Introduction

Twin pregnancies are more commonly affected by poor growth, as the human uterus is less capable of fulfilling the requirements of more than one fetus. As such, one in four twins is born small for dates (birthweight below the 10th percentile according to singleton nomograms) . Up to 32 weeks’ gestation, growth curves for twins are similar to those of singletons, at which point the growth velocity begins to slow down . In twins, the average birth weight crosses the 10th centile at 38 weeks, which implies that all twins born at 38 weeks will be on average growth restricted according to singleton standards. Some investigators consider this growth restriction physiological, and advocate the use of twin-specific nomograms to better predict intrauterine demise . The use of twin-specific standards, however, has so far failed to gain widespread acceptance for fear of creating a false sense of security that it is normal for a twin to be small.

Ultrasound scanning plays an indispensable role in the diagnosis and management of poor growth in a twin pregnancy. At first, an accurate pregnancy dating and knowledge of chorionicity are mandatory, and both must be established on the first-trimester ultrasound scan. Growth in a twin pregnancy may be described in two ways: the estimated fetal weight of each individual twin and the discordance in estimated fetal weight between the twins. Discordant growth is determined as (A − B) × 100/A, where A is the weight of the larger and B is the weight of the smaller twin. In each pair of growth-discordant twins, the larger usually grows appropriately, whereas the smaller twin eventually becomes growth restricted. The average growth discordance in monochorionic as well as dichorionic twin pregnancies is 10%. The threshold for clinically relevant discordant growth is set at a difference in estimated fetal weight of more than 25%, which affects about 10% of twin pregnancies and is as common in monochorionic as in dichorionic twins . Although gestational age at birth and birthweight are the most important predictors of postnatal outcome, discordant growth is also independently associated with adverse outcome . As such, even if both twins are appropriately grown, discordant growth confers an added risk, especially if it concerns a monochorionic twin pregnancy . Antenatal factors identified as having an association with discordant growth are a peripheral cord insertion in one twin , monochorionicity , sex-discordant twins , nulliparity , and assisted conception . Although discordant growth affects monochorionic and dichorionic twins equally, the risk of adverse perinatal outcome is higher for monochorionic than for dichorionic twins at every level of discordance . Also, monochorionic twins are more commonly small for dates than their dichorionic counterparts , revealing that having to share a single placenta is a disadvantage.

The antenatal detection of severe discordant growth by ultrasound scan is far from ideal, with sensitivities varying between 23 and 61% . The difference in abdominal circumference rather than the actual estimated fetal weight may be a better estimate to assess discordant growth. As such, a ratio of larger and smaller abdomen circumference of more than 1.3 seems a more accurate predictor of discordant growth . Also, it is helpful to put the two images of both abdominal circumferences with similar magnification next to one another to establish any significant discordance, if in doubt, about a significant growth difference ( Fig. 1 ). Obviously, other parameters of placental function (amniotic fluid volume and Doppler indices) should also be evaluated whenever aberrant growth is suspected.

Ultrasound image of the abdominal circumferences of both twins put next to one another to show important discordant growth (44%).

In a monochorionic pair

In monochorionic twins, isolated discordant growth must be distinguished from twin-to-twin transfusion syndrome (TTTS). Most monochorionic twins with discordant growth also have discordant amniotic fluid volumes, with a decreased amount in the smaller and not uncommonly a mild increase in the larger twin. Yet, to be labelled isolated discordant growth, the fluid discordance must be mild and not comply with the criteria of TTTS (<2 cm and >10 cm deepest vertical pocket). As such, a growth difference with a severe oligohydramnios of less than 2 cm in the smaller twin, but only a deepest vertical pocket of 6 cm in the larger, should be labelled and managed as isolated discordant growth and not as TTTS. In fact, discordant growth in a monochorionic pair is always a provisional diagnosis. As long as the twins are in utero , they may still develop TTTS or twin anaemia sequence (TAPS) on top of their initial growth discordance.

By definition, all monochorionic twins are monozygotic, and thus differences in genetic growth potential cannot explain a difference in size. Three factors determine the venous return upon which each twin depends for its oxygen and nutritional supply: the degree of sharing of the single placenta between the twins, the vascular anastomoses, and the quality of implantation of each individual part. An unequal sharing of the placenta is the main cause of discordant growth in a monochorionic twin pregnancy . Unfortunately, on prenatal ultrasound, the size of each individual territory cannot be estimated. Yet, the site of cord insertion can be reliably determined , and the combination of a velamentous central cord insertion and central cord insertion may serve as a proxy marker for unequal placental sharing, and thus identify a group of monochorionic twins at high risk for discordant growth . Vascular anastomoses also influence growth in monochorionic twins. As such, an unbalanced net arteriovenous transfusion, as in TTTS and TAPS, may restrict and promote the growth of the donor and recipient twin, respectively . Also, unequally shared placentas have larger arterio–arterial anastomoses and a more elaborate intertwin blood exchange than equally shared placentas. Most probably, this extensive blood exchange fulfills a beneficial and life-saving role by increasing the availability of oxygen and nutrients to the twin with the smaller placental part . Finally, a suboptimal implantation of part of the placenta may also lead to growth restriction of the corresponding twin . One has to bear in mind that monochorionic twins may rarely be discordant for chromosomal anomalies and genetic syndromes. With discordant growth, it seems important to consider the possibility of trisomy 18, triploidy, and Beckwith–Wiedemann syndrome in one of the twins, the latter being more common in monozygotic female twins after assisted conception .

Isolated discordant growth is as common as TTTS, but has a much better outcome and usually does not require invasive treatment. In one-half of cases, the growth discordance is present from before 20 weeks and in one-half the growth discordance arises only in the latter half of pregnancy. Monochorionic pairs with discordant growth from early on in pregnancy have different placental characteristics and different outcome than those where discordant growth arises later on. Twin pregnancies with early onset discordant growth typically have an unequally shared placenta with large anastomoses. Intrauterine demise occurs in about 20% of twins, and most have an abnormal umbilical artery Doppler pattern from 16 weeks onwards. On the other hand, pregnancies with progressively increasing growth discordance after 26 weeks have more equally shared placentas with smaller anastomoses. Doppler examination in the umbilical artery of the smaller twin usually shows a positive end-diastolic flow pattern in the umbilical artery, and the survival rate is nearly 100%. Nevertheless, late-onset discordant growth in a monochorionic twin pair should raise the suspicion of twin anaemia–polycythaemia sequence, and is an indication for measurements of the peak systolic velocities in the middle cerebral artery to detect any discordance .

Next to classification based on gestational age at first presentation, growth-discordant monochorionic pairs can also be classified according to the Doppler characteristics of the umbilical artery of the smaller twin ( Fig. 2 ). In singletons, absent end-diastolic flow indicates poor trophoblastic invasion and reduced uteroplacental blood flow. In contrast, in monochorionic twins, the blood flow across the anastomoses and the degree of hypovolaemia also influence the umbilical artery Doppler pattern. The presence of a large arterio-arterial anastomosis may result in a cyclical variation in the end-diastolic flow component, and thus in an intermittent absent or reversed end-diastolic flow pattern in the umbilical artery of the smaller twin. The smaller twin of a growth-discordant monochorionic pair may either a positive (Type I), a persistent absent or reversed (Type II) or an intermittent absent or reversed end-diastolic flow pattern (Type III), depending on the presence or absence of a large artery-to-artery anastomosis. Each of these types has distinct placental characteristics and different outcomes. Large arterio-arterial anastomoses (>2 mm) are present in 70%, 18%, and 98% of Type I, Type II, and Type III, respectively. Also, an inverse relationship exists between the degree of unequal sharing and the type of growth discordance, Type III cases having the most unequally shared placentas. Pregnancies with positive end-diastolic flow in the umbilical artery of the smaller twin (Type I) have the best outcome, with a low-risk of deterioration and a survival rate of nearly 100%. Twin anaemia sequence should be ruled out in these cases as a cause of discordant growth by peak systolic velocities in the middle cerebral artery measurement. Cases with a persistent absent end-diastolic flow in the umbilical artery of the smaller twin (Type II) is the least common type, but carries the worst prognosis, as 90% eventually deteriorate and survival rates are only 60%. Type II is likely caused by deficient placentation, with moderate unequal sharing and absence of a compensating large artery-to artery anastomosis. Because these cases usually have only small or no artery-to artery anastomoses, the larger twin has less or no opportunity to help the smaller twin by inter-twin blood exchange. The absence of an inter-twin rescue transfusion also explains why these cases behave like singletons with early onset growth restriction (i.e. Doppler patterns show a similar pattern of progressive deterioration of the arterial, middle cerebral artery and ductus venosus Dopplers and a derangement of the biophysical profile score so that demise can be predicted). Finally, pregnancies with an intermittent absent end-diastolic flow pattern (Type III) have an intermediate prognosis with an 80% survival, but are the most unpredictable. Type III cases have the most unequally shared placentas, but large anastomoses and an extensive intertwin blood exchange rescues the twin on the smaller placental part. Demise of the smaller twin occurs in about 15% of pregnancies, which is usually unexpected as with monoamniotic twin gestations and not preceded by signs of progressive hypoxic deterioration. Because of the large artery-to-artery anastomoses, in about one-half of the cases with demise of the smaller twin, concomitant demise of the larger twin occurs. In Type III, the larger twin especially may be at increased risk of antenatal brain injury, owing to larger arterio-arterial anastomosis, short episodes of bradycardia, or hypotension in the smaller twin, leading to considerable volume shifts to the smaller twin, and increasing the risk of ischaemic brain lesions in the larger twin . Finally, in one out of five Type III cases, the larger twin has cardiomegaly, with increased ventricular wall thickness, most likely caused by the burden of perfusing part of the placental territory of the smaller twin. These hypertrophic cardiac changes seem to be reactive, and are not associated with a poorer neonatal outcome . Pulmonary artery stenosis in the larger twin and aortic coarctation in the smaller twin is also a common association in these Type III discordant monochorionic pairs. The large inter-twin blood exchange in Type III is, therefore, a double-edged sword, because it promotes the survival and growth of the twin with the small placental part, but this is not without risks for the twin that has the larger part, and massive shifts can occur in the event of bradycardia or demise of the smaller twin.

(a) Typical placenta from a twin pair with Type I discordant growth (30%) electively delivered at 37 weeks. Mild unequal sharing takes place, and the pattern of anastomoses is quite similar to that of uncomplicated monochorionic twin pregnancies. Twin 2 with the smaller placental share has a velamentous cord insertion. The star marks the bi-directional artery-to-artery anastomosis. The full circle marks a uni-directional arterio-venous anastomosis from twin 1 to 2, whereas the dotted circle marks an oppositely directed arterio-venous anastomosis from twin 2 to 1; (b) typical placenta from a twin pair with type II discordant growth delivered for worsening of fetal condition at 30 weeks. Moderate unequal sharing and is evident, with only a small arterio-arterial anastomosis and several small arteriovenous anastomosis. The growth discordance is severe. Owing to the lack of important inter-twin blood exchange, the small twin can die, or does not benefit from a rescue transfusion, and has to cope on its own, which probably explains why the pattern of deterioration is similar to a singleton pregnancy. The full circle marks a uni-directional arterio-venous anastomosis from twin 1 to 2, whereas the dotted circle marks an oppositely directed arterio-venous anastomosis from twin 2 to 1; (c) typical placenta from a twin pair with Type III discordant growth delivered at 30 weeks after spontaneous preterm premature rupture of membranes. The placenta is severely unequally shared. The red vessels are the venous territory of twin 1, whereas the white vessels drain to twin 2. A large artery-to-artery anastomosis can be seen, and several large arterio-venous anastomosis. Growth was significantly discordant from 16 weeks onward. The discordance, however, decreased so that, at birth, the discordance was only 13%. The extensive blood exchange between the twins benefitted the growth of the twin with the smaller placental part. The larger twin showed signs of cardiac hypertrophy and tricuspid regurgitation, which regressed postnatally. The full star marks a uni-directional arterio-venous anastomosis from twin 1 to 2, whereas the dotted circle marks an oppositely directed arterio-venous anastomosis from twin 2 to 1.

In contrast to TTTS, the natural history and management of discordant growth in monochorionic twin pairs is not well established. Without treatment, a pregnancy complicated by TTTS will result in either miscarriage, in a severe preterm birth of two sick neonates, or in the demise of one or both twins; fetoscopic laser coagulation is the best treatment option . In contrast, discordant growth has a much more favourable prognosis, and usually results in two surviving infants, provided the pregnancy is monitored carefully and an elective preterm birth is planned accordingly. In our institution, we follow pregnancies with early onset discordant growth and an abnormal umbilical artery Doppler evaluation (Type II and III) on a weekly basis to detect the evolution to TTTS in a timely manner, and to look for signs of imminent demise. In the pre-viable period, we offer a laser coagulation of the vascular anastomoses if progression to TTTS or a selective reduction in the presence of discordant anomalies or signs of imminent demise of the smaller twin, such as arrest of growth and anuria with anhydramnios (<1 cm) . Expectant management is not an option in a monochorionic twin pregnancy with imminent demise of the smaller twin, because of the risks of exsanguination of the larger twin and possible co-twin death or brain damage in the survivor . A selective reduction by coagulation of the smaller twin’s cord aims to protect the normally growing twin against exsanguination. After cord occlusion, the survival rate for the normally growing twin is about 80–90% . At 28 weeks, Type II and III patients are usually hospitalised for inpatient monitoring with cardiotocography three times a day, and biophysical profile score with Doppler assessment twice a week. We administer a first course of corticosteroids for lung maturation on admission at 28 weeks, and deliver electively by caesarean section at 32–33 weeks after a repeat course of steroids as per our local protocol. Pairs with late-onset discordant growth, or with a normal umbilical artery Doppler evaluation in the smaller twin, are managed as outpatients, and are monitored on a weekly basis, with measurement of the peak systolic velocities in the middle cerebral artery to exclude evolution of TAPS and electively delivered around 34–35 weeks. It is important to bear in mind that the larger twin, in particular, is at risk for respiratory distress syndrome, when an elective preterm birth is considered . It would seem reasonable to offer a course of steroids, especially if the birth is by an elective preterm cesarean section.

We do not usually offer laser coagulation of the vascular anastomoses as first-line treatment for discordant growth with abnormal umbilical artery Dopplers, as, at least for Type III, most do well without intervention. Compared with expectant management, laser treatment improves neither the survival rates nor the neurologic outcome in Type II and III TTTS . Furthermore, the procedure is technically more challenging owing to the absence of polyhydramnios, and entry is always in the sac of the normally grown twin with the best chances of survival. Also, in Type III cases, the surgery is hindered by the multiple and large anastomoses. In addition, as Type III cases typically have severe unequal placental sharing, with a rescue inter-twin transfusion, laser separation actually causes the death of the smaller twin in about 75% of cases . Moreover, even if the small twin survives, it usually remains severely growth restricted, as there is little placenta left, which may increase the risk of long-term impairment. Finally, in about one out of 10 cases, the procedure may also result in the demise of the larger twin, so that a severely growth-restricted twin remains with too little placenta for a favourable outcome . Therefore, in cases with imminent demise of the smaller twin, we prefer a selective reduction by coagulation of the smaller twin’s cord, as it probably carries the lowest risk of handicap and the highest chance for parents to take home a healthy survivor.

As management of severe growth is less clear cut, patient’s preferences and technical aspects also play an important role in the decision process . As such, patients may prefer an elective reduction by cord occlusion (e.g. when they already have a large family to look after), whereas other patients have strong objections against selective reduction, in which case a laser separation may be the best option to protect the larger twin. A large anterior placenta in the absence of polyhydramnios may preclude a laser separation, whereas a large posterior placenta with the smaller twin stuck against the posterior wall may not be accessible for cord occlusion.

Although the classification helps to guide management, further refinement may be necessary. As such, the difference between the different types is not always so obvious, as sometimes Type III may evolve into Type I, and there may be doubt about the presence of a cyclical pattern so that Type II sometimes appears to be a Type III. In some cases, a pronounced cyclical pattern in the umbilical artery of the smaller twin may occur, with cardiac hypertrophy in the larger twin but without severe growth discordance, despite unequal sharing. Finally, sometimes in Type I, severe oligohydramnios and anuria may occur in the smaller twin. The best management and outcome of these atypical cases is even more uncertain. It would be helpful to have larger prospective studies on the outcome of growth restriction and unequal sharing, so that factors can be identified that more accurately predict adverse outcome in these cases.