- Top of page
Objective To investigate whether pre-eclampsia is associated with an exaggeration of the insulin resistance seen in normotensive pregnancy.
Design Minimal model analysis of a frequently sampled intravenous glucose tolerance test to assess insulin sensitivity.
Setting Royal Maternity Hospital, Belfast.
Participants Eleven women with pre-eclampsia and 11 matched normotensive pregnant women.
Results Insulin sensitivity (SI) was increased in the group with pre-eclampsia compared with the normotensive women (mean [±SEM]: 2.6 [0.4] vs 1.6 [0.2] 10−4 min−1 per mU/L; P= 0.028). This was accompanied by a decrease in glucose effectiveness (SG) (1*1 ±0.1 vs 1.7 ±0.1 10−2 mid, P= 0.006) in the pre-eclamptic women. In the normotensive group there was a significant inverse correlation between S, and mean arterial blood pressure (Y=−0.65; P= 0.03), but no such relation existed in the group with pre-eclampsia.
Conclusions As with other forms of secondary hypertension, and unlike essential hypertension, the pathophysiology of pre-eclampsia is not associated with insulin resistance.
Hypertensive disorders of pregnancy continue to be leading causes of maternal mortality. It is known that pre-eclampsia, the most common of these disorders, is associated with failure of trophoblast invasion of the spiral arteries of the placenta. However, other aspects of its pathophysiology are still poorly understood.
Elevated insulin levels in patients with hypertension were first reported in 19662, but it was not until the large epidemiological study of Modan et al.3 in 1985 that interest in this area began to grow. Since then, other clinical4–7 and metabolic8,9 studies have shown hyperinsulinaemia and insulin resistance to be linked with essential hypertension.
Pregnancy is itself an insulin-resistant state. The diabetogenic influence of pregnancy has been recognised for over 100 years10. As gestation advances, there is a progressive increase in the secretory response of the maternal pancreas to glucose, and the maintenance of euglycaemia is associated with progressively increasing levels of circulating insulin. Indirect evidence of insulin resistance in normal pregnancy is provided by the blunted response to an intravenous injection of exogenous insulin11. Direct evidence of insulin resistance in pregnancy has been provided by studies using both euglycaemic clamp12 and minimal model methodologies13,14.
To our knowledge there are no reported studies in which insulin sensitivity has been measured in women with pre-eclampsia. In the 1950s, Burt15 demonstrated that women with pregnancy-induced hypertension had higher glucose levels on intravenous glucose tolerance testing and an attenuated glucose response to intravenous insulin16. More recently, insulin concentrations were shown to be higher in response to an intravenous glucose load in women with hypertension in pregnancy than normotensive controls17. These studies suggest an association between pregnancy-induced hypertension and insulin resistance.
The aim of this study was to investigate whether pre-eclampsia is associated with an increase in insulin resistance over and above that seen in normotensive pregnancy. Insulin sensitivity was determined using the minimal model technique originally described by Bergman et al.18. We considered that the minimal model approach was better suited than the euglycaemic glucose clamp to investigating pregnant and potentially ill women. The experimental protocol is simpler and does not involve the infusion of insulin. There is a strong correlation between model-based and clamp-based insulin sensitivities19.
- Top of page
The study was approved by the Research Ethical Committee of the Queen's University of Belfast, and written informed consent was obtained prior to each intravenous glucose tolerance test.
The study group comprised 11 women in the third trimester of pregnancy who were admitted to the Royal Maternity Hospital, Belfast, because of pre-eclampsia. These were women with no previous history of hypertension or renal disease who were known to be normotensive during the earlier stages of this pregnancy. Inclusion criteria were proteinuria (> 0.3 g/L) on dipstick testing in at least two samples on the day before the study, and diastolic blood pressure ≥ 90 mmHg on at least three out of five readings taken using an automated blood pressure recorder (Dinamap, Critikon, Tampa, Florida, USA) at 15-minute intervals during a one-hour rest period on the day before the proposed study. Women taking anti-hypertensive or steroid therapy were excluded.
A control group of 11 normotensive pregnant women were recruited from the antenatal clinic at the same hospital. These were each paired with one of the study group using three criteria: age (±two years); body mass index (±10%); and gestation (±one week).
Insulin sensitivity was determined by minimal model analysis of the glucose and insulin concentrations obtained by frequent blood sampling during the three hours after an intravenous bolus of glucose. The plasma glucose and insulin dynamics were modelled by digital computer using the Simulation Analysis And Modelling (SAAM) program, and its interactive batch version CONSAM20. The modelling process determines values for parameters from which can be derived insulin sensitivity (SI), glucose effectiveness (SG) and pancreatic beta cell responsiveness (first and second phase; Ø1, and Ø2). SG is a measure of insulin-mediated glucose disposal and SG a measure of glucose-mediated glucose disposal.
The frequently sampled intravenous glucose tolerance test was performed following an overnight fast. Before the test an intravenous cannula was inserted into a vein in each arm. A slow-running saline infusion was connected to one cannula to facilitate venous sampling throughout the study. The other cannula was used to inject an intravenous bolus of 50% dextrose at a dose of 300 mg/kg over one minute, starting at time zero; 4 mL blood samples for glucose and insulin were taken at −20, −10, −5, 2, 3, 4, 5, 6, 8, 10, 12, 14, 16, 19, 22, 25, 30, 40, 50, 60, 70, 80, 90, 100, 120, 140, 160 and 180 minutes.
Samples for plasma glucose were analysed using a glucorotor immobilised glucose oxidase membrane and oxygen electrode. The insulin samples were kept at 4°C during the study, and then transferred to the endocrine laboratory for separation of the plasma which was stored at −20°C. Insulin levels were determined using a radioimmunoassay with polyclonal antiserum.
The glucose and insulin values were entered into a program of the minimal model on an IBM-compatible personal computer. CONSAM was used to perform manual fitting of curves to the data by appropriate adjustment of the variable parameters. Subsequently, CONSAM and then SAAM were used to perform least sum of squares iterations to achieve the closest possible fits and determine the final parameter values, as described by Martin et al.21. Values for SI, SG, Ø1 and Ø2 were then derived from these parameters.
The intravenous glucose clearance rate (KG) was determined as the least square slope of the natural logarithm of the glucose concentrations between 12 and 30 minutes after the glucose bolus. Acute insulin response to intravenous glucose (AIRglucose) was calculated as the mean of the incremental plasma insulin concentrations from 0–10 minutes following the intravenous glucose bolus22. Insulin-mediated, glucose-mediated and total glucose disposal during the glucose tolerance test were calculated as described by Henriksen et al.23. Insulin-mediated glucose disposal was calculated as the product of the insulin sensitivity index SI and AIRglucose (SI× AIRglucose); this calculation has the unit min−1 in common with glucose-mediated glucose disposal SG. Total glucose disposal during the tolerance test was calculated as the sum of insulin-mediated and glucose-mediated glucose disposal.
The data obtained from the glucose tolerance tests were found to be normally distributed and were analysed by parametric statistics. The derived measurements of glucose metabolism of the women in the matched study and control groups were compared using Student's t test. Pearson's correlation analysis was used to examine the relation between mean arterial blood pressure and both fasting insulin and insulin sensitivity.
- Top of page
The success of matching the study (those with pre-eclampsia) and control (normotensive) groups of women for factors likely to influence insulin sensitivity was checked retrospectively by comparing age, gestation, body mass index (BMI) and percentage of fat; no significant differences were noted (Table 1). There were also no significant differences in mean lipid, progesterone, prolactin or androgen concentrations between the two groups (Table 1). Mean [±SEM] arterial blood pressure at the time of the intravenous glucose tolerance test was 116/66 [3/2] mmHg in the control group and 153/94 [5/2] mmHg in the study group. The women with pre-eclampsia had mild disease as none had a significant disturbance of liver transaminases or coagulation. Three of the 11 women had a platelet count < 150 × 109/L.
Table 1. Comparison of demographic features and biochemical factors which might influence insulin sensitivity between women with pre-eclampsia (PE) and those in the normotensive control group. Values are given as mean [SEM]. BMI = body mass index.
|Age (years)||26.4 [1.91||28.2 [2.0]|
|BMI (kg/m2)||28.8 [1.21||29.2 [0.9]|
|Percentage fat||30.2 [0.9]||29.5 [1.0]|
|Gestational age (weeks)||36.4 [0.5]||35.9 |
|Cortisol (nmollL)||871 ||734 |
|Progesterone (nmoYL)||670 ||605 [1041|
|Testosterone (nmoYL)||3.3 [0.4]||4.1 [0.6]|
|Cholesterol (mmoVL)||6.3 [0.3]||6.3 [0.4]|
|Triglyceride (mmol/L)||2.7 [0.3]||3.0 [0.3]|
Mean [±SEM] fasting glucose levels were not significantly different in the control and study groups (3.9 [0.11 vs 3.6 [0.2] mmol/L; P = 0.21). However, mean fasting insulin was significantly lower in the women with pre-eclampsia compared with the control group (10.1 [1.2] vs 16.6 [2.1] mU/L, P = 0.016). Median plasma glucose and insulin levels during the course of the intravenous glucose tolerance test are shown in Figs 1 and 2. There was virtually no difference between the glucose decay curves (Fig. l), and there was no significant difference between the glucose clearance rate K, in the two groups (pre-eclamptic vs normotensive: 1.86 [0.21] vs 2.02 [0.16]; P= 0.56). In contrast, there was a small but prolonged decrease in the insulin response of the women with pre-eclampsia compared with the women in the normotensive group (Fig. 2).
Figure 2. The profile of median insulin concentrations during the intravenous glucose tolerance tests in the pre-eclamptic (•) and normotensive (▵) pregnant women.
Download figure to PowerPoint
Table 2 compares the derived parameters in the two groups of women. Mean [±SEM] insulin sensitivity was significantly higher in women with pre-eclampsia than in the controls (P = 0.028). This was accompanied by a significant decrease in mean glucose effectiveness in the pre-eclamptic group (P= 0.006). There was no significant difference in the first or second phase insulin response, or in AIRglucose(Table 2).
Table 2. Comparison of mean glucose effectiveness (SG), insulin sensitivity (SI), first and second phase insulin response (Ø1, and Ø2), and acute insulin response to glucose (AIRglucose) between the control group and the group with pre-eclampsia (PE). Values are given as mean [SEM].
|SG (10−2 min−1)||1.7 [0.1]||1.1 [0.1]||0.006|
|S1 (10−4 min−1 per mU/L)||1.6 [0.2]||2.6 [0.4]||0.028|
|Ø1 (mU/L. min−1 permg/dL)||15.8 [1.8]||14.2 [3.1]||0.65|
|Ø1, (mU/L. min−2 per mg/dL)||57.4 [20.6]||54.4 [21.0]||0.92|
|AIRglucose(mUW||139.5 [15.7]||138.5 [29.4]||0.98|
Total glucose disposal correlated with the intravenous glucose clearance K, in the combined group of women (r = 0.72, P= 0.001). The relative contributions of insulin and glucose-mediated glucose disposal to total glucose disposal are shown in Table 3. In the group with pre-eclampsia the relative contribution of glucose-mediated glucose disposal to total glucose disposal was significantly lower (29.1% [3–0] vs 47.2% [5.0]; P= 0.005) compared with the women in the control group. The mean arterial blood pressure (MAP) in the normotensive women was significantly correlated with both fasting insulin (r = 0.68, P= 0.02) (Fig. 3) and S, (r =−0.65, P= 0.03) (Fig. 4). No such correlations were found with blood pressure in the women with pre-eclampsia (MAP vs fasting insulin r =−0–21, P= 0.55; MAP vs S, r = 0.24, P = 0.47).
Table 3. Comparison of glucose disposal between the control group and those women with pre-eclampsia (PE). Values are given as mean [SEM].
|Total glucose disposal 10−2 min−1||3.78 [0.27]||4.14 [0.41]|
|Insulin mediated glucose disposal min−1||2.09 [0.31]||3.00 [0.38]|
|Proportion of total glucose disposal (%)||53 ||71 |
|Glucose mediated glucose disposal 10−2 mid||1.69 [0.13]||1.14 [0.12]|
|Proportion of total glucose disposal (%)||47 151||29 |
Figure 3. Correlation between mean arterial blood pressure and fasting insulin in the normotensive (▵) and pre-eclamptic (•) women. Normotensive: r= 0.68, P= 0.02; pre-eclamptic: r=−0.21, P= 0.55.
Download figure to PowerPoint
Figure 4. Correlation between mean arterial blood pressure and insulin sensitivity in the normotensive (▵) and pre-eclamptic (•) women. Normotensive: r= 46.5, P= 0.03; pre-eclamptic: r= 0.24, P= 0.41.
Download figure to PowerPoint
- Top of page
Hyperinsulinaemia and insulin resistance are closely associated with essential hypertension. No such relation is apparent with secondary forms of hypertension such as that due to renal failure or impaired renal blood flow24. Furthermore, lowering of blood pressure with most antihypertensive agents does not lead to an increase in insulin sensitivity25. It thus appears that hypertension per se does not cause insulin resistance. Postulated mechanisms by which hyperinsulinaemia and insulin resistance could lead to hypertension are an associated increase in sympathetic nervous system activity or sensitivity to catecholamines, increased sodium reabsorption in the proximal renal tubule, or altered cellular cation transport.
We have previously used minimal model analysis of the intravenous glucose tolerance test to study glucose metabolism in the proliferative menstrual phase in a group of 22 nonpregnant women with regular menstrual cycles (unpublished data). Mean [±SEM] SI was 6.4 [0.78] 104 min−1per mU/L, which is significantly higher than that in either the normotensive or pre-eclamptic group of pregnant women (P < 0.01), thus confirming using this methodology the insulin resistance of pregnancy reported by other authors12–14. Mean fasting insulin concentration (5.1 [0.5] mU/L) and mean first and second phase insulin response (Ø1:3.0 [0.3] mU/L. min−1 per mg/dL, Ø2:10.7 [3.3] mU/L. min−1 per mg/dL) were all significantly lower in nonpregnant women than in either group of pregnant women (P < 0.01). Mean SG in our nonpregnant women was 1.6 [0.1] 10−2 min−1 which is not significantly different to the women in this study who had normotensive pregnancies (1.7 [0.1] 10−2 min−1). In normal pregnancy glucose tolerance appears to be maintained by a compensatory increase in insulin secretion, which counteracts the marked reduction in insulin sensitivity.
The present study has shown that, contrary to our hypothesis, pre-eclampsia is not associated with an increase in insulin resistance beyond that of normal pregnancy; indeed the reverse is true. The relatively increased insulin sensitivity is also reflected in a lower fasting insulin concentration in the women with pre-eclampsia compared with pregnant controls. However, the glucose clearance rate was very similar in the two groups, as was total glucose disposal during the intravenous glucose tolerance test. The total glucose disposal rate in both groups was very close to that previously reported (3.63 [0.33] 10−2 mi d) in a population of young healthy non-pregnant women20. This indicates that neither group of pregnant women had any perturbation of glucose tolerance. In the normotensive pregnant women the relative contribution of insulin-mediated and glucose-mediated glucose disposal to total glucose disposal was also very close to that previously reported in healthy nonpregnant women (58  and 42  10−2 min−1, respectively)23. However, in the women with pre-eclampsia, glucose disposal was predominantly via the insulin-mediated pathway, with a reciprocal reduction in glucose-mediated glucose disposal. It is impossible to know whether the change in insulin-mediated or glucose-mediated glucose disposal is the prime mover in pre-eclampsia, but it would appear that they compensate for each other in maintaining euglycaemia.
A possible criticism of our study is the fact that the women were matched by their body mass index at the time of the intravenous glucose tolerance test rather than their body mass index before they became pregnant. This was done because pre-pregnancy weights were not known. It would be expected for the women with pre-eclampsia to have had more oedema fluid and therefore to have been leaner than their matched normotensive counterparts. We investigated this potential source of bias by comparing body mass indices at matched gestations between 20 and 28 weeks, before pre-eclampsia was manifest. Such data were available for nine of the eleven pairs of women. Six of the nine pairs remained matched to within 10% of their body mass index. The mean body mass index of the nine women with pre-eclampsia was 24.2 kg/m2 compared with 24.4 kg/m2 for the normotensive group (P = 0.84). It thus seems unlikely that matching at the time of the intravenous glucose tolerance test has introduced a significant bias.
If we accept that insulin resistance plays a role in the pathophysiology of essential hypertension, and that the transition from normal blood pressure to elevated blood pressure is a continuum, the inverse relation between mean arterial blood pressure and insulin sensitivity seen in normotensive pregnancies is not unexpected. In contrast to this, our findings of reduced insulin resistance during pre-eclampsia, together with the lack of correlation between mean arterial blood pressure and insulin sensitivity, indicate that insulin resistance is not involved in the pathophysiology of raised blood pressure in pre-eclampsia. This appears to contradict the earlier literature on glucose and insulin metabolism in pregnancy-induced hypertension15–17. However, in addition to the fact that these earlier studies did not measure insulin sensitivity, they did not specifically exclude women with pre-existing hypertension and did not distinguish between proteinuric and non-pro teinuric hypertension.
It has been suggested that hypertension occurring for the first time in pregnancy can be divided into two aetiological groups: pre-eclampsia and gestational hypertension26. Pre-eclampsia or proteinuric pregnancy-induced hypertension, appears to be an immunologically mediated condition and is seen most commonly in young women in their first pregnancy. Gestational hypertension is not associated with proteinuria and tends to occur in older multi-parae. There is evidence that it is latent essential hypertension which is unmasked by the pregnancy in much the same way as pregnancy unmasks a tendency to type 2 diabetes in the form of gestational diabetes27. As essential hypertension is an insulin-resistant condition, this raises the question of whether, in contrast to pre-eclampsia, gestational hypertension is associated with an increase in insulin resistance. Indirect evidence to support this hypothesis is provided by Solomon et al.27 who found a significantly higher incidence of elevated glucose concentrations one hour after a 50 g oral glucose load in women with gestational hypertension, but not pre-eclampsia, when compared with normotensive pregnant women.
We conclude from our study that, as with other forms of secondary hypertension, pre-eclampsia is not associated with insulin resistance. We would postulate that the relative lack of insulin resistance in pre-eclampsia compared with normal pregnancy may represent a failure of the usual metabolic adaptation to pregnancy due to abnormal placentation.
The authors would like to thank Mr B. Sheridan and the staff of the Regional Endocrinology Laboratory at the Royal Victoria Hospital for performing the insulin assays, Dr A. I. Traub for his support, and the consultant obstetricians at Royal Maternity Hospital Belfast for access to their patients. This study was supported by a Royal Group of Hospitals Research Fellowship and a grant from the British Diabetic Association.