Cervicovaginal fibronectin and cervical length at 23 weeks of gestation: relative risk of early preterm delivery

Authors


Correspondence: Professor K. H. Nicolaides, Harris Birthright Research Centre for Fetal Medicine, King's College Hospital, Denmark Hill, London SE5 8RX, UK.

Abstract

Objectives To establish the prevalence of cervicovaginal fetal fibronectin positivity at 23 weeks of gestation in a routine population of singleton pregnancies and determine the relative risk of spontaneous delivery before 33 weeks in women with a fibronectin positive result.

Design Prospective clinical study.

Setting Inner city antenatal clinic.

Population Singleton pregnancies attending for routine antenatal care.

Methods Cervicovaginal fetal fibronectin and cervical length were measured at 23 weeks of gestation. The distribution of fibronectin positivity within subgroups according to maternal characteristics was calculated and the relative risk of spontaneous delivery before 33 weeks was estimated.

Main outcome measures Prevalence of a fibronectin positive result and its relation to cervical length measurement and spontaneous preterm delivery before 33 weeks.

Results Of 5146 women participating in the study, 182 (3.5%) had a fibronectin positive result and 76 (1.5%) had a cervical length of ≤ 15 mm. Fibronectin positive women were more likely to be Afro-Caribbean in origin, to have had a previous second trimester miscarriage and to have a short cervix. In the 5068 women who were managed expectantly, the significantly independent relative risk of spontaneous delivery at < 33 weeks was 46.2 (95% CI 18.8–113.6), for cervical length of ≤ 15 mm, 8.1 (95% CI 3.8–17.5) for a fibronectin positive result, and 4.4 (95% CI 2.2–9.1) for cigarette smoking.

Conclusion Fibronectin positivity at 23 weeks of gestation provides useful prediction of pregnancies at risk of spontaneous preterm delivery before 33 weeks, with a relative risk that is twice as high as cigarette smoking, but is a sixth of that of cervical length.

INTRODUCTION

Preterm delivery is the leading cause of neonatal death1, which is about 80% for pregnancies delivering at 24 weeks, decreasing to < 2% for those delivering after 33 weeks. The risk of preterm delivery varies with maternal characteristics, such as ethnic origin, age, ponderal index, cigarette smoking and drug abuse, as well as previous obstetric history2. This has led to the development of risk scoring systems, but such systems have a high false positive rate and a low sensitivity. Recent interest in the prediction of severe preterm delivery has focused on the presence of fetal fibronectin in cervicovaginal secretions and the ultrasonographic finding of a short cervix. Thus, a screening study in which cervicovaginal fibronectin was examined in 2929 singleton pregnancies at 24 weeks of gestation reported that 4% of the population were fibronectin positive and this group contained 21% of those delivering before 35 weeks3. In terms of screening by ultrasound measurement of cervical length, one study that examined 2915 singleton pregnancies at 24 weeks, reported that the cervical length was < 20 mm in 3% of the population and this group contained 23% of those delivering spontaneously before 35 weeks4. Another study of 2702 singleton pregnancies, examined at 23–24 weeks, reported that the prevalence of cervical length of ≤ 15 mm was 1.6%, and it was estimated that the sensitivity of this cervical length cut off for spontaneous delivery before 33 weeks was 58%5.

This study examines the prevalence of a fetal fibronectin positive result at 23 weeks of gestation in a routine population of 5146 singleton pregnancies in relation to maternal characteristics and cervical length and determines the relative risks of spontaneous delivery at 1276 ≤ 33 weeks for maternal characteristics, fibronectin positivity and cervical length.

METHODS

At King's College Hospital, London, women attending for routine antenatal care are offered the option of having two ultrasound examinations; the first is at 10.14 weeks of gestation and the second at 23 weeks. During a 26-month period (January 1997 to March 1999) women attending for the 23-week scan were offered the option of having preterm delivery screening. Written informed consent was obtained from those agreeing to participate in the study, which was approved by the hospital's ethics committee.

The women were asked to empty their bladder and were placed in the dorsal lithotomy position. A speculum examination was first carried out and a dry swab was taken from the posterior vaginal fornix with a Dacron swab. These samples were frozen at −60° C and the concentration of fetal fibronectin antigen was measured with a sensitive immunoassay that used the fetal fibronectin-specific monoclonal antibody FDC-6 (Azeda Biomedical Corporation, California, USA). A positive test for fibronectin was defined as any value of ≥ 50 ng/mL. A second swab was also taken from the posterior vaginal fornix for assessment of bacterial vaginosis as described by Nugent et al.6. Cervical length was measured by transvaginal sonography as previously described7,8.

Patients with a cervical length > 15 mm had normal antenatal care. Those with a length of ≤15 mm were either managed expectantly or they had a Shirodkar cerclage; in the first year of the study the decision was primarily based on the preference of the managing obstetrician, but subsequently the patients were recruited into an ongoing randomised study comparing the two management options. The managing clinicians were unaware of the results of testing for fibronectin and bacterial vaginosis.

Patient characteristics, including demographic data and previous obstetric and medical history, were obtained from the patients at their first antenatal visit to the hospital by midwives and were entered into a computer database. Similarly, the ultrasound findings and the results of the swabs taken were recorded in the database. Gestational age was determined from the menstrual history and confirmed from the measurement of fetal crown-rump length at the first trimester scan. A computer search was made to identify all singleton pregnancies that had 23-week preterm delivery screening and a known pregnancy outcome. Women with a multiple gestation, or major fetal anomaly were excluded.

Statistical analysis

To determine the significance in differences between subgroups (according to demographic characteristics, previous obstetric history, cervical length and the presence or absence of bacterial vaginosis) in the percentage of fibronectin positivity, χ2 test or Fisher's exact test were used. Multiple logistic regression analysis was used to determine those variables that provided a significant independent contribution in explaining a positive fibronectin result.

In the calculation of relative risks for spontaneous preterm delivery before 33 weeks, patients with a short cervix who were treated by cervical cerclage and those who had iatrogenic delivery before 33 weeks were excluded. The χ2 test or Fisher's exact test were used to calculate the significance of differences in the percentage of spontaneous delivery before 33 weeks between subgroups according to demographic characteristics, previous obstetric history, cervical length measurement and fibronectin or bacterial vaginosis positivity. Multiple logistic regression analysis was used to determine those variables (from the demographic characteristics, past obstetric history, gram stain result, fibronectin result and cervical measurement) that provided a significant independent contribution in explaining the rate of spontaneous preterm delivery before 33 weeks.

In the multiple logistic regression analysis, maternal age was given a score of one if < 20 years and zero if ≥ 20 years; ponderal index was scored one if ≤ 19.8 and zero if > 19.8; cigarette smoking and cervical surgery were scored as one (for yes) or zero (for no); ethnic group was scored as zero for non Afro-Caribbean and one for those of Afro-Caribbean origin; obstetric history was scored as zero for primigravidae and those with previous termination of pregnancy or a miscarriage at < 16 weeks, one for those with previous deliveries at term, two for those with a previous spontaneous delivery at 33–36 weeks, three for those with a previous spontaneous delivery at 24–32 weeks, and four for those with a previous spontaneous miscarriage at 16–23 weeks; the gram stain result and fibronectin measurement were described as one if positive and zero if negative; the cervical length measurement was described as one if it was ≤ 15 mm and zero for > 15 mm.

RESULTS

The entry criteria for the study were fulfilled by 5146 women. Determination of fibronectin status and measurement of cervical length, at 22–24 weeks (median 23), were successfully achieved in all cases. The fibronectin test was positive in 182 women (3.5%), the cervical length (which was approximately normally distributed with a median value of 36 mm) was ≤ 15 mm in 76 women (1.5%) and bacterial vaginosis was present in 847 women (16.5%). The prevalence of fibronectin positivity was significantly increased in women of Afro-Caribbean origin, those who had a previous second trimester miscarriage and those with cervical length of ≤ 15 mm (Fig. 1), but there was no significant association with age, ponderal index, cigarette smoking, cervical surgery or bacterial vaginosis (Table 1). Multiple regression analysis demonstrated significant independent contributions in explaining fibronectin positivity by ethnic origin (β= 1.82, 95% CI 1.33–2.49, P < 0.0005) and cervical length of ≤ 15 mm (β= 9.41, 95% CI 5.49–16.13, P < 0.0001).

Figure 1.

Prevalence of a positive fetal fibronectin result in relation to cervical length.

Table 1.  Incidence of a positive fibronectin result at 23 weeks of gestation. The relative risk (RR) was calculated as a proportion of the percentage of women with a positive fibronectin result within each subgroup, compared with that of the rest of the population. Values are given as n or n (%), unless otherwise indicated. BV = bacterial vaginosis.
Patient characteristicsTotal populationPositive fibronectin resultRRComparison of subgroupsdfP
All51461823.5   
Ethnic group      
 Caucasian2575 (50.0)52 (2.0)0.4 2<0.0001
 Afro–Caribbean2145 (41.7)109 (5.1)2.1   
 Other426 (8.3)21 (4.9)1.4 1 
Age (years)      
 < 20329 (6.4)10 (3.0)0.9 2NS
 20–353750 (72.9)135 (3.6)1.1   
> 351067 (20.7)37 (3.5)1.0   
Ponderal index      
 < 19.8423 (8.2)13 (3.1)0.9 2NS
 19.8–262661 (51.7)84 (3.2)0.8   
> 262062 (40.1)85 (4.1)1.3   
Cigarette      
 Smoker758 (14.7)27 (3.6)1.0 1NS
 Nonsmoker4388 (85.3)155 (3.5)1.0   
Caesarean section      
 Yes405 (7.9)17 (4.2)1.2 1NS
 No4741 (92.1)165 (3.5)0.8   
Obstetric history      
 Primigravidae1614 (31.4)48 (3.0)0.8 6NS
 Fetal loss at < 16 wks973 (18.9)35 (3.6)1.01 vs 21NS
 Delivery at ≥ 37 wks2261 (43.9)82 (3.6)1.01 vs 31NS
 Delivery at 33–36 wks130 (2.5)7 (5.4)1.51 vs 41NS
 Delivery at 24–32 wks78 (1.5)4 (5.1)1.51 vs 51NS
 Termination at 16–23 wks55 (1.1)2 (3.6)1.01 vs 61NS
 Miscarriage at 16–23 wks35 (0.7)4 (11.4)3.31 vs 71<0.05
Length of cervix      
 ≤ 15 mm76 (1.5)21 (27.6)8.7 1<0.0001
 ≥ 15 mm5070 (98.5)161 (3.2)0.1   
BV      
 Positive847 (16.5)30 (3.5)1.0 1NS
 Negative4299 (83.5)152 (3.5)1.0   

The cervical length was ≤ 15 mm in 76 women; 39 were managed expectantly and 37 had a Shirodkar suture. These 37 women with a suture and another 41 who had an iatrogenic delivery before 33 weeks were excluded from further analysis. In the remaining 5068 women there were 43 women (0.8%) who had a spontaneous preterm delivery before 33 weeks. The relative risk of spontaneous preterm delivery was significantly higher in women of Afro–Caribbean origin, cigarette smokers, those who had a previous preterm delivery before 33 weeks, those with cervical length of ≤ 15 mm and those with a fibronectin positive result, but there was no significant association with age, ponderal index, cervical surgery or bacterial vaginosis (Table 2). The predictive properties of the tests that were significantly associated with preterm delivery are shown in Table 3 and those of the combination of fibronectin and cervical length tests are shown in Table 4. However, the true sensitivity of cervical length of ≤ 15 mm in predicting preterm delivery is likely to be higher than 28%, because half of the women with a very short cervix were treated with placement of a cervical suture. Multiple regression analysis demonstrated significant independent contributions in explaining spontaneous preterm delivery before 33 weeks by cigarette smoking (β= 4.44, 95% CI 2.17–9.12, P < 0.0001), fibronectin positivity (β= 8.13, 95% CI 3.78–17.50, P < 0.0001), and cervical length of ≤ 15 mm (β= 46.15, 95% CI 18.76–113.56, P < 0.0001).

Table 2.  Relation between rate of spontaneous delivery at less than 33 weeks of gestation and patient characteristics. The relative risk (RR) was calculated as a proportion of the percentage of women delivering at less than 33 weeks within each subgroup, compared with that of the rest of the population. Values are given as n or n (%), unless otherwise indicated. BV = bacterial vaginosis; FFN = fetal fibronectin.
Patient characteristicsTotal populationDelivery < 33 weeksRRComparison of subgroupsdfP
All5068430.8   
Ethnic group      
  Caucasian255616 (0.6)0.6 2< 0.05
  Afro-Caribbean209026 (1.2)2.21 vs 21< 0.05
  Other4221 (0.2)0.3   
Age (years)      
  < 203286 (1.8)2.3 2NS
  20–35369627 (0.7)0.6   
  > 35104410 (1.0)1.2   
Ponderal index      
  < 19.84183 (0.7)0.8 2NS
  19.8–26263718 (0.7)0.7   
> 26201322 (1.1)1.6   
Cigarette      
  Smoker74914 (1.9)2.8 1< 0.005
  Nonsmoker431929 (0.7)0.4   
Caesarean section      
  Yes3975 (1.3)1.5 1NS
No467138 (0.8)0.6   
Obstetric history      
  Primigravidae159413 (0.8)0.9 6< 0.005
  Fetal loss at < 16 wks9508 (0.8)1.01 vs 21NS
  Delivery at ≥ 37 wks223816 (0.7)0.71 vs 31NS
  Delivery at 33–36 wks1281 (0.8)0.91 vs 41NS
  Delivery at 24–32 wks734 (5.5)7.01 vs 51< 0.005
  Termination at 16–23 wks520 1 vs 61NS
  Miscarriage at 16–23 wks331 (3.0)3.61 vs 71NS
Length of cervix      
  ≤ 15 mm3912 (30.8)49.9 1< 0.0001
  > 15 mm502931 (0.6)0.02   
FFN      
  Positive17214 (8.1)13.7 1< 0.0001
  Negative489629 (0.6)0.1   
BV      
Positive8247 (0.8)1.0 1NS
Negative424436 (0.8)1.0   
Table 3.  The predictive properties of fibronectin positivity, a cervical length of less than or equal to 15 mm, cigarette smoking, a previous delivery at 24 to 32+6 weeks of gestation and being of Afro-Caribbean origin in the calculation of risk of delivery before 33 weeks. Values are given as %.
   Predictive value
 SensitivitySpecificityPositiveNegative
Fibronectin positive32.696.98.199.4
Cervical length <15 mm27.999.530.899.4
Smoking32.685.41.999.3
Previous delivery at 24–32+6 wks9.398.65.599.2
Afro-Caribbean60.558.91.299.4
Table 4.  Spontaneous preterm delivery before 33 weeks, according to cervical length and fibronectin results. Values are given as n/ntotal, (%).
Test resultsDelivery < 33 weeks.
Fibronectin positive 
 Cervical length 
≤ 15 mm9/12 (75)
> 15 mm5/160 (3)
Fibronectin negative 
 Cervical length 
≤ 15 mm3/27 (11)
> 15 mm26/4869 (0.5)

The relative risk of spontaneous preterm delivery has also been analysed in relation to cervical length of ≤ 25 mm, to allow comparison with results from previous studies which have used this cut off to define the group at high risk of preterm delivery. In our study, the cervical length was ≤ 25 mm in 428 (8.4%) of the women who were managed expectantly. Multiple regression analysis using this cut off of cervical length to define those women with a ‘short cervix’ demonstrated significant independent contributions in explaining spontaneous preterm delivery before 33 weeks by cigarette smoking (β= 4.07, 95% CI 2.00–8.31, P < 0.001), fibronectin positivity (β= 12.16, 95% CI 5.97–24.76, P < 0.0001) and cervical length of ≥ 25 mm or less (β= 12.52, 95% CI 6.60–23.74, P < 0.0001).

DISCUSSION

The findings of this study suggest that fibronectin positivity at 22–24 weeks of gestation is strongly related to cervical length and is more common in women of Afro-Caribbean origin. Furthermore, a positive fibronectin result at this gestation is associated with an eight-fold increase in risk for spontaneous preterm delivery before 33 weeks. However, the main risk factor for early preterm delivery is a cervical length of ≤ 15 mm, which is associated with a 46-fold increase in risk. In this study half of the women with a very short cervix were treated with placement of cervical suture and therefore the true spontaneous preterm delivery rate and sensitivity of the test can not be calculated.

When cervical length of ≤ 25 mm is used, the contributions of cervical length and fibronectin in the prediction of preterm delivery are similar. These results are compatible with those of a multicentre collaborative study9 in the United States on 2929 pregnancies at 24 weeks, which reported that the odds ratio for delivery before 32 weeks was 9.8 for fibronectin positivity and 8.7 for cervical length of ≤ 25 mm in nulliparous women and 10.0 and 4.6, respectively, for multiparous women.

In the multiple regression analysis for the prediction of preterm delivery, in addition to a short cervical length and fibronectin positivity, there was a significant independent contribution from cigarette smoking. It has been suggested that smoking may induce labour by increasing the amniotic fluid concentration of the inflammatory mediator platelet activating factor; cigarette smoke is a potent inhibitor of the enzyme that degrades platelet activating factor10.

There was a strong association between cervical length and fibronectin positivity, which increased exponentially with decreasing cervical length, while the overall prevalence of a positive fibronectin result of 3.5% rose to nearly 60% in those with a cervix of ≤ 5 mm. A possible link between a short cervix and fibronectin positivity is ascending infection from the lower genital tract. Cervical shortening and the accompanying loss of cervical mucous, which through its physical properties and intrinsic antibacterial activity normally acts as a barrier to ascending infection, may promote bacterial colonisation of the fetal membranes. Intrauterine infection, either directly or through the host response with the production of cytokines, could cause release of fibronectin and also stimulate synthesis of prostaglandins that induce uterine contractions1,12. In this respect, fetal fibronectin, a product of the fetal membranes thought to play a role in binding the placenta and its membranes to the decidua13 is an early marker of ascending infection.

There is extensive evidence linking infection with the risk of spontaneous preterm delivery12. However, in our study there was no significant association between positive results for fibronectin and bacterial vaginosis, nor between bacterial vaginosis and the risk of preterm delivery. Furthermore, randomised controlled trials with antibiotics in women with bacterial vaginosis have not demonstrated a significant reduction in the rate of preterm delivery14. In addition, studies of women with preterm prelabour rupture of membranes have shown the presence of intrauterine infection in only one third of the cases15. Consequently, if there is a causative association between bacterial vaginosis, fibronectin positivity and preterm labour, this cannot be the explanation for the majority of cases. This issue will hopefully be resolved by the results of multicentre randomised studies which are investigating the potential value of antibiotics in the prevention of preterm delivery, both in women presenting with preterm labour16 and in asymptomatic women with a positive fibronectin result17.

An alternative mechanism for the inter-relation between a positive fibronectin result, cervical shortening and preterm delivery is a noninfective, inflammatory process. Although, the initiating mechanism for such a process is uncertain, and indeed may be multifactorial, the final pathway involving the release of cytokines and prostaglandins and the activation of proteolytic enzymes could cause release of fibronectin from the fetal membranes, shortening of the cervix and uterine contractions. Anti-inflammatory drugs, such as indomethacin, have been shown to be effective tocolytic agents18. The extent to which such drugs can effectively prevent the onset of preterm labour in women with a short cervix and a fibronectin positive result needs to be determined. Our cross sectional study could not define the temporal relation between fibronectin positivity, cervical shortening and preterm delivery. However, preliminary data suggest that mechanical reconstitution of the cervix by the placement of a Shirodkar suture can interrupt this process19. The effectiveness of this method in the prevention of preterm delivery is the subject of an ongoing multi-centre randomised trial.

Acknowledgements

This study was supported by a grant from the Fetal Medicine Foundation (Charity No: 1037116).

Ancillary