Objective
High maternal concentrations of thrombin-antithrombin (TAT) complexes have been associated with adverse outcomes. The objective of this study was to evaluate the relationship between TAT in asymptomatic subjects at 24 and 28 weeks and spontaneous preterm birth (SPTB).
Study Design
A secondary analysis of the National Institute of Child Health and Human Development Preterm Prediction Study was performed. Subjects with SPTB were matched to controls. Maternal TAT concentrations were previously measured at 24 and 28 weeks. Differences between cases and controls were analyzed with Mann-Whitney U and logistic regression.
Results
TAT was lower in cases than controls at 28 weeks ( P = .01). The odds ratio for SPTB with TAT less than 25% was 2.55 (95% confidence interval, 1.34–4.89) when adjusted for clinical variables.
Conclusion
Early third-trimester TAT was lower in subsequent cases of SPTB. In some patients, low TAT concentrations may represent impaired thrombin activation and be pathologic.
Spontaneous preterm birth (SPTB), defined as nonindicated birth before 37 weeks’ estimated gestational age, remains a major source of neonatal morbidity and mortality despite continuing research efforts. Deficits in our understanding of the underlying causes of SPTB pose significant impediments to the development of treatment or prevention strategies. Most cases of SPTB are believed to arise from 1 or a combination of 4 classic pathways: infection/inflammation, excessive myometrial mechanical stretch (ie, multifetal gestation), fetal or maternal stress, and chorion-decidual hemorrhage.
The study of preterm labor secondary to chorion-decidual bleeding is complicated by the fact that clinical bleeding is heterogeneous, ranging from minor vaginal spotting to overt placental abruption. Furthermore, many of the cases of chorion-decidual bleeding that contribute to SPTB are asymptomatic. For example, although clinically recognized placental abruption constitutes only about 5% of preterm births, histologic evidence of prior bleeding, as represented by hemosiderin-laden macrophages in the chorion decidua, is present in 30–40%.
The heterogeneity of chorion-decidual bleeding presents a significant obstacle to clinical study, and more quantitative means of assessment would be necessary prior to future mechanistic or interventional trials. To that end, researchers have sought to correlate peripheral markers of thrombin activation with adverse pregnancy outcomes.
Elovitz et al and Chaiworapongsa et al demonstrated increased thrombin-antithrombin (TAT) concentrations in subjects who presented with active preterm labor vs normal controls, with concentrations being higher in those subjects who subsequently failed tocolysis and delivered. Although placental pathologic data were not available, these subjects were presumably experiencing subclinical chorion-decidual bleeding. Additional studies by us and others have demonstrated an association between elevated TAT concentrations in asymptomatic subjects and the later development of preterm premature rupture of membranes (PPROM) and SPTB.
The purpose of this study was to analyze the relationship between peripheral markers of thrombin activation among women in the early third trimester and subsequent spontaneous preterm birth through a secondary analysis of the Preterm Prediction Study. Our hypothesis was that thrombin-antithrombin complexes would be higher in asymptomatic subjects at 24 and 28 weeks who subsequently experienced a spontaneous preterm birth.
Materials and Methods
The Preterm Prediction Study (PPS) was a prospective cohort of 2929 subjects recruited between 1992–1994 conducted by the Maternal-Fetal Medicine Units Network of the National Institute of Child Health and Human Development. Previous publications have described the study protocol and population in detail. In brief, subjects were recruited from 10 centers in which institutional review board approval had been obtained and participants had provided consent.
Exclusion criteria included major fetal anomalies, multifetal gestation, current cervical cerclage, and placenta previa. Baseline demographic data were collected at the initial study visit at 23–24 weeks of gestation, and subjects were then seen at 2-week intervals for 3 additional visits.
Outcome data were collected at the conclusion of the pregnancy. If the last menstrual period was known, it was used to determine the gestational age, unless it disagreed with the first ultrasound by more than 10 days. If the last menstrual period was not known, the earliest ultrasound was used for gestational age determination. Samples of maternal blood and cervical fluid were collected at 23–24 and 27–28 weeks. Maternal plasma and serum were separated within 2 hours of the blood draw and stored at –70°C.
To study the effects of serum and cervical fluid analytes on major outcomes, a planned nested case-control study was performed within the original cohort, and the detail has also been previously described. Cases were defined as subjects experiencing spontaneous preterm birth before 37 weeks’ estimated gestational age and were matched on the basis of race, parity, and center to subjects who had delivered at longer than 37 weeks.
A number of serum analytes were previously measured in the serum samples of cases and controls, as described elsewhere. An immunoassay for TAT had been previously performed using a commercial kit from American Diagnostica (Greenwich, CT). Sample absorbances were measured using a microtiter plate reader and Soft Max software program from Molecular Devices (Menlo Park, CA). All of the serum analyses had been performed prior to the dataset having been made available to the study investigators for secondary analysis.
TAT concentrations at 24 and 28-weeks in cases and controls, as well as subjects with or without vaginal bleeding or PPROM, were analyzed with a Mann-Whitney U test. The relationship between maternal TAT concentrations less than the 25th percentile or greater than the 75th percentile and preterm birth was determined with crude and adjusted odds ratios (ORs). Backward stepwise logistic regression was used to control for clinical variables and the results of other analytes. A P value of .05 was used for the selection of variables.
The clinical and laboratory variables included in the initial regression were identical to previous publications: smoking, prior preterm birth, vaginal bleeding, fetal fibronectin (FFN), bacterial vaginosis (BV), body mass index (BMI), cervical length (CL), α-fetoprotein (AFP), alkaline phosphatase, and granulocyte colony-stimulating factor (GCSF). In addition, C-reactive protein (CRP) and cortisol concentrations were included in the regression model as well. When analyzing TAT samples at 24 or 28-weeks, we used the FFN, BV, CL, and analyte results from the same visit. FFN was dichotomized with a cutoff of 50 ng/mL, CL was dichotomized with a cutoff of 25 mm, CRP at 8 ng/mL, and cortisol at the 75th percentile. The definitions of the other clinical variables are described in previous publications.
Survival analysis was performed for the time in days from the 28-week laboratory draw until delivery. Kaplan-Meier survival estimates were obtained for subjects with TAT concentration greater than or less than 25% at 28 weeks. A Cox regression was additionally performed with the same clinical and laboratory variables utilized in the logistic regression.
All statistical evaluations were performed with STATA 10 (Stata Corp, College Station, TX).
Results
Maternal plasma concentrations of TAT were available from 491 subjects at 24 weeks and 351 subjects at 28 weeks. To maintain the original matching algorithm, cases and controls without a race- or parity-matched pair were removed from the evaluations, resulting in 169 and 136 matched pairs at 24 and 28 weeks, respectively. Demographic data for these subjects are outlined in Table 1 . Because some subjects without matched pairs were removed, these values differ slightly from those that have previously been published for this group.
| Demographic | Case | Control | P value |
|---|---|---|---|
| Age, y | 23.3 | 23.8 | .36 |
| Black, % | 72.1 | 72.1 | Matched |
| BMI | 24.4 | 26.4 | < .001 |
| Parity | 1.7 | 1.7 | Matched |
| Smoking, % | 29.2 | 27.4 | .71 |
| Prior vaginal bleeding, % | 35.7 | 24.4 | .02 |
| History of prior SPTB, % | 33.8 | 17.8 | < .001 |
| EGA at delivery, wk | 34.0 | 39.3 | < .001 |
| Infant birthweight, g | 2218 | 3234 | < .001 |
TAT concentrations were nonparametrically distributed. The correlation between concentrations at 24 and 28 weeks was significant (Spearman’s rho = 0.67; P < .001). Among subjects with any vaginal bleeding in the first or second trimester, there were no differences in TAT concentrations at either 24 ( P = .58) or 28 ( P = .96) weeks.
Among subjects with persistent vaginal bleeding, there was no difference in TAT concentrations at 28 weeks ( P = .86), although a nonsignificant trend was present at 24 weeks ( P = .085). Interestingly, the trend at 24 weeks was toward lower, as opposed to higher, median TAT concentrations in the group with persistent vaginal bleeding (11 vs 17.8 ng/mL). When TAT concentrations at less than 25% were analyzed as a dichotomized variable, the association between persistent vaginal bleeding and low TAT concentrations was significant ( P = .025) by Fisher’s exact test.
TAT concentrations in cases and controls are presented at 24 and 28 weeks in Figures 1 and 2 , respectively, as well as Table 2 . Concentrations of TAT were lower in cases than controls at 28 weeks ( P = .01), and a nonsignificant trend toward lower TAT concentrations was present at 24 weeks ( P = .08). Seventy-two of the cases with TAT available at 28 weeks and 82 of the cases with TAT available at 24 weeks experienced PPROM prior to delivery.
