Diabetes, also referred to as diabetes mellitus, is an enduring health condition that has had a profound and far-reaching effect on the lives of millions of individuals globally. The disease is a collection of disorders that share hyperglycemia as a characteristic. Hyperglycemia is caused by low insulin secretion and action, or both (Maraschin, 2013). Based on the figures on diabetes mellitus in the year 2021 from International Diabetes Federation, approximately 537 million individuals, aged between 20 and 79 years, are currently grappling with the disease, translating to one in every 10 people affected by this health condition (Atlas). This figure might reach 643 million by 2030 and 783 million by 2045, according to predictions (Atlas). In addition, diabetes turned out to be the cause of death for 6.7 million people in 2021, equivalent to one death every five seconds (Atlas). If not managed properly, diabetes can lead to severe complications and premature deaths (Rahman et al., 2021). Type 2 diabetes mellitus (T2DM) is markedly further prevalent than type 1 diabetes mellitus (T1DM) or gestational diabetes, accounting for more than 90% of all cases. One of the main characteristics of type 2 diabetes (T2DM) is the impaired metabolism of proteins, fats, and carbohydrates as a result of either insulin resistance or inadequate insulin secretion, or both.

In recent decades, our comprehension of the onset and advancement of Type 2 Diabetes Mellitus (T2DM) has undergone rapid expansion. The principal factor driving the disease is the gradual deterioration of insulin secretion by pancreatic β cells (DeFronzo et al., 2015). Type 2 diabetes (T2DM) has its causes rooted in both hereditary and environmental factors. Obesity plays a vital role in T2DM (L Tuck & B Corry, 2010; Habib F., 2022). The Pathophysiological alterations include beta-cell dysfunction, insulin resistance, and chronic inflammation, all of which impede the regulation of blood sugar and hasten onset of micro- and macrovascular problems (DeFronzo, 2009). 

Evidence suggests that IR is inherited and is caused by a number of intrinsic factors. Insulin resistance is capable of being identified in patients containing a mutation in the insulin receptor gene. It was recently identified that a glycoprotein with the name ectoenzyme nucleotide pyrophosphate phosphodiesterase 1 (ENPP1) (ENPP1; likewise referred to as plasma cell glycoprotein 1, PC-1) is strongly expressed in the muscle, skin, and fat of people who have type 2 diabetes. As a consequence of this, insulin signal transduction is disrupted, and insulin resistance is brought about (Maddux & Goldfine, 2000). As a direct consequence of this, type 2 diabetes begins to develop in an organized manner.

Serosa and endoplasmic reticulum membranes contain type II transmembrane glycoprotein ENPP1. The following are possible mechanisms for the ENPP1 121Q alleles induction of IR: (i) changes to the insulin receptors tyrosine kinase active area affect the serine and threonine autophosphorylation sites, preventing the downstream cascade from receiving insulin signals. (Costanzo et al., 2001); (ii) inhibiting the activity of tyrosine kinase prevents post-receptor signal trans-duction (Kumakura et al., 1998) as well as (iii) elevated serum insulin levels subsequently induce ENPP1 expression via a system for sending signals quickly and accurately (Menzaghi et al., 2003). The ENPP1 gene, which spans 80 kilobases and is located on chromosome 6q22-23, has 24 introns and 25 exons. The missense mutation at position 121 of the ENPP1 gene (rs1044498) results in a polymorphism known as K121Q.  The change in the 121st codon from adenine (A base) to cytosine (C base) results in the matching amino acid sequence being altered to have glutamine (Q) instead of lysine (K) (Grarup et al., 2006). It has been reported that the ENPP1 K121Q (rs1044498) polymorphism has been linked to type 2 diabetes in many countries (Badaruddoza et al., 2015; Hsiao & Lin, 2016; Marchenko et al., 2018; Mtiraoui et al., 2012; Yako et al., 2015). In this analysis, we attempted to get a more precise understanding of the link between ENPP1 (rs1044498) and type 2 diabetes by doing an updated meta-analysis study utilizing the earlier studies that had been done on a range of different ethnic groups. This study through meta-analysis, will help us to summarize the overall association of the SNP with type 2 diabetes. Moreover, to understand the unknown effect size, we can also be able to compare and contrast the findings of several studies, identify patterns among studies, and find sources of disagreement among those results.

Literature Search Strategy

A thorough search of Google Scholar, PubMed, Science Direct, and Medline was conducted up until the end of 2022 to retrieve the literature on the relationship between ENPP1 polymorphisms and T2D susceptibility. The investigation was conducted using the following keywords: (K121Q OR rs1044498 OR polymorphism) AND (ENPP1 OR "PC-1" OR "plasma cell membrane glycoprotein 1" OR "ectonucleotidase pyrophosphatase/phosphodiesterase 1") AND (Diabetes OR T2D OR T2DM OR "type 2 diabetes mellitus"). The included researchs cited works and other pertinent papers were also read. Alongside we retrieved multiple studies from the previous meta-analysis research. 

Inclusion and exclusion criteria of the study

The subsequent inclusion measures were used to select studies for inclusion in this meta-analysis: (1) case-control studies; (2) consideration of ENPP1 polymorphisms and type 2 diabetes susceptibility; (3) allele and genotype counts in great detail between case and controls; and (4) Value of Hardy-Weinberg Equilibrium (HWE) conforming controls. We technically omitted studies that were basically- i) case studies or reviews that did not include any controls or differentiate case and control data; ii) reports with no available data; and iii) reports that are already on file.

Data extraction 

Data extraction was done after the literatures were screened and the inclusion and exclusion criteria were followed. Specifically, the following data were retrieved for each study which included: the list of authors, the year the study was published, the ethnicity of the participants, the sample size, the genotype of each gene variant, and the HWE.

Statistical analysis

Using odds ratios (ORs) and 95% confidence intervals (CIs), the degree of the relationship between ENPP1 rs1044498 polymorphisms and Type 2 Diabetes (T2D) was evaluated. The pooled ORs for ENPP1 rs1044498 (K121Q) K > Q were determined using five distinct genetic models: homozygous (QQ vs. KK), heterozygous (KQ vs. KK), dominant (KQ + QQ vs. KK), recessive (QQ vs. KK + KQ), and allelic (Q vs. K). Heterogeneity was evaluated using I2, with a preference for I2 values exceeding 50% to indicate significant heterogeneity. In instances where I2 exceeded 50%, a random-effects model was the utilized. (DerSimonian & Laird, 1986), and when homogeneity was present (I2 ≤ 50%), we used a fixed-effects model (Mantel & Haenszel, 1959). In addition, subgroup analyses based on ethnicity were carried out to calculate ORs that were specific to each ethnic group. Finally, Begg-Mazumdars test, Eggers test, and funnel plots were used to evaluate publication bias (Begg & Mazumdar, 1994; Egger, Smith, Schneider, & Minder, 1997), with a P-value for statistical significance of less than 0.05. For each study, to assess the Hardy-Weinberg equilibrium (HWE), a comparison was made between the expected and the observed genotype frequencies of the control group. Using a two-tailed P-value, statistical analyses were done in Stata (StataCorp., College Station, TX, USA) version 14.2, and Review Manager (5.4.1). The cutoff for significance was set at P<0.05.

Features of the study

A total of forty-four articles were selected after the inclusion and exclusion criteria were applied. The selected articles contained a total of 56 studies. 8 of those studies did not fulfill the HWE value criteria (Abate et al., 2005; Weedon et al., 2006; Willer et al., 2007; Bhatti et al., 2010; Saberi et al., 2011; Barna et al., 2018; Golbon et al., 2018; Gohari-Lasaki et al., 2020) (P<0.05). Therefore, they were excluded from the finalized meta-analysis. The final meta-analysis contained 48 studies from 37 articles comprising 24979 cases and 33005 controls. For the meta-ana-lysis, the features of each study are presented in Table 2 and Fig. 1 shows a selection process flowchart.

Table 1: Genotypic and descriptive details according to the chosen study for rs1044498 meta-analysis.

Meta-analysis

Table 2 and Fig. 2 illustrate that K>Q polymorphism in ENPP1 (K121Q) is related to an enhanced risk of developing type 2 diabetes. Overall, population susceptibility to type 2 diabetes was strongly correlated with the polymorphism. 

Each genetic variation was linked to an enhanced chance of evolving into type 2 diabetes, and the associations were statistically significant. (QQ vs. KK: OR = 1.53, 95% CI = 1.23-1.90, P = 0.0001; KQ vs. KK: OR = 1.22, 95% CI = 1.08-1.37, P = 0.001; KQ + QQ vs. KK: OR = 1.15, 95% CI = 1.11-1.41, P = 0.0003; QQ vs. KK + KQ: OR = 1.38, 95% CI = 1.17-1.64, P = 0.0002; and Q vs. K; OR = 1.22, 95% CI = 1.10-1.36, P = 0.0003) (P<0.05). 

Table 2: Meta-analysis of the association between rs1044498 and T2DM.

In the African population, the ENPP1 rs1044498 K>Q polymorphism was not shown to be related to an enhanced risk of evolving into type 2 diabetes, according to the subgroup analysis (P>0.05), but there was a link corresponding to the Asian population in all genetic models (Table 2) and the Caucasian population for four genetic models (Homozygous, Dominant, Recessive and Allelic model) where P value was less than 0.05. However, there was no evidence of any associations between the heterozygous model and type 2 diabetes in the Caucasian group (P>0.05). 

Fig. 2: Forest plot of type 2 diabetes and rs1044498 polymorphism for homozygous model (QQ vs KK).

Eggers test and Begg-Mazumdars test below (Table 3), as well as the funnel plot (Fig. 3), were utilized in this study to investigate whether or not each genetic model had been affected by publication bias. However, these analyses and figures did not reveal any signs of publishing bias. 

Table 3: Evolution of Publication Bias.

A leave-one-out investigation was completed to determine the impact of each individual study on the pooled odds ratios (Fig. 4).  dependability of the meta-analysis findings.

Fig. 3: Forest plot of type 2 diabetes and rs1044498 polymorphism for heterozygous model (KQ vs KK).

The results showed that the pooled effect did not change much when a study was taken out of a meta-analysis and the remaining studies were used instead. This demonstrated the stability and

Fig. 4: Forest plot of type 2 diabetes and rs1044498 polymorphism for dominant model (KQ+QQ vs KK).

The worldwide spread of Type 2 Diabetes Mellitus (T2DM) is a big cause for concern in terms of public health. Approximately 6.9 million adults in Bangladesh are living with diabetes. Deaths related to diabetes make up 3% of the total mortality rate in the country (Yasmin et al., 2020). The probability of developing T2DM over time increases in those inflicted with insulin resistance (IR) (Bacci et al., 2005). A connection between insulin resistance and the ENPP1 gene has been found by (Bacci et al., 2005). Based on these findings, we chose this gene to include in this meta-analysis. 

There are a total of 24979 cases representing type 2 diabetes patients and 33005 healthy people serving as controls in this study. These case-control were collected from 48 case-control studies that were published in 37 articles. According to the findings of the meta-analysis, the rs1044498 polymorphism in ENPP1 shows a noteworthy connection to type 2 diabetes. In each of the models, the rs1044498 polymorphism in ENPP1 was observed to be associated with an enhanced risk of evolving into type 2 diabetes. The odds ratio (OR) for the homozygous model was 1.53 (95% confidence interval = 1.23-1.90, P = 0.0001), while the odds ratio (OR) for the heterozygous model was 1.22 (95% confidence interval = 1.08-1.37, P = 0.001). The odds ratio (OR) for the dominant model was 1.15 (95% confidence interval = 1.11-1.41), P = 0.0003; for the recessive model, the rs1044498 polymorphism was 1.38 (95% confidence interval = 1.17-1.64), P = 0.0002; and for the allelic model, the OR was 1.22 (95% confidence interval = 1.10-1.36), P = 0.0003 (P<0.05). According to the heterogeneity (I2) among the studies, we generated forest plots to quantify the findings. These plots were produced using either a model with random effects or a fixed effect. If the I2 value was lower than 50%, the fixed effect model was favored over the random effect model (DerSimonian & Laird, 1986; Higgins & Thompson, 2002; Mantel & Haenszel, 1959). 

It was found that there was heterogeneity among the studies and it used subgroup analysis. Except for the African population, all other models showed considerable heterogeneity. However, we recognized that a limited number of studies on African ethnicity were found in databases included in this meta-analysis (Table 1) and we expect that including more research from this region in the future may produce a different conclusion. In every genetic model for Asian populations, the ENPP1 rs1044498 mutation was found to have a highly substantial link with type 2 diabetes. In the case of the Caucasian population, the association could be detected in the homozygous, dominant, recessive, and allelic models; not in the heterozygous model. This confirms what has been found in prior research (McAteer et al., 2008; Tang et al., 2014).

Fig. 5: Forest plot of type 2 diabetes and rs1044498 polymorphism for recessive model (QQ vs KK+KQ).

Three other meta-analysis studies on the same poly-morphism issue were conducted in the distant past (Li, 2012; McAteer et al., 2008; Tang et al., 2014). However, there were severe limitations in those meta-analyses that we have addressed in ours. The meta-analysis study on this particular polymorphism published in 2008 was centered around only the European studies (McAteer et al., 2008). Another meta-analysis was conducted in 2012 with the Chinese studies on the polymorphism only (Li, 2012). Both of these studies were conducted only on a selective ethnicity; not on all the ethnic populations available. However, a meta-analysis study in 2014 considered studies on all available ethnic studies (Tang et al., 2014). They carried out a meta-analysis of 51 studies retrieved from 40 articles. But major drawbacks can be pointed out from that study. For instance, they undertook the meta-analysis with four genotypic models in view whereas in our study, we executed a meta-analysis for five genotypic models. Also, they used some insufficient study data, some of which contained no control data present at all. In our updated meta-analysis, we excluded insufficient study data so that our data are more acceptable and reliable despite being quantitatively compromised. Moreover, in our meta-analysis, the studies included are of the most recent years to allow our results to be more up-to-date and robust. Our research revealed symmetry in the funnel plot, indicating that there was no publishing bias. However, funnel plot alone is not an effective tool for assessing publication bias. We also conducted Begg Mazumdar and Egger tests and found no proof of publication bias across the analysis (P>0.05). Furthermore, sensitivity analysis demonstrated the consistency and reliability of our analysis. Because of the limited size of the sample, the connection between the ENPP1 rs1044498 variant with the African population was not established. This result may change if more genetic association studies with large sample data from African populations are incorporated, which was considered the first limitation of this meta-analysis. 

However, the limitation mentioned may serve more as a direction through which the study can be further sophisticated. Including more subjects would strengthen the sample pool. To attain a more comprehensive comprehension of the connection between the ENPP1 gene and the threat of developing type 2 diabetes, it is suggested that further investigations be carried out to examine additional single nucleotide polymorphisms (SNPs) of this gene. These future studies would serve to more precisely elucidate the function of ENPP1 in the progress of type 2 diabetes and could shed light on the mechanisms by which specific SNPs of the gene may increase an individuals predisposition to developing the disease.

Fig. 6: Forest plot of type 2 diabetes and rs1044498 polymorphism for allelic model (Q vs K).

Fig. 7: Examination of publication bias using funnel plots.

Fig. 8:  Sensitivity analysis plot for the Homozygous model (QQ vs. KK) of the association between the rs1044498 polymorphism and type 2 diabetes.

Fig. 9:  Sensitivity analysis plot for the Heterozygous model (KQ vs. KK) of the association between the rs1044498 polymorphism and type 2 diabetes.

Fig. 10: Sensitivity analysis plot for the Dominant model (KQ+QQ vs. KK) of the association between the rs1044498 polymorphism and type 2 diabetes.

Fig. 11:  Sensitivity analysis plot for the Recessive model (QQ vs. KK+KQ) of the association between the rs1044498 polymorphism and type 2 diabetes.

Fig. 12:  Sensitivity analysis plot for the Allelic model (Q vs. K) of the association between the rs1044498 polymorphism and type 2 diabetes.

This meta-analysis report supports the connection between the ENPP1 rs1044498 polymorphism and Type 2 diabetes, especially in Caucasian and Asian populations. The significance of the ENPP1 gene has been proposed in the initiation of Type 2 diabetes. To completely comprehend the mechanism by which this gene contributes to the onset of T2DM and to identify potential therapeutic targets, more rigorous research into the avenue is necessary. Overall, the data could help create strategies to prevent or treat T2DM in people with this genetic variant. To verify and understand this preliminary finding in the Caucasian and Asian populations, large-scale prospective studies may be required.

The authors thank the Department of Biotechnology and Genetic Engineering, Noakhali Science and Technology University for providing the opportunity to conduct the research.

The authors declare that there is no conflict of interest.