Page 1 of 19
Accepted Preprint first posted on 15 April 2015 as Manuscript EJE-15-0153
1
β-cell dysfunction in women with previous gestational diabetes is associated with visceral adipose
2
tissue distribution
3 4
Tove Lekva, PhD 1, Jens Bollerslev, MD, PhD ,
2,3
, Kristin Godang, BSc 2, Marie Cecilie Paasche
5
Roland, MD, PhD 4, Camilla Margrethe Friis, MD, PhD 4, Nanna Voldner, PhD 4, Tore Henriksen,
6
MD, PhD 3,4 Thor Ueland, PhD 1,3
7 8
1 Research Institute for Internal Medicine, Oslo University Hospital, 2 Section of Specialized
9
Endocrinology, Department of Endocrinology, Oslo University Hospital 3 Faculty of Medicine,
10
University of Oslo, 4 Department of Obstetrics, Oslo University Hospital Rikshospitalet, Oslo,
11
Norway.
12 13
Tables 2, Figures 1, word count: 3486
14 15
Abbreviated title: Visceral fat linked with β-cell dysfunction
16 17
Keywords: GDM, follow-up study, visceral fat, β-cell function, prediabetes
18 19
Corresponding author:
20
Tove Lekva, PhD
21
Research Institute for Internal Medicine
22
Oslo University Hospital, Rikshospitalet
23
University of Oslo
24
P.b 4950 Nydalen, 0424 Oslo, Norway
25
Phone: 47-23072782; E-mail: [email protected]
26 27 28
Copyright © 2015 European Society of Endocrinology.
Page 2 of 19
29
Abstract
30
Context: Glucose intolerance in pregnancy predicts an increased risk of future type 2 diabetes.
31
Objective: The aim of the study was to evaluate glucose metabolism in women with and without
32
gestational diabetes mellitus (GDM) at 5 years follow-up and identify risk factors associated with
33
disturbed glucose metabolism post partum.
34
Design: This follow-up study included 300 consecutively enrolled women from a previous population-
35
based cohort study. The participants underwent oral glucose tolerance test under pregnancy and at the
36
follow-up study, in addition to dual-energy x-ray absorptiometry at the follow-up study.
37
Results: Fifty-two women (17.7 %) were found to have GDM in pregnancy with an odds ratio of 4.8
38
developing prediabetes 5 years later. β-cell function, but not insulin resistance or sensitivity, was
39
reduced in the follow-up study after adjusting for known risk factors. Furthermore, visceral fat content
40
at follow-up was increased in GDM women compared to non-GDM women, and the β-cell function
41
declined with increasing visceral fat in both groups but was more pronounced in the women with
42
previous GDM.
43
Conclusions: Women with GDM are at increased risk of developing pre-diabetes and have a decreased
44
β-cell function 5 years post-partum that is associated with increased visceral fat mass.
45 46 47 48 49 50 51 52 53 54 55 56
Page 3 of 19
57
Abbreviations
58
AUC- Area under the curve
59
BP – Blood pressure
60
CV – Cardiovascular
61
DXA – Dual energy x-ray absorptiometry
62
FPG- Fasting plasma glucose
63
GDM- Gestational diabetes mellitus
64
IADPSG- International Association of Diabetes and Pregnancy Study Groups
65
ISSI-2 - Insulin secretion-sensitivity index
66
OGTT- Oral glucose tolerance test
67
ROI- Region of interest
68
VAT- Visceral adipose tissue
69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84
Page 4 of 19
85
Introduction
86
Women with a history of previous gestational diabetes mellitus (GDM) are at increased risk of future
87
impaired glucose tolerance with some studies reporting a seven-fold increased risk of developing type
88
2 diabetes [1]. However, also women with milder glucose intolerance, but without the diagnosis of
89
GDM, have an increased risk of developing prediabetes and type 2 diabetes in the future [2;3].
90
Prediabetes is an intermediate form of dysglycemia and identifies individuals at risk of diabetes,
91
cardiovascular (CV) disease and mortality [4-6]. The prevalence of prediabetes increased from 27.4 %
92
in 1999-2002 to 34.1 % in 2007-2010 [7] implying that this is an important group for risk
93
modification.
94
During pregnancy there is a uniform 50-60 % decrease in insulin sensitivity with progression
95
of gestation [8]. GDM is a disease of the pancreatic β-cells, which produce inadequate amounts of
96
insulin to meet the increased insulin needs of late pregnancy [9]. Many women with GDM seem to
97
have a β-cell defect that is chronic rather than acquired during pregnancy [9]. Longitudinal studies of
98
glucose regulation after GDM reveal falling β-cell compensation for chronic insulin resistance, which
99
might also worsen over time [10]. Risk factors for attenuated β-cell function that cause type 2 diabetes
100
include weight gain, insulin resistance, rising levels of C-reactive protein and falling levels of
101
adiponectin [11]. Weight loss through dietary intervention and/or physical exercise may improve
102
adipose tissue distribution and provide protection against development of type 2 diabetes following
103
GDM [9].
104
As follow-up of women with a history of GDM and pre-diabetes states are crucial for targeted
105
intervention to prevent development of overt type 2 diabetes, it is important to determine early
106
predictors of these complications in normal pregnancies. Several studies demonstrate increased BMI
107
during long-term follow-up in women with previous GDM. However, there are no studies
108
investigating specific adipose tissue distribution and associations with indices of glucose metabolism.
109
Recently, a new dual-energy X-ray absorptiometry (DXA) application for quantifying visceral adipose
110
tissue (VAT) in the android region of the body was developed [12]. This allowed us to measure total
111
body composition and VAT in a population of women 5 years after the index pregnancy in a simple
112
and cost effective manner.
Page 5 of 19
113
The STORK (STORe barn og Komplikasjoner; translated “Large Babies and
114
Complications”) study is a prospective cohort study performed in the period 2002–2008 following
115
1031 Norwegian healthy pregnant women. The follow up study was performed 5 years after the index
116
pregnancy in 300 women. The aims of the study was to 1) investigate the association of GDM (with
117
the new IADPSG criteria) in pregnancy with indices of glucose metabolism and rates of prediabetes
118
after 5 years follow-up 2) evaluate visceral fat distribution, and association with indices of glucose
119
metabolism, in GDM vs. non-GDM at follow-up.
120 121
Subjects and Methods
122
Study design and subjects
123
The STORK study is a prospective cohort study with a longitudinal design including 1031 healthy
124
women of Scandinavian heritage who gave birth at Oslo University Hospital, Rikshospitalet between
125
2002 and 2008. Exclusion criteria were multiple pregnancies, known pre-gestational diabetes and
126
severe chronic diseases (lung, cardiac, gastrointestinal or renal), and pregnancies with fetal
127
malformations discovered at the routine scan in week 18 of pregnancy. Details about the study have
128
been published [13]. The follow-up study was performed 5 years after the index pregnancy, and 300
129
women participated (Figure 1). All 1031 participants from the STORK study got an invitation letter.
130
Exclusion criteria were pregnancy and last birth had to be at least one year ago. Written informed
131
consent was obtained from all participants in the study. All clinical investigations were conducted
132
according to the principles expressed in the Declaration of Helsinki. The study was approved by the
133
Regional Committee for Medical Research Ethics, Southern Norway, Oslo, Norway.
134 135
Each pregnant woman had four antenatal visits, gestational age (GA) 14–16, 22-24, 30-32, and 36-38
136
weeks. Clinical data and blood samples were collected at each visit and stored at -80 °C. Parity was
137
divided into primipara for nulliparous women and multipara for those with one or more previous
138
births. Birth weight and placental weight was measured by the midwife within one hour after the
139
delivery. BMI (kg/m2) was calculated by height and weight. Weight was measured by a calibrated
140
scale at each visit. Height were self-reported at antenatal GA visit 14-16 weeks and measured at the
Page 6 of 19
141
follow up visit. Gestational weight gain was calculated as the difference between weights measured at
142
antenatal GA visit 36-38 and 14-16 weeks. Total body composition was determined by dual-energy X-
143
ray absorptiometry (DXA; GE Lunar Prodigy Densitometer (software version 12.10; GE Medical
144
Systems, Lunar Corp., Madison, WI, USA) for the 300 participants at follow up visit. No hardware
145
changes were made during the study period, but two upgrading software version (12.10 and 12.20)
146
were used. In the end of the study all scans were imported into an updated version of the software
147
(enCORE 14.10, GE Medical Systems) and reanalyzed. Standard imaging and positioning protocols
148
were used to scan the subjects. All scans were performed by the same densitometer technologist.
149
CoreScan computes VAT within the android region, located over the abdomen, of a total body DXA
150
scan. CoreScan has been validated against volumetric computed tomography [12;14]. For measuring
151
android fat, a region of interest (ROI) was defined with the caudal limit at the top of the iliac crest and
152
the cephalic limit at the base of the skull. The height of the android ROI was automatically set to 20%
153
of the distance from the iliac crest to the base of the skull. Android ROI contains both visceral (VAT)
154
and subcutaneous adipose tissue (SAT). The CoreScan algorithm works through detection of two key
155
parameters: the width of the SAT layer on the lateral aspects of the abdomen, and the anterior-
156
posterior thickness of the abdomen, which can be attained using the DXA tissue attenuation image.
157
The software estimates the quantity of SAT in the android ROI. VAT was computed by subtracting
158
android SAT from the total android fat. The fat mass data from DXA was transformed to volume using
159
a constant correction factor (0.94g/cm3) consistent with the density of adipose tissue [12]. All VAT
160
under 50g was set to 50g since the DXA measurement is unreliable in the low range visceral fat
161
content (precision error (with the iDXA) is approximately 50g) [15]. The short- and long-term
162
coefficients of variation for our densitometer are 0.8% and 1.4%, respectively.
163 164 165
Laboratory measurements
166
All oral glucose tolerance tests (OGTTs) were performed in the morning after an overnight fast at
167
antenatal visit GA 30-32 weeks and follow-up visit. Venous EDTA blood was analyzed on site by the
168
Accu-Check Sensor glucometer (Roche Diagnostics GmbH, Mannheim, Germany). Venous blood was
Page 7 of 19
169
also drawn in gel tubes, allowed to clot for 30 min, thereafter centrifuged for 10 min 3000 g, serum
170
separated and stored at -80. Glucose was measured from frozen serum samples at antenatal visit GA
171
30-32 weeks, with the hexokinase method at an accredited clinical chemistry laboratory at Oslo
172
University Hospital, Rikshospitalet (Hitachi Modular P800 with reagents from Roche) and results are
173
used further for this study. The median difference in glucose values for the 300 women at antenatal
174
visit GA 30-32 weeks between on-site measurements [16] and the measurements performed at the
175
clinical chemistry laboratory was 0.09, 0.17, 0.24, 0.19, and 0.21 mmol/l for the timepoints
176
0,30,60,90,120 min in the OGTT test, respectively, and highly correlated R2=0.70, 0.84, 0.90, 0.91,
177
0.90, consistent with what others have found [17]. For the follow up visit we used the glucose data
178
from the Accu-check Sensor glucometer. Insulin samples were assayed in duplicate (RIA, DPC, Los
179
Angeles, CA) as previously reported [16] and the same method were used for the follow-up samples.
180 181
Diagnosis of GDM and prediabetes
182
GDM was diagnosed with the IADPSG criteria; fasting plasma glucose (FPG) 5.1-6.9 mmol/l, 1-h
183
plasma glucose ≥10.0 mmol/l or 2-h plasma glucose 8.5-11.0 mmol/l following a 75 g oral glucose
184
load [18]. Prediabetes was diagnosed at follow-up visit with the criteria of FPG 5.6-6.9 mmol/l or 2-h
185
plasma glucose 7.8-11.0 mmol/l after 75 g OGTT [19]. GDM (IADPSG) and prediabetes were
186
diagnosed after the study was finished. The women who were diagnosed with GDM during pregnancy
187
(WHO) did not get any medication but did receive nutrition guidance.
188 189
Insulin sensitivity was measured with the Matsuda index 10,000/square root of [fasting glucose
190
(mmol/l) x fasting insulin (mU/l] x [mean glucose (mmol/l) x mean insulin (mU/l)] during OGTT.
191
This index is a measure of whole body insulin sensitivity that has been validated against the
192
euglycemic-hyperinsulinemic clamp [20]. β-cell function was assessed with the insulin secretion-
193
sensitivity index (ISSI-2) (area under the curve (AUC) insulin (mU/l) 0-120 /glucose(mmol/l) 0-120 x
194
Matsuda), validated against the disposition index from the intravenous glucose tolerance test [21] and
195
HOMA–IR was calculated as fasting insulin (mU/l) × fasting glucose (mmol/l)/22.5, as described by
196
Matthews et al [22].
Page 8 of 19
197 198
Statistics
199
Statistical analyses were conducted using SPSS for Windows, version 21.0 (Chicago, IL, USA). Data
200
are expressed as mean±SD when normally distributed and median (25th, 75th percentile) when skewed.
201
Comparison between women with GDM and non-GDM were performed using t-test or Mann-Whitney
202
U depending on distribution, and Chi-square test for categorical variables. Univariate and stepwise
203
(probability of F to-enter 0.1 -remove 0.15) linear regression analyses was carried out on log
204
transformed variables (if skewed) and results are given as standardized regression coefficients.
205
Interactions between VAT and indices of glucose metabolism were evaluated by univariate general
206
linear model with glucose indices as dependent and VAT, glucose indices, GDM and the interaction
207
term (VAT*GDM) as independent variables.
Accepted Preprint first posted on 15 April 2015 as Manuscript EJE-15-0153
1
β-cell dysfunction in women with previous gestational diabetes is associated with visceral adipose
2
tissue distribution
3 4
Tove Lekva, PhD 1, Jens Bollerslev, MD, PhD ,
2,3
, Kristin Godang, BSc 2, Marie Cecilie Paasche
5
Roland, MD, PhD 4, Camilla Margrethe Friis, MD, PhD 4, Nanna Voldner, PhD 4, Tore Henriksen,
6
MD, PhD 3,4 Thor Ueland, PhD 1,3
7 8
1 Research Institute for Internal Medicine, Oslo University Hospital, 2 Section of Specialized
9
Endocrinology, Department of Endocrinology, Oslo University Hospital 3 Faculty of Medicine,
10
University of Oslo, 4 Department of Obstetrics, Oslo University Hospital Rikshospitalet, Oslo,
11
Norway.
12 13
Tables 2, Figures 1, word count: 3486
14 15
Abbreviated title: Visceral fat linked with β-cell dysfunction
16 17
Keywords: GDM, follow-up study, visceral fat, β-cell function, prediabetes
18 19
Corresponding author:
20
Tove Lekva, PhD
21
Research Institute for Internal Medicine
22
Oslo University Hospital, Rikshospitalet
23
University of Oslo
24
P.b 4950 Nydalen, 0424 Oslo, Norway
25
Phone: 47-23072782; E-mail: [email protected]
26 27 28
Copyright © 2015 European Society of Endocrinology.
Page 2 of 19
29
Abstract
30
Context: Glucose intolerance in pregnancy predicts an increased risk of future type 2 diabetes.
31
Objective: The aim of the study was to evaluate glucose metabolism in women with and without
32
gestational diabetes mellitus (GDM) at 5 years follow-up and identify risk factors associated with
33
disturbed glucose metabolism post partum.
34
Design: This follow-up study included 300 consecutively enrolled women from a previous population-
35
based cohort study. The participants underwent oral glucose tolerance test under pregnancy and at the
36
follow-up study, in addition to dual-energy x-ray absorptiometry at the follow-up study.
37
Results: Fifty-two women (17.7 %) were found to have GDM in pregnancy with an odds ratio of 4.8
38
developing prediabetes 5 years later. β-cell function, but not insulin resistance or sensitivity, was
39
reduced in the follow-up study after adjusting for known risk factors. Furthermore, visceral fat content
40
at follow-up was increased in GDM women compared to non-GDM women, and the β-cell function
41
declined with increasing visceral fat in both groups but was more pronounced in the women with
42
previous GDM.
43
Conclusions: Women with GDM are at increased risk of developing pre-diabetes and have a decreased
44
β-cell function 5 years post-partum that is associated with increased visceral fat mass.
45 46 47 48 49 50 51 52 53 54 55 56
Page 3 of 19
57
Abbreviations
58
AUC- Area under the curve
59
BP – Blood pressure
60
CV – Cardiovascular
61
DXA – Dual energy x-ray absorptiometry
62
FPG- Fasting plasma glucose
63
GDM- Gestational diabetes mellitus
64
IADPSG- International Association of Diabetes and Pregnancy Study Groups
65
ISSI-2 - Insulin secretion-sensitivity index
66
OGTT- Oral glucose tolerance test
67
ROI- Region of interest
68
VAT- Visceral adipose tissue
69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84
Page 4 of 19
85
Introduction
86
Women with a history of previous gestational diabetes mellitus (GDM) are at increased risk of future
87
impaired glucose tolerance with some studies reporting a seven-fold increased risk of developing type
88
2 diabetes [1]. However, also women with milder glucose intolerance, but without the diagnosis of
89
GDM, have an increased risk of developing prediabetes and type 2 diabetes in the future [2;3].
90
Prediabetes is an intermediate form of dysglycemia and identifies individuals at risk of diabetes,
91
cardiovascular (CV) disease and mortality [4-6]. The prevalence of prediabetes increased from 27.4 %
92
in 1999-2002 to 34.1 % in 2007-2010 [7] implying that this is an important group for risk
93
modification.
94
During pregnancy there is a uniform 50-60 % decrease in insulin sensitivity with progression
95
of gestation [8]. GDM is a disease of the pancreatic β-cells, which produce inadequate amounts of
96
insulin to meet the increased insulin needs of late pregnancy [9]. Many women with GDM seem to
97
have a β-cell defect that is chronic rather than acquired during pregnancy [9]. Longitudinal studies of
98
glucose regulation after GDM reveal falling β-cell compensation for chronic insulin resistance, which
99
might also worsen over time [10]. Risk factors for attenuated β-cell function that cause type 2 diabetes
100
include weight gain, insulin resistance, rising levels of C-reactive protein and falling levels of
101
adiponectin [11]. Weight loss through dietary intervention and/or physical exercise may improve
102
adipose tissue distribution and provide protection against development of type 2 diabetes following
103
GDM [9].
104
As follow-up of women with a history of GDM and pre-diabetes states are crucial for targeted
105
intervention to prevent development of overt type 2 diabetes, it is important to determine early
106
predictors of these complications in normal pregnancies. Several studies demonstrate increased BMI
107
during long-term follow-up in women with previous GDM. However, there are no studies
108
investigating specific adipose tissue distribution and associations with indices of glucose metabolism.
109
Recently, a new dual-energy X-ray absorptiometry (DXA) application for quantifying visceral adipose
110
tissue (VAT) in the android region of the body was developed [12]. This allowed us to measure total
111
body composition and VAT in a population of women 5 years after the index pregnancy in a simple
112
and cost effective manner.
Page 5 of 19
113
The STORK (STORe barn og Komplikasjoner; translated “Large Babies and
114
Complications”) study is a prospective cohort study performed in the period 2002–2008 following
115
1031 Norwegian healthy pregnant women. The follow up study was performed 5 years after the index
116
pregnancy in 300 women. The aims of the study was to 1) investigate the association of GDM (with
117
the new IADPSG criteria) in pregnancy with indices of glucose metabolism and rates of prediabetes
118
after 5 years follow-up 2) evaluate visceral fat distribution, and association with indices of glucose
119
metabolism, in GDM vs. non-GDM at follow-up.
120 121
Subjects and Methods
122
Study design and subjects
123
The STORK study is a prospective cohort study with a longitudinal design including 1031 healthy
124
women of Scandinavian heritage who gave birth at Oslo University Hospital, Rikshospitalet between
125
2002 and 2008. Exclusion criteria were multiple pregnancies, known pre-gestational diabetes and
126
severe chronic diseases (lung, cardiac, gastrointestinal or renal), and pregnancies with fetal
127
malformations discovered at the routine scan in week 18 of pregnancy. Details about the study have
128
been published [13]. The follow-up study was performed 5 years after the index pregnancy, and 300
129
women participated (Figure 1). All 1031 participants from the STORK study got an invitation letter.
130
Exclusion criteria were pregnancy and last birth had to be at least one year ago. Written informed
131
consent was obtained from all participants in the study. All clinical investigations were conducted
132
according to the principles expressed in the Declaration of Helsinki. The study was approved by the
133
Regional Committee for Medical Research Ethics, Southern Norway, Oslo, Norway.
134 135
Each pregnant woman had four antenatal visits, gestational age (GA) 14–16, 22-24, 30-32, and 36-38
136
weeks. Clinical data and blood samples were collected at each visit and stored at -80 °C. Parity was
137
divided into primipara for nulliparous women and multipara for those with one or more previous
138
births. Birth weight and placental weight was measured by the midwife within one hour after the
139
delivery. BMI (kg/m2) was calculated by height and weight. Weight was measured by a calibrated
140
scale at each visit. Height were self-reported at antenatal GA visit 14-16 weeks and measured at the
Page 6 of 19
141
follow up visit. Gestational weight gain was calculated as the difference between weights measured at
142
antenatal GA visit 36-38 and 14-16 weeks. Total body composition was determined by dual-energy X-
143
ray absorptiometry (DXA; GE Lunar Prodigy Densitometer (software version 12.10; GE Medical
144
Systems, Lunar Corp., Madison, WI, USA) for the 300 participants at follow up visit. No hardware
145
changes were made during the study period, but two upgrading software version (12.10 and 12.20)
146
were used. In the end of the study all scans were imported into an updated version of the software
147
(enCORE 14.10, GE Medical Systems) and reanalyzed. Standard imaging and positioning protocols
148
were used to scan the subjects. All scans were performed by the same densitometer technologist.
149
CoreScan computes VAT within the android region, located over the abdomen, of a total body DXA
150
scan. CoreScan has been validated against volumetric computed tomography [12;14]. For measuring
151
android fat, a region of interest (ROI) was defined with the caudal limit at the top of the iliac crest and
152
the cephalic limit at the base of the skull. The height of the android ROI was automatically set to 20%
153
of the distance from the iliac crest to the base of the skull. Android ROI contains both visceral (VAT)
154
and subcutaneous adipose tissue (SAT). The CoreScan algorithm works through detection of two key
155
parameters: the width of the SAT layer on the lateral aspects of the abdomen, and the anterior-
156
posterior thickness of the abdomen, which can be attained using the DXA tissue attenuation image.
157
The software estimates the quantity of SAT in the android ROI. VAT was computed by subtracting
158
android SAT from the total android fat. The fat mass data from DXA was transformed to volume using
159
a constant correction factor (0.94g/cm3) consistent with the density of adipose tissue [12]. All VAT
160
under 50g was set to 50g since the DXA measurement is unreliable in the low range visceral fat
161
content (precision error (with the iDXA) is approximately 50g) [15]. The short- and long-term
162
coefficients of variation for our densitometer are 0.8% and 1.4%, respectively.
163 164 165
Laboratory measurements
166
All oral glucose tolerance tests (OGTTs) were performed in the morning after an overnight fast at
167
antenatal visit GA 30-32 weeks and follow-up visit. Venous EDTA blood was analyzed on site by the
168
Accu-Check Sensor glucometer (Roche Diagnostics GmbH, Mannheim, Germany). Venous blood was
Page 7 of 19
169
also drawn in gel tubes, allowed to clot for 30 min, thereafter centrifuged for 10 min 3000 g, serum
170
separated and stored at -80. Glucose was measured from frozen serum samples at antenatal visit GA
171
30-32 weeks, with the hexokinase method at an accredited clinical chemistry laboratory at Oslo
172
University Hospital, Rikshospitalet (Hitachi Modular P800 with reagents from Roche) and results are
173
used further for this study. The median difference in glucose values for the 300 women at antenatal
174
visit GA 30-32 weeks between on-site measurements [16] and the measurements performed at the
175
clinical chemistry laboratory was 0.09, 0.17, 0.24, 0.19, and 0.21 mmol/l for the timepoints
176
0,30,60,90,120 min in the OGTT test, respectively, and highly correlated R2=0.70, 0.84, 0.90, 0.91,
177
0.90, consistent with what others have found [17]. For the follow up visit we used the glucose data
178
from the Accu-check Sensor glucometer. Insulin samples were assayed in duplicate (RIA, DPC, Los
179
Angeles, CA) as previously reported [16] and the same method were used for the follow-up samples.
180 181
Diagnosis of GDM and prediabetes
182
GDM was diagnosed with the IADPSG criteria; fasting plasma glucose (FPG) 5.1-6.9 mmol/l, 1-h
183
plasma glucose ≥10.0 mmol/l or 2-h plasma glucose 8.5-11.0 mmol/l following a 75 g oral glucose
184
load [18]. Prediabetes was diagnosed at follow-up visit with the criteria of FPG 5.6-6.9 mmol/l or 2-h
185
plasma glucose 7.8-11.0 mmol/l after 75 g OGTT [19]. GDM (IADPSG) and prediabetes were
186
diagnosed after the study was finished. The women who were diagnosed with GDM during pregnancy
187
(WHO) did not get any medication but did receive nutrition guidance.
188 189
Insulin sensitivity was measured with the Matsuda index 10,000/square root of [fasting glucose
190
(mmol/l) x fasting insulin (mU/l] x [mean glucose (mmol/l) x mean insulin (mU/l)] during OGTT.
191
This index is a measure of whole body insulin sensitivity that has been validated against the
192
euglycemic-hyperinsulinemic clamp [20]. β-cell function was assessed with the insulin secretion-
193
sensitivity index (ISSI-2) (area under the curve (AUC) insulin (mU/l) 0-120 /glucose(mmol/l) 0-120 x
194
Matsuda), validated against the disposition index from the intravenous glucose tolerance test [21] and
195
HOMA–IR was calculated as fasting insulin (mU/l) × fasting glucose (mmol/l)/22.5, as described by
196
Matthews et al [22].
Page 8 of 19
197 198
Statistics
199
Statistical analyses were conducted using SPSS for Windows, version 21.0 (Chicago, IL, USA). Data
200
are expressed as mean±SD when normally distributed and median (25th, 75th percentile) when skewed.
201
Comparison between women with GDM and non-GDM were performed using t-test or Mann-Whitney
202
U depending on distribution, and Chi-square test for categorical variables. Univariate and stepwise
203
(probability of F to-enter 0.1 -remove 0.15) linear regression analyses was carried out on log
204
transformed variables (if skewed) and results are given as standardized regression coefficients.
205
Interactions between VAT and indices of glucose metabolism were evaluated by univariate general
206
linear model with glucose indices as dependent and VAT, glucose indices, GDM and the interaction
207
term (VAT*GDM) as independent variables.
