TVS-guided cervical strain elastography is more effective than measuring cervical length as an independent predictor of spontaneous preterm delivery in asymptomatic, high-risk women during the mid-trimester

Abstract

Introduction TVS (Transvaginal Sonography) guided Cervical strain elastography (CSE) is now available in tertiary referral centers of LMICs (Low- and Middle-Income Countries). TVS cervical length (CL) assessment is being used routinely. Still, elastography is not used in clinical settings, although enough evidence suggests good predictive value towards sPTD (spontaneous Preterm Delivery). The clinical utility of elastography has not been tested among high-risk populations of LMICs for the prediction of sPTD. Aim To test the performance of TVS-CSE in predicting sPTD among asymptomatic women in the mid-trimester at risk of sPTD either due to clinical factors or due to a short cervix. Method Prospective observational study performed at a tertiary hospital in South India. Asymptomatic pregnant women between 16 and 24 weeks who had one or more clinical risk factors for sPTD or CL <2.5 cm were included. GE Voluson E-8 ultrasound machine was used. After CL measurement, elastography color coding was noted around the internal-os in the sagittal view. The strain ratio (SR) was calculated using the trace method on three ROIs (Region of Interest): Internal-os in sagittal view (IN), whole cervix in sagittal view (WN), and internal-os in axial view (AN). Reference Tissue (RT) of similar size and depth was chosen in the darkest blue region on elastography (stiffest area) outside the cervix, posterior/lateral to the cervix over the ligament insertion. Lower the SR – softer the cervix. Two trained fetal medicine consultants performed the initial 57 cases until intra/inter-observer correlation was satisfactory. Delivery before 37 weeks (after 26 weeks), in which the process of labor has begun spontaneously, or labor was induced after PPROM—was considered as sPTD. SRs were assessed to determine how well they could predict sPTD independently or combined with cervical length. Results Out of 221 recruited,17 were lost to follow-up after 32 weeks; 204 were delivered in our hospital. Irrespective of the route of delivery, 71 (34.8%) had sPTD. Of the remaining 133, 106 delivered at term, and 27 underwent medically indicated PTD. Apart from multiple pregnancies, no other preterm-related risk factors (including CL < 2.5 cm) showed significant association with sPTD. Red CSE pattern around internal-os was associated with a significantly higher (54.5%) incidence of sPTD. CLs were similar (3.63 ± 0.67 vs. 3.63 ± 0.80, p = .981) whereas SRs in all three ROIs were significantly lower among sPTD group versus no sPTD group (IN:0.65 ± 0.29 vs 0.79 ± 0.30 p = .001, WN:0.34 ± 0.13 vs 0.39 ± 0.15, p = .013, AN:0.37 ± 0.16 vs 0.48 ± 0.26, p = .002, respectively). Using ROC curves, while CL was not predictive (AUROC 0.49, p = .81), SRs showed moderate predictive value toward sPTD with the best AUC of 0.624 (p = .003) at IN. Prediction was slightly better for early sPTD <32 weeks (AUC 0.653 p = 0.03). The best cutoff for SR at IN was 0.72, below which there was a moderate accuracy in predicting sPTD (sensitivity 52.11%, specificity 60.9%, PPV 41.57%, NPV 70.44%, diagnostic OR 1.69 and overall accuracy of 57.84%). A weak positive correlation is seen between IN and CL (Pearson’s correlation R = 0.181). Multi-variable binary logistic regression analysis suggested that SRs at IN (Adjusted OR − 0.259 CI 0.079–0.850), AN (Adjusted OR 0.182 CI 0.034–0.963), Multiple Pregnancy (Adjusted OR 3.5 CI 1.51–8.13) and previous sPTD/PPROM (Adjusted OR 2.72 CI 0.97–7.61) independently predicted sPTD. Conclusions TVS CSE performed better than CL as an independent predictive tool toward sPTD, although predictive efficacy was modest at best. Since technology is now available in high-end USG machines in tertiary care centers, we propose optimal utilization of CSE in LMICs to triage at-risk populations since low SRs are strongly associated with sPTD.