, Volume 51, Issue 5, pp 347–356

Effects of failed commissuration on the septum pellucidum and fornix: implications for fetal imaging


    • Academic Unit of RadiologyUniversity of Sheffield
    • Academic Unit of Radiology, C FloorRoyal Hallamshire Hospital
  • Ruth Batty
    • Academic Unit of RadiologyUniversity of Sheffield
    • Department of RadiologySheffield Teaching Hospitals NHS Foundation Trust
  • Dan A. J. Connolly
    • Academic Unit of RadiologyUniversity of Sheffield
    • Department of RadiologySheffield Teaching Hospitals NHS Foundation Trust
  • Michael J. Reeves
    • Academic Unit of RadiologyUniversity of Sheffield
Paediatric Neuroradiology

DOI: 10.1007/s00234-009-0507-x

Cite this article as:
Griffiths, P.D., Batty, R., Connolly, D.A.J. et al. Neuroradiology (2009) 51: 347. doi:10.1007/s00234-009-0507-x


In the previous article, we considered the normal appearances of the midline stuctures of the brain as they appear on high-resolution magnetic resonance imaging. In this article, we discuss the effects of failed commissuration on the midline structures. We highlight some of the misconceptions of this process that may lead to misdiagnosis of agenesis of the corpus callosum in utero.


Magnetic resonance imagingCorpus callosumSeptum pelucidumFornixFailed commissuration


In our previous paper, we reviewed the normal development and anatomy of the corpus callosum, septum pellucidum, and fornix along with some of the commoner anatomical variants as shown on MR imaging. In this paper, we build on those topics by discussing failed commissuration. One of the most clinically relevant of those abnormalities is agenesis of corpus callosum (ACC) as it is one of the most commonly detected brain abnormalities in fetal life and is associated with a high risk of developmental/neurological sequelae [1]. The risk is particularly high when other brain abnormalities are present but ACC is considered to indicate a poor prognosis by many groups even when it is an isolated finding [1, 2].

Recent work has compared the results of fetal ultrasonography and in utero MR (iuMR) imaging and has shown that ACC (and hypoplasia) is diagnosed with greater accuracy when iuMR is included in the diagnostic process [35]. This appears to be true for both second and third trimester imaging. It is relevant to ask why these differences occur, and one pertinent observation is that it is often difficult to show the corpus callosum directly on fetal ultrasonography; therefore, clinicians in the field often rely on surrogate anatomical indicators of its presence. Leading amongst those is the cavum septum pellucidum (CSP), which is said to be absent in ACC in most textbooks on fetal ultrasonography. Conversely, examples of the conditions that feto-maternal experts consider if the CSP is not seen on fetal imaging include: anencephaly, holoprosencephaly, septo-optic dysplasia, ACC, isolated absence of the CSP, any extensive destructive process of the brain (e.g., hydranencephaly), and destruction of the CSP secondary to raised intraventricular pressure, i.e., hydrocephalus.

The latter two scenarios imply that the septum pellucidum did form normally but was destroyed at a later stage, while the other developmental abnormalities usually implicate non-formation of the septum pellucidum. This point is highly relevant as the reverse of the argument is frequently applied in clinical practice, i.e., if the CSP is seen, the corpus callosum is assumed to be present. The purpose of this paper is to judge if that supposition is true.

The anatomy and embryology of the normal corpus callosum were reviewed previously [6]. If the corpus callosum does not form, the axons that should cross the midline remain in the ipsilateral hemisphere and run parallel to the interhemispheric fissure as the so-called bundles of Probst [7]. Those paired structures are thickest anteriorly, where they are comma-shaped and bulge into the medial part the frontal horns of the lateral ventricles. This produces the “steer-horn” deformity of the frontal horns of the lateral ventricles seen on coronal sections in cases of ACC. Colpocephaly or selective enlargement of the trigones and occipital horns of the lateral ventricles frequently occurs in ACC. The absence of the splenium of the corpus callosum results in the posterior portions of the ventricles being restricted only by loosely packed white matter, which allows expansion of the adjacent ventricles. This is another sonographic indicator that ACC may be present in a fetus although colpocephaly is found in other developmental abnormalities such as Chiari 2 deformities. In some cases, the corpus callosum may have formed in part but not in entirety. The distribution of abnormality in hypogenesis of the corpus callosum can be explained by the cranial to caudal direction of the development of the corpus callosum as shown on Fig. 1. The genu forms first [8] and is invariably present in hypogenesis, whereas the last forming parts (rostrum and splenium) are most commonly affected. Table 1 shows a classification system of commissural disorders proposed by Raybaud [9]. That classification recognizes that the theory of a cranio-caudal pattern of callosal hypogenesis is not perfect as, for example, type IB includes cases in which the genu is absent but the splenium and part of the body is present (although this is very unusual).
Fig. 1

A line diagram demonstrate the development of the midline commissural structures, particularly the cranio-caudal development of the corpus callosum. This explains the pattern of hypogenesis seen in the majority of clinical cases. (Modified from Ref. [11])

Table 1

Classification of commissural disorders (From reference 9).

I Classical commissural agenesis

A. Complete commissural agenesis

   Corpus callosum and hippocampal commissure absent

   Anterior commissure absent in 50%, present in 50% (small or normal)

B. Partial commissural agenesis

   Corpus callosum: typically the posterior portion is variably absent; rarely the agenesis may affect the genu instead of the posterior part of the corpus callosum

   Hippocampal commissure is absent

II Commissural agenesis with meningeal dysplasia

   Less common

   Partial or complete absence of the commissures

    Presence of multiloculated, non-communicating cyst within the meninges of the interhemispheric fissure, not communicating the ventricular lumen. Extensive deformity of the brain due to mass effect of the cyst

   Parencyhmal disorders such as dysplastic cortex bordering the cysts and masses of heterotopic gray matter

III Agenesis of a single commissure

A. Corpus callosum agenesis

  1 Complete agenesis

    Very uncommon

  2 Partial agenesis

   More common than complete agenesis

   The intermediate portion of the corpus callosum is lacking, whereas the anterior portion and splenium are present

 B. Hippocampal commissure agenesis

 C. Anterior commissure agenesis

IV Other commissural dysplasia

   Hyperplasia (associated with neurofibromatosis I)


V Septo-commissural dysplasia

A. Septo-optic dysplasia with partial commissural agenesis

    In 25% of cases of septo-optic dysplasia, a partial agenesis of the commissures is associated with the septal agenesis and the anterior optic defects

B. Septo-optic dysplasia with complete commissural agenesis or septo-commissural agenesis

    Very uncommon

    No bundle of Probst and no sheath of white matter closing the medial aspect of the ventricles. Instead, the ventricle is closed with tela choroidea, which may expand as an interhemispheric cystic cavity continuous with the ventricles

On superficial analysis, it appears reasonable to assume that agenesis of the corpus callosum must, by necessity, be accompanied by absence of the septum pellucidum. The argument would run: The septum pellucidum is an unpaired midline structure connected in normal circumstances to the undersurface of the corpus callosum as it crosses the midline—therefore, the septum pellucidum must be absent in cases of agenesis of the corpus callosum. We know, however, from information reviewed here and in the previous paper [6], that this argument cannot be correct. Three embryological facts are of particular relevance.
  1. 1.

    In “agenesis” of the corpus callosum, the crossing fibers are not actually absent, they merely stay within the ipsilateral hemisphere as the bundles of Probst.

  2. 2.

    The fornices are invariably present in cases of agenesis of the corpus callosum as the fornix forms much earlier in fetal life than the corpus callosum.

  3. 3.

    The septum pellucidum is a paired, parasagittal structure in the fetus linking the fornix to the axons of the corpus callosum.


On this analysis, the septum pellucidum should be present in ACC although not in its normal position. In order to study this further, we have investigated the relationship between the bundles of Probst and the fornix in cases of failed commissuration: Is there any structure that links the two that could be designated “septum pellucidum”? For this, we have reviewed 18 children who underwent MR imaging at Sheffield Children's Hospital for clinical purposes and were shown to have either classical commissural agenesis or commissural agenesis with meningeal dysplasia according to Raybaud's classification. This included eight cases of complete commissural agenesis (type IA), six with partial (type IB), and four with associated meningeal dysplasia (type II). All had the same high-resolution MR examination that included T1-weighted volume imaging acquired with 1.0-mm partitions acquired at 1.5 T.

Complete classical commissural agenesis

The course of the fornix in children with complete commissural agenesis appears to be remarkably constant, and one example is shown in Fig. 2. The commencement of the crus of the fornix from the tail of the hippocampus and its vertical passage appears to be relatively normal although the fornix travels more laterally than usual away from its partner. The fornix joins with the inferior aspect of the posterior part of bundles of Probst but remains clearly delineated from it.
Fig. 2

MR imaging of a child with agenesis of the corpus callosum. There is no evidence of any part of the corpus callosum on sagittal T1- (a) and axial T2-weighted images (b), which also show colpocephaly. Coronal T1-weighted images from the volume data set (cf) show the abberant passage of the fornix and the thickened leaves of the septum pellucidum (arrowed). This anatomical arrangement is consistent with the previously published work of Barkovich et al. (g; published with permission from Ref. [7])

The fornix remains connected to the Probst bundle underneath the everted cingulate gyrus by a thin leaf, and we believe this should considered to be the septum pellucidum on embryological grounds and is labeled as such in Barkovich's textbooks and peer-reviewed publications [10] (Fig. 3). As the fornix passes more anteriorly, the connecting stalk lengthens, and the fornix frequently is not in direct contact with the cingulate gyrus. At the level of the anterior commissure (which was present in all of our cases although frequently small), relatively normal anatomy was restored, and fibers connecting the two fornices were seen in the majority of cases, indicating a degree of belated commissuration (Fig. 2f).
Fig. 3

MR imaging of a child with hypogenesis of the corpus callosum. The sagittal T1- (a) and axial T2-weighted images (b) show that the genu and a small portion of the body are present. This constitutes the typical pattern of hypogenesis, compare with Fig. 4. Coronal T1-weighted images from the volume data set (cf) show a similar anatomical arrangement compared with agenesis of the corpus callosum posteriorly (c, d), but anteriorly (e, f), the leaves of the septum pellucidum are well formed under the small, poorly formed corpus callosum. The septum pellucidum/fornix complex is arrowed on cf

Partial commissural agenesis

This covers a wide spectrum of abnormalities. We included the mildest form (absence of the rostrum) in our first paper discussing normal and near normal appearances as the small number of cases we have seen have been in normal volunteers. At the other extreme, we have studied five cases in which the majority of the corpus callosum was missing, and we show the MR imaging of one child in which only the genu and a small portion of the body were present (Fig. 3). In all of those cases, the posterior trajectory of the fornix was the same as in ACC, the fornix contacting the bundle of Probst. As the fornix turns anteriorly, there was clear separation between the fornix and the bundle of Probst just behind the hypoplastic corpus callosum with thickened leaves of the CSP clearly shown (Fig. 3c). Those structures were also demonstrated underneath the hypoplastic corpus callosum (Fig. 3d) where the two leaves remain separate. The passage of the fornix around the anterior commissure appeared to be normal but the CSP was in continuity with the anterior inter-hemispheric fissure because of the absence of the rostrum.

Figure 4 shows a case of hypoplasia of the corpus callosum, one of the rare cases in which the genu is absent but the posterior parts of the corpus callosum are relatively normal. The path of the fornix out of the hippocampus and toward the splenium was relatively normal; however, it was difficult to recognize the fornix as separate from the splenium (unlike the normal situation). Anteriorly, at the point that the corpus callosum became deficient, the fornix was visually separable from the bundles of Probst and was connected to it by a stalk of tissue that we recognize as the CSP, as in cases of ACC. In the one case that we have been able to study, the anterior commissure was exceptionally hypoplastic, and the fornix appeared to enter the basal forebrain without the normal division that is usually seen on MR imaging. There was a bulky connection between the anterior parts of the fornices producing a very prominent hippocampal commissure, a feature described in Barkovich's earlier work [10].
Fig. 4

MR imaging of a child with hypogenesis of the corpus callosum. The sagittal T1- (a) and axial T2-weighted images (b) show that the posterior part of the corpus callosum is present but the anterior portion is not. This is a rare variety of hypogenesis of the corpus callosum. Coronal T1-weighted images from the volume data set (cf) show normal passage of the fimbria and fornix as far as the splenium of the corpus callosum (carrowed) but the fornix cannot be delineated as separate from the corpus callosum on (d). The fornix reappears underneath the bundles of Probst anteriorly (earrowed) and connects with its partner via a large commissural structure (farrowed)

Commissural agenesis with meningeal dysplasia

Such cases are characterized by the presence of inter-hemispheric cysts (which do not communicate with the ventricular system) and neocortical formation abnormalities. The precise path of the fornix in these cases is highly variable and posteriorly appears to be influenced by the anatomy of the interhemispheric cysts to a major extent (Fig. 5). The anterior portion, however, seems to be more constant but different from the classical commissural cases listed above. The fornix maintains a high-riding path as it courses cephalad and does not appear to give a post-commissural branch; instead, the fornix passes more anteriorly than usual before passing posteriorly to enter the basal forebrain.
Fig. 5

MR imaging of a child with commissural dysgenesis and meningeal dysplasia. The sagittal T1-weighted images (a) and axial T2-weighted images (b) show agenesis of the corpus callosum and multiple midline cystic abnormalities secondary to meningeal dysplasia. The fornices remain separate and asymmetric throughout their course (arrowed) due to the presence of the cysts (cf). Note the extensive abnormality of neocortical formation in the cerebral hemispheres anteriorly

Implications for fetal imaging

In our experience, problems relating to the corpus callosum are amongst the commonest referrals for iuMR. At our institution, we only take iuMR referrals when brain abnormalities have been shown (or are suspected) on ultrasonography. There are, therefore, three situations where ultrasonographic findings and iuMR findings can be compared:
  1. 1.

    ACC suspected on ultrasonography and confirmed on iuMR.

  2. 2.

    ACC suspected on ultrasonography but not shown on iuMR.

  3. 3.

    ACC shown on iuMR when the fetal brain was being investigated for reasons other than ACC.


Between 2004 and 2007, we performed 50 iuMR examinations on fetuses with ACC suspected on ultrasonography. ACC was confirmed by iuMR in 32 of 50 (64%) and refuted in 18 of 50 (36%). During the same period, we studied 136 fetuses referred for iuMR as part of a study of fetuses diagnosed with isolated ventriculomegaly on ultrasonography. ACC was present as an unexpected finding in 11 of 136 of cases (8%) and constituted the largest group of disagreements between the two techniques (manuscript in preparation). We believe that this indicates the breadth of the problem in our practice although we acknowledge that other groups maintain that their figures of misdiagnosis of ACC are lower (in some cases considerably lower or non-existent).

If we accept that iuMR has a role in the pre-natal diagnosis of brain malformations, and in particular, if it is helpful in diagnosing ACC, then we need to consider why. Current iuMR methods produce images with a lower anatomical resolution than ultrasonography but better contrast resolution in most situations. The ultrasonographic appearance of the CSP is highly characteristic, two hyperechoic stripes on either side of the midline and its presence is often taken to imply the presence of a normal corpus callosum. When sagittal images of the fetal brain cannot be obtained with ultrasound (e.g., in cases with unfavorable lie of the fetus or if the mother has a high body mass index), it will be difficult to obtain good views of the full length of the corpus callosum, but this is rarely a problem for iuMR.

We have shown in this article that the assumption of a normal corpus callosum from identification of a CSP is flawed and should be used with great caution. First, it should be appreciated that many cases of failed commissuration are hypogenesis of the corpus callosum, in which case some of the CSP will be present in a normal position. More importantly, our imaging of the pediatric brain confirms that the leaves of the septum pellucidum are present in ACC albeit in an unusual site. In some cases, the abnormally located leaves of the septum pellucidum and the fornices produce bilateral, parasagittal structures that could produce echogenic structures very similar to a “normal” CSP on fetal ultrasound. For example, Fig. 6 shows two cases of iuMR performed because of “isolated” ventriculomegaly on ultrasound but in both cases iuMR showed ACC (and other abnormalities). In one of those cases, the leaves of the septum pellucidum are seen in a relatively normal position, and this highlights a significant potential pitfall if the septum pellucidum is used as a surrogate indicator of normality of the corpus callosum.
Fig. 6

Coronal images from two fetal MR studies showing agenesis of the corpus callosum that was not suspected on ultrasonography. The case in (a) shows similar anatomy to the pediatric case of agenesis of the corpus callosum shown in Fig. 2 with the fornices closely applied to the bundles of Probst and the septum pellucidum not visible but presumably present but closely applied to the undersurface of the bundles of Probst. The case shown in (b) has the fornix hanging below the bundles of Probst (arrowed) and taking a discernible septum pellucidum with it. It is possible that those structures were shown on fetal ultrasonography, and normality of the corpus callosum was assumed because of that

In summary, review of the embryology of the cerebral commissures, septum pellicidum, and fornix does not reveal any reason why the septum pellucidum should be absent in cases of ACC. We have shown that children with ACC or hypoplasia of the corpus callosum have a structure that links the fornix and bundles of Probst, and that should be recognized as the deformed leaves of the septum pellucidum. It is highly likely that in some cases those structures would be detected on antenatal ultrasound leading to a false assumption that the corpus callosum is normal.

Conflict of interest statement

We declare that we have no conflict of interest.

Copyright information

© Springer-Verlag 2009