2.1. Metabolic profile of the cohort
A complete characterization was conducted on 1,262 individuals, comprising 1,035 women (mean age 34.7 ± 10.23 years; mean BMI 24.7 ± 5.30 kg/m2) and 227 men (mean age 36.5 ± 13.13 years; mean BMI 25.6 ± 4.55 kg/m2). Among women, 454 participants had a positive history of gestational diabetes mellitus (GDM) and 189 were diagnosed with polycystic ovary syndrome (PCOS) according to the European Society of Human Reproduction and Embryology consensus [25]. Based on our examination, 17 participants (12 women, 1.2% and 5 men, 2.2%; p = 0.21) were newly diagnosed with T2DM. Impaired fasting glucose was present in 100 individuals (76 women, 7.4% and 24 men, 10.8%; p = 0.10), impaired glucose tolerance based on 120th min of the OGTT was detected in 83 participants (69 women, 6.7% and 14 men, 6.3%; p = 1.00). Impaired glucose metabolism (IGM), encompasing either IFG, IGT, both disorders concurrently or overt T2DM, was present in 167 participants (131 women, 12.6% and 36 men, 15.9%; p = 0.19). According to the NCEP_ATPIII criteria [26], 139 participants (102 women, 10.0% and 37 men, 16.4%; p = 0.01) suffered from metabolic syndrome. Table 1 reports detailed anthropometric data and biochemical characteristics of the examined subjects. In the study, special attention was paid to glucose metabolism. Therefore, indices of insulin sensitivity were calculated for fasting as well as for dynamic conditions following the glucose load. Pancreatic beta cell function was also characterized for fasting state and for dynamic conditions. Details on the calculation of the indices listed in Table 1 are reported in methodological section 4.2. “Anthropometric and metabolic characterization of the subjects“.
2.2. Representation of individual types of curves
Among the 1,262 glycemic curves evaluated, 633 (50%) were monophasic, 221 (17.5%) biphasic, 351 (28%) triphasic, whereas 57 (4.5%) showed more complex patterns.
People pertaining to the monophasic and triphasic groups were on average 3 years older than people in the biphasic group. BMI was on average 1.3 kg/m2 higher in people within the monophasic group compared to the other groups. Data presenting age and BMI as the basic characteristics of the given groups, based on the glycemic trajectories, are shown in Table 2.
Table 1
Anthropometric and metabolic characterization of the subjects.
parameter (units) | median [95% LCL; 95% UCL] |
n | 1,262 |
age (years) | 33.3 [32.6; 34.0] |
systolic blood pressure (mmHg) | 115.0 [114.0; 115.0] |
diastolic blood pressure (mmHg) | 72.0 [71.0; 73.0] |
BMI (kg/m2) | 23.9 [23.5; 24.3] |
WHR women | 0.764 [0.759; 0.768] |
WHR men | 0.864 [0.849; 0.877] |
basal glycemia (mmol/l) | 4.8 [4.7; 4.8] |
AUCGlycemia(mmol*min/l) | 1,046 [1,031; 1,059] |
basal insulinemia (mIU/l) | 6.16 [5.90; 6.30] |
AUCInsulin (pmol*min/l) | 33,885 [32,598; 35,091] |
basal C-peptide (nmol/l) | 0.59 [0.58; 0.60] |
AUCC−peptide (pmol*min/l) | 3.6*105 [3.5*105;3.7*105] |
hepatic insulin extraction (%) | 68.1 [67.5; 68.8] |
insulin sensitivity/resistance |
HOMA IR | 1.30 [1.26; 1.36] |
OGIS 120 min (ml/min/m2) | 457.7 [453.5; 462.1] |
ISICOMP ([(mg/dl)2(µU/ml)2]−1/2) | 8.48 [8.18; 8.81] |
MCRest (ml/min/kg) | 9.87 [9.75; 9.99] |
Si(oral) ((ml/min/kg)/(µU/ml)) | 0.15 [0.15; 0.16] |
PREDIM (mg/min/kg) | 6.97 [6.83; 7.17] |
beta cell function |
HOMA B (mIU/mmol) | 103.5 [98.8; 106.2] |
IGI (pmol/mmol) | 88.5 [84.2; 95.1] |
Ins0/Glc0 (pmol/mmol) | 7.68 [7.40; 8.00] |
AUCInsulin/AUCGlc (pmol/mmol) | 32.7 [31.9; 33.9] |
IGI*ISICOMP | 264.9 [257.9; 270.2] |
lipid spectrum |
cholesterol (mmol/l) | 4.59 [4.54; 4.65] |
HDL cholesterol women (mmol/l) | 1.57 [1.54; 1.60] |
HDL cholesterol men (mmol/l) | 1.26 [1.21; 1.32] |
LDL cholesterol (mmol/l) | 2.56 [2.52; 2.63] |
TAG (mmol/l) | 0.86 [0.82; 0.89] |
liver enzymes |
ALT (ukat/l) | 0.30 [0.30; 0.31] |
AST (ukat/l) | 0.35 [0.34; 0.36] |
GGT (ukat/l) | 0.23 [0.22; 0.24] |
thyroid hormones |
TSH (mIU/l) | 2.25 [2.17; 2.33] |
fT3 (pmol/l) | 4.90 [4.85; 4.97] |
fT4 (pmol/l) | 15.2 [15.0; 15.3] |
BMI—body mass index; WHR—waist-hip ratio; AUCGlycemia—area under the glycemic curve; AUCInsulin—area under the insulinemic curve; AUCC−peptide—area under the C-peptide curve; HOMA IR—Homeostatic Model Assessment of Insulin Resistance; OGIS—Oral Glucose Insulin Sensitivity; ISICOMP—Insulin Sensitivity Composite index; PREDIM—Peripheral Insulin Sensitivity Index; HOMA B—Homeostatic Model Assessment of Beta-cell function; IGI—Insulinogenic index; IGI*ISICOMP—product of the IGI and the ISICOMP index; HDL—high-density lipoprotein; LDL—low-density lipoprotein; TAG—triacylglycerols; ALT—alanine aminotransferase; AST—aspartate aminotransferase; GGT—γ-glutamyl transferase; TSH—thyrotropin; fT3—free triiodothyronine; fT4—free thyroxine.
Table 2
Age and BMI of the groups based on the shape of glycemic curves.
n = 1,262 | monophasic n = 633 | biphasic n = 221 | triphasic n = 351 | multiphasic n = 57 | P-level |
age (years) | 34.1[33.2; 35.1] | 31.4[28.3; 32.9] | 34.0[32.4; 34.7] | 28.7[26.6; 30.6] | < 0.01 |
BMI (kg/m2) | 24.6[24.1; 25.1] | 23.3[22.6; 24.1] | 23.2[22.7; 23.8] | 23.5[22.2; 24.2] | < 0.01 |
Data are given as median [95% LCL; 95% UCL], P-level according to Kruskal Wallis Z-value test
The representation of men and women within individual types of curves was significantly different with the most noticeable difference in distribution of the bi- and triphasic curves. Statistical comparison showed that a higher percentage of men had a biphasic curve (33% vs. 14% of women) and a higher percentage of women had a triphasic curve (30% vs. 19% of men), Chi2 = 46; p < 0.0001, power of the test > 0.99.
The monophasic shape of the curve was predominant among newly diagnosed diabetics (15 out of 17; i.e. 88%), but also 70% of subjects with IFG and 78% with IGT in 120th min of the test. Furthermore, 75% of subjects meeting the criteria for metabolic syndrome also fell into the group with a monophasic curve. Furthermore, monophasic group included 53% of women with a history of GDM and 46% of women diagnosed with PCOS. These findings suggest that monophasic group shows worse metabolic health compared to individuals with biphasic, triphasic or more complex curves and is associated with the presence of many components of the metabolic syndrome. In contrast, the biphasic trajectory (and the rarer multiphasic pattern) appears most beneficial in terms of metabolic health as it is associated with the best glycemic and lipid profile. We have previously published a detailed analysis of the anthropometric and metabolic differences between curve-derived groups, as well as gender-specific variations based on glycemic trajectory shape [24]. In this study, we will focus on the genetic background possibly underlying these observations.
2.3. Genetic determination of glycemic curve shapes
Of the 37 SNPs listed in methodological section 4.4. that were tested in candidate genes associated with glucose metabolism (NME7, ATP1B1, BLZF1, GCK, KCNJ11, LRP5, PPARGC1A, PPARG, SLC30A8, TCF7L2, FTO, ZBTB16, THADA, PICALM, BIN1, CLU, CR1, MTNR1B, PNPLA3), only five variants in the NME7 gene were associated with the shape of the glycemic curve during the 3-hour OGTT. Four polymorphisms in this gene were significantly associated with the shape of the curves, while a fifth polymorphism showed borderline significance. Table 3 presents the proportions of individual NME7 genotypes for each glycemic trajectory. In Supplementary Material 1, we report detailed Chi-squared statistical outputs with exact numbers of these five genetic NME7 variants and the result of the analyses carried out separately for men and women (Supplementary Tables S1-S5).
Table 3
Genotype and allele frequencies of NME7 gene polymorphisms in individual groups of glycemic trajectories.
GENOTYPES of NME7_rs4656659 (%) | ALLELES (%) | | | |
| TT | CT | CC | T | C | | STATGenotypic distribution: |
monophasic | 45 | 43 | 12 | 67 | 33 | | Chi2 = 16.43 | |
biphasic | 34 | 54 | 12 | 61 | 39 | | Power = 0.88 | |
triphasic | 39 | 53 | 8 | 66 | 34 | | P-level = 0.012 | |
multiphasic | 48 | 40 | 12 | 68 | 32 | | | |
GENOTYPES of NME7_rs2157597 (%) | ALLELES (%) | | | |
| CC | CT | TT | C | T | | STATGenotypic distribution: |
monophasic | 52 | 40 | 8 | 71 | 29 | | Chi2 = 12.64 | |
biphasic | 42 | 48 | 10 | 66 | 34 | | Power = 0.76 | |
triphasic | 48 | 47 | 5 | 72 | 28 | | P-level = 0.049 | |
multiphasic | 52 | 41 | 7 | 73 | 27 | | | |
GENOTYPES of NME7_rs10732287 (%) | ALLELES (%) | | | |
| CC | CT | TT | C | T | | STATGenotypic distribution: |
monophasic | 46 | 43 | 11 | 67 | 33 | | Chi2 = 19.39 | |
biphasic | 59 | 33 | 8 | 76 | 24 | | Power = 0.93 | |
triphasic | 43 | 48 | 9 | 67 | 33 | | P-level = 0.004 | |
multiphasic | 37 | 48 | 15 | 61 | 39 | | | |
GENOTYPES of NME7_rs4264046 (%) | ALLELES (%) | | | |
| CC | CT | TT | C | T | | STATGenotypic distribution: |
monophasic | 29 | 49 | 22 | 53 | 47 | | Chi2 = 16.42 | |
biphasic | 39 | 44 | 17 | 61 | 39 | | Power = 0.88 | |
triphasic | 28 | 56 | 16 | 56 | 44 | | P-level = 0.012 | |
multiphasic | 25 | 53 | 22 | 52 | 48 | | | |
GENOTYPES of NME7_rs10800438 (%) | ALLELES (%) | | | |
| GG | GT | TT | G | T | | STATGenotypic distribution: |
monophasic | 31 | 49 | 20 | 56 | 44 | | Chi2 = 14.21 | |
biphasic | 42 | 43 | 15 | 64 | 36 | | Power = 0.82 | |
triphasic | 32 | 53 | 15 | 59 | 41 | | P-level = 0.027 | |
multiphasic | 24 | 57 | 19 | 53 | 47 | | | |
Data are given as percentages, P-level according to Chi-square test
Proceeding from the arrangement of individual NME7 SNPs on the chromosome 1q24.2 (Fig. 1), we found an association of the biphasic type of curves with five tightly linked polymorphisms. More precisely, as shown in Table 3, the wild-type alleles as well as wild-type homozygotes of the rs4656659 and rs2157597 SNPs were less frequent in the biphasic group of curves (p = 0.01 and 0.05, respectively). Conversely, the wild-type alleles and wild-type homozygotes of three other linked NME7 SNPs, rs10732287, rs4264046 and rs10800438, were more frequent in the biphasic group (p < 0.01, p = 0.01, and p = 0.03, respectively).
As demonstrated by the results of linkage analysis in Fig. 2, all these five NME7 variants are in very strong linkage disequilibrium (LD) and form one LD block marked as Block 1. The sixth tested polymorphism of the NME7 gene, rs4656671, already falls into a different LD block (marked as Block 2) and does not show a significant association with the shape of the glycemic curve.
2.4. Haplotype analysis
Carefully performed haplotype analysis showed that the two most abundant haplotypes, which are inherited as intact blocks with very rare recombination events, are precisely the ones indicated by association analysis. The first haplotype carries the wild-type variants of SNPs rs4656659 and rs2157597 in combination with the minor variants of SNPs rs10732287, rs4264046 and rs10800438, forming the TCTTT haplotype. It was calculated to be, either in the heterozygous or homozygous constellation, highly likely present in 52.2%. Presence of this haplotype was statistically significantly lower in individuals with biphasic course of the curve (40%) in comparison with other types of curves (55%), Chi2 = 15.41, p < 0.01. The second haplotype, essentially the inverse of the first, includes the minor variants of SNPs rs4656659 and rs2157597 in combination with wild-type variants of SNPs rs10732287, rs4264046 and rs10800438, forming the CTCCG combination. Its presence - again either in heterozygous or homozygous constellation - was calculated as most probable in 49.7% individuals in our cohort. Presence of this haplotype was significantly higher in individuals with biphasic course of the curve (57%) compared to those with other curve types (48%), Chi2 = 5.05, p = 0.02. The estimated frequency of other haplotypes with different SNP combinations was already significantly lower, as demonstrated in Table 4, which is a consequence of a very tight genetic link between the polymorphisms in a given haplotype constellation. Three individuals had to be excluded from the haplotype analysis due to insufficient genotyping data, as indicated in Table 4.
Table 4
Block 1 haplotypes with a higher than rare (2.5%) representation sorted by frequency.
haplotype | number of carriers women (n = 1,033) | number of carriers men (n = 226) | number of carriers all (n = 1,259) |
1. TCTTT | 550 (53.2%) | 107 (47.3%) | 657 (52.2%) |
2. CTCCG | 509 (49.3%) | 117 (51.8%) | 626 (49.7%) |
3. TCCCG | 377 (36.5%) | 91 (40.3%) | 468 (37.2%) |
4. TCCTT | 209 (20.2%) | 47 (20.8%) | 256 (20.3%) |
5. CCCCG | 110 (10.6%) | 29 (12.8%) | 139 (11.0%) |
6. TCCTG | 54 (5.2%) | 11 (4.9%) | 65 (5.2%) |
Individuals who carry at least one copy of a given haplotype are listed, some may carry two identical copies of the same haplotype.
2.5. Haplotypes and biochemical parameters
Finding that the biphasic course of the glycemic curve is indicative of considerable health benefits compared to the other curves, especially the monophasic one, led us to consider whether a better biochemical profile can also be found in individuals who were assigned to the CTCCG haplotype by haplotype analysis, as this haplotype was shown to be linked with biphasic course of glycemic curve. We compared carriers of this haplotype (either in heterozygous or homozygous constellation) with non-carriers, and we focused on glucose metabolism and components of the metabolic syndrome. In connection with the data presented, it is important that the representation of the CTCCG haplotype was the same in both sexes (49.3% in women and 51.8% in men; p = 0.51), therefore the cohort was evaluated as a whole except for parameters with different reference ranges for women and men (WHR, HDL cholesterol). Table 5 demonstrates anthropometric and metabolic characterization of the subjects depending on the presence of the CTCCG haplotype as estimated by haplotype analysis.
Although no significant differences between carriers and noncarriers of a given haplotype were observed in terms of anthropometry or blood pressure, glucose metabolism and calculated indices of insulin sensitivity clearly indicate that carrying of the CTCCG haplotype is systematically reflected in biochemical profile and confirms our hypothesis that those carrying the haplotype have a metabolically healthier profile. This is evident not only when it comes to fasting levels of glucose, insulin and C-peptide, but it is also reflected in dynamic parameters as AUCGlycemia, AUCInsulin, AUCC−peptide and, which is remarkable, all presented indices of insulin sensitivity.
Table 5
Anthropometric and metabolic characterization of the subjects according to the presence of the CTCCG haplotype as estimated by haplotype analysis.
parameter (units) | CTCCG haplotype + | CTCCG haplotype - | P-level |
n | 626 | 633 | n/a |
age (years) | 33.2 [32.3; 34.3] | 33.5 [32.4; 34.1] | 0.58 |
systolic blood pressure (mmHg) | 114 [113; 115] | 115 [114; 116] | 0.65 |
diastolic blood pressure (mmHg) | 72 [71; 73] | 72 [71; 74] | 0.78 |
BMI (kg/m2) | 23.7 [23.2; 24.1] | 24.2 [23.6; 24.5] | 0.10 |
WHR women | 0.760 [0.751; 0.765] | 0.769 [0.760; 0.777] | 0.11 |
WHR men | 0.862 [0.845; 0.877] | 0.867 [0.848; 0.888] | 0.49 |
basal glycemia (mmol/l) | 4.7 [4.7; 4.8] | 4.8 [4.8; 4.8] | 0.01 |
AUCGlycemia(mmol*min/l) | 1,029 [1,008; 1,052] | 1,059 [1,041; 1,079] | < 0.01 |
basal insulinemia (mIU/l) | 5.9 [5.5; 6.2] | 6.3 [6.0; 6.7] | 0.04 |
AUCInsulin (pmol*min/l) | 31,757 [30,510; 33,309] | 35,793 [34,209; 38,421] | < 0.01 |
basal C-peptide (nmol/l) | 0.57 [0.56; 0.59] | 0.61 [0.59; 0.63] | < 0.01 |
AUCC−peptide (pmol*min/l) | 3.5*105 [3.4*105;3.6*105] | 3.7*105 [3.6*105;3.8*105] | < 0.01 |
hepatic insulin extraction (%) | 68.5 [67.7; 69.4] | 67.7 [66.8; 68.7] | 0.14 |
insulin sensitivity/resistance |
HOMA IR | 1.24 [1.17; 1.32] | 1.37 [1.29; 1.46] | 0.02 |
OGIS 120 min (ml/min/m2) | 462.5 [455.8; 468.6] | 453.0 [448.0; 459.3] | < 0.01 |
ISICOMP ([(mg/dl)2(µU/ml)2]−1/2) | 8.91 [8.56; 9.24] | 7.98 [7.52; 8.38] | < 0.01 |
MCRest (ml/min/kg) | 9.99 [9.81; 10.17] | 9.76 [9.54; 9.93] | < 0.01 |
Si(oral) ((ml/min/kg)/(µU/ml)) | 0.16 [0.15; 0.17] | 0.15 [0.14; 0.16] | < 0.01 |
PREDIM (mg/min/kg) | 7.18 [6.90; 7.38] | 6.83 [6.64; 7.00] | 0.03 |
beta cell function |
HOMA B (mIU/mmol) | 103.7 [95.7; 108.3] | 103.2 [97.8; 107.3] | 0.78 |
IGI (pmol/mmol) | 89.1 [83.1; 95.8] | 88.3 [83.2; 98.3] | 0.59 |
Ins0/Glc0 (pmol/mmol) | 7.40 [7.00; 7.85] | 7.92 [7.56; 8.37] | 0.11 |
AUCInsulin/AUCGlc (pmol/mmol) | 31.3 [29.9; 32.7] | 34.5 [32.7; 35.8] | < 0.01 |
IGI*ISICOMP | 267.9 [257.7; 275.5] | 260.6 [253.3; 269.9] | 0.21 |
lipid spectrum |
cholesterol (mmol/l) | 4.61 [4.55; 4.70] | 4.58 [4.47; 4.65] | 0.54 |
HDL chol. women (mmol/l) | 1.58 [1.56; 1.64] | 1.54 [1.50; 1.59] | 0.03 |
HDL chol. men (mmol/l) | 1.30 [1.23; 1.35] | 1.21 [1.15; 1.31] | 0.28 |
LDL cholesterol (mmol/l) | 2.57 [2.47; 2.64] | 2.56 [2.52; 2.64] | 0.35 |
TAG (mmol/l) | 0.84 [0.79; 0.88] | 0.89 [0.83; 0.93] | 0.09 |
liver enzymes |
ALT (ukat/l) | 0.30 [0.29; 0.31] | 0.30 [0.29; 0.31] | 0.59 |
AST (ukat/l) | 0.36 [0.35; 0.37] | 0.34 [0.34; 0.35] | 0.07 |
GGT (ukat/l) | 0.23 [0.22; 0.25] | 0.23 [0.22; 0.24] | 0.65 |
thyroid hormones |
TSH (mIU/l) | 2.19 [2.11; 2.33] | 2.26 [2.19; 2.38] | 0.99 |
fT3 (pmol/l) | 4.89 [4.80; 4.99] | 4.91 [4.84; 5.00] | 0.71 |
fT4 (pmol/l) | 15.0 [14.8; 15.2] | 15.4 [15.1; 15.6] | 0.13 |
Data are given as median [95% LCL; 95% UCL], P-level according to Mann-Whitney test. BMI—body mass index; WHR—waist-hip ratio; AUCGlycemia—area under the glycemic curve; AUCInsulin—area under the insulinemic curve; AUCC−peptide—area under the C-peptide curve; HOMA IR—Homeostatic Model Assessment of Insulin Resistance; OGIS—Oral Glucose Insulin Sensitivity; ISICOMP—Insulin Sensitivity Composite index; PREDIM—Peripheral Insulin Sensitivity Index; HOMA B—Homeostatic Model Assessment of Beta-cell function; IGI—Insulinogenic index; IGI*ISICOMP—product of the IGI and the ISICOMP index; HDL—high-density lipoprotein; LDL—low-density lipoprotein; TAG—triacylglycerols; ALT—alanine aminotransferase; AST—aspartate aminotransferase; GGT—γ-glutamyl transferase; TSH—thyrotropin; fT3—free triiodothyronine; fT4—free thyroxine.