Schizencephaly revisited

Schizencephaly is a rare congenital abnormality of the brain that has a prevalence of approximately 1.5 per 100,000 live-born babies as estimated from a review of four million births in California [1]. Howe and colleagues reported a combined live birth/stillbirth rate of 1.48 per 100,000 from a population of over 2.5 million in the UK and less than half of those cases were detected antenatally [2]. In spite of recent interest in a range of genetic abnormalities that are associated with schizencephaly [3‐5], the majority of cases of schizencephaly are sporadic and non-familial, and in most cases, an aetiological cause is not found [6].

Magnetic resonance (MR) imaging is central to the diagnosis of children with congenital brain abnormalities, and in utero MR (iuMR) imaging is now used for antenatal detection of brain abnormalities, including schizencephaly. The primary purpose of this paper is to assess the range of appearances of schizencephaly in children and fetuses by review of a 10-year experience from a single centre. I will also detail the classification systems used to describe different forms of schizencephaly and re-consider the possible aetiological/mechanistic causes.

Twenty-one consecutive cases of pediatric schizencephaly and 11 cases of fetal schizencephaly are reported in this article in order to illustrate the range of appearances that can be expected in clinical practice. For most radiologists, schizencephaly is a term used to describe congenital (present at birth) brain abnormality in which the defining lesion is a full-thickness CSF cleft lined by abnormal grey matter (polymicrogyria and/or heterotopia). The grey matter lining distinguishes it from porencephaly, which results from a destructive lesion in the fetal brain white matter, lined with variable degrees of gliosis, (Fig. 2), some of which can be minimal. CSF-containing clefts can be missed on MR imaging, particularly if imaging in the three orthogonal planes is not used, and 3D T1 volume imaging with thin partitions is ideal for this role once myelination is mature. In spite of using best quality imaging, a cleft can still be overlooked and a dimple or ‘nipple’ on the ventricular surface may be the only indicator of trans-mantle abnormality on MR imaging. I have noted another definition of schizencepahly that does not require a cleft to be present [10‐12], and recent editions of Barkovich and Raybaud’s textbook acknowledge ‘…transmantle heterotopia may be an extreme closed lip schizencephaly’ [8]. Those differences in definition of schizencephaly can cause confusion when attempting to sub-classify types of schizencephaly, hence my attempt to produce the unifying system shown in Fig. 1.

Allowing for the fact that many authors consider the schizencephaly (type 1) cases in this paper to be trans-mantle heterotopion, there are inevitable differences in the anatomical descriptions of schizencephaly found in textbooks and previous papers when compared with the current cohort. It is noteworthy that there were no fetuses with schizencephaly (type 1) in this cohort, which may be due to the limitations of iuMR imaging in terms of anatomical resolution. Barkovich reports bilateral schizencephaly in 40–50% of pediatric cases [8], which is comparable with the findings reported here (just less than 50% in children and 36% in fetuses), but there were substantial differences in the category of schizencephaly. ‘Closed lip’ schizencephaly is found in only 15–20% of children according to the textbook in comparison with schizencephaly (type 2) being found in 67% of children and 50% of fetuses in this study. The frontal and parietal preponderance reported by Barkovich (about 75%) was re-demonstrated and strengthened here (94% of all schizencephalies in both the pediatric and fetal cohorts). Textbook estimates of the absence of the septum pellucidum in children with schizencephaly are around 70% (optic nerve hypoplasia in approximately 30%) and were similar in this report (absent or disrupted septum pellucidum in 76% of pediatric cases and 64% of fetuses). The tendency for absent septum pellucidum to be associated with ‘open lip’ (schizencephaly (type 3) reported previously was also shown in our cases. Other brain abnormalities (not including the septum pellucidum/fornix) were found in approximately two thirds of the children which is comparable to previous reports from pediatric populations.

There are limited reports of fetal schizencephaly using iuMR imaging—for example, small case series [16] and case reports [17‐19] with the largest case series to date by Nabavizadeh et al. [20] who reported 10 fetuses with schizencephaly on iuMR imaging and confirmed on post-natal imaging including eight bilateral cases. One interesting finding in that study was a morphological change in the type of cleft between the pre-natal and post-natal imaging as 47% of the ‘open-lip’ schizencephalies detected in utero had become ‘closed lip’ on post-natal imaging. Another important point was how frequently polymicrogyria was missed on iuMR imaging in that study, and this may account for the low rate of other brain abnormalities shown on iuMR in the present study (55% compared with 81% in children).

One mechanistic explanation of the aetiology of schizencephaly arises from the work of Yakovlev and Wadsworth in 1946 [11, 12]. In their opinion, schizencephaly is due to ‘failure of growth and differentiation of a circumscribed part of the cerebral wall’ as quoted in Greenfield’s neuropathology [21]. As such, it has its origins in the first trimester, specifically the first 2 months post-conception. Naidich [22] elaborated and refined the theory by proposing pinning of the ependymal and pia, which are anatomically very close in the first trimester (Fig. 12). The focal non-separation leads to an ependymal/pial seam that lies between the lips of the abnormal developing cleft. The ependymal/pial pinning theory may explain some features of schizencephaly (type 2) lesions where the lips are closely opposed, but intuitively, it seems difficult to explain the widely spaced lips and volume loss frequently present in schizencephaly (type 3). The proponents of this idea merely suggest that larger areas of ependyma and pia fail to separate in those cases.

There is opposition to the pinning theory in recent neuropathology texts, and Harding and Copp, writing in Greenfield’s neuropathology, state ‘… interpretation of the apposed walls of the cleft and the bridging membrane over the defect is no longer convincing, although for obscure reasons it lingers in the radiological literature. There is now ample clinical, morphological and experimental evidence favouring a destructive origin for these lesions’ [21]. Those authors go on to say that further evidence against the pinning theory is the consistent presence of abnormal cortex lining the clefts, which points to an aetiological cause in the second trimester: ‘…timing is suggested by the coincident polymicrogyria to be between the fourth and sixth months’. Most textbooks now consider sporadic schizencephaly (the huge majority of cases) to be the result of a destructive lesion(s): Chen, for example, states that the primary aetiology is due to ‘in utero vascular insufficiency’ [9]. The data emerging from iuMR imaging studies also favour destructive aetiologies in most cases. Circumstantial evidence comes from the work of Nabavizadeh et al. [20] who found that 7/10 fetuses with schizencephaly had hemorrhage and/or hemosiderin deposition in the cleft or in the cerebral ventricles on iuMR imaging. The finding reported in the current paper that schizencephaly in the fetus is accompanied by a small head size in 6/11 fetuses (below the 10th centile) and in 4/11 the head size was below the 3rd centile is relevant because the most rational explanation for the high risk of microcephaly is brain destruction. It is interesting to note that the four fetuses with the head size below the 3rd centile all had schizencephaly (type 3), i.e. cases in which greater volume of brain injury might be predicted. Direct evidence comes from the fetus illustrated in Fig. 13, which is not from the cohort as presented but has been reported previously as web extra material [25]. That case shows the evolution of schizencephaly in a second trimester fetus with a normally formed but injured cerebral mantle. That fetus was studied because it was the result of a monochorionic pregnancy complicated by the death of the co-twin, a known risk factor for schizencephaly [9, 21].

Consistent with many previous descriptions, the septum pellucidum was absent in many cases of schizencephaly, although the reason for the association is not known with certainty. In some of the pediatric and fetal cases reported here, remnants of the septum pellucidum were visualised, suggesting that the structure had formed but was damaged, perhaps by raised intraventricular pressure arising from hydrocephalus secondary to intraventricular blood. It is not possible to say if this mechanism accounts for the case in which no portion of the septum pellucidum was seen.

I explained the need to distinguish between schizencephaly and porencephaly earlier, but from the discussion above, it seems likely that destructive brain parenchymal lesions are the most likely cause for most cases of schizencephaly as well as porencephaly. Why would one fetus develop schizencephaly and another develop porencephaly? It is attractive to propose that the gestational age at which the brain injury occurs is the deciding factor. If the insult occurs in the ‘4–6-month’ time period proposed by Harding and Copp [21], further neuroglial migration is possible and likely after the damaging event. That ‘late migration’ neurons could extend as far as the margins of the injured brain but are not likely to organise properly and result in polymicrogyria (Fig. 13). In contrast, a brain injury occurring after 6 months is most likely to produce an abnormality where the damaged brain is resorbed, and because no further neuroglial migration is possible, the margins are not lined by normal (or abnormal) grey matter. The exact appearances of injuries in that time period will depend on the ability of the brain to produce a gliotic reaction, porencephaly resulting if there is no gliosis (less mature fetuses) or encephalomalacia if there is gliosis (more mature fetuses) [8].

If most cases of schizencephaly are the result of destructive lesions in the second trimester, the damage to the cerebrum must be trans-mantle but it is interesting to consider cases in which a focal brain injury is superficial rather than full thickness. If the injury occurs at the time that further neuroglial migration is possible, the damaged cortical plate may become re-populated by grey matter (Fig. 14). Again, it is not likely to result in normal six-layered cortex but more likely to result in what has been called ‘reparative polymicrogyria’. This is a possible explanation for the association of unilateral schizencephaly (full-thickness vascular injury) and contralateral polymicrogyria (superficial vascular injury) in the same fetus as suggested in a previous work [26].

Other recognised risk factors for the development of schizencephaly include transplacental infections, classically by cytomegalovirus [21] with more recent reports of infection by the Zika virus [27]. Transplacental infection by cytomegalovirus may involve the ventricular layer (germinal matrix) where the future neurons and glia are formed and mature before migration. It is possible that in these circumstances schizencephaly may result from abnormal migration/organisation. Alternatively, cytomegalovirus infection is known to cause arteritis, which potentially could injure previously normal brain and results in schizencephaly by ‘late migration’ mechanism described above. Schizencephaly can also be a feature of many congenital anomaly/mental retardation conditions such as Adam-Oliver, Aicardi, Arima, Delleman, Galloway-Mowat and micro syndromes [21].

Discrete genetic causes of schizencephaly have been difficult to confirm and it is generally considered to be a non-genetic, non-familial ‘sporadic’ abnormality. Early reports linking schizencephaly to mutations in the EMX2 gene have not been borne out [24]. Two cases in our pediatric cohort did have relevant histories: one child had a known mutation of the EMX2 gene and the other had a strong family history of schizencephaly but no known genetic abnormality at the time of the MR study. The MR imaging features of those two children had striking similarities (Fig. 9)—both were microcephalic, both had bilateral schizencephaly (type 2) involving the paracentral lobes and both had extensive bilateral polymicrogyria and intact septum pellucidum with normally situated fornices. The presence of normal septum pellucidums in these cases is particularly noteworthy as the structure was either absent or disrupted in 16/19 (84%) of the other children with schizencephaly.

In conclusion, I have proposed a new system for classifying schizencephaly that takes into account all definitions of the abnormality. Using that approach, I have described the appearances and associations of pediatric and fetal cases of schizencephaly from a single centre. Review of the current literature appears to favour an acquired destructive aetiology for most cases of schizencephaly, and I have proposed a mechanism to explain the cortical formation abnormalities (predominantly polymicrogyria) found consistently in and around areas of schizencephaly.