Hypertension 2014 Verlohren 346 52

346

Preeclampsia is a multisystem disorder in pregnancy. With an incidence of 2% to 5% of all pregnancies, it is a frequent

pregnancy-related disease and a major cause of maternal and fetal morbidity and mortality.1 Preeclampsia is defined as the new-onset of hypertension (≥140/90 mm Hg on 2 separate occasions ≥4 hours apart) and proteinuria (≥300 mg/24 h).2 The simplicity of this definition contrasts with the complexity of the disease. The pathophysiology of preeclampsia is not conclusively resolved up to now. Early-onset preeclampsia, defined as the onset before 34 weeks of gestation, comprises the mal implantation of the pla-centa, insufficient spiral-artery remodeling, prevalently result-ing in intrauterine growth restriction and an altered expression of placental proteins such as soluble fms-like tyrosine kinase-1

(sFlt-1) and placental growth factors (PlGFs).3 In contrast, late-onset preeclampsia, defined as the onset after 34 weeks of ges-tation, is not necessarily accompanied by placental dysfunction4 but angiogenic and antiangiogenic factors are also dysregulated, though to a less dramatic extent.5

We and others have previously shown that the sFlt-1/PlGF ratio is elevated in patients with preeclampsia.6–10 The auto-mated measurement of the sFlt-1/PlGF ratio detects early-onset preeclampsia with a sensitivity of 89% and a specificity of 97% when the single, gestation-wide cutoff of 85 is used.8 Recently, Rana et al9 have shown that in patients with signs and symptoms for the disease, an sFlt-1/PlGF ratio ≥85 predicts the occurrence of preeclampsia-related adverse maternal and fetal outcomes,

Abstract—To establish gestational phase adapted cutoffs for the use of the soluble fms-like tyrosine kinase-1 (sFlt-1)/placental growth factor (PlGF) ratio as a diagnostic tool for preeclampsia in the clinical setting, a multicenter case–control study including a total of 1149 patients was performed. We report normal values of sFlt-1, PlGF, and the sFlt-1/PlGF ratio based on the analysis of a total of 877 patients with uneventful pregnancy outcome. A total of 234 patients with preeclampsia and a matched cohort consisting of 468 patients with normal pregnancy outcome were compared, and sFlt-1 and PlGF were measured on an automated platform. Separate cutoffs for the sFlt-1/PlGF ratio were determined for the early (20+0–33+6 weeks) and the late gestational phase (34+0 weeks–delivery). For each of the 2 gestational phases, 2 independent cutoffs framing an equivocal zone were determined: the first cutoff with focus on high sensitivity, and the second focusing on high specificity. Between 20+0 and 33+6 weeks, the cutoffs at ≤33 and ≥85 resulted in a sensitivity/specificity of 95%/94% and 88%/99.5%, respectively. An sFlt-1/PlGF ratio of ≤33 had the lowest likelihood of a negative test (0.05; 95% confidence interval, 0.02–0.13), whereas values ≥85 had the highest likelihood of a positive test (176; 95% confidence interval, 24.88–1245). After 34+0 weeks, the cutoffs at ≤33 and ≥110 yielded a sensitivity/specificity of 89.6%/73.1% and 58.2%/95.5%, respectively. The approach to use multiple cutoffs for the early and late gestational phase enhances the diagnostic accuracy of the sFlt-1/PlGF ratio as a diagnostic tool for preeclampsia. (Hypertension. 2014;63:346-352.) • Online Data Supplement

Key Words: angiogenesis ◼ angiogenic factors ◼ antiangiogenic factors ◼ hypertension ◼ preeclampsia ◼ pregnancy ◼ sFlt1-PlGF ratio

Received June 15, 2013; first decision July 7, 2013; revision accepted October 1, 2013.From the Department of Obstetrics, Campus Virchow-Clinic, Charité University Medicine Berlin, Berlin, Germany (S.V., K.S.); Fetal Medicine Unit,

Department of Obstetrics and Gynecology, Hospital Universitario 12 de Octubre, Universidad Complutense de Madrid, Madrid, Spain (I.H., A.G.); Department of Obstetrics and Gynecology, University of Basel, Basel, Switzerland (O.L.); Department of Obstetrics and Gynecology, University of Jena, Jena, Germany (D.S.); Department of Obstetrics and Gynecology, University of Vienna Medical School, Vienna, Austria (H.Z.); Department of Obstetrics and Gynecology, Charles University, Prague, Czech Republic (P.C.); Department of Obstetrics and Gynecology, University Hospital Sant Joan de Déu, Espluges, Spain (J.S.); Department of Obstetrics and Gynecology, University of Freiburg, Freiburg, Germany (F.M.-E.); Roche Diagnostics, Clinical Operations Professional Diagnostics, Penzberg, Germany (B.D.); and Department of Obstetrics, University Hospital Leipzig, Leipzig, Germany (H.S.).

The online-only Data Supplement is available with this article at http://hyper.ahajournals.org/lookup/suppl/doi:10.1161/HYPERTENSIONAHA. 113.01787/-/DC1.

Correspondence to Stefan Verlohren, Department of Obstetrics, Charité University Medicine Berlin, Charitéplatz 1, D-10117 Berlin, Germany. E-mail [email protected]

New Gestational Phase–Specific Cutoff Values for the Use of the Soluble fms-Like Tyrosine Kinase-1/Placental Growth

Factor Ratio as a Diagnostic Test for PreeclampsiaStefan Verlohren, Ignacio Herraiz, Olav Lapaire, Dietmar Schlembach, Harald Zeisler,

Pavel Calda, Joan Sabria, Filiz Markfeld-Erol, Alberto Galindo, Katharina Schoofs, Barbara Denk, Holger Stepan

© 2013 American Heart Association, Inc.

Hypertension is available at http://hyper.ahajournals.org DOI: 10.1161/HYPERTENSIONAHA.113.01787

See Editorial Commentary, pp 210–211

Preeclampsia

by guest on August 26, 2015http://hyper.ahajournals.org/Downloaded from by guest on August 26, 2015http://hyper.ahajournals.org/Downloaded from by guest on August 26, 2015http://hyper.ahajournals.org/Downloaded from by guest on August 26, 2015http://hyper.ahajournals.org/Downloaded from by guest on August 26, 2015http://hyper.ahajournals.org/Downloaded from by guest on August 26, 2015http://hyper.ahajournals.org/Downloaded from by guest on August 26, 2015http://hyper.ahajournals.org/Downloaded from by guest on August 26, 2015http://hyper.ahajournals.org/Downloaded from by guest on August 26, 2015http://hyper.ahajournals.org/Downloaded from by guest on August 26, 2015http://hyper.ahajournals.org/Downloaded from by guest on August 26, 2015http://hyper.ahajournals.org/Downloaded from

Verlohren et al Preeclampsia and sFlt-1/PlGF Ratio 347

irrespective of the presence of a diagnosis of preeclampsia according to the gold standard.

With accumulating evidence of the importance of the sFlt-1/PlGF ratio as a diagnostic and prognostic marker, the limitations and benefits of its clinical use have to be assessed. Up to now, only 1 cutoff (85) exists, irrespective of the gestational week tested, with limited diagnostic accuracy especially in late-onset preeclampsia. This cutoff was based on a preliminary analysis of a multicenter case–control study consisting of 71 patients with preeclampsia and a matched cohort of 280 control patients.

Here, we report the final analysis of a multicenter case–con-trol study including a total of 1149 patients. We report normal values of sFlt-1, PlGF, and the sFlt-1/PlGF ratio based on the analysis of a total of 877 patients with uneventful pregnancy outcome. To better reflect the putatively different pathophysi-ology of early-onset and late-onset preeclampsia, diagnostic accuracy of 2 separate cutoffs framing an equivocal zone was determined for the early and late phase of gestation.

MethodsStudy PopulationSingleton pregnancies were enrolled at 9 European perinatal care centers. An identical study protocol and data collection form was used at each center. The local ethics committees and institutional review boards approved the study, and all subjects gave their written informed consent before participation.

Preeclampsia was defined according to the National High Blood Pressure Education Program Working Group on High Blood Pressure in Pregnancy.11 The syndrome of hemolysis, elevated liver enzymes, and low platelets (HELLP syndrome), superimposed preeclampsia, and in-trauterine growth restriction were defined as previously published.8 In this study, patients with any form of preeclampsia or HELLP were com-bined in the preeclampsia/HELLP group (preeclampsia/HELLP) for analysis. A pregnancy outcome was defined as normal if the mother was not diagnosed with any form of preeclampsia/HELLP and the infant was not diagnosed with intrauterine growth restriction. Investigators were blinded to sFlt-1 and PlGF levels which prevented an influence of this information on decision making and defining time point of delivery.

A total of 1149 individuals were enrolled in the study: 915 singleton pregnancies with normal pregnancy outcome and 234 singleton preg-nancies with preeclampsia outcome. At least 100 preeclampsia/HELLP cases for each early and late gestational phase were determined by study protocol as target numbers for the nested case–control popula-tion. Recruitment was performed in a prospective manner until these targets were obtained with the consequence of enrolling 1149 patients.

Two subsets of this data collective were built (Figure S1 in the online-only Data Supplement). For analysis of reference ranges, maximum 1 visit from each subject per gestational age window was included in case of follow-up visits in the control group. Therefore, this subset contains 1685 control visits from 877 subjects. An additional 38 control subjects either contributed samples only before week 10 or were not normoten-sive and were therefore excluded from the reference range analysis. For the determination of gestational age–dependent reference values, we analyzed 7 gestational age windows: window 1: 10+0 to 14+6 weeks of gestation; window 2: 15+0 to 19+6 weeks of gestation; window 3: 20+0 to 23+6 weeks of gestation; window 4: 24+0 to 28+6 weeks of gestation; window 5: 29+0 to 33+6 weeks of gestation; window 6: 34+0 to 36+6 weeks of gestation; and window 7: 37+0 weeks of gestation to delivery. All gestational age windows comprise 5 weeks except window 3 (20–23+6 weeks of gestation) and window 6 (34+0–36+6 weeks of gestation), acknowledging clinical thresholds at 24+0 (viability), 34+0 (early/late preeclampsia), and 37+0 weeks of gestation (preterm birth).

For receiver operating characteristics (ROC) curves and analysis of the cutoff, a total of 234 subjects with a preeclampsia/HELLP outcome were matched to 468 subjects with uneventful pregnancy outcomes. Only visits from subjects with a gestational age of ≥20+0 weeks were included. In patients with preeclampsia/HELLP, only the first visit after confirmation

of the diagnosis was included in the analysis. A preeclampsia pregnancy was defined as early onset if preeclampsia was diagnosed at a visit before week 34 (≤33+6 weeks of gestation). In case preeclampsia was diagnosed at >34+0 weeks of gestation, the patient was assigned to the late-onset preeclampsia group. Patients of the preeclampsia/HELLP group were matched pairwise by gestational week to a healthy control in a 1:2 manner, resulting in a double sample size of the control group. A control patient only donated samples either to the early- or the late-onset preeclampsia group, and no repeated samples were used in the case–control cohort.

In a separate descriptive subgroup analysis, patients with super-imposed preeclampsia (n=14), isolated HELLP (n=15), preeclampsia not being severe or complicated by HELLP syndrome (n=99), and patients with severe preeclampsia (n=106) were analyzed.

Samples and ImmunoassaysSerum samples were collected according to a common standard operat-ing procedure at each center. Single measurements were performed for sFlt-1 and PlGF on the fully automated Roche Elecsys system (Elecsys PlGF, human PlGF, and Elecsys sFlt-1, sFlt-1) as described previously, and the sFlt-1/PlGF ratio was calculated for each sample.12,13 Only serum samples were analyzed in this study, results in plasma samples may differ.

StatisticsBasic statistics (mean, median, SD, quartiles, and range) were performed for sFlt-1, PlGF, and the sFlt-1/PlGF ratio. To compare clinical sub-groups, descriptive statistics (median±interquartil ranges) were gener-ated. Gestational age–dependent reference values were determined per time window as quantiles. For the statistical comparison of marker levels in the respective clinical groups, data were transformed as appropriate and either parametric (ANOVA, t test) or nonparametric (Wilcoxon/Kruskal–Wallis) tests were applied. All P values are 2-tailed. Statistical significance was assigned when a P value was <0.05. All statistical analy-ses were performed using SAS. ROC analysis was used for the evaluation of the area under curve and the sensitivity and specificity as a function of cutoff for markers sFlt-1, PlGF, and sFlt-1/PlGF ratio. Positive and nega-tive likelihood ratios (LR+/LR−) were calculated. Gerhard plots were generated to find optimized cutoff values for the single markers and the ratio. A generalized linear regression model with a γ-distributed response regarding the positively skewed, continuous distribution was applied to test for confounding effects of body mass index and maternal age.

ResultsDemographic and Clinical Baseline CharacteristicsNo significant differences were found in age, height, and weight between patients with early- or late-onset preeclamp-sia and their gestational age–matched controls. As expected, patients with preeclampsia had a significantly earlier date of delivery, higher systolic and diastolic blood pressure, higher body mass index, and lower neonatal birth weight (P<0.05 or P<0.001 where appropriate, Table S1).

Gestational Age–Dependent Normal ValuesNormal values for the sFlt-1/PlGF ratio as well as the single markers were generated in 7 gestational age windows from women with uneventful pregnancy outcome (Table 1). As ex -pected, the sFlt-1/PlGF ratio showed an U-shaped curve with median sFlt-1 concentrations rising and PlGF values exhibiting a bell-shaped curve in the course of pregnancy (Figure S2a–S2c).

sFlt-1 and PlGF Values in Different Clinical Subsets of preeclampsiaIn patients with preeclampsia/HELLP, circulating serum concen-trations of sFlt-1 were elevated, PlGF levels were decreased, and the sFlt-1/PlGF ratio was increased when compared to patients with uneventful pregnancy outcome. The median sFlt-1/PlGF ratio in all patients with preeclampsia/HELLP was 185 (n=234;

by guest on August 26, 2015http://hyper.ahajournals.org/Downloaded from

348 Hypertension February 2014

25th–75th centile, 92.5–427) versus 7.78 (n=468; 25th–75th cen-tile, 3.29–23.9; P<0.001) in women with uneventful pregnancy outcome (Figure 1A, left). In patients with superimposed pre-eclampsia, the median sFlt-1/PlGF ratio was 75.4 (n=14; 25th–75th centile, 18.0–141); in patients with isolated HELLP, 420 (n=15; 25th–75th centile, 192–814); in patients with preeclamp-sia not being severe or complicated by HELLP syndrome, 116 (n=99; 25th–75th centile, 51.7–202); and in patients with severe preeclampsia, 350 (n=106; 25th–75th centile, 170–574; P<0.001 Kruskal–Wallis for subgroup comparison; Figure 1A, right).

The median sFlt-1/PlGF ratio in patients with early-onset pre-eclampsia was 424 (n=100; 25th–75th centile, 186–717) versus 129 (n=134; 25th–75th centile, 59–207; P<0.001) in late-onset cases. Patients with uncomplicated pregnancies ≤33+6 weeks of gestation (n=200) had a median sFlt-1/PlGF ratio of 3.68 (25th–75th centile, 2.03–7.50) versus 16.2 (25th–75th centile, 6.50–37.0; P<0.001) in control patients ≥34+0 weeks of gestation (n=268; Figure 1B).

Diagnostic Accuracy of sFlt-1, PlGF, and the sFlt-1/PlGF Ratio in Early- and Late-Onset preeclampsiaIn the ROC analysis, sFlt-1/PlGF ratio exhibited a superior per-formance to the single parameters in diagnosing preeclampsia

(Figure 2). For all patients with preeclampsia/HELLP, the ROC area under curve was 0.94 (95% confidence interval [CI], 0.92–0.96). For sFlt-1 ROC analysis showed a diagnostic accu-racy of 0.92 (95% CI, 0.89–0.94), whereas PlGF yielded an area under curve of only 0.91 (95% CI, 0.88–0.93) (Figure 2).

An optimized cutoff was evaluated based on the ROC anal-ysis. For the whole gestational phase, the optimized cutoff was found to be 34.1, resulting in a sensitivity of 91% and a specificity of 83.6%. When analyzing the preeclampsia/HELLP group divided by early- and late-onset preeclamp-sia, for early-onset preeclampsia an optimized cutoff of 38.5 resulted in a sensitivity of 93% and a specificity of 97%. For late-onset preeclampsia, a cutoff of 34.6 resulted in a sensitiv-ity of 88.8% and a specificity of 74.6% (Tables 2 and 3).

Definition of Gestational Phase–Specific Cutoff ValuesSeparate cutoffs were determined for the early gestational phase (20+0–33+6 weeks of gestation) and for the late gestational phase (34+0 weeks of gestation–delivery). Moreover, to enhance the diagnostic accuracy of the sFlt-1/PlGF ratio, for each of the gestational phases, an equivocal zone between 2 cutoff values

Table 1. Norm Values for the sFlt-1/PlGF Ratio, sFlt-1, and PlGF

Centiles

Completed Weeks of Gestation

10–14 15–19 20–23 24–28 29–33 34–36 ≥37

sFlt-1/PlGF ratio

Q 2.5 7.39 3.18 1.29 0.88 0.84 0.90 1.67

Q 5 9.27 3.51 1.82 0.95 0.94 1.23 2.18

Q 10 11.6 4.67 2.22 1.22 1.22 2.15 3.81

Median 24.8 10.5 4.92 3.06 3.75 9.03 19.6

Q 90 46.6 20.5 11.0 7.49 16.1 43.4 85.7

Q 95 54.6 25.7 14.6 10.0 33.9 66.4 112

Q 97.5 65.0 28.7 17.5 14.9 62.8 89.9 134

sFlt-1

Q 2.5 586 532 477 520 695 912 1262

Q 5 652 708 572 618 773 992 1533

Q 10 776 844 718 722 967 1220 1899

Median 1328 1355 1299 1355 1742 2552 3485

Q 90 2174 2453 2605 2557 3650 5620 7901

Q 95 2501 2807 2997 3205 5165 7363 9184

Q 97.5 2707 3382 3592 3912 5985 8214 11471

PlGF

Q 2.5 26.8 57.2 106 145 91 68.0 48.9

Q 5 28.8 66.2 119 169 114 78.0 54.4

Q 10 31.3 80.9 143 200 139 98.2 68.6

Median 52.6 135 264 465 471 284 191

Q 90 100 251 500 921 1073 831 620

Q 95 122 289 605 1117 1297 984 862

Q 97.5 136 322 694 1668 1567 1378 1006

n 246 157 217 346 319 224 176

Serum values of the sFlt-1/PlGF ratio, sFlt-1, and PlGF in patients with normal pregnancy outcome. The median as well as the centiles (Q) 2.5, 5, 10 and 90, 95, and 97.5 in the 7 gestational age windows sorted by completed weeks of gestation are displayed. n indicates number of patients; PlGF, placental growth factor; and sFlt-1, soluble fms-like tyrosine kinase-1.

by guest on August 26, 2015http://hyper.ahajournals.org/Downloaded from

Verlohren et al Preeclampsia and sFlt-1/PlGF Ratio 349

was established. For the early gestational phase, the rationale for choosing the cutoffs was to reach a ≥95% sensitivity (at the low cutoff) and ≥95% specificity (at the high cutoff; Figure 3). In this group, a cutoff of 33 resulted in a sensitivity of 95% and a specific-ity of 94%, corresponding to an LR− of 0.05 (95% CI, 0.02–0.13) and an LR+ of 15.8 (95% CI, 9.13–27.5). In total numbers, 95 of 100 patients were correctly classified as having preeclamp-sia, whereas 188 of 200 were correctly diagnosed as not having preeclampsia. However, in the same early phase, a cutoff of 85 resulted in a sensitivity of 88% and a specificity of 99.5% and yielded an LR+ of 176 (95% CI, 24.9–1245) and an LR− of 0.12 (95% CI, 0.07–0.21). All 200 patients bar one were correctly clas-sified as healthy, using the cutoff of 85. In combining the cutoffs ≤33/≥85, 95 of 100 preeclamptic patients were correctly classified and only 5 were incorrectly diagnosed as not being preeclamptic, whereas only 1 in 100 healthy pregnancies was incorrectly classi-fied as having preeclampsia (Tables 2 and 3).

For the late gestational phase, the cutoffs were chosen to reach a minimum of 95% specificity for the early cases of late-onset preeclampsia/HELLP (≥34–< 37 weeks of gesta-tion). Here, a cutoff of 33 resulted in a sensitivity of 89.6%, a

specificity of 73.1%, an LR− of 0.14 (95% CI, 0.09–0.24), and an LR+ 3.33 (95% CI, 2.71–4.10). However, a cutoff of 110 resulted in a sensitivity of 58.2% and a specificity of 95.5%, corresponding to an LR+ of 13 (95% CI, 7.34–23.0) and an LR− of 0.44 (95% CI, 0.36–0.54). Therefore, when combin-ing the cutoffs of ≤33/≥110, a sensitivity of 89.6% and a spec-ificity of 95.5% were reached (Tables 2 and 3).

DiscussionThe aim of the study was to provide reference ranges and cut-offs for the clinical use of the sFlt-1/PlGF ratio as an aid in diag-nosis of preeclampsia. Previous work from our group as well as from others has consistently shown that the automated mea-surement of sFlt-1, PlGF and the calculation of the sFlt-1/PlGF ratio is a reliable tool for the diagnosis of preeclampsia.8,10,13–15

We reported earlier a cutoff of 85 for the sFlt-1/PlGF ratio as an aid in diagnosis of preeclampsia regardless of the gesta-tional week tested. This preliminary cutoff was based on a case–control cohort of 71 patients with preeclampsia and 280 control patients.8 In the preliminary analysis, a cutoff of 85 resulted in a sensitivity of 82% and a specificity of 95% for diagnosing

Figure 1. A, Box-and-whisker plots showing the distribution of the calculated soluble fms-like tyrosine kinase-1 (sFlt-1)/placental growth factor (PlGF) ratio of patients with preeclampsia (PE)/hemolysis, elevated liver enzymes, and low platelets (HELLP) in 20 to 33+6 weeks of gestations vs gestational phase–matched controls (early gestational phase) and patients with PE/HELLP in ≥34+0 weeks of gestation vs gestational phase–matched controls (late gestational phase). The boxes represent the interquartile range, the whiskers represent the range, and the error bars represent the median. B, Box-and-whisker plots showing the distribution of the calculated sFlt-1/PlGF ratio of all patients (20 weeks of gestation–term) with PE/HELLP vs controls (non-PE/HELLP, left) and subgroup analysis of the PE/HELLP group: superimposed PE, HELLP without concurrent PE, PE (not severe, not complicated by HELLP syndrome), and severe PE (right). The boxes represent the interquartile range, the whiskers represent the range, and the error bars represent the median. *P<0.001 for the comparison PE/HELLP vs non-PE/HELLP and #P<0.001 multigroup comparison PE subgroups.*P<0.001 where appropriate. Early phase/onset: 20+0 to 33+6 weeks of gestation; late phase/onset: >34+0 weeks of gestation.

by guest on August 26, 2015http://hyper.ahajournals.org/Downloaded from

350 Hypertension February 2014

preeclampsia. For the subgroup early-onset preeclampsia, the same cutoff value (85) for the sFlt-1/PlGF ratio resulted in a sen-sitivity of 89% and a specificity of 97%. The cutoff was selected to generate a high specificity in a trade-off with sensitivity of the test.

After concluding the multicenter case–control study, we were now able to look at a total of 234 patients with pre-eclampsia and 915 control subjects. In-depth analysis of the whole study cohort revealed a different performance of the sFlt-1/PlGF ratio. Now, the single cutoff of 85 resulted in a sensitivity of 75.6% and a specificity of 95.5%. For the sub-group of early-onset preeclampsia, the single cutoff of 85 resulted in a sensitivity of 88% and a specificity of 99.5%.

We calculated optimized cutoffs based on the ROC analy-sis. For the whole gestational phase, a single, optimized cutoff of 34.1 resulted in a sensitivity of 91% and a specificity of 83.6%. Preeclampsia is a heterogeneous disease with marked differences in presentation and outcome especially between early-onset (≤33+6 weeks of gestation) and late-onset disease (≥34+0 weeks of gestation). The putatively different under-lying pathophysiological mechanisms of the early- and late-onset preeclampsia are reflected by a different performance of sFlt-1, PlGF, and the sFlt-1/PlGF ratio. We therefore aimed to adequately reflect the putative 2 diseases of preeclampsia in defining different cutoffs for the early- and late-onset disease.

Introducing 2 cutoffs for the different disease entities resulted in a higher accuracy in identifying the disease.

Gestational Phase–Dependent Cutoffs Framing an Equivocal Zone Enhance Diagnostic AccuracyTo accurately classify a patient with early-onset preeclampsia, it is necessary to have a high sensitivity. The number of false-negative test results has to be low in order not to miss a diseased patient. When applying the cutoff of 33 with a sensitivity of 95%, we would have correctly classified 95 of 100 patients in our cohort. The associated LR− is 0.05. On the contrary, the demand to accu-rately rule out the disease in patients with unspecific symptoms requires a high specificity. When applying the upper cutoff of 85, the specificity of 99.5 would correctly classify 199 of 200 patients as healthy. The corresponding LR+ is 176. Therefore, the use of the 2 cutoffs in our setting would result in a correct diagnosis of all bar 1 as not having preeclampsia and in an incorrect diagnosis of only 5 of 100 as falsely not having the disease. The equivocal zone contains 7% of the patients with preeclampsia/HELLP and 5.5% of the controls in the early gestational phase. However, the equivocal zone for late-onset preeclampsia contains 31.3% of the preeclampsia/HELLP and 22.4% of the control patients in the late gestational phase. The establishment of the equivocal zone allows for the identification of patients at intermediate risk. A precise diagnosis at the outer borders, combined with a necessity

Figure 2. Receiver operating characteristics curves for the discrimination of healthy and preeclamptic pregnancies by means of the soluble fms-like tyrosine kinase-1 (sFlt-1)/placental growth factor (PlGF) ratio (left), sFlt-1 (middle), and PlGF (right). The black line represents all patients, the green line represents patients from the early gestational phase, and the blue line represents patients from the late gestational phase. On the x axis, false positive/1−specificity, and on the y axis, true positive/sensitivity are expressed.

Table 2. Sensitivities and Specificities With 95% CIs for the Whole as Well as for Early and Late Gestational Phases

Whole Gestational Phase (Wk 20+0–Delivery) Early Gestational Phase (Wk 20+0–33+6) Late Gestational Phase (Wk 34+0–Delivery)

Sens/Spec Absolute Relative 95% CI Sens/Spec Absolute Relative 95% CI Sens/Spec Absolute Relative 95% CI

Cutoff 85 Cutoff 33 Cutoff 33

Sens 177/234 75.6% 70.6–100 Sens 95/100 95.0% 89.8–100 Sens 120/134 89.6% 84.2–100

Spec 447/468 95.5% 93.6–100 Spec 188/200 94.0% 90.5–100 Spec 196/268 73.1% 68.3–100

Cutoff 85 Cutoff 110

Sens 88/100 88.0% 81.3–100 Sens 78/134 58.2% 50.7–100

Spec 199/200 99.5% 97.7–100 Spec 256/268 95.5% 92.9–100

CI indicates confidence interval; Sens, sensitivities; and Spec, specificities. Bold values indicate the outer borders of the cut-off.

by guest on August 26, 2015http://hyper.ahajournals.org/Downloaded from

Verlohren et al Preeclampsia and sFlt-1/PlGF Ratio 351

of timely retesting in patients inside the equivocal zone, allows for maximum diagnostic safety.

LimitationsOur study has limitations. We have performed a case–control study with a 2:1 matching in the ROC group. Because of the design, with a constructed incidence of preeclampsia of 33.3%, we were not able to calculate positive or negative predictive val-ues. Individual cutoffs could be established only for 2 gestational phases. The data set is too small to establish cutoffs for shorter time intervals, which resulted in a sudden change of cutoffs at 34+0 weeks of gestation. For the late gestational age group, there is a significant proportion of patients inside the equivocal zone. Almost 30% of the patients with preeclampsia/HELLP and almost 25% of the controls need to be retested. For these patients, data are lacking on when and how often they have to be retested. We evaluated singleton pregnancies in this study, there-fore our results only apply to this group of patients. It is known, however, that multifetal gestations exhibit different patterns of angiogenic factor serum levels with healthy twin pregnancies exhibiting a 3-fold increase in the sFlt-1/PlGF ratio as compared

with singleton pregnancies.16 Therefore, caution must be taken when applying the reference ranges and cutoffs found in this study to multifetal gestations. Furthermore, it is noteworthy that only a small proportion of patients in this European study are of other than white origin. Therefore, results may vary in popula-tions with a different racial background and thus different disease prevalence. Further studies are needed to clarify these points.

Clinical ValueThe clinical presentation of preeclampsia is diverse, and the diagnosis of atypical cases of preeclampsia is a daily challenge to the practicing obstetrician.17 Current definitions and guide-lines rely solely on the measurement of blood pressure and pro-teinuria to diagnose the disease.2,18 However, the measurement of blood pressure and proteinuria has low predictive accuracy of adverse maternal and fetal outcomes. Recently, Rana et al9 have shown that in women with suspected preeclampsia, the sFlt-1/PlGF ratio helps to identify those who will develop a pre-eclampsia-related pregnancy complication. In addition to the gold standard of preeclampsia diagnostics, the measurement of blood pressure and proteinuria, the sFlt-1/PlGF ratio is able to improve prediction of adverse preeclampsia-related outcomes.9

Various groups have shown that 1 single cutoff at 85 is not optimal for diagnosing preeclampsia or other hypertensive preg-nancy disorders in all clinical settings.14,19,20 The new gestational phase–specific 2 cutoffs allow maximized accuracy of diagno-sis. A more precise diagnosis and gestational phase–dependent analysis might have future therapeutic consequences. Recently, a pilot study has shown a possible therapy for women with early-onset preeclampsia. Preeclamptic women who underwent sFlt-1 apheresis had longer remaining pregnancy duration.21 Removal of sFlt-1 resulted in amelioration of kidney function and prolongation of pregnancy. Therefore, the sFlt-1/PlGF ratio might have importance as a parameter to monitor disease sever-ity and hence therapeutic progress.

Figure 3. Gerhard plots displaying the decision limits for ruling in and out as a basis for finding cutoffs and equivocal zone. Values were taken from receiver operating characteristics analysis: the light blue line represents the sensitivity, and the dark blue line represents the specificity. Left, The localization of the present cutoff of 85 in the Gerhard plot for the whole gestational phase. Middle, The gestational phase–specific cutoffs for the early phase. Right, The gestational phase–specific cutoffs for the late phase. The color fills are chosen to represent interpretations of the test result: below the (lower) cutoff, green indicates low risk; above the (upper) cutoff, red stands for high risk; inside the equivocal zone, yellow indicates intermediate risk.

Table 3. Positive and Negative Likelihood Ratios (LR+/LR−) With 95% CIs for Early and Late Gestational phases

Early Gestational Phase (Wk 20+0–33+6)

Late Gestational Phase (Wk 34+0–Delivery)

LR Estimate 95% CI LR Estimate 95% CI

Cutoff 33 Cutoff 33

LR+ 15.8 9.13–27.5 LR+ 3.33 2.71–4.10

LR− 0.05 0.02–0.13 LR− 0.14 0.09–0.24

Cutoff 85 Cutoff 110

LR+ 176 24.9–1245 LR+ 13 7.34–23.0

LR− 0.12 0.07–0.21 LR− 0.44 0.36–0.54

CI indicates confidence interval; LR, likelihood ratio. Bold values indicate the outer borders of the cut-off.

by guest on August 26, 2015http://hyper.ahajournals.org/Downloaded from

352 Hypertension February 2014

PerspectivesThe use of an equivocal zone and focus on the outer borders of this zone to rule in or rule out patients with preeclampsia enhances the diagnostic accuracy of the sFlt-1/PlGF ratio with the potential to reduce maternal and fetal morbidity and mortal-ity. Below 34 weeks of gestation, an sFlt-1/PlGF ratio of ≤33 is associated with an LR− of 0.05, whereas values ≥85 have a like-lihood of a positive test of 176. For late-onset preeclampsia, the LR− at the cutoff ≤33 is 0.14, the LR+ at ≥110 is 13. Looking at the clinical feasibility, a simple diagnostic algorithm is desirable for the user. However, the establishment of differential cutoffs for different gestational phases instead of using 1 single cutoff value better reflects the pathophysiology of the disease. In our opinion, this approach of gestational phase–specific cutoff values provides a reasonable balance/compromise between both needs.

Sources of FundingP. Calda has received funding by grant RVO-VFN64165/2012. The study was supported by Roche Diagnostics, Germany.

DisclosuresB. Denk is employed by Roche Diagnostics, Penzberg, Germany. H. Stepan has received consultancy payments from Roche regarding ad-vice on clinical trial design. S. Verlohren has received consultancy payments from Roche regarding advice on data analysis and publica-tion. S. Verlohren and H. Stepan received lecture fees from Roche. The other authors report no conflicts.

References 1. Steegers EA, von Dadelszen P, Duvekot JJ, Pijnenborg R. Pre-eclampsia.

Lancet. 2010;376:631–644. 2. Brown MA, Lindheimer MD, de Swiet M, Van Assche A, Moutquin JM.

The classification and diagnosis of the hypertensive disorders of pregnancy: statement from the International Society for the Study of Hypertension in Pregnancy (ISSHP). Hypertens Pregnancy. 2001;20:IX–XIV.

3. Verlohren S, Stepan H, Dechend R. Angiogenic growth factors in the diag-nosis and prediction of pre-eclampsia. Clin Sci (Lond). 2012;122:43–52.

4. Egbor M, Ansari T, Morris N, Green CJ, Sibbons PD. Morphometric placen-tal villous and vascular abnormalities in early- and late-onset pre-eclamp-sia with and without fetal growth restriction. BJOG. 2006;113:580–589.

5. Soto E, Romero R, Kusanovic JP, Ogge G, Hussein Y, Yeo L, Hassan SS, Kim CJ, Chaiworapongsa T. Late-onset preeclampsia is associated with an imbalance of angiogenic and anti-angiogenic factors in patients with and without placental lesions consistent with maternal underperfusion. J Matern Fetal Neonatal Med. 2012;25:498–507.

6. Levine RJ, Maynard SE, Qian C, Lim KH, England LJ, Yu KF, Schisterman EF, Thadhani R, Sachs BP, Epstein FH, Sibai BM, Sukhatme VP, Karumanchi SA. Circulating angiogenic factors and the risk of pre-eclampsia. N Engl J Med. 2004;350:672–683.

7. Levine RJ, Lam C, Qian C, Yu KF, Maynard SE, Sachs BP, Sibai BM, Epstein FH, Romero R, Thadhani R, Karumanchi SA; CPEP Study Group. Soluble endoglin and other circulating antiangiogenic factors in preeclampsia. N Engl J Med. 2006;355:992–1005.

8. Verlohren S, Galindo A, Schlembach D, Zeisler H, Herraiz I, Moertl MG, Pape J, Dudenhausen JW, Denk B, Stepan H. An automated method for the determination of the sFlt-1/PIGF ratio in the assessment of preeclamp-sia. Am J Obstet Gynecol. 2010;202:161.e1–161.e11.

9. Rana S, Powe CE, Salahuddin S, Verlohren S, Perschel FH, Levine RJ, Lim KH, Wenger JB, Thadhani R, Karumanchi SA. Angiogenic factors and the risk of adverse outcomes in women with suspected preeclampsia. Circulation. 2012;125:911–919.

10. Verlohren S, Herraiz I, Lapaire O, Schlembach D, Moertl M, Zeisler H, Calda P, Holzgreve W, Galindo A, Engels T, Denk B, Stepan H. The sFlt-1/PlGF ratio in different types of hypertensive pregnancy disor-ders and its prognostic potential in preeclamptic patients. Am J Obstet Gynecol. 2012;206:58.e1–58.e8.

11. Report of the National High Blood Pressure Education Program Working Group on High Blood Pressure in Pregnancy. Am J Obstet Gynecol. 2000;183:S1–S22.

12. Schneider E, Gleixner A, Hänel R, Leyhe Y, Kleinschmidt C, Beck G, Steinberg M, Denk B, Gassner D. Technical performance of the first fully automated assays for soluble fms-like tyrosine kinase 1 and human pla-cental growth factor. Z Geburtsh Neonatol. 2009;213:69.

13. Ohkuchi A, Hirashima C, Suzuki H, Takahashi K, Yoshida M, Matsubara S, Suzuki M. Evaluation of a new and automated electrochemilumines-cence immunoassay for plasma sFlt-1 and PlGF levels in women with preeclampsia. Hypertens Res. 2010;33:422–427.

14. Schiettecatte J, Russcher H, Anckaert E, Mees M, Leeser B, Tirelli AS, Fiedler GM, Luthe H, Denk B, Smitz J. Multicenter evaluation of the first automated Elecsys sFlt-1 and PlGF assays in normal pregnancies and pre-eclampsia. Clin Biochem. 2010;43:768–770.

15. Stepan H, Unversucht A, Wessel N, Faber R. Predictive value of maternal angiogenic factors in second trimester pregnancies with abnormal uterine perfusion. Hypertension. 2007;49:818–824.

16. Rana S, Hacker MR, Modest AM, Salahuddin S, Lim KH, Verlohren S, Perschel FH, Karumanchi SA. Circulating angiogenic factors and risk of adverse maternal and perinatal outcomes in twin pregnancies with sus-pected preeclampsia. Hypertension. 2012;60:451–458.

17. Sibai BM, Stella CL. Diagnosis and management of atypical preeclamp-sia-eclampsia. Am J Obstet Gynecol. 2009;200:481.e1–481.e7.

18. ACOG Committee on Practice Bulletins--Obstetrics. ACOG practice bul-letin. Diagnosis and management of preeclampsia and eclampsia. Number 33, January 2002. Obstet Gynecol. 2002;99:159–167.

19. Gómez-Arriaga PI, Herraiz I, López-Jiménez EA, Gómez-Montes E, Denk B, Galindo A. Uterine artery Doppler and sFlt-1/PlGF ratio: usefulness in diagnosis of pre-eclampsia. Ultrasound Obstet Gynecol. 2013;41:530–537.

20. Engels T, Pape J, Schoofs K, Henrich W, Verlohren S. Automated measure-ment of sFlt1, PlGF and sFlt1/PlGF ratio in differential diagnosis of hyper-tensive pregnancy disorders. Hypertens Pregnancy. 2013;32:459–473.

21. Thadhani R, Kisner T, Hagmann H, et al. Pilot study of extracorporeal removal of soluble fms-like tyrosine kinase 1 in preeclampsia. Circulation. 2011;124:940–950.

What Is New?•An imbalance of angiogenic and antiangiogenic factors is involved in the

pathophysiology of preeclampsia. •Automated measurement of antiangiogenic soluble fms-like tyrosine ki-

nase-1 and angiogenic placental growth factor (PlGF) and the calculation of the soluble fms-like tyrosine kinase-1/placental growth factor ratio allows for detection of the hypertensive pregnancy disorder.

What Is Relevant?•Up to now only 1 cutoff for the clinical use of the soluble fms-like tyrosine

kinase-1/placental growth factor ratio as an aid in diagnosis has been evaluated.

• In the final evaluation of a multicenter study, we show here that the use of 2 gestational phase–specific cutoffs framing an equivocal zone improved the diagnostic accuracy of the test.

SummaryThe use of multiple cutoffs better reflects the pathophysiological and clinical difference of early and late preeclampsia and might thus improve clinical diagnostics and management.

Novelty and Significance

by guest on August 26, 2015http://hyper.ahajournals.org/Downloaded from

Online supplement HYPE201301787: New gestational phase specific cut-off values for the use of the sFlt-1/PlGF-ratio as a diagnostic test for preeclampsia Stefan VERLOHREN, MD1; Ignacio HERRAIZ, MD2; Olav LAPAIRE, MD3; Dietmar SCHLEMBACH, MD4; Harald ZEISLER, MD5; Pavel CALDA, MD6; Joan SABRIA, MD7; Filiz MARKFELD-EROL, MD8; Alberto GALINDO, MD2; Katharina SCHOOFS, MD1; Barbara DENK, PhD9 and Holger STEPAN, MD10 Supplementary Methods: Patients: Briefly, hypertension was defined as the repeated measurement of systolic ≥ 140 mmHg (Korotkoff I) and diastolic ≥ 90 mmHg (Korotkoff V) blood pressure. Proteinuria was defined as the excretion of ≥300 mg protein per day in the 24 hour (h) urine collection or a repeated dipstick of ≥1+. PE was defined as severe in contrast to mild, when hypertension ≥160/≥110 mmHg, proteinuria ≥5g/24h or the occurrence of organ failure (kidney, lung) or neurologic symptoms were observed. Inclusion criteria for the reference range cohort were the presence of informed consent, maternal age ≥16 years, and a normal pregnancy outcome. A normal pregnancy outcome refers to the mother not diagnosed with any form of PE and the infant not diagnosed with intrauterine growth retardation (IUGR). The exclusion criteria comprise the diagnoses of PE, HELLP syndrome, IUGR, and all those for the PE group as stated below. Samples Maternal blood was collected by venipuncture in tubes without anticoagulant. After clotting, the samples were centrifuged with 2000g and serum was pipetted, and stored at –80°C until testing. The sFlt-1 and PIGF concentrations of each sample were determined in parallel. Statistics Gestational age-dependent reference values were determined per time window as quantiles according to the respective protocols by International Federation of Clinical Chemistry and Laboratory Medicine (IFCC) and Clinical and Laboratory Standards Institute (CLSI),Smooth curves for gestational age-dependent quantiles were determined using a robust quantile regression method.

1

Table S1 Patient Characteristics

PE/HELLP & 20+0-33+6

controls & 20+0-33+6

PE/HELLP & >34+0

controls & >34+0

Age [y] – median (IQR) 32.0 (28.0 - 35.0) 31.0 (27.0 - 35.0) 30.5 (25.0 - 35.0) 31.0 (27.0 - 34.0) Height [cm] – median (IQR) 164 (160 - 170) 165 (160 - 169) 165 (160 - 169) 165 (161 - 170) Weight [kg] – median (IQR) 68.0 (56.0 - 81.0) 63.0 (57.0 - 75.0) 66.5 (60.0 - 77.2) 64.0 (56.0 - 76.8) BMI [kg/m^2] – median (IQR) 24.2 (21.5 - 29.9)* 23.0 (20.6 - 27.9) 24.5 (21.6 - 29.0)* 23.5 (20.7 - 27.1)

Birth Weight [g] – median (IQR) 1270 (861 - 1670)**

3125 (2779 - 3490)

2800 (2200 - 3291)** 3305 (2939 - 3665)

Gestational Week of Delivery – median (IQR) 31.1 (27.9 - 33.2)** 38.6 (37.4 - 39.9) 37.9 (35.9 - 39.4)** 39.4 (38.4 - 40.4) Max. diastolic BP [mmHg] – median (IQR) 100 (90.0 – 110)** 73.0 (70.0 - 80.0) 95.0 (90.0 - 101)** 78.0 (70.0 - 85.0) Max. systolic BP [mmHg] – median (IQR) 160 (149 - 180)** 120 (110 - 130) 150 (142 - 161)** 123 (113 - 135) Smoking: Current [N (%)] 7 (7%) 29 (14.5%) 10 (7.5%) 41 (15.3%) Smoking: Past [N (%)] 14 (14%) 30 (15%) 19 (14.2%) 42 (15.7%) Smoking: Never [N (%)] 69 (69%) 128 (64%) 82 (61.2%) 147 (54.9%) Smoking: Unknown / NA [N (%)] 10 (10%) 13 (6.5%) 23 (17.2%) 38 (14.2%) Race: White / Caucasian [N (%)] 82 (82%) 173 (86.5%) 107 (79.9%) 229 (85.4%) Race: Black / African American [N (%)] 5 (5%) 5 (2.5%) 8 (6%) 8 (3%) Race: Other [N (%)] 11 (11%) 20 (10%) 13 (9.7%) 15 (5.6%) Race: Unknown / NA [N (%)] 2 (2%) 2 (1%) 6 (4.5%) 16 (6%) Family History of PE: Yes [N (%)] 9 (9%) 6 (3%) 7 (5.2%) 9 (3.4%) Family History of PE: No [N (%)] 81 (81%) 163 (81.5%) 105 (78.4%) 228 (85.1%) Family History of PE: Unknown / NA [N (%)] 10 (10%) 31 (15.5%) 22 (16.4%) 31 (11.6%) N 100 200 134 268

2

Legend: Baseline Characteristics: PE/HELLP patients 20+0 – 33+6 wks and ≥34+0 wks were compared to gestational age matched controls (* p<0.05). Data is expressed in median and interquartile range (IQR). BMI = body mass index, BP = blood pressure, NA = not applicable. IQR (interquartilrange): 25. percentile – 75. Percentile, * p<0.05 vs. gestational phase matched controls, **p<0.001 vs. gestational phase matched controls

3

Figure  1      

1149  (2403) Singleton  

pregnancies

   234  (308)                PE/HELLP  

outcome            

Case-­‐Control  Study

Reference  Range  Study

877  (1685) normotensive,  gest.  wk  ≥  10 multiple    visits    allowed

but  max.  1  visit  per  window

268 late  phase

200 early  phase  

134 late  onset

100 early  onset  

Case-­‐Control  Study 1.  visit  w

ith  diagnosis  

915  (2095) normal  pregnancy  

outcome

468 1  visit  per  patient

matched  for  gest.  wk

38 not  normotensive gest.  wk    ≤  10

  gest  age  window  1 gest  age  window  2 gest  age  window  3 gest  age  window  4 gest  age  window  5 gest  age  window  6 gest  age  window  7

Figure S1

4

Legend: Study Design: A total of 1149 patients yielding 2403 visits were analyzed in two separate study collectives. For the establishment of reference ranges 915 singleton pregnancies with normal pregnancy outcome yielding 2095 visits were analyzed. After exclusion of 38 patients due to not being normotensive or yielding samples before wk 10, a total of 877 patients yielding 1685 visits remained. Analysis was undertaken in seven gestational age windows (gest age window): Window 1: 10+0 - 14+6 week of gestation (wks), window 2: 15+0 - 19+6 wks, window 3: 20+0 - 23+6 wks, window 4: 24+0 - 28+6 wks, window 5: 29+0 - 33+6 wks, window 6: 34+0 - 36+6 wks and window 7: 37+0 wks – delivery. At most one visit from each subject per gestational age window was included in case of follow-up visits in the control group. For receiver operating characteristics curves (ROC) and analysis of the cut-off, only visits with a subjects' gestational age of at least 20 weeks were included. A total of 234 subjects with a PE/HELLP-outcome was matched to 468 subjects with uneventful pregnancy outcome. Legend: early phase / onset: 20+0 – 33+6 wks; late phase / onset: >34+0 wks.

Figure S2 a

S8 6

Figure S2 b

7

Figure S2 c

8

Legend: Scatter plots of maternal serum concentrations of sFlt-1 (a), PlGF (b) and calculated sFlt-1/PlGF-ratio (c) of uncomplicated pregnancies (controls). The dots represent individual serum values, the smooth curves represent the median, 2.5 / 5 / 10 as well as 90 / 95 / 97.5 centile of the serum values of sFlt-1, PlGF and the sFlt-1/PlGF-ratio.  

Holger StepanandCalda, Joan Sabria, Filiz Markfeld-Erol, Alberto Galindo, Katharina Schoofs, Barbara Denk

Stefan Verlohren, Ignacio Herraiz, Olav Lapaire, Dietmar Schlembach, Harald Zeisler, PavelTyrosine Kinase-1/Placental Growth Factor Ratio as a Diagnostic Test for Preeclampsia

Specific Cutoff Values for the Use of the Soluble fms-Like−New Gestational Phase

Print ISSN: 0194-911X. Online ISSN: 1524-4563 Copyright © 2013 American Heart Association, Inc. All rights reserved.

is published by the American Heart Association, 7272 Greenville Avenue, Dallas, TX 75231Hypertension doi: 10.1161/HYPERTENSIONAHA.113.017872014;63:346-352; originally published online October 28, 2013;Hypertension. 

http://hyper.ahajournals.org/content/63/2/346World Wide Web at:

The online version of this article, along with updated information and services, is located on the

http://hyper.ahajournals.org/content/suppl/2013/10/28/HYPERTENSIONAHA.113.01787.DC1.htmlData Supplement (unedited) at:

  http://hyper.ahajournals.org//subscriptions/

is online at: Hypertension Information about subscribing to Subscriptions: 

http://www.lww.com/reprints Information about reprints can be found online at: Reprints:

  document. Permissions and Rights Question and Answer this process is available in the

click Request Permissions in the middle column of the Web page under Services. Further information aboutOffice. Once the online version of the published article for which permission is being requested is located,

can be obtained via RightsLink, a service of the Copyright Clearance Center, not the EditorialHypertensionin Requests for permissions to reproduce figures, tables, or portions of articles originally publishedPermissions:

by guest on August 26, 2015http://hyper.ahajournals.org/Downloaded from