Haptoglobin Phenotype, Preeclampsia and Response to Vitamin C and E Supplementation in Pregnant Women with Type 1 Diabetes

Abstract

Objective

The phenotype of the anti-oxidant and pro-angiogenic protein haptoglobin (Hp) predicts cardiovascular disease risk and treatment response to antioxidant vitamins in individuals with diabetes. Our objective was to determine whether the Hp phenotype influences preeclampsia risk, or the efficacy of vitamins C and E in preventing preeclampsia, in women with Type 1 diabetes.

Design

This is a secondary analysis of a randomized controlled trial in which women with diabetes received daily vitamins C and E, or placebo, from 8–22 weeks gestation until delivery.

Setting

25 antenatal metabolic clinics across the UK (northwest England, Scotland, and Northern Ireland).

Population

Pregnant women with Type 1 diabetes.

Methods

Hp phenotype was determined in white women who completed the study, and had plasma samples available (n=685). \

Main Outcome Measures

Preeclampsia.

Results

Compared to Hp 2-1, Hp 1-1 (odds ratio: 0.59 (95% confidence interval: 0.30–1.16)) and Hp 2-2 (0.93 (0.60–1.45)) were not associated with significantly decreased preeclampsia risk after adjusting for treatment group and HbA1c at randomization. Our study was not powered to detect an interaction between Hp phenotype and treatment response. However, our preliminary analysis suggests that vitamins C and E did not prevent preeclampsia in women of any Hp phenotype (Hp 1-1: 0.77 (0.22–2.71); Hp 2-1: 0.81 (0.46–1.43); Hp 2-2: 0.67 (0.34–1.33)) after adjusting for HbA1c at randomization.

Conclusions

Hp phenotype did not significantly affect preeclampsia risk in women with Type 1 diabetes.

Introduction

Preeclampsia affects 15–18% of women with Type 1 diabetes,^1,2 leading to increased maternal³ and fetal⁴ morbidity and mortality. Oxidative stress is associated with preeclampsia,^5,6 and the antioxidant vitamins C and E lowered preeclampsia incidence by 60% among high-risk women in a small randomized controlled trial (RCT).⁷ Unfortunately, subsequent RCTs in high^8–12 and low¹³ risk women, and women with Type 1 diabetes¹⁴ were negative. While these divergent results likely reflect low power in the small trial, differences could also be explained by the greater diversity of patients in multicenter trials masking a subset of responsive women.

Haptoglobin (Hp) is an anti-oxidant¹⁵ and pro-angiogenic¹⁶ protein with three generically-determined phenotypes (1-1, 2-1, 2-2).¹⁷ Hp 1-1 is the strongest antioxidant.¹⁵ Hp 2-2 is the most angiogenic.¹⁶ Little is known about the function of Hp 2-1, which is structurally distinct from Hp 1-1 and 2-2.¹⁸ Hp phenotype predicts cardiovascular risk^19,20 and responsiveness to vitamin E^21–23 or C and E²⁴ in individuals with diabetes.

We examined whether Hp phenotype might affect preeclampsia risk, or identify women with Type 1 diabetes who would respond to vitamin supplementation, for three reasons. First, angiogenic imbalance and oxidative stress contribute to preeclampsia,^5,6 and Hp’s pro-angiogenic and antioxidant properties are phenotype-dependant.^16–18 Hp 2-1 was associated with a two-fold greater preeclampsia risk among white women without diabetes in our recent study.²⁵ Second, all three phenotypes are common in whites (17%–48%); therefore any effect would affect a large proportion of women.²⁶ Third, Hp phenotype influences cardiovascular risk, and responsiveness to vitamin E^21–23 or C and E²⁴ in individuals with diabetes. Cardiovascular event risk is doubled in Hp 2-2 individuals with diabetes, compared to Hp 1-1 and 2-1 individuals with diabetes,^21,24,27 and vitamin E eliminates this increased risk.^21,22,28 In contrast, vitamin C combined with vitamin E is either beneficial or harmful depending on phenotype.²⁴ In post-menopausal women with coronary artery disease, vitamin C and E decreased coronary artery diameter in Hp 2-2 women with diabetes, but benefited Hp 1-1 women by increasing coronary artery diameter.²⁴

We performed a secondary analysis of an RCT¹⁴ of antioxidants to prevent preeclampsia in women with Type 1 diabetes to determine whether Hp phenotype is associated with preeclampsia risk, or antioxidant response, in these women. We hypothesized that Hp 2-1 would be associated with increased preeclampsia risk, in accordance with our previous data from women without diabetes. We also posited that phenotype would affect treatment response, although the nature of the effect was difficult to predict using data from non-pregnant individuals.

Methods

Study Population

This was a secondary analysis of a multicenter RCT (ISRCTN 27214045) in which 762 women with Type 1 diabetes received 1000 mg of vitamin C and 400 IU of vitamin E, or placebo, daily from 8–22 weeks gestation until delivery. The trial was conducted from 2003 to 2008 in 25 antenatal metabolic clinics across the UK (northwest England, Scotland, and Northern Ireland). Full details have been reported previously.¹⁴ The West Midlands multicenter research ethics committee approved the study (MREC 02/7/016). All subjects provided written, informed consent prior to participating. Hp phenotypes were determined at the University of Pittsburgh (Institutional Review Board exempt approval PRO10090150; retrospective analysis of samples in existence).

The distribution of Hp phenotypes depends upon race.²⁶ 96.5% of women in the DAPIT study were white (n = 735), 0.9% were black (n = 7), 1.4% were asian (n = 11) and 1.2% were of other or unknown race (n = 9). Sample sizes were too small to draw conclusions about non-whites; therefore these women were excluded from the analysis. Fifty white women were excluded because plasma samples were not available. Hp phenotype was determined in the remaining 685 white women. Ten women who experienced fetal loss before 20 weeks gestation were excluded from analyses examining the relationship between Hp phenotype, preeclampsia risk, and vitamin supplementation.

The primary outcome for the trial was preeclampsia, defined as gestational hypertension and proteinuria according to the International Society for the Study of Hypertension in Pregnancy Guidelines at the time of the trial.²⁹ Gestational hypertension was defined as two diastolic blood pressure readings ≥90 mmHg separated by at least 4 hours, or a single diastolic blood pressure reading ≥110 mmHg, between 20 weeks gestation and 48 hours post-partum, excluding labor. Proteinuria was defined as a dipstick reading ≥1+ on at least 2 occasions, or a 24-hour urinary protein ≥300 mg. Among women with pre-existing proteinuria, preeclampsia was diagnosed if women had dipstick readings of ≥2+, 24-hour urinary protein ≥600 mg, or one or more other features of preeclampsia such as those defining the HELLP syndrome or eclampsia. Secondary outcomes for this analysis were severe preeclampsia and early onset preeclampsia. Severe preeclampsia was defined as preeclampsia accompanied by one or more of the following: diastolic blood pressure ≥ 110 mmHg, the highest urine protein dipstick reading ≥3+, urine output <500 ml/24h, grand mal seizures, blurred vision, headache, pulmonary edema, HELLP syndrome, or epigastric pain. Data for early onset preeclampsia are presented as preeclampsia with delivery before 34, or 37 weeks gestation.

Hp Phenotyping

Hp phenotype was determined as described previously.²⁵ Native polyacrilamide gel electrophoresis (PAGE) was performed on 5 μl of citrated plasma supplemented with 3 μl of 25 μM human hemoglobin (Sigma-Aldrich, St. Louis, MO). Samples were run on 6% tris-glycine gels (Invitrogen, Carlsbad, CA) for 2 hours at 120 volts; then transferred to a polyvinylidene fluoride (PVDF) membrane (Millipore, Billerica, MA). Peroxidase activity of the hemoglobin/Hp complex was used to visualize Hp phenotype.

Samples that were hemolyzed or had low Hp concentrations were phenotyped by SDS PAGE.²⁵ 2 μl of β-mercaptoethanol was added to 1 to 6 μl of serum. After heating for 7 minutes at 82°C, samples were run on 15% tris-glycine gels at 120 volts for 1.75 hours. Proteins were transferred to PVDF. Membranes were incubated with blocking solution (tris-buffered saline containing 5% non-fat milk, 0.1% Tween 20), primary antibody (1:5,000, Polyclonal Rabbit Anti-Human Haptogloin, DakoCytomation, Carpinteria, CA) and secondary antibody (1:25,000, Goat anti-Rabbit IgG horseradish peroxidase, Millipore) at room temperature for one hour each. Antibodies were dissolved in blocking solution, and membranes were washed in tris-buffered saline containing 0.1% Tween 20 between incubations.

Membranes from native and SDS PAGE were stained for peroxidase activity (SuperSignal West Pico Chemiluminescent Substrate, Fisher Scientific, Pittsburgh, PA) and imaged (FlouroChem Q System, Cell Biosciences, Santa Clara, CA). Hp phenotypes were identified by characteristic banding patters (Figure 1).

Figure 1. Hp phenotypes by Native and SDS PAGE.

Panel A: Hp phenotyping of Hb-supplemented plasma by Native PAGE on a 6% gel using Hb peroxidase detection. Hp 1-1 has a single fast migrating band (Lane 2). Hp 2-2 has a series of slowly migrating bands (Lane 6). Hp 2-1 (Lanes 1, 3–5) has one band between the Hb/Hp 1-1 and Hb/Hp 2-2 bands, and several slowly migrating bands in the same region as the Hb/Hp 2-2 bands.

Panel B: Hp phenotyping of plasma by SDS PAGE on a 12% gel with Western blotting. All phenotypes have the Hp β band. The α₁ band indicates that the Hp 1 allele is present. The α₂ band indicates that the Hp 2 allele is present. Hp 2-2 has α₂ and β bands (Lane 1), Hp 2-1 has α₁, α₂ and β bands (Lane 2), and Hp 1-1 has α₁ and β bands (Lane 3).

Statistical Analysis

Quantitative variables were summarized using means and standard deviations unless distributions were heavily skewed in which case medians and interquartile ranges were used. Comparisons of characteristics between Hp phenotype groups were performed using one-way analysis of variance, Kruskal-Wallis analysis of variance of ranks or the χ² test. Logistic regression analysis was used to compare the risks of preeclampsia in the phenotype groups. The 2-1 phenotype was selected as the reference category, as this was the phenotype with the largest sample size. Logistic regression was done both before and after adjustment for the potential confounders treatment group (vitamins or placebo) and HbA1c category at randomization (≤6.0%, 6.0–6.9%, 7.0–7.9%, ≥8.0% or unknown). Logistic regression was also used to check if the effect of vitamin supplementation on preeclampsia risk differed between the three Hp phenotype groups by adding the interaction between Hp phenotype and treatment group to the model.

Results

Subject Characteristics

The prevalence of the Hp 1-1 (14.7%), 2-1 (48.7%) and 2-2 (36.6%) phenotypes were similar to previously reported values for Caucasians.²⁶ Women with the Hp 2-1 phenotype were randomized slightly earlier than women with the Hp 1-1 or 2-2 phenotypes, and this small difference was the only statistically significant difference (P<0.05) found in 19 comparisons (Table 1). Hp phenotype groups did not differ with respect to maternal demographic characteristics (age, parity, education), physical characteristics (BMI, blood pressure, and albumin-creatinine ratio at randomization), health behaviors (smoking, multivitamin or aspirin consumption prior to randomization), diabetes history (duration, HbA1c and insulin dose at randomization) or hypertension (hypertension or antihypertensive treatment before pregnancy, previous preeclampsia).

Table 1.

Maternal demographic characteristics

Variable	Hp 1-1 (n = 99)	Hp 2-1 (n = 335)	Hp 2-2 (n = 251)	p
Age (years)a	29.7 ± 5.7	29.6 ± 5.5	29.8 ± 5.6	0.96
Gestational age at randomization (weeks)a	14.2 ± 3.3	13.7 ± 3.3	14.4 ± 3.6	0.05
Body mass index at randomization (kg/m2)a	28.0 ± 5.2	27.8 ± 5.5	27.1 ± 4.2	0.15
Current smokerc	21 (21%)	75 (22%)	40 (16%)	0.14
Primiparousc	44 (44%)	170 (51%)	118 (47%)	0.46
12 years or fewer of full time educationc	36 (36%)	132 (39%)	96 (38%)	0.85
Taking multivitamin at randomizationc	12 (12%)	33 (10%)	26 (10%)	0.81
Taking aspirin at randomizationc	5 (5%)	23 (7%)	18 (7%)	0.77
Antihypertensive treatment before pregnancyc	7 (7%)	28 (8%)	24 (10%)	0.74
Hypertension before pregnancyc	18 (18%)	49 (15%)	37 (15%)	0.68
Blood pressure at entry (8–22 weeks; mmHg)
Systolica	120.1 ± 12.3	118.0 ± 11.6	119.9 ± 12.1	0.11
Diastolica	76.1 ± 8.8	74.1 ± 8.1	75.1 ± 9.0	0.09
Preeclampsia in previous pregnancyc	14 (14%)	40 (12%)	22 (9%)	0.28
Diabetes
Duration (years)a	14.6 ± 7.9	14.1 ± 8.0	15.1 ± 8.5	0.38
HbA1c at randomizationa	7.3 ± 1.1	7.2 ± 1.0	7.1 ± 0.9	0.47
Insulin at randomization (units/day)b	60 (43–74)	58 (42–74)	54 (44–68)	0.24
Renal status before index pregnancy b
Normal	87 (88%)	307 (92%)	230 (92%)	0.56
Microalbuminuria	7 (7%)	14 (4%)	11 (4%)
Macroalbuminuria	1 (1%)	3 (1%)	0 (0%)
Urinary protein >3g/24 hrs	0 (0%)	2 (1%)	2 (1%)
Not known	4 (4%)	9 (3%)	8 (3%)
Albumin:creatinine at randomization (mg/mmol)b	0.81 (0.46–1.98)	0.69 (0.36–1.47)	0.70 (0.41–1.53)	0.20

Values are mean ± SD, median (interquartile range) or n (%). Abbreviations: hr, hour.

^aOne way ANOVA;

^bKruskal-Wallis ANOVA of ranks;

^cχ² test.

Hp Phenotype and Preeclampsia Risk

There was no significant difference in the risk of preeclampsia, severe preeclampsia, or early onset preeclampsia between the three phenotype groups (Table 2). Compared to Hp 2-1, Hp 1-1 and 2-2 were not associated with significantly decreased risk or preeclampsia, severe preeclampsia, or early onset preeclampsia (Table 3). Adjustment for treatment group (vitamins C and E vs. placebo) and HbA1c category at randomization had a minimal effect on odds ratios.

Table 2.

Preeclampsia Incidence According to Hp Phenotype

Outcome	Hp 1-1 (n = 99)	Hp 2-1 (n = 329)	Hp 2-2 (n = 247)	p
Preeclampsia	12 (12.1%)	59 (17.9%)	42 (17.0%)	0.39
Severe Preeclampsia	10 (10.1%)	53 (16.1%)	33 (13.4%)	0.29
Early Onset Preeclampsia
<34 weeks	2 (2.0%)	12 (3.6%)	6 (2.4%)	0.58
<37 weeks	10 (10.1%)	41 (12.5%)	24 (9.7%)	0.55

Values are n (%). Groups were compared using the χ² test.

Table 3.

Odds ratios for preeclampsia

Outcome	Phenotype	n (% within group)		OR (95% CI)a	Adjusted OR (95% CI)a,b
		Preeclampsia	Controls
Preeclampsia	1-1	12 (10.6%)	87 (15.5%)	0.63 (0.32, 1.23)	0.59 (0.30, 1.16)
	2-1	59 (52.2%)	270 (48.0%)	1	1
	2-2	42 (37.2%)	205 (36.5%)	0.94 (0.61, 1.45)	0.93 (0.60, 1.45)
Severe Preeclampsia	1-1	10 (10.4%)	87 (15.5%)	0.59 (0.29, 1.20)	0.55 (0.27, 1.13)
	2-1	53 (55.2%)	270 (48.0%)	1	1
	2-2	33 (34.4%)	205 (36.5%)	0.82 (0.51, 1.31)	0.81 (0.50, 1.31)
Early Onset Preeclampsia (<34 weeks)	1-1	2 (10.0%)	87 (15.5%)	0.52 (0.11, 2.36)	0.47 (0.10, 2.15)
	2-1	12 (60.0%)	270 (48.0%)	1	1
	2-2	6 (30.0%)	205 (36.5%)	0.66 (0.24, 1.78)	0.68 (0.25, 1.85)
Early Onset Preeclampsia (<37 weeks)	1-1	10 (13.3%)	87 (15.5%)	0.76 (0.36, 1.57)	0.72 (0.35, 1.51)
	2-1	41 (54.7%)	270 (48.0%)	1	1
	2-2	24 (32.0%)	205 (36.5%)	0.77 (0.45, 1.32)	0.77 (0.45, 1.32)

Statistical analysis was performed by logistic regression.

^aHp 2-1 was used as the reference group for all comparisons

^bAdjusted for treatment group and HbA1c at randomization by category (≤6.0%, 6.0–6.9%, 7.0–7.9%, ≥8.0% and unknown).

Hp Phenotype and Treatment Response

There was no significant interaction between the effect of Hp phenotype and the effect of vitamin supplementation on preeclampsia risk (p=0.87). Odds ratios for the development of preeclampsia with vitamin supplementation were 0.67 (95% confidence interval: 0.15, 2.67) in Hp 1-1, 0.82 (0.45, 1.51) in Hp 2-1, and 0.66 (0.32, 1.36) in Hp 2-2 (Table 4). Vitamin C and E supplementation did not significantly reduce preeclampsia risk in white women with Type 1 diabetes of any Hp phenotype.

Table 4.

Incidence of preeclampsia among women in the placebo and treatment groups

Phenotype	Treatment	Placebo	OR (95% CI)	p	Adjusted OR (95% CI)a	p
1-1	5/50 (10.0%)	7/49 (14.3%)	0.67 (0.20,2.26)	0.52	0.77 (0.22, 2.71)	0.69
2-1	26/158 (16.5%)	33/171 (19.3%)	0.82 (0.47, 1.45)	0.50	0.81 (0.46, 1.43)	0.46
2-2	18/127 (14.2%)	24/120 (20.0%)	0.66 (0.34, 1.29)	0.23	0.67 (0.34, 1.33)	0.25

Statistical analysis was performed by logistic regression.

^aAdjusted for HbA1c at randomization by category (≤6.0%, 6.0–6.9%, 7.0–7.9%, ≥8.0% and unknown).

Discussion

This secondary analysis of an RCT¹⁴ of daily vitamin C and E to prevent preeclampsia reveals two important findings. First, Hp phenotype was not associated with preeclampsia risk in white women with Type 1 diabetes. Second, although our study was not powered to detect an interaction between Hp phenotype and treatment, we found no evidence that vitamins C and E significantly affect preeclampsia risk in women of any Hp phenotype.

Hp Phenotype and Preeclampsia Risk

When we began this investigation, two small, underpowered case-control studies in women without diabetes had reported that preeclampsia risk was either increased³⁰ or not different³¹ in Hp 1-1 women. Cohort studies, adequately powered studies, and studies in women with diabetes were needed. A subsequent Israeli study in women without diabetes reported a lower preeclampsia risk among Hp 1-1 women, compared to Hp 2-1 and 2-2 women (5.8% vs. 12.5%).³² Our larger case-control study indicated that Hp 1-1 was protective in comparison to Hp 2-1, but not Hp 2-2, in white women without diabetes.²⁵ Racial differences between populations may have contributed to these divergent results in women without diabetes. However, the small samples sizes of early studies suggest that spurious findings may also have been a factor.

The present study extends the existing literature to include women with Type 1 diabetes. This is particularly important, as the impact of Hp phenotype on cardiovascular disease risk^19,20 and responsiveness to antioxidant vitamins^21–24 in non-pregnant populations is primarily confined to individuals with diabetes. Hp phenotype was not associated with the risk of preeclampsia, severe preeclampsia, or early onset preeclampsia in white women with Type 1 diabetes in the present study. Rates of preeclampsia tended to be lower in Hp 1-1 women, however this effect was clearly not significant. Although the sample size for this secondary analysis was limited by the number of participants in the RCT, there were sufficient subjects to detect a 14% difference in preeclampsia risk between Hp 1-1 and 2-1 (12% vs. 26%; 80% power; α=0.05). The study was not powered to detect smaller differences. Despite this limitation, this is the only study which we are aware of to examine women with Type 1 diabetes. The cohort design is also a significant strength, as all studies in women without diabetes have used a case-control design.

Hp Phenotype and Vitamin C and E

Emerging evidence suggests that there is a strong relationship between preeclampsia and future cardiovascular disease. Preeclampsia and cardiovascular disease share many risk factors (i.e. obesity, diabetes, black race, hyperlipidemia) and underlying pathophysiological processes (endothelial dysfunction, angiogenic imbalance, inflammation, and oxidative stress).^33,34 Preeclampsia is now recognized as a cardiovascular disease risk factor by the American Heart Association.³⁴ Hp phenotype affects both cardiovascular disease risk and treatment response to antioxidant vitamins in individuals with diabetes.^21,22,24,27 Therefore, we examined the potential for Hp to modify the efficacy of vitamins C and E in preventing preeclampsia in women with Type 1 diabetes.

Our study did not have sufficient power to detect an interaction between Hp phenotype and treatment response. However, our preliminary analysis provided no evidence that vitamin C and E supplementation prevented preeclampsia among white women with Type 1 diabetes of any Hp phenotype. It is extremely unlikely that larger studies will ever be conducted. Several factors could contribute to the difference between our study and previous studies of cardiovascular disease risk in individuals with diabetes.^21,22,24,27 First, although preeclampsia and cardiovascular disease share many risk factors and pathophysiological processes, there are important differences between these two conditions. Preeclampsia is a pregnancy-specific syndrome, in which one or more factors released by the placenta are believed to contribute to maternal vascular dysfunction.³³ Factors that are only produced by the placenta, or are released from the placenta in larger amounts than from other tissues, are unlikely to play a major role in cardiovascular disease.

Second, cardiovascular disease studies suggest that the effects of vitamin E alone differ from the effects of combined vitamin C and E supplementation.^21,22,28 Among post-menopausal women with coronary artery disease, vitamins C and E may be beneficial or harmful depending on Hp phenotype.²⁴ Supplementation increased coronary artery diameter in Hp 1-1 women, and this beneficial effect was stronger in Hp 1-1 women with diabetes.²⁴ In Hp 2-2 women with diabetes, however, the weak antioxidant capacity of Hp 2-2 may have interacted with and the pro-oxidant effects of vitamin C³⁵ to accelerate coronary artery narrowing.²⁴ Both Type 1 diabetes³⁶ and pregnancy³³ increase oxidative stress, suggesting that a similar mechanism could potentially be active in our study population. However, we did not find evidence to support this hypothesis.

Third, previous studies examining the interaction between Hp phenotype and responsiveness to antioxidant vitamins have focused on Type 2 diabetes.^22,27 Women in the present study had Type 1 diabetes. Hp 2-2 increases cardiovascular disease risk in both Type 1¹⁹ and Type 2²⁰ diabetes. The purported mechanisms by which vitamin E reduces risk (by modifying adverse effects of glycosylated hemoglobin on high density lipoprotein dysfunction³⁵) would likely be similar in both conditions. However, the relationship between Hp phenotype, cardiovascular disease risk and vitamin supplementation in Type 1 diabetes has not yet been investigated.

Conclusions

In contrast to results of previous studies in women without diabetes,^25,32 Hp phenotype did not significantly affect preeclampsia risk in white women with Type 1 diabetes. Researchers have suggested that Hp phenotype may identify a subgroup of women who would benefit from antioxidant supplementation to prevent preeclampsia.^25,37 Although our study had limited power to detect an interaction between Hp phenotype and treatment response, our preliminary analysis did not suggest that vitamins C and E supplementation would modify preeclampsia risk in women with Type 1 diabetes of any Hp phenotype. It remains possible that other types or combinations of antioxidants may benefit women of specific Hp phenotypes. However, the absence of a relationship between Hp phenotype and preeclampsia risk makes this possibility unlikely.