This study was performed in a nulliparous population, with collection of a comprehensive clinical data set and with a standardized definition of preeclampsia that allowed for a rigorous evaluation. Despite evaluation of multiple first trimester clinical and biochemical parameters, we were unable to identify an algorithm that could predict subsequent preeclampsia with clinically useful sensitivity and specificity. When calculating sensitivity with specificity set at 80% we correctly identified less than half of those women who developed preeclampsia, i.e. less than the flip of a coin, and just more than half of those who developed severe preeclampsia. The small number of patients with early onset preeclampsia unfortunately precluded analysis for this group with a high incidence of maternal and fetal morbidity.
Multivariable analysis demonstrated that African American or Hispanic race, systolic blood pressure, BMI at enrollment, and education level were the strongest clinical predictors of development of preeclampsia but with only 36% sensitivity at 80% specificity. In contrast to previous findings diastolic blood pressure at enrollment (11
) was not significantly higher. Hispanic ethnicity is not traditionally recognized as a risk factor for preeclampsia but is related to an increased incidence of obesity, metabolic syndrome and diabetes (12
). Recent reports utilizing clinical variables collected at less than 15 or 16 weeks gestation similarly noted their low sensitivity for prediction of preeclampsia (13
). When predicting severe preeclampsia the multivariable analysis of clinical predictors was improved to a sensitivity of 42% at 80% specificity, although still not clinically useful.
Preeclampsia is proposed to have an immune component with a protective effect of exposure to paternal antigens (15
) but we found no relationship between number of partners or the duration of the sexual relationship and development of preeclampsia. Nor was there an association of smoking, a family history of preeclampsia or cardiovascular disease and development of the syndrome. This contrasts with the reports from Chesley (17
) and more recently that a positive family history was associated with a 20–30% risk of developing preeclampsia (18
). Our failure to find such associations may reflect our racially and ethnically heterogeneous population with a smaller proportion of smokers (16.9%) and a larger number of obese (22.5%) and overweight (26.0%) women.
Of the blood parameters only MPV, a sensitive indicator of platelet activation and consumption (19
) was different at enrollment, being modestly greater in those who subsequently developed preeclampsia compared with those who remained normotensive. MPV is higher in patients with hypertension and pre-hypertension (20
), in insulin resistant, non-obese, non-diabetic patients with coronary artery disease (21
) and in those with diabetes mellitus, hypertension, hypercholesterolemia, smoking and obesity, which are all reported risk factors for preeclampsia. Cross-sectional studies of MPV in women with established preeclampsia revealed high sensitivity and specificity (8
). We found that MPV MoM was significantly higher in the first trimester in women destined to develop preeclampsia preceding the clinical onset by 16–18 weeks, but with poor sensitivity and specificity. MPV MoM was however not significantly higher in women who later developed severe preeclampsia, suggesting platelet activation is not common to all cases of preeclampsia. Unlike previous reports (22
) the ratio of platelet volume/number did not relate to development of preeclampsia (data not shown). The association of platelet volume with African American and Hispanic race, first-trimester BMI and systolic blood pressure indicates we may be identifying susceptible women with subclinical vascular dysfunction (23
Microarray analysis revealed ADAM12 to be the most highly upregulated gene transcript in placental tissue (24
) from women with established preeclampsia, with corresponding increases in ADAM12 in maternal serum. Our finding of significantly increased serum concentrations of ADAM12 in the first trimester contradicts reports of decreased ADAM12 (25
) in women who subsequently developed preeclampsia. This may be due to differing patient populations; however we agree with previous studies, that another proteinase, PAPP-A, was significantly lower in the first trimester (27
) in women who developed preeclampsia compared with those who did not. Although significantly different at 9–12 weeks of gestation, the sensitivity, either singly or combined of ADAM12 or PAPPA, for prediction of subsequent preeclampsia did not have clinical utility in agreement with previous smaller studies (26
In contrast to the reports that first trimester PP13 is significantly reduced (29
) in those patients who went on to develop preeclampsia we find no significant difference. A smaller retrospective study reported similar negative findings for PP13 although it found, as we did, a positive association for PAPP-A (32
) Our finding of no increase in predictive power by combined measurement of PP13 and PAPP-A again agrees with recent studies (33
Significant changes in the ratio of pro-angiogenic (PlGF) to antiangiogenic (sFlt-1 and endoglin) markers precede the clinical presentation of preeclampsia by several weeks (35
). PlGF has the best predictive power in the first trimester but with low sensitivity and combination of angiogenic and antiangiogenic markers i.e. PlGF/sFlt and PlGF/endoglin ratios did not appreciably increase the sensitivity (data not shown). The low predictive capability of suites of trophoblast function or placental pro-angiogenic/antiangiogenic markers indicates that either the involvement of these agents in the pathophysiology of preeclampsia is a late event or that there are several pathologic phenotypes leading to the syndrome of preeclampsia (3
The low sensitivity (46.1%) from this multivariable analysis of clinical risk factors and biochemical markers points to the heterogeneity of preeclampsia, the difficulty of identifying at-risk patients from among a low-risk group and of defining an enriched population for study. When new biomarkers are advanced they may similarly lack sensitivity if the syndrome has several pathologic phenotypes. Defining the clinical outcomes of patients identified by abnormal biomarker values in the first trimester may prove useful in elucidating if preeclampsia has several different underlying pathologies. Phenotyping and predicting disease based on biomarkers may prove more useful for diagnosis of disease and outcome in individuals rather than a clinical diagnosis of the preeclampsia syndrome.