Postpartum Adiponectin Concentration, Insulin Resistance and Metabolic Abnormalities Among Women With Pregnancy-Induced Disturbances

Authors


Tina Costacou, PhD, University of Pittsburgh, Department of Epidemiology, 3512 Fifth Avenue, Pittsburgh, PA 15213
E-mail: costacout@edc.pitt.edu

Abstract

The authors compared postpartum adiponectin levels among women with prior pregnancy-induced disturbances and assessed their association with homeostasis model assessment for insulin resistance (HOMA-IR), the metabolic syndrome (MS), and the Framingham risk score (FRS). Women delivering in 1998 through 2001 and who had gestational diabetes mellitus (n=22), gestational hypertension (n=32), or preeclampsia (n=34) were examined 1 to 2 years after delivery and were grouped-matched to controls (n=29) by age and prepregnancy body mass index. HOMA-IR was increased, adiponectin values were decreased, and there was a higher MS prevalence in women with prior gestational diabetes mellitus (all P<.05). Adiponectin levels were inversely related to HOMA-IR (r=−0.45; P<.0001) and FRS (r=−0.25; P=.007), and a significant trend for decreasing adiponectin values with increased number of MS components was noted (P trend <.0001). Adiponectin concentration remained a significant correlate of FRS and MS irrespective of pregnancy history; a concentration <10.5 µg/mL provided the optimal cutoff to distinguish those with or without MS. Thus, a lower postpartum adiponectin concentration identifies women at increased cardiovascular risk regardless of pregnancy history.

Hypertensive disorders, including preeclampsia, affect approximately 6% to 8% of pregnancies and are the second leading cause of maternal mortality in the United States.1 Furthermore, approximately 7% of all pregnancies are complicated by gestational diabetes mellitus, resulting in around 200,000 cases annually nationwide.2 Although insulin resistance gradually develops during a normal pregnancy as a way to facilitate the transfer of glucose to the fetus and is known to peak in the third trimester,3–5 it has been suggested that this process may be exaggerated among women with these two pregnancy-induced disturbances.6 Moreover, new-onset blood pressure elevations during pregnancy may lead to subsequent hypertension or cardiovascular disease in women, although study findings have been conflicting.7,8 A history of gestational diabetes is also well known to increase the risk of future development of type 2 diabetes.9 However, to our knowledge, there is lack of data regarding the role of gestational diabetes in the prediction of cardiovascular disease risk after pregnancy, despite limited evidence of subclinical inflammation and early vascular dysfunction postpartum.10

Recently, considerable interest has developed concerning adiponectin, an adipose-tissue specific protein possessing both antiatherogenic and anti-inflammatory properties.11–13 Adiponectin has been inversely related to the severity of insulin resistance,14 glucose intolerance,15,16 and coronary artery disease.15,17 Thus, the hypothesis that decreasing adiponectin concentrations relate to increasing insulin resistance in pregnancy and therefore enhanced subsequent cardiovascular resistance is appealing. However, study findings have thus far produced inconsistent results.18–22 Conflicting findings also exist regarding adiponectin levels in gestational hypertensive disorders or preeclampsia.18,19,23–25 However, the concentration of this adipokine appears decreased in gestational diabetes.19,26

We therefore aimed to determine the prevalence, 2 years postpartum, of insulin resistance and metabolic abnormalities (as defined by the metabolic syndrome [MS] and the Framingham 10-year coronary heart disease [CHD] risk score) among women with 3 pregnancy-induced disturbances (namely, gestational diabetes mellitus, gestational hypertension, and preeclampsia) and healthy women with normal pregnancies. We also assessed whether plasma adiponectin concentration differed among these groups of women and whether adiponectin improves the identification of women at increased cardiovascular disease risk by virtue of having MS or a higher Framingham 10-year CHD risk score.

MATERIALS AND METHODS

Study Population

The Prenatal Exposures and Preeclampsia Prevention (PEPP) study enrolled approximately 3000 women who delivered at the Magee-Women's Hospital in Pittsburgh, Pennsylvania. The PEPP study protocol was approved by the Magee-Women's Hospital Institutional Review Board. The design and methods have been described previously.27 Briefly, women underwent a standardized interview at their first prenatal visit administered by trained, professional interviewers. Medical records during pregnancy (blood pressure profile, laboratory results of blood and urine testing) were registered on a medical data abstraction form.

Eligibility criteria for the current study included being between 18 and 39 years of age at the index delivery and having delivered between October 1998 and March 2001, at least 12 months before the recruitment (December 2001-October 2002). Women with preexisting diabetes or hypertension and those reporting regular use of steroid drugs (except birth control pills) or serious endocrine disease were excluded, as were those having a further pregnancy between the index delivery and recruitment. The sampling frame (n=1509) was stratified into 4 groups based on the patient's pregnancy medical data and/or PEPP chart abstraction.

Preeclampsia (n=122) was defined as blood pressure elevation (a single blood pressure value of >140/90 mm Hg after 20 weeks of gestation) plus proteinuria (>1+ on a catheterized or >2+ on a voided urine sample or >300 mg/d on 24-hour urine specimen results). The same elevations in blood pressure in the absence of proteinuria defined gestational hypertension (n=142). The diagnosis of gestational diabetes (n=44) was based on the American Diabetes Association criteria: initially a 50-g oral glucose load (glucose challenge test [GCT]) was performed between 24 and 28 weeks of gestation and a diagnostic 100-g oral glucose tolerance test was done on the subset of women with a 1-hour serum glucose value >140 mg/dL on the GCT. Women with no signs of elevated blood pressure or proteinuria and with a GCT result <130 mg/dL were classified as potential controls (n=554). During the enrollment period, initially patients with preeclampsia, gestational hypertension, and gestational diabetes were contacted and invited to the follow-up visit. After a substantial number of cases (n≈60) were scheduled and examined, controls were group-matched for age and prepregnancy body mass index (BMI; measured before the index pregnancy or at the first prenatal visit) using 2-year age categories (range, 18–39 years) and 4 kg/m2 BMI (range, 16–52 kg/m2). As a result of this procedure, 82 women were identified as eligible controls. Compared with the larger cohort of 1509 women participating in the PEPP study, women selected for the present study were slightly older (27.7 vs 26.0 years; P<.0001) with a greater BMI (28.6 vs 24.9 kg/m2; P<.0001) and higher blood pressure levels at the clinic visits, but a similar number of prior pregnancies (2.3 vs 2.1; Wilcoxon 2-sided P=.09). These differences are consistent with our oversampling of women with preeclampsia, gestational diabetes, and gestational hypertension.

Follow-Up Procedures

The study procedure was fully approved by both the University of Pittsburgh and the Magee-Women's Hospital Institutional Review Board. Initially, a letter was sent to eligible patients informing them of and inviting them to the study. Approximately 7 to 10 days later, they were contacted by phone to answer any questions regarding the study; obtain information on current pregnancy, serious medical problems, medication use (eg, steroids); confirm date of last delivery (index delivery); and arrange an appointment if the patient was agreeable. Patients were scheduled for available appointments at the Magee Clinical Research Center (CRC). Thirty-six women with prior preeclampsia, 33 with previous gestational hypertension, 24 with previously diagnosed gestational diabetes, and 31 controls were successfully enrolled during the planned recruitment period.28

All women were examined after an 8-hour overnight fast. A urine pregnancy test was conducted to exclude the possibility of current gestation. Participants were questioned regarding recent medical problems, medications, and family history of high blood pressure and diabetes. Clinical (weight, height, waist and hip circumference, blood pressure, pulse rate) as well as laboratory parameters (serum insulin, plasma glucose, hemoglobin A1c, low-density lipoprotein cholesterol, high-density lipoprotein (HDL) cholesterol, lipoprotein(a), triglyceride, urine albumin, and creatinine values) were measured during the visit.

Clinical and Laboratory Measures

Weight and height were measured using a standard calibrated scale. BMI was calculated as body weight in kilograms divided by height in meters squared. Waist circumference was measured while standing (light clothing, empty pockets, and no shoes) midway between the lowest rib and the iliac crest. Hip circumference was measured over the widest part of the gluteal region. Measurements were repeated, and if a >5-cm difference was found, a third measurement was performed. Average results and waist-to-hip ratio (WHR) were calculated. Blood pressure readings were measured by a random-zero sphygmomanometer according to the Hypertension Detection Follow-up Protocol29 after a 5-minute rest period. The average blood pressure was calculated.

Plasma free insulin levels were measured using radioimmunoassay (Linco Research, Inc, St Charles, MO) (cross-reactivity of the antibody with human proinsulin is <0.2%). Plasma glucose was measured by glucose oxidase method; lipoprotein(a) was determined using enzyme-linked immunosorbent assay; total cholesterol, HDL cholesterol, and triglycerides were assessed by an enzymatic, colorimetric method;30 and low-density lipoprotein cholesterol was calculated using the Friedewald equation.31 Hemoglobin A1c and urine creatinine and albumin concentrations were determined using the DCA2000 (Bayer Co, Elkhart, IN). The albuminto-creatinine ratio (ACR) was calculated to assess microalbuminuria (ACR >30 mg/g).

Total adiponectin level was measured using an radioimmunoassay procedure developed by Linco Research, Inc. Briefly, samples (citrated plasma) were diluted 1:500 in assay buffer, mixed with 125I-adiponectin and adiponectin antibody, and then incubated at room temperature for 20 to 24 hours. The adiponectin-antibody complex was precipitated with precipitating agent at 4° for 20 minutes and then sedimented by centrifugation at 3,000 × g for 45 minutes. Finally, the supernatant was decanted and the pellets counted. Under these conditions, the limit of sensitivity is 1 ng/mL and the response is linear up to 200 ng/mL. Standards, blanks, quality control, and a control pool were run simultaneously with all samples. The coefficient of variation between runs was 8.0%.

A thyroid-stimulating hormone test was conducted in women thought to have possible thyroid dysfunction by clinical signs, medication or medical history data, and to rule out current endocrine insufficiency (n=9). An abnormal (elevated) thyroid-stimulating hormone value was found only in one person who was excluded from the final analyses, reducing the number of participants with prior gestational hypertension to 32.

Clinical Definitions at Follow-Up

Insulin resistance was determined using the homeostasis model assessment method: ([serum insulin × fasting plasma glucose]/22.5). Follow-up hypertension was diagnosed when the average of the 3-time measured blood pressure values reached or exceeded 140/90 mm Hg and/or the patient received antihypertensive medication.

The definitions of MS were based on the most recent National Cholesterol Education Program Expert Panel on Detection, Evaluation, and Treatment of High Blood Cholesterol in Adults (NCEP ATP III) criteria,32 as modified by the American Heart Association/National Heart, Lung, and Blood Institute (AHA/NHLBI)33 and International Diabetes Federation (IDF) criteria.34 Thus, a diagnosis of MS by NCEP-ATP III criteria was defined as having any 3 of the following: waist circumference >88 cm, triglycerides ≥1.7 mmol/L, HDL cholesterol <1.3 mmol/L, blood pressure ≥130/85 mm Hg, or fasting plasma glucose ≥5.6 mmol/L. By IDF, MS was defined as having central obesity (ethnicity-specific values measured by waist circumference: we used the Europid cutoffs) plus ≥2 of the following: raised triglyceride level (≥1.7 mmol/L) or treatment for hypertriglyceridemia; reduced HDL cholesterol value (<1.3 mmol/L) or treatment for this lipid abnormality; raised blood pressure level (≥130/85 mm Hg) or antihypertensive treatment; or raised fasting plasma glucose level (≥5.6 mmol/L). The Framingham 10-year CHD risk score was calculated as described by Wilson and associates,35 with the exception that no data were available on smoking status.

Statistical Analyses

Data analysis was performed using SAS version 9.1 (SAS Institute, Inc, Cary, NC). For continuous variables not normally distributed (ie, homeostasis model assessment for insulin resistance [HOMA-IR] and values of serum insulin, plasma glucose, lipoprotein(a), triglycerides, urinary creatinine and albumin, and ACR), logarithmic transformation was performed. To assess differences in continuous variables by pregnancy status, univariable analysis of variance models were constructed and the Bonferroni post hoc test for multiple comparisons was used. The chi-square test or Fisher's exact test, as applicable, were applied to evaluate differences in categoric variables. Bivariate associations between adiponectin and measured continuous covariates were assessed with Pearson's correlation coefficients. The sensitivity, specificity, and positive predictive value of adiponectin and the three pregnancy-induced disturbances for MS were calculated from 2×2 tables, whereas the Youden's index was defined as sensitivity plus specificity minus 1. Finally, logistic regression models with MS as the outcome variable were constructed to obtain the area under the curve, whereas the association between adiponectin and the Framingham 10-year CHD risk score was assessed by linear regression models.

RESULTS

Participants with gestational hypertension (n=32) and gestational diabetes (n=24) examined at the follow-up visit had similar prepregnancy age and BMI to those who were identified as eligible but not enrolled into the study (not shown). Women with prior preeclampsia who participated in the study (n=36) were significantly older (29.3±5.6 vs 26.7±6.2 years; P=.02) but had a similar BMI compared with the women with preeclampsia identified but not recruited (not shown). The mean follow-up time from the patients' index delivery (postpartum period) was 2.2±0.7 years. The mean age of the participants was 31.5±5.5 years (range, 20–40 years), and the mean BMI was 31.3±7.7 kg/m2 (range, 17–51 kg/m2) at the follow-up. One hundred twenty women had adequate plasma for the measurement of adiponectin. Furthermore, because this study included a very small number of Hispanic (n=2) and Asian/Pacific Islander (n=1) participants, the present analyses were restricted to African American (n=25) and non-Hispanic white women (n=92).

Table I presents participant characteristics by study group. No demographic differences were seen between the 4 groups of women studied, although women with prior gestational diabetes had a greater BMI compared with women with a normal pregnancy after adjusting for race. As expected, higher blood pressure values (especially diastolic) were seen in all 3 groups compared with controls, and in those with prior preeclampsia only, a non-statistically significant higher ACR was observed. Among the latter, 14.7% had microalbuminuria. In addition, those with prior gestational diabetes had elevated HOMA-IR and glucose, insulin, hemoglobin A1c, and lipid values compared with controls. The prevalence of MS by either definition also appeared higher in women with prior gestational diabetes compared with normal controls and those with prior preeclampsia or gestational hypertension (45.5 vs 24.2; P=.046 by the NCEP ATP III criteria and 45.5 vs 22.1; P=.03 by the IDF criteria), although no differences were observed among the other groups and the overall P value did not reach statistical significance (Fisher's exact P values, .26 and .18 for the NCEP ATP III and IDF definitions, respectively). Of note, the NCEP ATP III and IDF definitions provided almost identical classifications of women with and without MS and, thus, all subsequent analyses were conducted using the NCEP ATP III definition. Those with prior gestational hypertension had higher HDL cholesterol concentrations and nonsignificantly lower lipoprotein(a) values compared with controls. Mean adiponectin concentrations were lower among participants with prior gestational diabetes compared with all other women (P=.08); however, formal statistical significance was not reached at α=.0083. No differences in plasma adiponectin levels were observed between women with previous preeclampsia or gestational hypertension vs their control counterparts, even after adjustment for ACR, which slightly reduced the adiponectin concentration in women with prior pregnancy-induced hypertensive disorders (10.7 and 10.9 µg/mL for women with prior preeclampsia and gestational hypertension, respectively). Only one person had a positive score on the Framingham 10-year CHD risk scale, and no differences were observed by prior pregnancy-induced abnormality.

Table I.  Characteristics and Cardiovascular Risk Factor Profile by Study Group at Follow-Up
Participant CharacteristicsControls (n=29) Mean (SD)Preeclampsia (n=34) Mean (SD)Gestational Hypertension (n=32) Mean (SD)Gestational Diabetes (n=22) Mean (SD)
Age, y28.5 (5.2)29.0 (5.6)28.8 (6.6)28.9 (4.8)
Follow-up, y2.3 (0.63)2.3 (0.76)2.2 (0.66)2.1 (0.81)
African Americans, % (No.)27.6 (8)23.5 (8)15.6 (5)18.2 (4)
Prepregnancy BMI, kg/m2 [No.]28.7 (5.9) [29]28.7 (6.4) [27]28.2 (5.3) [25]32.3 (7.2) [20]
BMI, kg/m230.4 (6.8)30.6 (7.1)31.0 (8.3)35.4 (8.3)
Waist circumference, cm88.7 (13.5)87.4 (14.2)88.2 (12.7)100.9 (22.4)a,b,c
Waist-to-hip ratio0.80 (0.07)0.80 (0.06)0.81 (0.06)0.84 (0.08)
Systolic blood pressure, mm Hg109.8 (14.4)117.6 (11.2)116.5 (10.9)119.1 (10.6)a
Diastolic blood pressure, mm Hg72.6 (8.4)82.0 (9.4)a78.8 (8.2)a80.2 (7.1)a
Pulse, bpm [No.]73.1 (10.4) [27]76.9 (12.9) [32]77.0 (9.4) [31]76.4 (10.5) [22]
Hemoglobin A1c, % [No.]5.0 (0.39) [28]5.0 (0.42) [34]5.0 (0.32) [31]5.4 (0.81) [22]a,b,c
Serum insulin, µU/mL14.7 (14.7)17.9 (11.4)14.7 (5.6)25.8 (18.7)a,c
Plasma glucose, mg/dL [No.]86.7 (9.8) [29]90.9 (7.6) [34]91.0 (10.1) [31]102.5 (16.2) [22]a,b,c
HOMA-IR [No.]3.2 (3.1) [29]4.1 (2.8) [34]3.4 (1.6) [31]6.9 (6.1) [22]a,b,c
HOMA-IR, highest tertile (%) [No.]17.2 (5) [29]41.2 (14) [34]29.0 (9) [31]59.1 (13) [22]d
MSe, % (No.)27.6 (8)23.5 (8)21.9 (7)45.5 (10)
MSf, % (No.)24.1 (7)23.5 (8)18.8 (6)45.5 (10)
Total cholesterol, mg/dL [No.]190.4 (39.2) [29]188.5 (37.1) [33]192.2 (31.0) [32]212.3 (47.1) [22]
HDL cholesterol, mg/dL47.1 (7.8)50.5 (10.1)54.4 (11.5)a51.8 (11.6)
Non-HDL cholesterol, mg/dL [No.]143.4 (40.2) [29]138.2 (37.9) [33]137.8 (33.2) [32]160.5 (42.0) [22]
Triglycerides, mg/dL [No.]117.8 (68.6) [29]112.2 (49.7) [33]117.3 (68.6) [32]132.9 (48.4) [22]
Lipoprotein(a), mg/dL38.9 (38.1)41.4 (38.9)25.1 (35.3)40.9 (54.2)
Adiponectin, µg/mL10.7 (4.6)11.2 (4.5)11.3 (3.3)9.3 (3.8)
Urine creatinine, mg/dL [No.]167.1 (122.6) [27]127.5 (88.8) [34]155.3 (79.3) [31]198.9 (68.5) [22]b
Urine albumin, mg/L [No.]14.9 (14.6) [27]16.4 (21.2) [34]10.7 (5.2) [31]27.7 (60.1) [21]
Albumin-to-creatinine ratio, mg/g [No.]10.2 (8.9) [27]19.9 (32.2) [34]9.2 (6.7) [31]11.9 (18.1) [21]
Microalbuminuria, % [No.]7.4 (2) [27]14.7 (5) [34]3.2 (1) [31]4.8 (1) [21]
Framingham risk score [No.]−9.0 (3.0) [29]−8.2 (4.2) [33]−8.8 (3.4) [32]−8.4 (3.0) [22]
Abbreviations: bpm, beats per minute; BMI, body mass index; HDL, high-density lipoprotein; HOMA-IR, homeostasis model assessment for insulin resistance; MS, the metabolic syndrome; aSignificantly different from control group at α=.0083. bSignificantly different from gestational hypertension group at α=.0083. cSignificantly different from the preeclampsia group at α=.0083. dChi-square test/Fisher's exact test P value <.05. eDiagnosed by the Third Report of the National Cholesterol Education Program Adult Treatment Panel criteria. fDiagnosed by International Diabetes Federation criteria.

As expected, in all groups combined adiponectin concentration was inversely related to BMI, waist circumference, hemoglobin A1c value, serum insulin level, fasting plasma glucose level, and HOMA-IR (Table II). Inverse correlations were also observed between adiponectin and non-HDL cholesterol level, triglyceride concentration, urinary creatinine level, albumin level, and Framingham 10-year CHD risk score. Conversely, a direct correlation with HDL cholesterol was observed. Women with MS by either definition had a lower adiponectin concentration, and there was a linear decrease in adiponectin levels with an increasing number of components of the NCEP ATP III MS definition. This relationship was not dependent on prior pregnancy complication status, as it held even within the 4 prior pregnancy-induced abnormality groupings. Reduced adiponectin concentrations were also observed among women in the upper tertile of HOMA-IR (P<.0001). Generally, these associations remained significant after adjustment for waist circumference (Table III), with the exception of subgroup analyses, potentially due to the decrease in sample size but also because abdominal obesity is likely to mediate some of these associations, especially among women with gestational diabetes in whom the lower adiponectin levels among those with MS were completely explained by abdominal obesity.

Table II.  Pearson Correlation Coefficients Between Participant Characteristics and Adiponectin Levels
Participant CharacteristicsNo.Adiponectin ConcentrationaP Value
Age at study entry, y1170.08.42
Prepregnancy BMI, kg/m2101−0.29.003
BMI, kg/m2117−0.32.0005
Waist circumference, cm117−0.43<.0001
Waist-to-hip ratio117−0.47<.0001
Systolic blood pressure, mm Hg117−0.08.38
Diastolic blood pressure, mm Hg117−0.07.46
Pulsea112−0.05.63
Hemoglobin A1c, %115−0.45<.0001
Insulin, µU/mLa117−0.42<.0001
Fasting plasma glucose, mg/dLa116−0.34.0002
HOMA-IRa116−0.45<.0001
Total cholesterol, mg/dL116−0.06.52
HDL cholesterol, mg/dL1170.37<.0001
Non-HDL cholesterol, mg/dL116−0.16.08
Triglycerides, mg/dLa116−0.35.0001
Lipoprotein(a), mg/dLa117−0.05.58
Urinary creatinine, mg/dLa114−0.19.05
Urinary albumin, mg/La113−0.26.006
ACR, mg/g1130.03.75
Framingham risk score116−0.25.007
Abbreviations: ACR, albumin-to-creatinine ratio; BMI, body mass index; HDL, high-density lipoprotein; HOMA-IR, homeostasis model assessment for insulin resistance. aLogarithmically transformed.
Table III.  Adiponectin Concentration by Participant Characteristics
Participant CharacteristicsMean (SD)CrudeP ValueAdjustedP Valuea
Race/ethnicity
 African American (n=25)9.6 (3.6)  
 Non-Hispanic white (n=92)11.1 (4.2).12.35
IDF definition of the metabolic syndrome
 No (n=86)11.7 (4.2)  
 Yes (n=31)8.1 (2.6)<.0001.03
NCEP ATP III definition of the metabolic syndrome
 No (n=84)11.8 (4.2)  
 Yes (n=33)8.1 (2.5)<.0001.009
Number of components of the metabolic syndrome (by NCEP ATP III)
 0 (n=24)13.6 (5.1)Overall P value:Overall P value:
 1 (n=32)11.3 (3.2)<.0001.04
 2 (n=28)10.7 (3.8)  
 3 (n=24)8.5 (2.6)bP trend: <.0001P trend: .001
 4 (n=5)7.9 (2.6)  
 5 (n=4)6.1 (1.8)
Among controls (n=29)
 No metabolic syndrome (NCEP ATP III [n=21])11.7 (4.9)  
 Metabolic syndrome (NCEP ATP III [n=8])8.0 (2.5).06.28
Among prior preeclampsia (n=32)
 No metabolic syndrome (NCEP ATP III [n=26])12.1 (4.6)  
 Metabolic syndrome (NCEP ATP III [n=8])8.2 (2.6).03.10
Among prior gestational hypertension (n=32)
 No metabolic syndrome (NCEP ATP III [n=25])12.2 (3.1)  
 Metabolic syndrome (NCEP ATP III [n=7])8.4 (2.3).007.03
Among prior gestational diabetes (n=22)
 No metabolic syndrome (NCEP ATP III [n=12])10.6 (4.0)  
 Metabolic syndrome (NCEP ATP III [n=10])7.9 (3.0).09.99
HOMA-IR
 <3.9 (n=75)12.1 (4.3)  
 ≥3.9 (n=41)8.4 (2.5)<.0001.002
Family history of diabetes
 No (n=40)11.0 (4.5)  
 Yes (n=73)10.6 (4.0).60.91
Family history of hypertension
 No (n=34)11.3 (3.7)  
 Yes (n=80)10.4 (4.3).30.42
Abbreviations: HOMA-IR, homeostasis model assessment for insulin resistance; IDF, International Diabetes Federation; NCEP ATP III, Third Report of the National Cholesterol Education Program Adult Treatment Panel. aAdjusted for waist circumference. bSignificantly different from 0 at P=.003.

To evaluate whether a cut point exists in the distribution of adiponectin concentration by which individuals at high risk for MS can be identified, we used logistic regression with MS as the outcome and adiponectin as the continuous independent variable. The area under the curve was 0.773 for adiponectin as a continuous variable, suggesting good separation between women with and without MS. Using Youden's index, a concentration of 10.5 was selected as the optimal cut point on the receiver operating characteristic curve. Table IV presents the sensitivity, specificity, positive predictive value, Youden's index, and area under the receiver operating characteristic curve for the dichotomized adiponectin, preeclampsia, gestational hypertension, gestational diabetes, and combinations of low adiponectin and prior pregnancy-induced complication. The dichotomized adiponectin data provided the best results in distinguishing women with and without MS, even when adiponectin was used in addition to the presence of any of the 3 pregnancy-induced disturbances (eg, the area under the receiver operating characteristic curve increased from 0.713 with adiponectin as the main independent variable to only 0.739 with the addition of gestational diabetes into the model). The relationship between adiponectin concentration and the Framingham 10-year CHD risk score was also assessed. Adiponectin was significantly, inversely related to the Framingham risk score after adjusting for race/ethnicity, BMI, and pregnancy-induced disturbances (β estimate= −1.75; P=.04).

Table IV.  Sensitivity, Specificity, Positive Predictive Value, Youden's Index, and Area Under the Curve for the NCEP ATP III Definition of the Metabolic Syndrome
 SensitivitySpecificityPPVNPVYouden's IndexAUCP Valuea
Adiponectin <10.5 µg/mL81.860.745.089.50.4250.713 
Preeclampsia24.269.123.569.9−0.0670.534.005
Gestational hypertension21.270.221.969.4−0.0860.543.003
Gestational diabetes30.385.745.575.80.1600.580.03
Adiponectin <10.5 µg/mL plus:
 Preeclampsia21.289.343.874.30.1050.552.001
 Gestational hypertension15.291.741.773.30.0690.534.0003
 Gestational diabetes24.291.753.375.50.1590.580.008
Abbreviations: AUC, area under the receiver operating characteristic curve; NCEP ATP III, Third Report of the National Cholesterol Education Program Adult Treatment Panel; NPC, negative predictive value; PPV, positive predictive value. aP value for the comparison with the AUC when adiponectin is the only independent variable.

DISCUSSION

In this study, we demonstrated that compared with controls, a greater proportion of women with previous pregnancy-induced disturbances were insulin-resistant (defined as the top tertile of HOMA-IR). However, differences in the proportion having MS (or in the mean Framingham 10-year CHD risk score) were less striking and nonsignificant 2 years after delivery. Although adiponectin was strongly, inversely related to the prevalence of insulin resistance, MS, and the Framingham risk score, women with previous preeclampsia or gestational hypertension had similar adiponectin concentrations to controls, and a slight decrease in adiponectin levels was only observed among those with previous gestational diabetes. Nevertheless, adiponectin concentration remained a significant correlate of MS even within the category of previous pregnancy-induced disturbance, although adjustment for waist circumference reduced the significance of these findings. Furthermore, a concentration <10.5 µg/mL provided the optimal cut point to distinguish women with MS and was a better marker for the presence of metabolic abnormalities than prior preeclampsia, gestational hypertension, or gestational diabetes. Adiponectin level also correlated with the Framingham 10-year CHD risk score independent of race/ethnicity, BMI, and pregnancy-induced disturbances.

Pregnancy-induced hypertension, which includes both gestational hypertension and preeclampsia, is a common and serious pregnancy complication. Unfortunately, the pathophysiology of pregnancy-induced hypertension is still poorly understood.36 Emerging evidence suggests that insulin resistance, which has been linked to essential hypertension, may play a role in both forms of hypertension developing during pregnancy,37 although preeclampsia is more likely a systemic disease characterized not only by hypertension but also by increased vascular resistance, diffuse endothelial dysfunction, proteinuria and coagulopathy. Our study suggests that though preeclampsia may be linked with insulin resistance, a less clear association is present for gestational hypertension approximately 2 years postpartum. Unlike previous reports, however,38,39 we were unable to demonstrate increased prevalence of metabolic abnormalities in women with either prior preeclampsia or gestational hypertension postpartum, despite elevations in diastolic blood pressure and ACR in those with prior preeclampsia. It is possible that matching on prepregnancy BMI in the present study may have led to similar proportions of women with increased postpartum waist circumference (>88 cm) and, therefore, failure to detect a higher prevalence of MS among women with prior pregnancy-induced hypertension.

Studies on insulin resistance (insulin action) and insulin secretion (β-cell function) in normoglycemic women with a history of gestational diabetes have reached conflicting conclusions, reporting either defective insulin secretion with intact insulin action,40,41 intact insulin secretion with defective insulin action,42 or defective insulin secretion with a concomitant defect in insulin action only in an obese subgroup.43 Later reports using the frequently sampled intravenous glucose tolerance test have suggested defects in both insulin action and secretion44,45 that might be evident in the years following the gestational diabetes complicated pregnancy, even among women who return to prolonged normal glucose homeostasis postpartum.46 Conversely, it is generally agreed on that women with a history of gestational diabetes are at increased risk for MS postpartum.47–51 Our results are consistent with earlier studies reporting insulin resistance in women with prior gestational diabetes 1 to 3 years after the index delivery. Findings from the present study also concur that compared with normal controls and women with pregnancy-induced hypertensive disorders, those with prior gestational diabetes are more likely to have MS.

Insulin resistance is known to progressively increase during a normal pregnancy,3–5 and plasma adiponectin concentration, an adipokine inversely related to insulin resistance,14 glucose intolerance,15,16 and cardiovascular disease,15,17 has been shown to decline, most notably after the third trimester19,21,22 and reaching its lowest level during lactation.21 Although two reports have failed to show a decline in adiponectin concentration, researchers only measured levels of this adipokine during the third trimester of pregnancy and 4 months postpartum.18,20 Recently, it has been suggested that hypoadiponectinemia during lactation may be attributed to prolactin affecting maternal metabolism through regulation (suppression) of adiponectin.21

Thus, the evaluation of the hypothesis that the concentration of adiponectin in plasma is further disturbed in states of pregnancy-induced insulin resistance, such as gestational hypertension, preeclampsia, and gestational diabetes, came as a natural next step. Indeed, two reports demonstrated a >30% reduction in adiponectin (total or the high molecular weight form) concentrations during pregnancy among women with gestational diabetes compared with women with an unaffected pregnancy.19,26 However, a third report failed to show significant declines in adiponectin levels, despite a slight trend for the healthy control women to exhibit higher adiponectin concentrations.20 In the present study, we observed lower adiponectin levels and an adverse metabolic profile among women with previous gestational diabetes 2 years postpartum. Together, these results suggest that adiponectin may be an early marker for the development of type 2 diabetes or cardiovascular disease among women with gestational diabetes.

Results concerning the plasma adiponectin concentration in hypertensive disorders of pregnancy have been contradictory. Thus, although lower adiponectin levels were observed in women with preeclampsia compared with normal healthy pregnant women matched for gestational age in a study by Cortelazzi and associates,19 markedly elevated adiponectin concentrations have been reported by others.18,23,24 A possible explanation for these reported increases might be the extent of renal disease among cases of preeclampsia, as adiponectin levels are known to be elevated in macroalbuminuria.16,52 However, published studies have not provided information on differences adjusting for albuminuria. Two years postpartum, we did not observe a difference in the concentration of adiponectin between women with either gestational hypertension or preeclampsia and women with normal pregnancies even after adjustment for ACR. Nonetheless, ACR was only slightly elevated among women with a history of preeclampsia, and the difference did not reach statistical significance. Contrary to our results, Girouard and colleagues25 reported lower adiponectin concentrations in gestational hypertension and preeclampsia compared with controls almost 8 years postpartum. Differences in the populations studied (ie, women in the present study were younger, with a higher BMI and waist-to-hip ratio), as well as differences in the duration of follow-up from the index pregnancy (2 years in the present study vs 8 years in the study by Girouard and associates25) might contribute to this discrepancy. Nevertheless, similar to our findings, Girouard and associates25 also noted greater insulin resistance in women with prior pregnancy-induced hypertensive disorders. None of the other studies mentioned above assessed insulin resistance among persons with pregnancy-induced hypertensive disorders.

These results confirm prior findings linking insulin resistance and metabolic disturbances postpartum to a history of gestational diabetes and suggest that preeclampsia may also have an insulin resistance component. Gestational hypertension is less strongly associated with decreased insulin sensitivity. Adding to current knowledge, our data now suggest a lower adiponectin concentration postpartum, further identifying those at increased cardiovascular risk by virtue of having MS or by scoring higher in the Framingham CHD risk scale. This is true irrespective of pregnancy history. Moreover, plasma adiponectin correlated equally strongly with insulin resistance as with the presence of MS or the Framingham risk score, further underscoring the close association of these conditions. An adiponectin concentration <10.5 provides the optimal cut point to distinguish women with MS, while values above this threshold suggest low probability of MS, irrespective of pregnancy history. Adiponectin may therefore be helpful in further characterizing risk for women experiencing any of these 3 disturbances during pregnancy. Future studies hopefully will provide evidence on long-term outcomes (eg, incidence of type 2 diabetes and cardiovascular disease among this subgroup of women) and further address the issue of whether adiponectin is a helpful early marker for subsequent disease risk.

Acknowledgments:

This study was supported by the American Diabetes Association Mentor-Based Fellowship Program and partly by National Institutes of Health NIH-5MO1-RR00056 (Magee-Women's CRC) and NIH-2PO1-HD30367 (Preeclampsia Program Project). We thank the Magee-Women's Research Institute and PEPP staff, especially Ms. Jennifer M. Eicher for contributing to the patient recruitment and laboratory work. We also thank the study participants for their willing assistance.

Ancillary