European Journal of Clinical Investigation (1977) 7,465-472
Prenatal diagnosis of neural tube defects* Based on European Mack-Forster Award Lecture, 1977 D. J. H. BROCK, University Department of Human Genetics, Western General Hospital, Edinburgh
For most of the past decade prenatal diagnosis has been concerned largely with cytogenetic disorders [ l ] and in particular with the early detection of Down’s syndrome in mothers who are at risk because of advanced age. A quantitatively less important, but nonetheless scientifically significant, corollary to these studies has been the diagnosis of a variety of metabolic disorders early in pregnancy . Recently it has become possible to detect the neural tube defects through measurements of alphafetoprotein in the amniotic fluid  and in some parts of the world this has become of equal importance to the determination of fetal karyotype. Furthermore, since assay of alphafetoprotein concentration in maternal blood allows the early recognition of most cases of both spina bifida and anencephaly, it has now become possible, at least in theory, to apply prenatal diagnosis to all pregnant women and thus to make an attack on the incidence of one of the more serious groups of congenital malformations. This review will be concerned with assessing the current state of knowledge in the early prenatal diagnosis of neural tube defects. The neural tube defects
Defects of closure of the neural tube account for the great majority of the central nervous system malformations. The commonest anomalies are anencephaly with or without spina bifida and the various forms of spina bifida cystica. Spina bifida occulta, though apparently common, is rarely associated with neurological or musculo-skeletal disturbances. Less common forms of neural tube defect are hydrocephaly, exencephaly, iniencephaly and encephalocele  . The aetiological relationship between anencephaly and spina bifida cystica is well established, but their relationship to the less common neural tube defects is not yet clear. From the point of view of prenatal diagnosis it is convenient to divide the neural tube defects into open and closed lesions. A closed lesion is defined as a malformation in which there is a full-thickness skin cover. An open lesion is defined as one in which meninges *Based on the 1977 European Mack-Forster Award Lecture given to the Meeting of the European Society for Clinical Investigation, Rotterdam, April 1977. Correspondence: Dr D. J. H. Brock, University Department of Human Genetics, Western General Hospital, Edinburgh 4, Scotland.
or neural elements lie exposed on the surface, whether these be covered by a membrane or not. Most cases of anencephaly and myelocele spina bifida represent open lesions, while the majority of meningocele spina bifidas and encephaloceles are closed lesions. However, the correlation between the open and closed terminology required by prenatal diagnosticians and the various subdivisions of the neural tube defects is not an exact one and remains t o be clarified by further careful pathological studies. The geographical distribution of the neural tube defects varies widely. Incidence rates are complicated by seasonal fluctuations, by long-term trends and by doubtful statistics from areas where compulsory registration of congenital malformations is not demanded. Highest frequencies are found in the North and West of the British Isles, particularly in Scotland, Wales and Northern Ireland where between six and ten births per 1000 may be affected. Conversely, low frequencies (less than one per 1000) have been shown in most black populations, whether in West Africa, the United States, the West Indies or Britain . Low rates are also found among Asians (with the exception of Sikh Indians) and in South America. Intermediate rates (one to three per 1000 births) have been reported from most parts of Europe outside the British Isles . In the United States the highest rates appear to occur in the North-East with an incidence of 1.6/1000 for anencephaly being reported from Rhode Island . In estimating recurrence risks from empirical studies it has been shown that anencephaly, iniencephaly, encephalocele, myelocele and meningocele may be treated as having a common aetiology . Recurrence risks after a single affected sib range from 4.5% to 5.6% , and there is little difference whether the index patient had spina bifida or anencephaly. The recurrence risk is lower when the birth frequency of neural tube defects in the population is lower, and it is probable that the relative risks based on British data (Table 1) are overestimates for most other populations. The recurrence risk after two affected sibs is of the order of 10%. If either of the parents is a surviving patient with spina bifida cystica, the risk to their children is about 4.5% . Other empirical risks are shown in Table 1 and include calculations for more complicated family histories [lo]. It should be noted that because there is a Ushaped relationship between both birth order and maternal age and the incidence of neural tube defects, the older mother has a slightly increased risk of bearing 465
D. J. H. BROCK
an affected infant, though the magnitude of this effect is not large.
by the closed lesions but is generally reckoned at between 5% and 10% . It is also not yet clear how serious these failures will be. Many will be represented by the comparatively harmless meningocele spina bifidas, but
Amniotic fluid studies
Table 1. Empirical risk of neural tube defect
The discovery of greatly increased concentrations of alphafetoprotein (AFT)in the amniotic fluids surrounding fetuses with neural tube defects stemmed from earlier observations of elevated bilirubin levels in term fluids of anencephalic infants Ell]. Bilirubin is an ambiguous marker in the sense that it could derive from either the maternal or fetal circulations, and as a comparatively low molecular weight compound would have a fairly rapid turnover. There seemed little profit to be gained in pursuing bilirubin concentrations as a parameter for the early prenatal diagnosis of neural tube defects, and indeed subsequent studies have shown this to be the case . A more effective marker would have to be of high molecular weight, unambiguously fetal in origin and measurable by a simple and specific procedure. An obvious candidate was AFP, a major serum protein of early fetal life and measurable by conventional immunodiffusion and immunoelectrophoretic techniques using a well-tried and readily available commercial antiserum. In their original publication Brock & Sutcliffe 131 collected thirty-seven third-trimester amniotic fluids from pregnancies where the outcome had been an infant with spina bifida, anencephaly or hydrocephaly and showed that AFP concentrations were greatly increased in a large majority of these. One amniotic fluid from a myelocele spina bifida at 13 weeks  and one from a anencephalic at 18 weeks [ 131 had AFP concentrations which were more than five times the upper limit of the normal range, thus predicting the usefulness of the method in early diagnosis. These frndings were rapidly confirmed in other laboratories and amniotic fluid AFP assay is now widely offered at genetic counselling clinics thoughout the developed world. AFP concentrations in amniotic fluids from pregnancies where the fetus has a neural tube defect are usually very high indeed (Fig. 1). Anencephalic values tend to be higher than those in spina bifida, in accordance with the theory that the protein leaks from the lesion into the surrounding fluid [ 141. If ‘rocket’ electrophoresis is the assay method used, neural tube defects often reveal themselves as open peaks, making precise quantitation largely redundant (Fig. 2). However, less clear cut, but nonetheless abnormal values may sometimes be observed with small, open spina bifidas [ 151 . Amniotic fluid AFP will not allow the detection of all types of neural tube defect. This was first pointed out in a report of an occipital encephalocele with a normal amniotic fluid AFP value at 16 weeks [ 161. There is now a general belief that a closed neural tube defect will not be detectable through assay of AFP in amniotic fluid. It is not yet clear what proportion of the total number of cases of neural tube defects are represented
One child with NTD Two children with NTD Three children with NTD Parent with NTD
5 10 21 4.5
One child with multiple vertebral anomalies One child with spinal dysraphism
o a0 O
0 0 8 0
0 O 0
Figure 2. Rocket immunoelectrophoresis of amniotic fluid AFP. Open topped rocket indicates a neural tube defect.
PRENATAL DIAGNOSIS OF NEURAL TUBE DEFECTS as Laurence [I71 has pointed out, encephaloceles and small, skin-covered myelocele spina bifidas are serious disorders, and may cause great distresss to mothers who were expecting to be protected against the birth of a child with a neural tube defect. Measurements of amniotic fluid AFP have been criticized on the grounds of their relative non-specificity, it being suggested that the association of other fetal conditions with raised values compromises the usefulness of this diagnostic tool. A close inspection of the reported facts does not entirely support this criticism. It must be remembered that the objective of AFP analysis is tomake diagnoses early enough in pregnancy to permit termination should the fetus be shown to be abnormal. Thus the observation that amniotic fluid AFP is raised in oesophageal atresia in the third trimester of pregnancy  has little bearing on the specificity of AFP assay until such a condition can be shown to be associated with an elevated value at or around the sixteenth week of pregnancy. A more valid comment relates to the variability of AFP elevations in many fetal conditions. Exomphalos (omphalocele) and Turner’s syndrome are cases in point. Both normal and raised AFP concentrations have been observed early in pregnancy (Table 2) Tabla 2. Fetal conditions, other than anencephaly and spina bifida, in which elevated amniotic fluid AFP concentrations have been reported
Condition 1. Intrauterine death (missed abortion)
3rd R 2nd R 1st R 2. Rh isoimmunization 3rd R 2nd RandN 3. Hydrocephalus 3rd N 3rd R 4. Turner’s syndrome 1st R 2nd R 2nd R N 2nd 5. Exomphalos 1st and 2nd N 3rd R R 2nd 2nd R 6. Congenital nephrosis 3rd R 2nd R 2nd R 7. Duodenal atresia 3rd R 2nd R 8. Oesophageal atresia 3rd R 9. Sacrococcygeal teratoma 2nd R 10. Meckel syndrome 2nd R 11. Annular pancreas 3rd R 12. Fdot’s tetralogy 3rd R 13. Pilonidal sinus 2nd R 14. Congenital skin defects 3rd R R 2nd
* R,raised; N, normal.
Reference [561 1571 [581
 131 1591
[601 [611 [621 ~ 3 1 [641 [65 1 (661 [671 1561 [681 ~ 9 1 [701 [711 [I81 1721 1731 I231
P I [751 [701 ~761
and there is as yet no data on the correlation of the AFP values and the severity of the conditions. Of the examples shown in Table 2, only intrauterine death and congenital nephrosis are always likely to be associated with raised AFP early in pregnancy. A common source of apparent false positive amniotic fluid AFP values are those deriving from fluids contaminated with fetal blood. AFP levels in fetal serum parallel those in amniotic fluid quite closely, but with concentrations between 100 and 200 fold higher [ 191 This means that the admixture of a comparatively small amount of fetal blood with the amniotic fluid can increase the AFP value to a point where it mimics the value found in neural tube defects. Current practice in many laboratories is to set aside an aliquot of whole amniotic fluid from any sample where there is visible blood contamination. Should the AFP value be elevated, the cell button is then examined for the presence of fetal red cells either by Kleihauer test or by electrophoresis which will show the characteristic band of haemoglobin F. Contaminated samples with moderately raised AFP values which contain a major proportion of fetal blood may be rejected and a fresh amniocentesis called for. Alternatively an attempt may be made to calculate the expected contribution of AFP from fetal blood, by estimating the number of fetal erythrocytes in the fluid and the mean level of AFP in fetal serum at the appropriate stage in gestation . When allowance has been made for fetal bloodcontaminated samples it is still difficult to estimate the proportion of genuine false positives. Much depends on the accepted upper limit of the normal range. The limited data in Table 3 suggest that between 3 and 4 standard deviations above the mean for the gestational period is the critical point. The precise figure is of considerable importance in view of the fact that it is now widely recommended that a l l amniotic fluid samples should be subjected to AFP analysis, whatever the reason for amniocentesis . Since the most common indications for amniocentesis remain advanced maternal age or a previous history of chromosome abnormality, neural tube defects will be encountered infrequently in these samples. It is therefore of the greatest importance
Table 3. Cumulative false positive and false negative results in
amniotic fluid AFP at various cut-off points False negatives Cut-off in SD above mean