Research www. AJOG.

org

OBSTETRICS
Diagnostic utility of soluble fms-like tyrosine kinase 1 and
soluble endoglin in hypertensive diseases of pregnancy
Saira Salahuddin, MD; Young Lee, MD; Mary Vadnais, MD; Benjamin P. Sachs, MB, BS, DPH; S. Ananth Karumanchi, MD;
Kee-Hak Lim, MD

OBJECTIVE: The objective of this pilot study was to evaluate the clin-
ical utility of soluble fms-like tyrosine kinase 1 (sFlt 1) and soluble
endoglin (sEng) in the differential diagnosis of hypertension in late
pregnancy.
STUDY DESIGN: We analyzed serum levels of sFlt 1 and sEng in
women with gestational hypertension (GHTN; n ⫽ 17), chronic hyper-
tension (CHTN; n ⫽ 19), preeclampsia (n ⫽ 19), and normal preg-
nancy (n ⫽ 20) in the third trimester. We calculated the sensitivity,
specificity, and positive and negative likelihood ratio (LR) for each fac-
tor in diagnosing preeclampsia.
RESULTS: The sensitivity and specificity of sFlt 1 in differentiating pre-
eclampsia from normal pregnancy were 90% and 90%, respectively,
and 90% and 95% for sEng. In women with GHTN, they were 79% and
88% for sFlt 1; 84% and 88% for sEng; 90% and 63% for uric acid. In
women with CHTN, they were 84% and 95% for sFlt 1; 84% and 79%
for sEng; 68%; and 78% for uric acid. The positive LR for preeclamp-
sia was 9 for sFlt 1 and 7 for sEng in women with normal pregnancy; in
women with GHTN; 6.7 for sFlt 1 and 7.2 for sEng; in CHTN, 16 for sFlt
1 and 4 for sEng. Serum uric acid had a positive LR of only 2.4 in
women with GHTN and 3.1 in women with CHTN.
CONCLUSION: Both sFlt 1 and sEng may prove useful in differentiating
preeclampsia from other hypertensive diseases of pregnancy. A pro-
spective cohort study should be performed determine the clinical utility
of measuring these proteins.
Key words: hypertension in pregnancy, preeclampsia, soluble
endoglin, soluble fms-like tyrosine kinase 1

Cite this article as: Salahuddin S, Lee Y, Vadnais M, et al. Diagnostic utility of soluble fms-like tyrosine kinase 1 and soluble endoglin in hypertensive diseases
of pregnancy. Am J Obstet Gynecol 2007;197;28.e1-28.e6.

P reeclampsia occurs in 2-7% of preg-

nancies1,2 and is a leading cause of
maternal and neonatal morbidity and cause the onset of hypertension and pro-
Recently, it has been shown that 2 an-
tiangiogenic peptides produced by the
placenta, soluble fms-like tyrosine ki-
mortality.3,4 The diagnosis is based on teinuria can be variable and the disorder nase 1 (sFlt 1) and soluble endoglin
the onset of hypertension and protein- may present without the pathogno- (sEng), contribute to the pathogenesis of
uria, usually after 20 weeks’ gestation as monic signs and symptoms, differentiat- preeclampsia.7-11 By sequestering the
well as a sudden rise in blood pressure ing preeclampsia from other forms of proangiogenic proteins, vascular endo-
and the appearance or aggravation of hypertensive diseases of pregnancy can thelial growth factor, and placental
proteinuria in women with known be challenging and time consuming, of- growth factor, sFlt-1 has been shown to
chronic hypertension.5,6 However, be- ten delaying appropriate care. cause hypertension, proteinuria, and
glomerular endotheliosis in rats.7 En-
doglin, a coreceptor for transforming
From the Departments of Obstetrics, Gynecology, and Reproductive Biology (Drs
Salahuddin, Vadnais, Sachs, Karumanchi, and Lim) and Medicine (Dr Karumanchi), Beth growth factor (TGF)-␤1 and TGF-␤3,
Israel Deaconess Medical Center, Harvard Medical School, Boston, MA, and the has been shown to be upregulated in the
Department of Obstetrics and Gynecology, St. Mary’s Hospital, College of Medicine, The placenta in preeclampsia, leading to in-
Catholic University of Korea, Seoul, Korea (Dr Lee). creased secretion of the soluble form into
Received Aug. 28, 2006; revised Dec. 9, 2006; accepted April 12, 2007. maternal circulation.8 We have shown
Reprints not available from the authors. that the serum level of sEng is increased
This work was supported in part by the Obstetrics and Gynecology Foundation, Beth Israel in patients with preeclampsia and corre-
Deaconess Medical Center (Boston, MA). lates with the disease severity. In addi-
S.A.K. is a coinventor on multiple patents filed by the Beth Israel Deaconess Medical Center for tion, sEng, in the presence of elevated
the diagnosis of preeclampsia. S.A.K. is also a consultant to Abbott Diagnostics, Beckman sFlt 1, has been shown to cause hemoly-
Center, and Johnson & Johnson. sis, hepatic ischemia, and necrosis, and
0002-9378/$32.00 extensive vascular damage of the pla-
© 2007 Mosby, Inc. All rights reserved.
doi: 10.1016/j.ajog.2007.04.010
centa, including infarction at the mater-
nal–fetal junction as well as signs of se-
See related editorial, page 1 vere maternal vascular damage in rats.8
In humans, the rise in serum levels of

28.e1 American Journal of Obstetrics & Gynecology JULY 2007

www.AJOG.org Obstetrics Research

TABLE 1
Admission demographics
Preeclampsia
Control (n ⴝ 20) GHTN (n ⴝ 17) CHTN (n ⴝ 19) (n ⴝ 19) P (by ANOVA)
Age (y) 33.7 ⫾ 5.3 31.2 ⫾ 5.5 35.2 ⫾ 6.3 30.6 ⫾ 5.6 .05
................................................................................................................................................................................................................................................................................................................................................................................
BMI 26.9 ⫾ 6.0 27.6 ⫾ 12 31.8 ⫾ 10 28.1 ⫾ 5.4 .40
................................................................................................................................................................................................................................................................................................................................................................................
SBP (mm Hg) 118.4 ⫾ 9.5 145.4 ⫾ 8.0 142.2 ⫾ 17 146.0 ⫾ 16.5 ⬍.001*
................................................................................................................................................................................................................................................................................................................................................................................
DBP (mm Hg) 74.7 ⫾ 8.2 93.0 ⫾ 9.4 92.6 ⫾ 26 91.8 ⫾ 10 .001*
................................................................................................................................................................................................................................................................................................................................................................................
GA (wks) 39.1 ⫾ 1.3 36.4 ⫾ 1.9 35.7 ⫾ 3.2 34.6 ⫾ 3.3 .05
................................................................................................................................................................................................................................................................................................................................................................................
Hematocrit (%) 34.8 ⫾ 3.0 34.2 ⫾ 4.2 34.6 ⫾ 3.0 35.1 ⫾ 3.3 .9
................................................................................................................................................................................................................................................................................................................................................................................
BMI, body mass index; CHTN, chronic hypertension; DBP, diastolic blood pressure; GA, gestational age; GHTN, gestational hypertension; SBP, systolic blood pressure.
* Compared with control.

these 2 peptides seems to precede the on- lege of Obstetrics and Gynecology (R&D Systems Inc, Minneapolis, MN).
set of hypertension and proteinuria by a (ACOG) practice bulletin (Jan. 2002).5 Briefly, various samples for ELISA mea-
few weeks9,11 and correlates with severity For proteinuria, 0.3 g protein or higher surement were diluted in respective cal-
of the disease. in a 24-hour urine specimen was used as ibrator diluent. After adding assay di-
Elevated serum levels of both sFlt 1 much as possible. When the 24-hour luent and the diluted sample in a 96-well
and sEng have been documented in pre- urine was not available, either a dipstick plate precoated with captured antibodies
eclampsia and gestational hypertension of ⫹1 or greater or a protein/creatinine directed against human sFlt-1 or human
retrospectively in samples stored for ratio of 0.3 or greater was used. We used endoglin, the plates were incubated for 2
more than a decade,9,11 and the clinical the criteria outlined by ACOG rather hours. The wells were washed 4 times in
utility of these 2 factors in differentiating than more stringent criteria of ⫹2 dip- wash buffer and incubated with second-
the 2 forms of hypertensive diseases in stick because we were interested in the ary polyclonal antibody against sFlt-1
pregnancy has not been examined. Fur- performance of these peptides in a typi- and endoglin conjugated to horseradish
thermore, we could not locate reports of cal clinical setting. peroxidase for an additional 2 hours.
serum levels of these peptides in women Severe preeclampsia was diagnosed if 1 The plates were then washed 4 times in
with chronic hypertension in the third or more of the following was present: (1) wash buffer.
trimester. Therefore, we performed this blood pressure of 160 mm Hg or greater Substrate solution containing hydro-
pilot study in advance of larger ones to systolic or greater than 110 mm Hg dia- gen peroxide and tetramethylbenzadine
evaluate the sensitivity, specificity, and stolic on 2 occasions at least 6 hours were added to each well and incubated
positive and negative likelihood ratios apart while on bed rest; (2) proteinuria for 30 minutes under protection from
(LRs) of sFlt1 and sEng in patients with of 5 g or higher in a 24-hour urine col- light. Stop solution was added to each
various hypertensive diseases in lection or greater than positive 3 on 2 well. The optical density was then deter-
pregnancy. random urine samples collected at least 4 mined by subtracting readings at 540 nm
hours apart; (3) oliguria of less than 500 from the reading at 450 nm. All assays
M ATERIALS AND M ETHODS mL in 24 hours; (4) cerebral or visual dis- were performed in duplicate, and the
We performed a prevalence case-control turbances; (5) pulmonary edema or cya- protein levels were calculated using a
study measuring serum sFlt 1 and sEng nosis; (6) epigastric or right upper-quad- standard curve derived from a known
in consenting pregnant women present- rant pain; (7) impaired liver function; concentration of respective recombinant
ing to labor and delivery at Beth Israel (8) thrombocytopenia; or (9) fetal proteins. The minimum detectable doses
Deaconess Medical Center (Boston, growth restriction.5 Pregnant women in the assay for sFlt-1 and endoglin were
MA) from 2004-2006 with hypertension without hypertension, proteinuria, or 3.5 and 7 pg/mL, respectively, with in-
in the third trimester. The Committee on underlying medical conditions were re- traassay and interassay coefficients of
Clinical Investigations at Beth Israel cruited as controls. variation of 3.5% and 5.5%, respectively,
Deaconess Medical Center approved the The serum samples were stored at for sFlt-1 and 3.2% and 6.5%, respec-
study (protocol 2003-P-000342/5; Oct. –70oC in a freezer before use. Assays tively, for endoglin.
14, 2004). The pregnancy outcomes were were performed by personnel who were The data were analyzed using analysis of
analyzed by reviewing each case and unaware of the outcome of the preg- variance (ANOVA) and Student t test as
were categorized as gestational hyper- nancy. Enzyme-linked immunosorbent appropriate using the Systat 8.0 program
tension, chronic hypertension, or pre- assays (ELISAs) for sFlt 1 and endoglin (SPSS Inc, Chicago, IL). Receiver operator
eclampsia, according to the diagnostic were performed with commercially characteristic (ROC) curves were drawn
criteria outlined by the American Col- available kits, as previously described and a full analysis performed to detect the

JULY 2007 American Journal of Obstetrics & Gynecology 28.e2

Research Obstetrics
highest sensitivity, specificity, positive
(sensitivity/[1-specificity]) and negative
([1-sensitivity]/specificity) likelihood ra-
tios using MedCalc software (Broekstraat,
Mariakerke, Belgium).
R ESULTS
Nineteen women with preeclampsia, 17
women with gestational hypertension,
and 19 women with chronic hyperten-
sion were recruited for the study. In ad-
dition, 20 women with normal preg-
nancy were enrolled as controls. Basic
clinical and demographic information is
shown in Table 1. We enrolled both mul-
tiparous and nulliparous women in the
study. There were no significant differ-
ences between the groups in terms of
maternal age, body mass index, or gesta-
tional age. The women with preeclamp-
sia, gestational hypertension, and
chronic hypertension had significantly
higher systolic (P ⬍ .001) and diastolic
(P ⫽ .001) blood pressure than the con-
trol group. Five of 19 patients with pre-
eclampsia met the criteria for severe dis-
ease as defined by ACOG.5 Two of the 5
patients had HELLP (hemolysis, ele-
vated liver enzymes, and low platelet
count) syndrome, and the remaining 3
met only the blood pressure criteria.
Three women in the preeclampsia group
had small-for-gestational-age babies as
defined by birthweight less then 10% for
gestational age.
The serum values of sFlt 1 in women
with different hypertensive diseases of
pregnancy are shown in Figure 1. The
mean serum value of sFlt 1 was signifi-
cantly higher in women with preeclamp-
sia (74.7 ⫾ 83.3 ng/mL) than in women
with gestational hypertension (23.5 ⫾
14.9 ng/mL), chronic hypertension (15.4
⫾ 12.7 ng/mL), or normal pregnancy
(16.6 ⫾ 11.0 ng/mL, P ⬍ .01 by
ANOVA). Unlike women with chronic
hypertension, the mean serum level of
sFlt 1 in women with gestational hyper-
tension was statistically higher than the
control group (23.5 ⫾ 14.9 ng/mL vs
16.6 ⫾ 11.0 ng/mL, P ⫽ .04 by t test).
Similarly, the mean serum level of
sEng was significantly higher in women
with preeclampsia (69.2 ⫾ 42.5 ng/mL)
than in women with gestational hyper-
28.e3 American Journal of Obstetrics & Gynecology JULY 2007
www.AJOG.org
FIGURE 1
Serum levels of sFlt 1 in various hypertensive diseases of pregnancy
1000
* **
100
sFlt 1 (ng/ml)
10
1
Control cHTN gHTN Preeclampsia
(n=19) (n=20) (n=17) (n=19)
CHTN, chronic hypertension; GHTN, gestational hypertension; sFlt 1, soluble
fms-like tyrosine kinase 1
*P = .04 compared with control
**P < .01 compared with control
The mean serum level of sFlt 1 in women with preeclampsia was significantly higher than that of
the control group (P ⬍ .01 by t test). Interestingly, the mean serum level of women with gestational
hypertension was significantly higher than the control group as well (P ⫽ .04 by t test). ANOVA
showed that the mean serum value in women with preeclampsia (74.7 ⫾ 83.3 ng/mL) was
significantly higher than those seen in control (16.6 ⫾ 11.0 ng/mL), gestational hypertension (23.5
⫾ 14.9 ng/mL), or chronic hypertension (15.4 ⫾ 12.7 ng/mL) (P ⬍ .01).
tension (23.6 ⫾ 15.3 ng/mL) or chronic ity (90%) and specificity (90%) with the
hypertension (22.7 ⫾ 19.9 ng/mL) or the highest positive LR (9.0) and the lowest
control group (15.5 ⫾ 6.9 ng/mL; P ⬍ negative LR (0.1) in differentiating
.01 by ANOVA) (Figure 2). Similar to women with preeclampsia from normal
sFlt 1, the mean value of sEng was signif- pregnancy. The area under the ROC
icantly higher in women with gestational curve was 0.94 with SE of 0.04 (P ⬍ .01).
hypertension when compared with that For differentiating women with pre-
of the control group (23.6 ⫾ 15.3 ng/mL eclampsia from those with gestational
vs 15.5 ⫾ 6.9 ng/mL; P ⫽ .03 by t test) but hypertension, the serum sFlt-1 level of
not in women with chronic hypertension 41.8 ng/mL had the highest sensitivity
(23.6 ⫾ 15.3 ng/mL vs 22.7 ⫾ 19.9 ng/ (79%) and specificity (88%) with posi-
mL; P ⫽ .13 by t test). tive LR and negative LR of 6.7 and 0.2,
The sensitivity, specificity, positive LR, respectively. The area under the ROC
negative LR, and area under the ROC curve was 0.89 with SE of 0.06 (P ⬍ .01).
curve for sFlt 1, sEng, and uric acid in For diagnosing preeclampsia in women
various hypertensive diseases in preg- with chronic hypertension, the sFlt-1
nancy are shown in Table 2. According to level of 35.8 ng/mL showed the best sen-
the ROC curve analysis, the serum sFlt 1 sitivity (84%) and specificity (95%) with
level of 23.5 ng/mL had the best sensitiv- positive LR of 16 and negative LR of 0.2.
www.AJOG.org Obstetrics Research

The area under the curve (AUC) was

FIGURE 2
0.94 with SE of 0.04 (P ⬍ .01).
The serum levels of sEng in various hypertensive diseases of
Similarly, sEng had high sensitivity
pregnancy
and specificity in differentiating women
1000 with preeclampsia from those with nor-
mal pregnancy as well as various hyper-
tensive diseases of pregnancy. Analysis of
* ** the ROC curve showed that the serum
level of 24.8 ng/mL had the highest sen-
sitivity (90%) and specificity (95%) with
100 a positive LR of 18 and negative LR of 0.1
sEndoglin (ng/ml)

in differentiating women with pre-

eclampsia from those with normal preg-
nancy. The AUC was 0.93 with SE of 0.04
(P ⬍ .01). For diagnosing preeclampsia
in women with gestational hypertension,
10 the serum sEng level of 33 ng/mL showed
a sensitivity of 84% and specificity of
88% with a positive LR of 7.2 and nega-
tive LR of 0.2. The AUC was 0.87 with SE
of 0.06 (P ⬍ .01). For diagnosing pre-
eclampsia in women with chronic hyper-
1 tension, the serum level of 31.5 ng/mL
Control CHTN GHTN Preeclampsia had a sensitivity of 79% and specificity of
(n=19) (n=20) (n=17) (n=19) 99% with a positive LR of 4 and negative
LR of 0.2. The AUC was 0.87 with SE of
0.6 (P ⬍ 0.01).
CHTN, chronic hypertension; GHTN, gestational hypertension; sEndoglin, soluble
Consistent with previously published
endoglin.
*P = .03 compared with control reports,12 we found the mean serum
**P < .01 compared with control level of uric acid to be significantly
higher in women with preeclampsia (6.4
The mean serum levels of sEng in women with preeclampsia was significantly higher than that of
⫾ 1.1 mg/dL, P ⫽ .02) than that of
the control group (P ⬍ .01, t test). Similar to sFlt 1, the mean serum level of sEng in women with
women with gestational hypertension
gestational hypertension was statistically higher than that of the control group (P ⫽ .03, t test).
(5.0 ⫾ 1.4 mg/dL). However, the mean
ANOVA showed that the mean serum value in preeclampsia (69.2 ⫾ 42.5 ng/mL) was higher than
serum uric acid value in women with
those noted in control (15.5 ⫾ 6.9 ng/mL), gestational hypertension (23.6 ⫾ 15.3 ng/mL), or
chronic hypertension (5.7 ⫾ 1.5 mg/dL;
chronic hypertension (22.7 ⫾ 19.9 ng/mL) (P ⬍ .01).
P ⫽ .2) was not statistically different
from that noted in women with pre-
eclampsia (Figure 3). The ROC curve
showed that the serum level of 4.9 mg/dL
has 90% sensitivity and 62.5% specificity
in diagnosing preeclampsia from those

TABLE 2
Test performance of serum sFlt 1, sEng, and uric acid in diagnosing preeclampsia
Control/PRE Gest HTN/PRE Chronic HTN/PRE
sFlt 1 sEng sFlt 1 sEng Uric acid sFlt 1 sEng Uric acid
Sens (%) 90 90 79 84 90 84 84 68
................................................................................................................................................................................................................................................................................................................................................................................
Spec (%) 90 95 88 88 63 95 79 78
................................................................................................................................................................................................................................................................................................................................................................................
LR (positive) 9 17.9 6.7 7.2 2.4 16 4 3.1
................................................................................................................................................................................................................................................................................................................................................................................
LR (negative) 0.1 0.1 0.2 0.2 0.2 0.2 0.2 0.4
................................................................................................................................................................................................................................................................................................................................................................................
ROC 0.93 0.93 0.88 0.87 0.75 0.94 0.87 0.70
................................................................................................................................................................................................................................................................................................................................................................................
Chronic HTN, chronic hypertension; gest HTN, gestational hypertension; PRE, preeclampsia; Sens, sensitivity; Spec, specificity.

JULY 2007 American Journal of Obstetrics & Gynecology 28.e4

Research Obstetrics
with gestational hypertension. However,
the positive LR was only 2.4 with nega-
tive LR or 0.2. In the setting of chronic
hypertension, the serum level of 5.9
mg/dL was associated with a sensitivity
of 68% and specificity of 78% with pos-
itive LR of 3.1 and negative LR of 0.4 in
diagnosing preeclampsia.
C OMMENT
Our data show that the serum levels of
sFlt 1 and sEng are significantly higher in
women with preeclampsia than women
with normal pregnancy, gestational hy-
pertension, and chronic hypertension.
Interestingly, consistent with previously
published reports our data show that, al-
though to a lesser degree, the mean se-
rum level of these peptides in women
with gestational hypertension is higher
than in the control group.8-11 Because it
is estimated that approximately 25% of
patients with gestational hypertension
will develop preeclampsia, it is possible
that some of the patients with gestational
hypertension in our study may have had
preeclampsia without proteinuria. It will
be interesting to determine whether
there is a correlation between perinatal
outcome and serum levels of these fac-
tors in women with gestational hyper-
tension in a larger study.
Furthermore, our findings suggest
that the serum levels of these antiangio-
genic factors may be sensitive and spe-
cific markers for preeclampsia. More im-
portantly, the positive LRs of these
serum markers can increase the pretest
probability of having preeclampsia
enough to alter clinical decision mak-
ing.13 Because these angiogenic factors
have been shown to cause glomerular en-
dotheliosis in rats,7 to date, they may be
the best serologic indicators of glomeru-
lar endotheliosis. In fact, the AUC com-
parison showed that serum sFlt 1 is a sig-
nificantly better test than serum uric acid
in differentiating preeclampsia from
chronic hypertension (AUC 0.94 vs. 0.7,
P ⫽ .02). For differentiating preeclamp-
sia from gestational hypertension, the
ROC curve for sFlt-1 was also better,
compared with that of uric acid (AUC of
0.88 vs 0.75), but the difference was sta-
tistically not significant (P ⫽ .13).
28.e5 American Journal of Obstetrics & Gynecology JULY 2007
FIGURE 3
The serum levels of uric acid
11
10
9
Uric Acid (mg/dl)
8
7
6
5
4
3
CHTN
(n=18)
CHTN, chronic hypertension; GHTN, gestational hypertension; UA, uric acid.
* P=.02 compared to GHTN
The mean serum level in patients with preeclampsia was statistically higher than that of the women
with gestational hypertension (P ⫽ .02, ANOVA) but not with chronic hypertension.
In the past few years, the work done by
our group and others7-11 established the
possible pathogenic role of sFlt 1 and
sEng in preeclampsia. This work repre-
sents an attempt to determine the clini-
cal application of these markers. Because
the serum levels of these markers rise ap-
proximately 6 weeks before the onset of
the symptoms of the disease, increased
serum levels of these factors in the setting
of labile blood pressures or equivocal
proteinuria may identify patients who
will develop the full symptoms and signs
of preeclampsia imminently. Depending
on the gestational age, this type of infor-
mation may be highly valuable in devel-
oping rational management plans for
these patients. In addition, the serum
levels of these factors may correlate bet-
ter with various perinatal outcomes than
blood pressure or 24-hour urine protein.
The correlation between disease severity
and serum levels of these antiangiogenic
peptides has been demonstrated.7,8 In
our study, the sample size was too small
www.AJOG.org
*
GHTN Preeclampsia
(n=16) (n=19)
to correlate various perinatal outcomes
with the serum levels of these peptides. A
larger study to examine the possibility of
these angiogenic factors predicting poor
perinatal outcome in women with hy-
pertension in pregnancy will be of great
clinical interest.
A limitation of this study was the sam-
ple size. This study was conducted to as-
sess the rationale and feasibility of de-
signing a larger, prospective study to
determine the clinical utility of these an-
giogenic factors. A sample size estima-
tion was difficult because we were not
able to locate sensitivity, specificity, and
likelihood ratios for these factors. How-
ever, based on our previous published
result of serum levels of sFlt1 and sEng in
patients with preeclampsia and control
population (sFlt1: 20 ⫾ 8 ng/mL [con-
trol] vs 40 ⫾ 10 ng/mL [preeclampsia]
and sEng: 19 ⫾ 5ng/mL [control] vs 38
⫾ 12 ng/mL [preeclampsia]),8 we
needed less than 10 patients in each arm
to show a 50% difference in the mean
www.AJOG.org
serum values between the 2 groups, with
an alpha of 0.05 and a beta of 0.2. Assum-
ing that the mean serum levels in patients
with gestational hypertension and
chronic hypertension were similar to
control population, we estimated the
same number of patients will be needed
to show a 50% difference in the mean
values, compared with patients with
preeclampsia.
Given our findings (ie, area under the
curves of the ROC), we now estimate
that we would need approximately 34
patients to show that sFlt 1 is a better
marker than serum uric acid for distin-
guishing women with preeclampsia
from those with chronic hypertension
with a beta of 0.2. For sEng, we would
need 69 women. For differentiating
women with gestational hypertension
from preeclampsia, we would need 103
women for sFlt1 and 121 for sEng.
Despite its small size, however, we
were able to demonstrate and confirm
the differences in serum levels of these
peptides in various hypertensive diseases
in pregnancy. This finding was true, even
if we excluded the 5 patients with severe
preeclampsia for both sFlt 1 (83.1 ⫾ 95.2
ng/mL [preeclampsia] vs 16.6 ⫾ 11.0
ng/mL [control], 15.4 ⫾ 12.8 ng/mL
[chronic hypertension], 23.5 ⫾ 14.9
ng/mL [gestational hypertension], P by
ANOVA ⬍ .01) and sEng (62.0 ⫾ 34.0
ng/mL [preeclampsia] vs 15.5 ⫾ 6.9
ng/mL [control], 22.7 ⫾ 19.9 ng/mL
[chronic hypertension], 23.6 ⫾ 15.3
ng/mL [gestational hypertension], P by
ANOVA ⬍ .01). In addition, our find-
ings were unchanged when we excluded
the highest serum value noted in sFlt 1
(60.8 ⫾ 27.0 ng/mL [preeclampsia] vs
16.6 ⫾ 11.0 ng/mL [control], 15.4 ⫾ 12.8
ng/mL [chronic hypertension], 23.5 ⫾
14.9 ng/mL [gestational hypertension],
P by ANOVA ⬍ .001).
Another limitation was that we were
not able to control for gestational age.
However, all of the deliveries took place
after 28 weeks of gestation. In addition,
the control group had the mean gesta-
tional age that is greater than that of the
study population. It is unclear whether
there is statistical difference in the serum
levels of these angiogenic factors be-
tween early third trimester and term.
Nevertheless, previously published data
suggest that the serum level of both sFlt 1
and sEng rise in normal pregnant popu-
lation with increasing gestational
age.9,11,14,15 We believe that this would
have biased the data against our findings
because the serum levels of our control
group were measured later in gestation.
Despite this possibility, our data show
that the serum levels of these antiangio-
genic proteins are higher in women with
preeclampsia. A larger study will allow us
to determine whether the clinical utility
of these factors will vary, depending on
gestational age.
Although preliminary, our findings
raise the possibility that measuring se-
rum levels of sFlt 1 and sEng in women
with various hypertensive diseases of
pregnancy may be clinically useful. Hav-
ing a sensitive and specific serum bi-
omarker for preeclampsia that can be
used in conjunction with urine protein
analysis will not only improve the accu-
racy but also expedite the diagnosis of
preeclampsia. Our findings provide a ra-
tionale for a large prospective study to
define normal reference ranges for these
factors and to determine their clinical
utility and ability to predict perinatal
outcome in the setting of various hyper-
tensive diseases in pregnancy. f fms-like tyrosine kinase 1 and free placental
REFERENCES
1. Hauth JC, Ewell MG, Levine RJ, et al. Preg-
nancy outcome in healthy nulliparas who devel-
oped hypertension. Obstet Gynecol 2000;
95:24-8.
2. Sibai BM. Diagnosis and management of
gestational hypertension and preeclampsia.
Obstet Gynecol 2003;102:181-92.
JULY 2007 American Journal of Obstetrics & Gynecology
Obstetrics Research
3. Walker JJ. Pre-eclampsia. Lancet 2000;
356:1260-5.
4. Redman CW, Sargent IL. Latest advances in
understanding preeclampsia. Science 2005;
308:1592-4.
5. Diagnosis and management of preeclampsia
and eclampsia: ACOG practice bulletin 33; Jan-
uary 2002.
6. Report of the National High Blood Pressure
Education Program. Working Group on High
Blood Pressure in Pregnancy. Am J Obstet Gy-
necol 2000;183:s1-22.
7. Maynard SE, Min JY, Merchan J, et al. Ex-
cess placental soluble fms-like tyrosine kinase 1
(sFlt1) may contribute to endothelial dys-
function, hypertension, and proteinuria in
preeclampsia. J Clin Invest 2003;111:
649-58.
8. Venkatesha S, Toporsian M, Lam C, et al.
Soluble endoglin contributes to the patho-
genesis of preeclampsia. Nat Med 2006;12:
642-9.
9. Levine RJ, Maynard SE, Qian C, et al. Circu-
lating angiogenic factors and the risk of pre-
eclampsia. N Engl J Med 2004;350:672-83.
10. Shibata E, Rajakumar A, Powers RW, et al.
Soluble fms-like tyrosine kinase 1 is increased in
preeclampsia but not in normotensive pregnan-
cies with small-for-gestational-age neonates:
relationship to circulating placental growth fac-
tor. J Clin Endocrinol Metab 2005;90:
4895-903.
11. Levine RJ, Lam C, Cong Qian C, et al. Sol-
uble endoglin, a novel circulating anti-angio-
genic factor in preeclampsia. N Engl J Med
2006;355:992-1005.
12. Lim KH, Friedman SA, Ecker JL, et al. The
clinical utility of serum uric acid measurements
in hypertensive diseases of pregnancy. Am J.
Obstet Gynecol 1998;178:1067-71.
13. Jaeschke R, Guyatt G, Sackett D. Users’
guide to the medical literature, III: how to use an
article about a diagnostic test. JAMA 1994;
271:703-7.
14. Hirashima C, Ohkuchi A, Arai F, et al. Es-
tablishing reference values for both total soluble
growth factor in pregnant women. Hypertens
Res 2005;28:727-32.
15. Powers RW, Roberts JM, Cooper KM, Gal-
laher MJ, Frank MP, Harber GF. Maternal serum
soluble fms-like tyrosine kinase 1 concentra-
tions are not increased in early pregnancy and
decrease more slowly postpartum in women
who develop preeclampsia. Am J Obstet Gy-
necol 2005;193:185-91.
28.e6