Concordance Rates of Birth Defects After Assisted Reproductive Technology Among 17 258 Japanese Twin Pregnancies: A Nationwide Survey, 2004–2009

Background Most twins after assisted reproductive technology (ART) are dizygotic. Analysis of dizygotic twin pairs is useful in assessing familial aggregation in the development of birth defects. Methods Using nationwide post-ART data from the Japan Society of Obstetrics and Gynecology, recurrence risk ratios (RRRs)—defined as probandwise concordance rates of birth defects in twins divided by the prevalence of birth defects in the general population—were calculated as indicators of familial aggregation. Birth defects were then reclassified according to the ICD-10 categories corresponding to codes Q00–Q99. From 2004 to 2009, there were 17 258 twin pregnancies. Results At least 1 birth defect was noted in 236 twin pairs: 11 concordant and 225 discordant pairs. Regarding major organ systems, high probandwise concordance rates were observed for congenital malformations of eye, ear, face, and neck (11.8%), cleft lip and cleft palate (10.5%), congenital malformations of the nervous system (9.8%), and other congenital malformations of the digestive system (9.5%). High RRRs were observed for congenital malformations of eye, ear, face, and neck (RRR = 233), specifically other congenital malformations of the ear (RRR = 449); congenital malformations of the great arteries (RRR = 235), specifically those of the patent ductus arteriosus (RRR = 530); and for cleft lip and cleft palate (RRR = 208), specifically cleft palate with cleft lip (RRR = 609). The probandwise concordance rate of any birth defect (8.9%) was nearly identical to the approximated recurrence risk of sib-pairs (8.8%), which assumed multifactorial inheritance. Conclusions The present findings suggest that familial aggregation is a factor in some birth defects.


INTRODUCTION
According to data on assisted reproductive technology (ART) and vital statistics in Japan, the percentage of ART live births was 2.49% (26 680/1 070 035) in 2009, which indicates that ART is becoming widespread in Japan. 1 To date, most population-based epidemiologic studies of twinning and birth defects view twins as individuals, not twin pairs.Zygosity determination of same-sex twin pairs is rarely performed at birth, and same-sex pairs are often regarded as monozygotic (MZ) twin pairs.This assumption necessarily underestimates the resemblance of MZ pairs according to the proportion of dizygotic (DZ) pairs.Given these circumstances, ART data present a unique opportunity for twin studies, as most twins after ART are DZ.
The first step in genetic epidemiologic analyses is to clarify familial aggregation of targeted traits.To identify familial aggregation, it is important to compare the concordance rate of birth defects in DZ twin pairs (ie, siblings that develop together in the same womb) 2,3 with the prevalence of birth defects in the general population.The present study used nationwide data on ART to calculate the concordance rate of twin pairs and examine familial aggregation of birth defects.

Outline of Japanese ART data
The method for collecting data has been described elsewhere. 4lmost all medical institutions that perform ART are registered with the Japan Society of Obstetrics and Gynecology (JSOG).
Starting in 2004, an annual list of all ART pregnancies resulting in birth defects has been presented in the JSOG annual ART reports (in Japanese).The author used these case report data from 2004-2009 as initial information.The items included are ART method, blastocyst transfer, maternal age, perinatal outcome and gestational week, plurality, sex, early neonatal infant death up to day 6, and disease name.Within the study period, there were 159 451 singleton pregnancies, 17 258 twin pregnancies, and 839 triplet/+ pregnancies.
Birth defects were reclassified according to the International Classification of Diseases, 10th edition (ICD-10), 2003 version.Diseases that were classified in the categories corresponding to ICD-10 codes Q00-Q99 (ie, congenital malformations, deformations, and chromosomal abnormalities) were selected and analyzed.In total, 1502 abortions, stillbirths, and live births with birth defects were included.
The present author paired twins, using information on birth year, maternal age, gestational weeks, ART method, blastocyst implantation, and plurality.Other information on twin status was also considered; for example, firstand second-born twins were clearly described and listed.

Statistical analyses
All concordant pairs were listed with their demographic data and neonatal outcome.The pairwise and probandwise concordance rates 5 were then calculated for each major organ system category, each subcategory, and, in some cases, each disease.
In the present study, the terms "concordant pair" and "discordant pair" are used to describe the disease condition of a given twin pair.The pairwise concordance rate is the probability that both members of a twin pair are affected if at least 1 member of the pair is affected.The probandwise concordance rate is the probability that a twin is affected if his/ her co-twin is affected.Only probandwise concordance rates can be directly compared with risk rates reported for other familial pairings and with population prevalence figures. 5airwise concordance rates were calculated as C/(C + D), and probandwise concordance rates as 2 × C/(2 × C + D), where C denotes the number of affected concordant pairs and D denotes the number of discordant pairs. 5ecurrence risk ratios (RRRs) 6 were used as indicators of familial aggregation of birth defects and were calculated as the ratio of the risk of disease manifestation (which, given that one's relative is affected, correspond to the probandwise concordance rate of twin pairs in the present study) to the disease prevalence in the general population.
Moreover, the author estimated the recurrence risk of DZ pairs, which have the same genetic resemblance as sib-pairs.According to Edwards, 7 if a targeted disease is determined by multifactorial inheritance, its frequency in sib-pairs or DZ twin pairs approximates the square root of disease prevalence in the general population.Thus, the present study compared the probandwise concordance rate of any birth defect in twin pairs with the estimated recurrence risk in sib-pairs and DZ pairs.

RESULTS
From among 247 twins with any birth defect, a total of 236 twin pairs were identified: 11 concordant and 225 discordant pairs.Thus, 1.4% (236/17 258) of twin pairs (pregnancies) had at least 1 affected member.
Demographic and perinatal outcome data for all concordant pairs are presented in Table 1.They comprise 6 male-male, 1 female-female, and 4 opposite-sex pairs.Two of the 11 pairs were stillbirth-stillbirth.The records for 6 of the 9 live-birth pairs showed no early neonatal infant death, although the outcome of the other 3 pairs is not known.
The calculated concordance rates and RRRs are shown in Table 2. Regarding classification by major organ system, relatively high probandwise concordance rates were observed for congenital malformations of eye, ear, face, and neck (11.8%), cleft lip and cleft palate (10.5%), congenital malformations of the nervous system (9.8%), and other congenital malformations of the digestive system (9.5%).
The probandwise concordance rate of any birth defect was 8.9%, which was nearly identical to the estimated recurrence risk among sib-pairs and DZ pairs, namely, 8.8% (the square root of 0.77, see Table 2).

Method of analysis
Correct zygosity diagnosis is rare in most studies of birth defects.Researchers have often compared resemblance between same-sex pairs (as a proxy for MZ pairs) and opposite-sex DZ pairs, on the assumption that there is no sex difference in prevalence.In the present study, information was obtained only from probands.In such a situation, the probandwise concordance rate is the second-best measure of resemblance in twin pairs.
Although most subjects in the present study were DZ pairs, some MZ pairs may well have also been included.According to a recent meta-analysis by Vitthala et al, 8 the incidence of MZ twins after ART is 2.25 times that after natural conception.Caution is warranted in interpreting these values because most previous studies used the pairwise rather than the probandwise concordance rate.Ascertainment bias in the identification of twin pairs would be small in the present sample, since birth defects during pregnancy or the neonatal period (at the latest) were reported in the same hospital.However, this ascertainment period may underestimate the  M: male, F: female; OS: opposite-sex; LB: live birth, SB: stillbirth; N: no, U: unknown.
See Table 2 for more information on classification codes.

Birth defects in twin pairs
The number of concordant pairs was clearly higher for male-male pairs than for female-female and opposite-sex pairs.No previous study reported that concordance rates differed according to the sex combination of twin pairs.
The concordance rates of birth defects in twin pairs, as determined in previous large studies, are shown in Table 3.The concordance rate for any birth defect is higher in MZ pairs and all twin pairs as compared with DZ pairs and oppositesex DZ pairs, [9][10][11][12] respectively, which suggests a genetic contribution to birth defects.The probandwise concordance rate of any birth defect (8.9%) was nearly identical to the estimated recurrence risk among sib-pairs (8.8%) and much higher than the prevalence in the general population (0.77%).These results suggest familial aggregation of birth defects and that the origin of this aggregation is multifactorial inheritance.
The probandwise concordance rate of congenital malformations of the circulatory system was 30-fold higher than the prevalence in the general population.Kenna et al 13 found 2 concordant pairs out of 15 MZ pairs and 2 concordant pairs with different heart lesions out of DZ 12 pairs.According to Hardin et al, 14 the probandwise concordance rate for oppositesex DZ twin pairs was 14.0%.A small number of studies directly compared the recurrence risk of twin pairs with that of first-degree relatives.Caputo et al 2 compared recurrence risk in DZ pairs and sib-pairs and concluded that the higher recurrence and concordance of congenital heart disease in DZ twins might depend on a poorly understood environmental risk during pregnancy.However, Øyen et al 15 found that intrauterine conditions had no effect, after comparing the RRRs of opposite-sex twin pairs and first-degree relatives.
7][18] Layde et al 19 found that a high concordance rate was observed in same-sex pairs, which suggests both a strong genetic component to the etiology of PDA and high rates of prematurity in twin pairs.The present finding of a high concordance rate and RRR for PDA supports the genetic/ shared environmental hypothesis.When concordance rates were calculated after excluding PDA cases, the results were not markedly different, as shown in Table 2.
One male-male concordant pair showed both micrognathia and low-set ear (subcategory Q17).There is no populationbased twin study of these defects.
There was one male-male concordant pair who had cleft palate with cleft lip (subcategory Q37), with a 15.4% probandwise concordance rate.This value is similar to that of DZ pairs (16.7%), as reported by Lin et al. 26 According to Grosen et al, 3 the probandwise concordance rate for cleft lip/cleft palate was higher for MZ pairs than for DZ pairs.However, they also found that the recurrence risk for both types of clefts was greater in DZ twins than in nontwin siblings, suggesting intrauterine environmental effects on these traits.The fact that the RRR for cleft palate with cleft lip was highest in the present study supports their results.
There was no concordant pair with esophageal atresia (subcategory Q39).David and O'Callaghan 27 found that the probandwise concordance rate for this condition was 19.0%, although Orford et al 28 reported a low pairwise concordance rate (4.1%).There was 1 concordant opposite-sex DZ pair of imperforate anus (subcategory Q42).Källén 17 reported that for alimentary atresia, including imperforate anus, concordance was rather common among same-sex pairs.
There was 1 concordant opposite-sex DZ pair with brachymelia (subcategory Q73).Métneki et al 29 studied the occurrence of congenital limb reduction defects in twins and concluded that genetic factors have a limited role in pathogenesis.Pober et al 30 conducted a large twin study of Bochdalek diaphragmatic hernia, including 8 twin cases with no concordant pairs.The concordance rate of congenital malformations and deformations of the musculoskeletal system was low in the present study, in contrast to the relatively high prevalence among the general population.

Limitations
Most limitations of this study are due to deficiencies in the data collection system.The first limitation was the lack of a Most probandwise concordance rates were recalculated by using the number of concordant and discordant pairs in the literature.MZ: monozygotic, DZ: dizygotic, OS: opposite-sex a blood type and placenta, b maternal report, c placentation and chorionicity, d blood type, chorionicity, physical characteristics, e placenta and reported type, f blood type, physical resemblance, chorionicity, g questionnaire.
zygosity classification for same-sex pairs.Second, pairing of twins was not necessarily complete.Some concordant pairs might not have been real twin pairs.Third, the statistical power of the analyses was limited.The present concordance rates might be strongly influenced by chance factors, since most disease concordance rates were calculated on the basis of no or 1 concordant pair.

Conclusions
The present results provide a good overview of concordance rates among twin pairs with birth defects after ART.Strong familial aggregation was observed for some birth defects.
) Ventricular septal defect Q22 Congenital malformations of pulmonary and tricuspid valves Q27 Other congenital malformations of peripheral vascular system Order of pairs does not necessarily reflect birth order (ie, first-and second-born).

Table 1 .
List of concordant pairs

Table 2 .
Concordance rates in twin pairs and recurrence risk ratios (RRRs) for birth defects Concordance rates and RRRs with their 95% CI were calculated for disease classifications that had >10 total concordant/discordant twin pairs and a concordance rate not equal to 0.
a Only twin pairs with at least 1 affected member are listed.b Total prevalence was calculated per fetuses/neonates.c

Table 3 .
Concordance rates of birth defects in previously published twin studies