Role of acoustic radiation force impulse and shear wave velocity in prediction of preterm birth: a prospective study

Abstract

Background

Preterm birth is one of the important causes of neonatal morbidity where we rely on subjective criteria such as modified Bishop’s scoring and contemporary sonographic measurement of cervical length. Acoustic radiation force impulse (ARFI) is a technological advancement in elastography that can be employed in prediction of cervical softening and preterm labor.

Purpose

To evaluate the role of ARFI technique and shear wave velocity (SWV) estimates as a predictor of preterm birth and its comparison with other clinical and sono-elastographic measures.

Material and Methods

Thirty-four pregnant women (gestation age = 28–37 weeks age) showing features suggestive of preterm labor were included and evaluated with modified Bishop’s score, cervical length by ultrasound (US), ARFI to derive Elastography index (EI), and SWV of the cervix. The patients were later divided into two groups, using the clinical outcome of preterm or term delivery.

Results

Twenty patients delivered at term (gestational age > 37 weeks) and 14 were preterm. Receiver operating characteristics (ROC) curves showed SWV with highest sensitivity and specificity (93% and 90%, respectively) for the prediction of preterm birth at a cutoff value of 2.83 m/s. EI and modified Bishop’s score were comparable to each other, but were less sensitive techniques.

Conclusion

Elastographic assessment of antenatal cervix is a novel technique of virtual palpation of internal os and can be utilized as an objective criterion for preterm birth prediction.

Keywords

Cervix elastography preterm birth ARFI imaging tissue elasticity imaging

Introduction

Preterm birth is defined as delivery after the gestational age of viability (28 weeks) but prior to 37 completed weeks or 259 days of gestation (1). It is estimated to affect approximately 13 million of annual births worldwide (2) and is a cause of nearly 75% of neonatal deaths. Hence, early detection of preterm labor is crucial in the prevention of neonatal and maternal morbidity. Existing pieces of evidence rely on clinical examination, modified Bishop’s score, and cervical length estimation by ultrasound (US) for preterm assessment; however, there is no standard and objective criterion for prediction of cases at risk of preterm birth.

Sono-elastography is a well-known technique for estimation of elasticity properties of any organ, based on its tissue milieu; however, there is a dearth of literature regarding its role in prediction of cervical ripening during pregnancy. Acoustic radiation force impulse (ARFI) is a technique of dynamic elastography (independent of manual compression) (3 –5) which utilizes short duration pushing pulses leading to tissue displacements (6,7). These displacements can be utilized to generate color elastograms (Elastography index [EI]) and calculate shear wave velocity (SWV in m/s), proportional to tissue stiffness.

The current use of ARFI is seen in the evaluation of the liver, pancreas, breast, thyroid, and prostate (8 –12), with a recent study emphasizing its role in cervical malignancy (13). The role of elastography in cervical ripening in preterm birth is discussed under various studies with very few of them based on shear wave elastography. The present study aimed to evaluate the sensitivity and clinical significance of ARFI for prediction of preterm birth by detecting changes in cervical tissue elasticity. The elastographic measures were also correlated with conventional methods to establish their role in virtual palpation of cervix.

Material and Methods

The study group of 34 primigravida clinically diagnosed with premature uterine contractions (≥4 in 20 min or >8 in 60 min, duration >40 s) at gestational ages of 28–37 weeks, was selected from the Department of Obstetrics and Gynecology. This was a prospective study, approved by the institutional review board and was conducted over a one-year period. Patients presenting with ruptured membranes, advanced stage of labor, and with confounding factors like the previous history of preterm birth, cervical surgery (cerclage), multiple gestation, and polyhydramnios were excluded.

Per vaginal assessment was done for every patient by the same obstetrician. Modified Bishop’s score was used to predict favorable or ripened cervix for induction of labor. Ordinal values of 0 and 1 were assigned for the score <6 (unfavorable cervix) and ≥6 (favorable cervix), respectively. The score for cervical consistency was charted separately for every case. Next, the patients were sent for standard obstetrical US and cervical length assessment, where all the scans were done by the same radiologist, using an Acuson S2000 diagnostic ultrasound system (Siemens Healthcare, Erlangen, Germany) with abdominal probe of 3.5-MHz frequency.

Sono-elastography assessment

Cervical Elastographic assessment was also done on the same ultrasound system by the same radiologist, using ARFI-based (eSie Touch EI and Virtual Touch tissue quantification [VTQ]) elasticity models. eSie Touch EI is a method which generates color scale elastograms, with increasing tissue stiffness from red to violet and intermediate color patterns. Fig. 1 shows the EI, which is a numerical value of 0–4 for a specific color (0: violet = hardest; 1: blue = medium hard; 2: green = medium soft; 3: yellow = soft; 4: red = softest). In cases of color overlapping, the softer alternative was used.

The VTQ model employs ARFI US imaging, where the rectangular region of interest (ROI) box of size 10 × 6 mm was placed on the anterior wall of the internal os (only in cases where depth from skin surface was <80 mm), which gives SWV or VTQ value in m/s (Fig. 2). The average of three values was charted for each case.

Statistical analysis

The clinical, sonographic, and sono-elastographic data were charted for all cases in Microsoft Excel 2010 (Microsoft Corp., Redmond, WA, USA). All statistical analysis was done using SPSS Statistics software for Macintosh, (Version 24.0. IBM Corp., Armonk, NY, USA). The ordinal variables were analyzed using non-parametric Mann–Whitney U test, while Student’s t-test was used for interval/ scale variables. Receiver operating characteristics (ROC) curves were designed for all the variables and an individual cutoff value of each variable was derived for the best sensitivity and specificity in prediction of preterm birth. Comparison of ROC curves of different variables was done using Hanley and McNeil methodology (14). The level of statistical significance was determined at P value < 0.05.

Results

Thirty-four patients were categorized into preterm (n = 14) and term (n = 20) groups, based on the timing of delivery. The mean age of the preterm patients was 27 years and of the term group 25 years. The age range of the cohort was 22–33 years with mean age of 26 years. Mean gestational age was comparable in both sets of patients and was 30.5 weeks in the preterm and 30.70 weeks in the term group. Mean values of the scale variables such as SWV (VTQ) were 2.113 m/s in preterm and 3.641 m/s in term-delivered women, while the cervical length was 2.25 cm and 2.53 cm, respectively (Table 1). Student’s t-test revealed that preterm women had statistically significant lower SWV, (t(32) = −5.357, P < 0.001) than the patients who delivered at term, while the results were insignificant for the cervical length (t(32) = −1.105, P = 0.277).

Table 1.

Sensitivities, specificities and AUC of the test variables under study at their ROC curve derived cutoffs.

Test variables	AUC	Cutoff value	Sensitivity (%)	Specificity (%)
SWV (m/s)	0.895	2.83	93	90
Modified Bishop’s score	0.793	6	78.6	80
Cervical consistency	0.602	1.50 (medium to soft)	50	80
Cervical length (cm)	0.668	2.45	57	65
EI	0.745	2.50 (yellow–green)	71.4	75

ROC, receiver operating characteristics; AUC, area under ROC curve; SWV, shear wave velocity; EI, elastography index.

Medians of the ordinal variables under study like modified Bishop’s score, cervical consistency scoring, and EI were compared using Mann–Whitney U test, which showed a statistically significant higher Bishop’s score (U = 58, P = 0.003) and EI (U = 71.5, P = 0.015) for preterm women with insignificance of scoring of cervical consistency in isolation (U = 93, P = 0.104).

ROC curves were framed for each of the variables (Fig. 3) with Table 2 showing ROC curve-derived cutoff values and their corresponding sensitivities and specificities. SWV showed the highest area under the curve (AUC) for prediction of preterm birth, with the highest sensitivity of 93% and specificity of 90% (Table 1). Positive predictive value of SWV for preterm birth was calculated to be 87.5%. Inter-ROC curve comparison for SWV and cervical length was done to give statistically significant P value of 0.047. Similar results could be seen through clustered box-and-whisker plots (Fig. 4) of all the variables (except modified Bishop’s score, having only binomial distribution). These box plots reflect the significantly lower median value and interquartile range (IQR) of the SWV for preterm than for women delivered at term; with no overlap in their IQR. The cervical length and EI values showed considerable overlap.

Table 2.

Correlation matrix in between test variables of all the patients under study, irrespective of their grouping.

Test variables		SWV (VTQ in m/s)	Modified Bishop’s score	Cervical consistency	Cervical length (cm)	EI (eSie)
SWV (m/s)	r-value*	1	−0.223	−0.114	0.036	−0.0363^†
SWV (m/s)	P value		0.205	0.522	0.839	0.035
Modified Bishop’s score	r-value	−0.223	1	0.613^†	−0.026	0.264
Modified Bishop’s score	P value	0.205		0.000	0.886	0.131
Cervical consistency	r-value	−0.114	0.613^†	1	0.030	0.181
Cervical consistency	P value	0.522	0.000		0.867	0.307
Cervical length (cm)	r-value	0.036	−0.026	0.030	1	−0.176
Cervical length (cm)	P value	0.839	0.886	0.867		0.319
EI	r-value	−0.0363^†	0.264	0.181	−0.176	1
EI	P value	0.035	0.131	0.307	0.319

r-value = Pearson correlation coefficient.

†

Correlation is significant at P value < 0.05 (two-tailed).

SWV, shear wave velocity; EI, elastography index.

Correlations between the parameters of preterm birth prediction are shown in Table 2, which revealed inverse correlation of SWV with modified Bishop’s score and EI, indicative of progressive decline in SWV with cervical softening. Positive correlation of EI with Bishop’s score and cervical consistency was also seen, thus reflecting the role of EI as a technique of virtual palpation of cervix. However, statistical significance was only shown between SWV and EI (P = 0.035).

Discussion

The primary findings of this study are, first, dynamic elastography can be used for the assessment of cervix during pregnancy even with a transabdominal approach. Second, ARFI can be applied to predict risk of preterm birth. Third, the elastographic parameters can be correlated with conventional clinical and sonographic parameters as labor progresses. Finally, and the most important, SWV can be used as an objective criterion for the prediction of preterm birth, even in low-risk pregnancy.

Risk assessment of preterm birth is a major challenge in modern-day obstetrics. Advancements in diagnostic modalities have been helpful in detecting a vast majority of intrauterine fetal abnormalities; however, preterm birth prediction is still inaccurate in more than 50% of the cases (15). Hence, this field demands substantial research to reduce fetal and maternal morbidity. A complex pathophysiology is the cause of preterm birth, where multiple overlapping pathways are converging towards the remodeling of the cervix. Hence, research should be targeted towards the estimation and prediction of the end process of these complex pathways, i.e. studying the pre-labor properties of the cervix. In vivo study of the histopathological and biochemical properties of the cervix is impractical and thus requires non-invasive techniques to study its microstructural properties.

Existing literature and routine obstetrical practice still rely on cervical length of 25 mm as a criterion for preterm risk; however, it may lead to underestimation in 62.7% of preterm deliveries where the cervical length was >25 mm (16,17). Our study shows 57% sensitivity of the cervical length (using 24–26 mm cutoffs) with specificity in the range of 35–45%. These results justify the need of additional and standard criterion for preterm birth prediction in cases of normal cervical length. Since the pathogenesis of preterm labor is multifactorial, we have included women carrying low risk of preterm birth, to exclude the confounding factors.

During pregnancy and labor, the cervix undergoes dynamic changes in its tissue properties, which can be broadly studied as cervical softening and ripening. Softening refers to the gradual change in the extracellular matrix of the cervix during pregnancy and precedes the dilatation and effacement while ripening is equivalent to dilatation and effacement of the cervix during labor, which is attributed to the release of pro-inflammatory cytokines in cervical stroma (18). Yamaguchi et al. were the first to apply elastography in obstetrics using color scale elastograms (19), which was later modified by Pries et al. into the elastography index (20). Cervical elastography is primarily based on assessment of the softening while elevated Bishop’s score correlates well with ripening and labor (15). Thus, in this study we have charted the score of cervical consistency separately, however with statistically insignificant correlations with SWV and EI. Similar results were seen in a study by Swiatkowska-Freund et al. (21) where the EI of the internal os showed statistically significant correlation with Bishop’s score, preterm delivery, and time to delivery except for score of cervical consistency. The sensitivity of EI was estimated by Wozniak et al. in a study using four-scale color index (17), which showed a 68.6% sensitivity for red color and a 85.7% for a combination of red and yellow (warm colors) with specificity in the range of 97–99%. Our study showed 71% sensitivity and 75% specificity at an EI cutoff of 2.5, indicative of red, yellow, and yellow–green as warm colors and predictors of soft cervix. The combined sensitivity and specificity of the colors were out of the scope of this study. These abovementioned studies (17, 21) were based on static elastography using physiological cardiovascular movements for generation of elastograms which are less standardized and more operator-dependent than ARFI, employed in our study.

Studies using strain ratio (static elastography) (22 –24) and shear wave speed (dynamic elastography) (25 –27) are summarized in Table 3 with their corresponding applications. These studies have stressed that the shear wave elastography is feasible in pregnant cervix (25), independent of depth from transducer surface (26), and can distinguish between a ripe and unripe cervix (27). The study by Muller et al. (25) highlighted the statistically significant lower SWV values in preterm women with no remarkable association between SWV and cervical length (r = 0.46), which were similar to our results, stressing upon the fact that long cervices can be soft and hence cervical length cannot be used as a one-stop-shop objective criterion for preterm birth prediction. Ex vivo studies by Carlson et al. showed spatial biological variability within and between cervices, with increased tissue homogeneity in ripe cervix and increased shear wave speed in proximal location of cervix. The study also showed statistically significant difference in SWV of ripe and unripe cervices, i.e. 2.11 to 2.68 m/s and 3.45 to 3.56 m/s, respectively, similar to the pattern showed in our study for preterm and term cervices. However, the spatial variability and location-based differences in SWV were not covered in our study.

Table 3.

Studies on elastography of cervix with methods employed and their applications.

Previous studies, year	Sample size (n)	Vendor equipment	Elastography mode (impulse generation)	Assessment method	Study purpose
Yamaguchi et al., 2007 (19)	Not available	Hitachi EUB-8500	Static (probe)	Elastography index (color assessment)	Elastography in obstetrics
Molina et al., 2012 (22)	112	Toshiba Aplio-MX	Static (probe)	Strain ratio	Quantification of elastographic colors
Hernandez-Arnande et al., 2013 (23)	262	Hitachi Hi Vision 9000	Static (probe)	Strain ratio	Anatomical plane for cervix elastography
Fruscalzo et al., 2014 (24)	74	Toshiba Aplio-XG	Static (probe)	Strain ratio	Internal os as site of elastography
Swiatkowska-Freund et al., 2014 (21)	282	Medison Accuvix V10	Static (cardiovascular movement)	Elastography index (color assessment)	Preterm prediction
Wozniak et al., 2014 (17)	333	Medison Accuvix V20	Static (cardiovascular movement)	Elastography index (color assessment)	Preterm prediction
Hernandez-Arnande et al., 2014 (26)	154	SuperSonic Imagine	Dynamic (shear wave)	Shear wave velocity	Effect of depth on shear wave estimation
Muller et al., 2015 (25)	157	Aixplorer SuperSonic Imagine	Dynamic (shear wave)	Shear wave velocity	Feasibility
Carlson et al., 2014 (27)	22	Siemens Acuson S2000	Dynamic (shear wave)	Shear wave velocity	Ex-vivo study on ripe and unripe cervix

The main limitation of this study is its small sample size which leads to a higher percentage of cases with preterm delivery; however, the study is too small to be generalized to a whole population. We have included only the primigravida women and the ones without any risk factor for preterm labor. This may lead to inappropriateness of the results in patients with risk factors for preterm delivery. Another limitation is the use of the transabdominal scan probe which leads to difficulty in assessment of deeper tissue and obese patients; however, it may be beneficial over the endocavitary transducer which is unacceptable to a number of women and can lead to bleeding or infection. Elastography methods lack standardization and are variable due to changes in position and size of the ROI; hence, a single examiner performed all the measurements to avoid inter-observer bias. Variability due to size of the ROI box was removed by default, as the ROI size of 10 × 6 mm is standardized by the vendor and cannot be altered by the radiologist, thus improving the reproducibility of the results. This is an initial attempt to evaluate the role of dynamic elastography and to use SWV as a standard method for virtual palpation of cervix and for prediction of preterm labor. Future studies with more number of patients and using other objective criteria, e.g. fetal adrenal gland biometry (28) can be done to standardize elastography methods and to open way for early diagnoses of preterm delivery.

In conclusion, ARFI-based dynamic elastography appears to be promising in the evaluation of the cervix during pregnancy and it also provides shear wave speed as a standard and objective criterion for prediction of preterm birth.

Footnotes

Acknowledgments

The authors thank Dr. Mosam Sinha for his assistance in the language editing of the manuscript.

Declaration of Conflicting Interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) received no financial support for the research, authorship, and/or publication of this article.

References

International classification of diseases and related health problems. 10th revision. Geneva: World Health Organization, 1992.

Svigos JM, Dodd JM, Robinson JS. Threatened and actual preterm labor including mode of delivery. In: James D, editor. High Risk Pregnancy Management Options. 4th ed. Elsevier India Pvt Ltd, 2012:1075–1090.

Sporea

Sirli

Bota

et al.

Comparative study concerning the value of acoustic radiation force impulse elastography (ARFI) in comparison with transient elastography (TE) for the assessment of liver fibrosis in patients with chronic hepatitis B and C. Ultrasound Med Biol 2012; 38: 1310–1316.

Colombo

Buonocore

Del Poggio

et al.

Head-to-head comparison of transient elastography (TE), real-time tissue elastography (RTE), and acoustic radiation force impulse (ARFI) imaging in the diagnosis of liver fibrosis. J Gastroenterol 2010; 47: 461–469.

Sporea I, Sirli R, Popescu A, et al: Is it better to use together transient elastography (TE) and acoustic radiation force impulse elastography (ARFI) for fibrosis evaluation in patients with chronic HCV hepatitis? Gastroenterology2011;140 (Suppl):S968.

Bamber

Cosgrove

Dietrich

et al.

EFSUMB guidelines and recommendations on the clinical use of ultrasound elastography. Part 1: Basic principles and technology. Ultraschall Med 2013; 34: 169–184.

Torr

. The acoustic radiation force. Am J Phys 1984; 52: 402–408.

D’Onofrio

Gallotti

Salvia

et al.

Acoustic radiation force impulse (ARFI) ultrasound imaging of pancreatic cystic lesions. Eur J Radiol 2011; 80: 241–244.

Meng

Zhang

et al.

Preliminary results of acoustic radiation force impulse (ARFI) ultrasound imaging of breast lesions. Ultrasound Med Biol 2011; 37: 1436–1443.

10.

Gallotti

D’Onofrio

Romanini

et al.

Acoustic radiation force impulse (ARFI) ultrasound imaging of solid focal liver lesions. Eur J Radiol 2012; 81: 451–455.

11.

Friedrich-Rust

Romenski

Meyer

et al.

Acoustic radiation force impulse-imaging for the evaluation of the thyroid gland: a limited patient feasibility study. Ultrasonics 2012; 52: 69–74.

12.

Zheng

Mao

et al.

A comparison of virtual touch tissue quantification and digital rectal examination for discrimination between prostate cancer and benign prostatic hyperplasia. Radiol Oncol 2012; 46: 69–74.

13.

et al.

Evaluation of cervical cancer detection with acoustic radiation force impulse ultrasound imaging. Exp Ther Med 2013; 5: 1715–1719.

14.

Hanley

McNeil

. A method of comparing the areas under receiver operating characteristic curves derived from the same cases. Radiology 1983; 148: 839–843.

15.

Swiatkowska-Freund

Preis

. Cervical elastography during pregnancy: clinical perspectives. Int J Womens Health 2017; 9: 245–254.

16.

Celik

Gajewska

et al.

Cervical length and obstetric history predict spontaneous preterm birth: development and validation of a model to provide individualized risk assessment. Ultrasound Obstet Gynecol 2008; 31: 549–554.

17.

Wozniak

Czuczwar

Szkodziak

et al.

Elastography in predicting preterm delivery in asymptomatic, low-risk women: a prospective observational study. BMC Pregnancy Childbirth 2014; 14: 238–238.

18.

House

Kaplan

Socrate

. Relationships between mechanical properties and extracellular matrix constituents of the cervical stroma during pregnancy. Semin Perinatol 2009; 33: 300–307.

19.

Yamaguchi

Kamei

Kozuma

et al.

Tissue elastography imaging of the uterine cervix during pregnancy. J Med Ultrason 2007; 34: 209–210.

20.

Preis

Swiatkowska-Freund

Pankrac

. Elastography in the examination of the uterine cervix before labor induction. Ginekol Pol 2010; 81: 757–761.

21.

Swiatkowska-Freund

Pankrac

Preis

. Intra- and inter-observer variability of evaluation of uterine cervix elastography images during pregnancy. Ginekol Pol 2014; 85: 360–364.

22.

Molina

Gómez

Florido

et al.

Quantification of cervical elastography: a reproducibility study. Ultrasound Obstet Gynecol 2012; 39: 685–689.

23.

Hernandez-Andrade

Hassan

Ahn

et al.

Evaluation of cervical stiffness during pregnancy using semiquantitative ultrasound elastography. Ultrasound Obstet Gynecol 2013; 41: 152–161.

24.

Fruscalzo

Londer

Frohlich

. Quantitative elastography for cervical stiffness assessment during pregnancy. Biomed Res Int 2014; 2014: 826535–826535.

25.

Muller

Aït-Belkacem

Hessabi

et al.

Assessment of the cervix in pregnant women using shear wave elastography: a feasibility study. Ultrasound Med Biol 2015; 41: 2789–2797.

26.

Hernandez-Andrade

Aurioles-Garibay

Garcia

et al.

Effect of depth on shear-wave elastography estimated in the internal and external cervical os during pregnancy. J Perinat Med 2014; 42: 549–557.

27.

Carlson

Feltovich

Palmeri

et al.

Estimation of shear wave speed in the human uterine cervix. Ultrasound Obstet Gynecol 2014; 43: 452–458.

28.

Turan

Funai

et al.

Ultrasound measurement of fetal adrenal gland enlargement: an accurate predictor of preterm birth. Am J Obstet Gynecol 2011; 204: 311–311.