUC >0.85 in NAFLD. Next, we figured out whether these miRNAs can function as biomarkers of T2DM as well. The results of this study indicate that miR-122, miR-34a and miR-375 are the most promising biomarkers of both NAFLD and T2DM. miR-122 is up regulated in NAFLD, pre-diabetic and diabetic patients, and has greater potential than other miRNAs. miRNA-34a was up regulated in patients with NAFLD and diabetes, but not altered in those with prediabetes. Notably, its expression level is found to be associated with NAFLD severity[35] and it can distinguish patients with Non-Alcoholic Steatohepatitis (NASH) from those with NAFLD[24]. Several studies acknowledge that miR-375 is up regulated in diabetic individuals; however, reported data on NAFLD and pre-diabetic individuals are insufficient and inconsistent.

The functions of the target genes of these three miRNAs can partly explain our conclusion. We searched and analyzed the target genes of miR-122, miR-34a and miR-375, and their mediated biological processes in the Gene Ontology database. The target genes of these miRNAs are involved in various processes, such as cell proliferation, pancreatic β-cell development, fatty acid transport, oxidative stress and other processes. Except for clinical research, two recently published reports revealed that miR-122 and miR-34a regulate the development of NAFLD by inducing lipid absorption, adipogenesis, inflammation and apoptosis while inhibiting fatty acid oxidation in mice[73,74]. Additionally, a series of animal experiments support the modulating roles of miR-122, miR-34a and miR-375 in the pathogenesis and development of diabetes[75-77].

Although we identified miR-122, miR-34a and miR- 375 as common biomarkers of NAFLD and T2DM, they exhibit limited effectiveness in identifying diabetics among patients with NAFLD. The primary reason is that only a few studies focus on the diagnostic effectiveness of miRNA in diabetes, which makes it difficult to reach a firm and reliable conclusion. Even though the mean fold-changes in different researches differ considerably, we observe that the expression levels of miR-122 and miR-34a are lower in patients with diabetes than in patients with NAFLD, which may be associated with different stages of the metabolic syndrome. Several studies suggest that miR-375 plays a crucial role in diabetic pathophysiology; however, there are limited data concerning its role in NAFLD. Besides, a study of 90 individuals found that miR-375 had an AUC >0.7 in separating diabetic and pre-diabetic conditions from healthy controls[43]. Interestingly, numerous studies have found that miRNA panels exhibit better diagnostic performance than individual miRNAs, which provides a further studying direction for this topic[34,36,57].

The inconsistency in miRNA expression profiling studies is one of its major drawbacks[28] and the included studies had conflicting results. For example, Krauskopf et al.[58] found that miR-375 was down regulated in T2DM while other studies claimed that it was up regulated. Kokkinopoulou et al.[56] also found that miR-34a was down regulated in T2DM while other studies claimed that it was up regulated. This inconsistency may be attributed to a lack of representativeness caused by a small sample size and an unknown bias in patient selection. Considering that the few anomalous data in one or two studies shall have limited effect on the final results, we take the most consistent results as final results. The studies on miR-375 expression in pre-diabetic patients varied greatly, indicating a low level of credibility and requiring further evaluation.

Inevitably, our study has several limitations. First, even though we limited the sample type, the results of miRNA expression profiling vary largely due to different expression profile platforms, methods of detection and normalization control types[78,79]. We ranked all target miRNAs based on a vote-counting strategy and discarded data not validated by RT- qPCR to reduce the concomitant bias. Still, the levels of circulating miRNAs vary under specific physiological and pathological conditions. Second, some articles about T2DM that met inclusion criteria were excluded for not reporting the target miRNAs, which may be because the target miRNAs were not statistically significant. Undoubtedly, publication bias might affect our results to some extent. However, due to the small amount of reference literature, we did not calculate the publication bias of each target miRNA. Finally, the strength of our conclusion might also be affected by the relatively small sample size of patients with NAFLD and T2DM. Nevertheless, our results may provide a new perspective on existing controversy and for future research. MiRNAs serve as potential biomarkers of both NAFLD and T2DM. MiR-122 and miR-34a can distinguish patients with NAFLD from healthy controls. Additionally, miR- 122, miR-34a and miR-375 are common biomarkers of both NAFLD and T2DM. However, whether these miRNAs can be used for early prediction of diabetes in patients with NAFLD needs further investigation.

Conflict of interests:

The authors declared no conflict of interests.

References