Multiple pregnancy usually results from the ovulation and subsequent fertilization of more than one oocyte. In such cases, the fetuses are genetically different (polyzygotic or non-identical). Multiple pregnancy can also result from the splitting of one embryonic mass to form two or more genetically identical fetuses (monozygotic).

In all cases of polyzygotic multiple pregnancy, each zygote develops its own amnion, chorion and placenta (polychorionic). In monozygotic pregnancies, there may be sharing of the same placenta (monochorionic), amniotic sac (monoamniotic) or even fetal organs (conjoined or Siamese). When the single embryonic mass splits into two within 3 days of fertilization, which occurs in one-third of monozygotic twins, each fetus has its own amniotic sac and placenta (diamniotic and dichorionic) (Figure 1). When embryonic splitting occurs after the third day following fertilization, there are vascular communications within the two placental circulations (mono chorionic). Embryonic splitting after the 9th day following fertilization results in monoamniotic, monochorionic twins and splitting after the 12th day results in conjoined twins.

Twins account for about 1% of all pregnancies, with two-thirds being dizygotic and one-third monozygotic. The incidence of dizygotic twins varies with ethnic group (up to 5 times higher in certain parts of Africa and half as high in parts of Asia), maternal age (2% at 35 years), parity (2% after four pregnancies) and method of conception (20% with ovulation induction). The incidence of monozygotic twins is similar in all ethnic groups and does not vary with maternal age or parity, but may be 2–3 times higher following in vitro fertilization procedures, possibly because with these methods the architecture of the zona pellucida is altered^1,2.

In the last 20 years, the rate of twinning has increased (Figure 2). The increase in dizygotic twins is mainly due to the widespread use of assisted reproductive techniques and the increasing maternal age. There has also been an increase in the rate of mono zygotic twinning, particularly in those countries in which there is widespread use of oral contraceptives.

Zygosity can only be determined by DNA fingerprinting. Prenatally, such testing would require an invasive procedure to sample amniotic fluid (amniocentesis), placental tissue (chorionic villus sampling) or fetal blood (cordocentesis).

Determination of chorionicity can be performed by ultrasonography and relies on the assessment of fetal gender, number of placentas and characteristics of the membrane between the two amniotic sacs. Different-sex twins are dizygotic and therefore dichorionic, but in about two-thirds of twin pregnancies the fetuses are of the same sex and these may be either monozygotic or dizygotic. Similarly, if there are two separate placentas, the pregnancy is dichorionic, but, in the majority of cases, the two placentas are adjacent to each other and there are often difficulties in distinguishing between dichorionic-fused and monochorionic placentas.

In dichorionic twins, the intertwin membrane is composed of a central layer of chorionic tissue sandwiched between two layers of amnion, whereas in monochorionic twins there is no chorionic layer. Consequently, the intertwin membrane tends to be thicker and more echogenic in dichorionic than monochorionic pregnancies, but this is a subjective and quite unreliable finding. For example, one study reported that dichorionicity is associated with an inter-twin septum thickness of 2 mm or more⁴, but the reproducibility of this measurement was poor and is dependent on such technical aspects as the angle of insonation and gestational age⁵.

The best way to determine chorionicity is by an ultrasound examination at 6–9 weeks of gestation, when in dichorionic twins there is a thick septum between the chorionic sacs (Figure 3)^6–8. After 9 weeks, this septum becomes progressively thinner to form the chorionic component of the intertwin membrane, but it remains thick and easy to identify at the base of the membrane as a triangular tissue projection, or lambda sign^9–11.

At 11–14 weeks of gestation, sonographic examination of the base of the inter-twin membrane for the presence or absence of the lambda sign (Figure 4) provides reliable distinction between dichorionic and monochorionic pregnancies. In an ultrasound study of 368 twin pregnancies at 10–14 weeks of gestation, pregnancies were classified as monochorionic if there was a single placental mass in the absence of the lambda sign at the inter-twin membrane–placental junction, and dichorionic if there was a single placental mass but the lambda sign was present or the placentas were not adjacent to each other¹¹. In 81 (22%) cases, the pregnancies were classified as monochorionic and in 287 (78%) as dichorionic. All pregnancies classified as monochorionic resulted in the delivery of same-sex twins and all different-sex pairs were correctly classified as dichorionic¹¹. 

The perinatal mortality rate in twins is around 6 times higher than in singletons^13–17. This increased mortality, which is mainly due to prematurity-related complications, is higher in monochorionic than dichorionic twin pregnancies. In monochorionic twins, an additional complication to prematurity is twin-to-twin transfusion syndrome. Thus, retrospective studies in which both zygosity and chorionicity were determined after birth reported that the perinatal mortality rate is about 3–4 times higher in mono chorionic compared to dichorionic twins, regardless of zygosity^18,19.

A prospective study, in which chorionicity was assessed by ultrasound examination at 10–14 weeks of gestation, compared pregnancy outcome in 102 monochorionic and 365 dichorionic twin pregnancies²⁰. There was at least one fetal loss before 24 weeks of gestation in 12.7% of monochorionic and 2.5% of dichorionic pregnancies. Additionally, there was at least one perinatal loss (at or after 24 weeks) in 4.9% of monochorionic and 2.8% of dichorionic pregnancies²⁰.

This study confirmed that perinatal mortality in twins, especially those that are monochorionic, is higher than in singleton pregnancies. However, perinatal statistics underestimate the importance of monochorionic placentation to fetal death since the highest rate of mortality is before 24 weeks of gestation (Figure 5)²⁰. This hidden mortality confined to monochorionic pregnancies is likely to be the consequence of the underlying chorioangiopagus and severe early-onset twin-to-twin transfusion syndrome. Therefore, reduction of the excess fetal loss in twins, compared to singletons, can only be achieved through early identification of monochorionic pregnancies by ultrasound examination at 11–14 weeks of gestation, and the development of appropriate methods of surveillance and intervention during the second trimester of pregnancy.

The most important complication of any pregnancy is delivery before term, and especially before 32 weeks. Almost all babies born before 24 weeks die and almost all born after 32 weeks survive. Delivery between 24 and 32 weeks is associated with a high chance of neonatal death and handicap in the survivors. In a singleton pregnancy, the chance of delivery between 24 and 32 weeks is 1–2%. In a study of 467 twin pregnancies in which chorionicity was assessed during the 11–14-week scan, the median gestation at delivery of live births was only marginally earlier in monochorionic (36 weeks), compared to dichorionic (37 weeks) pregnancies²⁰. However, the proportion of pregnancies delivering very preterm (before 32 weeks) was nearly twice as high in monochorionic (9.2%) compared to dichorionic (5.5%) twins (Figure 6)²⁰.

The extent to which close monitoring of cervical length and the insertion of cervical sutures in those with a short cervix will reduce the risk of severe preterm delivery remains to be determined.

In singleton pregnancies, the main factors determining fetal growth are genetic potential and placental function, which is thought to be due mainly to the effectiveness of trophoblastic invasion of the maternal spiral arteries.

In monochorionic twin pregnancies, both the genetic constitution and the factors which govern trophoblastic invasion should be the same for the two fetuses. Consequently, inter-twin disparities in growth are likely to reflect the degree of unequal splitting of the initial single cell mass or the magnitude of imbalance in the bidirectional flow of fetal blood through placental vascular communications between the two circulations. In contrast, since about 90% of dichorionic pregnancies are dizygotic, inter-twin disparities in size would be due to differences in genetic constitution of the fetuses and their placentas.

In twin pregnancies, the risk of delivering growth-restricted babies is about 10 times higher than in singleton pregnancies²¹. In a study of 467 twin pregnancies in which chorionicity was assessed at the 11–14-week scan, the chance of growth restriction (birth weight below the 5th centile for gestation in singletons) of at least one of the fetuses was 34% for monochorionic and 23% for dichorionic twins²⁰. Furthermore, the chance of growth restriction of both twins was about four times as high in monochorionic (7.5%) compared to dichorionic (1.7%) pregnancies (Figure 7)²⁰.

Ultrasonographic studies in the first trimester have examined inter-twin disparities in crown–rump length to determine if this measurement is useful in the prediction of pregnancy outcome. One study examined 180 pregnancies at less than 8 weeks of gestation (median crown–rump length of 8.4 mm) and reported that in those pregnancies resulting in two live births, the median inter-twin disparity in crown–rump length was about 10% (0.9 mm); a difference of more than 3 mm was associated with a 50% chance of intrauterine death of the smaller twin²². There are also three studies reporting on a total of seven pregnancies discordant for growth restriction or congenital abnormalities that demonstrated large inter-twin disparities in crown–rump length at 6–11 weeks of gestation^23–25.

In a study of 123 monochorionic and 416 dichorionic twin pregnancies, there were no significant differences in inter-twin disparity in crown–rump length at the 11–14-week scan or birth weight between monochorionic and dichorionic twins (Figure 8)²⁶. In addition, there was no significant correlation between inter-twin disparities in crown–rump length and inter-twin disparities in birth weight. In dichorionic pregnancies with chromosomally abnormal fetuses, and in those which ended in miscarriage or intrauterine death of one or both fetuses, the inter-twin disparity in crown–rump length was significantly higher than in pregnancies resulting in two live births. However, in the monochorionic twins with adverse pregnancy outcome, there was no significant difference in inter-twin disparity in crown–rump length from pregnancies resulting in two live births²⁶.

In twin pregnancies resulting in live births the median inter-twin disparity in fetal size increases with gestation from about 3% at 12 weeks to 10% at birth²⁶. In monochorionic twins, this increasing disparity may be a consequence of the degree of imbalance in fetal nutrition as a result of chronic twin-to-twin transfusion syndrome. Similarly, in dichorionic twins, the increasing disparity in size may also be due to differences in fetal nutrition, but in this case such differences may be a consequence of discordancy in the effectiveness of trophoblastic invasion of the maternal spiral arteries and therefore placental function. The finding, of no significant association between inter-twin disparity in crown–rump length and inter-twin disparity in birth weight²⁶, suggests that assessment in early pregnancy cannot provide useful prediction of the subsequent development of either mild chronic twin-to-twin transfusion syndrome in monochorionic twins or growth restriction in dichorionic twins.

The findings in dichorionic twins (that adverse pregnancy outcome or chromosomal abnormalities are associated with large inter-twin disparities in crown–rump length)²⁶, suggest that, in such pregnancies, there is early-onset growth restriction in one of the fetuses, either due to a genetic defect or impaired placentation. In addition, the association between large inter-twin disparities in crown–rump length and mis carriage are compatible with observations that, in multiple pregnancies, spontaneous or iatrogenic death of one of the fetuses can destabilize the whole pregnancy, resulting in miscarriage or severe preterm delivery.

The finding in monochorionic twins, that adverse pregnancy outcome is not associated with large inter-twin disparities in crown–rump length at the 11–14-week scan, suggests that, at this early gestation, fetal growth may not be affected by impaired nutrition through such conditions as chronic fetal hemorrhage. It is possible that at this stage there is programmed fetal growth that may only be affected by serious genetic abnormalities, such as chromosomal defects, or extreme degrees of placental impairment that will subsequently result in fetal death.

Splitting of the embryonic mass after day 9 of fertilization results in monoamniotic twins. In these cases, there is a single amniotic cavity with a single placenta and the two umbilical cords insert close to each other. Monoamniotic twins are found in about 1% of all twins or about 5% of monochorionic twins. In a series of 1288 twin pregnancies (including 317 monochorionic) examined at the 11–14-week scan at the Harris Birthright Research Centre, King’s College Hospital, there were 14 mono amniotic pregnancies (including four with conjoined twins and two with twin reversed arterial perfusion sequence).

In monoamniotic twins, the fetal loss rate is about 50–75%, due to fetal malformations, preterm delivery and complications arising form the close proximity of the two umbilical cords. Cord entanglement is generally thought to be the underlying mechanism for the majority of fetal losses, and attempts have been made to prevent this complication by the administration to the mother of sulindac during the second trimester to stabilize the fetal lie by reducing the amniotic fluid volume⁴³. However, cord entaglement is found in most cases of monoamniotic twins and this is usually present from the first trimester of pregnancy^44–46. Therefore, a more likely cause of fetal death in monoamniotic twins, which occurs suddenly and unpredictably, is acute twin-to-twin transfusion syndrome. The close insertion of the umbilical cords into the placenta is associated with large-caliber anastamoses between the two fetal circulations^46,47. Consequently, an imbalance in the two circulations could not be sustained for prolonged periods of time (which is necessary for the development of the classic features of twin-to-twin transfusion syndrome), but would rather have major hemodynamic effects, causing sudden fetal death.

On the basis of existing data, the diagnosis of monoamniotic twins at the 11–14- week scan should lead to counselling of the parents as to the high risk of sudden, unexpected and non-preventable fetal death. In our series of eight monoamniotic pregnancies with two separate fetuses diagnosed by the early scan, there were four discordant for major fetal abnormality and these resulted in termination of pregnancy or death of both fetuses. In the four cases where both fetuses were normal, one resulted in survival of both twins, another in survival of one twin and two in intrauterine death of both fetuses (Table 1).

Splitting of the embryonic mass after day 12 of fertilization results in conjoined twins, which are found in about 1% of monochorionic pregnancies. Conjoined twins are classified, according to the dominant site of interfetal body part connection, into five major types: thoracopagus (thorax, 30–40%), omphalopagus (abdomen, 25–30%), pygopagus (sacrum, 10–20%), ischiopagus (pelvis 6–20%) and craniopagus (head, 2–16%). The prognosis depends on the site and extent of conjoining, but, in general, about 50% are stillborn and one-third of those born alive have severe defects for which surgery is not possible. In the live-born cases in whom planned surgery is carried out, about 60% of infants survive^48,49.

In our four cases of conjoined twins diagnosed at the 11–14-week scan, the nuchal translucency was increased in six of the eight fetuses (0.5 mm and 6.5 mm at 11 weeks, 4.2 mm and 7.5 mm at 11 weeks, 2.4 mm and 3.6 mm at 13 weeks and 3.5 mm and 7.0 mm at 13 weeks, respectively). However, the extent to which nuchal translucency provides useful prediction as to the outcome of such pregnancies is uncertain. In cases diagnosed in the first trimester, the patients usually elect termination of pregnancy. There are no series reporting on the natural history of the condition.