TWIN PREGNANCIES CONCIEVED BY ASSISTED REPRODUCTION – EARLY PREDICTION OF PRETERM BIRTH

Background/Aim. Twins conceived by assisted reproduction techniques (ART) are the most susceptible for perinatal complications. The aim of this study was to examine the role of prenatal noninvasive fetal screening of the first and second trimester in prediction of delivery time of ART conceived twins. Methods. Prospective cohort study of all ART conceived twin pregnancies was conducted at the Clinic for Obstetrics and Gynecology, Clinical Center of Serbia, during the period from January 1, 2016 to December 31, 2017. In the 12th gestational week (GW) twins crown-rump lenght (CRL) and thickness nuchal translucency (NT) were measured ultrasonographically. Moreover, serum levels of beta subunit of human chorionic gonadotropin (? hCG) and pregnancy-associated plasma protein A (PAPPA) were assessed. In the 17th GW twins biparietal diameter (BPD), abdominal circumference (AC) and femur length (FL) were determined ultrasonografically. Additionally, ? hCG, alphafetoprotein (AFP), estriol (E3) and inhibin (INH) were measured in mothers serum. The GW of delivery was noted for each pregnancy. Results. Study included 100 pregnant women with mean age 35.44 ? 5.82. In the examined sample of ART conceived twins significantly more (51%) were delivered in term (? 35 GW) (p = 0.001). Delivery time correlated negatively with NT and first trimester ? hCG serum levels, while it correlated positively with FL of the smaller twin, second trimester ? hCG, AFP and E3 concentrations. According to obtained model for prediction of delivery time in ART conceived twin pregnancies based on first trimester diagnostic tests the significant predictors were PAPPA and ? hCG in the 12th GW as well as NT of the first larger twin. Nevertheless, reliability (sensitivity 50%?75%, specificity 30%?40%) of these diagnostic tests was moderate. Conclusion. Prenatal noninvasive fetal screening of the first and second trimester (ultrasonography and laboratory testing) can be used for prediction of delivery time of ART conceived twins.


ABSTRACT
Background/Aim: Twins concieved by asisted reproduction techniques (ART) are the most susceptible for perinatal complications.The study aim was to examine the role of prenatal noninvasive fetal screening of the first and second trimester in prediction of delivery time of ART conceived twins.
Methods: Prospective cohort study of all ART conceived twin pregnancies was conducted at the Clinic for Ob/Gyn Clinical Center of Serbia, during the period from 01.01.2016.to 31.12.2017.In the 12 th gestational week twins crown-rump lenght (CRL) and nuchal translucency thickness (NT) were measured ultrasonographically.Moreover, serum levels of beta subunit of human horionic gonadotropin (Beta HCG) and pregnancy-associated plasma protein A (PAPPA) were assessed.In the 17 th gestational week twins biparietal diameter (BPD), abdominal circumference (AC) and femur length (FL) were determined ultrasografically.Additionally, Beta HCG, alpha feto protein (AFP), estriol (E3) and inhibin (INH) were measured in mothers serum.The gestational week (GW) of delivery was noted for each pregnancy.
Results: Study included 100 pregnat women with mean age 35.44 +/-5.82.In the examined sample of ART concieved twins significantly more (51%) were delivered in term (≥35 GW) (p=0.001).Delivery time correlated negatively with NT and frst trimester Beta HCG serum levels, while it correlated positively with FL of the smaller twin, second trimester Beta HCG, AFP and E3 concentrations.According to obtained model for prediction of delivery time in ART conceived twin pregnancies based on first trimester diagnostic tests the significant predictors were PAPPA and Beta HCG in the 12 th SAŽETAK Uvod/Cilj: Blizanci nastali tehnikama asistirane reprodukcije (ART) su najosetljiviji za pojavu perinatalnih komplikacija.Cilj studije bio je ispitati ulogu prenatalnog neinvazivnog fetalnog skrininga prvog i drugog trimestra u predviđanju vremena porođaja blizanaca nastalih metodama ART.

INTRODUCTION
Application of assisted reproduction technologies (ART) as infertility treatment, especially in-vitro fertilisation (IVF), intra-cytoplasmic sperm injection (ICSI) and oocyte donation (DO), has lead to increase in the number of twin pregnancies 1,2 .
Despite advances in obstetrical and neonatal care, it is well kown that twin pregnancies still carry a high risk for numerous gestational complications and adverse pregnancy outcomes [1][2][3] .Twin pregnancies are associated with low birth weight of fetuuses, caesarean section, prematurity and associated perinatal morbidity and mortality.Literature data show that in twin pregnancies there is also an increased risk of gestational hypertensive disorders of the mother 2,3 .Nevertheless, there are conflicting reports on whether conception by ART further increases these risks when compared to spontaneously conceived pregnancies 4,5 .Pregnancies concieved by ART are found to be more often complicated with preterm delivery and fetal low birth weigh, placental abruption, gestational diabetes and hypertension [2][3][4] .Consequently, ART concieved twins are the most susceptible various perinatal complications among which the most severe is premature pregnancy ending 6 .
One of the most important issues in perinatology is early and non-invasive prediction of adverse pregnacy outcomes 1,6 .Therfore, the aim of this study was to examine the role of prenatal noninvasive fetal screening of the first and second trimester in prediction of delivery time of ART conceived twins.

METHODS
A prospective cohort study of all pregnant women with ART conceived twin pregnancies was conducted at the Clinic for Gynecology and Obstetrics Clinical Center of Serbia in Belgrade, during the period from January 1 st 2016.to December 31 2017.
Exclusion criteria for this study were miscarriage before 8 th gestational week, genetic disorders of twins, monochorionicity and monoamnionicity of twins and existence of and chronic ilnesses in mothers that could influence the pregnancy course and outcome.
The study was approved by the Institutional review board.All patients signed a written informed consent form.
Already during the first examination for pregnancy confirmation, a complete history regarding mothers age, chronic ilnesses, previous parity and ART type (IVF, ICS, DO) was taken from each patient.Investigated women were regularly checked-up at least once per trimester in our Clinic.
In the first trimester, during the 12 th gestational week, for each twin, the crownrump lenght (CRL) and the thickness of nuchal translucency (NT) were measured and chorionicity determined ultrasonographically.Moreover, biochemical markers that are incorporated in Double test screening such as beta subunit of human horionic gonadotropin (Beta HCG) and pregnancy-associated plasma protein A (PAPPA) were assessed.In the 17 th gestational week, ultrasound examination for both twins determined their biometric parameters such as biparietal diameter (BPD), abdominal circumference (AC) and femur length (FL).Additionally, in the second trimester we evaluated markers used for Triple testing i.e.Beta HCG, alpha feto protein (AFP), estriol (E3) and inhibin (INH).
Monitoring of twins was performed by ultrasound examinations.Ultrasound biometry and monitoring were performed through ACCUVIX device (Samsung Medison, Seoul, Sought Korea), with 3.75 MHz abdominal and vaginal probe.
For laboratory diagnostic tests 5 to 10 milliliters of blood was drawn by venipuncture into nonheparinized tubes and centrifuged for 15 minutes.After serum separation concentrations of investigated biochemical markers were measured using BRAHMS KRYPTOR analyzer and applying fluorocytometric immunoassay with SsdwLab 5 software.The measured serum concentrations (IU/L) of biomarkers were then converted into multiples of median (MoM) and adjusted for gestational week for easier comparison and analysis.The standard referral range of all tested pregnancy markers, most widely used in literature and adopted in this study as well, is from 0.5 to 2 MoM.
All complications of pregnancy were registered regularly throughout pregnancy.
The gestational week (GW) in which delivery occurred as well as the delivary type (vaginal or cesarean section -SC) were noted for each pregnancy.Time of delivery was classified as miscariage (before 12 GW), early pretem (before 25 th GW), late pretem (before 35 GW) and term delivery (in and after the 35 th GW).
Upon delivery, all data collected throughout pregnancy were compared with delivery time and analyzed by methods of descriptive number, percent, mean, standard deviation) and analytical statistics using the SPSS 20 software.Correlations of biochemical parameters and occurrence of hypertension were tested using Spearman correlation.Significance of differences between categories of assessed parameters was examined by χ 2 test.
Enter method of binary logistic regression was applied to construct models for prediction of delivery time of ART concieved twins based on investigated ultrasonographic and biochemical markers of the first and second pregnancy trimester.All models were adjusted for potential confoundings (mothers' age, and parity).
Finaly, study authors performed the ROC (Receiver Operating Characteristics) analysis to set the cut-off values of ultrasonographic and biochemical parameters of the first and second trimester that could imply on the delivery time in ART concieved twin pregnancies.
In 88% of cases investigated women did not have previous deliveries, while 29% of them had previous miscarriages and/or abortions.Significantly more (59%) women had no gestational illnesses and pregnancy complications (p=0.001).The most common pregnancy complication was gestational hypertension (21% of mothers).In our sample there were no significant differences in frequency of different ART types (53% had only IVF, while 42% had IVF/ICSI; p=0.259).Still, only five women had oocyte donation.
In the examined sample of ART concieved twins significantly more (51%) were delivered in term (after the 35 gestational week) (p=0.001).We registered 8 cases of miscarriage before the 12 th gestational week, 8 early preterm births before 24 th gestational week, while 33 twins were born between 24 th and 33 th gestational week (p=0.001).
Majority of twins were liveborn (86% of first larger and 87% of second smaller twins).
Delivery was mostly by Caesarean Section (p=0.001).
On the Table 1 we presented diagnostic parameters (ultrasonografy and laboratory) of the first and second pregnancy trimester.Majority of twins had all values in the refferal range for gestational week.
Correlations of delivery time with general patients data and first trimester diagnostic testing are presented on the Table 2, while correlations with the second trimester ultrasound and laboratory findings are presented on the Table 3.  4.
The significant predictors were Beta HCG in the 12 th , NT of the first larger twin in the 12 th gestational week as well as CRL in the 12 th and FL in the 17 th gestational week of the second smaller twin.

DISCUSSION
Twin pregnancies in general are at higher risk of numerous pregnancy complications including miscariage or preterm birth caused by a development of two fetuses instead of one 1,2 .Adverse obstetric outcome after ART might be caused by the very procedure 4 .Moreover, ART concieved twins can be at even higher risk of pregnancy complications and adverse obstetric outcomes due to maternal age, pre-existing maternal medical conditions such as polycystic ovary syndrome or thyroid diseases that are associated with miscariages and aneuploidies [7][8][9] .
Studies show that the mean gestational age at delivery was significantly lower in the ART group compared with the spontaneous conception group regardless of the cerclage application 3 .ART conceived twins are at significantly increased risks of preterm birth both early and late, althouhg according to majority of studes ART twins were mostly delivered moderately preterm (32-34 gestational weeks) 2,10 .However, the rate of preterm birth of ART concieved twins in several investigations was similar to that of spontaneously conceived ones or that very preterm deliveries even occurred more frequently in spontaneous twin pregnancies 4 .In our study, simillarly, more women delivered at term than preterm (51 vs 49 % respectively), while late preterm birt was predominant time of delivery of preterm twins (33%).Moreover, although somewhat unexpectidely, older mothers delivered at term more often.A possible explanation could be that older women who had ART as infertility treatment are more concerened for the pregnancy sucess and therefore comply better to their therapy and have more regular check-ups that could have prevented preterm labor.
Consequently, ART twin pregnancies require additional antenatal care. 7Potential early prediction of advrese outcomes could enable adequate and timely management of pregnancy complications allowing preventing premature pregnancy ending 11 .Prenatal noninvasive fetal screening in patients with ART conceived twin pregnancies is essential part of appropriate control of these high-risk pregnancies.It is performed at the time of standard fetal screening for chromosomal anomalies (Double, Triple or Q test) by ultrasound examination of both fetuses and the analysis of mothers biomarkers (hormones) serum levels has been established during the first (11 to 14 gestational weeks) and/or second (16 to 18 gestational weeks) pregnancy trimester.Literature data show that regular and thorough check-ups of twins can show apart from potential risk of aneuploidies also the risk of adverse pregnancy outcomes 11 .
Data from current literature are still contradictory regarding the strength of association between pregnancy biochemical markers and twin pregnancy outcomes 12,13 .Biochemical markers are associated with a variety of external factors, such as patient's and partner's age, demographic data, course and outcome of previous pregnancies, especially premature births, spontaneous abortion, perinatal morbidity and mortality, type of performed ART (IVF, ICSI, donation of oocyte), number of previous unsuccessful ART attempts, the course of current pregnancy and the presence of symptoms that indicate imminent abortion or premature delivery 12,13 .There is still an on going debate how could serum biomarkers be used in the prediction of pregnancy course and outcome.It has been found that in the case of twin pregnancies the values of biomarkers depend on the horionicity as well as the gestational age 6 .Studies that evaluated Beta HCG serum levels for prediction of twins preterm birth found that reliability and predictive performance of high Beta HCG levels was low 11,14 .Similar findings were reported for PAPPA serum levels (senisitivity mostly below 50%).Nevertheless, specificity of PAPPA as well as first and second trimester Beta HCG serum concentrations were significantly higher reaching above 90% in some cases for prediction of twins premature birth 11,14 .Although several studies found a negative correlation between elevated maternal serum AFP levels and twins gestational age at birth predictive ability of elevated maternal AFP levels was poor (sensitivity of 30%, specificity of 88%) 11,14 .
The role of ultrasound in control of ART concieved twin pregnancies is essential as it for diagnosing chorionicity and placentation, discordinant growth, twin-to-twin transfusion syndrome 15 .Both early first and second trimester sonographic examinations are investigated in recent studies for predicting adverse perinatal outcomes 16 .It is postulated that fetuses suffering from reduced oxygenation and/or nutrient limitation caused by progressed pregnancy complications during early pregnancy tend to be smaller when their biometrical measurements are evaluated 17,18 .Several studies have examined the relationship between the parameters obtained during ultrasonographic surveillance of twins with the course, delivery time and type as well as the outcome of both dichorionic and monochorionic twin pregnancies 5,19 .According to literature data significant correlations of parameters measured and determined during the ultrasound screening such as inter-twin differences in CRL and NT, horionicity, amniotic fluid quantity, abdomeninal circumference of both fetuses in the second trimester with the pregnancy outcome and numerous gestational complications are commonly found 15,18 .According to some investigations abnormal measurements of CRL in the first trimester were found to be associated with early fetal loss 10,20 .A meta-analysis showed that twin pregnancies with CRL discordance ≥10% were at significantly higher risk of fetal loss and premature delivery after the 24 th until 34 th gestational week, but not before 24 weeks of gestation 21 .
Only few studies have assessed nuchal translucency in predicting preterm birth in twin gestations.Still, they reported sensitivity of 75% and specificity of 94% in prediction of preterm birth of twins 11,22 .Inter-twin AC discordances were proven by difrferent authors to reliably imply on of adverse obstetric outcomes.A single biometric assessment of twins at 16 weeks could adequately predict subsequent pregnancy complications and premature pregnancy ending in almost 50% of cases 23,24 .On the other hand, high reliability of biometrical measurements of the first and second trimester was not confirmed in all available studeis.Moreover, cut-off values of ultrasonographically determined twins biomety vary among studies.Further studies on this matter asre still needed 20,21 .
In our study we found that out of assessed ultrasonographic fetal measures the thickness of nuchal translucence of both twins and femur lenght of the smaller second twin in the second trimester could be used for prediction of preterm birth of ART concieved twins.When laboratory analyses were assessed Beta HCG serum levels measured in the first and the second trimester, as well as PAPPA and AFP concentrations could imply on the gestational week of ART concieved twins delivery.We obtained a significant model for prediction of delivery time in ART conceived twin pregnancies based on first trimester diagnostic testsultrasound and laboratory.Moreover, we set the cut-of levels for ultrasonografic and laboratory diagnostic tests of first and second trimester.Performed ROC anlysis showed that sensitivity was better than specificity for prediction of delivery time of ART concieved twins.Nevertheless, reliability of these diagnostic testst in our study was not high (sensitivity 50 to 75%, specificity 30 to 40%), similarly as in the literature.
Consequently it can be seen that in the examined sample of one hundred ART concieved twins significantly more (51%) were delivered in term (after the 35 gestational week), while more than 80% of them were liveborn.Based on the results of our study it can be concluded that prenatal noninvasive fetal screening of the first and second trimester (ultrasonografy and laboratory testing) can be used for prediction of delivery time of ART conceived twins.The significant predictors were PAPPA and Beta HCG of the first trimester, larger twin NT as well as smaller twin CRL and FL.Nevertheless, reliability of these diagnostic testst in our study was moderate.Therefore, further studies are needed to more thoroughly investigate how could we potentially increase the reliability of these predictors as well as to test the obtained prediction model for delivery time of ART concieved twins based on ultrasonografy and laboratory diagnostic tests of the first and second trimester.
Delivery time correlated positively with mothers age and negatively with previous parity.Out of all assessed ultrasonographic fetal measures in the first trimester only the thickness of nuchal translucence for both twins was negatively correlated with the delivery time.Moreover, femur lenght of the smaller second twin measured ultrasonographically in the second trimester correlated positively with the delivery time, but was not correlated with the exact gestational week of delivery.When laboratory analyses were assessed it was determined that Beta HCG serum levels in the first trimester correlated positively with gestational week of delivery.Contrary, Beta HCG as well as AFP concentrations in the second trimester correlated negatively with the exact gestational week of delivery.Serum levels of Estriol in the second trimester correlated positively with term delivery.We obtained a significant model for prediction of delivery time in ART conceived twin pregnancies based on first trimester diagnostic testsultrasound and laboratory (B=0.041;Wald=0.140;Nagelkerke R 2 =0.628; χ 2 =38.664; p=0.001; explained variance=73.0%).The significant predictors were PAPPA and Beta HCG in the 12 th gestational week as well as NT of the first larger twin.Mothers age and previous parity were confoundings in this model.ART twins term delivery = 3.054 -1.142 x PAPPA 12 gw -1.263 x Beta HCG 12gw -0.940 x NT gemelus I -2.128 x previous parity + 0.104 x mothers age We did not manage to obtain the significant binary logistic regression for prediction of delivery time in ART conceived twin pregnancies based on second trimester diagnostic tests (χ 2 =22.355; p=0.217).ROC analysis of ultrasonografic diagnostic tests is presented on the Chart 1 while on the Chart 2 we presented laboratory tests of first and second trimester.Newly established cut off values with their sensitivity and specificity for potential ultrasonographic and laboratory predictors of twins delivery time is presented on Table

Table 1 .
Ultrasound and laboratory diagnostic findings of first and second trimester

Table 2 .
Correlation of delivery time with general patients data and first trimester diagnostic testing CRLcrown rump lenght; NTnuchal translucence; HCGhuman chorionic gonadothropin; PAPPApregnancy associated plazma protein A; ART type -IVF, ICSI, DO

Table 4 .
ROC analyzis of potential predictors for ART concieved twins delivery time