The value of fetal fibronectin in cervical and vaginal secretions and of ultrasonographic...

The value of fetal fibronectin in cervical and vaginal secretions and of ultrasonographic examination of the uterine cervix in predicting premature delivery for patients with preterm labor and intact membranes

Ginseppe Rizzo, MD, Alessandra Capponi, MD, Domenico Arduini, MD, Carmine Lorido, MD, and Carlo Romanini, MD

Rome, Italy

OBJECTIVE: We compared the diagnostic performances of fetal fibronectin assay of cervical and vaginal secretions and of transvaginal ultrasonographic evaluation of the uterine cervix in the prediction of preterm delivery of patients presenting with preterm labor and intact membranes. STUDY DESIGN: One hundred eight patients admitted to the hospital for preterm labor and with intact membranes underwent assay of fibronectin in the cervical and vaginal secretions and transvaginal ultrasonographic evaluation of the uterine cervix. The ultrasonographic parameters evaluated were cervical length, presence of funneling, and cervical index ([Funnel length ÷ 1]/Cervical length). Outcome measures were occurrence of preterm delivery, defined as birth before the thirty-seventh week of gestation, and the admission-to-delivery interval. RESULTS: Forty-seven patients (43.5%) were delivered preterm. Receiver characteristic curve analysis showed that a level of fetal fibronectin in cervical secretions _>60 ng/ml had the highest diagnostic performance in predicting preterm delivery (sensitivity 80.9%, specificity 83.6%). Multiple stepwise logistic regression analysis indicated that the cervical index significantly improved the prediction of preterm delivery achieved by the fetal fibronectin assay. In patients with cervical secretion fibronectin levels >60 ng/ml, survival analysis showed a shorter admission-to-delivery interval in the presence of an abnormal cervical index (/3 < 0.001). CONCLUSIONS; The assay of fetal fibronectin in cervical secretions is more accurate than ultrasonographic evaluation of the uterine cervix in the prediction of preterm delivery. Combined use of the fetal fibronectin assay and the cervical index improves the diagnostic efficiency and allows prediction of the admission-to-delivery interval, identifying a subgroup of patients who may require aggressive treatment. (Am J Obstet Gyneco11996;175:1146-51)

Key words: Preterm labor, uterine cervix, fetal fibronectin, transvaginal ultrasonography

Although different mechanisms may promote preterm delivery, increasing evidence suggests that these processes share a common pathway characterized by changes in ex-

tracellular matrix that lead to cervical modifications and

to the disruption of the chorionic-decidual interface)' 2 Fetal fibronectin, a marker of the disruption of chorionic-

decidual interface, was found to be predictive of preterm delivery. ~-5 The ultrasonographic detection of a short uter-

From the Fetal Medicine Cent~ Department of Obstetrics and Gynecol- ogy, Universita' di Roma "Tot Vergata." Supported by a grant from the Italian National Council of Research (CNR grant No. 95.00895 PF41). A.C. is the recipient of a "Ugo Tropea "fellowship awarded by the Italian Society of Perinatal Medicine. Received for publication January 25, 1996; revised March 25, 1996; accepted May 17, 1996. Reprint requests: Dr. Giuseppe Rizzo, Centro di Medicina Fetale, Clinica Ostetrica e Ginecologica, Universita ' di Roma "Tor Vergata," Policlinico Nuovo S. Eugenio, P. le Umanesimo, 10, 00144 Roma, Italy. Copyright © 1996 by Mosby-Year Book, Inc. 0002-9378/9655.00+ 0 6/1/75021

ine cervix or of dilatation of the internal os also is associated with an increased risk of preterm labor. ~-1°

Despite the link between these findings, no compara-

tive analysis of the two has been performed. This study compares the performances of the assess-

ment of vaginal and cervical fetal fibronectin levels and of

the ultrasonographic measurement of the uterine cervix in the prediction of preterm delivery for patients in preterm labor with intact membranes. It also tests the hy- pothesis that the combined use of both techniques may improve the prediction of preterm delivery.

Material and method

Study patients. The study population consisted of 108 patients selected among those admitted to our institution with the diagnosis of preterm labor and intact membranes between January 1993 and September 1995.

Preterm labor was defined as the presence of uterine

1146

Volume 175, Number 5 RJZZO et al. 1147 Am J Obstet Oynecol

Table I. Characteristics of patients studied

Maternal age (yr, mean + SD) Nulliparous (No.) Multiparous (No.) Previous preterm delivery (No.) Gestational age at admission (wk, mean -+ SD) Admission-to-delivery interval (days, median and range) Gestational age at delivery (days, mean _+ SD) Uterine contractions at admission (No./hr, median and range) Cervical dilatation (cm, median and range) Effacement (%, median and range)

31.4 _+ 2.7 44 (40.7%) 64 (59.3%) 15 (13.8%) 30.1 _+ 2.9

35 (0-111) 35.5 _+ 4.1

1] (8-]8) 1.5 (0-2.5) 60 (0-90)

contractions on external tocodynamometry, occurring at

a frequency of four in 20 minutes or eight in 1 hour

despite lateral recumbency and hydration associated with

changes in cervical effacement and dilatation by serial observation. Criteria for inclusion in the study were (1)

singleton gestation accurately dated by early-second-tri- mester ultrasonographic examination; (2) gestational

age between 24 and 36 weeks; (3) cervical dilatation

<3 cm; (4) intact amniotic membranes (i.e., no obvious leakage of amniotic fluid, no pooling, and negative fern-

ing and Nitrazine paper reaction test results) ; (5) absence

of other maternal or fetal complications at admission and

during pregnancy; (6) cervical and vaginal secretion sam- pling and ultrasonographic assessment of the cervix ob-

tained at admission. All patients gave their informed con-

sent before entering into the study. Their characteristics are reported in Table I.

Fibronectin assay. Two specimens were obtained from each patient, one from the external cervical os and the

second from the posterior fornix of the vagina by means of polyester fiber swabs left in place for 10 seconds. The

swabs were then inserted into polypropylene tubes con- taining 750 p l /ml of assay buffer (Fibronectin collection

kit, Adeza Biomedical)2. Samples were then filtered, and

aliquots were stored at -80 ° C until assay. Fetal fibronec-

tin was assayed with a commercially available solid-phase,

enzyme-linked immunosorbent assay (Adeza Biomedb cal). Assay sensitivity was 50 ng/ml. The intraassay and

interassay coefficients of variation were <10%. Ultrasonographic cervical measurement. Ultrasono-

graphic scans of the uterine cervix were performed with a 6.5 MHz transvaginal probe (Ansaldo Esaote, Idea AUC

4) and an empty bladder, according to the technique described by Andersen et al. 6 The internal cervical os was identified by a sagittal plane view, and the probe was

moved until the entire cervical canal was visualized. Three parameters were measured: cervical length as tile distance between the internal and external os, identified by the sonolucency of the cervical canal; presence or absence of funneling, defined as a dilatation of internal os with a width >5 ram; and the cervical index, calculated as (Funnel length + 1)/Cervical length). 8 All the ultraso-

nographic measurements were performed by two of the authors (G.R. and A.C.). The mean interobserver and

intraobserver coefficients of variation initially assessed in

30 patients were 9.8% and 7.6%, respectively, for the

cervical length and 10.5% and 8.3%, respectively, for the cervical index.

Data analysis. Results of fetal fibronectin assays and of

ultrasonographic measurement of the cervix were not

provided to the clinician in charge and were not used in

patient management. Outcome variables for the study were preterm delivery, defined as birth before the thirty-

seventh week of gestation, and admission-to-delivery interval.

Receiver-operator characteristic curves were generated for fetal cervical .fibronectin levels and the ultrasono-

graphic indexes analyzed for the prediction of preterm labor. The areas under the receiver-operator characteris-

tic curves were analyzed and compared with the proce- dure of Metz et al. n

Logistic regression analysis was used to investigate the relationship between the occurrence of preterm delivery

and different potential explanatory variables (e.g., fetal

fibronectin levels, ultrasonographic parameters, gestational age, parity, history of preterm delivery, number of

contractions per hour at admission, cervical dilatation,

cervical effacement). With stepwise forward selection the variable with a higher significant predictive ability in

the univariate model was entered first into the model, followed by the other significant explanatory variables.

The same analysis was run with each of the significant

explanatory variables entered first in the model, followed

by the others. This analysis was designed to evaluate whether any combination of variables added significantly to the prediction of preterm delivery achieved with univariate analysis.

Survival analysis was applied to determine the relationship between the admission-to-delivery interval and the

association between fetal fibronectin levels and uhrasonographic cervical parameters. The Wilcoxon test was used to compare the differences in the generated survival curves. A value ofp > 0.05 was considered significant.

Results

The prevalence of preterm delivery was 43.5% (47/108). Fig. 1 shows the results of receiver-operator characteristic curve analysis for fibronectin levels and the

1148 Rizzo et al. November 1996 Am J Obstet Gynecol

0,4

/•• @ cervical fibronectin

20

/

0,2

1

0,8

0,6

- - O vaginal fibronectin

, i , - - cervical index

- - £ , cervical length

I I I I

0 0,2 0,4 0,6 0,8 1

1 -Specificity

Fig. 1. Comparison of receiver-operator characteristic curves constructed for fibronectin levels in cervical and vaginal secretions, cervical length, and cervical index in prediction of preterm delivery.

different ultrasonographic parameters. All the curves

were significantly above the 45-degree diagonal line of

unity, indicating a significant relationship between these variables and preterm delivery. The area under the curve

for the cervical fetal fibronectin level was significantly

greater than for those constructed for cervical index (z= 1.9, p< 0.01) and cervical length (z= 2.1, p< 0.01)

but not for vaginal fibronectin (z = 1.0, not significant). The area under the curve for vaginal fibronectin levels was significantly greater than those for cervical index

(z= 1.5, p< 0.05) and cervical length (z= 1.6, p< 0.05). No significant differences were found between the areas constructed for cervical index and cervical length

(z = 0.7, not significant). Fetal fibronectin levels in cervical secretions had the highest diagnostic index in predicting preterm delivery, and a value >60 ng /ml resulted in the best cutoff (Table II and see Fig. 1). Using a cutoff value of 50 ng /ml for fibronectin levels in cervical secre-

tions reduced the specificity of the test (specificity 73.7%, sensitivity 82.9%, positive predictive value 70.9%, negative predictive value 84.9%) (Fig. 1). Logistic regression analysis indicated significant correlations between the occurrence of preterm delivery and cervical fibronectin levels (Z ~ = 38.4, p< 0.001), cervical length (Z 2 = 30.4, p < 0.001), cervical index (Z 2 = 28.7, p_< 0.001), presence

of funneling (Z2= 8.3, p_< 0.01), and gestational age at admission (Z 2 = 6.4, p _< 0.05) but not parity (Z 2 = 2.2, not

significant), history of preterm deliveries (Z2=2.8,

not significant), number of uterine contractions per hour (Z ~ = 1.7, not significant), cervical dilatation (Z 2 = 1.2, not

significant), and cervical effacement at admission (Z2=

0.8, not significant). Stepwise multiple logistic regression analysis indicated

that when the presence of fetal fibronectin levels in cer-

vical secretions was entered first into a model the addi- tion of cervical length (increment ofz 2 from 38.4 to 45.7,

p < 0.01) and cervical index (increment of% ~ from 38.4 to 58.1, p_< 0.01) or the presence of funneling (increment of Z 2 from 38.4 to 43.7, p _< 0.05) resulted in a significant improvement. Entering the cervical length and the presence of funneling after cervical index did not significantly improve the prediction obtained by fetal fibronectin and cervical index (Table III). No further improve- ments in the model were gained by adding other variables or by entering variables with different sequences. The odds ratio, 95% confidence interval, and p value for explanatory variables in the multiple logistic model are shown in Table III.

When survival analysis was applied to values of fetal fibronectin levels in cervical secretions >60 ng/ml , a

Volume 175, Number 5 Rizzo et al. ] 149 AmJ Obstet Gynecol

Table II. Efficacy of fetal fibronectin determination and uhrasonographic cervical measurement for prediction

of preterm delivery

Sensitivity Specificity Positive predictive Negative predictive (%) (%) value (%) value (%)

Cervical fetal fibronectin 260 ng/ml 80.85 Vaginal fetal fibronectin >50 ng/ml 74.47 Cervical index _>0.50 70.21 Cervical length -<20 mm 68.09 Funneling present 70.2

83.61 79.17 85.00 86.89 81.40 81.54 80.33 73.33 77.78 78.69 71.11 76.19 67.2 62.2 74.5

Table III. Relationship between occurrence of preterm delivery and variables analyzed by multiple

logistic regression

Odds ratio 95 % Confidence intervals } Significance [

Cervical fetal fibronectin 21.3 2.8-42.4 p < 0.001 Cervical index 9.9 1.5--19.7 p < 0.005 Cervical length 5.2 0.8-14.9 NS Funneling present 3.4 0.6--9.4 NS Vaginal fetal fibronectin 1.6 0.5-3.8 NS Gestational age at admission 1.3 0.4-1.6 NS

NS, Not significant.

shorter admission-to-delivery interval correlated with cervical length < 22 mm (median 7 days, range 0 to 41 vs

median 21 days, range 6 to 84; Wilcoxon test, p_< 0.01),

cervical funneling (median 7 days, range 0 to 34 vs me-

dian 24 days, range 7 to 84; Wilcoxon test, p _< 0.005), or a

cervical index 2 0.50 (median 6.5 days, range 0 to 24 vs

median 29 days, range 7 to 84; Wilcoxon test, p < 0.001)

(Fig. 2).

Comment

Current methods of identifying patients at risk for pre-

term delivery among those presenting with premonitory

symptoms of preterm labor rely on obstetric history, de- mographic factors, or evaluation of uterine contractions and cervix by digital examination.12' i3 These methods are

neither sensitive nor specific) 4' 1~ As a consequence some

patients may be unnecessarily hospitalized and receive

treatments that may be dangerous for the mother and the fetus, 16 and others may have late institution of tocolytic

therapy, adversely influencing the likelihood of successful treatment. Objective methods that could more accurately and rapidly identify patients at risk for preterm labor are

required, The presence of fetal fibronectin in cervicovaginal se-

cretions and ultrasonographic examination of the uterine cervix have been proposed as specific and sensitive predictors of preterm delivery, promising to be more efficacious than current diagnostic methods. Lockwood et al.,3 studying patients with preterm labor, found that

the presence of fibronectin in cervicovaginal secretions had a sensitivity of 81.7% and a specificity of 82.5% in predicting preterm delivery, and these findings have been confirmed by studies that followed. 4' ~ Similarly, the

predictive power of a short cervical length or of dilatation

of the internal cervical os, as assessed by transvaginal uhrasonography, had a sensitivity ranging between 73%

and 100% and a specificity ranging between 44% and 94% in patients admitted for preterm labord -1°

Our data confirm these findings by showing that fibronectin levels and the uhrasonographic evaluation of

the uterine cervix identify, among patients with preterm labor and intact membranes, those at risk for preterm

delivery with a higher degree of efficiency than current diagnostic parameters such as the obstetric history, digital

evaluation of the cervix, or number of uterine contrac-

tions at admission. The cutoff values for fibronectin and ultrasonographic measurements resulting from receiver-

operator characteristic curve analysis are in the same

ranges as those obtained in previous studies?' 4. ~ Our data

confirm the higher sensitivity of cervical fibronectin concentrations compared with vaginal concentrations¢

This study was designed to compare the diagnostic performance of fibronectin levels and of the ultrasonographic evaluation of the uterine cervix. Receiver-opera-

tor characteristic curve analysis shows a higher performance for cervical fibronectin levels compared with the uhrasonographic parameters describing the cervix. Al-

though some caution is necessary in comparing indexes that may have similar pathophysiologic significance in an exploratory analysis, our data suggest a higher clinical efficiency for fetal fibronectin compared with ultrasonographic evaluation of the uterine cervix. The reasons for the better performance of fetal fibronectin is unclear and cannot be explained by findings in this study. A possible explanation is that the presence of fetal fibronectin in cervical secretions implies disruption of the chorionic-

1150 Rizzo et al. November 1996 Am J Obstet Gynecol

~a

100

i.,!,,, . ..................... <20ram 80.

i • 60- "i,,,

40 . . . . . . . i..,,,,..1..,,.i ........................... ~ _>20

••••••••••••••••••••••••••••••••••••••••i•••••••••••••••••••••••••••••••••••••••••• 0 7 14 21 28 35 42 49 56 63 70 77 84

admission to delivery interval (days)

100 .

80 i.,

60 'i..i.

....................... present

~ I absent

*a 40- "''t'-"L.,i,.~ ~_~ i e ~ 20-- ' ...................... ill. '

O ~ 1 ~ ~ } ~ i ~ ~ ~ ~ 0 7 14 21 28 35 42 49 56 63 70 77 84


100 .

.................... >0.50

i 60

~a 40 2

2O

"[i h <0.50 7 i,,...,!,,_, L

....... i. LL~

••••••••••••T•••••••••••i••••••••••••••••••••••••••••••••••••••••••••••••••••••••••• 0 7 14 21 28 35 42 49 56 63 70 77 84


Fig. 2. Cumulative percentage of undelivered patients in pregnancies with fibronectin levels in cervical secretions >60 ng/ml, divided according to ultrasonographic characteristics of uterine cervix. Top, Cervical length; middle, funneling; bottom, cervical index.

decidual extracellular matr ix interface, a p h e n o m e n o n

strictly l inked to p re te rm delivery and that the modifica-

tion of the ul t rasonographic characteristics of the uter ine

cervix may occur independent ly of p re t e rm labor. An

alternative explanat ion could be h igher reproducibil i ty

of the f ibronect in assay compared with ul t rasonographic

measu remen t of the uter ine cervix. The h igher diagnos-

tic efficiency of cervical f ibronect in is in ag reemen t with

the data of Lockwood et al., 4 who found h igher perfor-

mance of the test compared with the assay of vaginal

secretions.

An impor tan t f inding of this study was that the assay of

f ibronect in in cervical secretions and the ul trasono-

graphic evaluation of the cervix independent ly improve

the diagnostic predict ion o f p r e t e r m delivery. Our results

show that the most efficient and pars imonious mode l to

explain the occur rence of p re t e rm delivery is the associa-

tion between cervical f ibronect in levels and cervical in-

Volume 175, Number 5 Rizzo et al. 1151 AmJ Obstet GynecoI

dex, a parameter reflecting the presence and length of

funne l ing and the cervical length. However, a significant improvement in the diagnostic efficiency of f ibronect in was also achieved when the cervical length or the presence of funne l ing was added in the model. The higher

diagnostic efficiency of the cervical index is no t surpris-

ing, because this measu remen t reflects the presence and

length of funne l ing and the cervical length, and it con- firms the results obta ined by Gomez et al.~a who com-

pared different uhrasonographic measurements of the

cervix in the predict ion of pre term delivery.

For patients with fibronectin-positive cervical secretions, the admission-to-delivery interval is significantly

shorter if they also have abnormal ul trasonographic cer-

vical parameters. The highest predictive per formance was achieved for cervical index values.

These data may have clinical significance. For fibronectin levels _>60 n g / m l in cervical secretions there is a wide

range in the admission-to-delivery interval. The interval

in this study ranged between 0 and 84 days, and this is in

agreement with previously reported series. However, by categorizing data according to cervical index values, it is

possible to identify a subgroup of p regnan t patients for whom the median admission-to-delivery interval is 6.5

days and ano ther group of patients for whom delivery

occurs at a median interval of 29 days. Different manage- m e n t protocols should be p l anned for these groups and

more aggressive tocolytic t rea tment or fetal lung maturi ty

induc t ion should be considered for those with positive f ibronect in values associated with abnormal cervical in-

dexes.

For patients with pre term labor and intact membranes

the assay of fetal f ibronect in in cervical secretions showed

the highest diagnostic per formance in identifying those who will have pre te rm deliveries. However, transvaginal

uhrasonographic evaluation of the cervix improves the efficiency of the f ibronect in assay in predict ing pre term

delivery. For patients with f ibronect in levels in cervical secretions >60 n g / m l , an abnormal cervical index pre-

dicts the admission-to-delivery interval. These findings

suggest that combined use of biochemical and ul trasono-

graphic markers of pre term delivery can help identify the pregnancies at risk and to select proper m a n a g e m e n t

protocol.

REFERENCES

1. Lockwood cj. Recent advantages in elucidating the patho- genesis of preterm delivery, the detection of patient at risk and preventative therapies. Curr Opin Obstet Gyneco11994; 6:7-18.

2. Romero R, Gomez R, Sepulveda W. The uterine cervix, ultrasound and prematurity. Ultrasound Obstet Gynecol 1992;2:385-8.

3. Lockwood CJ, Senyei A, Disehe MR, Casal D, Shah K, Jones L, et al. Fetal fibronectin in cervical and vaginal secretions as a predictor of preterm delivery. N EnglJ Med 1991;325:669- 74.

4. Lockwood CJ, Wein R, Lapinski R, Casal D, Berkowitz G, Alvarez M, et al. The presence of cervical and veaginal fetal fibronectin predicts preterm delivery in an inner-city obstetric population, kanJ Obstet Gynecol 1993;169:798-804.

5. Iams JD, Casal D, McGregorJA, Goodwin TM, Kreaden US, Lowensobn R, et al. Fetal fibronectin improves the accuracy of diagnosis of preterm labor. AmJ Obstet Gynecol 1995; 173:141-5.

6. Andersen HF, Nugent CE, Wanty SD, Hayashi Rid. Predic- tion of risk for preterm delivery by ultrasonographic measurement of the cervical length. Am J Obstet Gynecol 1990; 163:859-67.

7. Okitsu O, Mimura T, Nakayama T, Aono T. Early prediction of preterm delivery by transvaginal ultrasonography. Ultra- sound Obstet Gynecol 1992;2:402-9.

8. Gomez R, Galasso M, Romero R, Mazor M, Sorokin Y, Bau- mann R, et al. Uhrasonographic examination of the cervix is better than cervical digital examination as a predictor of the likelihood of premature delivery in patients with preterm labor and intact membranes. AmJ Obstet Gyneco11994;171: 956-64.

9. Iams JD, Paraskos J, Landon MB, Teteris JN, Johnson FE Cervical sonography in preterm labor. Obstet Gyneco11994; 84:40-6.

10. Timor-Tritsch I, Boozarjomehri F, Masakowski Y, Mon- teagudo A, Chao CR. Can a "snapshot" sagittal view of the cervix by transvaginal uhrasonography predict active preterm labor? AmJ Obstet Gynecol 1996;174:990-5.

11. Metz CE. Basic principles of ROC analysis. Semin Nucl Med 1979;8:283-98.

12. Creasy RK, Gummer BA, Liggings GC. System for predicting spontaneous preterm birth. Obstet Gynecol 1980;55:692-5.

13. Gonik B, Creasy RK. Preterm labor: its diagnosis and management. AmJ Obstet Gynecol 1986;154:3-8.

14. Creasy RK. Preterm birth prevention: where we are? Am J Obstet Gynecol 1993;168:1223-30.

15; Main DM, Gabbe SG, Richardson D, Strong S. Can preterm deliveries be prevented? Am J Obstet Gynecol 1985;151: 892-8.

16. Benedetti T. Maternal complications ofparenteral beta-sym- pathomimetic therapy for premature labor. Am J Obstet Gynecol 1983;145:1-6.

The value of fetal fibronectin in cervical and vaginal secretions and of ultrasonographic...

Documents

Transcript of The value of fetal fibronectin in cervical and vaginal secretions and of ultrasonographic...