Introduction:
Pregnancy is characterized by decreasing insulin sensitivity and increasing insulin resistance, a physiological change required to provide glucose for fetus. But when it exceeds the physiological limit or when insulin secretion is impaired due to pancreatic cell dysfunction, gestational impaired glucose tolerance (GIGT) or GDM sets in.
Universal screening for GDM with oral Glucose tolerance test (OGTT) is now the standard of care in most hospitals in Asia. However OGTT without insulin levels does not depict insulin sensitivity or resistance. The gold standard for estimating insulin sensitivity is euglycemic-hyper-insulinemic clamp technique, which is not practical in routine clinical setup. Several workers in the past have tried to compare the results of insulin sensitivity from OGTT or fasting insulin and glucose levels (homeostasis assessment model of insulin resistance [HOMA-IR] and quantitative insulin sensitivity check index [QUICKI]) with that of hyper-insulinemic clamps and indicated that these are probably the close indicators of insulin metabolism in fasting state [1,2]. However, these insulin indices (HOMA and QUICKI) are governed by hepatic insulin sensitivity. Peripheral insulin sensitivity is better measured by OGTT coupled with insulin levels, reflecting both hepatic as well as peripheral insulin sensitivity which correlates well with glucose clamp values [3,4]. The present study was conducted to find differences in the insulin sensitivity and secretion indices among three subsets, of antenatal women with normal glucose tolerance test (NGT), gestational impaired glucose tolerance (GIGT), and GDM. Our objective is to explore if there is any association of these Insulin indices with glucose tolerance test in our women and determine what causes GDM in our women – altered insulin sensitivity, resistance or decreased secretion. This study will also help us to find out which of the indices is most useful to study the altered glucose metabolism in our population.
Material and Methods
This prospective study was conducted in the Department of Obstetrics of a tertiary care hospital. Sixty-nine non-diabetic pregnant women from antenatal clinic were enrolled for the study before 12 weeks of gestation. The exclusion criteria included pre-existing diabetes, abnormal fasting glucose values at the time of first registration. Institutional Ethical Committee clearance was obtained.
Sample Size Calculation
To study fasting plasma insulin levels in patients with gestational diabetes and healthy pregnant women we referred to the published mean and standard deviation (75.1 ± 38.1 Pico mole/L for normal subjects and 101.3 103.3 ± 48.3 Pico mole/L for GDM subjects, which is equivalent to 10.82 ± 5.49 µU/mL & 14.59 ± 6.95 µU/mL respectively) for fasting plasma insulin in second trimester of pregnancy by Ley SH et al [5]. With a desired level of power of 80% and level of significance 0.05, the minimum sample size for GDM patients was calculated using the formula for one group sample size;
| N= |
(Ζα+Ζβ)2 X σ2 |
P(μ1+μ2)2 |
Where in N is number of patients; P is prevalence of gestational diabetes in Indian population (0.162 or 16.2%) [6].
Accordingly, a total of 47 patients (having 8 patients in GDM group) were required.
Subject enrolment:
After an informed consent 3-hour 100-g oral glucose tolerance test was performed following 10- to 12-hour overnight fast (ACOG criteria) as per the standard protocol of our hospital. Detailed history regarding parity, previous diabetes or in ?rst-degree relatives was noted. Body mass index (BMI) was determined by the formula; weight/height2 (kg/m2) at the booking visit. All blood samples drawn during fasting hour, 1, 2 and 3 hours after 100 gms glucose ingestion were sent to laboratory within half hour for estimation of glucose and insulin values. A glucose hexokinase method (cobas(®) c111; Roche Diagnostics ) was used to determine plasma glucose levels. Insulin levels were measured by radioimmunoassay method (Board of Radiation & Isotope Technology, BARC, Mumbai, India) using the PC-RIA-MAS Radioimmunoassay analyzer (STRATEC Biomedical AG, Germany)
Normal values for OGTT were defined as the fasting plasma glucose level <95mg/dl, 1-hour plasma glucose level < 180 mg/dl, 2-hour <155mg/dl, and 3-hour plasma glucose <140mg/dl (Carpenter & Coustan criteria) [7]. Based on the OGTT results 3 study groups were defined -1) NGT (normal OGTT), 2) GIGT (only one abnormal value); and 3) GDM (at least two abnormal values). Various Insulin indices were then computed and compared in these 3 groups.
Formulae for Insulin indices:
HOMA1-IR (Homeostasis Model Assessment) = (fasting blood glucose [mg/dl] X fasting plasma insulin [µU / ml]) / 405
HOMA1-B (Homeostasis Model Assessment – Beta-Cell Function) = (fasting plasma insulin [µU / ml] X 360) / (fasting blood glucose [mg/dl] – 63)
HOMA1-S (Homeostasis Model Assessment – Insulin Sensitivity) = (100/HOMA-IR)
The QUantitative Insulin Sensitivity Check Index (QUICKI) = 1 / (log[fasting insulin µU/mL] + log[fasting glucose mg/dL])
Composite Insulin Index (ISOGTT – also known as Matsuda Index) = 10000/√([fasting glucose mg/dL x fasting insulin µU / ml] x [mean blood glucose mg/dL x mean plasma insulin [during OGTT] (1)
All HOMA1 indices reflect hepatic insulin sensitivity.
These have been replaced recently by HOMA2 indices by Diabetes Trials Unit (Oxford) and the indices can be calculated only by a software which can be downloaded at http://www.dtu.ox.ac.uk [8]
Measurements of insulin secretion and sensitivity index
HOMA1-IR and HOMA2-IR were used to calculate insulin resistance with an assumption that normal non-pregnant women will have the ratio of one. Other indices such as HOMA ß-cell secretion (HOMA1-B) which indicates insulin secretion capacity were also computed. Insulin sensitivity was calculated by Fasting Glucose/ insulin ratio (FGIR),QUICKI , HOMA1- and Composite insulin sensitivity index from OGTT were calculated using the formulae mentioned earlier.
Statistical analysis
Mean ± SD of important variables were calculated in three different groups. Statistical differences between groups were assessed using one-way analysis of variance (ANOVA). Receiver Operator Characteristic (ROC) analysis was carried out to determine the best cut off values for various insulin indices to detect abnormal results of 3 hour OGTT tests. Areas under ROC curves (AUC) along with their 95% confidence intervals were compared to determine diagnostic abilities of these indices to detect significant hyperglycemia.
Results
The demographic characteristics including clinical and biochemical variables of the study subjects are shown in Table 1. Though statistically not significant, the subjects in the normal group were younger than those in the other two groups. Normal patients had low BMI and mean weight gain of 12 kg during pregnancy which was significantly lower compared to altered glucose tolerance groups. However, no measurable differences of statistical importance were seen between three groups with respect to parity index, family history of diabetes and hypertension, progression to preeclampsia, period of gestation at the time of birth and birth weight of the newborns.
Table 1. Comparison of baseline demographic characteristics of the three study groups |
Parameter |
Normal (n=52) |
Gestational Impaired Glucose Tolerance(n=8) |
Gestational Diabetes (n=9) |
P value |
Age* |
26.9(4.23) |
28.57(5.26) |
28.11(3.22) |
0.49 |
BMI* |
21.14(3.75) |
25.79(6.83) |
22.98(2.91) |
0.01 |
Weight gain during pregnancy* |
11.99(3.64) |
18.33(12.04) |
14.96(2.67) |
0.04 |
Gestational age at delivery* |
38.29(1.66) |
37.38(2.07) |
37.67(2.24) |
0.31 |
Birth weight in Kg* |
2.88(0.49) |
2.77(0.77) |
2.66(0.5) |
0.48 |
Primigravida# |
38(73.1) |
5(62.5) |
6(66.7) |
0.79 |
History of Diabetes in family# |
5(9.6) |
2(25) |
1(11.1) |
0.449 |
Family history of hypertension# |
6(11.5) |
2(25) |
0(0) |
0.275 |
Progression to preeclampsia# |
4(7.7) |
0(0) |
1(11.1) |
0.657 |
*Mean(SD), #n(%) |
From figure 1, it is evident that at 4 different hours of 100-g OGTT (0, 1, 2 & 3 hrs), both plasma glucose and insulin levels gradually increased as the severity of glucose intolerance increased (GDM> GIGT>NGT). Mean levels have been indicated by markers and vertical bar shows standard error for each mean value in the same graph. Exact mean and standard deviations of these hourly parameters have been shown in Table 2 with statistical significance. It is interesting to note that plasma glucose values started declining 1-hr onwards, whereas persistent increase in the insulin response was noted till 2-hrs following glucose ingestion, following which insulin levels declined in all the three groups. However even after 3-hrs the insulin response was maintained much above the base line.
 |
Figure 1: Mean glucose and insulin levels in three groups |
Table 2. Comparison of plasma glucose and insulin values of the three study groups during oral GTT (Mean & SD) |
Parameter |
NGT (n=52) |
GIGT(n=8) |
GDM (n=9) |
F value |
P* value |
Fasting |
Glucose (mg/dL)
Insulin (µU/mL) |
80.92 (8.16)
21.14 (11.17) |
95.13 (12.59)
28.03 (13.75) |
95.11 (12.36)
35.78 (12.66) |
14.8
6.6 |
<0.0001
0.002 |
1-hour |
Glucose (mg/dL)
Insulin (µU/mL) |
124 (22.99)
135.13 (76.19) |
163 (30)
181.05 (96.07) |
195.11 (24.35)
204.27 (75.54) |
38.7
3.7 |
<0.0001
0.030 |
2-hour |
Glucose (mg/dL)
Insulin (µU/mL) |
104.69 (20.64)
128.68 (81.05) |
139.38 (31.96)
202.62 (121.18) |
169.78 (30.1)
254.77 (82.77) |
33.8
9.6 |
<0.0001
<0.001 |
3-hour |
Glucose (mg/dL)
Insulin (µU/mL) |
81.35 (19.65)
73.84 (64.38) |
111.63 (25.72)
156.37 (106.49 |
116.11 (35.45)
201.52 (94.76) |
13.2
14.1 |
<0.0001
<0.001 |
*Analysis of variance (ANOVA) |
Table 3 shows comparison of various insulin indices such as sensitivity index, resistance index and beta cell function. It can be seen that NGT group had higher sensitivity index whereas resistance index was more in abnormal group. The differences were statistically significant except that FGIR is not a very good index during pregnancy. The beta cell activity was increased in all the groups, but interestingly the differences in three groups did not reach statistical significance.
Table 3. Comparison of various insulin indices across three groups |
Insulin Parameters |
Index |
NGT (n=52) |
GIGT(n=8) |
GDM (n=9) |
F value |
P* value |
A. Sensitivity Indices |
Matsuda Index |
3.32 (1.65) |
1.9 (1.14) |
1.19 (0.34) |
9.5 |
<0.001 |
HOMA1-S |
31.81 (19.46) |
19.08 (11.6) |
13.39 (4.65) |
5.4 |
0.007 |
QUICKI |
0.32 (0.03) |
0.3 (0.02) |
0.29 (0.01) |
7.7 |
0.001 |
HOMA2 -S |
50.15 (28.17) |
33.89 (16.09) |
24.96 (7.99) |
4.6 |
0.013 |
B. Resistance Indices |
HOMA1-R |
4.32 (2.58) |
6.62 (3.16) |
8.34 (3.04) |
9.9 |
<0.001 |
HOMA2 -IR |
2.59 (1.34) |
3.51 (1.55) |
4.37 (1.31) |
7.5 |
<0.001 |
FGIR |
4.91 (2.67) |
3.98 (1.5) |
2.99 (1.16) |
2.6 |
0.080 |
C. Beta Cell function |
HOMA1-B |
507.5 (364.33) |
347.5 (194.26) |
557.37 (529.89) |
0.77 |
0.466 |
HOMA2-B |
226.78 (72.05) |
207.29 (76.78) |
258.11 (110.15) |
0.95 |
0.391 |
*Analysis of variance (ANOVA) |
The next picture and table (*Figure 2A-2D, Table 4) show the results of diagnostic analysis using ROC curves and statistics. Among insulin sensitive indices Matsuda Index was superior to HOMA1-S, QUICKI, HOMA2 -S, whereas when insulin resistance was considered, HOMA1-IR had better diagnostic ability compared to HOMA2 -IR and FGIR. However, again beta cell functions (HOMA1-B & HOMA2-B) did not show any significant results. Figure 2D shows among all the indices analysed Matsuda Index had the most pickup rate.
 |
Table 4. Diagnostic parameters of various insulin indices by ROC analysis |
Insulin Indices |
Index |
Criterion |
Sensitivity (95% CI) |
Specificity (95% CI) |
+LR (95% CI) |
-LR (95% CI) |
AUC (95% CI) |
Z statistics |
Significance |
Sensitivity Indices |
Matsuda Index |
≤2.2 |
88.24 (63.5 - 98.2) |
80.77 (67.5 - 90.4) |
4.59 (3.7 - 5.7) |
0.15 (0.04 - 0.6) |
0.875 (0.773 to 0.942) |
8.97 |
0.0001 |
HOMA1-S |
≤21.7 |
94.12 (71.2 - 99.0) |
71.15 (56.9 - 82.9) |
3.26 (2.6 - 4.0) |
0.083 (0.01 - 0.6) |
0.818 (0.706 to 0.900) |
6.11 |
0.0001 |
QUICKI |
≤0.3 |
82.35 (56.6 - 96.0) |
71.15 (56.9 - 82.9) |
2.85 (2.2 - 3.8) |
0.25 (0.08 - 0.8) |
0.8 (0.687 to 0.887) |
5.48 |
0.0001 |
HOMA2 -S |
≤39.1 |
94.12 (71.2 - 99.0) |
67.31 (52.9 - 79.7) |
2.88 (2.3 - 3.6) |
0.087 (0.01 - 0.6) |
0.798 (0.683 to 0.885) |
5.39 |
0.0001 |
Resistance Indices |
HOMA1-IR |
>4.35 |
94.12 (71.2 - 99.0) |
71.15 (56.9 - 82.9) |
3.26 (2.6 - 4.0) |
0.083 (0.01 - 0.6) |
0.818 (0.706 to 0.900) |
6.11 |
0.0001 |
HOMA2 -IR |
>2.5 |
94.12 (71.2 - 99.0) |
67.31 (52.9 - 79.7) |
2.88 (2.3 - 3.6) |
0.087 (0.01 - 0.6) |
0.798 (0.683 to 0.885) |
4.29 |
0.0001 |
FGIR |
≤4.78 |
88.24 (63.5 - 98.2) |
40.38 (27.0 - 54.9) |
1.48 (1.0 - 2.1) |
0.29 (0.08 - 1.1) |
0.674 (0.550 to 0.782) |
2.46 |
0.0138 |
Beta Cell function |
HOMA1-B |
≤263.2 |
47.06 (23.0 - 72.1) |
84.62 (71.9 - 93.1) |
3.06 (1.8 - 5.1) |
0.63 (0.3 - 1.4) |
0.614 (0.489 to 0.729) |
1.51 |
0.1325 |
HOMA2-B |
≤183.2 |
52.94 (27.9 - 77.0) |
73.08 (59.0 - 84.4) |
1.97 (1.2 - 3.2) |
0.64 (0.3 - 1.3) |
0.533 (0.409 to 0.654) |
0.41 |
0.6822 |
Discussion
Insulin resistance in pregnancy is physiological to some extent but can give rise to GDM when altered. Though GIGT is an intermediate stage between NGT and GDM, women affected already show biochemical evidences of lowered insulin sensitivity and rising insulin resistance. Here among the various insulin indices used, Matsuda index showed better correlation with altered glucose metabolism. This has been observed by another group from New Delhi where Matsuda index in the first trimester predicted abnormal OGTT in the second trimester [9]. This composite insulin index is dynamic being composed of both fasting and postprandial levels thereby reflecting both hepatic and peripheral insulin sensitivity. This is especially useful in pregnancy where peripheral tissue insulin resistance is more marked [3,4].
The euglycemic-hyperglycemic clamp considered as the gold standard test for insulin resistance is difficult to use in a clinical setting. A modified HOMA model (HOMA2), now in use helps to study hepatic and peripheral glucose resistance and is recalibrated using recent assays for insulin and C peptide to obtain cent percent %B and %S values [8,10]. This is also useful for calculation of beta cell function, and is independent of linear correlation.
Yang SJ et al (2013) showed decreasing beta cell function in Korean pregnant women who had GIGT and GDM, those with GDM exhibited lesser beta cell function [11].
A study from Bulgaria also indicated lowered beta cell function in their pregnant women with GDM group [12]. However we could not come to any statistically significant conclusion regarding beta cell function in altered glucose metabolism.
Ismail NA et al (2013) from Malaysia correlated HOMA-IR score with maternal and fetal outcome in GDM women. They noted a HOMA-IR cut off of 2.92 predicted risks of caesarean delivery and neonatal hypoglycemia [13]. We did not correlate insulin indices with perinatal outcome.
The cut off values for various insulin indices vary from country to country. Mřrkrid K et al (2012) conducted population-based cohort study of 695 healthy pregnant women from Western Europe (41%), South Asia (25%), Middle East (15%), East Asia (6%) and elsewhere (13%) [14]. Blood samples for glucose & insulin were taken in early second trimester (around 15 weeks), early third trimester (around 29 weeks) and 3 months postpartum. They concluded that pregnant women from East and South Asia were more insulin resistant and showed poorer HOMA-ß-cell function than Western Europeans.
It is well-known that racial factors play an important role in pathogenesis of type 2 diabetes, Asians being the more vulnerable group compared to Western countries. Even though belonging to low BMI category, many of the Asians develop non-insulin dependent diabetes at a younger age. Impaired capacity of pancreas to secrete insulin hormone appears to be a causative factor probably due to a smaller pancreas with lesser ß-cell population [14, 15]. However we noted altered insulin sensitivity and Insulin resistance as the problem in our subset rather than insulin secretion as the cause for impaired glucose metabolism. In a comparative study of insulin sensitivity measures in different populations, Subbulaxmy et al noted that south Asians had a greater insulin resistance and lower insulin sensitivity compared to Caucasians [4]. Though ethnic differences exist in insulin levels it may be partly due to differences in the insulin assays used [3].
Incidence of GDM varies and Insulin indices are population specific and therefore every country should have customized reference ranges.
Wei et al reported a high incidence of GIGT & GDM in their pregnant population in China [16]. In addition to high insulin resistance and reduced insulin functionality index their patients demonstrated abnormal lipid profile. The prevalence of GDM was 13% in our study population and GIGT being 11%. Other workers from South India have noted prevalence varying from 9%-17% from rural to urban areas [6].
The main limitation of this study was the small number of patients. Though sample size calculation was adequate, larger number of GDM and GIGT women would have been better. Another limitation is the cost of estimation of insulin which prevents its use for routine clinical use. We just studied the association of Insulin indices with abnormal OGTT but using it for prediction would be more clinically useful. Impairment of glucose homeostasis starts much earlier than fully established GDM. This can only be identified when insulin levels are studied along with GTT. High HOMA IR, low sensitivity indicated by Matsuda index and low HOMA B indicating impaired insulin secretion are associated with impaired glucose tolerance. These tests may predict GDM earlier in pregnancy even when GTT is normal. Estimating Insulin indices in the first trimester as a predictor of GDM in the second trimester is good point as illustrated in the study of Grewal and Ozgu-erdinc [9, 17].
Further insulin indices along with GTT will not only predict GIGT or GDM but also help in planning therapeutic interventions. When insulin resistance is the primary problem, insulin sensitizers like metformin and when B cell dysfunction is the cause, Insulin may be used for treatment.
To conclude low insulin sensitivity and high insulin resistance was seen in South Indian women with altered glucose tolerance test. Among resistance indices HOMA-1 IR seemed slightly better than HOMA-2 IR. The Matsuda index seemed better than other insulin indices in estimating insulin sensitivity. It is primarily because it takes into consideration of all glucose values of GTT along with insulin levels.
Potential benefits of present study in clinical practice (for prediction and management)
The practical application of this paper is to do insulin indices along with GTT to determine the cause of GDM, so that it can be treated appropriately. As we have mentioned earlier in our South Indian women increased Insulin resistance is the cause & not insulin secretion. So treatment like metformin can be directed to the cause.
Diagnosis of GDM is made with OGTT but without concurrent insulin levels sensitivity and resistance can’t be known. With evolving automated machines, these indices can be determined easily unlike the method we followed. It is also essential to know the reference values for these surrogate indices of insulin sensitivity/ resistance in a given population for future application. Hence we have derived the same for our antenatal women.
If we can device a test which predicts occurrence of the disease before it is clinically evident, we may be able to reduce the associated perinatal and maternal morbidity. Now that we know the abnormal values, we also intend to predict GDM by doing this in first trimester. So starting with diet restriction early enough may decrease the severity.
References
- Gutch M, Kumar S, Razi SM, Gupta KK, Gupta A. Assessment of insulin sensitivity/resistance. Indian J Endocrinol Metab. 2015;19(1):160-4.
- Tam CS, Xie W, Johnson WD, Cefalu WT, Redman LM, Ravussin E. Defining insulin resistance from hyperinsulinemic-euglycemic clamps. Diabetes Care. 2012;35(7):1605-10.
- Anwar Borai, Livingstone C, Kaddam I, Ferns G Selection of appropriate method for the assessment of Insulin resistance. BMC Medical Research Methodology 2011;11:158 doi:10.1186/1471-2288-11-158
- Subbulaxmi T, Annaswamy R, Bindu C, Ellen WS, Donald CS Comparison of Insulin sensitivity measures in south Asians. Metabolism 2013;62(10):1448-1454 doi: 10.1016/j.metabol 2013.05.016
- Ley SH, Hanley AJ, Retnakaran R, Sermer M, Zinman B, O'connor DL. Effect of macronutrient intake during the second trimester on glucose metabolism later in pregnancy. Am J Clin Nutr. 2011;94(5):1232-40.
- Mithal A, Bansal B, Kalra S. Gestational diabetes in India: Science and society. Indian J Endocrinol Metab. 2015;19(6):701-4.
- Berggren EK, Boggess KA, Stuebe AM, Jonsson funk M. National Diabetes Data Group vs Carpenter-Coustan criteria to diagnose gestational diabetes. Am J Obstet Gynecol. 2011;205(3):253.e1-7.
- Oxford Diabetes Trial unit, HOMA Calculator (2013), retrieved from http://WWW.dtu.ox.ac.uk
- Grewal E, Kansara S, Kachhava G, Ammini AC, Kripalani A, Agarwal N et al. Prediction of gestational diabetes mellitus at 24 to 28 weeks of gestation by using first-trimester insulin sensitivity indices in Asian Indian subjects. Metabolism 2012;61(5):715-720.
- Song YS, Hwang YC, Ahn HY, Park CY. Comparison of the Usefulness of the Updated Homeostasis Model Assessment (HOMA2) with the Original HOMA1 in the Prediction of Type 2 Diabetes Mellitus in Koreans. Diabetes Metab J. 2016;40(4):318-25.
- Yang SJ, Kim TN, Baik SH, et al. Insulin secretion and insulin resistance in Korean women with gestational diabetes mellitus and impaired glucose tolerance. Korean J Intern Med. 2013;28(3):306-13.
- Genova MP, Todorova-Ananieva K, Atanasova B, Tzatchev K. Assessment of Beta-Cell Function During Pregnancy and after Delivery. Acta Medica Bulgarica. 2014;41(1):5-12 DOI 10.2478/amb-2014-0001
- Ismail NA, Kasim MM, Noor Aizuddin A, Umar NA. Homeostatic indices of insulin resistance among gestational diabetics in anticipating pregnancy complications. Gynecol Endocrinol. 2013;29(7):691-4.
- Mřrkrid K, Jenum AK, Sletner L, et al. Failure to increase insulin secretory capacity during pregnancy-induced insulin resistance is associated with ethnicity and gestational diabetes. Eur J Endocrinol. 2012;167(4):579-88.
- Ma RC, Chan JC. Type 2 diabetes in East Asians: similarities and differences with populations in Europe and the United States. Ann N Y Acad Sci. 2013;1281:64-91.
- Wei J, Gao J, Cheng J Gestational diabetes mellitus and impaired glucose tolerance pregnant women. Pak J Med Sci 2014;30(6):1203-1208
- Ozgu-erdinc AS, Yilmaz S, Yeral MI, Seckin KD, Erkaya S, Danisman AN. Prediction of gestational diabetes mellitus in the first trimester: comparison of C-reactive protein, fasting plasma glucose, insulin and insulin sensitivity indices. J Matern Fetal Neonatal Med. 2015;28(16):1957-62.
|
