Association of Elevated Serum Ferritin Levels and the Risk of Gestational Diabetes Mellitus in Pregnant Women: The Camden Study Free

The study was conducted in the Osborn Family Health Center, Our Lady of Lourdes Hospital, in Camden, New Jersey. Data were collected as part of the Camden Study, a prospective cohort study of maternal nutrition and pregnancy outcome in young, generally healthy women residing in one of the poorest cities in the continental U.S. (23). The institutional review board at the University of Medicine and Dentistry of New Jersey–SOM approved the study protocol. Informed written consent was obtained from each participant after explanation of the nature and purpose of the study.

Study participants were enrolled between October 1996 and June 2003 for prenatal care. Less than 5% of the women screened for eligibility were excluded from participation because of serious nonobstetric problem (e.g., lupus, type 1 or 2 diabetes, seizure disorders, malignancies, acute or chronic liver disease, drug or alcohol abuse, and psychiatric problems). Eighty percent of the patients who were eligible agreed to participate in this study. A final total of 1,456 pregnant women were included in this analysis after the exclusions of 3 patients who had hepatitis during pregnancy.

Data on socioeconomic status, demographics, and lifestyle were obtained by interview at entry to care (∼15 weeks of gestation) and updated at weeks 20 and 28 of gestation. Ethnicity was self-defined. Gestational duration was based upon gestation from participants’ last normal menstrual period confirmed or modified by ultrasound. Height was measured at entry to prenatal care, and prepregnant weight was obtained by recall. BMI was computed as prepregnant weight for height (weight in kilograms divided by the square of height in meters). Skinfold thickness was measured by a skinfold caliper (Cambridge Scientific Industries, Cambridge, MA), and upper arm fat was computed with measures of arm circumference and triceps thickness as described (24). Three different points of the skinfold thickness, including triceps, subscapular, and suprailiac, were summed.

The diagnosis of GDM was made using a two-step approach. Patients were initially screened by measuring the plasma glucose concentration 1 h after a 50-g oral glucose challenge test at ∼28 weeks of gestation. A diagnostic oral glucose tolerance test was performed on the subset of women whose plasma glucose concentrations exceeded the glucose threshold value (>140 mg/dl). The diagnostic criteria for GDM were the Carpenter/Coustan conversion as recommended by the American Diabetes Association (16).

A nested case-control study included 35 patients with GDM who had sufficient blood specimens available and 137 control subjects without GDM. Control subjects were randomly selected from each tertile of serum ferritin concentration to assure that the distribution of serum ferritin was the same as in the cohort as a whole (SAS: proc surveyselect).

Blood samples collected at entry to care (entry, 15.6 ± 0.1 weeks of gestation) from study participants were refrigerated and centrifuged at 4°C. Plasma and serum samples were stored at −70°C until assayed.

Serum ferritin was measured using a two-site immunoradiometric assay kit with a minimum detectable concentration of 2.55 pmol/l (Bio-Rad, Hercules, CA). Serum CRP levels were measured by an ultrasensitive enzyme-linked immunoabsorbent assay (Diagnostic Systems Laboratories, Webster, TX) with a minimum detectable concentration of 1.6 ng/ml. Hb was assayed by the commercial cyanmethemoglobin method (Sigma Diagnostics, St. Louis, MO), and hematocrit (HCT) was determined by microhematocrit capillaries after centrifugation. The coefficients of variation within and between assays were 3.0 and 6.2% for serum ferritin and 4.2 and 6.3% for serum CRP.

Univariate statistics were calculated for continuous variables, and a χ² test was used for categorical variables. ANOVA was used to generate and compare means for the measurements. Quintiles of serum ferritin were used in the following analyses because of its skewed distribution. Multiple linear regression analyses were used to examine the relations between elevated serum ferritin (independent variable, coded as highest quintile versus quintiles 1–4) and prepregnant BMI and body fatness by skinfold measurements (as the dependent variable). The logistic regression analyses were performed to determine the influence of elevated serum ferritin on the risk of GDM. Odds ratios (ORs), adjusted odds ratios (AORs) and their 95% CIs from the logistic regression coefficients and their corresponding covariance matrices, and the P for trend were computed. Potential confounding variables known to be associated with GDM were included in multivariable models (maternal age, ethnicity, parity, and family history of diabetes in a first-degree relative). Prepregnant BMI was tested in separate models because it correlated with serum ferritin and was highly correlated with CRP levels. Similar methods were used to analyze the data from the nested case-control study using tertiles of serum ferritin and CRP because of the smaller sample size. All statistical procedures were performed using SAS version 9.0 (SAS Institute, Cary, NC).

The characteristics of participants were tested according to serum ferritin quintiles (Table 1). Linear relationships between serum ferritin with maternal age (P < 0.0001) and prepregnant BMI (P < 0.0002) were observed, but Hb and HCT levels were not significantly different. In addition, women who developed GDM during pregnancy (n = 45, 3.1%) were older (P < 0.0001) and had higher prepregnant BMI (P < 0.0001) and higher serum ferritin concentrations (131.11 ± 17.00 pmol/l) than women who did not develop GDM (87.53 ± 2.15 pmol/l, P < 0.001).

The highest quintile of serum ferritin (>131 pmol/l) was significantly and positively associated with prepregnant BMI (P < 0.0001) and skinfold measurements determined at entry including the sum of three skinfolds (P < 0.01) and upper arm fat area (P < 0.01, Table 2). This relationship changed after GDM, and obese women were excluded (prepregnant BMI P < 0.0001, skinfold measures NS).

The highest quintile of serum ferritin at entry was associated with a twofold increased risk of developing GDM when compared with all other quintiles (Table 3) (OR 2.05 [95% CI 1.10–3.86], P for trend <0.05). After multivariable adjustment, the AOR was 2.02 (1.04–3.92, P for trend <0.05). The results were moderately modified after an additional adjustment for prepregnant BMI (AOR 1.84 [0.95–3.58]).

We further examined the association of high serum ferritin at entry and prepregnant BMI on risk of GDM. Among obese women (n = 333), the highest quintile of serum ferritin was associated with 3.5-fold increased risk of development of GDM (95% CI 1.35–9.27, P < 0.01), but there was little association between high ferritin levels and GDM in the nonobese subjects (n = 1,123, AOR 1.42 [95% CI 0.55–3.69]).

In addition, we did not find any association between high Hb level (Hb >130 g/l, highest quintile) and risk of GDM for the cohort (multivariable adjusted, AOR 0.81 [95% CI 0.36–1.81], P > 0.05) or among nonanemic patients (0.97 [0.42–2.24], P > 0.05) (25).

Women with GDM (n = 35) were older (mean ± SE, 26.9 ± 1.0 years) than randomly selected control subjects (n = 137, 22.7 ± 0.5 years, P < 0.001) and had higher prepregnant BMI (29.5 ± 1.0 vs. 25.2 ± 0.5, P < 0.001). Case patients had a significantly higher concentration of serum ferritin (141 ± 17 vs. 86 ± 9 pmol/l) and CRP (13.9 ± 1.5 vs. 9.0 ± 0.8 mg/l) levels than the control subjects (P < 0.01 for both). However, serum ferritin and CRP levels were not highly correlated (r = 0.136, P = 0.08).

Table 4 presents the results of the nested case-control study. Women in the highest tertile of serum ferritin had a greater than twofold increased risk of development of GDM (AOR 2.35 [95% CI 1.06–5.22], P for trend <0.05). The association remained positive but became nonsignificant after serum CRP level was added to the model (2.10 [0.92–4.79]), although the change was small. In a separate model, women within the highest tertile of serum CRP level had a 2.7-fold increased risk of GDM (2.67 [1.16–6.17]). The result did not change appreciably with additional adjustment of serum ferritin (2.53 [1.08–5.94]). In addition, women who were in the highest tertiles of both serum ferritin and CRP levels had the highest risk of GDM in univariate (OR 4.98 [95% CI 2.02–12.26]) and multivariate models (AOR 3.30 [1.10–8.41]). However, the effects of serum ferritin and CRP levels on GDM became nonstatistically significant after the adjustment for prepregnant BMI.

In this prospective cohort with a nested case-control study, we have demonstrated associations between moderate elevations in serum ferritin levels during early gestation, measures of body fatness, and the risk of GDM in healthy young pregnant women. To our knowledge, the current study is the first large prospective study of serum ferritin and risk of GDM in young and healthy pregnant women from the U.S.

Data in men and nonpregnant women have showed that elevated serum ferritin was significantly associated with several CVD risk factors including BMI, waist circumference or waist-to-hip ratio, and serum lipid levels (7,14). Fernandez-Real et al. (11) found that BMI and high serum ferritin levels are independent predictors of insulin sensitivity and suggested that a high serum ferritin level could be a component of the insulin resistance syndrome in healthy subjects and an indicator of diabetes glycemic control. In contrast, studies have failed to find a positive association between serum ferritin and CVD risk factors including BMI (26). Among Chinese pregnant women, serum ferritin concentration was not correlated with the mother’s BMI (r = 0.0455) (22), whereas we observed that the highest quintile of serum ferritin was significantly correlated with prepregnant BMI and skinfold measurements during pregnancy (Table 2). This relationship was modified significantly after patients with obesity and GDM were excluded (Table 2). Thus, an elevated serum ferritin level can be a biomarker of body fat mass before and during pregnancy, especially in obese women.

Positive associations between mildly increased serum ferritin concentrations and indexes of insulin resistance in both healthy subjects and patients with type 2 diabetes have been reported (9–10,12). Elevated serum ferritin levels were associated with a greater than twofold increased risk of development of type 2 diabetes in the Finnish population (10). A strong association between higher serum ferritin concentration and newly diagnosed type 2 diabetes was observed among a U.S. population as well (12). Data from a large prospective nested case-control study in healthy women indicated a significantly increased risk of type 2 diabetes in women whose serum ferritin level was >107 ng/ml (13).

Data on the relation of serum ferritin and risk of GDM are limited. Reports by Lao et al. (21,22) among Chinese gravidas indicated that mean serum ferritin concentrations at 28–30 weeks of gestation was increased significantly in women with impaired glucose tolerance (P < 0.0001) and in patients with GDM compared with control subjects (P < 0.0001). We observed that serum ferritin levels were significantly increased in patients with GDM compared with participants without GDM in the cohort and the nested case-control study at entry. Gravidas with higher serum ferritin levels in early gestation (highest quintile >131 pmol/l) had a twofold increased risk of GDM after a multivariable adjustment without prepregnant BMI (Table 3) (P < 0.05). This effect persisted in the nested case-control study (AOR 2.35 [95% CI 1.06–5.22], Table 4), thus suggesting that moderate elevations of serum ferritin in normal pregnancy are associated with an increased risk of GDM.

Whether or not a high serum ferritin level is a risk factor for type 2 diabetes or GDM independent of known risk factors is controversial. Jiang et al. (13) suggested that the increased risk of type 2 diabetes associated with an elevated serum ferritin level is independent of known diabetes risk factors (confirmed after adjustment for BMI) in apparently healthy women. Our data showed that the association between high serum ferritin levels and risk of GDM remained positive but was moderately modified after adjustment for prepregnant BMI in the cohort (Table 3) (AOR 1.84 [95% CI 0.95–3.58]) and in the case-control study (Table 4) (AOR 2.10 [0.92–4.79]). In addition, obese women with high ferritin levels had a 3.5-fold increased risk of developing GDM (95% CI 1.35–9.27, P < 0.01), whereas nonobese women did not. These data thus suggest that the impact of high serum ferritin on the risk of GDM is, at least in part, mediated by obesity.

There is an extensive body of data suggesting that higher iron stores are associated with risk of type 2 diabetes in nonpregnant subjects (10,12, 13,27,28). In pregnant women, Lao et al. found that higher Hb (>13 g/dl) was an independent risk for GDM (29) and that women with iron deficiency anemia had a reduced risk of GDM (30). Our findings that higher Hb (>130 g/l) during early pregnancy was not associated with increased risk for GDM did not support the hypothesis that higher serum ferritin reflects excess iron stores in patients with GDM.

The inflammatory cytokines, including tumor necrosis factor-α and interleukin-1α, have been shown to induce ferritin synthesis in the experimental models (2). Thus, serum ferritin may be an unreliable indicator of iron stores when there is infection or inflammation (5,31). Pregnancy itself is an inflammatory state; serum CRP levels are raised as early as 4 weeks of gestation (32). Studies suggested that adipose tissue is an important determinant of a low level of chronic inflammation as indicated by a strong correlation between serum CRP levels and body fat mass in both nonpregnant and pregnant subjects (20,33). Wolf et al. (18) and Retnakaran et al. (33) demonstrated that a higher serum CRP level associated with a risk of GDM was mediated by increased BMI. In contrast, Qiu et al. (19) reported that the increased risk of GDM associated with a higher CRP level was independent of maternal adiposity. A few studies examined serum CRP and ferritin simultaneously and found them not to be highly correlated (13–15). Jiang et al. (13) found that high serum ferritin and CRP levels are independent risk factors for type 2 diabetes. Our data showed that higher serum ferritin or CRP levels are associated with increased risk of GDM (Table 4). Women who had both high CRP and serum ferritin had the greatest risk of GDM (Table 4), which confirmed prior findings with type 2 diabetes in nonpregnant women (13). ORs were modified with an additional adjustment for prepregnant BMI in the serum ferritin model and were significantly reduced in the CRP model. These data suggest that there may be a systemic inflammation involved in the pathophysiology of GDM, indicated by elevations of both serum ferritin and CRP levels. To some extent, the inflammation is mediated by body fat mass or obesity before and during pregnancy.

This study was supported by grants HD18269 and HD38329 from the National Institute of Child Health and Human Development.

We thank the staff of the Osborn Family Health Center, Our Lady of Lourdes Hospital, for providing access to patients; Melissa Sims, Janet Mead, and SaTonya Jones for laboratory assays; and Deborah Cruz for typing the manuscript.