The cerebrofacial metameric syndromes: An embryological review and proposal of a novel classification scheme

Abstract

The cerebrofacial metameric syndromes are a group of congenital syndromes that result in vascular malformations throughout specific anatomical distributions of the brain, cranium and face. Multiple reports of patients with high-flow or low-flow vascular malformations following a metameric distribution have supported this idea. There has been much advancement in understanding of segmental organization and cell migration since the concept of metameric vascular syndromes was first proposed. We aim to give an updated review of these embryological considerations and then propose a more detailed classification system for these syndromes, predominately incorporating the contribution of neural crest cells and somitomeres to the pharyngeal arches.

Introduction and background

Cerebrofacial metameric syndromes (CMS) are congenital disorders caused by embryological defects in craniofacial vasculogenesis, which results in vascular malformations within predictable anatomical patterns of the brain, cranium and face. ¹ Syndromic presentation can manifest with either high-flow or low-flow malformations, termed cerebrofacial arteriovenous metameric syndrome (CAMS) or cerebrofacial venous metameric syndrome (CVMS), respectively. High-flow vascular malformations in CAMS are most commonly arteriovenous malformations (AVM), though arteriovenous fistulae in context of presumed CAMS have also been described. ² Low-flow vascular malformations in CVMS include developmental venous anomalies (DVA), capillary malformations (CM), and cavernous malformations.^3–5 When discussing the common features of CAMS and CVMS, we will refer to both using the term CMS.

Given significant imaging and management implications, recognition of CMS by the neuroradiologist is important. We aim to give an updated review of the segmental organization of the brain and face and propose a more detailed classification system for metameric distribution of vascular malformations. A discussion of various subtypes of both CAMS and CVMS will follow in the hope to better inform the neuroradiologist of this less-ubiquitous group of syndromes as they are presently defined in the literature.

History of cerebrofacial metameric syndrome (CMS)

The story of CMS begins in 1874 when Magnus first described a retinal arteriovenous malformation, which was thought to be an isolated ophthalmological occurrence. ⁶ Although Yates and Paine ⁷ would describe a case of concurrent retinal and cerebral AVM in a single patient in 1930, an association between facial, retinal, and brain AVMs would not be proposed until 1937 by Bonnet, Dechaume and Blanc, ⁸ who reported two similar cases.Wyburn-Mason would add nine additional cases in 1943. ⁹ The condition of concomitant facial, retinal and cerebral AVMs became known as either Bonnet–Dechaume–Blanc syndrome or Wyburn-Mason syndrome and is considered a neurocutaneous disorder. ¹

In a case series and review in 2001, Bhattacharya et al. ¹ suggested that vascular malformations occurring within certain regions in the brain and face may be based on defect-affected specific segments or “metameres” in embryonic development. The concept of CMS was proposed. Shortly thereafter, this same classification system was illustrated with low-flow vascular malformations with similar anatomical patterns. ¹⁰ Several case series of patients with cerebrofacial vascular malformations in metameric distributions have been reported, further strengthening an underlying embryological basis for pathogenesis.^3,11–13 There has been much advancement in the understanding of segmental organization and cell migration since the concept of metameric vascular syndromes was first proposed. Details of the rostro-caudal segmental organization of the vertebrate brain have been advanced with fate-mapping, progeny analysis, and studies of developmental gene expression.

Embryological basis for cerebrofacial metameric syndromes (CMS)

Cerebrofacial development is a complex orchestration of orderly, rapid, and often simultaneous processes involving ectoderm, endoderm, mesoderm, and neural crest. Brain and face morphogenesis has been extensively studied in multiple species including avian models, mouse and zebrafish; these data have predominately been extrapolated to humans as patterns of development are highly conserved across species. In this section, we will review the embryology of the brain and face and also highlight recent changes.

During gastrulation the process of cell differentiation begins, forming the three primary germ layers around the 16–17th day of gestation. Subsequently, on approximately day 18, a thickened area of dorsal ectoderm forms the neural plate and undergoes infolding to become the neural tube. Mesodermal cells flanking the neural tube, termed paraxial mesoderm, condense into loose masses of tissue called somitomeres at the end of the third gestational week. Three main divisions of the brain (prosencephalon, mesencephalon, and rhombencephalon) become distinguishable along the neural tube at this time. The seven (cranial) somitomeres are situated along the caudal forebrain, the midbrain, and the cranial hindbrain. They remain incompletely partitioned, never forming into somites.^14,15

Shortly after gastrulation begins, neural crest progenitor cells can be identified along the lateral margins of the neural plate. Neural crest cells (NCCs) are transient and multipotent stem cells that are crucial to brain, head, and neck development with cellular derivatives including neurons, glial cells, craniofacial skeleton, smooth muscle, skin pigment cells, and a number of ocular/periocular structures. Emerging from the dorsal neural tube, cranial NCCs begin migrating to their destinations before tube closure and progress along pathways predominately between outer ectoderm and underlying mesoderm. Arising in close proximity to the neural crest at the border of the neural plate are precursors to the neural sensory placodes. ¹⁶ The patterns of the NCC streams and their eventual differentiation have been found to be extremely flexible and highly dependent on complex environmental signals.^17–19 Reciprocal cell–cell interactions between the neural crest and placode cells help to drive head and face morphogenesis. ¹⁶

By the fourth week, the three divisions of the neural tube have further formed transient segments or neuromeres. It is currently recognized that the mesencephalon has two mesomeres, which are separated from the hindbrain by an isthmus (designated r0) with the rhombencephalon caudal to this having 11 rhombomeres (r1–r11). ²⁰

During this same time, the first pharyngeal arch (PA) begins to form, followed by the second and the third. The process is rapid and transient such that the first two arches are already indistinct by the time arches 4 and 6 develop. ¹⁴ Each arch has a surface covering of ectoderm, a core of mesenchymal tissue derived from neural crest, and an inner covering of endoderm. Multiple series of segments become evident in the developing brain and face with somitomeres situated alongside the neuromeres and PAs situated alongside the somitomeres (Figure 1). By the fifth week of gestation, the three main neural tube divisions begin to further differentiate with the prosencephalon developing into telencephalon (future cerebral hemispheres) and diencephalon and mesencephalon remaining as midbrain. In the rhombencephalon, the cerebellum arises from r0 and r1, the pons proper arises from r2 to r4 and the medulla arises from r7 to r11. In a recent paper proposing changes in brain stem nomenclature based on developmental gene patterns, Watson et al. ²⁰ explain that structures from rhombomeres 5 and 6 (which include abducens and facial nuclei) should be designated as ‘retropontine’. For the purposes of this paper, we will continue to refer to this as dorsal pons. Watson et al. also explain that the midbrain has been found to possess similar gene expression patterns as the diencephalon and hypothalamus and embryologically appears distinct from the hindbrain. ²⁰

Figure 1.

Simplified schematic of the segmental arrangement of the developing head and neck. The neural tube (central structures in tan) is divided into neuromeres. The two mesomeres of the mesencephalon are designated m1 and m2, rhombomeres of the rhombencephalon are divided into r1 to r11 (only r1–r8 are shown here) and the isthmus is labeled as r0. Somitomeres (shown in pink) arise from paraxial mesodermal cells, which flank the neural tube. While some authors leave the paraxial mesoderm unsegmented, many have designated seven pairs of somitomeres, which form the head mesenchyme. Both neural crest cells (NCCs) and somitomeres contribute to the pharyngeal arches (shown in gray). Neuromeres and pharyngeal arches appear to be the primary drivers of segmentation in the head. Neural crest from posterior mesencephalon and the upper metencephalon (r1–r3) as well as the fourth and fifth cranial somitomere supply the first pharyngeal arch. Neural crest from the upper myelencephalon (r4–r5) and the sixth cranial somitomere migrate to the second pharyngeal arch. Neural crest from the mid-myelencephalon (primarily r6) and seventh cranial somitomere migrate to the third pharyngeal arch and neural crest from the lower myelencephalon (r7–r8) and second–fourth occipital somites migrate to the fourth pharyngeal arch.^21,22

Neural crest migration streams have been tracked to their destinations. Relatively few cranial NCCs originate from the anterior prosencephalon, whereas an abundance of NCCs from the diencephalon and rostral mesencephalon migrate into the frontonasal process. These cells migrate around the optic stalks and olfactory placodes within the prosencephalon. In the face, this distribution of cells corresponds to the forehead, periorbital region, nose and philtrum. ²³

At the hindbrain level, NCCs migrate in streams predominately out of the segmented rhombomeres to colonize the PAs. The first stream primarily involves cells from the posterior mesencephalon and first two rhombomeres (r1 and r2) and colonizes the first PA, the second stream primarily arises from r4 and populates the second PA, and the third stream arises in r6 and 7, which later separate to colonize the third and fourth PAs.^17,24–26

Embryonic folding allows first and second arches to merge with fronto–orbito–nasal prominences. Structures that arise include hypothalamus, thalamus, hypophysis, optic nerve, retina , and eye. Mesodermal cells from the somitomeres also stream into PAs, contributing to the striated muscle of the face and neck, though the pattern of muscle formation appears to be controlled by the cranial neural crest. While somites drive segmental patterns in the trunk, NCCs of the neuromeres and pharyngeal pouches appear to be the major driving force in segmental patterning in the head with the somitomeric mesoderm following their lead.^21,27,28

The structures of the PAs are well known. Derivatives of the first PA include muscles of mastication, mylohyoid and anterior belly of the digastric, maxilla and palate, mandible, incus and malleus of the ossicular chain, trigeminal nerve, maxillary artery, and external carotid artery. Derivatives of the second PA include the muscles of facial expression, the stapes, part of the hyoid bone, facial nerve, and ascending pharyngeal and stapedial arteries. Derivatives of the third PA include the stylopharyngeus muscle, part of the hyoid, glossopharyngeal nerve, and internal and common carotid arteries. The fourth and sixth arches contribute to the cricothyroid muscle, thyroid and laryngeal cartilages, and portions of the vagus nerve. The tongue notably arises from multiple PAs with the anterior two-thirds arising primarily from the first PA, the dorsal one-third arising from the third PA, and epiglottis and adjacent regions arising from the fourth PA.

There is unique encoded information in the crest cells by Homeobox (Hox) genes and other developmental genes such as DLX involved in patterning of the PAs.^29,30 Additionally, there are numerous molecules involved in neural crest migration, including those assisting with cell–matrix interactions, chemotaxis and chemokinesis, and distinguishing NCC route margins. ²⁶ In depth detail of these molecules is beyond the scope of this paper; however, several genes are integrally involved in successful rostrocaudal segmentation of the central nervous system and deserve mention. Pax 6 is expressed in the alar diencephalon, Otx2 in the forebrain and midbrain, Gbx2 in the rostral hindbrain, Fgf8 in the isthmus, and Hox genes are expressed from r2 to the caudal end of the spinal cord. ²⁰

The developing brain is supported by the collaterally occurring development of brain vasculature. Perineural plexuses initially surround the neural tube and then begin sprouting radially to form the subventricular vascular plexus. This process does not simply occur in response to increased parenchymal mass and oxygen demand but follows a similar pattern as telencephalic brain regions, proceeding along a posterior-to-anterior and ventral-to-dorsal gradient. Pericytes are multi-functional mural cells of microvascular capillaries, which are critical for the regulation of angiogenesis, maintaining the integrity of the blood–brain barrier and vessel stability.^31–34 These cells can differentiate into and are often considered synonymous with vascular smooth muscle cells. Patients with arteriovenous malformations have been found to exhibit reduced brain vessel coverage by pericytes, contributing to vascular instability. ³¹ Quail-chick chimera analysis and lineage-tracing experiments have demonstrated that neural crest contributes to pericytes destined for the face, forebrain, retina, neck tissues, and thymus, thyroid, and parathyroid glands while mesoderm-derived pericytes supply the midbrain, hindbrain, and spinal cord.^34–36 These two systems contact at different points, including the circle of Willis anastomoses. ³⁵ The endothelial cells of cephalic vasculature, in contrast, derive from the mesoderm.^1,37 It should also be noted that cranial NCC contribute to cardiac structures, including the tunica media of the great vessels, and also help orchestrate septation of the outflow tracts.^38–40 Preotic NCCs have also been shown to contribute to coronary artery smooth muscle. ⁴⁰ Vessel formation requires interaction between neuroepithelium, endothelial cells, and extracellular matrix including stimulation of endothelial cells by vascular endothelial growth factor (VEGF) or Wnt secreting by the neuroepithelium. Factors that influence this process remain to be completely elucidated, though Fox1C is a transcription factor that has been found to play an essential role in early stages of vascular formation in the telencephalon.^41,42 The Notch cell surface receptor regulates cell proliferation and differentiation; deficient Notch signaling in pericytes has been found to result in arteriovenous malformations. ³² Disruptions in any of these processes or gene expressions could conceivably cause alteration in vascular development. Prior to cellular migration from each metamere to its respective distribution, an insult may result in defects during formation of the vascular system. ¹ Dysfunction in neural crest and mesodermal cells within a specific metamere may manifest as downstream vascular malformations. ⁴³ It is also conceivable that an insult occurring at an earlier developmental time point may involve more than one cephalic metamere, thereby resulting in a phenotype that involves multiple CMS subtypes.^3,10

A New classification system

Based upon an updated understanding of segmental organization of the vertebrate brain and face, we propose a more detailed classification system (Figure 2). This essentially divides the prior CAMS/CVMS 3 into three distinct subtypes, taking into consideration the rhombomeric contribution to hindbrain structures and the contribution of NCCs and somitomeres to the PAs. The first two subtypes remain similar to the initial proposition of structures arising in medial or lateral distributions of the prosencephalon. CMS 3 includes cerebellum, ventral pons and 1st PA structures, CMS 4 includes pons and 2nd PA structures and CMS 5 includes dorsal pons, upper medulla, and both third and fourth PA structures. Although more detailed, these categories provide a generalized concept of metameric development, knowing that some overlap occurs. For example, given the minute contribution of NCC from rhombomere 3 to the first and second PAs and small supply from rhombomere 5 to the second and third PAs, vascular malformations could conceivably occur within the pons in CMS 3, 4, or 5.

Figure 2.

Illustration of the proposed classification system for cerebrofacial metameric syndromes.

Features of cerebrofacial arteriovenous metameric syndrome (CAMS)

The clinical features of CAMS occur as the result of vascular malformations that correspond to the specific metamere(s) involved. In general, the intracranial manifestations observed in each subtype of CAMS may be secondary to altered hemodynamics as a result of the AVM (i.e. steal phenomenon), or secondary to compressive dysfunction of adjacent neural structures, particularly in cases of larger AVMs. Hemorrhage may also occur, though may be less commonly observed than in non-syndromic AVMs. ⁴⁴ It is likely that the high-flow nature of CAMS leads to earlier symptom manifestations, and hence younger ages at initial presentation are more common. Furthermore, the relatively severe symptoms that are evident at presentation (i.e. visual loss) are also likely explained by the presence of disruptive, high-flow vascular malformations.

Each CAMS subgroup may have overlapping clinical features, though subtle distinctions in symptomatology may exist based on the unique anatomical distribution and are therefore key in recognizing each subgroup of CAMS. Further complicating the nature of CAMS is the potential for overlapping phenotypes in cases of involvement of multiple metameres. It is also of note that the AV shunts associated with CAMS may appear metachronously rather than at the same time.

In the initial classification scheme, CAMS1 corresponded to the medial prosencephalic group; this is unchanged in our proposed scheme. This group therefore may include clinical presentations related to the nose, forehead, hypothalamus, corpus callosum, and hypophysis. In cases thought to be isolated CAMS1 occurrences, epistaxis is a commonly observed presentation given the presence of nasal AVMs.^1,44 Intracranial AVMs in CAMS1 that involve the hypothalamus and hypophysis may conceptually result in dysfunction of these two structures. ⁴⁵ However, though the optic apparatus may not be directly involved in CAMS1, compression of the adjacent optic chiasm or optic nerve—for example related to venous pouches—may result in visual disturbances. ¹

The previous CAMS2 group included the temporal, parietal, and occipital lobes, optic nerve, retina, thalamus, eye, cheek, and maxilla. One notable difference in our proposed scheme is that first arch structures including cheek, maxilla, and palate are a part of CAMS3. Visual complaints, in CAMS2 and in all CAMS subtypes, have been the most commonly observed presenting complaint as demonstrated here. ⁴⁴ Such complaints in the context of CAMS2 may result from direct involvement of the retina, or may result from involvement of the optic nerve, optic chiasm, or occipital cortex.^1,12,46,47 Other, more rare presentations in presumed CAMS2 include hemiparesis and convulsions.^1,44

The previously described CAMS3 included all hindbrain structures. Besides cosmetic complaints resulting from AVMs in the affected areas, patients presenting with ataxia as a result of hindbrain involvement, as well as gingival hemorrhage, have also been reported.^1,48 To our knowledge, three prior cases of isolated CAMS3 have been documented, suggesting isolated CAMS3 is a relatively rare occurence.^49–51

Patients presenting with mixed CAMS phenotypes are more common than patients with isolated CAMS subtypes. Clinical and imaging findings of such patients may consist of multiple of those seen in the individual CAMS phenotypes. Examples of CAMS1 and CAMS 1 + 2 are found in Figures 3 and 4, respectively. Figure 5 demonstrates an example of CAMS 3 and 4, and Figure 6 demonstrates an example of CAMS 3–5.

Figure 3.

CMS 1. Axial contrast-enhanced CT images of the brain demonstrate a midline arteriovenous malformation (AVM) involving the sella and suprasellar cistern (a, b) with involvement of hypophysis and hypothalamus as well as corpus callosum (c). Diagnostic external carotid artery angiogram (d) shows a high-flow nasal AVM fed by facial and maxillary artery branches.

CT: computed tomography; AVM: arteriovenous malformation; CMS1: cerebrofacial metameric syndromes1.

Figure 4.

CMS 1 and 2. Case 1: A 17-year-old female diagnosed with the Wyburn-Mason syndrome who presented with left eye blindness. Coronal angiographic Time-Resolved Imaging of Contrast Kinetics image (a) shows an AVM within the suprasellar cistern (arrow) and in the region of the left optic canal. Coronal T2W image (b) further shows the extent of this lesion into the third ventricle and left thalamus. Left internal carotid artery angiogram (c) shows extension along the left optic pathway with a supply from an enlarged left ophthalmic artery. Case 2: Young adult male with an extensive right facial AVM involving the buccal, nasal, and orbital region and right visual deficit. Coronal post-gadolinium T1-weighted image (d) shows infiltration of the medial right orbit (arrow). Sagittal pre-gadolinium T1W image (e) shows additional AVM in the hypothalamic/chiasmatic region (arrow), which extended to the third ventricle and thalami. The patient was subsequently diagnosed with complete Wyburn-Mason syndrome. Post-contrast CT (f) years later show persistent right face and nasal AVM. There has been right eye enucleation.

AVM: arteriovenous malformation; CT: computed tomography; CMS: cerebrofacial metameric syndromes

Figure 5.

CMS 3 and 4. Axial post-contrast CT image (a) shows a portion of a residual AVM in the left cerebellopontine region (arrow). There has been a left occipital craniectomy for resection. Both axial and coronal post-contrast CT images (a and b) show a large left preauricular component. Lateral view from a left vertebral artery angiogram shows the posterior fossa malformation with arterial feeders arising from left posterior cerebral artery branches. Coronal view from a superselective angiogram of the posterior auricular artery redemonstrates the high-flow AVM in the preauricular region. There is an extension to the left mandibular angle. Given potential involvement of both first and second pharyngeal arches with involvement of the preauricular/periparotid region as well as the cerebellopontine region, findings are most consistent both CMS 3- and 4 distribution.

CMS: cerebrofacial metameric syndromes 3 and 4; CT: computed tomography; AVM: arteriovenous malformation;

Figure 6.

CMS 3–5. A 16-year-old female with transspatial cervicofacial slow-flow vascular malformations. Axial T2W images with fat saturation through the lower face and neck (a–c) show mixed venolymphatic malformation which involves first PA structures including muscles of mastication, mylohyoid, anterior belly of the digastric, mandible and anterior two-third of the tongue, second PA structures including facial nerve branches and hyoid bone, third PA structures including dorsal one-third of the tongue and part of the hyoid and fourth PA structures including valleculae, lingual surface of epiglottis, preepiglottic fat, and laryngeal cartilages. Axial T1W post-gadolinium images (d–f) demonstrate a complex developmental venous anomaly (arrows) involving the right greater than left pons, cerebellum and inferior midbrain with drainage into right perimesencephalic, anterior pontomesencephalic and transverse pontine veins. A sagittal T1W post-gadolinium image (g) shows both the facial and brainstem vascular malformations (arrows), partially visualized. Given the distribution, findings would be compatible with CMS 3–5.

CMS 3–5: cerebrofacial metameric syndromes 3–5; PA: pharyngeal arch.

Features of cerebrofacial venous metameric syndrome (CVMS)

In contrast to the high-flow AVMs and AVFs of CAMS, CVMS consists of low-flow vascular malformations, including facial capillary or venous malformations, DVAs, dural sinus malformations, and cavernous malformations.^3,11 The anatomical distribution of vascular malformations depends on the subtype of CVMS; as with CAMS, patients present with combinations of facial and intracranial vascular malformations. ³ In contrast to CAMS, acute hemorrhage is exceedingly rare in CVMS, but have been reported in the context of a cavernous malformation. ³

Similar to CAMS, multiple groupings of CVMS are commonly observed. Sturge–Weber syndrome (SWS) is the best-known example. Based on prior observations that port wine stains generally occur in the V1 and V2 trigeminal nerve distributions, they have been classified according to facial neural innervation. However, abnormalities of SWS are more likely traced back to a disruption of the embryonic vascular plexus. As the neural crest contributes to pericytes destined for the face and forebrain as well as to nerve sheaths, the distribution of findings is likely related to territories corresponding to NCC migration.^34–36 In their study reviewing records of 192 children with facial port wine stains, Waelchli et al. looked for patterns in the malformations that were seen in conjunction with cerebral and ocular manifestations of SWS as well as clinical outcomes including seizure and glaucoma. ²³ They conclude that the most reliable predictor of SWS was found using embryologic vascular distributions as opposed to trigeminal nerve innervation. Port wine stains involving any part of the forehead was highly predictive of patient having an abnormal magnetic resonance imaging and adverse clinical outcomes. Notably, their definition of the forehead was demarcated inferiorly by a line from the outer canthus to the upper ear and incorporating the upper eyelid, a region which involves all three trigeminal nerve divisions. ²³ The authors point out that the frontonasal prominence and optic vesicle area make up this region; as described above, these structures are formed from NCC migration from the anterior prosencephalon, diencephalon, and rostral mesencephalon. Simultaneous occurrence of malformations involving the forehead, brain and eye indicates an early mutation affecting the NCC.

Adding to this, many cases of Sturge–Weber syndrome extend beyond the territories of the trigeminal nerve with capillary malformations involving neck, upper chest, and back, though are still within the distribution of neural crest migration streams. ¹⁰ In addition to venous capillary malformations of the face, the impaired superficial cortical venous drainage characteristic of SWS leads to low-flow angiomatosis involving the leptomeninges, enlarged transmedullary veins in that region, and engorgement of the ipsilateral choroid plexus. ⁵² The venous capillary malformations of the face most commonly encompass a distribution consistent with CVMS1 + 2, though mandibular distributions consistent with CVMS3 occur less frequently.^10,11

Outside of SWS, a wide variety of CVMS presentations have been described with various metameric distributions. In a review of 63 patients with cervicofacial venous malformations, 36.5% were found to have DVAs and 9.5% had dural venous sinus ectasia. In a majority of these patients (73.9%), intracranial venous anomalies were within the same metamere as facial venous abnormalities and in 82.6% they were ipsilateral. Cavernous malformations are relatively rarer but reported within metameric distributions matching the facial venous malformations. ³

Implications and future research

The primary avenue of future study related to CMS involves elucidating the potential metameric insults that take place prior to cellular migration. How these potential insults result in defects in vascular formation remains uncertain, though several genes are of interest. Hox genes play an important role in signaling the appropriate embryonic segmentation for brain and craniofacial development.^30,50 Specific Hox genes have also been shown to play important roles in endothelial and capillary development, potentially implying their involvement in CMS. ⁵⁰ Other genes with well-established functions, including GNAQ (associated with Sturge-Weber) and PIK3CA, may also play an underlying role in CVMS pathogenesis. ³

Conclusions

Since its original description in 2001, the concept of CMS has continued to be reinforced by several cases of patients with vascular malformations in metameric distributions. Clinically recognizing patients who may have CAMS or CVMS is important given that such patients likely require neuroimaging to characterize the extent of their disease. In patients with CAMS, recognizing intracranial high-flow vascular malformations is particularly important given that such malformations may require intervention. Our understanding of the potential genetic underpinnings of the CMS remains in its infancy, though several potential genetic players are intriguing targets for further research. Given the rarity of the CMS, reporting of such cases is crucial to gain a more comprehensive pathological understanding, in addition to providing confirmation of the novel classification scheme presented here.