Posterior-Anterior Cephalometric Study of Neurofibromatosis Type 1 Patients With Facial Plexiform Neurofibroma: Analysis of Skeletal Symmetry Concerning Midfacial and Skull Base Reference Points (Zygomatic Arch, Mastoid, and Juga)

Discussion

This study described symmetry deviations of the facial skull in patients with a distinct tumor suppressor gene disease. The study focused on bilateral midfacial and skull base reference points. Most skeletal findings were topographically related to the facial spread of a characteristic neurogenic tumor, PNF. Despite the known highly variable phenotypes of NF1 patients, patterns of skeletal deformation of the facial skull are defined for the first time, which can be expected with statistically confirmed frequency in patients who have developed FPNF. The FPNF is typically developed unilaterally (5-8, 15). The study showed that the tumor-associated deformations in certain cephalometric measurement points may also alter the non-affected side. It seems plausible to assume that some trophic effects on the tumor side cause adaptive processes on the opposite side, for example, to compensate for tumor-related reduced muscular activity when chewing food. The studies confirm the concept of NF1 gene functioning both as a tumor suppressor and a histogenesis control gene (31).

NF1 and FPNF. The basis of the investigation was clinical and involved morphological findings of the face, which justified the group distinction of NF1 patients. The defining evidence for the diagnostic group “FPNF” is a characteristic neoplasm of the nerve sheath cells, the PNF, which is pathognomonic for the disease. In principle, PNF can develop in all regions of the body that are innervated by the peripheral nervous system. The assessment of PNF as a precancerous condition is well proven by numerous studies (3). In principle, MPNST can grow into neighboring tissues such as the bones and destroy hard tissues. However, transformation of FPNF to MPNST is rare (32). FPNF typically grows in a diffuse infiltrative pattern (33), destroys masticatory and mimic muscles and other organs to an extent that varies greatly from individual to individual (15). Frequently, FPNF is associated with deformities of adjacent bones (34-38), which can extend to dissolution of the bone adjacent to the tumor (39). In this respect, the skeletal changes often resemble the pattern of displacing tumors, which in individual cases leads to bone destruction (15, 39). However, incomplete bone development is also part of the spectrum of the disease, namely unilateral hypoplasia of the mandible or dysplasia of the sphenoid bone. These findings are related to FPNF (2). This destructive feature of the tumor, which is usually unilaterally manifested, has disastrous lifestyle consequences for many patients (9). Both soft tissue and skeletal deformities of the face having provoked generations of medical appraisal, and the number of individual reports with detailed descriptions of the often-grotesque phenotypes is incalculable. However, the hyperplasia of the affected facial region caused by the tumor only rarely corresponds to hyperplasia of the adjacent facial bones. Rather, hypoplasia of the facial skeleton associated with FPNF is the rule (40).

In this study, an attempt was made to demonstrate that the skeletal deformation of the facial skull follows a quantifiable pattern. For this purpose, patients who had both clinically developed FPNF and histological evidence of the specific tumor manifestation were selected. In the selection of patients, a distinction was made based on imaging and clinical assessment as to which parts of the face (unilaterally) had been affected by PNF. The examination did not differentiate between the variable tumor masses (volume). In addition, the assignment to the diagnostic group was based solely on tumor detection in the trigeminal dermatomes without further specification of the clinical findings, e.g., organ function. In most patients, the tumor had been known since childhood. However, the classification of the lesion as a congenital manifestation of the disease was not certain in every case. It was also not possible to identify growth phases of the tumor in the evaluation. This temporospatial variability in clinical expression offers the opportunity to identify a basic pattern of skeletal change.

Facial asymmetry and FPNF. The assessment of facial symmetry is a basic interpersonal assessment based on the bilateral symmetrical physical constitution of the human being and registers visible deviations from symmetry as an individual characteristic (41, 42). In NF1 patients, the question arises as to whether there is any visible deviation from facial symmetry beyond the known minor biological deviations from the geometric ideal of symmetrical body structure and methodological errors of measurement in the range of one to two millimeters in left/right comparisons (43-46), which may be associated with the disease. However, in the current study “difference of distances” is a question of calculated side and tumor-specific differences in the measurement points to reference planes. The tumor-associated skeletal change despite the variable skeletal characteristics proves the influence of the local tumors on the skeleton. The finding “facial asymmetry” in the diagnostic spectrum of the NF1 patient is clinically relevant. Facial asymmetry is a negative prognostic factor in NF1 patients (47). The criteria for this finding must therefore be disclosed to enable comparative investigations and to validate the statement.

The previous studies either describe a unilateral, tumor-related hyperplasia of the face because of the nerve sheath tumor and do not consider the skeletal component of the deformity (48) or focus on parts of the skull, especially the orbit and jaw (49-51). Metrics or angle measurements are rarely used to quantify facial asymmetries (15, 30, 36).

An early study on neurofibromatosis phenotype differentiates facial asymmetries caused by sphenoid dysplasia (4% of cases) from those summarily called “other skull anomalies” (19%) (52). However, these findings are not specified further.

An analysis of the United States National Neurofibromatosis Foundation (NNFF) International database on 1,728 patients (1,479 probands and 249 relatives) with NF1, registered facial asymmetry in 8% of probands and 3.3% of relatives. The diagnostic criteria of the diagnosis “facial asymmetry” are not mentioned. In the discussion of the results, biases of the NF1-associated findings are listed, which also included a suspected frequency of severely ill NF1 patients recorded in the register. For the diagnosis of sphenoid wing dysplasia, an X-ray examination is indispensable. Patients with facial asymmetry were probably X-rayed more frequently than other patients (frequency of sphenoid wing dysplasia in probands/relatives: 11.3%/3.3%). The frequency of facial asymmetry and sphenoid dysplasia in this large NF1 population is striking. A preferred selection of orbito-facial PNF was considered in the evaluation of NNFF data influencing the reported frequency of asymmetric patients (53).

A review of published geno-phenotype correlations in microdeletions of the NF1 gene describes “facial asymmetries” in 5/118 cases (4.2%) (54). The criteria for this finding are not reported. Interestingly, the authors also analyze the literature on skeletal anomalies in microdeletion patients. Changes in the skull are not included in the classification of skeletal anomalies (54).

In a study on the correlation of clinical findings and the probability of internal plexiform neurofibromas in NF1 children (n=357), facial asymmetries were registered in 6.4%. However, the criterion for assessing facial symmetry was not given (55). In a recent study on the geno-phenotype of NF1 patients with large deletions of the NF1 gene, 28% of the patients with facial asymmetries were even registered. However, the examination criteria for this judgment were not mentioned in this study either (56). As a rule, facial dysmorphism in these studies is reported as finding in a symmetrical face. The descriptions of the dysmorphia focus on the glabella/forehead and orbit region considering the genetic profile of the NF1 patient (57).

In summary, the information on the frequency of facial asymmetries in NF1 varies considerably. These differences may be due to the different inclusion criteria of the analysis, for example the selection of patients based on their genetic status. Some studies suggest that skeletal findings of the examination region were included in the assessment. However, the studies do not provide any comprehensible criteria for assessing how facial asymmetry was determined.

Method errors in the radiological examination. The basic requirement of the current investigation is the assumption of a bilaterally symmetrical development of the skull, which can be recorded under standardized radiological parameters in the posterior-anterior projection. For the evaluation of X-ray images of the skull of different provenance, deviations in measurement accuracy of around 2 mm are considered tolerable (so-called deviation from the ground truth of direct measurement on the object) (43-46). Deviations from facial symmetry are more common in the lower face than in the midface. The visual perception threshold of object asymmetries also lies in this value range (58). For this reason, we assessed the statistically significant left-right asymmetry of cephalometric measuring points in patients with facial DNF as a finding that is not a counterargument for the development of a symmetrical face in NF1 patients not affected by FPNF, provided that both the biological variability and the instrumental limitations of the measuring method are considered (30). The present study demonstrates the relationship between the infiltration and spread of FPNF and the change in cephalometric measurement points relative to two standardized reference planes that define the orientation of the skull to the horizontal plane and object symmetry relative to the median sagittal plane.

Limitations of study. The study reveals the bony changes on plain radiographs as deviations from two reference planes. Although PNFs are considered congenital tumors and the skeletal findings described here are often typical and diagnostic in the unilaterally manifest combination of soft and hard tissue lesions, the analysis cannot determine whether a bone alteration is a developmental disorder of the bones, induced by the tumor during bone formation, a passive process following the expansile and invasive tumor growth, or combinations thereof. However, this study clearly shows the association of severe jaw changes with the adjacent tumor. Bone remodeling due to pressure from adjacent benign tumors has been described in detail and suspected for NF1-associated bone deformities (13). On the other hand, invasive and bone-destructive growth of facial PNF has been documented in follow-up observations (39), confirming older reports that have detailed bone-dissolving capacities of PNF (59).

The two-dimensional radiological representation of the skull allows only indirect conclusions to be drawn about the position of relevant skeletal reference points in the sagittal plane. It is well known that FPNF in the midface can cause flattening of the zygomaticomaxillary complex and torsion of zygomaticomaxillary complex to the unaffected side. The distortion defines so-called facial scoliosis considerably contributing to facial disfigurement (15). More interesting is the finding that posterior-anterior cephalometry helps to identify changes in the skeletal outline of the midface, both on the affected and non-affected side of the skull.

The investigations show the change in the measurement points from the reference planes on the side affected by the tumor. The differences of measurements to the unaffected side, to the DNF group and to the control group, are very often statistically significant. However, the measured value differences are usually only a few millimeters apart and a clinical/visual abnormality of the findings cannot generally be asserted. This assessment is based on the inclusion criterion of the diagnostic group, that the extent of neurofibroma is based solely on the histological findings and topographical diagnosis, but not on the extent and size of the soft tissue lesion in the respective dermatome (or even tumor spread beyond the facial region). Despite variable manifestations of the disease in the individual diagnostic groups, the skeletal effect of the lesion can be demonstrated. Evidence for the variable size and extent of FPNF also reflects previous assessments of the facial phenotype, according to which the bulky, tumorous asymmetry of the face is regularly associated with skeletal hypoplasia of the affected side, in other words, bone dystrophy. Malformation of facial bones is frequently covered by the tumor mass. Physical inspection reveals facial hyperplasia on the tumor side due to soft tissue masses but usually ignores the hypoplastic skeleton to which the tumorous tissue is fixed (40).

Neurofibromin and bone. Several studies have shown that constitutive mutation of the NF1 gene impairs bone metabolism in NF1 patients. The gene product neurofibromin is apparently significantly involved in bone homeostasis. The early onset of osteoporosis in NF1 patients is thought to be a direct consequence of insufficient neurofibromin production (60). These effects affect the skeleton in general and thus early osteoporosis occurs symmetrically in the skeleton of patients. A local influence of osteoporosis on the phenotype is, for example, the early occurrence of bone fractures in NF1 patients who have not received vitamin D supplementation. An influence of the disease on general bone growth with disturbances of bilateral symmetry is observed when local homeostasis disturbances occur, for example in pseudarthrosis of long bones or deformation of the spine. In the updated diagnostic guideline for NF1, sphenoid wing dysplasia is no longer considered an independent skeletal malformation feature of the disease, but a bone formation disorder associated with FPNF (2). In a previous study, a tendency to developing a long face was determined in DNF patients. It is plausible to assume that the asymmetric osseous changes of FPNF patients presented here resulted from interactions between the local tumor and bone. However, when assessing the skull, the morphology of the brain and the influence of the constitutive mutation on brain development with its internal shaping of the skull must also be considered.

Zygomatic arch. For PNF developed in and around orbit, the working term orbital/periorbital plexiform neurofibroma (OPPN) was introduced to capture the variable phenotype in a diagnostic group (17). The variable spread of OPPN corresponds to the variable facial spreads of the tumor in the supply area of the trigeminal nerve and results in the differentiation of individual FPNF subgroups of this study. In these tumor manifestations, the development of the adjacent brain lobes is frequently altered, whether by arachnoid cysts, displaced and/or enlarged brain regions or atypical cortical structures (14). These local changes in the brain may also be associated with changes in the cranium (61). In several cases, it was observed that the skull base on the affected side extended further caudally than that on the unaffected side. Associated with this may be a malposition of the cranium including its processus, for example the mastoid process or both temporal and zygomatic processes constituting the ZA (62). The study presented here shows that in the DNF group, such potential cerebral changes did not affect the transversal dimension of the measurement point ZA to the M-plane, which noticeably influenced the symmetry [but reveals a tendency for long face (30)]. In contrast, statistically significant changes in this dimension were demonstrated for FPNF. This change, often known as a noticeable deformation (and narrowing) of the ZA, is well documented. This study specifies the associated change in the ZA as a tumor-associated bone finding with statistical certainty in a study group despite the highly variable clinical expression of FPNF. The statistically significant increased distance of the ZA to the Z-plane on the affected side in patients with hemifacial PNF very likely is the result of the tumor manifestation in these patients. The tumor infiltrates and in many cases destroys the musculature of the masticatory muscles. The predominantly passive attachment of the soft tissue to the ZA causes a caudal distortion of the bone in accordance with the effect of gravity. Accordingly, the results of the relationship between the measurement point ZA and the median-sagittal plane show that patients with hemifacial PNF are also primarily affected. However, the shorter distances on the unaffected side are particularly noticeable here, which can contribute to the midface appearing somewhat narrower. This effect can be demonstrated for both hemifacial NF1 patients and those affected in the 1^st and/or 2^nd branch compared to the DNF group.

At first sight, the changes of the measuring point ZA in patients with FPNF limited to the 1^st or 2^nd branch fit into the concept of local interaction between tumor and adjacent bone. Measurement results that cannot determine this relationship with statistical certainty could be explained because of the variable tumor spread in individual cases and the low number of cases of a subgroup in individual evaluations. However, clinical assessment of tumor spread is not proof of tumor boundary determination despite surgical treatment and histologic diagnosis confirmation performed in almost all cases. FPNF expansion may also escape magnetic resonance tomography, the gold standard for imaging neurofibromas (63). This reference is important because it cannot be excluded with complete certainty that relevant portions of the third trigeminal branch of the same side may have been tumor-altered in cases with apparent extension confined to the first and/or second trigeminal branch. In these cases of unrecognized affection of the third branch, the barely musculature could also have been affected and thus the musculoskeletal unit altered, so that the position of the measuring point ZA has been changed.

Mastoid. In this study, the cephalometric measuring point ‘mastoid’ is a stable reference value positioned symmetrically to the median sagittal reference plane. The slightly more caudal position of the mastoid tip in DNF patients recorded as symmetrical may be a partial aspect of the tendency to the long face of NF1 patients (30). It is remarkable that the diagnosis group FPNF had only minor impact on the comparisons. In this study group, with focus on FPNF, patients could be included whose tumors extended further dorsally, transgressing the trigeminal dermatomes, and were in topographical relation to the lateral skull base. However, these patients were rare, and this characteristic had only minor influence on the measuring point mastoid. On the other hand, a discrete narrowing of mastoid to M plane was recorded in some PNF subgroups. It can be concluded that in this study, potential structural cerebral changes (61) very likely had no influence on the lateral measuring point of the skull base.

Juga. The measuring point ‘juga’ hardly changed on the affected side of the FPNF group compared to the control group and DNF group. On the other hand, on the contralateral side, both an approximation of the measurement point to the M-plane and a cranialization relative to the Z-plane could be noted. This change was likely a skeletal adaptation to tumor-induced facial scoliosis, which can cause a helical twisting of the midface.