Materials and Methods
Study population
Study participants were recruited from 2 patient populations as reported previously. The longitudinal cohort was a nested case-control study of 60 women who were sampled serially throughout pregnancy, with the first sample taken between 6-14 weeks of gestation and an additional sample obtained in each trimester. In this longitudinal cohort, 15 women who developed preeclampsia at various gestational ages were matched with 45 women who remained normotensive and had measurements within approximately 2 weeks of their preeclampsia counterparts. A clinical preeclampsia cohort of 47 patients who were diagnosed with preeclampsia at various gestational ages was analyzed to measure the rate of change in GlyFn levels during the course of their preeclampsia. Preeclampsia status was defined having a systolic blood pressure ≥140 mm Hg or a diastolic blood pressure ≥90 mm Hg with proteinuria ≥300 mg/d. In all, 207 serum samples were analyzed using a plate assay and 86 serum samples were also analyzed using a GlyFn POC device. A total of 26 participants included in the analysis had 1 measurement, 62 had 2 measurements, and 19 had 3 measurements. Severe and mild preeclampsia were defined based on National High Blood Pressure Education Program Working Group on High Blood Pressure in Pregnancy criteria.
The study participants were recruited from the Department of Obstetrics and Gynecology, Oulu University Hospital, Oulu, Finland, and the Finnish maternity cohort serum bank at the National Institute for Health and Welfare from 2004 through 2006. The research protocol was approved by the Oulu University Hospital Ethics Committee, and all participants provided informed consent.
Analyte assays
Maternal serum was spun, aliquoted, and stored at –80°C until subjected to the assays described below.
GlyFn plate assay
Reacti-Bind plates (Thermo Scientific, Rockford, IL) were coated with an Fc fragment-specific goat antimouse IgG (catalog no. 115-005-071; Jackson ImmunoResearch Laboratories Inc, West Grove, PA) in carbonate buffer, pH 9.6, and incubated at 4°C overnight followed by washing with phosphate-buffered saline (PBS)-0.05% Tween 20. Plates were blocked with 3% bovine serum albumin in PBS, pH 7.2, for 1 hour at room temperature. Plates were then washed with PBS-0.05% Tween 20 buffer and an Fn monoclonal antibody was added and incubated for 45 minutes at room temperature. This Fn monoclonal antibody replaced our previous SNA lectin-coupled assay, as we found that the reactivity of this monoclonal with a glycosylated fraction of Fn was similar to that exhibited by the SNA lectin, allowing a simplified assay protocol. Samples and standard protein (human Fn isolated from serum, catalog no. 1918-FN-02M; R&D Systems, Minneapolis, MN) were incubated for 45 minutes, washed, and a biotinylated antihuman Fn polyclonal antibody (catalog no. A0245; DAKO, Carpinteria, CA) was added. Labeling was performed using high-sensitivity streptavidin-horseradish peroxidase (catalog no. 21130; Thermo Scientific). After incubation for 45 minutes followed by washing the plate with PBS-0.05% Tween 20 buffer, the plate was developed with 100 μL of K-Blue TMB substrate (catalog no. 304177; Neogen, Glasgow, Scotland) and quenched by the addition of 2N sulfuric acid (H 2 SO 4 ). The interassay coefficient of variation for the plate assay was 7.1% and the intraassay coefficient of variation was 3.1%.
GlyFn POC assay
A fluorescence immunoassay, comprising an automated cassette reader (cPoC reader; LRE Medical, Oceanside, CA) and a disposable, single-use, plastic assay cartridge was developed that employs standard immunoassay techniques to specifically and quantitatively detect GlyFn in serum specimens. The polyclonal anti-Fn antibody employed in the plate assay described above was conjugated to a Tide Fluor 5WS succinimidyl ester fluorescent tag (catalog no. 2281; AAT Bioquest, Sunnyvale, CA) and served as the detection antibody. The monoclonal Fn antibody employed in the plate assay described above served as the capture antibody and was immobilized on a solid phase (test zone). Goat polyclonal antirabbit IgG, Fc antibody (catalog no. 111-045-046; Jackson ImmunoResearch Laboratories Inc) was immobilized in a separate capture zone to act as a reference for the test zone and to provide assurance that the device performed properly.
Serum was diluted in assay buffer and applied to the test strip. The serum flows down the diagnostic lane via capillary action, taking the fluorescent detection antibody into suspension. GlyFn in the specimen binds to the fluorescent antibody to form a multivalent complex that is captured by the antibody immobilized in the test zone. The cartridge is inserted into the cassette reader and quantitative measurements of GlyFn concentration in the range from 10–2000 μg/mL are displayed on the meter screen and/or printout after 10 minutes.
sFlt1 levels were determined by enzyme-linked immunosorbent assay as previously described. Due to the large amount of serum needed for this assay, 13 participants were unable to be assayed for this analyte.
PlGF levels were determined using a commercial kit (human PlGF Quantikine enzyme-linked immunosorbent assay kit, catalog no. DPG00; R&D Systems). Due to inadequate serum sample, this analysis was subset to 57 subjects.
Plates were read using an Epoch plate reader at 450 nm, and data were processed using Gen5 software, version 1.10.8 (BioTek, Winooski, VT) and analyzed as described below.
Statistical analysis
Analyses were performed on the normotensive, longitudinal preeclampsia, and clinical preeclampsia samples separately, and were combined or compared only where indicated. Maternal characteristics were compared across study groups using Kruskal-Wallis nonparametric analysis of variance for continuous variables and Fisher exact test for categorical variables. Comparisons of GlyFn levels between longitudinal participants with and without preeclampsia were performed with parametric and nonparametric Wilcoxon t tests using measures matched by gestational age within approximately 2 weeks for each subject within a trimester. For analysis of sFlt1, PlGF, and the sFlt1/PlGF ratio, samples were subset to the third trimester and the normotensive group was compared to the clinical preeclampsia cohort via nonparametric Wilcoxon t tests.
To determine the average weekly change in GlyFn, linear regression with repeated measurements was used. Weekly change in controls was found to be stable across the span of pregnancy and was assessed by using a subset of patients with ≥2 repeated measures between 7-40 weeks. To determine the change in GlyFn in the progression of mild and severe preeclampsia, average change over time in GlyFn levels for these participants was assessed in a subset of participants with ≥2 repeated measurements between 33-38 weeks. All repeated measurements were taken within 2 weeks of the first measurement and measurements were on average 5 days after preeclampsia diagnosis, ranging between 5 days before to 19 days after preeclampsia diagnosis.
Receiver operating characteristic (ROC) curves for preeclampsia diagnosis were generated using predicted probabilities from simple logistic regression models using a single third-trimester measure for each subject from all cohorts. The area under the ROC curve (AUROC) and corresponding 95% confidence limits were calculated using simple logistic regression. Sensitivity and specificity were reported based on thresholds chosen, and 95% confidence limits calculated by the score method with a continuity correction are reported. Statistical tests of differences in ROC curves were calculated using contrast matrices of differences. Hypothetical predictive values and 95% confidence intervals (CIs) for diagnosis of preeclampsia were calculated using the standard logit method using a population prevalence of 3%, 5%, or 7%.
A post hoc analysis of maternal and fetal clinical characteristics and outcomes vs GlyFn values determined by the plate assay was performed. Within all cohorts, GlyFn was compared to gestational age at delivery, birthweight, systolic blood pressure, and diastolic blood pressure. Within clinical preeclampsia participants, comparative analyses were performed with GlyFn with respect to gestational age of preeclampsia start, uric acid, alanine transaminase, proteinuria, HELLP syndrome, small-for-gestational age, and placental insufficiency. For continuous variables, Pearson correlation coefficients and linear regression slopes were calculated. For interpretability, linear regression slopes were calculated to reflect a change in GlyFn of 100 μg. For categorical variables, Fisher exact tests were performed by categorizing participants as those with or without GlyFn levels ≥500 μg.
A comparison of the GlyFn plate assay to the GlyFn POC was performed on samples assayed by both methods to assess the POC test’s ability to distinguish between participants with and without preeclampsia as well as to assess the ability of GlyFn to monitor progression of preeclampsia during the second and third trimesters. Correlation coefficients were calculated to compare the 2 measures. ROC curves were generated separately for GlyFn plate and POC data for classification of control vs preeclampsia, control vs mild preeclampsia, and mild vs severe preeclampsia, and the AUROC for each was compared between GlyFn plate and POC assays. Reported P values are 2-sided, and P < .05 was considered statistically significant. Statistical analysis was performed using software (SAS, version 9.3; SAS Institute Inc, Cary, NC).
Results
Patients in the clinical preeclampsia group were more likely to give birth earlier ( P < .01) and have lower neonatal birthweights ( P < .01) ( Table 1 ). There was no difference in maternal age ( P = .14) and nulliparity ( P = .31) between the cohorts. Median gestational age at diagnosis of preeclampsia was significantly later in the longitudinal preeclampsia group than in the clinical preeclampsia cohort. A comparison of the longitudinal normotensive and preeclampsia groups found that, within each trimester, GlyFn levels were significantly higher in patients with preeclampsia than in controls ( P < .01) ( Table 2 and Figure 1 ). To assess the change in serum biomarkers during the third trimester, levels of GlyFn, sFLt1, PlGF, and the sFLT1/PlGF ratio were compared between age-matched samples from the normotensive control and clinical preeclampsia cohorts ( Table 3 ). There was a significant difference in all serum biomarkers between participants with and without preeclampsia ( P < .01) ( Table 3 ).
| Clinical characteristic | Normotensive, n = 45 | Longitudinal preeclampsia, n = 15 | Clinical preeclampsia, n = 47 | Group difference P value a |
|---|---|---|---|---|
| Maternal age at last menstrual period, y b | 26.5 (19.0–35.0) | 28.0 (21.0–34.0) | 29.0 (20.0–40.0) | .14 |
| Gestational age at delivery, wk b | 40.1 (38.4–42.0) | 40.2 (36.7–42.0) | 36.3 (21.7–40.6) | < .01 |
| Gestational age at diagnosis of preeclampsia, wk b | NA | 38.9 (32.0–40.0) | 32.7 (21.0–37.3) | < .01 |
| Neonatal birthweight, g b | 3510 (2690–4488) | 3260 (2520–4100) | 2250 (315–4200) | < .01 |
| Nulliparity, n (%) b | 29 (83) | 11 (73) | 32 (68) | .31 |
a Group differences were determined using Kruskal-Wallis nonparametric analysis of variance for continuous variables and Fisher exact tests for categorical variable
b Maternal age, gestational age at delivery, birthweight, and parity data were unavailable for 13, 12, 12, and 10 normotensive patients, and for 2, 1, 1, and 0 longitudinal preeclampsia participants.
| Biomarker concentrations | First trimester | Second trimester | Third trimester | ||||||
|---|---|---|---|---|---|---|---|---|---|
| Normotensive cohort (n = 24) | Longitudinal preeclampsia cohort (n = 11) | Group difference P value a | Normotensive cohort (n = 28) | Longitudinal preeclampsia cohort (n = 12) | Group difference P value a | Normotensive cohort (n = 34) | Longitudinal preeclampsia cohort (n = 13) | Group difference P value a | |
| Gestational age at sample collection, wk | 9.7 (6.4–13.0) | 9.3 (7.3–11.0) | .79 | 23.3 (17.4–26.6) | 22.4 (21.6–26.3) | .17 | 35.4 (27.0–39.7) | 36.4 (28.1–38.7) | .13 |
| GlyFn, μg/mL | 62 (7–198) | 184 (30–387) | < .01 | 41 (9–144) | 161 (6–848) | < .01 | 54 (1–199) | 239 (111–522) | < .01 |
a Group differences were determined using Wilcoxon nonparametric t tests.
| Biomarker concentrations | Third trimester | ||
|---|---|---|---|
| Normotensive cohort (n = 34) | Clinical preeclampsia cohort (n = 44) | Group difference P value a | |
| Gestational age at sample collection, wk | 34.8 (27.0–39.0) | 35.0 (27.0–39.0) | .21 |
| GlyFn, μg/mL | 55 (1–199) | 517 (151–1703) | < .01 |
| sFlt1 | 6.1 (0–15.6) | 23.2 (0–71.4) | < .01 |
| PlGF | 361.7 (115.3–673.7) | 105.6 (14.6–260) | < .01 |
| sFlt1/PlGF | 0.021 (0.000–0.066) | 0.208 (0.061–2.781) | < .01 |
a Data are median (range). Group differences were determined using Wilcoxon nonparametric t test. Due to insufficient sample volume, data are missing for 7, 22, and 22 normotensive patients and for 1, 14, and 14 clinical preeclampsia patients for Flt1, PlGF, and Flt1/PlGF ratio, respectively.
A repeated-measures analysis of change across all cohorts over time in biomarkers found that, in controls, there was not a significant change in GlyFn ( P = .83) across the span of pregnancy. In patients with preeclampsia, the weekly change between 33-38 weeks was 81.7 (SE 94.1) μg/mL for participants with mild preeclampsia and 195.2 (SE 88.2) μg/mL for participants with severe preeclampsia ( Table 4 ).
| Preeclampsia status | No. of subjects (no. of total measurements) | Gestational week | Weekly change GlyFn, μg/mL |
|---|---|---|---|
| Normotensive | 38 (87) | 7-40 | 0.1 ± 0.6 |
| Mild preeclampsia | 10 (5) | 33-38 | 81.7 ± 94.1 |
| Severe preeclampsia | 8 (4) | 33-38 | 195.2 ± 88.2 |
The clinical utility of these biomarkers for detection of preeclampsia was tested via ROC curves. The AUROCs for GlyFn, sFlt1, PlGF, and the sFlt1/PlGF ratio are show in Table 5 and the respective ROC curves in Figure 2 . Since the sFlt1 assay requires significant serum quantities, this analysis was restricted to 15 control and 39 preeclampsia participants between 20-39 weeks of gestation with sufficient serum for sFlt1 analysis. The AUROC for GlyFn was 0.99, and trended toward being significantly different from the AUROC for sFlt1 (AUROC, 0.96; P = .11) and PlGF (AUROC, 0.94; P = .10). Upon categorization at a threshold of 176.4 μg/mL, GlyFn demonstrated a sensitivity of 0.97 (0.85-1.00) and a specificity of 0.93 (0.66-1.00). At this threshold, and with an estimated population prevalence of 5% for preeclampsia, the positive and negative predictive values for diagnosis of preeclampsia were 47% (95% CI, 23–72%) and 89% (95% CI, 80–98%), respectively ( Table 6 ).
| Biomarker | AUROC (95% CI) | P value for comparison to GlyFn ROC |
|---|---|---|
| GlyFn, μg/mL | 0.99 (0.98–1.00) | NA |
| sFlt1, ng/mL | 0.96 (0.89–1.00) | .11 |
| PlGF, pg/mL | 0.94 (0.86–1.00) | .10 |
| Flt1/PlGF | 0.98 (0.94–1.00) | .29 |
| Threshold of GlyFn POC | Sensitivity and specificity | Predictive value | Predictive value (95% CI) with varying prevalence estimates | ||
|---|---|---|---|---|---|
| 3% | 5% | 7% | |||
| 176.4 | Sensitivity: 0.97 | Positive predictive value | 0.41 (0.19–0.67) | 0.47 (0.23–0.72) | 0.50 (0.26–0.75) |
| Specificity: 0.93 | Negative predictive value | 0.95 (0.83–0.99) | 0.94 (0.80–0.98) | 0.93 (0.77–0.98) | |
Stay updated, free articles. Join our Telegram channel
Full access? Get Clinical Tree
