Materials and Methods
Between March 1997 and June 1999, nonfasting blood samples were collected between 8:00 AM and 4:00 PM during second-trimester pregnancies of women receiving prenatal care at Northwestern Memorial Hospital. At the time of blood sampling, patients noted the times of their last meals. After centrifugation, sera were stored at −20C until further analysis. From that cohort, a nested case-control study was done in which women who subsequently developed preeclampsia were identified through chart reviews and matched to two controls who remained normotensive throughout their pregnancies.
Patients were diagnosed with preeclampsia based on the occurrence of new-onset hypertension after 20 weeks' gestation (two readings, at least 6 hours apart, of systolic pressure of at least 140 mmHg and diastolic pressure of at least 90 mmHg) with new-onset proteinuria (2+ or more on dipstick). Potential control subjects were matched according to gestational ages at blood sampling (±2 weeks) and delivery (±2 weeks) and birthweight (±300 g). From that group, two subjects were selected randomly by computerized assignment as the controls for each preeclamptic subject. Women were excluded if they had multifetal gestations, abnormal 1-hour glucose tolerance tests, or preexisting medical conditions such as hypertension, diabetes mellitus, or renal disease.
All assays were done in duplicate, in a single assay, and according to the manufacturer's instructions. Insulin was measured with a solid-phase radioimmunoassay provided by Diagnostic Products Corp. (Los Angeles, CA). The minimum detection limit was 1.2 μIU/mL, and the intraassay coefficient of variation was 8.7%. Insulin-like growth factor-I and IGFBP-1 were measured with a two-step enzyme-linked immunosorbent assay (Diagnostic Systems Laboratories Inc., Webster, TX). The IGF-I assay had a minimum detection limit of 0.01 ng/mL and an intraassay coefficient of variation of 4.6%. The IGFBP-1 assay had a minimum detection limit of 0.25 ng/mL and an intra-assay coefficient of variation of 4.6%. All assays were performed by individuals masked to the assignment of the control and study subjects.
For the power analysis, we used α = .05 and β = .2. Expected means and standard deviations of controls were derived from the published literature (79 ± 14 ng/mL IGFBP-1, 179 ± 28 ng/mL IGF-I).10,11 Based on that formulation, nine preeclamptic subjects and 18 controls were necessary to have 80% power to detect a 20% difference between groups with regard to the analytes under consideration. That magnitude of difference is consistent with alterations found in preeclamptic populations.
Continuous variables were compared with Student t test, and categorical variables were compared with χ2 analysis or Fisher exact test. The normality of the distribution of the serum analytes was assessed with the Kolmogorov–Smirov test for goodness of fit. Because those serum analytes were collected at different gestational ages, we used two-way analysis of variance to evaluate the interaction of gestational age at times of serum collection with the clinical outcome under consideration. For that analysis, the second trimester was divided into a first half (14–21 weeks) and a second half (22–28 weeks). P < .05 was considered significant for all tests. To find whether analyte distributions could predict which women would become preeclamptic, we used receiver operating characteristic (ROC) curve statistics. With a logistic regression equation that used preeclampsia as the dependent variable and analyte values as the independent variable, we contructed a variables ROC curve for early and late samplings. Statistics were done with Minitab 11 (Minitab Inc., State College, PA) and Stata 6 (Stata Corp., College Station, TX).
This study was approved by the Institutional Review Board of Northwestern University Medical School, and informed consent was obtained from all subjects before enrollment.
Twelve cases with preeclampsia and 24 controls were identified. Of the women with preeclampsia, one met criteria for severe disease and 11 had mild disease.13 Cases and controls were similar with respect to many demographic factors including, as expected from the matching criteria, birthweight and gestational age at delivery (Table 1). The only significant difference was mean arterial pressure at first prenatal visits and deliveries.
The Kolmogorov–Smirnov goodness-of-fit test showed that IGF-I and IFGBP-1 values were normally distributed for cases and controls (P > .05). Insulin values were normally distributed only after logarithmic transformation, and consequently those transformed values were evaluated by parametric statistical tests.
Serum concentrations of insulin, IGF-I, and IFGBP-1 with respect to gestational age at blood sampling and ultimate clinical outcome are presented in Table 2. For the entire patient population, serum concentrations of insulin and IGF-I were significantly higher in the later half of the second trimester (P < .001), whereas IGFBP-1 (P = .07) concentrations were not significantly different at the later gestational age. Comparison of women who became preeclamptic with normotensive controls showed no difference in insulin concentrations, although serum concentrations of IGF-I were significantly higher (P < .001) and those of IGFBP-1 were lower (P < .05). The interaction between gestational age at sampling and clinical outcome was not statistically significant for those analytes.
Given that the serum concentrations of IGF-I and IGFBP-1 were significantly different, we compared the case and control distributions and investigated whether a discriminatory point could be found that distinguished women who developed preeclampsia from those who remained normotensive. Because of the substantial overlap, no adequate cutpoint was found. We postulated that the ratio of IGF-I to IGFBP-1 might decrease this overlap and found that a decrease was apparent when the distributions of that ratio were compared (Figure 1A and B). Based on a logistic regression equation that used preeclampsia as the dependent variable and the ratio of IGF-I to IGFBP-1 as the independent variable, we constructed a ROC curve for early and late samplings. With respect to the sampling period of 14–21 weeks, the curve showed that a ratio greater than 1.3 optimized sensitivity (83%; 95% confidence interval: 36, 100) and specificity (83%; 95% confidence interval: 52, 98). For the sampling period of 22–28 weeks, a ratio greater than 2.3 resulted in a sensitivity of 100% (95% confidence interval: 54, 100) and a specificity of 75% (95% confidence interval: 43, 96).
Some controversy exists concerning the relation between insulin, the insulin-like growth factor family of proteins, and preeclampsia. Whereas many investigators found evidence of hyperinsulinemia and greater insulin resistance in women with preeclampsia, others did not and, in some cases, found that preeclamptic women were more sensitive to insulin.14,15 Many investigators also noted that women who had developed preeclampsia had higher serum concentrations of IGFBP-1.7,8,10 Giudice et al theorized that higher decidual concentrations of the protein, related to those serum elevations, might have contributed to the shallow placental invasion that is characteristic of preeclampsia, although the exact mechanisms of inhibition remain to be determined.16 In contrast, two investigators found that before women showed clinical evidence of preeclampsia, serum concentrations of IGFBP-1 also were altered, but in a direction opposite to that found in women who had clinical evidence of preeclampsia.11,12 Although Giudice et al also found that IGF-I decreased in women with preeclampsia, the peptide had not been studied before the clinical development of preeclampsia.10
In this study, we simultaneously investigated serum concentrations of insulin, IGF-I, and IGFBP-1 in second-trimester pregnancies of women who had not yet developed clinical evidence of preeclampsia. Because the concentrations of those analytes were reported to change with advancing gestational age,11,17 cases were matched with controls according to gestational age at blood sampling. Controls also were selected according to gestational age and birthweight. Because IGF-I and IGFBP-1 were correlated with fetal growth restriction,18 those matching criteria allowed any alterations in those proteins to be more specifically related to preeclampsia and not the simultaneous occurrence of growth restriction. Analysis of the demographic factors of women in this study showed that the cases and controls were quite similar; there were no discernable differences in matched or other characteristics that could introduce bias. Glucose tolerance, for example, as assessed by the 1-hour tolerance test, was similar between groups, as was the timing of last meals before blood samplings. As expected, mean arterial pressure was significantly greater at delivery among preeclamptic women. Although mean arterial pressure also was greater at first prenatal visits, the magnitude of this difference was less and, we believe, more statistically significant than clinically relevant. A slightly elevated mean arterial pressure before development of preeclampsia is consistent with reported findings.19
No difference in mean insulin concentrations was found between groups. One issue of potential consequence is that single insulin concentrations alone cannot reliably indicate the clinical state of insulin resistance. Methods such as area under the curve of sequential insulin values showed that insulin resistance does exist in preeclampsia even when mean insulin concentrations are not significantly different.20 Therefore, we do not believe that this lack of statistical significance of a single insulin concentration negates the possibility of a relation between preeclampsia and reduced insulin sensitivity. Conversely, consistent with previous reports,11,12 we found that women had significantly lower serum concentrations of IGFBP-1 before the appearance of clinical signs associated with preeclampsia. Serum concentrations of IGF-I also were significantly higher than in women who remained normotensive. As with IGFBP-1, that alteration of IGF-I is in a direction opposite to that reported in women who had developed preeclampsia. It remains unknown if those proteins are altered after initiation of the pathophysiologic process that culminates in preeclampsia or are essential determinants of preeclampsia.
Given the role those proteins have in placental development, it is reasonable to believe that they might contribute to the development of preeclampsia. Patients with preeclampsia had impaired trophoblast invasion into maternal deciduas.21 Both IGF-I and IGFBP-1 are modulators of placental growth and might regulate one another. In fact, IGFBP-1 is prevalent in maternal decidua and might have an inhibitory effect on the invading trophoblast.16,18 Interestingly, molar pregnancies, which have abnormal placentation and a significantly higher incidence of preeclampsia, had serum concentrations of IGF-I and IGFBP-1 that were lower and higher, respectively, than those of normal pregnancies.22,23 Therefore, alterations in the concentrations of those serum proteins might reflect essential alterations in the uterine environment that contribute to the deranged placentation found in preeclampsia.
Regardless of the exact roles of IGF-I and IGFBP-1 in the pathophysiologic cascade of preeclampsia, that they were detectably different in women destined to become preeclamptic suggests that these analytes may be useful as markers to identify women at high risk of becoming preeclamptic later in their pregnancies. Although the means of each analyte differed between women who became preeclamptic and those who remained normotensive, the overlap of the distributions of these analytes, when each was considered alone, precludes their usefulness as potential markers. Other investigators reported similar findings.11 However, combining these analytes into a single ratio appeared to decrease inter-patient variability of the serum concentrations and achieve a more discriminatory distribution. That finding must be interpreted with caution. Although this study had adequate power to detect a difference in mean concentration of the serum analytes, the number of patients was too small to establish the adequacy of a screening test, and the discriminatory point was empirically determined by ROC curves after the data were obtained. The large confidence intervals associated with the sensitivities and specificities also emphasize the preliminary nature of that finding. The potential use of that ratio can be confirmed only with a larger sample that has been prospectively evaluated after establishing an ante hoc discriminatory point.
1. Grunewald C. Biochemical prediction of preeclampsia. Acta Obstet Gynecol Scand 1997;76:104–7.
2. Ferrannini E, Buzzigoli G, Bonadonna R, Giorico MA, Oleggini M, Graziadei L, et al. Insulin resistance in essential hypertension. N Engl J Med 1987;17:350–7.
3. Sermer M, Naylor CD, Farine D, Kenshole AB, Ritchie JW, Gare DJ, et al. The Toronto trihospital gestational diabetes project: A preliminary review. Diabetes Care 1998;174:1032–7.
4. Siddiqi T, Rosen B, Mimouni F, Khouri J, Miodovnik M. Hypertension during pregnancy in insulin-dependent diabetic women. Obstet Gynecol 1991;77:514–9.
5. Fuh MT, Yin CS, Pei D, Sheu WH, Jeng CY, Chen YD, et al. Resistance to insulin-mediated glucose uptake and hyperinsulinemia in women who had preeclampsia during pregnancy. Am J Hypertens 1995;8:768–71.
6. Laivouri H, Tikkanen MJ, Ylikorkala O. Hyperinsulinemia 17 years after preeclamptic first pregnancy. J Clin Endocrinol Metab 1996; 81:2908–11.
7. Iino K, Sjoberg J, Seppala M. Elevated circulating levels of a decidual protein, placental protein 12, in preeclampsia. Obstet Gynecol 1986;68:58–60.
8. Howell RJS, Economides D, Teisner B, Farkas AG, Chard T. Placental proteins 12 and 14 in preeclampsia. Acta Obstet Gynecol Scand 1989;68:237–40.
9. Halhli A, Bourges H, Carrillo A, Garabedian M. Lower circulating insulin-like growth factor I and 1,25-dihydroxyvitamin D levels in preeclampsia. Rev Invest Clin 1995;47:259–66.
10. Giudice LC, Martina NA, Crystal RA, Tazuke S, Druzin M. Insulin-like growth factor binding protein-1 at the maternal–fetal interface and insulin-like growth factor-I, insulin-like growth factor-II, and insulin-like growth factor binding protein-1 in the circulation of women with severe preeclampsia. Am J Obstet Gynecol 1997;176:751–8.
11. De Groot CJM, O'Brien TJ, Taylor RN. Biochemical evidence of impaired trophoblastic invasion of decidual stroma in women destined to have preeclampsia. Am J Obstet Gynecol 1996;175:24–9.
12. Hietala R, Pohja-Nylander P, Rutanen EM, Laatikainen T. Serum insulin-like growth factor binding protein-1 at 16 weeks and subsequent preeclampsia. Obstet Gynecol 2000;95:185–9.
13. American College of Obstetricians and Gynecologists. Hypertension in pregnancy. ACOG technical bulletin no. 219. Washington DC: American College of Obstetricians and Gynecologists, 1996.
14. Roberts RN, Henfiksen JE, Hadden DR. Insulin sensitivity in preeclampsia. Br J Obstet Gynaecol 1998;105:1095–100.
15. Caruso A, Ferrazzani S, DeCarolis S, Lucchese A, Lanzone A, DeSantis L, et al. Gestational hypertension but not preeclampsia is associated with the insulin resistance syndrome characteristics. Hum Reprod 1999;14:219–3.
16. Giudice LC, Mark SP, Irwin JC. Paracrine actions of insulin-like growth factors and IGF binding protein-1 in non-pregnant human endometrium and at the decidual–trophoblast interface. J Reprod Immunol 1998;39:133–48.
17. Hills FA, English J, Chard T. Circulating levels of IGF-I and IGF-binding protein-1 throughout pregnancy: Relation to birth weight and maternal weight. J Endocrinol 1996;148:303–9.
18. Giudice LC. Multifaceted roles for IGFBP-1 in human endometrium during implantation and pregnancy. Ann NY Acad Sci 1997;828:146–56.
19. Caritis S, Sibai B, Hauth J, Lindheimer M, VanDorsten P, Klebanoff M, et al. Predictors of preeclampsia in women at high-risk. Am J Obstet Gynecol 1998;179:946–51.
20. Lorentzen B, Birkeland KI, Endresen MJ, Henriksen T. Glucose intolerance in women with preeclampsia. Acta Obstet Gyncol Scand 1998;77:22–7.
21. Khong TY, DeWolf F, Robertson WB, Brosens I. Inadequate maternal vascular response to placentation in pregnancies complicated by preeclampsia and by small-for-gestational age infants. Br J Obstet Gynecol 1986;93:1049–59.
22. Rutanen EM, Bohn H, Seppala M. Radioimmunoassay of placental protein 12: Levels in amniotic fluid, cord blood, and serum of healthy adults, pregnant women, and patients with trophoblastic disease. Am J Obstet Gynecol 1982;144:460–3.
23. Wihman IL, Carlstrom K, Faxen M. Insulin-like growth factor-I in women with hydatidiform mole and in normal pregnancy. Obstet Gynecol 1998;92:431–4.