Background
First-trimester cervical length for the prediction of spontaneous preterm delivery remains controversial. A better method for the measurement of the first-trimester cervical length and additional cervical ultrasound parameters for the identification of women at high risk for spontaneous preterm delivery are needed.
Objective
This study aimed to compare the predictive value of cervical length measured by 2 different methods in the first trimester of pregnancy to predict spontaneous preterm delivery and to explore the potential value of first-trimester cervical shear-wave elastography for the prediction of spontaneous preterm delivery.
Study Design
This was a prospective study in unselected singleton pregnancies at 11 +0 to 13 +6 weeks’ gestation. Cervical length was measured by the following 2 methods in the base-cohort population: (1) a linear distance between the 2 ends of the glandular area around the endocervical canal (single-line method: cervical length-s) and (2) a sum of the linear distance from the internal os to the greatest cervical curvature and the linear distance from this point to the external os (2-line method: cervical length-t). In a substudy, cervical shear-wave elastography scores for 9 regions of interest (inner, middle, and external parts of anterior lip, endocervical canal, and posterior lip) in midsagittal plane were also obtained by transvaginal ultrasonography. The screening performance of the first-trimester cervical length measured by the 2 different methods for the prediction of spontaneous preterm delivery was assessed by receiver operating characteristics curve analysis. The areas under the curves were compared using a DeLong test. The predictive performance of a soft cervix (mean elastography scores with multiple of median <5th, 10th, 15th, 20th, and 25th percentile) for spontaneous preterm delivery was also determined.
Results
Among a total of 2316 included pregnancies, spontaneous delivery at <37 and <34 weeks’ gestation occurred in 111 cases (4.8%) and 20 cases (0.9%), respectively. In the total study population, when compared with the term delivery group, the median cervical length-t was shorter in women with spontaneous delivery at <34 weeks’ gestation (36.9 mm vs 35.1 mm; P =.015), but there was no clear correlation for cervical length-s. Receiver operating characteristics curves demonstrated that cervical length-t achieved better performance in predicting spontaneous delivery at <34 weeks’ gestation (area under the curve, 0.658 vs 0.573; P <.01) than cervical length-s. The best combined model to predict spontaneous delivery at <34 weeks’ gestation was provided by cervical length-t and history of preterm delivery (area under the curve, 0.692). In the substudy, a soft cervix with a mean elastography scores multiple of median <10th percentile had a relative risk of 7.8 (95% confidence interval, 2.1–28.6) for spontaneous delivery at <34 weeks’ gestation; the detection rate was 44.4% at a false-positive rate of 9.0%.
Conclusion
The 2-line approach provides a better estimate of the actual first-trimester cervical length and achieves better performance as a screening tool for spontaneous preterm delivery at <34 weeks’ gestation than the conventional measurement. A soft cervix as determined by shear-wave elastograpthy in the first trimester is associated with an increased risk for subsequent spontaneous preterm delivery.
Introduction
Preterm delivery is the leading cause of perinatal morbidity and mortality. Every year, an estimated 15 million babies are born prematurely. Spontaneous preterm delivery, including those with premature onset of labor or preterm prelabor rupture of membranes (PPROM), accounts for approximately two-thirds of all premature births. Thus far, cervical length in the second trimester has been demonstrated to be associated with an increase risk for spontaneous preterm delivery. Evidence suggests that high-risk women with a short cervix would benefit from vaginal progesterone prophylaxis, reducing the risk for spontaneous preterm delivery by 40%. However, the administration of vaginal progesterone is not as effective in women with a very short cervical length of <10 mm as in those with a length of 10 to 25 mm. Hence, it may be preferable to measure cervical length in early pregnancy before the critical length of 10 mm is reached.
Why was this study conducted?
This study aimed to explore an alternative method for the measurement of first-trimester cervical length and to determine the potential value of first-trimester cervical shear-wave elastography for the screening of spontaneous preterm delivery.
Key findings
The 2-line approach provided a better estimate of the actual first-trimester cervical length and achieved better performance for the screening of early spontaneous preterm delivery than the conventional approach. A soft cervix as determined by shear-wave elastography in the first trimester was associated with an increased risk for subsequent spontaneous preterm delivery.
What does this add to what is known?
The cervical length, measured by the 2-line approach, and cervical shear-wave elastography measurements obtained in the first trimester have the potential to help improve early prediction of spontaneous preterm delivery.
Cervical length assessment in the first trimester as a screening tool for spontaneous preterm delivery remains controversial, which may be explained by the fact that the first-trimester cervix is usually curved, thus leading to an underestimation of the cervical length when measured by the conventional method with a straight line from the internal os to external os. There is therefore a need to develop an alternative method for the measurement of the first-trimester cervical length and to explore its performance as predictive of spontaneous preterm delivery.
Apart from the cervical length, cervical softness has been proven to be associated with an increased risk for spontaneous preterm delivery. Together with cervical shortening, cervical softening begins in the first trimester and accelerates before onset of labor. In recent years, ultrasonographic elastography, including strain elastography and shear-wave elastography (SWE), has been introduced as a noninvasive technique to assess the cervical stiffness in the form of elasticity. , Strain elastography measures the percentage of tissue deformation before and after a mechanical force is applied manually, and SWE automatically generates an acoustic force that creates a mechanical impulse, leading to tissue displacement in the form of shear-waves. The SWE technique can overcome the limitations of strain elastography because the generation of the mechanical impulse is independent of the operator. Given that previous studies have demonstrated that transvaginal cervical SWE measurements at midgestation are potentially predictive of spontaneous preterm delivery, , it is reasonable to explore the role of first-trimester cervical SWE measurements as predictors of spontaneous preterm delivery, in addition to measurement of the cervical length, to improve the performance of early prediction models.
The aims of this study were to (1) compare the performance of the single-line and 2-line methods of cervical length measurement in the first trimester of pregnancy in predicting spontaneous preterm delivery and (2) explore the potential value of the first-trimester cervical SWE measurements in predicting spontaneous preterm delivery.
Material and methods
Study design and participants
We are reporting on the expanded data from a prospective study of unselected asymptomatic women with a viable singleton pregnancy who were attending a screening appointment for Down syndrome at 11 +0 to 13 +6 weeks’ gestation, from June 2018 to July 2020, at the Prince of Wales Hospital, Hong Kong SAR. , We excluded women aged <18 years or with fetal defect(s) detected during first-trimester ultrasound scan. All eligible women provided written informed consent before participation.
In this study, maternal history (including age, weight, height, body mass index, racial origin, smoking habit, method of conception, parity, obstetrical history, previous preterm delivery, and cervical surgery history) and cervical length were recorded in all cases in the base-cohort population (n=2433). In a substudy, cervical SWE assessment was performed between January 2019 and January 2020 (n=1172).
Ultrasound assessment
The transvaginal ultrasound measurements were conducted by 4 operators (Q.F., P.C., H.D., and X.J.) accredited for the measurement of cervical length by the Fetal Medicine Foundation ( www.fetalmedicine.org ). All operators underwent training with a radiologist (Stella S.Y. Ho) with expertise in cervical elastography. Before commencement of the study, all operators performed 30 SWE assessments of the cervix as part of the training course.
A 2-dimensional transvaginal scan was performed to acquire the image of the sagittal plane of the cervix according to the Fetal Medicine Foundation protocol ( www.fetalmedicine.org ). In brief, all women had an empty bladder and were in a modified lithotomy position. The transducer was introduced into the anterior fornix. After having identified the entire length of the endocervical canal in the sagittal plane, the position and pressure of the transducer were adjusted, and the ultrasound image was magnified by ensuring that the cervix occupied 50% to 75% of the image. Care was taken to visualize the endocervical canal clearly without excessive pressure being applied to the cervix to avoid artificial lengthening. Cervical length was then measured by 2 methods. The first method was the conventional approach recommended by the Fetal Medicine Foundation in which the linear distance between the internal and external cervical os (2 ends of the glandular area around the endocervical canal) was measured (single-line method: cervical length-s) ( Figure 1 , A). The second method was the sum of the following 2 linear measurements: first, from the internal os to the greatest cervical curvature and second, from the end of the first line to the external os (2-line method: cervical length-t) ( Figure 1 , B).
During the same transvaginal scan, the SWE scores of the cervix were measured using a SuperSonic Aixplorer – Innovative UltraFast Ultrasound Imaging (Aix-en-Provence, France), which was equipped with the SWE tool. SWE measurements were performed using a color-coded map in the region of interest (ROI), which represents Young’s modulus (E) in meters per second at each pixel. Under certain assumptions, the shear-wave speed (c) is related to Young’s modulus (E), which is a measure of stiffness, as follows: E = 3ρc 2 , where ρ is tissue density. Therefore, by estimating the shear-wave speed, the underlying tissue stiffness can be quantified. A low speed corresponds to a soft medium, whereas a high speed indicates a stiff medium. The shear-wave speed can be used directly as a proxy for stiffness or it can be converted to Young’s modulus. In the same midsagittal plane of the cervix that was used for the measurement of the cervical length, the transducer was maintained at the anterior fornix without excessive pressure. The SWE tool was switched on and the elastography map was placed over the entire length of the cervix. Once the elastography map was stable for at least 3 seconds, 3 cervical elastographic images were obtained. The measurement of the SWE scores, which were automatically calculated using Young’s modulus in kilopascal (kPa), was performed post hoc, and the mean values of the 3 elastographic images were used for analysis. The ROI was set at 5 mm and placed on the inner (A i ), middle (A m ), and external (A e ) parts of the anterior lip at an equal distance between the cervical serosa and cervical canal. A similar process was repeated along the endocervical canal (C i , C m , C e ) and the posterior cervical lip (P i , P m , P e ) ( Figure 2 ). The machine setting that was used for shear-wave frequency was 50 to 250 Hz with a 1 Hz frame per second. The acquisition frame rate was 1000 to 3000 Hz. A standard deviation (SD) of >30% for the SWE score was used as a cutoff to define which cervical SWE scores were considered unreliable. , Unreliable measurements were excluded for statistical analysis.
Outcome measures
The primary outcome measure was spontaneous preterm delivery at <37 weeks’ gestation (<259 days). The secondary outcome measure was spontaneous preterm delivery at <34 weeks’ gestation (<238 days). The obstetrical records for all cases delivering at <37 weeks’ gestation were reviewed to determine whether the preterm delivery was medically indicated or spontaneous. The latter included those with spontaneous onset of labor and those with PPROM.
Statistical analysis
The sample size calculation was based on an expected difference in the mean cervical length of 3 mm between the term and preterm delivery groups and a common SD of 4.7 mm. , At a power of 80% and a hypothetical dropout rate of 15%, a total of 2111 pregnant women were needed for inclusion. Because there was no previous first-trimester study evaluating the role of the cervical SWE measurements in predicting spontaneous preterm delivery, the sample size calculation for this part of the study was based on a previous study that evaluated cervical shear-wave speed (in m/s), which correlates with Young’s modulus, measured at 24 to 35 weeks’ gestation, with an expected difference in the cervical shear-wave speed of 0.2 m/s between preterm and term delivery groups and an SD in the term delivery group of 0.15 m/s. At a power of 95% and a 95% confidence interval (CI), the required sample size to detect a marked difference was 682 when using a t test. If a hypothetical dropout rate of 15% was assumed, a total of 803 pregnant women were needed for inclusion.
Normality of the data was tested using the Kolmogorov-Smirnov test. Descriptive data were presented as mean (SD) or median (interquartile range [IQR]) for continuous variables and as numbers and percentages for categorical variables. Comparisons between the outcome groups were done using chi-square tests, continuity correction tests, and Fisher exact tests for categorical variables and using Student t tests or Mann-Whitney U tests for continuous variables.
To represent the overall elasticity of the cervix, the mean cervical SWE score of the 9 ROIs was used for analysis. Distributions of the cervical length-s, cervical length-t, and mean cervical SWE score were logarithmically transformed if these were not normally distributed. Histograms and normal probability plots were used to assess normality.
A multivariate backward linear regression analysis was used to determine which maternal factors were marked predictors of the log 10 values of cervical length by the 2 methods and of the mean cervical SWE score in the term delivery group. The distributions of log 10 cervical length and log 10 mean cervical SWE score, expressed as the multiple of the median (MoM) of the term delivery group, were determined for all cases. Correlation between the mean cervical SWE MoM and cervical length MoM was assessed using a Spearman correlation. Comparisons of log 10 cervical length MoM and log 10 mean cervical SWE MoM between the outcome groups were done using Mann–Whitney U tests or Student t tests.
In the base-cohort population, backward stepwise multivariate logistic regression analysis was used to determine the best predictive models for spontaneous preterm delivery. The performance of first-trimester cervical length and the performance of the cervical length when combined with maternal characteristics for the prediction of spontaneous preterm delivery was determined using receiver operating characteristic (ROC) curve analysis. Differences between the areas under the ROC curves (AUROCs) were assessed using the DeLong test.
In the substudy, the relative risks (RR) and detection rates of a soft cervix with different cut offs (mean SWE MoM <5th, 10th, 15th, 20th, and 25th percentile) for spontaneous delivery at <37 and <34 weeks’ gestation were derived by 2×2 contingency tables. A 2-tailed P value <.05 was considered statistically significant.
Data analyses were performed using the statistical software package SPSS 23.0 (SPSS Inc, Chicago, IL) and Medcalc (Medcalc Software, Mariakerke, Belgium).
Results
A first-trimester cervical length assessment was performed for a total of 2433 women with a singleton pregnancy. We excluded 117 cases (4.8%) because they had missing outcome data or because the pregnancies ended in miscarriage before 24 weeks’ gestation, in stillbirth, or in termination. In the remaining 2316 pregnancies, delivery at <37 weeks’ gestation occurred in 159 (6.9%) cases with more than two-thirds of those being spontaneous (111/159; 4.8%). Among the 111 spontaneous preterm delivery cases, 20 (0.9%) women delivered <34 weeks’ gestation. A total of 1114 of 2316 women had their cervical SWE score measured, and of these, 54 (4.8%) and 9 cases (0.8%) delivered spontaneously at <37 and <34 weeks’ gestation, respectively.
Maternal demographic characteristics for each outcome group are shown and compared in Table 1 . There were no marked differences in the maternal characteristics and the gestational age at the time of the scan between the term delivery and spontaneous preterm delivery groups, except that among women with spontaneous preterm delivery in the index pregnancy, there was a higher rate of previous preterm delivery than among those who delivered at term.
Characteristics | Delivery ≥37 wk | sPTD <37 wk | sPTD <34 wk | |||
---|---|---|---|---|---|---|
Base cohort (n=2157) | Substudy (n=1035) | Base cohort (n=111) | Substudy (n=54) | Base cohort (n=20) | Substudy (n=9) | |
Maternal age (y) | 32.6 (4.4) | 32.6 (4.3) | 32.6 (4.0) | 33.2 (4.0) | 32.9 (4.5) | 34.8 (2.8) |
Maternal height (cm) | 159.3 (5.6) | 159.7 (5.6) | 159.1 (5.8) | 159.9 (5.5) | 157.2 (5.6) | 160.2 (5.0) |
Maternal weight (kg) | 55.3 (49.9–61.6) | 55.8 (50.2–62.0) | 55.7 (50.9–61.0) | 56.5 (51.6–63.7) | 53.9 (49.2–62.4) | 55.9 (50.2–62.1) |
Body mass index (kg/m 2 ) | 21.8 (19.9–24.1) | 21.9 (19.9–24.2) | 21.8 (20.4–24.1) | 22.2 (20.6–24.9) | 22.0 (19.9–25.0) | 22.2 (21.6–25.5) |
Cigarette smoker | 273 (12.7%) | 120 (11.6%) | 12 (10.8%) | 6 (10.9%) | 1 (5.0%) | 1 (11.1%) |
In vitro fertilization | 42 (1.9%) | 23 (2.2%) | 3 (2.7%) | 3 (5.5%) | 1 (5.0%) | 1 (11.1%) |
Previous cervical surgery | 17 (0.8%) | 14 (1.4%) | 0 | 0 | 0 | 0 |
Gestational age (wk) | 12.3 (12.0–12.6) | 12.3 (12.0–12.6) | 12.1 (12.0–12.4) | 12.3 (12.0–12.4) | 12.4 (12.0–12.6) | 12.3 (11.9–12.5) |
Parity | ||||||
Nulliparous | 1197 (55.5%) | 574 (55.4%) | 66 (59.5%) | 30 (54.5%) | 9 (45.0%) | 4 (44.4%) |
Parous without previous PTD | 926 (42.9%) | 449 (43.4%) | 38 (34.2%) | 21 (38.2%) | 8 (40.0%) | 3 (33.3%) |
Parous with previous PTD | 34 (1.6%) | 12 (1.2%) | 7 (6.3%) a | 4 (7.4%) a | 3 (15.0%) a | 2 (22.2%) a |
Ethnicity | ||||||
Chinese | 2154 (99.9%) | 1035 (100.0%) | 110 (99.1%) | 54 (100.0%) | 20 (100.0%) | 9 (100.0%) |
Non-Chinese | 3 (0.1%) | 0 | 1 (0.9%) | 0 | 0 | 0 |
Base-cohort study analysis
The median cervical length-s at 11 to 13 weeks’ gestation was 33.6 mm (IQR, 31.3–36.2), which was significantly shorter than the cervical length-t (36.9 mm; IQR, 34.0–40.0; Wilcoxon signed-rank test, P <.001). Multivariate linear regression analysis in the term delivery group demonstrated that for the prediction of log 10 cervical length-s, independent contributions were provided by maternal weight and gestational age at the time of the scan, but the log 10 cervical length-t was not affected by maternal characteristics or gestational age at the time of the scan ( Supplemental Table 1 ).
Compared with the term delivery group, the median cervical length-t was significantly shorter in women who delivered spontaneously at <34 weeks’ gestation (36.9 mm vs 35.1 mm; P =.015), whereas no marked difference was observed in women with spontaneous delivery at <37 weeks’ gestation. The cervical length-s was not markedly different between the spontaneous preterm and term delivery groups. The same findings were observed for the MoM values of cervical length measured by the 2 methods ( Table 2 , Figure 3 ).
Variables | Delivery ≥37 wk | Spontaneous delivery | |
---|---|---|---|
<37 wk | <34 wk | ||
Total (n) | 2157 | 111 | 20 |
Cervical length (mm) | |||
Single-line method | 33.6 (31.2–36.2) | 33.3 (31.5–35.6) | 33.0 (31.7–34.7) |
2-line method | 36.9 (33.9–40.0) | 36.2 (33.9–39.2) | 35.1 (33.1–36.7) a |
Cervical length (MoM) | |||
Single-line method | 1.000 (0.930–1.079) | 0.991 (0.939–1.059) | 0.983 (0.944–1.033) |
2-line method | 1.000 (0.920–1.086) | 0.983 (0.919–1.063) | 0.951 (0.898–0.995) a |
Substudy (n) | 1035 | 54 | 9 |
Cervical length (2-line) (mm) | 36.7 (33.9–39.6) | 36.4 (33.4–38.8) | 35.7 (33.1–37.2) |
Cervical length (2-line) (MoM) | 0.998 (0.923–1.078) | 0.991 (0.909–1.055) | 0.971 (0.901–1.013) |
Mean cervical SWE (kPa) | 30.6 (21.9–41.4) | 28.0 (17.6–36.4) a | 21.3 (14.0–41.8) |
Mean cervical SWE (MoM) | 1.014 (0.726–1.369) | 0.926 (0.582–1.203) a | 0.705 (0.462–1.383) a |