This study showed that common SLC2A10 genetic variants were associated with the development of PAD in type 2 diabetic patients independent of all known risk factors for PAD including age, sex, smoking, lipids, and blood pressures. The association was observed at baseline and was further replicated during follow-up study. These data suggest the SLC2A10 gene plays a significant role in the pathogenesis of PAD in diabetic patients.
gene encodes GLUT10, a facilitative glucose transporter. The facilitated glucose transporter (GLUT) mediated the uptake of several monosaccharides including glucose, fructose, mannose, galactose, and glucosamine [17
]. The physiological function of GLUT10 was unclear until recently, when a genetic form of arterial tortuosity syndrome was identified by a loss-of-function mutation in the SLC2A10
], with a pathology characterized by upregulation of TGF-β signaling in the vascular walls. TGF-β signaling is one of the final common pathways linking hyperglycemia and vascular complications in individuals with diabetes mellitus [9
]. In addition, impaired uptake of monosaccharides, a function of GLUT10, may hinder glycosylation, which is an important step for the production of functional glycoproteins and proteoglycans [5
]. Glycoproteins and proteoglycans are essential structural components of the arterial wall and connective tissue. In support of this notion, we recently demonstrated that mutation in the SLC2A10
gene in mice causes irregular shape of large and medium-sized arteries, characterized by markedly increased elastic fibers and intimal endothelial hypertrophy [18
]. These data further support the role of SLC2A10
gene in the pathogenesis of diabetic vascular complications.
The SNPs with significant associations with PAD in our study are located in non-coding regions within a single LD block spanning from intron 1 to the 3' downstream of SLC2A10
gene. One non-synonymous SNP rs2235491 (Ala260Thr, exon 2) was in strong LD with rs2179357 (D' = 0.82) but did not show significant association with PAD, which may be attributed to its low frequency in the Taiwanese population. Previous study has provided evidence that the rs2235491 Ala/Ala carriers exhibit higher fasting plasma insulin level and higher area-under-curve of insulin levels after glucose loading [19
]. The rs2235491 Ala allele might contribute to the vascular complication in type 2 diabetes via associated hyperinsulinemia. A larger sample size would be required to clarify whether this putatively functional variation is the causal variant. In our study, the association is strongest for the SNPs that are located near the 3' UTR, downstream of the SLC2A10
gene. However, bioinformatics analysis of the sequences surrounding this region failed to reveal any conserved regulatory motif among different species to support the functional consequences [20
Our results demonstrate that variation of the SLC2A10
gene not only results in a rare vascular syndrome but also is associated with susceptibility to common vascular complications in type 2 diabetes. It is interesting to note that prevalence of PAD and amputation rate in type 2 diabetes is lower in Chinese populations as compared to Caucasians and Indians [21
]. Using data from the HapMap database, the risk at the rs2179357 T allele, which is in strong LD with other risk alleles in our study, is present in 81% of Caucasians but only in 44% of Han Chinese [24
]. Whether this difference explains the lower incidence of PAD in type 2 diabetic Asians remains to be further studied.
There are certain limitations of this study. First, although a total of 1,967 person-years of follow-up were accumulated in this study, the sample size is relatively small and the incident cases were limited. Therefore, the association may be seriously influenced by random fluctuation. We calculated the power of this study to detect variants with various frequencies and odds ratio. For 345 controls and 27 cases, we will have a statistical power of 0.8 to detect an odds ratio of 3.0, 3.6, 4.6 and 10.0 for a target allele or haplotype frequency of 0.2, 0.1, 0.05 and 0.01, respectively, assuming a type I error rate of 0.05 (Additional File 1
: Supplemental Figure S1). Second, the result has not been replicated in an independent study. Although the association in baseline was replicated in longitudinal follow-up, it cannot be viewed as a true independent replication. Thirdly, population stratification may arise when samples are drawn from structure populations. However, our study participants were all Han Chinese by self-report. Previous research has demonstrated that genotype distribution in Han Chinese in Taiwan is highly homogenous [25
]. The high homogeneity mitigates the possibility of spurious association. Lastly, the drop-out rate in this study is substantial (31.11% in H4 haplotype carriers and 31.11% in non-H4 haplotype carriers). Although there is no difference in drop-out rate or drop-out reasons between two groups (data not shown), whether the high drop-out rate introduces bias to this study is not known.