Neural tube defects cannot be diagnosed prenatally by electrophoresis of amniotic fluid transferrin isoforms.

To the Editor: Prenatal diagnosis of neural tube defects (NTDs) is based on ultrasonography; but uncertainty exists in a few cases, and biochemical analysis of amniotic fluid (AF) is required. Electrophoresis of AF acetylcholinesterase is a specific and sensitive method for diagnosis of myelomeningocele (1, 2). Although false-positive results are rare, another biochemical technique would be of great value. Adult and infant cerebrospinal fluid (CSF) is characterized by a specific marker, asialotransferrin or 2-transferrin, because it migrates more slowly in electrophoresis than 1-tetrasialotransferrin, the main isoform in all biological fluids (3). The specific behavior of 2-transferrin is used for the detection of CSF leakage from the subarachnoid space into the nasal or aural cavity (4, 5). Assuming that CSF would leak from a myelomeningoceleaffected fetus into the AF, we screened AF for asialotransferrin. We undertook a retrospective study of 12 AF samples from NTDaffected fetuses (16–35 weeks of gestation) and 36 AF gestational agematched controls. Serum controls were from newborns or fetuses (collected by in utero puncture at 22 and 32 weeks for fetal karyotyping), and CSF controls were from newborns or infants (normal biochemical and bacteriologic findings). Electrophoresis was performed on agarose with the Hydragel 6 CSF assay (Sebia), followed by immunostaining with a polyclonal anti-transferrin antiserum conjugated to peroxidase (Sebia). Dilutions were used to obtain a transferrin concentration of 10 mg/L. Control sera and AF (regardless of gestational age) produced the classic 1-transferrin band corresponding to the tetrasialotransferrin isoform, whereas control CSF produced 2 bands, a major band of 1 mobility and a lighter band of 2 mobility, corresponding to the asialo isoform (Fig. 1). AF from fetuses with NTDs gave the same pattern as controls without detectable 2-transferrin. To find an explanation for this unexpected result, we compared fetal and newborn CSF electrophoretic patterns. Fetal CSF (n 8; 20 to 35 weeks of gestation) was collected by cranial puncture after medical termination of pregnancy (terminal renal failure, skeletal abnormalities, or trisomy 21). We observed that the fetal CSF pattern depended on gestational age. Before 27 weeks, there was a single 1 band, whereas from 32 weeks, a smear appeared between 1 and 2. However, CSF from a newborn delivered at 37 weeks of gestation displayed the classic CSF pattern, with 2 distinct transferrin bands, rather than this peculiar pattern. This study indicates that electrophoresis of AF transferrin isoforms is inadequate for prenatal diagnosis of NTD. There are 2 possible explanations for the absence of the 2 band: (a) inadequate sensitivity of the technique because of unknown in vivo dilution of CSF leaks in AF; and (b) an absence of asialotransferrin in fetal CSF. The CSF fetal patterns agree with this hypothesis, showing the progressive appearance of the 2 brain isoform from the 32nd week of gestation. The processing of brain glycosylation appears to be mature at birth as the same pattern can be observed in newborn and adult CSF. Similar findings have been reported for the abnormal glycosylation of transferrin in carbohydratedeficient glycoprotein syndrome (6 ). Blood samples from affected fetuses display the classic 1 pattern of unaffected fetuses, suggesting a false negative. The carbohydrate-deficient pattern appears only after birth. Although negative, the present findings may be of interest to pediatric physicians, particularly when they are using protein glycosylation abnormalities for prenatal diagnosis. Fig. 1. Electrophoretic patterns of human transferrin isoforms after agarose electrophoresis and enzyme immunostaining.