Of the 1175 TwPs followed between January 2010 and December 2022 in our center, the dataset included 572 TwPs that met the inclusion criteria: 450 (78.7%) DC and 122 (21.3%) MC. Of these, 464 completed first-trimester serum biochemical screening, 390 underwent UtA-PI evaluation, and 371 had complete data with all biomarkers. Maternal and pregnancy characteristics stratified per birthweight percentiles are summarized in Table 1.
Overall, pregnancies affected with one or both infants classified as SGA <3rd, <5th or <10th percentiles were 120/572 (20.9%), 157/572 (27.4%) and 190/572 (33.2%), respectively (Table 1). Newborns classified as SGA <3rd, <5th and <10th percentiles accounted for 145/1142 (12.7%), 191/1142 (16.7%) and 247/1142 (21.6%), respectively.
Considering maternal and pregnancy characteristics, MC pregnancies exhibited similar incidence of SGA <3rd and <10th percentile compared to DC: 22.1% vs 20.7%, OR 0.9 (95%CI: 0.5-1.5, p = 0.725) and 39.3% vs 31.6%, OR 0.7 (95%CI: 0.5-1.0, p = 0.105), respectively (Table 1). Women with low body mass index (BMI) (<20Kg/m2) showed increased odds of SGA <3rd and <10th percentiles, 31.1% vs 19.5%, OR 1.8 (95%CI: 1.0-3.1, p=0.022), and 51.4% vs 30.5% OR 2.4 (95%CI: 1.4-3.9, p<0.001), respectively. Furthermore, women who reported smoking habits in the first trimester had increased odds of SGA in the subgroup >3rd <10th percentile, 24.1% vs 11.0%, OR 2.5 (95%CI: 1.3-5.0, p=0.005), but not to SGA <3rd percentile: 20.4% vs 21.0%, OR 0.9 (95%CI: 0.5-1.9, p=0.908) (Table 1).
Within parous women, those with a previous history of SGA had increased odds of SGA <3rd and <10th percentile, 53.8% vs 16.7%, OR 5.8 (95%CI: 1.8-18.4, p=0.004) and 61.5% vs 25.8%, OR 4.6 (95%CI: 1.4-14.7, p=0.009). Additionally, women under aspirin prophylaxis revealed increased odds for SGA <5th percentile, 39.7% vs 25.7%, OR 1.9 (95%CI: 1.1-3.1, p=0.012) (Supplementary Table 1).
In univariable analysis, women with high UtA-PI ≥95th percentile (TwPs’ references) showed and increased odds of SGA <3rd and SGA <5th percentile, 54.2% vs 21.0 %, OR 4.4 (95%CI: 1.9-10.2, p<0.001) and 58.3% vs 26.8%, OR 3.8 (95%CI: 1.6-8.9, p<0.001), respectively (Table 2 and Supplementary Table 2). Additionally, women with low PAPP-A MoM (≤10th percentile, corresponding to 0.50 MoM) also had increased odds of SGA <3rd percentile, 34.8% vs 19.2% OR 2.2 (95%CI: 1.1-4.3, p<0.014) (Table 2).
The incidence of SGA <3rd or <10th percentile concurrent with PTB <32 weeks was found to be 4.0% and 4.9%, respectively, with more than half (71.8%) of premature births <32 weeks being associated with the presence of SGA <10th percentile in one or both twins. Before 32 weeks, SGA <3rd was present in 23 (59.0%) vs 97 (18.2%) cases, OR 6.4 (95%CI: 3.2-12.7, p<0.001); <34 weeks, in 40 (48.8%) vs 80 (16.3%) cases, OR 4.8 (95%CI: 2.9-8.0, p<0.001) and <36 weeks, in 73 (37.4%) cases vs 47 (12.5%) cases, OR 4.2 (95%CI: 2.7-6.4, p<0.001). Iatrogenic PTB <32, 34 and 36 weeks occurred in 46.2% (50.0% MC and 45.5% DC), 40.2% (47.7% MC and 38.1% DC) and 47.7% (62.5% MC and 41.7% DC), respectively. Among women with iatrogenic PTB <34 weeks, 30 (90.9%) had prenatal suspicion of FGR and 31 (93.9 %) had one or both SGA <10th percentile. Conversely, the incidence found of early onset-PE/HELLP <34 weeks was 1.0% (unshown data).
Two single fetal demises occurred in two DC pregnancies, one at 25 weeks and another at 36 weeks. Neither fetus was suspected to be growth-restricted, although in the latter case, the surviving co-twin was SGA <3rd percentile. Both neonatal and perinatal death rates were higher in the groups affected by SGA <5th and <3rd percentile, with observed incidences of 1.3% and 1.6%, and 1.7% and 2.1%, respectively; however without reaching statistical significance. Additionally, these groups experienced prolonged stays (≥ 8 days) in the neonatal care unit, particularly pronounced in MC twins, totaling 65.7% and 70.3%, p<0.001 in SGA <5th and <3rd percentiles, respectively (Table 3 and Supplementary Table 3).
In the multivariable LR analysis, the best association model was obtained for one or both SGA <3rd percentile concurrent with to PTB <32 weeks, with an AUC 0.765, a sensitivity rate of 70% and a false positive rate of 20% (Table 4). Lowering the false positive rate to 10% resulted in a decrease in the detection rate to approximately 35%. We also performed a similar LR model with the same covariates but using UtA-PI references in singletons of our population and observed that the sensitivity lowers to 65% and 29% for the same false positive rates considered (unshown data).
UtA-PI and PAPP-A were important independent risk factors found to be associated with SGA as well the composite outcome of SGA concurrent with PTB. The highest odds (OR 6.4, 95%IC 2.5-16.4, p<0.001) were observed in women with UtA-PI ≥ 95th percentile for SGA <3rd percentile concurrent with PTB <36 weeks (Table 4). In our cohort, there was no association between β-hCG levels and any of the evaluated outcomes (Table 2).