A total of 589 pregnant women with GDM were enrolled and finished their postpartum visit in this study. A total of 227 (38.5%) of these women were diagnosed with dyslipidemia and 209 (35.5%, 32.4% with prediabetes and 3.1% with diabetes) were diagnosed with abnormal glucose tolerance at 6–12 weeks after delivery. A total of 23.1% of participants had dyslipidemia with normal glucose tolerance, which accounted for up to 60% of dyslipidemia. A total of 15.49% (13.6% with prediabetes and 1.89% with diabetes) of participants had both postpartum glucose intolerance and dyslipidemia (Fig. 1a). Of these, 195 (33.1%) had abnormal TC levels, 33 (5.6%) had abnormal TG levels, 15 (2.5%) had abnormal HDL-C levels, and 127 (21.6%) had abnormal LDL-C levels. A total of 42.3% of the participants presented with only one type of dyslipidemia (Fig. 1b). Women with dyslipidemia had an older delivery age, higher levels of SBP, HbA1c, 1 h PG, TC, TG, and LDL-C, a higher TG/HDL-C ratio during pregnancy, higher postpartum glucose levels, and a higher incidence of postpartum glucose intolerance compared with women with normal postpartum lipids (Table 1).
Table 1 Basic characteristics of women with history of GDM
|
Postpartum NLG
(n=362)
|
Postpartum DLG
(n=227)
|
P value
|
Age (years) median (IQR)
|
33(30-36)
|
34(30.5-38)
|
0.019*
|
age ≥35, n/%
|
149 /41.2%
|
114/50.2%
|
0.057
|
BMI a (kg/m2) median (IQR)
|
21.49(19.59-23.89)
|
21.99(20.04 -23.82)
|
0.082
|
BMI a ≥24kg/m2, n /%
|
81/24.6%
|
57/23.9%
|
0.375
|
Weight gain during pregnancy (kg) median (IQR)
|
11(8-14)
|
11(8.8-14.42)
|
0.384
|
Family history of diabetes, n/%
|
93 /63.3%
|
54/36.7%
|
0.582
|
Insulin history, n/%
|
3/0.9%
|
5/2.4%
|
0.148
|
SBP (mmHg) median (IQR)
|
117 (110-125)
|
122 (122-129)
|
0.01*
|
DBP (mmHg) median (IQR)
|
74(68-79)
|
74(68-80)
|
0.299
|
24-28 weeks during pregnancy
|
FPG (mmol/L), median (IQR)
|
4.6(4.3-4.9)
|
4.6(4.3-5.0)
|
0.422
|
1 h PG (mmol/L), median (IQR)
|
9.8(8.8-10.4)
|
10.0(9.1-10.6)
|
0.01*
|
2 h PG (mmol/L), median (IQR)
|
8.8(8.4-9.3)
|
8.8(8.35-9.6)
|
0.194
|
HbA1C (%), median (IQR)
|
5.0 (4.8-5.2)
|
5.1 (4.8-5.3)
|
0.03*
|
TC (mmol/L), median (IQR)
|
6.0(5.3-6.6)
|
6.7(6.0-7.5)
|
0.00*
|
TG (mmol/l), median (IQR)
|
2.15(1.78-2.7)
|
2.47(1.91-3.17)
|
0.00*
|
HDL-C(mmol/L), median (IQR)
|
1.92(1.72-2.16)
|
1.96(1.69-2.27)
|
0.06
|
LDL-C(mmol/L), median (IQR)
|
3.28(2.82-3.69)
|
3.76(3.29-4.29)
|
0.00*
|
TG / HDL-C ratio
|
1.13(0.85-1.51)
|
1.28(0.91-1.75)
|
0.003*
|
At 6–12 weeks postpartum
|
FPG (mmol/l), median (IQR)
|
4.7(4.4-5.0)
|
4.8(4.6-5.2)
|
0.00*
|
1 h PG (mmol/l), median (IQR)
|
8.5(7.6-9.4)
|
9.1(8.0-9.9)
|
0.02*
|
2 h PG (mmol/l), median (IQR)
|
6.9(5.8-8.2)
|
7.4(6.4-8.4)
|
0.029*
|
* P<0.05;
a pre-pregancy
NLG, normal lipid group; DLG, dyslipidemia group;
IQR, inter-quartile range; BMI, body mass index; SBP, systolic blood pressure; DBP, diastolic blood pressure; TC, total cholesterol; TG, triglycerides. HDL-C high-density lipoprotein cholesterol; LDL-C, low-density lipoprotein cholesterol; FPG, fasting plasma glucose; PG, plasma glucose.
Logistic regression analysis showed that age and SBP, levels of 1 h PG and HbA1c, the lipid profile (TC, TG, HDL-C, and LDL-C), and the TG/HDL-C ratio during pregnancy were significantly associated with postpartum lipid outcome. The odds ratios (ORs) for these variables ranged 1.047–2.551. Multivariate logistic regression analysis further showed that age (OR = 1.06, 95% confidence interval [CI]: 1.014–1.109, P = 0.11), SBP (OR = 1.022, 95% CI: 1.006–1.038, P = 0.006), HbA1c (OR = 1.897, 95% CI: 1.119–3.215, P = 0.017), and LDL-C (OR = 3.671, 95% CI: 1.386–9.724, P = 0.009) were independent predictors of abnormal postpartum lipid metabolism. ROC curves were used to predict dyslipidemia. Sensitivity in the prediction of incident dyslipidemia of each predictor varied from 47.1% (age) to 63.9% (LDL-C), and specificity decreased from 69.6% (SBP) to 57.4% (HbA1c) across these categories. The cut-offs of age, SBP, HbA1c values, and LDL-C levels were 35 years, 123 mmHg, 5.1%, and 3.56 mmol/L, respectively. The AUCs ranged 0.56–0.696 (Table 2).
Table 2. Logistic analysis of the factors during pregnancy associated with postpartum abnormal lipid metabolism*
|
|
β value
|
OR (95%CI)
|
P value
|
sensitivity
|
specificity
|
AUC
|
Cut-off
|
Age (y)
|
crude
|
0.048
|
1.047 (1.007-1.088)
|
0.02
|
51.4
|
58
|
0.56
|
35
|
adjust
|
0.06
|
1.06 (1.014-1.109)
|
0.011
|
SBP (mmHg)
|
crude
|
0.023
|
1.023 (1.008-1.038)
|
0.02
|
47.1
|
69.6
|
0.593
|
123
|
adjust
|
0.022
|
1.022 (1.006-1.038)
|
0.006
|
1h PG
|
crude
|
0.159
|
1.172 (1.03-1.333)
|
0.016
|
-
|
-
|
-
|
-
|
HbA1c (%)
|
crude
|
0.717
|
2.069 (1.295-3.303)
|
0.002
|
55.8
|
57.4
|
0.576
|
5.1
|
adjust
|
0.601
|
1.897 (1.119-3.215)
|
0.017
|
TC
|
crude
|
0.669
|
1.949 (1.629-2.331)
|
0.00
|
-
|
-
|
-
|
-
|
TG
|
crude
|
0.46
|
1.566 (1.272-1.928)
|
0.00
|
-
|
-
|
-
|
-
|
HDL-C
|
crude
|
0.434
|
1.716 (1.11-2.654)
|
0.015
|
-
|
-
|
-
|
-
|
LDL-C (mmol/l)
|
crude
|
0.986
|
2.551 (1.969-3.304)
|
0.00
|
63.9
|
69.2
|
0.696
|
3.56
|
adjust
|
1.341
|
3.671 (1.386-9.724)
|
0.009
|
TG / HDL-C
|
crude
|
0.359
|
1.4 (1.085-1.087)
|
0.01
|
-
|
-
|
-
|
-
|
* The R2 of logistic model was 0.247.
crude= simple logistic regression analysis without adjust variables; adjust = Adjusted for the 1 h PG, TC, TG, HDL-C and TG/HDL-C ratio during pregnancy.
OR = odds ratio; CI = confidence interval; SBP = Systolic blood pressure; AUC = area under the receiver operating characteristic curve; TC = total cholesterol; TG = triglycerides. LDL-C = low-density lipoprotein cholesterol; 1 h PG= 1 h plasma glucose.
Generally, LDL-C achieved a balanced mix of high sensitivity and specificity, with the highest area under the AUC (0.696). To improve the overall predictability, the AUC was up to 0.733 when all of the independent predictors were combined (Figure 2).
Study strength and limitations
This study found that age, SBP before labor, HbA1c and LDL-C in the second trimester of pregnancy were the potential predictors in women with history of GDM, and also has a few limitations. First, the results were based on a fairly short follow-up period and all participants were from a single center. Second, a retrospective study may result in selective bias and incomplete clinical data. Finally, some characteristics of the subjects were lacking, such as breastfeeding, diet, and physical activity, which might affect glucolipid metabolism during pregnancy and postpartum. Therefore, a long-term follow-up and large-scale multicenter study for validation of our results should be performed in the future.