OBJECTIVE

A common nonsense mutation in TBC1D4 was recently found to substantially increase the odds of type 2 diabetes in Greenlandic Inuit, leading to exclusively increased postprandial glucose. We investigated the frequency and effect of the TBC1D4 mutation on glucose metabolism and type 2 diabetes diagnosis among Canadian and Alaskan Inuit.

RESEARCH DESIGN AND METHODS

Exome sequencing of the TBC1D4 variant was performed in 114 Inuit from Nunavik, Canada, and Sanger sequencing was undertaken in 1,027 Alaskan Inuit from the Genetics of Coronary Artery Disease in Alaskan Natives (GOCADAN) Study. Association testing evaluated the effect of the TBC1D4 variant on diabetes-related metabolic traits and diagnosis.

RESULTS

The TBC1D4 mutation was present in 27% of Canadian and Alaskan Inuit. It was strongly associated with higher glucose (effect size +3.3 mmol/L; P = 2.5 x 10−6) and insulin (effect size +175 pmol/L; P = 0.04) 2 h after an oral glucose load in homozygote carriers. TBC1D4 carriers with prediabetes and type 2 diabetes had an increased risk of remaining undiagnosed unless postprandial glucose values were tested (odds ratio 5.4 [95% CI 2.5–12]) compared with noncarriers. Of carriers with prediabetes or type 2 diabetes, 32% would remain undiagnosed without an oral glucose tolerance test (OGTT).

CONCLUSIONS

Disruption of TBC1D4 is common among North American Inuit, resulting in exclusively elevated postprandial glucose. This leads to underdiagnosis of type 2 diabetes, unless an OGTT is performed. Accounting for genetic factors in the care of Inuit with diabetes provides an opportunity to implement precision medicine in this population.

Precision medicine, as defined by the National Institutes of Health’s Precision Medicine Initiative Working Group, aims to tailor health care to individual variability in genes, environment, and lifestyles (1). While oncology has demonstrated some tangible success with precision medicine, few examples exist in other domains of health care (2). Genetic predisposition to disease may vary not only by individual but also by ancestry, and harnessing this information may help health care providers to better understand how to approach disease among different ethnicities. This may be particularly relevant in specific populations, such as the Inuit, that have had different natural selection pressures leading to different frequencies of genetic variants.

Moltke et al. (3) recently identified a population-specific nonsense variant in TBC1D4 which was both common (minor allele frequency [MAF] 17%) and had a large effect on type 2 diabetes risk among the Greenlandic population (homozygote odds ratio [OR] 10.3; P = 1.6 × 10−24 in a recessive model). Specifically, compared with noncarriers, homozygous carriers of the TBC1D4 p.Arg684.Ter variant had markedly higher concentrations of 2-h plasma glucose (3.8 mmol/L [68 mg/dL]; P = 2.5 × 10−35) and serum insulin (165 pmol/L [24 uIU/mL]; P = 1.5 × 10−20) after an oral glucose tolerance test (OGTT). Conversely, this variant decreased fasting glucose and had little effect on HbA1c. The same effects—although of smaller magnitude—were also observed in heterozygote carriers.

What is not known is whether this variant is present in other Inuit and whether it may, given the exclusive effect on postprandial glucose, influence diagnostic strategies for type 2 diabetes. Such information may have clinical utility given that the prevalence of type 2 diabetes has been increasing dramatically among the Inuit over the past 25 years (4).

Interestingly, the ancestors of the Inuit in Greenland are thought to have migrated from Siberia, across the Central and Eastern Canadian Arctic (5). If the TBC1D4 nonsense mutation arose earlier along migration routes out of Siberia, this mutation would also be present in North American Inuit. Therefore, in this study we assessed the prevalence of the TBC1D4 mutation and its effect on diagnostic strategies for type 2 diabetes in the North American Arctic Inuit population.

To test whether the TBC1D4 variant is indeed specific to Greenland or is shared by other Arctic populations, we assayed the p.Arg684Ter single nucleotide polymorphism in two distinct North American circumpolar populations. The first consisted of 114 Inuit from Nunavik in Northern Quebec, Canada (the Eastern North American Arctic), and the second of 1,027 individuals from a cohort study of Alaskan Inuit, called the Genetics of Coronary Artery Disease in Alaskan Natives (GOCADAN) study (6) (the Western North American Arctic).

Inuit from Nunavik were recruited originally for the purposes of a research project assessing genetic determinants of fatty acid metabolism (7). While that study did not include type 2 diabetes status or metabolic traits, it was used to quantify the prevalence of the TBC1D4 variant in Nunavik.

Informed consent for the current study was obtained from all participants. The TBC1D4 variant (rs61736969, a nonsense polymorphism in TBC1D4 that is also called p.Arg684Ter or c.2050C>T) was assayed using exome sequencing (see the Supplementary Data for a detailed description of the exome sequencing).

We next tested the effect of this variant on glucose metabolism in the GOCADAN study (6). This longitudinal, population-based study, conducted between October 2000 and April 2004, aimed to investigate the genetic determinants of cardiovascular disease among 1,214 Inuit from several coastal villages in the Norton Sound region of Western Alaska. All individuals older than 18 years residing in this region were invited to participate. Testing included a physical examination, laboratory determinations, and measures of subclinical disease (described in detail elsewhere [6]). Specifically, type 2 diabetes status was determined by self-report, medication use, and fasting glucose screening (Accu-Chek Advantage). A urine sample was collected and blood samples were taken when fasted and 2 h after a standard 75-g OGTT (Glutol; Paddock Laboratory Inc., Minneapolis, MN). Participants were excluded from the OGTT if they were already considered as having type 2 diabetes according to the following criteria: having known type 2 diabetes and requiring insulin, or using oral hypoglycemic agents with two recorded glucose values >13.9 mmol/L or a fasting blood glucose value >12.5 mmol/L.

For the purposes of this study, we used Sanger sequencing on the TBC1D4 variant in 1,027 GOCADAN participants (51 self-reported cases with type 2 diabetes and 976 controls) with available DNA samples. PCR and Sanger sequencing of the resulting amplicons were performed using established protocols (see the Supplementary Data for a detailed description of the method).

For comparison purposes, we queried the MAFs of the TBC1D4 variant in different ancestral populations with available data in phase 3 of the 1000 Genomes Project (8).

Next, we conducted a genetic association study of the GOCADAN cohort, applying both additive and recessive genetic models to test for associations between the TBC1D4 variant and reported type 2 diabetes status, as well as 14 traits related to type 2 diabetes (fasting and 2-h plasma glucose concentrations at OGTT, fasting and 2-h serum insulin concentrations at OGTT, Gutt insulin sensitivity index [ISI0,120] (9), serum HbA1c, triglycerides, total serum cholesterol, LDL cholesterol, HDL cholesterol, albuminuria, BMI, percentage fat mass, and waist circumference). For these quantitative traits, the association testing was done only in participants without diabetes. To evaluate the association between the TBC1D4 variant and type 2 diabetes status, we used the full data set (both patients with diabetes and those without diabetes). Because of skewed values for some traits, we performed additional association tests after a rank-based inverse normal transformation (10) of the values to ensure normality. The measured genotype association method used for the association testing was implemented in the Sequential Oligogenic Linkage Analysis Routines software program (available from https://www.nitrc.org/projects/se_linux; described in detail elsewhere [11]). Finally, we accounted for multiple testing by applying a Bonferroni correction (12), dividing the α level (0.05) by the number of traits tested (n = 14). We recognize that this correction is overly conservative because many of the traits are highly correlated.

To compare the risk of underdiagnosis of prediabetes and type 2 diabetes among TBC1D4 carriers and noncarriers, we applied the 2010 American Diabetes Association criteria (13) to the sample of patients with available OGTT values in order to classify them into 3 groups: type 2 diabetes, prediabetes, or neither. Specifically, type 2 diabetes was diagnosed if at least one of the following four criteria is present: 1) HbA1c ≥6.5% (48 mmol/mol); 2) fasting plasma glucose ≥7 mmol/L (126 mg/dL); 3) 2-h plasma glucose ≥11.1 mmol/L (200 mg/dL) during 75-g OGTT; or 4) random plasma glucose ≥11.1 mmol/L (200 mg/dL). Prediabetes was diagnosed if at least one of the following three criteria was present: 1) fasting plasma glucose between 5.6 and 6.9 mmol/L (100 and 125 mg/dL); 2) 2-h plasma glucose in 75-g OGTT between 7.8 and 11.0 mmol/L (140 and 199 mg/dL); or 3) HbA1c between 5.7% and 6.4% (39 and 46 mmol/mol).

Then we compared the ratio of TBC1D4 carriers to noncarriers who would have been diagnosed as having prediabetes and type 2 diabetes only by a 2-h OGTT. The OR of missed diagnosis of type 2 diabetes was calculated using the Fischer exact method (14,15) as applied in the epitools package of the R statistical software package (R Foundation for Statistical Computing, Vienna, Austria).

The study of Nunavik Inuit was approved by the Nunavik Nutrition and Health Committee and Comité d’éthique de la recherche du Centre hospitalier de l’Université de Montréal. The GOCADAN study was approved by the ethics review board at Norton Sound Health Corporation and all relevant institutional review boards. The analysis investigating the risk of type 2 diabetes underdiagnosis among GOCADAN participants was approved by the research ethics committee of the Jewish General Hospital of Montreal.

A map detailing the locations of the participating Nunavik and Alaskan Inuit cohorts appears in Fig. 1. Exome sequencing of the p.Arg684Ter variant (rs61736969, a nonsense polymorphism in TBC1D4 c.2050C>T) in the cohort of 114 Nunavik Inuit individuals (63 men and 51 women; mean age 52 years) revealed that the variant was present. Of the 114 sequenced samples, 10 were removed from the analysis because of high levels of admixture (see the Supplementary Data for a definition of the admixture). The genotyping of the C/T variant in the remaining 104 samples from Nunavik revealed an allele frequency of 16.3% for the T allele (the nonsense polymorphism) (Supplementary Table 1).

Figure 1

Map of the Arctic showing the regions of sampling for the Quebec Inuit (Nunavik) and GOCADAN (Norton Sound) cohorts.

Figure 1

Map of the Arctic showing the regions of sampling for the Quebec Inuit (Nunavik) and GOCADAN (Norton Sound) cohorts.

Close modal

Sanger sequencing of the p.Arg684Ter variant in the GOCADAN cohort also revealed the presence of the rs61736969 C/T variation, a nonsense mutation. The estimated allele frequency of the T allele among the sample of 1,027 individuals was 13.2% (Supplementary Table 1).

We next assessed the allele frequencies of the TBC1D4 variants among different ancestral populations in phase 3 of the 1000 Genomes Project (8). The TBC1D4 c.2050C>T variant had a MAF of 0% in Europeans, 0.6% in Latinos, 0% in African Americans, 0% in East Asians, and 0% in South Asians, and a global MAF of ∼1%. It is important to state that rs61736969 is a triallelic variant. There exists in individuals of African ancestry an alternate allele that has a MAF of 3.6% and is not the same allele as the one found in Inuit.

A detailed description of the metabolic traits assayed for the entire GOCADAN cohort, and separately for carriers, noncarriers, and nongenotyped participants, appears in Table 1. The results of the association between the TBC1D4 p.Arg684Ter variant and these metabolic traits for both the additive and recessive genetic models are reported in Table 2. When comparing homozygous carriers without diabetes of the variant with carriers of other genotypes, we found that homozygous carriers have increased concentrations of 2-h plasma glucose (β = +3.3 mmol/L; P = 2.5 × 10−6) and 2-h serum insulin concentrations (β = +175 pmol/L; P = 0.04) after an OGTT. Homozygous carriers also demonstrated decreased peripheral insulin sensitivity, as estimated by the ISI0,120 (β = −1.36 SDs; P = 0.001). Heterozygous carriers without diabetes also showed an increase in 2-h plasma glucose (β = +0.5 mmol/L; P = 2.8 × 10−4). There was no discernible effect on fasting plasma glucose or serum insulin concentrations and HbA1c in homozygotes without diabetes when applying a recessive model of transmission. Consistent with results found by Moltke et al. (3), heterozygote carriers without diabetes had lower fasting glucose and insulin concentrations (β = −0.13 mmol/L; P = 2.2 × 10−4 and β = −9 pmol/L; P = 7 × 10−4, respectively). When analyzing a group of 26 individuals with type 2 diabetes who were not using any medications for type 2 diabetes and had undergone an OGTT, we found that homozygous carriers of the TBC1D4 variant displayed a higher 2-h plasma glucose concentration than noncarriers (β = +14 mmol/L; P = 0.02), but there was no significant association with 2-h serum insulin and fasting glucose and insulin concentrations among these individuals (Supplementary Table 2).

Table 1

Characteristics of the entire GOCADAN cohort (metabolic traits) and specifically in TBC1D4 variant carriers, noncarriers, and nongenotyped participants

TraitNoncarriers (n = 771)
Heterozygote carriers (n = 241)
Homozygote carriers (n = 15)
Nongenotyped participants (n = 187)
All participants (n = 1,214)
Participants (n)Mean (SD)Participants (n)Mean (SD)Participants (n)Mean (SD)Participants (n)Mean (SD)Participants (n)Mean (SD)
Fasting plasma glucose (mmol/L) 771 5.3 (1.1) 240 5.1 (1.1) 15 5.0 (0.9) 133 5.2 (0.6) 1,159 5.2 (1.0) 
2-h plasma glucose (mmol/L) 573 5.2 (1.9) 156 5.6 (2.3) 8.9 (3.5) 92 5.6 (2.2) 827 5.3 (2.0) 
Fasting serum insulin (pmol/L) 768 65.1 (52.2) 241 52.2 (33.4) 15 50.2 (26.8) 130 67.2 (51.6) 1,154 62.5 (48.8) 
2-h serum insulin (pmol/L) 570 205.6 (234.0) 156 210.0 (193.4) 350.4 (300.4) 91 227.1 (228.5) 823 209.9 (226.8) 
HbA1c (%) 770 5.5 (0.5) 241 5.4 (0.4) 15 5.5 (0.3) 132 5.4 (0.3) 1,158 5.4 (0.5) 
HbA1c (mmol/mol) 770 36.0 (5.1) 241 35.0 (4.8) 15 37.0 (3.3) 132 35.0 (3.6) 1,158 36.0 (4.9) 
ISI0,120 (mg × L2/mmol × mU × min) 566 115.4 (53.9) 156 110.1 (53.1) 71.9 (47.7) 88 104.6 (46.0) 816 113.1 (53.1) 
Fasting serum triglyceride (mmol/L) 771 1.5 (1.1) 239 1.4 (0.6) 15 1.7 (0.6) 131 1.4 (0.9) 1,156 1.5 (1.0) 
Fasting serum total cholesterol (mmol/L) 771 5.2 (1.1) 239 5.2 (1.0) 15 5.5 (1.2) 131 5.3 (1.1) 1,156 5.2 (1.0) 
Fasting serum LDL cholesterol (mmol/L) 771 3.0 (0.9) 239 3.0 (0.9) 15 3.1 (1.0) 131 3.1 (0.9) 1,156 3.0 (0.9) 
Fasting serum HDL cholesterol (mmol/L) 771 1.5 (0.5) 239 1.6 (0.5) 15 1.6 (0.3) 131 1.6 (0.5) 1,156 1.5 (0.5) 
Albuminuria (mg/g) 739 0.1 (0.3) 234 0.0 (0.2) 15 0.1 (0.3) 129 0.1 (0.2) 1,117 0.1 (0.2) 
Waist circumference (cm) 709 88.0 (13.3) 227 84.2 (11.4) 12 87.8 (6.8) 169 88.7 (13.2) 1,117 87.3 (13.0) 
BMI (kg/m2717 27.5 (5.7) 229 26.4 (5.2) 12 26.5 (3.3) 174 27.9 (5.6) 1,132 27.3 (5.6) 
Fat mass (%) 711 38.1 (8.2) 226 37.0 (8.3) 12 32.1 (6.2) 172 37.4 (8.8) 1,121 37.7 98.3) 
Type 2 diabetes, n (%)      
 Self-reported 21 (2.7) 6 (2.5) 3 (20) 3(1.6) 33 (2.7) 
 Diagnosed 24 (3.1) 7 (2.9) 1(6.7) 2 (1.1) 38 (3.1) 
TraitNoncarriers (n = 771)
Heterozygote carriers (n = 241)
Homozygote carriers (n = 15)
Nongenotyped participants (n = 187)
All participants (n = 1,214)
Participants (n)Mean (SD)Participants (n)Mean (SD)Participants (n)Mean (SD)Participants (n)Mean (SD)Participants (n)Mean (SD)
Fasting plasma glucose (mmol/L) 771 5.3 (1.1) 240 5.1 (1.1) 15 5.0 (0.9) 133 5.2 (0.6) 1,159 5.2 (1.0) 
2-h plasma glucose (mmol/L) 573 5.2 (1.9) 156 5.6 (2.3) 8.9 (3.5) 92 5.6 (2.2) 827 5.3 (2.0) 
Fasting serum insulin (pmol/L) 768 65.1 (52.2) 241 52.2 (33.4) 15 50.2 (26.8) 130 67.2 (51.6) 1,154 62.5 (48.8) 
2-h serum insulin (pmol/L) 570 205.6 (234.0) 156 210.0 (193.4) 350.4 (300.4) 91 227.1 (228.5) 823 209.9 (226.8) 
HbA1c (%) 770 5.5 (0.5) 241 5.4 (0.4) 15 5.5 (0.3) 132 5.4 (0.3) 1,158 5.4 (0.5) 
HbA1c (mmol/mol) 770 36.0 (5.1) 241 35.0 (4.8) 15 37.0 (3.3) 132 35.0 (3.6) 1,158 36.0 (4.9) 
ISI0,120 (mg × L2/mmol × mU × min) 566 115.4 (53.9) 156 110.1 (53.1) 71.9 (47.7) 88 104.6 (46.0) 816 113.1 (53.1) 
Fasting serum triglyceride (mmol/L) 771 1.5 (1.1) 239 1.4 (0.6) 15 1.7 (0.6) 131 1.4 (0.9) 1,156 1.5 (1.0) 
Fasting serum total cholesterol (mmol/L) 771 5.2 (1.1) 239 5.2 (1.0) 15 5.5 (1.2) 131 5.3 (1.1) 1,156 5.2 (1.0) 
Fasting serum LDL cholesterol (mmol/L) 771 3.0 (0.9) 239 3.0 (0.9) 15 3.1 (1.0) 131 3.1 (0.9) 1,156 3.0 (0.9) 
Fasting serum HDL cholesterol (mmol/L) 771 1.5 (0.5) 239 1.6 (0.5) 15 1.6 (0.3) 131 1.6 (0.5) 1,156 1.5 (0.5) 
Albuminuria (mg/g) 739 0.1 (0.3) 234 0.0 (0.2) 15 0.1 (0.3) 129 0.1 (0.2) 1,117 0.1 (0.2) 
Waist circumference (cm) 709 88.0 (13.3) 227 84.2 (11.4) 12 87.8 (6.8) 169 88.7 (13.2) 1,117 87.3 (13.0) 
BMI (kg/m2717 27.5 (5.7) 229 26.4 (5.2) 12 26.5 (3.3) 174 27.9 (5.6) 1,132 27.3 (5.6) 
Fat mass (%) 711 38.1 (8.2) 226 37.0 (8.3) 12 32.1 (6.2) 172 37.4 (8.8) 1,121 37.7 98.3) 
Type 2 diabetes, n (%)      
 Self-reported 21 (2.7) 6 (2.5) 3 (20) 3(1.6) 33 (2.7) 
 Diagnosed 24 (3.1) 7 (2.9) 1(6.7) 2 (1.1) 38 (3.1) 

ISI0,120, Gutt insulin sensitivity index.

Table 2

Association of the TBC1D4 variant with metabolic traits in individuals without diabetes from the GOCADAN cohort

TraitPatients (n)Additive model
Recessive model
β95% CIP valueβ95% CIP value
Fasting plasma glucose (mmol/L) 976 −0.13 −0.19 to −0.06 2.21 × 10−4 −0.08 0.08–0.08 0.33 
2-h plasma glucose (mmol/L) 709 0.50 0.23–0.77 2.87 × 10−4 3.31 0.70–4.68 2.50 × 10−6 
HbA1c (%) 975 −0.03 −0.07 to 0.01 0.10 0.13 0.08–0.28 0.10 
HbA1c (mmol/mol) 975 −0.3 −0.8 to 0.1 0.10 1.4 0.9–3.1 0.10 
Fasting serum insulin (pmol/L) 974 −9.00 −14.20 to −3.79 7.03 × 10−4 −6.84 10.83–14.39 0.53 
2-h serum insulin (pmol/L) 706 13.8 −20.16 to 47.85 0.43 175.28 88.65–349.03 0.05 
ISI0,120 (SD) 703 −0.15 −0.31 to 0.01 0.07 −1.36 −2.18 to −0.53 1.21 × 10−3 
Fasting serum HDL cholesterol (mmol/L) 974 0.02 −0.05 to 0.08 0.64 0.09 0.13–0.35 0.46 
Fasting serum LDL cholesterol (mmol/L) 974 0.01 −0.12 to 0.13 0.93 0.09 0.26–0.60 0.73 
Fasting serum triglyceride (mmol/L) 974 −0.08 −0.20 to 0.04 0.18 0.32 0.25–0.80 0.20 
Fasting serum total cholesterol (mmol/L) 974 0.00 −0.13 to 0.14 0.98 0.35 0.28–0.89 0.21 
Waist circumference (cm) 919 −2.85 −4.70 to −0.99 2.47 × 10−3 −0.10 4.11–7.98 0.98 
Fat mass (%) 921 −0.99 −1.73 to −0.25 0.01 −0.61 1.64–2.60 0.71 
BMI (kg/m2931 −0.83 −1.63 to −0.04 0.04 −0.86 1.79–2.64 0.63 
Albuminuria (mg/g) 939 0.27 −0.06 to 0.60 0.11 3.38 3.31–9.86 0.31 
TraitPatients (n)Additive model
Recessive model
β95% CIP valueβ95% CIP value
Fasting plasma glucose (mmol/L) 976 −0.13 −0.19 to −0.06 2.21 × 10−4 −0.08 0.08–0.08 0.33 
2-h plasma glucose (mmol/L) 709 0.50 0.23–0.77 2.87 × 10−4 3.31 0.70–4.68 2.50 × 10−6 
HbA1c (%) 975 −0.03 −0.07 to 0.01 0.10 0.13 0.08–0.28 0.10 
HbA1c (mmol/mol) 975 −0.3 −0.8 to 0.1 0.10 1.4 0.9–3.1 0.10 
Fasting serum insulin (pmol/L) 974 −9.00 −14.20 to −3.79 7.03 × 10−4 −6.84 10.83–14.39 0.53 
2-h serum insulin (pmol/L) 706 13.8 −20.16 to 47.85 0.43 175.28 88.65–349.03 0.05 
ISI0,120 (SD) 703 −0.15 −0.31 to 0.01 0.07 −1.36 −2.18 to −0.53 1.21 × 10−3 
Fasting serum HDL cholesterol (mmol/L) 974 0.02 −0.05 to 0.08 0.64 0.09 0.13–0.35 0.46 
Fasting serum LDL cholesterol (mmol/L) 974 0.01 −0.12 to 0.13 0.93 0.09 0.26–0.60 0.73 
Fasting serum triglyceride (mmol/L) 974 −0.08 −0.20 to 0.04 0.18 0.32 0.25–0.80 0.20 
Fasting serum total cholesterol (mmol/L) 974 0.00 −0.13 to 0.14 0.98 0.35 0.28–0.89 0.21 
Waist circumference (cm) 919 −2.85 −4.70 to −0.99 2.47 × 10−3 −0.10 4.11–7.98 0.98 
Fat mass (%) 921 −0.99 −1.73 to −0.25 0.01 −0.61 1.64–2.60 0.71 
BMI (kg/m2931 −0.83 −1.63 to −0.04 0.04 −0.86 1.79–2.64 0.63 
Albuminuria (mg/g) 939 0.27 −0.06 to 0.60 0.11 3.38 3.31–9.86 0.31 

Results are shown for an additive and a recessive genetic model. The number of patients for each trait includes those with genotype data for the specific variant and phenotypic data for the specific trait. β is the effect size (change in the trait expressed in its unit of measurement if one copy of the effect allele is present) estimated using untransformed values. All P values <0.05 are considered statistically significant and appear in boldface. ISI0,120, Gutt insulin sensitivity index.

Among the rest of the type 2 diabetes–related traits, we found a marginal decrease in BMI, waist circumference, and percentage fat mass values among heterozygotes without diabetes (Table 2) and a lower BMI among heterozygote carriers with diabetes (β = −5.9 kg/m2; P = 1.7 × 10−2) (Supplementary Table 2). However, we caution that these results require further replication in additional cohorts. After Bonferroni correction for multiple testing, the P value cutoff for statistical significance decreased to 3.5 × 10−3. At this overly conservative threshold, all of our results remained significant, except for percentage fat mass and BMI. We observed consistent results when testing the association of the p.Arg684Ter variant with inverse normalized traits (Supplementary Tables 3 and 4). No association was found between the p.Arg684Ter variant and type 2 diabetes status among the 51 cases and 976 controls in the GOCADAN cohort in either the additive or recessive models (Supplementary Table 5).

Finally, we investigated how many TBC1D4 carriers and noncarriers with available 2-h OGTT results would have been diagnosed with prediabetes or type 2 diabetes only by meeting the 2-h OGTT diagnostic criterion. This is clinically relevant because the variant leads to increased 2-h OGTT but decreased fasting glucose, with no change in HbA1c. Among the TBC1D4 carriers, 50 had type 2 diabetes or prediabetes, of whom 16 (32%) would have been diagnosed only by using the 2-h OGTT (30% of heterozygote carriers and 50% of homozygote carriers) (Table 3). By contrast, failure to perform a 2-h OGTT would have missed a diagnosis of type 2 diabetes or prediabetes in 16 of 203 noncarriers (8%) (Table 3). The odds of missing a diagnosis of type 2 diabetes or prediabetes was 5.4 (95% CI 2.5–12) when comparing carriers with noncarriers. Table 3 describes the rate of missed diagnoses of type 2 diabetes and prediabetes separately. Supplementary Table 6 provides the prevalences of prediabetes and type 2 diabetes in the subsample of GOCADAN participants with available OGTT results, according to their number of TBC1D4 variant copies.

Table 3

Classification of TBC1D4 variant carriers and noncarriers according to their type 2 diabetes and prediabetes status and their diagnostic criteria for type 2 diabetes

Criteria for T2D diagnosisT2D or prediabetes
T2D and prediabetes cases diagnosed only by OGTT (%)T2D
Prediabetes
Only 2-h glucose (OGTT)FG and/or HbA1cOnly 2-h glucose (OGTT)FG and/or HbA1cOnly 2-h glucose (OGTT)FG and/or HbA1c
TBC1D4 noncarriers 16 187 11 176 
TBC1D4 heterozygote carriers 14 32 30 10 29 
TBC1D4 homozygote carriers 50 
TBC1D4 carriers (homozygote + heterozygote) 16 34 32 11 31 
Criteria for T2D diagnosisT2D or prediabetes
T2D and prediabetes cases diagnosed only by OGTT (%)T2D
Prediabetes
Only 2-h glucose (OGTT)FG and/or HbA1cOnly 2-h glucose (OGTT)FG and/or HbA1cOnly 2-h glucose (OGTT)FG and/or HbA1c
TBC1D4 noncarriers 16 187 11 176 
TBC1D4 heterozygote carriers 14 32 30 10 29 
TBC1D4 homozygote carriers 50 
TBC1D4 carriers (homozygote + heterozygote) 16 34 32 11 31 

Data are n unless otherwise indicated. The numbers of individuals with only abnormal 2-h plasma glucose on the OGTT (according to the American Diabetes Association criteria for type 2 diabetes [T2D] and prediabetes) represent cases of missed T2D or prediabetes diagnoses. FG, fasting glucose.

Genetic disruption of TBC1D4 is present in approximately 27% of Inuit across the North American Arctic and results in a high rate of missed diagnoses of type 2 diabetes and/or prediabetes unless 2-h OGTTs are performed. This is because TBC1D4 disruption is associated with an elevation in postprandial glucose and decreases fasting glucose, while having no effect on HbA1c. This suggests that since 2-h OGTTs are rarely performed in clinical practice (16), diagnostic strategies for type 2 diabetes and prediabetes in the Inuit should now always include OGTT and/or genotypic risk stratification. These findings demonstrate an opportunity to apply the principles of precision medicine.

The findings confirm and extend the main findings from Moltke et al. (3) among Greenlandic Inuit to the North American Inuit. While effect sizes and directions of the TBC1D4 mutation were similar in our populations, the uncertain effect on type 2 diabetes risk in our population may be a result of the relatively small number of participants with type 2 diabetes in the GOCODAN cohort (51 cases/976 controls in the GOCADAN cohort vs. 220 cases/1,810 controls in the Greenlandic cohort).

There is a plausible pathophysiologic explanation for the abovementioned findings. TBC1D4 acts as a mediator of insulin-stimulated glucose uptake through GLUT4 mobilization (17). Murine experiments have shown that knocking down Tbc1d4 causes a significant decrease in basal plasma glucose and insulin-stimulated glucose uptake in muscle and adipose tissue (18) In humans, these findings are supported by a case report of familial postprandial hyperinsulinemia associated with a TBC1D4 variant (19).

Carrying a mutation that solely increases postprandial glucose may influence type 2 diabetes complications. Interestingly, postprandial glucose seems to be a better predictor of cardiovascular disease and all-cause mortality than fasting glucose, and this association is linear (20,21). Specifically, postprandial glucose >7.8 mmol/L has been associated with all-cause mortality (22), and postprandial glucose >10.0 mmol/L has been associated with microvascular complications and myocardial infarctions (23). Some of these studies found that these effects of postprandial glucose occurred in addition to effects from fasting glucose and HbA1c. These findings and others have prompted the American and Canadian Diabetes Associations to include OGTT as part of the diagnostic criteria for type 2 diabetes (13,24). However, all the above-mentioned studies have been done in non-Inuit populations, leaving open the possibility that the impact of postprandial hyperglycemia might be different in Inuit carrying the TBC1D4 variant.

Our findings suggest that the TBC1D4 variant is relatively common among the Eastern Canadian Nunavik Inuit and the Alaskan Inuit and potentially in other Arctic populations. These findings are clinically relevant given the increasing prevalence of diabetes in these populations (6,25,26). With an expected allele frequency of approximately 14%, we anticipate that 2% of the North American Inuit population are homozygous and 25% are heterozygous for the TBC1D4 variant. Among the possible public health implications of this finding, we speculate that ascertainment of the TBC1D4 variant could serve as a useful screening test to predict patients with prediabetes and type 2 diabetes with pronounced insulin resistance and exclusively abnormal postprandial glucose concentrations.

The clinical relevance of these findings is magnified since 2-h OGTTs are rarely performed in clinical practice; type 2 diabetes and prediabetes are generally diagnosed using fasting glucose, random glucose, and HbA1c (13,24,27). Although the frequency of OGTT among the Inuit population has not been reported, a population-based registry of laboratory tests for type 2 diabetes diagnosis in Ontario, Canada (16), showed that over 20 years (1995–2005) there was a 28% increase in fasting glucose tests, which were prescribed for 37% of adults without known type 2 diabetes in 2005. During the same period, the percentage of HbA1c tests also increased from 1.7% in 1995 to 6.0%. Remarkably, less than 1% of this population underwent OGTT testing in any year between 1995 and 2005, although OGTT has been recommended for type 2 diabetes screening by the Canadian Diabetes Association since 2003 (24). Thus current clinical practice would miss an important proportion of the Inuit population with type 2 diabetes unless OGTTs are performed.

We did not observe a significant increase of missed type 2 diabetes diagnoses without OGTT among TBC1D4 carriers, likely because of the small sample of patients with type 2 diabetes and OGTT values (n = 26) in the GOCADAN cohort; nevertheless, we found a sevenfold increase of the odds of missing a diagnosis of prediabetes. Considering that TBC1D4 carriers with prediabetes and diabetes (27% of the Inuit population) have a fivefold increase in their combined risk to remain undiagnosed (or a 32% chance to remain undiagnosed) without measuring 2-h glucose, performing OGTTs for these Inuit individuals seems to be of clinical relevance. Given that performing an OGTT as a screening test for type 2 diabetes and prediabetes can be particularly cumbersome in the Inuit population, genetic screening for the TBC1D4 mutation might represent a clinically useful alternative, especially if the cost-effectiveness of this approach is shown. Homozygous and heterozygous carriers could then be further investigated by OGTTs. Of note, genetic screening tests for colon cancer have been successfully implemented in the Quebec Inuit population (28).

An important limitation of this study was that the degree of admixture of the Inuit population studied could not be quantified. Moltke et al. (3) showed an important influence of admixture on the MAF of the TBC1D4 variant among Greenlandic Inuit, but the degree of admixture among the GOCADAN population cannot be likewise assessed because of a lack of dense genotyping. Nonetheless, this does not change the clinical relevance of these findings, since an important proportion of the North American Inuit population carries the TBC1D4 variant.

Detecting the presence of the TBC1D4 variant in the Inuit population might be important not only for diagnostic purposes but also for therapeutic decision making. TBC1D4 carriers represent a subgroup of patients with type 2 diabetes who may be good candidates for specific pharmacologic treatments. This is because the mechanism of action of TBC1D4 could suggest a better response of carriers to insulin sensitizers (29,30) than to other first-line type 2 diabetes oral antihyperglycemic drugs, such as sulfonylureas. However, this hypothesis should be tested in clinical trials. Finally, our findings could be applied in pharmacologic research, as TBC1D4 may represent a potential drug target for type 2 diabetes.

In summary, genetic disruption of TBC1D4 is common across the North American Inuit and leads to a higher postprandial glucose. Type 2 diabetes and prediabetes should be assessed in the population using OGTT, either after genotypic risk stratification or by testing the entire at-risk population.

See accompanying articles, pp. 1854, 1858, 1870, 1874, 1879, 1896, 1902, 1909, and 1915.

Duality of Interest. No potential conflicts of interest relevant to this article were reported.

AuthorContributions. D.M., J.W.K., K.H., D.E.N., J.G.U., A.G.C., and J.B.R. conducted the analyses. D.M. and J.B.R. wrote the first draft of the manuscript. J.W.K., K.H., D.E.N., J.G.U., and A.G.C. conducted the sequencing of the TBC1D4 variant in the GOCADAN cohort. S.Z., P.X., and G.R. conducted the sequencing of the TBC1D4 variant in the Quebec Inuit cohort. J.B.R. conceived the experiment. C.M.G., P.B., and S.C. provided important feedback on the manuscript. J.B.R. is the guarantor of this work and, as such, had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

1.
Hudson KLR, Patrick-Lake B, Burchard EG, et al.; Precision Medicine Initiative (PMI) Working Group. The Precision Medicine Initiative Cohort Program - Building a Research Foundation for 21st Century Medicine. Bethesda, MD, National Institutes of Health, 2015
2.
Collins
FS
,
Varmus
H
.
A new initiative on precision medicine
.
N Engl J Med
2015
;
372
:
793
795
[PubMed]
3.
Moltke
I
,
Grarup
N
,
Jørgensen
ME
, et al
.
A common Greenlandic TBC1D4 variant confers muscle insulin resistance and type 2 diabetes
.
Nature
2014
;
512
:
190
193
[PubMed]
4.
Jørgensen
ME
,
Bjeregaard
P
,
Borch-Johnsen
K
.
Diabetes and impaired glucose tolerance among the inuit population of Greenland
.
Diabetes Care
2002
;
25
:
1766
1771
[PubMed]
5.
Helgason
A
,
Pálsson
G
,
Pedersen
HS
, et al
.
mtDNA variation in Inuit populations of Greenland and Canada: migration history and population structure
.
Am J Phys Anthropol
2006
;
130
:
123
134
[PubMed]
6.
Howard
BV
,
Devereux
RB
,
Cole
SA
, et al
.
A genetic and epidemiologic study of cardiovascular disease in Alaska natives (GOCADAN): design and methods
.
Int J Circumpolar Health
2005
;
64
:
206
221
[PubMed]
7.
Zhou
S
,
Xiong
L
,
Xie
P
, et al
.
Increased missense mutation burden of fatty acid metabolism related genes in Nunavik Inuit population
.
PLoS One
2015
;
10
:
e0128255
[PubMed]
8.
Abecasis
GR
,
Altshuler
D
,
Auton
A
, et al.;
1000 Genomes Project Consortium
.
A map of human genome variation from population-scale sequencing [published correction appears in Nature 2011;473:544]
.
Nature
2010
;
467
:
1061
1073
[PubMed]
9.
Gutt
M
,
Davis
CL
,
Spitzer
SB
, et al
.
Validation of the insulin sensitivity index (ISI(0,120)): comparison with other measures
.
Diabetes Res Clin Pract
2000
;
47
:
177
184
[PubMed]
10.
Blangero
J
,
Goring
HH
,
Kent
JW
 Jr
, et al
.
Quantitative trait nucleotide analysis using Bayesian model selection
.
Hum Biol
2005
;
77
:
541
559
[PubMed]
11.
Almasy
L
,
Blangero
J
.
Multipoint quantitative-trait linkage analysis in general pedigrees
.
Am J Hum Genet
1998
;
62
:
1198
1211
[PubMed]
12.
Gelman
A
,
Hill
J
,
Yajima
M
.
Why we (usually) don’t have to worry about multiple comparisons
.
J Res Educ Eff
2012
;
5
:
189
211
13.
American Diabetes Association
.
Diagnosis and classification of diabetes mellitus
.
Diabetes Care
2010
;
33
(
Suppl. 1
):
S62
S69
[PubMed]
14.
Mehta
CR
,
Patel
NR
.
A hybrid algorithm for Fishers exact test in unordered rxc contingency-tables
.
Commun Stat Theory Methods
1986
;
15
:
387
403
15.
Clarkson
DB
,
Fan
YA
,
Joe
H
.
A remark on algorithm-643 - Fexact - an algorithm for performing Fishers exact test in R X C contingency-tables
.
ACM Trans Math Software
1993
;
19
:
484
488
16.
Wilson
SE
,
Lipscombe
LL
,
Rosella
LC
,
Manuel
DG
.
Trends in laboratory testing for diabetes in Ontario, Canada 1995–2005: a population-based study
.
BMC Health Serv Res
2009
;
9
:
41
[PubMed]
17.
Sano
H
,
Kane
S
,
Sano
E
, et al
.
Insulin-stimulated phosphorylation of a Rab GTPase-activating protein regulates GLUT4 translocation
.
J Biol Chem
2003
;
278
:
14599
14602
[PubMed]
18.
Wang
HY
,
Ducommun
S
,
Quan
C
, et al
.
AS160 deficiency causes whole-body insulin resistance via composite effects in multiple tissues
.
Biochem J
2013
;
449
:
479
489
[PubMed]
19.
Dash
S
,
Sano
H
,
Rochford
JJ
, et al
.
A truncation mutation in TBC1D4 in a family with acanthosis nigricans and postprandial hyperinsulinemia
.
Proc Natl Acad Sci U S A
2009
;
106
:
9350
9355
[PubMed]
20.
DECODE Study Group
.
Age- and sex-specific prevalences of diabetes and impaired glucose regulation in 13 European cohorts
.
Diabetes Care
2003
;
26
:
61
69
[PubMed]
21.
Sorkin
JD
,
Muller
DC
,
Fleg
JL
,
Andres
R
.
The relation of fasting and 2-h postchallenge plasma glucose concentrations to mortality: data from the Baltimore Longitudinal Study of Aging with a critical review of the literature
.
Diabetes Care
2005
;
28
:
2626
2632
[PubMed]
22.
Cavalot
F
,
Pagliarino
A
,
Valle
M
, et al
.
Postprandial blood glucose predicts cardiovascular events and all-cause mortality in type 2 diabetes in a 14-year follow-up: lessons from the San Luigi Gonzaga Diabetes Study
.
Diabetes Care
2011
;
34
:
2237
2243
[PubMed]
23.
Sarwar
N
,
Gao
P
,
Seshasai
SR
, et al.;
Emerging Risk Factors Collaboration
.
Diabetes mellitus, fasting blood glucose concentration, and risk of vascular disease: a collaborative meta-analysis of 102 prospective studies [published correction appears in Lancet 2010;376:958]
.
Lancet
2010
;
375
:
2215
2222
[PubMed]
24.
Canadian Diabetes Association Clinical Practice Guidelines Expert Committee
.
Canadian Diabetes Association 2003 clinical practice guidelines for the prevention and management of diabetes in Canada
.
Can J Diabetes
2003
;
27(Suppl. 2)
:
S10
S13
25.
Carter
EA
,
MacCluer
JW
,
Dyke
B
, et al
.
Diabetes mellitus and impaired fasting glucose in Alaska Eskimos: the Genetics of Coronary Artery Disease in Alaska Natives (GOCADAN) study
.
Diabetologia
2006
;
49
:
29
35
[PubMed]
26.
Pelletier
C
,
Dai
S
,
Roberts
KC
,
Bienek
A
,
Onysko
J
,
Pelletier
L
.
Report summary. Diabetes in Canada: facts and figures from a public health perspective
.
Chronic Dis Inj Can
2012
;
33
:
53
54
[PubMed]
27.
Simmons
RK
,
Echouffo-Tcheugui
JB
,
Griffin
SJ
.
Screening for type 2 diabetes: an update of the evidence
.
Diabetes Obes Metab
2010
;
12
:
838
844
[PubMed]
28.
Li
L
,
Hamel
N
,
Baker
K
, et al
.
A homozygous PMS2 founder mutation with an attenuated constitutional mismatch repair deficiency phenotype
.
J Med Genet
2015
;
52
:
348
352
[PubMed]
29.
Della-Morte
D
,
Palmirotta
R
,
Rehni
AK
, et al
.
Pharmacogenomics and pharmacogenetics of thiazolidinediones: role in diabetes and cardiovascular risk factors
.
Pharmacogenomics
2014
;
15
:
2063
2082
[PubMed]
30.
Soccio
RE
,
Chen
ER
,
Lazar
MA
.
Thiazolidinediones and the promise of insulin sensitization in type 2 diabetes
.
Cell Metab
2014
;
20
:
573
591
[PubMed]
Readers may use this article as long as the work is properly cited, the use is educational and not for profit, and the work is not altered. More information is available at http://www.diabetesjournals.org/content/license.

Supplementary data