Objective
Impaired fetal growth might be better evaluated in twin pregnancies by assessing the intertwin discordance rather than the individual fetal size. The aim of this study was to investigate the prediction of perinatal loss in twin pregnancy using discordance in fetal biometry and Doppler.
Study Design
This was a retrospective cohort study in a tertiary referral center. The estimated fetal weight (EFW), umbilical artery (UA) pulsatility index (PI), middle cerebral artery (MCA) PI, cerebroplacental ratio (CPR), and their discordance recorded at the last ultrasound assessment before delivery or demise of one or both fetuses were converted into centiles or multiples of the median (MoM). The discordance was calculated as the larger value–smaller value/larger value. A logistic regression analysis was performed to identify, and adjust for, potential confounders. The predictive accuracy was assessed using receiver-operating characteristic curve analysis.
Results
The analysis included 620 (464 dichorionic diamniotic and 156 monochorionic diamniotic) twin pregnancies (1240 fetuses). Perinatal loss of one or both fetuses complicated 16 pregnancies (2.6%). The combination of EFW discordance and CPR discordance had the best predictive performance (area under the curve, 0.96; 95% confidence interval, 0.92–1.00) for perinatal mortality. The detection rate, false-positive rate, positive likelihood ratio, and negative likelihood ratio were 87.5%, 6.7%, 13.08, and 0.13, respectively. The EFW centile, EFW below the 10th centile (small for gestational age), UA PI discordance, MCA PI discordance, and MCA PI MoM were significantly associated with the risk of perinatal loss on univariate analysis, but these associations became nonsignificant after adjusting for other confounders ( P = .097, P = .090, P = .687, P = .360, and P = .074, respectively). The UA PI MoM, CPR MoM, EFW discordance, and CPR discordance were all independent predictors of the risk of perinatal loss, even after adjusting for potential confounders ( P = .022, P = .002, P < .001, and P = .010, respectively).
Conclusion
EFW discordance and CPR discordance are independent predictors of the risk of perinatal loss in twin pregnancies. Their combination could identify the majority of twin pregnancies at risk of perinatal loss. These findings highlight the importance of discordance in Doppler indices of fetal hypoxia, as well as fetal size, in assessing the risk of perinatal mortality.
Twin pregnancy is associated with an increased risk of perinatal mortality and morbidity; the risk is significantly higher in monochorionic compared with dichorionic pregnancies. In 2009, the stillbirth rate in the United Kingdom was 12.3 per 1000 twin births, compared with 5 per 1000 singleton births.
The incidence of twins is rising, mainly because of assisted reproduction techniques. In 1980, 10 women per 1000 giving birth in England and Wales had multiple births compared with 16 women per 1000 in 2009. Ultrasound assessment of fetal biometry is used to identify and monitor twin pregnancies at risk of adverse pregnancy outcome, such as selective fetal growth restriction (sFGR). In a singleton pregnancy, impaired fetal growth can be assessed by comparing biometry and Doppler indices against standards for uncomplicated pregnancies. However, this might be better evaluated in twin pregnancies by assessing the intertwin discordance.
Both birthweight and estimated fetal weight (EFW) discordance have been shown to be strongly associated with adverse pregnancy outcome. The UK National Institute for Health and Care Excellence guidelines for the clinical management of multiple pregnancy state that a 25% or greater difference in size between twins should be considered an important clinical indicator of fetal growth restriction (FGR) and that this finding should trigger referral to a tertiary level fetal medicine center. Similarly, the American College of Obstetricians and Gynecologists considers 15–25% EFW discordance as a cutoff for the diagnosis of significantly discordant fetal growth.
The ratio of middle cerebral to umbilical artery Doppler indices, also termed the cerebroplacental ratio (CPR), has an established role in predicting adverse outcome in singleton pregnancies complicated by small for gestational age. Recently assessment of CPR has also been shown to be of value in assessing average for gestational age fetuses that are compromised by placental insufficiency at term.
CPR is emerging as a better proxy marker of fetal compromise, in particular secondary to placental insufficiency and hypoxemia, than fetal size alone. Despite the fact that CPR has been investigated in many studies in singleton pregnancies, there is a paucity of data on the value of CPR in predicting adverse outcome in twin pregnancy. There is limited evidence suggesting that the value of CPR in twin pregnancies could be similar to that reported in singleton pregnancies. Gaziano et al have demonstrated that CPR has been shown to be superior to umbilical artery (UA) and middle cerebral artery (MCA) in predicting adverse fetal and neonatal outcomes, such as growth restriction and the length of stay in the neonatal unit.
The main aim of this study was to investigate the role of fetal size, Doppler indices, and their discordance in the prediction of perinatal loss in twin pregnancies.
Materials and Methods
This was a retrospective cohort study in a single tertiary referral center over a 14 year period from 2000 to 2013. Cases were identified by searching the ViewPoint database (ViewPoint 5.6.8.428; ViewPoint Bildverarbeitung GmbH, Weßling, Germany) in the fetal medicine unit (St George’s Hospital). The inclusion criteria were twin pregnancies in which routine fetal biometry, UA, and MCA Doppler were recorded at the last visit before delivery or the diagnosis of intrauterine death of one or both fetuses. Gestational age (GA) was determined by the crown-rump length of the larger twin at the 11-14 week scan or by head circumference if assessed after 14 weeks’ gestation.
Chorionicity was determined by ultrasound evaluation according to the number of placentas and the presence of the lambda or T signs, and confirmed after birth. Pregnancies complicated by structural or chromosomal abnormalities, twin to twin transfusion syndrome, twin reversed arterial perfusion, of unknown chorionicity, monochorionic monoamniotic, and high-order multiple gestations were not included in the analysis. Pregnancies ending in termination or loss to follow-up were also excluded.
The main outcome in this study was perinatal loss. Stillbirth was defined as the death of the fetus after 24 weeks and before birth, whereas perinatal loss included stillbirth and neonatal death within the first 28 days. The study was exempt from review by the Wandsworth Research Ethics Committee. There was an overlap between the population of this study and that of the Southwest Thames Obstetric Research Collaborative (STORK) cohort. However, the STORK cohort included 9 hospitals, the period was 2000–2009, and none of the STORK papers included fetal Doppler data.
The population in the current study is derived from a single center (St George’s Hospital), the period was between 2000 and 2013, and it includes Doppler data. Therefore, the data from the last 4 years were not included in the STORK cohort.
Routine fetal biometry was performed according to a standard protocol and the EFW calculated using Hadlock’s formula. Qualified and experienced sonographers performed all scans. Ultrasound examinations were performed with Voluson E8/E6/730 (GE Healthcare, Little Chalfont, Buckinghamshire, UK) and Philips IU22 ultrasound machines (Philips, Bothell, WA) using 2-8 MHz convex probes. Ultrasound data, including head circumference, abdominal circumference, and femur length, were measured according to a standard protocol. The UA and MCA Doppler waveforms were recorded using color Doppler, and the pulsatility index (PI) was calculated according to a standard protocol. In brief, the MCA PI values were obtained in the space in which the artery passes by the sphenoid wing close to the circle of Willis, and the UA PI values were obtained in free loops of umbilical cord. The measurements were obtained in the absence of fetal tachycardia, and keeping the insonation angle with the examined vessels less than 30°.
Most of the studies were performed after 20 weeks. The CPR was calculated as the simple ratio between the MCA PI and the UA PI. All Doppler indices were converted into multiples of median (MoM), correcting for GA using reference ranges, and birthweight values were converted into centiles. The EFW discordance was calculated as (larger EFW–smaller EFW)/larger EFW. UA PI discordance was calculated as (higher UA PI–lower UA PI)/higher UA PI. The MCA PI discordance was calculated as (higher MCA PI–lower MCA PI)/higher MCA PI.
The CPR discordance was calculated as (higher CPR–lower CPR)/higher CPR. Small for gestational age (SGA) was defined as either EFW or birthweight less than the 10th centile. Dichorionic diamniotic (DCDA) twin pregnancies are offered an ultrasound scan every 4 weeks from 20–22 weeks onward, whereas monochorionic diamniotic (MCDA) twin pregnancies are offered follow-up scans every 2 weeks from 16 weeks until 20–22 weeks (to monitor for the development of TTTS) and then every 3-4 weeks thereafter.
The frequency of repeating ultrasound scans in case of an abnormal Doppler waveform in the UA or MCA depends on the severity of these abnormalities (raised PI or absent EDF in the UA Doppler) and the GA. Because delivery prior to 32–34 weeks was not considered or recommended (to avoid the risks of prematurity) when one fetus was smaller or had abnormal Doppler, especially in DCDA twin pregnancies, the frequency of ultrasound scans might be high if the gestational age is 30 weeks or more, whereas it could be low if the gestational age is less than 28 weeks’ gestation. The frequency of repeating ultrasound scans in the case of a normal Doppler waveform in the UA or MCA is the same as in the routine clinical protocol.
Statistical analysis
Categorical data were presented as number (percentage) and compared using the Fisher exact test or χ 2 test. Continuous data were presented as median (interquartile range [IQR]). The D’Agostino and Pearson Omnibus test was used to assess the normality of the data. Nonparametric analysis using the Mann-Whitney U test was then used to compare continuous data between the study groups. Logistic regression analysis was performed to identify, and adjust for, potential confounders. Both unadjusted and adjusted odds ratios were calculated. The predictive accuracy was assessed using receiver-operating characteristic (ROC) curve analysis. Values of P < .05 were considered statistically significant. All P values were 2 tailed. The analysis was performed using the statistical software packages SPSS version 18.0 (SPSS Inc, Chicago, IL), Stata 11 (release 11.2; College Station, TX), and GraphPad Prism version 5.0 for Windows (InStata; GraphPad Software Inc, San Diego, CA).
Results
The study cohort included 683 twin pregnancies. Of those 33 (4.8%) were lost to follow-up and were not included in the analysis. We included 620 twin pregnancies (1240 fetuses) in the analysis, with both UA and MCA Doppler assessment at the last ultrasound assessment prior to delivery or the diagnosis of perinatal loss of one or both twins. Of those, 464 were DCDA and 156 were MCDA. The median interval between the ultrasound and outcome was 1.9 weeks (IQR, 1.0–2.9).
Perinatal loss of one or both fetuses complicated 16 pregnancies (2.6%) (15 pregnancies with single perinatal loss and 1 pregnancy with a double perinatal loss). The incidence of perinatal loss was the same in the DCDA (2.6%) and the MCDA (2.6%) pregnancies. The incidence of stillbirth in this cohort was 2.1% (13 pregnancies). Of the perinatal losses, 82.4% were SGA fetuses, of which 41.2% had absent or reversed flow in the UA.
The median (IQR) number of studies performed in each set of twins was 7 (range, 5–9). The GA at delivery was significantly lower in those pregnancies complicated by perinatal loss compared with those that were not (median, 31.50; IQR, 26.15–33.46 vs 37.00; IQR, 35.29–37.43; P < .001). The GA at delivery of those pregnancies complicated by neonatal death was significantly lower than those that were not (median, 30.71; IQR, 24.79–34.71 vs 37.00; IQR, 35.25–37.43; P = .002).
The maternal characteristics, ultrasound, and birth indices in the groups, according to whether or not the pregnancies were complicated by perinatal loss, are shown in Tables 1 and 2 . Maternal age, parity, body mass index, ethnic origin, and smoking did not differ between those pregnancies complicated by perinatal loss and those that were not ( P = .463, P = .448, P = .910, P = .870, and P = .586, respectively). There were similar proportions of MCDA twin pregnancies in both groups (25%), whereas the proportion of pregnancies conceived using assisted reproductive technology in the group complicated by perinatal loss was half that in the group that was not (13.0% vs 25.0%).
| Variable | Pregnancies without perinatal loss (n = 604) | Pregnancies complicated by perinatal loss (n = 16) | P value |
|---|---|---|---|
| Maternal age in years, median (IQR) | 33.00 (30.00–36.00) | 32.50 (29.25–35.75) | .463 |
| Body mass index (kg/m 2 ), median (IQR) | 24.30 (21.50–27.40) | 24.30 (22.40–25.60) | .91 |
| Nulliparous, n (%) | 359 (59.4) | 8 (50.0) | .448 |
| Ethnicity, n (%) | .870 | ||
| Caucasian | 458 (75.8) | 11 (68.8) | |
| African | 64 (10.6) | 3 (18.8) | |
| Asian | 75 (12.4) | 2 (12.5) | |
| Mixed | 3 (0.5) | 0 | |
| Other | 4 (0.6) | 0 | |
| Smoker, n (%) | 22 (3.6) | 1 (6.3) | .586 |
| Assisted conception, n (%) | 150 (25.0) | 2 (13.03) | .299 |
| Monochorionicity, n (%) | 152 (25.2) | 4 (25.0) | .988 |
| Gestational age at ultrasound (wks), median (IQR) | 34.57 (33.00–35.57) | 27.43 (23.65–29.00) | < .001 |
| Gestational age at delivery (wks), median (IQR) | 37.00 (35.29–37.43) | 31.50 (26.15–33.46) | < .001 |
| Estimated fetal weight discordance | 8.31 (4.03–15.69) | 36.12 (20.49–50.96) | < .001 |
| Umbilical artery pulsatility index discordance | 13.54 (6.74–26.36) | 25.58 (7.65–50.78) | .074 |
| Middle cerebral artery pulsatility index discordance | 13.08 (5.88–23.06) | 23.38 (14.91–31.88) | .003 |
| Cerebroplacental ratio discordance | 17.43 (8.65–30.54) | 44.26 (26.59–68.33) | < .001 |
| Criteria | No perinatal loss | Perinatal loss | P value |
|---|---|---|---|
| (n = 1223) | (n = 17) | ||
| Ultrasound and Doppler criteria | |||
| Gestational age at ultrasound (wks), median (IQR) | 34.57 (33.00–35.57) | 27.43 (23.65–29.00) | < .001 |
| Interval between scan and delivery (wks), median (IQR) | 1.86 (1.00–2.86) | 3.14 (1.36–4.64) | .015 |
| Estimated fetal weight (g), median (IQR) | 2259 (1917–2548) | 519 (353–788) | < .001 |
| Estimated fetal weight centile, median (IQR) | 30.75 (14.39–48.96) | 0.06 (0.01–5.96) | < .001 |
| Small for gestational age, n (%) | 215 (17.3) | 14 (82.4) | < .001 |
| Umbilical artery pulsatility index | 0.98 (0.85–1.10) | 2.00 (1.11–3.11) | < .001 |
| Umbilical artery pulsatility index MoM | 1.05 (0.91–1.18) | 1.59 (1.03–2.68) | < .001 |
| Absent or reversed umbilical artery end-diastolic flow, n (%) | 29 (2.4) | 7 (41.2) | < .001 |
| Middle cerebral artery pulsatility index | 1.71 (1.52–1.97) | 1.48 (1.29–1.72) | .011 |
| Middle cerebral artery pulsatility index MoM | 0.92 (0.82–1.04) | 0.77 (0.64–0.94) | .002 |
| Cerebroplacental ratio | 1.80 (1.49–2.12) | 0.84 (0.42–1.36) | < .001 |
| Cerebroplacental ratio MoM | 0.85 (0.71–1.01) | 0.54 (0.21–0.79) | < .001 |
| Criteria at birth | |||
| Gestational age at delivery (wks), median (IQR) | 37.00 (35.29–37.43) | 31.50 (26.15–33.46) | < .001 |
| Birthweight (g), median (IQR) | 2430 (2069–2740) | 1005 (360–1360) | < .001 |
| Birthweight centile, median (IQR) | 14.11 (4.7–28.70) | 0.10 (0–5.19) | < .001 |
Discordances in EFW, MCA PI, and CPR were all significantly higher in the pregnancies complicated by perinatal loss than in those that were not ( Tables 1 and 2 ). The individual EFW centile, prevalence of SGA, UA PI MoM, MCA PI MoM, and CPR MoM were also significantly different in the perinatal loss group ( Tables 1 and 2 ). There were 115 pregnancies in which both fetuses were SGA. Of those, only 1 pregnancy was complicated by perinatal loss. This was a DCDA twin pregnancy in which both fetuses were stillborn.
The results of the univariate regression analysis of the association between ultrasound parameters and the risk of perinatal loss are shown in Table 3 . All of the parameters examined were significantly associated with the risk of perinatal loss ( P < .05) except chorionicity ( P = .876). Binary variables, including SGA and absent or reversed end-diastolic flow in the UA Doppler had significantly higher odds ratios (21.81 and 28.82, respectively) than continuous variables such as EFW and Doppler PI.
| Ultrasound and Doppler criteria | Unadjusted odds ratio | 95% confidence interval | P value |
|---|---|---|---|
| Estimated fetal weight, g | 0.99 | 0.99–1.00 | < .001 |
| Estimated fetal weight centile | 0.92 | 0.88–0.96 | < .001 |
| Estimated fetal weight discordance | 1.09 | 1.06–1.12 | < .001 |
| Small for gestational age | 21.81 | 6.21–76.57 | < .001 |
| Umbilical artery pulsatility index | 4.14 | 2.53–6.78 | < .001 |
| Umbilical artery pulsatility index MoM | 4.14 | 2.46–6.97 | < .001 |
| Absent or reversed umbilical artery end-diastolic flow | 28.82 | 10.25–81.03 | < .001 |
| Umbilical artery pulsatility index discordance | 1.03 | 1.01–1.05 | .004 |
| Middle cerebral artery pulsatility index | 0.16 | 0.03–0.82 | .027 |
| Middle cerebral artery pulsatility index MoM | 0.01 | 0.00–0.18 | .002 |
| Middle cerebral artery pulsatility index discordance | 1.05 | 1.01–1.8 | .013 |
| Cerebroplacental ratio | 0.05 | 0.02–0.12 | < .001 |
| Cerebroplacental ratio MoM | 0.01 | 0.00–0.04 | < .001 |
| Cerebroplacental ratio discordance | 1.06 | 1.03–1.08 | < .001 |
| Monochorionicity | 0.91 | 0.30–2.82 | .876 |
| Gestational age at ultrasound, wks | 0.60 | 0.52–0.69 | < .001 |
The results of the multivariate regression analysis of the association between ultrasound parameters and the risk of perinatal loss are shown in Tables 4 and 5 . The EFW centile, EFW below the 10th centile (SGA) and MCA PI MoM were significantly associated with the risk of perinatal loss, but these associations became nonsignificant after adjusting for other confounders ( P = .097, P = .090, and P = .074, respectively). The UA PI MoM and CPR MoM were independent predictors of the risk of perinatal loss, even after adjusting for potential confounders ( P = .022 and P = .002, respectively).
| Potential confounding adjusted variables | Adjusted odds ratio | 95% confidence interval | P value |
|---|---|---|---|
| Estimated fetal weight centile | 0.97 | 0.94–1.00 | .097 |
| Small for gestational age | 3.84 | 0.81–18.21 | .090 |
| Umbilical artery pulsatility index multiple of the median | 1.98 | 1.10–3.55 | .022 |
| Middle cerebral artery pulsatility index multiple of the median | 0.05 | 0.00–1.34 | .074 |
| Cerebroplacental ratio multiple of the median | 0.03 | 0.00–0.26 | .002 |
| Estimated fetal weight discordance | 1.07 | 1.04–1.10 | < .001 |
| Umbilical artery pulsatility index discordance | 1.01 | 0.98–1.04 | .687 |
| Middle cerebral artery pulsatility index discordance | 1.02 | 0.98–1.06 | .360 |
| Cerebroplacental ratio discordance | 1.03 | 1.01–1.06 | .010 |
| Potential confounding adjusted variables | Adjusted odds ratio | 95% confidence interval | P value |
|---|---|---|---|
| Estimated fetal weight centile | |||
| None (unadjusted) | 0.92 | 0.88–0.96 | < .001 |
| Gestational age at ultrasound, wks | 0.96 | 0.92–0.99 | .008 |
| Umbilical artery pulsatility index MoM | 0.95 | 0.91–0.99 | .009 |
| Absent or reversed umbilical artery end-diastolic flow | 0.95 | 0.91–0.99 | .014 |
| Middle cerebral artery pulsatility index MoM | 0.93 | 0.89–0.97 | .002 |
| Cerebroplacental ratio MoM | 0.95 | 0.92–0.997 | .034 |
| Umbilical artery pulsatility index MoM plus middle cerebral artery pulsatility index MoM | 0.95 | 0.91–0.99 | .014 |
| Chorionicity | 0.92 | 0.88–0.96 | < .001 |
| Umbilical artery pulsatility index MoM plus middle cerebral artery pulsatility index MoM plus Gestational age at ultrasound, wks | 0.98 | 0.95–1.01 | .199 |
| Cerebroplacental ratio MoM plus gestational age at ultrasound, wks | 0.97 | 0.94–1.00 | .097 |
| Small for gestational age | |||
| None (unadjusted) | 21.81 | 6.21–76.57 | < .001 |
| Gestational age at ultrasound, wks | 11 | 2.89–41.89 | < .001 |
| Umbilical artery pulsatility index MoM | 13 | 3.45–48.94 | < .001 |
| Absent or reversed umbilical artery end-diastolic flow | 12.76 | 3.32–49.00 | < .001 |
| Middle cerebral artery pulsatility index MoM | 17.61 | 4.80–64.60 | < .001 |
| Cerebroplacental ratio MoM | 10.13 | 2.51–40.81 | .001 |
| Umbilical artery pulsatility index MoM plus middle cerebral artery pulsatility index MoM | 12.53 | 3.23–48.62 | < .001 |
| Chorionicity | 22.88 | 6.48–80.74 | < .001 |
| Umbilical artery pulsatility index MoM plus middle cerebral artery pulsatility index MoM plus | |||
| Gestational age at ultrasound, wks | 5.49 | 1.23–24.59 | .026 |
| Cerebroplacental ratio MoM plus gestational age at ultrasound, wks | 3.84 | 0.81–18.21 | .090 |
| Umbilical artery pulsatility index multiple of the median | |||
| None (unadjusted) | 4.14 | 2.46–6.97 | < .001 |
| Gestational age at ultrasound, wks | 2.54 | 1.48–4.33 | .001 |
| Estimated fetal weight centile | 2.79 | 1.62–4.82 | < .001 |
| Small for gestational age | 2.54 | 1.51–4.30 | .001 |
| Middle cerebral artery pulsatility index MoM | 3.42 | 1.93–6.07 | < .001 |
| Small for gestational age plus gestational age at ultrasound, wks | 1.81 | 1.01–3.25 | .046 |
| Estimated fetal weight centile plus gestational age at ultrasound, wks | 1.98 | 1.10–3.55 | .022 |
| Chorionicity | 4.32 | 2.54–7.36 | < .001 |
| Middle cerebral artery pulsatility index multiple of the median | |||
| None (unadjusted) | 0.01 | 0.00–0.18 | .002 |
| Gestational age at ultrasound, wks | 0.01 | 0.00–0.15 | .002 |
| Estimated fetal weight centile | 0.07 | 0.00–1.33 | .076 |
| Small for gestational age | 0.15 | 0.01–2.52 | .187 |
| Umbilical artery pulsatility index MoM | 0.13 | 0.01–3.27 | .215 |
| Absent or reversed umbilical artery end-diastolic flow | 0.25 | 0.01–5.06 | .370 |
| Small for gestational age plus gestational age at ultrasound, wks | 0.05 | 0.00–1.34 | .074 |
| Estimated fetal weight centile plus gestational age at ultrasound, wks | 0.02 | 0.00–0.57 | .023 |
| Chorionicity | 0.01 | 0.00–0.17 | .002 |
| Cerebroplacental ratio multiple of the median | |||
| None (unadjusted) | 0.01 | 0.00–0.04 | < .001 |
| Gestational age at ultrasound, wks | 0.02 | 0.00–0.12 | < .001 |
| Estimated fetal weight centile | 0.02 | 0.00–0.23 | .001 |
| Small for gestational age | 0.05 | 0.01–0.42 | .006 |
| Umbilical artery pulsatility index MoM | 0.04 | 0.00–0.72 | .029 |
| Absent or reversed umbilical artery end-diastolic flow | 0.06 | 0.00–0.91 | .043 |
| Middle cerebral artery pulsatility index MoM | 0.001 | 0.00–0.20 | < .001 |
| Small for gestational age plus gestational age at ultrasound, wks | 0.05 | 0.01–0.45 | .008 |
| Estimated fetal weight centile plus gestational age at ultrasound, wks | 0.03 | 0.00–0.26 | .002 |
| Umbilical artery pulsatility index MoM plus middle cerebral artery pulsatility index MoM | 0.01 | 0.00–0.99 | .049 |
| Chorionicity | 0.005 | 0.00–0.03 | < .001 |
Stay updated, free articles. Join our Telegram channel
Full access? Get Clinical Tree
