Association between MIC-1 and Type 2 Diabetes: A Combined Analysis

Background and Objectives Type 2 diabetes mellitus (T2DM) is an epidemic disease that endangers human health seriously. Recently, a large number of reports have revealed that macrophage-inhibiting cytokine-1 (MIC-1) is linked with T2DM, but the results were inconclusive. The aim of this study was to perform bioinformatics analysis of the association between MIC-1 and T2DM. Material and Methods Datasets and relevant literatures were searched in Gene Expression Omnibus (GEO), PubMed, Google Scholar, and Web of Science till September 20, 2019. Expression levels of MIC-1 were extracted, pooled, and compared between T2DM cases and controls. Results In summary, 11 GEO datasets and 3 articles with 421 T2DM cases and 711 controls were finally included. The expression level of MIC-1 was significantly higher in T2DM patients compared with controls, with a standard mean difference (SMD) of 0.54 and a 95% confidence interval (95% CI) of 0.24-0.83; in blood samples, the difference was still significant (SMD = 0.65; 95%CI = 0.24‐1.06). Meanwhile, the expression level of MIC-1 plays a significant role in differentiating T2DM cases from controls; the combined sensitivity, specificity, and odds ratio were 0.83 (95%CI = 0.72‐0.90), 0.59 (95%CI = 0.45‐0.72), and 1.64 (95%CI = 1.35‐1.99), respectively. The summary receiver operating characteristic (SROC) curve demonstrated that the area under the curve (AUC) was 0.81 (95%CI = 0.77‐0.84). Conclusion Our results suggested that the expression levels of MIC-1 were significantly higher in T2DM patients in multiple tissues including blood samples.


Introduction
Diabetes is a global disease which is described as a type of metabolic disorder distinguished by increased blood glucose concentration. At present, nearly half a billion people suffer from diabetes [1], and the number of people affected with diabetes has been rising for the last decades [2]. Type 2 diabetes mellitus (T2DM) is the most common type of diabetes and accounts for around 90% of total diabetes cases [3]. Overweight and obesity have been regarded as the main risk factors contributing to T2DM [4]. Approximately 50% of obese subjects will develop into T2DM [5]. Thus, pathway targeting energy metabolism may provide useful diagnosis and treatment information for T2DM [6].
Macrophage-inhibiting cytokine-1, also termed growth differentiation factor 15 (GDF-15), encodes a secreted protein of transforming growth factor beta (TGF-beta) family [7]. It is weakly expressed in multiple tissues under normal conditions [7]. The upregulated production of MIC-1 is triggered in response to inflammation such as tissue injury, biomechanical stress, and anoxia [8,9]. MIC-1 has been considered to play a pivotal role in the development and progression of many diseases such as cardiovascular diseases and malignant cancer [10][11][12]. Recently, MIC-1 is considered as a long-term metabolism regulator with a function of increasing lipolysis as an adipokine in a paracrine fashion [13][14][15]. There have been evident findings in mice that overexpression of MIC-1 could lower the preference for fat intake and improve glucose tolerance and insulin sensitivity compared with MIC-1-null mice which further manifests its protective role in energy homeostasis [16][17][18].
Although several studies have depicted the specific role of MIC-1 in energy expenditure and metabolic activities [13,14,[19][20][21][22], current findings clarifying the relationship between MIC-1 and T2DM are still rare. In our study, we conducted a meta-analysis based on a combination of GEO datasets and relevant clinical reports to illustrate the association between MIC-1 expression levels and T2DM.

Data Acquisition and Search Strategy.
A well-established database of microarray was searched for the current metaanalysis up to September 20, 2019: the Gene Expression Omnibus (GEO; https://www.ncbi.nlm.nih.gov/geo/). The search strategy was as follows: ("diabetes mellitus" OR "diabetes insipidus" OR DIABETES) AND "Homo sapiens".
Afterwards, a systematic literature was searched in PubMed, Google Scholar, and Web of Science, using the combination of key words "diabetes" AND "MIC-1". The language of relevant articles was restricted to English. In order to find more qualified literatures, we viewed the reference lists of publications selected for inclusion.
The process on how we conducted our data search and acquisition of the statistics was described in Figure 1. 2.3. Quality Control and Data Extraction. Two authors (Yue Zhang and Jianan Lu) extracted the information from all qualified datasets in line with the inclusion criteria independently, and any problem or ambiguity was discussed with the team. The following information was extracted for every dataset: last name of first author, country of origin, published year, study subjects (disease status and sample type), expression values, means, and SD of MIC-1.

Statistical Analysis.
At first, we extracted MIC-1 expression levels from datasets and qualified literature; mean and SD were calculated. Next, we performed a meta-analysis to describe expression difference. Forest plots were used to get pooled standard mean difference (SMD) and 95% confidence interval (95% CI). We tested the possibility of heterogeneity by Cochran's Q-statistic and I 2 statistics. A P value < 0.05 and an I 2 > 50% were considered as heterogeneous, and the random effect model would be chosen to calculate the pooled SMD [23]. Otherwise, the fixed effect model would be selected. Besides, funnel plots and Begg's test were used to check the potential publication bias. Obesity is usually classified by body mass index: BMI ≥ 30 as obesity and BMI < 30 as nonobesity [24]. As the original data from the literatures were not available, we only included MIC-1 expression data from GEO datasets and then conducted diagnostic analysis to assess the diagnostic possibility of MIC-1 in T2DM patients. We carried out a meta-analysis with SROC to check the expression level of MIC-1 in T2DM patients. Additionally, multivariate analysis of covariance (MAN-COVA) was used to test differences in MIC-1 between T2DM and non-T2DM groups with age and BMI as covariates in qualified dataset. A nominal level of significance P < 0:05 was accepted.
The data analyses were performed by SAS (SAS Institute Inc., NC, USA) and STATA (STATA Corporation, College Station, TX, USA).

Discussion
In our report, MIC-1 value increased in different tissues of T2DM patients compared with nondiabetes people, including blood samples. MIC-1 has been suggested to be linked with obesity and T2DM. Accumulating evidence in animals clarifies that the overexpression of MIC-1 leads to weight loss, enhanced insulin sensitivity, and higher glucose tolerance. It is assumed that MIC-1 might keep a protective factor in obese and obesity-related diseases like T2DM [25]. Firstly, previous publication has elucidated that MIC-1 acts on feeding centers in the hypothalamus and brainstem to regulate weight control [26]. Secondly, MIC-1 is a secretory product of adipocytes and plays a role in improving lipolysis which is reversely correlated with serum total cholesterol [20,27]. Moreover, elevated MIC-1 expression could invoke improved insulin activity and mediate Th2 cytokines like IL-13 function [14,16]. Previous studies in human etiology of T2DM have demonstrated that high glucose would promote cellular aging and apoptosis. Acting as a protective regulator, MIC-1 expression is accordingly elevated in endothelial cells by ROS-and p53-dependent pathway [28]. MIC-1 is acknowledged as an anti-inflammation cytokine and responds positively to chronic inflammatory disease like T2DM. This acknowledgment may be the reason of higher MIC-1 level in T2DM. Also, a hypothesis was proposed that MIC-1 may function as a self-guarded cytokine in T2DM accounting for reduced food intake and then weight loss. After the peak of MIC-1 expression, it is not sufficient to compensate in the long-term low-grade inflammation, leading to an unstoppable progression of weight gain and metabolism dysfunction [29]. This hypothesis is consistent with our finding that MIC-1 was found to have lower expression levels in obese patients compared with nonobese ones.
Importantly, evidence showed that the serum MIC-1 level ranks the highest in T2DM patients, intermediate in individuals with prediabetes, and lowest in patients without diabetes [20,[30][31][32]. Elevated serum MIC-1 expression also suggested a high risk for T2DM-related complications including cardiovascular risks [33,34] and renal diseases [35,36]. The current study found a positive correlation between T2DM and MIC-1 level in blood samples, suggesting that MIC-1 could be a practical research target in the future.
Interestingly, an experiment in mice has reported that treating wild-type mice with both MIC-1 and liraglutide, a long-term glucagon-like peptide 1 (GLP-1) agonist, shows a synergistic impact on weight loss [18]. Progress has been made in combined utility of MIC-1 treatment and mature clinical medicine which shows a promising route for MIC-1 as a therapy biomarker. A fair number of progress have been made in identifying GFRAL, an orphan receptor of glialderived neurotropic factor receptor α, as a receptor for MIC-1 and its neuronal circuits in which it acts [22,[37][38][39]. Based on this breakthrough, a class of MIC-1/GFRAL/-RET-based drugs can be highly anticipated for treatment in T2DM [40].
The limitation of our study should also be of concern. First, our study estimated the contribution of numbered risk factors accounting for T2DM, such as age and BMI. Due to the limited data we obtained, we were unable to remove other influential factors such as smoking, dietary habit, and sedentary lifestyle. Second, our conclusion is summarized by a meta-analysis, causal relationship cannot be confirmed. Therefore, laboratory experiments should be performed to address the comprehensive mechanism between MIC-1 and T2DM.

Conclusion
Our study suggested that the expression level of MIC-1 was significantly higher in T2DM patients in multiple tissues including blood samples.

Conflicts of Interest
No potential conflict of interest was reported by the authors.