In the examined population (n=171), a total of 49.1% (n = 84) were concordant, 19.3% (n = 33) TTTS and 21.6% (n = 37) sIUGR, of which 70.3% (n = 26) early-onset (EO) sIUGR and 29.7% (n = 11) late-onset (LO) sIUGR. Within the TTTS group, 48.5% (n = 16) received a conservative approach with regular follow-ups, 39.4% (n=13) a fetoscopic laser therapy and 12.1% (n=4) an amnioreduction. In the conservative treatment group, a single intrauterine fetal demise of the donor occurred in 4 cases, while a double demise occurred in 3 cases, resulting in overall mortality of 26.7%. In the group treated with laser therapy, one selective feticide was performed due to hydrops fetalis, resulting in a mortality rate of 7.7%.
The linear mixed model showed a positive coefficient of the gestational age (fixed effect), representing a linear increase in interfetal weight difference depending on the gestational age (P>0.001). The relationship of patient groups and progressive interfetal weight difference over time (gestational age) also showed a significant correlation (P<0,001), (Tab. 1.) Hence, interfetal weight discordance differs over time and in between patient groups, (Fig. 2.) The mixed-effects model seems appropriate to the data as estimates of variance and covariance were all significant (P<0.001). Analysis of normality plots (Q-Q-plot) and Shapiro-Wilk test (P<0,001) of the residuals of the mixed model showed that they are not normally distributed.
For the determination of the onset of weight discordance in twin pairs, the interfetal weight difference of all included twin pairs was compared at the corresponding gestational age. (Fig. 3, Fig. 3a-c).
While CC show a parallel growth course, TTTS and sIUGR show a progressively divergent weight difference starting at the 16th GW, which is statistically significant throughout the pregnancy (Tab. 2 and Fig. 4).
Although statistically significant, the absolute estimated weight discordance in the 16th week of gestation is <100 g. A weight difference to compensate for interobserver variability (>100g) is measurable starting the 20th weeks of gestation (TTTS: 99.86 ± 74.81 g; sIUGR: 93.36 ± 52.61 g). On average the initial time of diagnosis of a TTTS was thus in the 21st week of gestation (20.6 ± 3.9 weeks) and of sIUGR in the 20th week of gestation (19.6 ± 3.4 weeks, P = 0.969), some 4-5 weeks after a statistically significant intertwin estimated weight discordance (Fig. 5).
Comparing the main patient groups of CC, TTTS, EO sIUGR and LO sIUGR (n=154) there were no significant differences regarding conception, delivery mode, maternal age or parity. However, it showed significant differences in complications such as intrauterine fetal demise and prematurity, as well as postnatal parameters such as birth weight, percentiles, and the scope of neo-intensive care (Tab. 3).
Compared to CC, the EO sIUGR group showed a higher proportion of artificial insemination (4,8% vs. 16%; P = 0.061), an increased rate of cervical insufficiency (8.7% vs. 33.8%; P = 0.019), and a need for antenatal corticosteroids (40,5% vs. 65%; P = 0.038). Patients in the EO sIUGR group delivered 2 weeks earlier (36 ± 2 vs. 34 ± 4 weeks; P = 0.003) with lower birth weight (2342 ± 457 g vs. 1735 ± 705 g; P <0.001), APGAR and pH values and a higher need for neonatal intensive care (43.6% vs. 69,2%; P = 0.005) for longer periods (0 vs. 15 days; P = 0.002). Overall, they had an increased mortality rate both intrauterine (7.7% vs. 0%; P = 0.01) and postpartum (11.5% vs. 1.2%; P = 0.015).
Compared to CC, the TTTS group showed increased proportion of nulliparity (69.6% vs. 50.6%, P = 0.062) and artificial reproductive techniques (4,8% vs. 15.2%; P = 0.061), a rate of intrauterine fetal demise (0% vs. 18.2%; P <0.001). As a combination of fetal and maternal indications, the rate of antenatal corticosteroids was higher in the TTTS group (40,5% vs. 85.7%; P = <0.001). They were delivered 3 weeks earlier (36 ± 2 vs. 33 ± 4 weeks; P <0.001) due to intrauterine deterioration such as pathological flow in doppler or pathological cardiotocograph (8,5% vs. 30.8%; P = 0.006). They also had lower birth weights (2342 ± 457g vs. 1824 ± 694g; P <0.001), APGAR scores and a higher need for neonatal intensive care (85.1% vs. 43.6%; P <0.001), for longer periods (0 vs 23 days; P <0.001).
In the TTTS group compared to sIUGR cervical insufficiency was more frequent due to polyhydramnios (33.3% vs. 8.7%; P = 0.034), while preeclampsia and HELLP syndrome are frequent in sIUGR (0% vs. 17.4%; P = 0.018). The TTTS group showed a high proportion of antenatal corticosteroids (85.7% vs. 65.2%; P = 0.086). However, in the sIUGR group, this was more often administered due to fetal indication (sIUGR 80% vs. TTTS 54; P<0,001). The rate of fetal intrauterine demise was higher in the FFTS group (18.2% vs. 7.7%; P = 0.243). Twins with TTTS were also delivered one week earlier with lower APGAR scores and an increased need for neonatal intensive care (85.1% vs. 69.2%; P = 0.077) for longer periods (23 vs. 15 days; P = 0.035).
Comparing EO sIUGR and LO sIUGR, the latter showed cervical insufficiency more often (8,7% vs. 40%, P = 0.032) and had lower birth percentiles (9 vs. 5, P = 0.022). However, the EO sIUGR group had a significantly higher prematurity rate (34 ± 4 weeks vs. at 37 ± 1 weeks, P<0,001). Twins with EO sIUGR also displayed significant differences in the absolute birth weight (1735 ± 705 g vs. 2095 ± 281 g, P = 0.025), body length (42 ± 6.3 cm vs. 45 ± 2.4 cm, P = 0.39 ) and head circumference (29.4 ± 4.1 cm vs. 31.7 ± 1.3 cm, P = 0.015) with significantly lower APGAR scores, higher morbidity and longer neonatal intensive care (15 vs. 5 days, P = 0.094) and higher postnatal mortality (11.5% vs. 0%, P = 0.24). The postnatal pH values hardly differed.
Due to the increased prematurity, antenatal corticosteroids play a significant role in the care of monochorionic twin pregnancies. In the CC group, it was applied in 40.5%, almost exclusively (96.7%) due to maternal indications (cervical insufficiency, premature labor). However, it was significantly more common in the groups of TTTS (85.7%) and sIUGR (65.2%). While in the sIUGR group, almost 80% of the RDS prophylaxis was based on fetal indication, in TTTS, half of it is based on fetal (54%) and a half on maternal indication (45.7%) (polyhydramnios, cervical insufficiency).
To evaluate the relationship between MCA-PSV and EO sIUGR, we compared all CC (n = 84) with the EO sIUGR group (n = 26). From this patient population (n = 110), 32 CC cases and 7 EO sIUGR cases were excluded due to lack of data on MCA-PSV and birth weight, resulting in a study population of 71 patients (52 CC and 19 sIUGR).
The EO sIUGR group showed an increased intertwin difference in MCA-PSV compared to the CC (P = 0.017). The difference in MCA-PSV shows an AUC of 0.67 (95% CI: 0.53-0.84). The optimal cut-off point is 0.2 MoM, with a sensitivity of 74% and a specificity of 64%. (Fig. 6) Based on the cut-off point, twins with an MCA-PSV difference > 0.2 MoM showed a significantly lower birth weight (2109 ± 681g vs. 2305 ± 397 g; P = 0.041), a lower gestational age at birth (35 ± 3 GW vs. 36 ± 2 GW; P = 0.042) and longer neonatal intensive care stays (9.5 vs. 18.9 days, P = 0.012).
The relationship between the MCA-PSV and LO sIUGR was also examined. Of 61 patients enrolled, 49 had a normal birth weight, and 12 had a birth weight <5th percentile. The AUC of the difference in MCA-PSV to predict LO sIUGR is 0.65 (95% CI: 0.48-0.82). It is not significant (P = 0.101). (Fig. 6)
To compare the Type I and Type III (Gratacos) sIUGR, 22 pairs of twins type I and 4 pairs of Type III were examined. Type III twins were born significantly earlier vs. Type I (28 ± 1 vs. 35 ± 4, P <0.001) with lower birth weight (880 ± 223 vs. 1919 ± 632, P <0.001), length (34.3 ± 3.0 vs. 43.7 ± 5.5, P <0.001) and head circumference (23.7 ± 2.0 vs. 30.6 ± 3.3, P <0.001), but higher birth percentiles (22.5 vs. 8, P = 0.021). Besides, Type III showed significantly lower APGAR scores (P <0.001) with neonatal intensive care needed in 69.2% (P = 0.052) with significantly longer stays (91 vs. 8 days, P <0.001). The intrauterine or neonatal mortality showed no significant differences between the groups.
Comparison between artificial insemination and spontaneous conception
The group of artificial insemination had a higher proportion of TTTS (P = 0.061) and sIUGR (P = 0.061). Pooling these pathological pairs of twins showed a significant difference between pregnancies following artificial insemination vs. spontaneous conception (69.2% vs. 38.3%, P = 0.031). Consequently, artificial insemination results in a 3.6-fold increased risk of developing one of the MC complications.