- Email: [email protected]
The predictive value of quantitative fibronectin testing in combination with cervical length measurement in symptomatic women
Accepted Manuscript The predictive value of quantitative fibronectin testing in combination with cervical length measurement in symptomatic women Merel MC. Bruijn, MD, Esme I. Kamphuis, MD, Irene M. Hoesli, MD, PhD, Begoña Martinez De Tejada, MD, PhD, Anne R. Loccufier, MD, Maritta Kühnert, MD, PhD, Hanns Helmer, MD, PhD, Marie Franz, MD, Martina M. Porath, MD, PhD, Martijn A. Oudijk, MD, PhD, Yves Jacquemyn, MD, PhD, Sven M. Schulzke, MD, Grit Vetter, MD, Griet Hoste, MD, Jolande Y. Vis, MD, PhD, Marjolein Kok, MD, PhD, Ben WJ. Mol, MD, PhD, Gert-Jan van Baaren, MD, PhD PII:
S0002-9378(16)30577-4
DOI:
10.1016/j.ajog.2016.08.012
Reference:
YMOB 11266
To appear in:
American Journal of Obstetrics and Gynecology
Received Date: 6 June 2016 Revised Date:
21 July 2016
Accepted Date: 8 August 2016
Please cite this article as: Bruijn MM, Kamphuis EI, Hoesli IM, Martinez De Tejada B, Loccufier AR, Kühnert M, Helmer H, Franz M, Porath MM, Oudijk MA, Jacquemyn Y, Schulzke SM, Vetter G, Hoste G, Vis JY, Kok M, Mol BW, van Baaren G-J, The predictive value of quantitative fibronectin testing in combination with cervical length measurement in symptomatic women, American Journal of Obstetrics and Gynecology (2016), doi: 10.1016/j.ajog.2016.08.012. This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
ACCEPTED MANUSCRIPT
1
The predictive value of quantitative fibronectin testing in combination with cervical
2
length measurement in symptomatic women.
3 Merel MC BRUIJN1*, MD, Esme I KAMPHUIS1,2, MD, Irene M HOESLI3, MD, PhD, Begoña
5
MARTINEZ DE TEJADA4, MD, PhD, Anne R LOCCUFIER5, MD, Maritta KÜHNERT6, MD, PhD,
6
Hanns HELMER7, MD, PhD, Marie FRANZ7, MD, Martina M PORATH8, MD, PhD, Martijn A
7
OUDIJK1,9, MD, PhD, Yves JACQUEMYN10, MD, PhD, Sven M SCHULZKE11, MD, Grit
8
VETTER3, MD, Griet HOSTE5, MD, Jolande Y VIS12, MD, PhD, Marjolein KOK1, MD, PhD, Ben
9
WJ MOL13, MD, PhD, Gert-Jan VAN BAAREN1, MD, PhD
M AN U
TE D
27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47
* Corresponding author: Merel MC Bruijn, Department of Obstetrics and Gynecology, Academic Medical Centre, 1105 DE, Amsterdam, The Netherlands, Email: [email protected].
EP
26
1. Dept. Obstetrics & Gynecology, Academic Medical Centre, Amsterdam, Netherlands. 2. Dept. Obstetrics & Gynecology, VU University Medical Centre, Amsterdam, Netherlands. 3. Dept. Obstetrics & Gynecology, University Hospital Basel, Basel, Switzerland. 4. Dept. Obstetrics & Gynecology, University Hospitals of Geneva and Faculty of Medicine, Geneva, Switzerland. 5. Dept. Obstetrics & Gynecology, Hospital St Jan, Bruges, Belgium. 6. Dept. Obstetrics & Gynecology, University Hospital of Marburg, Germany. 7. Dept. Obstetrics & Maternal-Fetal Medicine, Medical University Vienna, Vienna, Austria. 8. Dept. Obstetrics & Gynecology, Maxima Medical Centre, Veldhoven, Netherlands. 9. University Medical Centre Utrecht, Netherlands. 10. Dept. Obstetrics & Gynecology, Antwerp University Hospital, Antwerp, Belgium. 11. Dept. Neonatology, University Children's Hospital Basel, Switzerland. 12. Clinical Chemistry and Hematology, University Medical Centre Utrecht, Netherlands. 13. School of Medicine, University of Adelaide, Adelaide, Australia.
AC C
11 12 13 14 15 16 17 18 19 20 21 22 23 24 25
SC
10
RI PT
4
Presented as a poster at the 35th Annual Meeting of the Society of Maternal Fetal Medicine, February 2 -7, 2015, San Diego, California. Financial disclosure: Ben WJ Mol is consultant for ObsEva. Payments go to his institute. The remaining authors report no conflict of interest. The study was conducted with financial support from Hologic USA (Marlborough, MA). They had no role in study design, interpretation of data or writing of the report. Word count abstract: 368 Word count manuscript: 3075
1
ACCEPTED MANUSCRIPT
48
Quantitative fetal fibronectin testing improves the prediction of spontaneous preterm birth
49
within seven days in symptomatic women undergoing cervical length measurement.
50
Short version of the article title: quantitative fFN and cervical length in the prediction of
52
preterm birth
RI PT
51
53 54
SC
55 56
M AN U
57 58 59 60 61 62
66 67 68 69 70 71 72 73 74
EP
65
AC C
64
TE D
63
75 76 77 78 79
2
ACCEPTED MANUSCRIPT
80
Abstract
81
Background: The combination of the qualitative fetal fibronectin test and cervical length
83
measurement has a high negative predictive value for preterm birth within seven days, but
84
positive prediction is poor. A new bedside quantitative fetal fibronectin test showed potential
85
additional value over the conventional qualitative test, but there is limited evidence on the
86
combination with cervical length measurement.
87
Objective: To compare quantitative fetal fibronectin and qualitative fetal fibronectin testing
88
in the prediction of spontaneous preterm birth within seven days in symptomatic women
89
undergoing cervical length measurement.
90
Study design: We performed a European multicenter cohort study in ten perinatal centers in
91
five countries. Women between 24 and 34 weeks of gestation with signs of active labor and
92
intact membranes underwent quantitative fibronectin testing and cervical length measurement.
93
We assessed the risk of preterm birth within seven days in predefined strata based on
94
fibronectin concentration and cervical length.
95
Results: Out of 455 included women, 48 (11%) delivered within seven days. A combination
96
of cervical length and qualitative fibronectin resulted in identification of 246 women at low
97
risk: 164 women with a cervix between 15 and 30 mm and a negative fibronectin test
98
(<50ng/mL) (preterm birth rate 2%), and 82 women with a cervix above 30 mm (preterm
99
birth rate 2%). Use of quantitative fibronectin alone resulted in a predicted risk of preterm
100
birth within seven days ranging from 2% in the group with the lowest fibronectin level
101
(<10ng/mL) to 38% in the group with the highest fibronectin level (>500ng/mL), with similar
102
accuracy as the combination of cervical length and qualitative fibronectin. Combining
103
cervical length and quantitative fibronectin resulted in identification of an additional 19
104
women at low risk (preterm birth rate 5%) using a threshold of 10ng/mL in women with a
AC C
EP
TE D
M AN U
SC
RI PT
82
3
ACCEPTED MANUSCRIPT
cervix below 15mm, and 6 women at high risk (preterm birth rate 33%) using a threshold of
106
>500ng/mL in women with a cervix above 30mm.
107
Conclusion: In women with threatened preterm birth, quantitative fibronectin testing alone
108
performs equal to the combination of cervical length and qualitative fibronectin. Possibly, the
109
combination of quantitative fibronectin testing and cervical length increases this predictive
110
capacity. Cost-effectiveness analysis as well as availability of these tests in a local setting
111
should determine the final choice.
SC
RI PT
105
M AN U
112 113
Keywords: cervical length, fetal fibronectin, prediction, predictive value, preterm birth,
114
threatened preterm labor
115
119
120
121
122
EP
118
AC C
117
TE D
116
123
124
4
ACCEPTED MANUSCRIPT
Introduction
126
Preterm birth, defined as birth before 37 weeks’ gestation, occurs worldwide in 15 million
127
babies and accounts for 11% of all live births.1 In 1.1 million of these births, prematurity
128
results in death of the child, thus making preterm birth the leading cause of neonatal
129
mortality, while in case of survival preterm birth is an important cause of morbidity.2
130
Accurate identification of women who will deliver in short time allows targeted interventions
131
that improve outcome among those who indeed deliver preterm, such as corticosteroids,
132
magnesium sulphate and in utero transfer to a perinatal center, thus avoiding unnecessary
133
interventions with potential side effects for women at low risk. However, the large majority of
134
women who present with symptoms of preterm labor will not deliver within seven days and
135
50% will deliver at term.3-5
M AN U
SC
RI PT
125
Among various methods to assess the risk of preterm birth, fetal fibronectin (fFN)
137
testing and cervical length measurement are widely used. Fetal fibronectin, a glycoprotein
138
found at the choriodecidual interface, is traditionally used as a binary bedside test providing a
139
positive or negative result based on a threshold of 50 ng/mL.6 The combination of cervical
140
length measurement and fFN testing has a high negative predictive value (>98%) for delivery
141
within seven days, but positive prediction is poor.7 The best strategy to combine the two tests
142
is additional fFN testing in women with a cervical length between 15 and 30 mm, reducing
143
the number of unnecessary referrals and admissions to perinatal centers.8 This approach
144
would also be cost saving without compromising neonatal health outcomes.9
EP
AC C
145
TE D
136
A new bedside quantitative fetal fibronectin test showed potential additional value
146
over the conventional qualitative test with a higher risk for preterm birth with increasing
147
levels of fibronectin. The positive predictive value for preterm birth on short term and preterm
148
birth before 34 weeks of gestation increased by changing the threshold from 50 ng/mL to 200
149
ng/mL or 500 ng/mL with minimal effect on the negative predictive value.10 11
5
ACCEPTED MANUSCRIPT
Here, we prospectively evaluated whether, in women with symptomatic preterm birth
151
the combination of quantitative fFN testing and cervical length measurement has better
152
predictive accuracy than qualitative fFN testing with cervical length measurement or than
153
quantitative fFN testing alone.
RI PT
150
154
Materials and methods
156
We conducted a European multicenter cohort study in 10 centers in five countries (four
157
centers in The Netherlands, two in Switzerland, two in Belgium, one in Germany and one in
158
Austria). They were all perinatal centers serving as tertiary referral centers for high-risk
159
obstetric patients with an annual number of women delivering between 1,000 (Antwerp
160
University Hospital) and 4,000 (University Hospitals of Geneva). The study protocol was
161
approved by the ethics committee of the Academic Medical Centre Amsterdam, the
162
University Hospitals of Geneva and the University Hospital Basel as well by the board of
163
directors of the other participating centers in Germany, Austria and Belgium. Written
164
informed consent was obtained from all participants prior to inclusion.
TE D
M AN U
SC
155
Women presenting with symptoms of preterm labor (more than three contractions per
166
30 minutes, vaginal blood loss or abdominal or back pain) who were between 24 and 34
167
weeks of gestation, and who had intact membranes were eligible for participation in the study.
168
Gestational age was based on first trimester ultrasound assessment. The women either
169
presented themselves directly at one of the participating centers or were referred by their
170
primary gynecologist from a general hospital (secondary care) or by a general practitioner or
171
midwife (primary care). Women who had received tocolytic treatment for more than 18 hours
172
were excluded from the study. Women with contraindications for tocolysis such as a lethal
173
congenital abnormality, suspected intrauterine infection, nonreassuring fetal status or
174
maternal distress requiring immediate intervention such as placental abruption or severe
AC C
EP
165
6
ACCEPTED MANUSCRIPT
vaginal blood loss were excluded. Other exclusion criteria were cervical dilatation of more
176
than three cm and triplet or higher order pregnancies. Also women who had iatrogenic
177
delivery within seven days after study entry for hypertensive disorders, fetal distress, or other
178
reasons for immediate delivery were excluded. We intended to include 500 women, i.e. 100
179
women per country. We derived a sample size with the aim to estimate the sensitivity of the
180
quantitative fFN test with sufficient precision. If the performance of the quantitative fFN test
181
would resemble the qualitative fFN test with a cut off value of 50 ng/ml, the expected
182
sensitivity would be 91% (based on a Dutch cohort study from 2009). Inclusion of 100
183
patients with subsequent preterm delivery would produce a two-sided 95% confidence
184
interval with a width equal to 0.112 around the expected sensitivity of 91%. In view of an
185
expected prevalence of preterm birth of 20%, we would have to include a total of 500 women
186
with complaints of threatened labor.
M AN U
SC
RI PT
175
After study enrolment, all women underwent cervical length measurement and fFN
188
testing. Cervical length was measured by transvaginal ultrasound. fFN testing was performed
189
using the quantitative Rapid fFN 10Q analyzer (Hologic®) according to manufacturer’s
190
instructions. A 50 ng/mL cut-off was used for the qualitative result of the fFN test (positive,
191
negative). The qualitative fFN result was available immediately for the use in patient care, but
192
the quantitative fFN result remained blinded until after delivery (the analyzer generated a
193
random 3-letter result code). When the study started, both cervical length measurement and
194
the qualitative fFN test were standard care in all participating centers for risk stratification in
195
women with symptoms of preterm labor. Clinicians were trained in transvaginal
196
ultrasonography and performed cervical length measurements themselves. Furthermore, all
197
involved personnel were trained in the use of the specimen collection and the fFN analyzer.
198
Preferably, a specimen had to be collected before the vaginal examination or cervical length
199
measurement. Furthermore, additional information on the maternal condition was obtained
AC C
EP
TE D
187
7
ACCEPTED MANUSCRIPT
including vaginal digital examination, maternal serum for CRP and leukocytes, urine and
201
vaginal culture, and blood pressure. In addition, data collection included information on
202
factors that might influence the fFN result, such as the use of vaginal soap or sexual
203
intercourse within 24 hours before performance of the fFN test and vaginal bleeding during
204
testing.
RI PT
200
Treatment recommendations were based on previous research.8 We recommended
206
starting tocolysis and steroids for fetal lung maturation in high-risk women, defined as
207
women with a cervical length below 15 mm and in women with a cervical length between 15
208
and 30 mm with a positive fFN result. For low-risk women with a cervical length above 30
209
mm, we recommended to withhold steroids and tocolysis. For women with a cervical length
210
between 15 and 30 mm in combination with a negative fFN test, the clinician on call decided
211
whether to start tocolysis and steroids or not. Tocolytic drugs that could be prescribed were
212
beta-sympathomimetics, nifedipine, atosiban, indomethacin. Magnesium sulphate for
213
neuroprotection and antibiotics were given on consideration of the attending physician.
TE D
M AN U
SC
205
The primary outcome was spontaneous preterm birth within seven days after study
215
entry. We used logistic regression analyses with an interaction term to assess the relationship
216
between potential influencing factors and the quantitative fFN result, such as transvaginal
217
ultrasound, vaginal examination, the use of vaginal soap or sexual intercourse within 24 hours
218
before performance of the fFN test and vaginal bleeding during testing. Quantitative fFN was
219
analyzed as a continuous variable, all other factors as dichotomous variables. Linearity of the
220
association between quantitative fFN and the risk of preterm birth was assessed using
221
restricted cubic spline analyses, using 3 knots. If one of the interaction terms was significant,
222
the factor was considered to have a confounding relation with the quantitative fFN result and
223
women with that factor were excluded from the analyses Descriptive statistics were calculated
224
for baseline demographics and obstetric characteristics.
AC C
EP
214
8
ACCEPTED MANUSCRIPT
For risk stratification, thresholds of 10, 50, 200 and 500 ng/mL for the quantitative
226
fFN test were predefined before analysis and cervical length was divided into groups of <15
227
mm, 15-29 mm and 30-50 mm. These thresholds were based on previous studies.8,10,12 The
228
rates of preterm birth within seven days were calculated within the corresponding strata. We
229
considered a risk of delivery less than 5% as low risk. This threshold has been derived from
230
the interpretation of cervical length, because this is currently used in clinical decision-making
231
in women with signs of preterm labor. A cervical length measurement with a threshold
232
between 20 and 30 mm corresponds to a post-test probability of delivery within seven days of
233
4-5%, compared to a pre-test probability of 11% if cervical length measurement is not
234
performed.5 To assess if more low-risk women could be identified using the quantitative fFN
235
test with various thresholds, we compared the accuracy to that of the qualitative fFN test
236
(threshold 50 ng/mL) in combination with cervical length measurement. As the fFN test may
237
behave differently according to parity and obstetric history, we additionally performed
238
subgroup analyses for nulliparous women and women without a history of preterm birth.
TE D
M AN U
SC
RI PT
225
We used logistic regression analyses to assess the univariable association between
240
cervical length, qualitative fFN and quantitative fFN, and preterm birth within seven days.
241
Furthermore, we constructed two multivariable prediction models to predict preterm birth
242
within seven days: one model including the variables cervical length and qualitative fFN and
243
one model including the variables cervical length and quantitative fFN. Both cervical length
244
and quantitative fFN were analyzed as continuous variables. Linearity of the association
245
between the variables cervical length and quantitative fFN and the risk of preterm birth was
246
assessed using restricted cubic spline analyses. We used bootstrapping techniques for internal
247
validation of the models to correct for overfitting. Two hundred samples with the same size of
248
the original data set were drawn from the original data set with replacement. A shrinkage
249
factor was derived from these analyses and used to adjust regression coefficients.13,14 We
AC C
EP
239
9
ACCEPTED MANUSCRIPT
compared the models in terms of overall fit using Nagelkerke R2 and discrimination using the
251
area under the receiver operating characteristics curve (AUC). To determine whether
252
quantitative fFN as compared to qualitative fFN in combination with cervical length improves
253
the capability of the model to identify low-risk women (<5% risk), we compared the two
254
models in terms of reclassification.
256
Data analyses were performed in SPSS version 20.0 (Chicago, IL, USA) and R version 2.10.0 (The R foundation for Statistical Computing, 2009).
SC
255
RI PT
250
257
Results
259
Between January 2013 and May 2014, a total of 532 women were potentially eligible for
260
study enrolment, of whom 52 women did not meet the inclusion criteria, 11 had one or more
261
exclusion criteria and seven did not give their consent to participate in the study. In the
262
remaining 462 participating women, we performed fFN testing and cervical length
263
measurement. The fFN result and cervical length were both not recorded in three women. We
264
excluded one woman who had an elective caesarean section within seven days after
265
enrollment because of fetal distress. In total, 455 women were available for analysis (Figure
266
1); 192 included in the Netherlands, 146 in Switzerland, 55 in Belgium, 47 in Germany and
267
15 in Austria. Baseline demographics and obstetric characteristics for the study participants
268
are shown in Table 1.
TE D
EP
AC C
269
M AN U
258
Logistic regression analysis showed no interaction between quantitative fFN and
270
cervical length at transvaginal ultrasound (n=211, p=0.98), vaginal examination (n=110,
271
p=0.44), the use of vaginal soap (n=90, p=0.21) or sexual intercourse (n=20, p=0.76) within
272
24 hours before performing the fFN test. Similarly, there was no interaction with vaginal
273
bleeding during testing (n=64, p=0.91). Therefore, women with one of these characteristics
274
were not excluded from the analyses.
10
ACCEPTED MANUSCRIPT
Table 2 shows the risk stratification. The risk of preterm birth within seven days
276
increased with increasing fFN concentration (from 2% when fFN was <10 ng/mL to 38%
277
when ≥500 ng/mL) and shortening of the cervix (from 2% when cervical length was ≥30mm
278
to 28% when cervical length was <15mm). Using the conventional threshold of 50 ng/mL
279
when cervical length was between 15 and 30 mm, 164 women were identified as low risk, of
280
whom three (2%) delivered within seven days. Changing of the threshold from 50 to 200
281
ng/mL in women with a cervical length between 15 and 30 m would label 45 women as high
282
risk, as four women (9%) delivered within seven days. In 103 women with a cervical length
283
below 15 mm, using a threshold of 10 ng/mL resulted in identification of 19 women as low
284
risk, of whom one (5%) delivered within seven days. Comparison of this proportion to
285
women with a fFN concentration above 10 ng/mL using a Fisher’s exact test, showed the
286
difference to be statistically significant (p=0.01). Two out of 82 women with a cervical length
287
above 30 mm (2%) delivered within seven days. Both had a fFN concentration of more than
288
500 ng/mL (33% of this group), which showed to be significantly different from the women
289
with a fFN concentration below 500 ng/mL (p=0.001).
TE D
M AN U
SC
RI PT
275
Table 3 and table 4 show the risk stratification for nulliparous women only (n=249)
291
and women without a history of preterm birth (n=383), respectively. The results were similar
292
to those of the complete group, showing an increased risk of preterm birth within seven days
293
with increasing fFN concentration and shortening of the cervix. Women with a long cervix
294
were still identified as high risk with a threshold of > 500 ng/mL, although this did not show
295
to be significant (table 3: p=0.14, table 4: p=0.07). A threshold of 10 ng/mL in women with a
296
short cervix did not help to identify more women at low risk.
AC C
EP
290
297
Cubic spline analyses showed a linear relationship between both quantitative fFN and
298
cervical length and the risk of preterm birth within seven days, and were therefore analyzed as
299
continuous variables. In univariable logistic regression analysis, a positive qualitative fFN test
11
ACCEPTED MANUSCRIPT
was associated with a higher risk of preterm birth within seven days (OR 11 (95% CI 4.7 –
301
27)). The same was the case for the quantitative fFN concentration (OR 1.006 per ng/mL
302
(95% CI 1.005 – 1.008)). A longer cervix was associated with a lower risk of preterm birth
303
within seven days (OR 0.90 per mm (95% CI 0.86 – 0.94)).
RI PT
300
Table 5 shows the two multivariable models. The model with qualitative fFN and
305
cervical length was presented after shrinkage with an average shrinkage factor of 0.96 and the
306
model with quantitative fFN and cervical length with an average shrinkage factor of 0.99. The
307
model with cervical length and qualitative fFN had an AUC of 0.83 (95% CI 0.77 – 0.88) and
308
a Nagelkerke R2 of 0.27. The model with CL and quantitative fFN had an AUC of 0.84 (95%
309
CI 0.78 – 0.89) and a Nagelkerke R2 of 0.32. Quantitative fFN alone had an AUC of 0.82
310
(95% CI 0.76 – 0.89) and a Nagelkerke R2 of 0.25.
M AN U
SC
304
Table 6 shows the reclassification of participants based on a multivariable model with
312
quantitative fFN and cervical length compared to a model with qualitative fFN and cervical
313
length. Using quantitative fFN and cervical length resulted in 62 women (14%) being
314
reclassified; 37 women (8%) were reclassified as low risk, of whom four (11%) delivered
315
within seven days. On the other hand, 25 women were reclassified as high risk, of whom three
316
(12%) delivered within seven days.
317
EP
TE D
311
Comment
319
This study assessed the diagnostic accuracy of the quantitative fFN test in combination with
320
cervical length measurement, as compared to the qualitative test in combination with cervical
321
length, and the quantitative fFN test alone, in the prediction of spontaneous preterm birth
322
within seven days in symptomatic women. The results demonstrated that quantification of
323
fFN provides additional information on risk stratification as the risk of preterm birth increases
324
with increasing fFN concentrations. In terms of reclassification, quantitative fFN in
AC C
318
12
ACCEPTED MANUSCRIPT
combination with cervical length does not improve the identification of women with a low
326
risk (<5%) of preterm birth within seven days compared to the qualitative fFN test in
327
combination with cervical length. However, when looking in detail using the risk
328
stratification, we see that with a threshold of 10 ng/mL in women with a short cervix (<15
329
mm), who are in current practice treated as high risk without additional fFN testing, more
330
low-risk women could be identified. On the other hand, a threshold of 500 ng/mL in women
331
with a cervix above 30 mm, who are now treated as low risk without additional fFN testing,
332
will help identifying women with a high risk to deliver within seven days.
SC
RI PT
325
A strength of this study is that data were derived from a well-described, large,
334
prospectively collected cohort of symptomatic women, representing a European population
335
with mostly Caucasian women. The qualitative fFN test was collected according to protocol
336
and the result was used to determine further management, whereas the quantitative result
337
remained blinded until the end of the study.
M AN U
333
We are aware that preterm delivery rates, sociodemographic characteristics and the
339
obstetrical management may differ from country to country, and even from center to center.
340
In our analysis, we did not account for these between-center differences and the assumption
341
that there are no differences may have serious implications if untrue. Therefore, we performed
342
additional analyses to investigate the potential between-center heterogeneity in our study. We
343
have included the centers in the models as dummy variables, to estimate a stratified intercept
344
for each center. This methodology has been described to account for heterogeneity between
345
study populations in individual participant data (IPD) meta-analysis, which can be compared
346
to the issue of different centers in our study.15 A random effects analysis of the intercepts
347
showed that there was no heterogeneity between centers (I2=0%). When comparing the
348
performance of these models (with centers) with our presented models (without centers) we
349
found that both model fit and discrimination were slightly better for the former models with
AC C
EP
TE D
338
13
ACCEPTED MANUSCRIPT
AUCs of 0.84 (qualitative fFN and cervical length), 0.86 (quantitative fFN and cervical
351
length) and 0.84 (quantitative fFN) compared with 0.83, 0.84, 0.82, respectively. However,
352
the increase in performance is only marginally improved given the increased complexity of
353
the models and the loss of generalizability. Therefore we decided to only present the results of
354
the models without adjustment for centers
RI PT
350
It is questionable whether tocolysis has influenced our results by reducing the number
356
of women who delivered within seven days after study entry. Tocolytics have a short half-life
357
and are in general only given for the first 48 hours after admission in order to delay delivery
358
long enough to administer antenatal corticosteroids causing improved neonatal outcomes.16,17
359
Good evidence that tocolysis, given during the first 48 hours after admission, delays delivery
360
after seven additional days have passed, is lacking.18 The World Health Organization
361
recommends not to use tocolytic treatments, other than to prolong pregnancy (up to 48 hours)
362
to provide a window for administration of antenatal corticosteroids and/or in-utero transfer,
363
because of the potential risks and lack of evidence of the efficacy.19 Therefore, we think that it
364
is unlikely that tocolysis have influenced the results of this study.
TE D
M AN U
SC
355
Previous studies have reported improved prediction of preterm birth using
366
quantification of fFN in both symptomatic and high-risk asymptomatic women.10,11,20 Abbott
367
et al. reported increasing spontaneous preterm birth rates within two weeks with increasing
368
fFN concentrations, ranging from 2% in the lowest fFN group (<10 ng/mL) to 46% in the
369
highest fFN group (>500 ng/mL), in symptomatic women. Changing the threshold from the
370
conventional 50 ng/mL to 200 ng/mL led to an increase in the positive predictive value from
371
20% to 37% for the prediction of preterm birth within two weeks with minimal effect on the
372
negative predictive value.10 A recently published study from Centra et al. supported these
373
results.11 Our results confirm that compared to qualitative fFN, quantification of fFN provides
374
better risk stratification across the risk range, showing an increasing preterm birth rate from
AC C
EP
365
14
ACCEPTED MANUSCRIPT
375
3% in the lowest fFN group (<10 ng/mL) to 38% in the highest fFN group (>500 ng/mL).
376
However, in those previous studies, the quantitative fFN test was not evaluated in
377
combination with cervical length measurement. We recently described similar analyses performed on data of a Dutch prospective
379
nationwide cohort study.21 In this study we evaluated the quantitative fFN test on frozen fFN
380
samples obtained from symptomatic women with a cervical length below 30 mm. The results
381
were in line with the results of the current study, showing an increasing risk of delivery within
382
seven days with increasing fFN concentrations. However, using a threshold of 10 ng/mL in
383
women with a short cervix (<15 mm), our previous study showed that no more women at low
384
risk of preterm birth within seven days were identified, whereas the present results showed
385
that 19 women were identified of whom one (5%) delivered within seven days, indicating that
386
these women could be considered as low risk. This difference could possibly be explained by
387
the smaller sample size in the former study.
M AN U
SC
RI PT
378
A previous model-based cost-effectiveness analysis suggested that additional fFN
389
testing in women with a cervical length between 10 and 30 mm is a cost-effective strategy,
390
without compromising neonatal health.9 Our results raise the interesting question whether the
391
use of the quantitative fFN test in addition to cervical length measurement is cost-effective,
392
since based on our results both in women with a short cervix (<15 mm) and women with a
393
long cervix (>30 mm) a quantitative fFN test should be performed as well. In the future, we
394
will perform a cost-effectiveness analysis to compare different strategies; cervical length
395
alone, quantitative fFN testing alone (for clinical settings where cervical length measurement
396
is not available), qualitative fFN testing in combination with cervical length, and quantitative
397
fFN in combination with cervical length. Moreover, further future research should focus on
398
the performance of the quantitative fFN test in subgroups such as nulliparous women, women
AC C
EP
TE D
388
15
ACCEPTED MANUSCRIPT
399
with a previous preterm birth and women with a multiple pregnancy, and in a wider range of
400
ethnicities.
401
Acknowledgements
403
First and most important, we would like to thank all women that participated in the study. We
404
also greatly acknowledge the efforts of all obstetrical residents, gynaecologists and midwifes
405
in the perinatal centres that helped us to include women during their shifts. The authors thank
406
Maureen TM Franssen, MD, PhD, for acquisition of data and coordination in the University
407
Medical Centre Groningen.
M AN U
408 409
410
415 416
417
EP
414
AC C
413
TE D
411 412
SC
RI PT
402
418
419
16
ACCEPTED MANUSCRIPT
References
1.
RI PT
2.
Blencowe H, Cousens S, Oestergaard MZ, et al. National, regional, and worldwide estimates of preterm birth rates in the year 2010 with time trends since 1990 for selected countries: a systematic analysis and implications. Lancet. 2012;379(9832):2162-2172. Liu L, Johnson HL, Cousens S, et al. Global, regional, and national causes of child mortality: an updated systematic analysis for 2010 with time trends since 2000. Lancet. 2012;379(9832):2151-2161. McPheeters ML, Miller WC, Hartmann KE, et al. The epidemiology of threatened preterm labor: a prospective cohort study. Am. J. Obstet. Gynecol. 2005;192(4):1325-1329. Sanchez-Ramos L, Delke I, Zamora J, Kaunitz AM. Fetal fibronectin as a short-term predictor of preterm birth in symptomatic patients: a meta-analysis. Obstet. Gynecol. 2009;114(3):631640. Sotiriadis A, Papatheodorou S, Kavvadias A, Makrydimas G. Transvaginal cervical length measurement for prediction of preterm birth in women with threatened preterm labor: a meta-analysis. Ultrasound Obstet. Gynecol. 2010;35(1):54-64. Matsuura H, Takio K, Titani K, et al. The oncofetal structure of human fibronectin defined by monoclonal antibody FDC-6. Unique structural requirement for the antigenic specificity provided by a glycosylhexapeptide. J. Biol. Chem. 1988;263(7):3314-3322. Defranco EA, Lewis DF, Odibo AO. Improving the screening accuracy for preterm labor: is the combination of fetal fibronectin and cervical length in symptomatic patients a useful predictor of preterm birth? A systematic review. Am. J. Obstet. Gynecol. 2012. van Baaren GJ, Vis JY, Wilms FF, et al. Predictive value of cervical length measurement and fibronectin testing in threatened preterm labor. Obstet. Gynecol. 2014;123(6):1185-1192. van Baaren GJ, Vis JY, Grobman WA, Bossuyt PM, Opmeer BC, Mol BW. Cost-effectiveness analysis of cervical length measurement and fibronectin testing in women with threatened preterm labor. Am. J. Obstet. Gynecol. 2013;209(5):436.e431-438. Abbott DS, Radford SK, Seed PT, Tribe RM, Shennan AH. Evaluation of a quantitative fetal fibronectin test for spontaneous preterm birth in symptomatic women. Am. J. Obstet. Gynecol. 2013;208(2):122-126. Centra M, Coata G, Picchiassi E, et al. Evaluation of quantitative fFn test in predicting the risk of preterm birth. J. Perinat. Med. 2016. Kurtzman J, Chandiramani M, Briley A, Poston L, Das A, Shennan A. Quantitative fetal fibronectin screening in asymptomatic high-risk patients and the spectrum of risk for recurrent preterm delivery. Am. J. Obstet. Gynecol. 2009;200(3):263-266. Harrell FE, Jr., Lee KL, Mark DB. Multivariable prognostic models: issues in developing models, evaluating assumptions and adequacy, and measuring and reducing errors. Stat. Med. 1996;15(4):361-387. Van Houwelingen JC, Le CS. Predictive value of statistical models. Stat. Med. 1990;9(11):1303-1325. Debray TP, Moons KG, Ahmed I, Koffijberg H, Riley RD. A framework for developing, implementing, and evaluating clinical prediction models in an individual participant data meta-analysis. Stat. Med. 2013;32(18):3158-3180. Crowley P. Prophylactic corticosteroids for preterm birth. The Cochrane database of systematic reviews. 2000(2):Cd000065. Tsatsaris V, Cabrol D, Carbonne B. Pharmacokinetics of tocolytic agents. Clin. Pharmacokinet. 2004;43(13):833-844. Haas DM, Imperiale TF, Kirkpatrick PR, Klein RW, Zollinger TW, Golichowski AM. Tocolytic therapy: a meta-analysis and decision analysis. Obstet. Gynecol. 2009;113(3):585-594.
3. 4.
SC
5.
8. 9.
10.
11. 12.
13.
EP
7.
TE D
M AN U
6.
AC C
420 421 422 423 424 425 426 427 428 429 430 431 432 433 434 435 436 437 438 439 440 441 442 443 444 445 446 447 448 449 450 451 452 453 454 455 456 457 458 459 460 461 462 463 464 465 466 467 468 469
14.
15.
16.
17. 18.
17
ACCEPTED MANUSCRIPT
20.
TE D
M AN U
SC
RI PT
21.
Organization WH. WHO Recommendations on Interventions te Improve Preterm Birth Outcomes. 2015; http://apps.who.int/iris/bitstream/10665/183055/1/WHO_RHR_15.16_eng.pdf?ua=1. Abbott DS, Hezelgrave NL, Seed PT, et al. Quantitative fetal fibronectin to predict preterm birth in asymptomatic women at high risk. Obstet. Gynecol. 2015;125(5):1168-1176. Bruijn M, Vis JY, Wilms FF, et al. Quantitative fetal fibronectin testing in combination with cervical length measurement in the prediction of spontaneous preterm delivery in symptomatic women. BJOG. 2015.
EP
478 479 480 481 482 483 484 485 486 487 488 489 490 491 492 493 494 495 496 497 498 499 500 501 502 503 504 505 506 507 508 509 510 511 512 513 514 515 516 517 518 519
19.
AC C
470 471 472 473 474 475 476 477
18
ACCEPTED MANUSCRIPT
Table 1. Baseline characteristics of the study population (n=455) Characteristics
Values 29.5 ± 5.2
Gestational age at study entry
29.6 (26.7 – 31.6)
Body-Mass Index kg/m2* (n=429)
24.5 (22.0 – 28.0)
Caucasian race
352 (77) 41 (9) 249 (55)
Previous preterm birth <37 wks
72 (16)
Previous preterm birth <34 wks
50(11)
Multifetal gestation
67 (15)
M AN U
Nulliparous
SC
Maternal smoking* (n=413)
RI PT
Maternal age – yr
Symptoms of preterm labor: Contractions
395 (87)
Vaginal blood loss
64 (14)
Abdominal or back pain
320 (70)
Last prenatal visit in:
225 (51)
General hospital Primary care practices Fetal fibronectin
EP
Qualitative positive result
TE D
Perinatal center
Quantitative result - ng/mL Cervical length – mm
120 (26) 108 (24)
197 (43) 34.0 (8.0 – 215) 21.3 ± 9.5
Data are n (%), mean ± standard deviation, or median (interquartile range). * Data are missing.
AC C
520 521
19
ACCEPTED MANUSCRIPT
Fetal fibronectin group 10-49 ng/mL
50-199 ng/mL
200-499 ng/mL
≥500 ng/mL
Total
<15mm
19 (1 PTB - 5%)
18 (1 PTB - 6%)
19 (5 PTB - 26%)
33 (13 PTB - 39%)
14 (9 PTB - 64%)
103 (29 PTB - 28%)
15mm-29mm
78 (2 PTB - 3%)
86 (1 PTB - 1%)
45 (4 PTB - 9%)
44 (7 PTB - 16%)
17 (3 PTB - 18%)
270 (17 PTB - 6%)
30mm-50mm
37 (0 PTB - 0%)
20 (0 PTB - 0%)
12 (0 PTB - 0%)
7 (0 PTB - 0%)
6 (2 PTB - 33%)
82 (2 PTB - 2%)
Total
134 (3 PTB - 2%)
124 (2 PTB - 2%)
76 (9 PTB - 12%)
84 (20 PTB - 24%)
37 (14 PTB - 38%)
455 (48 PTB - 11%)
AC C
EP
TE D
PTB = preterm birth within 7 days
SC
<10 ng/mL
M AN U
Cervical length group
RI PT
Table 2. Risk stratification of preterm birth within seven days using quantitative fetal fibronectin in combination with cervical length (n=455).
20
ACCEPTED MANUSCRIPT
10-49 ng/mL
50-199 ng/mL
200-499 ng/mL
≥500 ng/mL
Total
<15mm
14 (1 PTB - 7%)
14 (1 PTB - 7%)
10 (2 PTB - 20%)
23 (9 PTB - 39%)
12 (8 PTB - 67%)
73 (21 PTB - 29%)
15mm-29mm
47 (1 PTB - 2%)
43 (1 PTB - 2%)
22 (2 PTB - 9%)
22 (2 PTB - 9%)
14 (1 PTB - 7%)
148 (7 PTB - 5%)
30mm-50mm
11 (0 PTB - 0%)
7 (0 PTB - 0%)
3 (0 PTB - 0%)
3 (0 PTB - 0%)
4 (1 PTB - 25%)
28 (1 PTB - 4%)
Total
72 (2 PTB - 3%)
64 (2 PTB - 3%)
35 (4 PTB - 11%)
30 (10 PTB - 33%)
249 (29 PTB - 12%)
48 (11 PTB - 23%)
AC C
EP
TE D
PTB = preterm birth within 7 days
SC
<10 ng/mL
M AN U
Cervical length group
RI PT
Table 3. Risk stratification of preterm birth within seven days using quantitative fetal fibronectin in combination with cervical length in nulliparous women only (n=249).
21
ACCEPTED MANUSCRIPT
<10 ng/mL
10-49 ng/mL
50-199 ng/mL
200-499 ng/mL
≥500 ng/mL
Total
<15mm
17 (1 PTB - 6%)
17 (1 PTB - 6%)
16 (4 PTB - 25%)
28 (10 PTB - 36%)
14 (9 PTB - 64%)
92 (25 PTB - 27%)
15mm-29mm
67 (2 PTB - 3%)
69 (1 PTB - 1%)
33 (2 PTB - 6%)
36 (3 PTB - 8%)
222 (11 PTB - 5%)
30mm-50mm
33 (0 PTB - 0%)
17 (0 PTB - 0%)
8 (0 PTB - 0%)
SC
17 (3 PTB - 18%)
6 (0 PTB - 0%)
5 (1 PTB - 33%)
69 (1 PTB - 2%)
Total
117 (3 PTB - 3%)
103 (2 PTB - 2%)
57 (6 PTB - 11%)
70 (13 PTB - 19%)
36 (13 PTB - 36%)
383 (37 PTB - 10%)
AC C
EP
TE D
PTB = preterm birth within 7 days
M AN U
Cervical length group
RI PT
Table 4. Risk stratification of preterm birth within seven days using quantitative fetal fibronectin in combination with cervical length in women without a history of preterm birth only (n=383).
22
ACCEPTED MANUSCRIPT
Table 5. Multivariable models for predicting spontaneous preterm birth within seven days in symptomatic women, including cervical length and quantitative fetal fibronectin or qualitative fetal fibronectin as predictors.
OR (95% CI)
Intercept
-2.0
Cervical length (mm)
-0.08
0.92 (0.89 – 0.96)
2.1
7.9 (3.2 – 19)
Qualitative fFN
SC
*Coefficients were shrunken with an average shrinkage factor 0.96 AUC = 0.83 (95% CI 0.77 – 0.88) Nagelkerke R2 = 0.27 -1.9 Intercept Cervical length (mm)
-0.08
0.92 (0.89 – 0.96)
Quantitative fFN (ng/mL)
0.006
1.006 (1.004 – 1.007)
AC C
EP
TE D
M AN U
*Coefficients were shrunken with an average shrinkage factor 0.99 AUC = 0.84 (95% CI 0.78 – 0.89) Nagelkerke R2 = 0.32 fFN = fetal fibronectin
RI PT
Beta*
23
ACCEPTED MANUSCRIPT
Table 6. Reclassification table, showing the number of women reclassified from high risk to low risk and vice versa using the model with cervical length and qualitative fetal fibronectin compared to the model with cervical length and quantitative fetal fibronectin.
Low risk*
High risk
Low risk*
222 (3 PTB – 1%)
25 (3 PTB – 12%)
High risk
37 (4 PTB – 11%)
171 (38 PTB – 22%)
Total
259 (7 PTB – 3%)
196 (41 PTB – 21%)
Total
246 (6 PTB – 2%)
209 (42 PTB – 20%)
455 (48 PTB – 11%)
SC
CL & qualitative fFN
RI PT
CL & quantitative fFN
AC C
EP
TE D
M AN U
CL = cervical length, fFN = fetal fibronectin, PTB = preterm birth within 7 days * Low risk is defined as a risk of PTB <5%
24
ACCEPTED MANUSCRIPT
AC C
EP
TE D
M AN U
SC
RI PT
Figure 1. Participant flow diagram.
25
AC C
EP
TE D
M AN U
SC
RI PT
ACCEPTED MANUSCRIPT
S0002-9378(16)30577-4
DOI:
10.1016/j.ajog.2016.08.012
Reference:
YMOB 11266
To appear in:
American Journal of Obstetrics and Gynecology
Received Date: 6 June 2016 Revised Date:
21 July 2016
Accepted Date: 8 August 2016
Please cite this article as: Bruijn MM, Kamphuis EI, Hoesli IM, Martinez De Tejada B, Loccufier AR, Kühnert M, Helmer H, Franz M, Porath MM, Oudijk MA, Jacquemyn Y, Schulzke SM, Vetter G, Hoste G, Vis JY, Kok M, Mol BW, van Baaren G-J, The predictive value of quantitative fibronectin testing in combination with cervical length measurement in symptomatic women, American Journal of Obstetrics and Gynecology (2016), doi: 10.1016/j.ajog.2016.08.012. This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
ACCEPTED MANUSCRIPT
1
The predictive value of quantitative fibronectin testing in combination with cervical
2
length measurement in symptomatic women.
3 Merel MC BRUIJN1*, MD, Esme I KAMPHUIS1,2, MD, Irene M HOESLI3, MD, PhD, Begoña
5
MARTINEZ DE TEJADA4, MD, PhD, Anne R LOCCUFIER5, MD, Maritta KÜHNERT6, MD, PhD,
6
Hanns HELMER7, MD, PhD, Marie FRANZ7, MD, Martina M PORATH8, MD, PhD, Martijn A
7
OUDIJK1,9, MD, PhD, Yves JACQUEMYN10, MD, PhD, Sven M SCHULZKE11, MD, Grit
8
VETTER3, MD, Griet HOSTE5, MD, Jolande Y VIS12, MD, PhD, Marjolein KOK1, MD, PhD, Ben
9
WJ MOL13, MD, PhD, Gert-Jan VAN BAAREN1, MD, PhD
M AN U
TE D
27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47
* Corresponding author: Merel MC Bruijn, Department of Obstetrics and Gynecology, Academic Medical Centre, 1105 DE, Amsterdam, The Netherlands, Email: [email protected].
EP
26
1. Dept. Obstetrics & Gynecology, Academic Medical Centre, Amsterdam, Netherlands. 2. Dept. Obstetrics & Gynecology, VU University Medical Centre, Amsterdam, Netherlands. 3. Dept. Obstetrics & Gynecology, University Hospital Basel, Basel, Switzerland. 4. Dept. Obstetrics & Gynecology, University Hospitals of Geneva and Faculty of Medicine, Geneva, Switzerland. 5. Dept. Obstetrics & Gynecology, Hospital St Jan, Bruges, Belgium. 6. Dept. Obstetrics & Gynecology, University Hospital of Marburg, Germany. 7. Dept. Obstetrics & Maternal-Fetal Medicine, Medical University Vienna, Vienna, Austria. 8. Dept. Obstetrics & Gynecology, Maxima Medical Centre, Veldhoven, Netherlands. 9. University Medical Centre Utrecht, Netherlands. 10. Dept. Obstetrics & Gynecology, Antwerp University Hospital, Antwerp, Belgium. 11. Dept. Neonatology, University Children's Hospital Basel, Switzerland. 12. Clinical Chemistry and Hematology, University Medical Centre Utrecht, Netherlands. 13. School of Medicine, University of Adelaide, Adelaide, Australia.
AC C
11 12 13 14 15 16 17 18 19 20 21 22 23 24 25
SC
10
RI PT
4
Presented as a poster at the 35th Annual Meeting of the Society of Maternal Fetal Medicine, February 2 -7, 2015, San Diego, California. Financial disclosure: Ben WJ Mol is consultant for ObsEva. Payments go to his institute. The remaining authors report no conflict of interest. The study was conducted with financial support from Hologic USA (Marlborough, MA). They had no role in study design, interpretation of data or writing of the report. Word count abstract: 368 Word count manuscript: 3075
1
ACCEPTED MANUSCRIPT
48
Quantitative fetal fibronectin testing improves the prediction of spontaneous preterm birth
49
within seven days in symptomatic women undergoing cervical length measurement.
50
Short version of the article title: quantitative fFN and cervical length in the prediction of
52
preterm birth
RI PT
51
53 54
SC
55 56
M AN U
57 58 59 60 61 62
66 67 68 69 70 71 72 73 74
EP
65
AC C
64
TE D
63
75 76 77 78 79
2
ACCEPTED MANUSCRIPT
80
Abstract
81
Background: The combination of the qualitative fetal fibronectin test and cervical length
83
measurement has a high negative predictive value for preterm birth within seven days, but
84
positive prediction is poor. A new bedside quantitative fetal fibronectin test showed potential
85
additional value over the conventional qualitative test, but there is limited evidence on the
86
combination with cervical length measurement.
87
Objective: To compare quantitative fetal fibronectin and qualitative fetal fibronectin testing
88
in the prediction of spontaneous preterm birth within seven days in symptomatic women
89
undergoing cervical length measurement.
90
Study design: We performed a European multicenter cohort study in ten perinatal centers in
91
five countries. Women between 24 and 34 weeks of gestation with signs of active labor and
92
intact membranes underwent quantitative fibronectin testing and cervical length measurement.
93
We assessed the risk of preterm birth within seven days in predefined strata based on
94
fibronectin concentration and cervical length.
95
Results: Out of 455 included women, 48 (11%) delivered within seven days. A combination
96
of cervical length and qualitative fibronectin resulted in identification of 246 women at low
97
risk: 164 women with a cervix between 15 and 30 mm and a negative fibronectin test
98
(<50ng/mL) (preterm birth rate 2%), and 82 women with a cervix above 30 mm (preterm
99
birth rate 2%). Use of quantitative fibronectin alone resulted in a predicted risk of preterm
100
birth within seven days ranging from 2% in the group with the lowest fibronectin level
101
(<10ng/mL) to 38% in the group with the highest fibronectin level (>500ng/mL), with similar
102
accuracy as the combination of cervical length and qualitative fibronectin. Combining
103
cervical length and quantitative fibronectin resulted in identification of an additional 19
104
women at low risk (preterm birth rate 5%) using a threshold of 10ng/mL in women with a
AC C
EP
TE D
M AN U
SC
RI PT
82
3
ACCEPTED MANUSCRIPT
cervix below 15mm, and 6 women at high risk (preterm birth rate 33%) using a threshold of
106
>500ng/mL in women with a cervix above 30mm.
107
Conclusion: In women with threatened preterm birth, quantitative fibronectin testing alone
108
performs equal to the combination of cervical length and qualitative fibronectin. Possibly, the
109
combination of quantitative fibronectin testing and cervical length increases this predictive
110
capacity. Cost-effectiveness analysis as well as availability of these tests in a local setting
111
should determine the final choice.
SC
RI PT
105
M AN U
112 113
Keywords: cervical length, fetal fibronectin, prediction, predictive value, preterm birth,
114
threatened preterm labor
115
119
120
121
122
EP
118
AC C
117
TE D
116
123
124
4
ACCEPTED MANUSCRIPT
Introduction
126
Preterm birth, defined as birth before 37 weeks’ gestation, occurs worldwide in 15 million
127
babies and accounts for 11% of all live births.1 In 1.1 million of these births, prematurity
128
results in death of the child, thus making preterm birth the leading cause of neonatal
129
mortality, while in case of survival preterm birth is an important cause of morbidity.2
130
Accurate identification of women who will deliver in short time allows targeted interventions
131
that improve outcome among those who indeed deliver preterm, such as corticosteroids,
132
magnesium sulphate and in utero transfer to a perinatal center, thus avoiding unnecessary
133
interventions with potential side effects for women at low risk. However, the large majority of
134
women who present with symptoms of preterm labor will not deliver within seven days and
135
50% will deliver at term.3-5
M AN U
SC
RI PT
125
Among various methods to assess the risk of preterm birth, fetal fibronectin (fFN)
137
testing and cervical length measurement are widely used. Fetal fibronectin, a glycoprotein
138
found at the choriodecidual interface, is traditionally used as a binary bedside test providing a
139
positive or negative result based on a threshold of 50 ng/mL.6 The combination of cervical
140
length measurement and fFN testing has a high negative predictive value (>98%) for delivery
141
within seven days, but positive prediction is poor.7 The best strategy to combine the two tests
142
is additional fFN testing in women with a cervical length between 15 and 30 mm, reducing
143
the number of unnecessary referrals and admissions to perinatal centers.8 This approach
144
would also be cost saving without compromising neonatal health outcomes.9
EP
AC C
145
TE D
136
A new bedside quantitative fetal fibronectin test showed potential additional value
146
over the conventional qualitative test with a higher risk for preterm birth with increasing
147
levels of fibronectin. The positive predictive value for preterm birth on short term and preterm
148
birth before 34 weeks of gestation increased by changing the threshold from 50 ng/mL to 200
149
ng/mL or 500 ng/mL with minimal effect on the negative predictive value.10 11
5
ACCEPTED MANUSCRIPT
Here, we prospectively evaluated whether, in women with symptomatic preterm birth
151
the combination of quantitative fFN testing and cervical length measurement has better
152
predictive accuracy than qualitative fFN testing with cervical length measurement or than
153
quantitative fFN testing alone.
RI PT
150
154
Materials and methods
156
We conducted a European multicenter cohort study in 10 centers in five countries (four
157
centers in The Netherlands, two in Switzerland, two in Belgium, one in Germany and one in
158
Austria). They were all perinatal centers serving as tertiary referral centers for high-risk
159
obstetric patients with an annual number of women delivering between 1,000 (Antwerp
160
University Hospital) and 4,000 (University Hospitals of Geneva). The study protocol was
161
approved by the ethics committee of the Academic Medical Centre Amsterdam, the
162
University Hospitals of Geneva and the University Hospital Basel as well by the board of
163
directors of the other participating centers in Germany, Austria and Belgium. Written
164
informed consent was obtained from all participants prior to inclusion.
TE D
M AN U
SC
155
Women presenting with symptoms of preterm labor (more than three contractions per
166
30 minutes, vaginal blood loss or abdominal or back pain) who were between 24 and 34
167
weeks of gestation, and who had intact membranes were eligible for participation in the study.
168
Gestational age was based on first trimester ultrasound assessment. The women either
169
presented themselves directly at one of the participating centers or were referred by their
170
primary gynecologist from a general hospital (secondary care) or by a general practitioner or
171
midwife (primary care). Women who had received tocolytic treatment for more than 18 hours
172
were excluded from the study. Women with contraindications for tocolysis such as a lethal
173
congenital abnormality, suspected intrauterine infection, nonreassuring fetal status or
174
maternal distress requiring immediate intervention such as placental abruption or severe
AC C
EP
165
6
ACCEPTED MANUSCRIPT
vaginal blood loss were excluded. Other exclusion criteria were cervical dilatation of more
176
than three cm and triplet or higher order pregnancies. Also women who had iatrogenic
177
delivery within seven days after study entry for hypertensive disorders, fetal distress, or other
178
reasons for immediate delivery were excluded. We intended to include 500 women, i.e. 100
179
women per country. We derived a sample size with the aim to estimate the sensitivity of the
180
quantitative fFN test with sufficient precision. If the performance of the quantitative fFN test
181
would resemble the qualitative fFN test with a cut off value of 50 ng/ml, the expected
182
sensitivity would be 91% (based on a Dutch cohort study from 2009). Inclusion of 100
183
patients with subsequent preterm delivery would produce a two-sided 95% confidence
184
interval with a width equal to 0.112 around the expected sensitivity of 91%. In view of an
185
expected prevalence of preterm birth of 20%, we would have to include a total of 500 women
186
with complaints of threatened labor.
M AN U
SC
RI PT
175
After study enrolment, all women underwent cervical length measurement and fFN
188
testing. Cervical length was measured by transvaginal ultrasound. fFN testing was performed
189
using the quantitative Rapid fFN 10Q analyzer (Hologic®) according to manufacturer’s
190
instructions. A 50 ng/mL cut-off was used for the qualitative result of the fFN test (positive,
191
negative). The qualitative fFN result was available immediately for the use in patient care, but
192
the quantitative fFN result remained blinded until after delivery (the analyzer generated a
193
random 3-letter result code). When the study started, both cervical length measurement and
194
the qualitative fFN test were standard care in all participating centers for risk stratification in
195
women with symptoms of preterm labor. Clinicians were trained in transvaginal
196
ultrasonography and performed cervical length measurements themselves. Furthermore, all
197
involved personnel were trained in the use of the specimen collection and the fFN analyzer.
198
Preferably, a specimen had to be collected before the vaginal examination or cervical length
199
measurement. Furthermore, additional information on the maternal condition was obtained
AC C
EP
TE D
187
7
ACCEPTED MANUSCRIPT
including vaginal digital examination, maternal serum for CRP and leukocytes, urine and
201
vaginal culture, and blood pressure. In addition, data collection included information on
202
factors that might influence the fFN result, such as the use of vaginal soap or sexual
203
intercourse within 24 hours before performance of the fFN test and vaginal bleeding during
204
testing.
RI PT
200
Treatment recommendations were based on previous research.8 We recommended
206
starting tocolysis and steroids for fetal lung maturation in high-risk women, defined as
207
women with a cervical length below 15 mm and in women with a cervical length between 15
208
and 30 mm with a positive fFN result. For low-risk women with a cervical length above 30
209
mm, we recommended to withhold steroids and tocolysis. For women with a cervical length
210
between 15 and 30 mm in combination with a negative fFN test, the clinician on call decided
211
whether to start tocolysis and steroids or not. Tocolytic drugs that could be prescribed were
212
beta-sympathomimetics, nifedipine, atosiban, indomethacin. Magnesium sulphate for
213
neuroprotection and antibiotics were given on consideration of the attending physician.
TE D
M AN U
SC
205
The primary outcome was spontaneous preterm birth within seven days after study
215
entry. We used logistic regression analyses with an interaction term to assess the relationship
216
between potential influencing factors and the quantitative fFN result, such as transvaginal
217
ultrasound, vaginal examination, the use of vaginal soap or sexual intercourse within 24 hours
218
before performance of the fFN test and vaginal bleeding during testing. Quantitative fFN was
219
analyzed as a continuous variable, all other factors as dichotomous variables. Linearity of the
220
association between quantitative fFN and the risk of preterm birth was assessed using
221
restricted cubic spline analyses, using 3 knots. If one of the interaction terms was significant,
222
the factor was considered to have a confounding relation with the quantitative fFN result and
223
women with that factor were excluded from the analyses Descriptive statistics were calculated
224
for baseline demographics and obstetric characteristics.
AC C
EP
214
8
ACCEPTED MANUSCRIPT
For risk stratification, thresholds of 10, 50, 200 and 500 ng/mL for the quantitative
226
fFN test were predefined before analysis and cervical length was divided into groups of <15
227
mm, 15-29 mm and 30-50 mm. These thresholds were based on previous studies.8,10,12 The
228
rates of preterm birth within seven days were calculated within the corresponding strata. We
229
considered a risk of delivery less than 5% as low risk. This threshold has been derived from
230
the interpretation of cervical length, because this is currently used in clinical decision-making
231
in women with signs of preterm labor. A cervical length measurement with a threshold
232
between 20 and 30 mm corresponds to a post-test probability of delivery within seven days of
233
4-5%, compared to a pre-test probability of 11% if cervical length measurement is not
234
performed.5 To assess if more low-risk women could be identified using the quantitative fFN
235
test with various thresholds, we compared the accuracy to that of the qualitative fFN test
236
(threshold 50 ng/mL) in combination with cervical length measurement. As the fFN test may
237
behave differently according to parity and obstetric history, we additionally performed
238
subgroup analyses for nulliparous women and women without a history of preterm birth.
TE D
M AN U
SC
RI PT
225
We used logistic regression analyses to assess the univariable association between
240
cervical length, qualitative fFN and quantitative fFN, and preterm birth within seven days.
241
Furthermore, we constructed two multivariable prediction models to predict preterm birth
242
within seven days: one model including the variables cervical length and qualitative fFN and
243
one model including the variables cervical length and quantitative fFN. Both cervical length
244
and quantitative fFN were analyzed as continuous variables. Linearity of the association
245
between the variables cervical length and quantitative fFN and the risk of preterm birth was
246
assessed using restricted cubic spline analyses. We used bootstrapping techniques for internal
247
validation of the models to correct for overfitting. Two hundred samples with the same size of
248
the original data set were drawn from the original data set with replacement. A shrinkage
249
factor was derived from these analyses and used to adjust regression coefficients.13,14 We
AC C
EP
239
9
ACCEPTED MANUSCRIPT
compared the models in terms of overall fit using Nagelkerke R2 and discrimination using the
251
area under the receiver operating characteristics curve (AUC). To determine whether
252
quantitative fFN as compared to qualitative fFN in combination with cervical length improves
253
the capability of the model to identify low-risk women (<5% risk), we compared the two
254
models in terms of reclassification.
256
Data analyses were performed in SPSS version 20.0 (Chicago, IL, USA) and R version 2.10.0 (The R foundation for Statistical Computing, 2009).
SC
255
RI PT
250
257
Results
259
Between January 2013 and May 2014, a total of 532 women were potentially eligible for
260
study enrolment, of whom 52 women did not meet the inclusion criteria, 11 had one or more
261
exclusion criteria and seven did not give their consent to participate in the study. In the
262
remaining 462 participating women, we performed fFN testing and cervical length
263
measurement. The fFN result and cervical length were both not recorded in three women. We
264
excluded one woman who had an elective caesarean section within seven days after
265
enrollment because of fetal distress. In total, 455 women were available for analysis (Figure
266
1); 192 included in the Netherlands, 146 in Switzerland, 55 in Belgium, 47 in Germany and
267
15 in Austria. Baseline demographics and obstetric characteristics for the study participants
268
are shown in Table 1.
TE D
EP
AC C
269
M AN U
258
Logistic regression analysis showed no interaction between quantitative fFN and
270
cervical length at transvaginal ultrasound (n=211, p=0.98), vaginal examination (n=110,
271
p=0.44), the use of vaginal soap (n=90, p=0.21) or sexual intercourse (n=20, p=0.76) within
272
24 hours before performing the fFN test. Similarly, there was no interaction with vaginal
273
bleeding during testing (n=64, p=0.91). Therefore, women with one of these characteristics
274
were not excluded from the analyses.
10
ACCEPTED MANUSCRIPT
Table 2 shows the risk stratification. The risk of preterm birth within seven days
276
increased with increasing fFN concentration (from 2% when fFN was <10 ng/mL to 38%
277
when ≥500 ng/mL) and shortening of the cervix (from 2% when cervical length was ≥30mm
278
to 28% when cervical length was <15mm). Using the conventional threshold of 50 ng/mL
279
when cervical length was between 15 and 30 mm, 164 women were identified as low risk, of
280
whom three (2%) delivered within seven days. Changing of the threshold from 50 to 200
281
ng/mL in women with a cervical length between 15 and 30 m would label 45 women as high
282
risk, as four women (9%) delivered within seven days. In 103 women with a cervical length
283
below 15 mm, using a threshold of 10 ng/mL resulted in identification of 19 women as low
284
risk, of whom one (5%) delivered within seven days. Comparison of this proportion to
285
women with a fFN concentration above 10 ng/mL using a Fisher’s exact test, showed the
286
difference to be statistically significant (p=0.01). Two out of 82 women with a cervical length
287
above 30 mm (2%) delivered within seven days. Both had a fFN concentration of more than
288
500 ng/mL (33% of this group), which showed to be significantly different from the women
289
with a fFN concentration below 500 ng/mL (p=0.001).
TE D
M AN U
SC
RI PT
275
Table 3 and table 4 show the risk stratification for nulliparous women only (n=249)
291
and women without a history of preterm birth (n=383), respectively. The results were similar
292
to those of the complete group, showing an increased risk of preterm birth within seven days
293
with increasing fFN concentration and shortening of the cervix. Women with a long cervix
294
were still identified as high risk with a threshold of > 500 ng/mL, although this did not show
295
to be significant (table 3: p=0.14, table 4: p=0.07). A threshold of 10 ng/mL in women with a
296
short cervix did not help to identify more women at low risk.
AC C
EP
290
297
Cubic spline analyses showed a linear relationship between both quantitative fFN and
298
cervical length and the risk of preterm birth within seven days, and were therefore analyzed as
299
continuous variables. In univariable logistic regression analysis, a positive qualitative fFN test
11
ACCEPTED MANUSCRIPT
was associated with a higher risk of preterm birth within seven days (OR 11 (95% CI 4.7 –
301
27)). The same was the case for the quantitative fFN concentration (OR 1.006 per ng/mL
302
(95% CI 1.005 – 1.008)). A longer cervix was associated with a lower risk of preterm birth
303
within seven days (OR 0.90 per mm (95% CI 0.86 – 0.94)).
RI PT
300
Table 5 shows the two multivariable models. The model with qualitative fFN and
305
cervical length was presented after shrinkage with an average shrinkage factor of 0.96 and the
306
model with quantitative fFN and cervical length with an average shrinkage factor of 0.99. The
307
model with cervical length and qualitative fFN had an AUC of 0.83 (95% CI 0.77 – 0.88) and
308
a Nagelkerke R2 of 0.27. The model with CL and quantitative fFN had an AUC of 0.84 (95%
309
CI 0.78 – 0.89) and a Nagelkerke R2 of 0.32. Quantitative fFN alone had an AUC of 0.82
310
(95% CI 0.76 – 0.89) and a Nagelkerke R2 of 0.25.
M AN U
SC
304
Table 6 shows the reclassification of participants based on a multivariable model with
312
quantitative fFN and cervical length compared to a model with qualitative fFN and cervical
313
length. Using quantitative fFN and cervical length resulted in 62 women (14%) being
314
reclassified; 37 women (8%) were reclassified as low risk, of whom four (11%) delivered
315
within seven days. On the other hand, 25 women were reclassified as high risk, of whom three
316
(12%) delivered within seven days.
317
EP
TE D
311
Comment
319
This study assessed the diagnostic accuracy of the quantitative fFN test in combination with
320
cervical length measurement, as compared to the qualitative test in combination with cervical
321
length, and the quantitative fFN test alone, in the prediction of spontaneous preterm birth
322
within seven days in symptomatic women. The results demonstrated that quantification of
323
fFN provides additional information on risk stratification as the risk of preterm birth increases
324
with increasing fFN concentrations. In terms of reclassification, quantitative fFN in
AC C
318
12
ACCEPTED MANUSCRIPT
combination with cervical length does not improve the identification of women with a low
326
risk (<5%) of preterm birth within seven days compared to the qualitative fFN test in
327
combination with cervical length. However, when looking in detail using the risk
328
stratification, we see that with a threshold of 10 ng/mL in women with a short cervix (<15
329
mm), who are in current practice treated as high risk without additional fFN testing, more
330
low-risk women could be identified. On the other hand, a threshold of 500 ng/mL in women
331
with a cervix above 30 mm, who are now treated as low risk without additional fFN testing,
332
will help identifying women with a high risk to deliver within seven days.
SC
RI PT
325
A strength of this study is that data were derived from a well-described, large,
334
prospectively collected cohort of symptomatic women, representing a European population
335
with mostly Caucasian women. The qualitative fFN test was collected according to protocol
336
and the result was used to determine further management, whereas the quantitative result
337
remained blinded until the end of the study.
M AN U
333
We are aware that preterm delivery rates, sociodemographic characteristics and the
339
obstetrical management may differ from country to country, and even from center to center.
340
In our analysis, we did not account for these between-center differences and the assumption
341
that there are no differences may have serious implications if untrue. Therefore, we performed
342
additional analyses to investigate the potential between-center heterogeneity in our study. We
343
have included the centers in the models as dummy variables, to estimate a stratified intercept
344
for each center. This methodology has been described to account for heterogeneity between
345
study populations in individual participant data (IPD) meta-analysis, which can be compared
346
to the issue of different centers in our study.15 A random effects analysis of the intercepts
347
showed that there was no heterogeneity between centers (I2=0%). When comparing the
348
performance of these models (with centers) with our presented models (without centers) we
349
found that both model fit and discrimination were slightly better for the former models with
AC C
EP
TE D
338
13
ACCEPTED MANUSCRIPT
AUCs of 0.84 (qualitative fFN and cervical length), 0.86 (quantitative fFN and cervical
351
length) and 0.84 (quantitative fFN) compared with 0.83, 0.84, 0.82, respectively. However,
352
the increase in performance is only marginally improved given the increased complexity of
353
the models and the loss of generalizability. Therefore we decided to only present the results of
354
the models without adjustment for centers
RI PT
350
It is questionable whether tocolysis has influenced our results by reducing the number
356
of women who delivered within seven days after study entry. Tocolytics have a short half-life
357
and are in general only given for the first 48 hours after admission in order to delay delivery
358
long enough to administer antenatal corticosteroids causing improved neonatal outcomes.16,17
359
Good evidence that tocolysis, given during the first 48 hours after admission, delays delivery
360
after seven additional days have passed, is lacking.18 The World Health Organization
361
recommends not to use tocolytic treatments, other than to prolong pregnancy (up to 48 hours)
362
to provide a window for administration of antenatal corticosteroids and/or in-utero transfer,
363
because of the potential risks and lack of evidence of the efficacy.19 Therefore, we think that it
364
is unlikely that tocolysis have influenced the results of this study.
TE D
M AN U
SC
355
Previous studies have reported improved prediction of preterm birth using
366
quantification of fFN in both symptomatic and high-risk asymptomatic women.10,11,20 Abbott
367
et al. reported increasing spontaneous preterm birth rates within two weeks with increasing
368
fFN concentrations, ranging from 2% in the lowest fFN group (<10 ng/mL) to 46% in the
369
highest fFN group (>500 ng/mL), in symptomatic women. Changing the threshold from the
370
conventional 50 ng/mL to 200 ng/mL led to an increase in the positive predictive value from
371
20% to 37% for the prediction of preterm birth within two weeks with minimal effect on the
372
negative predictive value.10 A recently published study from Centra et al. supported these
373
results.11 Our results confirm that compared to qualitative fFN, quantification of fFN provides
374
better risk stratification across the risk range, showing an increasing preterm birth rate from
AC C
EP
365
14
ACCEPTED MANUSCRIPT
375
3% in the lowest fFN group (<10 ng/mL) to 38% in the highest fFN group (>500 ng/mL).
376
However, in those previous studies, the quantitative fFN test was not evaluated in
377
combination with cervical length measurement. We recently described similar analyses performed on data of a Dutch prospective
379
nationwide cohort study.21 In this study we evaluated the quantitative fFN test on frozen fFN
380
samples obtained from symptomatic women with a cervical length below 30 mm. The results
381
were in line with the results of the current study, showing an increasing risk of delivery within
382
seven days with increasing fFN concentrations. However, using a threshold of 10 ng/mL in
383
women with a short cervix (<15 mm), our previous study showed that no more women at low
384
risk of preterm birth within seven days were identified, whereas the present results showed
385
that 19 women were identified of whom one (5%) delivered within seven days, indicating that
386
these women could be considered as low risk. This difference could possibly be explained by
387
the smaller sample size in the former study.
M AN U
SC
RI PT
378
A previous model-based cost-effectiveness analysis suggested that additional fFN
389
testing in women with a cervical length between 10 and 30 mm is a cost-effective strategy,
390
without compromising neonatal health.9 Our results raise the interesting question whether the
391
use of the quantitative fFN test in addition to cervical length measurement is cost-effective,
392
since based on our results both in women with a short cervix (<15 mm) and women with a
393
long cervix (>30 mm) a quantitative fFN test should be performed as well. In the future, we
394
will perform a cost-effectiveness analysis to compare different strategies; cervical length
395
alone, quantitative fFN testing alone (for clinical settings where cervical length measurement
396
is not available), qualitative fFN testing in combination with cervical length, and quantitative
397
fFN in combination with cervical length. Moreover, further future research should focus on
398
the performance of the quantitative fFN test in subgroups such as nulliparous women, women
AC C
EP
TE D
388
15
ACCEPTED MANUSCRIPT
399
with a previous preterm birth and women with a multiple pregnancy, and in a wider range of
400
ethnicities.
401
Acknowledgements
403
First and most important, we would like to thank all women that participated in the study. We
404
also greatly acknowledge the efforts of all obstetrical residents, gynaecologists and midwifes
405
in the perinatal centres that helped us to include women during their shifts. The authors thank
406
Maureen TM Franssen, MD, PhD, for acquisition of data and coordination in the University
407
Medical Centre Groningen.
M AN U
408 409
410
415 416
417
EP
414
AC C
413
TE D
411 412
SC
RI PT
402
418
419
16
ACCEPTED MANUSCRIPT
References
1.
RI PT
2.
Blencowe H, Cousens S, Oestergaard MZ, et al. National, regional, and worldwide estimates of preterm birth rates in the year 2010 with time trends since 1990 for selected countries: a systematic analysis and implications. Lancet. 2012;379(9832):2162-2172. Liu L, Johnson HL, Cousens S, et al. Global, regional, and national causes of child mortality: an updated systematic analysis for 2010 with time trends since 2000. Lancet. 2012;379(9832):2151-2161. McPheeters ML, Miller WC, Hartmann KE, et al. The epidemiology of threatened preterm labor: a prospective cohort study. Am. J. Obstet. Gynecol. 2005;192(4):1325-1329. Sanchez-Ramos L, Delke I, Zamora J, Kaunitz AM. Fetal fibronectin as a short-term predictor of preterm birth in symptomatic patients: a meta-analysis. Obstet. Gynecol. 2009;114(3):631640. Sotiriadis A, Papatheodorou S, Kavvadias A, Makrydimas G. Transvaginal cervical length measurement for prediction of preterm birth in women with threatened preterm labor: a meta-analysis. Ultrasound Obstet. Gynecol. 2010;35(1):54-64. Matsuura H, Takio K, Titani K, et al. The oncofetal structure of human fibronectin defined by monoclonal antibody FDC-6. Unique structural requirement for the antigenic specificity provided by a glycosylhexapeptide. J. Biol. Chem. 1988;263(7):3314-3322. Defranco EA, Lewis DF, Odibo AO. Improving the screening accuracy for preterm labor: is the combination of fetal fibronectin and cervical length in symptomatic patients a useful predictor of preterm birth? A systematic review. Am. J. Obstet. Gynecol. 2012. van Baaren GJ, Vis JY, Wilms FF, et al. Predictive value of cervical length measurement and fibronectin testing in threatened preterm labor. Obstet. Gynecol. 2014;123(6):1185-1192. van Baaren GJ, Vis JY, Grobman WA, Bossuyt PM, Opmeer BC, Mol BW. Cost-effectiveness analysis of cervical length measurement and fibronectin testing in women with threatened preterm labor. Am. J. Obstet. Gynecol. 2013;209(5):436.e431-438. Abbott DS, Radford SK, Seed PT, Tribe RM, Shennan AH. Evaluation of a quantitative fetal fibronectin test for spontaneous preterm birth in symptomatic women. Am. J. Obstet. Gynecol. 2013;208(2):122-126. Centra M, Coata G, Picchiassi E, et al. Evaluation of quantitative fFn test in predicting the risk of preterm birth. J. Perinat. Med. 2016. Kurtzman J, Chandiramani M, Briley A, Poston L, Das A, Shennan A. Quantitative fetal fibronectin screening in asymptomatic high-risk patients and the spectrum of risk for recurrent preterm delivery. Am. J. Obstet. Gynecol. 2009;200(3):263-266. Harrell FE, Jr., Lee KL, Mark DB. Multivariable prognostic models: issues in developing models, evaluating assumptions and adequacy, and measuring and reducing errors. Stat. Med. 1996;15(4):361-387. Van Houwelingen JC, Le CS. Predictive value of statistical models. Stat. Med. 1990;9(11):1303-1325. Debray TP, Moons KG, Ahmed I, Koffijberg H, Riley RD. A framework for developing, implementing, and evaluating clinical prediction models in an individual participant data meta-analysis. Stat. Med. 2013;32(18):3158-3180. Crowley P. Prophylactic corticosteroids for preterm birth. The Cochrane database of systematic reviews. 2000(2):Cd000065. Tsatsaris V, Cabrol D, Carbonne B. Pharmacokinetics of tocolytic agents. Clin. Pharmacokinet. 2004;43(13):833-844. Haas DM, Imperiale TF, Kirkpatrick PR, Klein RW, Zollinger TW, Golichowski AM. Tocolytic therapy: a meta-analysis and decision analysis. Obstet. Gynecol. 2009;113(3):585-594.
3. 4.
SC
5.
8. 9.
10.
11. 12.
13.
EP
7.
TE D
M AN U
6.
AC C
420 421 422 423 424 425 426 427 428 429 430 431 432 433 434 435 436 437 438 439 440 441 442 443 444 445 446 447 448 449 450 451 452 453 454 455 456 457 458 459 460 461 462 463 464 465 466 467 468 469
14.
15.
16.
17. 18.
17
ACCEPTED MANUSCRIPT
20.
TE D
M AN U
SC
RI PT
21.
Organization WH. WHO Recommendations on Interventions te Improve Preterm Birth Outcomes. 2015; http://apps.who.int/iris/bitstream/10665/183055/1/WHO_RHR_15.16_eng.pdf?ua=1. Abbott DS, Hezelgrave NL, Seed PT, et al. Quantitative fetal fibronectin to predict preterm birth in asymptomatic women at high risk. Obstet. Gynecol. 2015;125(5):1168-1176. Bruijn M, Vis JY, Wilms FF, et al. Quantitative fetal fibronectin testing in combination with cervical length measurement in the prediction of spontaneous preterm delivery in symptomatic women. BJOG. 2015.
EP
478 479 480 481 482 483 484 485 486 487 488 489 490 491 492 493 494 495 496 497 498 499 500 501 502 503 504 505 506 507 508 509 510 511 512 513 514 515 516 517 518 519
19.
AC C
470 471 472 473 474 475 476 477
18
ACCEPTED MANUSCRIPT
Table 1. Baseline characteristics of the study population (n=455) Characteristics
Values 29.5 ± 5.2
Gestational age at study entry
29.6 (26.7 – 31.6)
Body-Mass Index kg/m2* (n=429)
24.5 (22.0 – 28.0)
Caucasian race
352 (77) 41 (9) 249 (55)
Previous preterm birth <37 wks
72 (16)
Previous preterm birth <34 wks
50(11)
Multifetal gestation
67 (15)
M AN U
Nulliparous
SC
Maternal smoking* (n=413)
RI PT
Maternal age – yr
Symptoms of preterm labor: Contractions
395 (87)
Vaginal blood loss
64 (14)
Abdominal or back pain
320 (70)
Last prenatal visit in:
225 (51)
General hospital Primary care practices Fetal fibronectin
EP
Qualitative positive result
TE D
Perinatal center
Quantitative result - ng/mL Cervical length – mm
120 (26) 108 (24)
197 (43) 34.0 (8.0 – 215) 21.3 ± 9.5
Data are n (%), mean ± standard deviation, or median (interquartile range). * Data are missing.
AC C
520 521
19
ACCEPTED MANUSCRIPT
Fetal fibronectin group 10-49 ng/mL
50-199 ng/mL
200-499 ng/mL
≥500 ng/mL
Total
<15mm
19 (1 PTB - 5%)
18 (1 PTB - 6%)
19 (5 PTB - 26%)
33 (13 PTB - 39%)
14 (9 PTB - 64%)
103 (29 PTB - 28%)
15mm-29mm
78 (2 PTB - 3%)
86 (1 PTB - 1%)
45 (4 PTB - 9%)
44 (7 PTB - 16%)
17 (3 PTB - 18%)
270 (17 PTB - 6%)
30mm-50mm
37 (0 PTB - 0%)
20 (0 PTB - 0%)
12 (0 PTB - 0%)
7 (0 PTB - 0%)
6 (2 PTB - 33%)
82 (2 PTB - 2%)
Total
134 (3 PTB - 2%)
124 (2 PTB - 2%)
76 (9 PTB - 12%)
84 (20 PTB - 24%)
37 (14 PTB - 38%)
455 (48 PTB - 11%)
AC C
EP
TE D
PTB = preterm birth within 7 days
SC
<10 ng/mL
M AN U
Cervical length group
RI PT
Table 2. Risk stratification of preterm birth within seven days using quantitative fetal fibronectin in combination with cervical length (n=455).
20
ACCEPTED MANUSCRIPT
10-49 ng/mL
50-199 ng/mL
200-499 ng/mL
≥500 ng/mL
Total
<15mm
14 (1 PTB - 7%)
14 (1 PTB - 7%)
10 (2 PTB - 20%)
23 (9 PTB - 39%)
12 (8 PTB - 67%)
73 (21 PTB - 29%)
15mm-29mm
47 (1 PTB - 2%)
43 (1 PTB - 2%)
22 (2 PTB - 9%)
22 (2 PTB - 9%)
14 (1 PTB - 7%)
148 (7 PTB - 5%)
30mm-50mm
11 (0 PTB - 0%)
7 (0 PTB - 0%)
3 (0 PTB - 0%)
3 (0 PTB - 0%)
4 (1 PTB - 25%)
28 (1 PTB - 4%)
Total
72 (2 PTB - 3%)
64 (2 PTB - 3%)
35 (4 PTB - 11%)
30 (10 PTB - 33%)
249 (29 PTB - 12%)
48 (11 PTB - 23%)
AC C
EP
TE D
PTB = preterm birth within 7 days
SC
<10 ng/mL
M AN U
Cervical length group
RI PT
Table 3. Risk stratification of preterm birth within seven days using quantitative fetal fibronectin in combination with cervical length in nulliparous women only (n=249).
21
ACCEPTED MANUSCRIPT
<10 ng/mL
10-49 ng/mL
50-199 ng/mL
200-499 ng/mL
≥500 ng/mL
Total
<15mm
17 (1 PTB - 6%)
17 (1 PTB - 6%)
16 (4 PTB - 25%)
28 (10 PTB - 36%)
14 (9 PTB - 64%)
92 (25 PTB - 27%)
15mm-29mm
67 (2 PTB - 3%)
69 (1 PTB - 1%)
33 (2 PTB - 6%)
36 (3 PTB - 8%)
222 (11 PTB - 5%)
30mm-50mm
33 (0 PTB - 0%)
17 (0 PTB - 0%)
8 (0 PTB - 0%)
SC
17 (3 PTB - 18%)
6 (0 PTB - 0%)
5 (1 PTB - 33%)
69 (1 PTB - 2%)
Total
117 (3 PTB - 3%)
103 (2 PTB - 2%)
57 (6 PTB - 11%)
70 (13 PTB - 19%)
36 (13 PTB - 36%)
383 (37 PTB - 10%)
AC C
EP
TE D
PTB = preterm birth within 7 days
M AN U
Cervical length group
RI PT
Table 4. Risk stratification of preterm birth within seven days using quantitative fetal fibronectin in combination with cervical length in women without a history of preterm birth only (n=383).
22
ACCEPTED MANUSCRIPT
Table 5. Multivariable models for predicting spontaneous preterm birth within seven days in symptomatic women, including cervical length and quantitative fetal fibronectin or qualitative fetal fibronectin as predictors.
OR (95% CI)
Intercept
-2.0
Cervical length (mm)
-0.08
0.92 (0.89 – 0.96)
2.1
7.9 (3.2 – 19)
Qualitative fFN
SC
*Coefficients were shrunken with an average shrinkage factor 0.96 AUC = 0.83 (95% CI 0.77 – 0.88) Nagelkerke R2 = 0.27 -1.9 Intercept Cervical length (mm)
-0.08
0.92 (0.89 – 0.96)
Quantitative fFN (ng/mL)
0.006
1.006 (1.004 – 1.007)
AC C
EP
TE D
M AN U
*Coefficients were shrunken with an average shrinkage factor 0.99 AUC = 0.84 (95% CI 0.78 – 0.89) Nagelkerke R2 = 0.32 fFN = fetal fibronectin
RI PT
Beta*
23
ACCEPTED MANUSCRIPT
Table 6. Reclassification table, showing the number of women reclassified from high risk to low risk and vice versa using the model with cervical length and qualitative fetal fibronectin compared to the model with cervical length and quantitative fetal fibronectin.
Low risk*
High risk
Low risk*
222 (3 PTB – 1%)
25 (3 PTB – 12%)
High risk
37 (4 PTB – 11%)
171 (38 PTB – 22%)
Total
259 (7 PTB – 3%)
196 (41 PTB – 21%)
Total
246 (6 PTB – 2%)
209 (42 PTB – 20%)
455 (48 PTB – 11%)
SC
CL & qualitative fFN
RI PT
CL & quantitative fFN
AC C
EP
TE D
M AN U
CL = cervical length, fFN = fetal fibronectin, PTB = preterm birth within 7 days * Low risk is defined as a risk of PTB <5%
24
ACCEPTED MANUSCRIPT
AC C
EP
TE D
M AN U
SC
RI PT
Figure 1. Participant flow diagram.
25
AC C
EP
TE D
M AN U
SC
RI PT
ACCEPTED MANUSCRIPT