Abstract
Neurofibromatosis type 1 (NF1) is an autosomal dominant inherited disease. Some stigmata of NF1 occur in the orbital region. The aim of this study was to reveal whether alterations of the orbital rim visible on plain radiographs may indicate the presence of a plexiform neurofibroma (PNF), a tumour almost exclusively diagnosed in NF1. Material and Methods: The plain orbital radiographs of 73 patients with NF1 (female: N=37, male: N=36) were investigated for alterations of the orbit. The group was further distinguished according to the presence of orbital PNF (N=53) and/or sphenoid wing dysplasia (N=30). Radiographs from patients with NF1 and with exclusion of PNF in the orbitofacial region were used for comparison (N=20). A special cephalometric analysis (Dental Vision™) was adapted to the demands of this study. Results: Patients with NF1 not affected by an orbitofacial PNF exhibited symmetrical orbits. Unilateral increase in orbital height was associated with ipsilateral PNF. The width of orbits affected by a PNF was often slightly increased compared to the non-affected side. The determination of cephalometrically-defined angles disclosed an erection of the PNF-affected orbit compared to the medio-sagittal plane. Conclusion: Plain radiographs are often the first diagnostic measure used to determine skeletal alterations. This study shows that certain parameters of the orbital rim are useful indicators of a PNF in patients who are unilaterally affected by this lesion in the orbital or orbitotemporal region.
- Orbit
- orbital rim
- orbital neurofibromatosis
- NF1
- plexiform neurofibroma
- sphenoid wing dysplasia
- orbital dysplasia
- plain radiographs
- cephalometry
Neurofibromatosis type 1 (NF1) is an autosomal dominant inherited disease affecting approximately 1:2500 to 1:3500 children at birth (1). About 50% of patients with NF1 have parents not affected by NF1 stigmata and are believed to represent cases arising from de novo mutations (2, 3). The gene for NF1 is located on chromosome 17q11.2 (4). In most cases diagnosis is clearly confirmed according to established clinical diagnostic criteria (5). Neurofibromas, i.e. benign nerve sheath tumours, are the hallmark of this disease (5). However, NF1 is also a bone disease (6, 7). In the skulls of patients with NF1 sphenoid wing dysplasia may be diagnosed and is a defining diagnostic feature of this entity (5). The prevalence of sphenoid wing dysplasia is not known (8, 9) and appears to be associated with other osseous abnormalities in the context of NF1 (8, 9).
Sphenoid wing dysplasia can be an isolated finding or be found in association with orbital or orbitotemporal plexiform neurofibroma (PNF) (10-16). Other skull base malformations have also been found in cases with sphenoid wing dysplasia of patients with NF1 (17-19). Increase of orbital volume and diameter were reported to be associated with sphenoid wing dysplasia (20, 21). Alterations of the orbit can severely impair the physical appearance of affected individuals (22-27).
The aim of this study was to analyze plain radiographs of the orbital region of patients with NF1 for alterations of the orbit associated with pathognomonic features of this disease.
Materials and Methods
Patient groups. The material of this retrospective study was occipitonasal skull plain radiographs collimated to the region of interest of 73 patients with established diagnosis of NF1 (5). The term ‘orbit’ refers to an exclusively skeletal structure throughout this article. The orbital radiographs were performed between 1968 and 2008. Out of this collective, 53 patients were affected by unilateral PNF of at least the first or second, or both branches of the trigeminal nerve. The PNF group denominates patients with facial (predominantly orbitotemporal) PNF. Patients with hemifacial PNF, i.e. PNF affecting all three trigeminal nerve branches, were also included in this group. A further 20 patients were affected by disseminated cutaneous neurofibroma (DNF) only. The DNF group comprises of patients with multiple neurofibroma nodules that may be found everywhere on the integument, including the skin of the orbital region. These patients had no PNF. Gender was equally distributed in the whole group (females: 37; males: 36). Age at the time of first radiograph of the orbital region was 16.7 years (range 1 year - 38 years) in females and 21 years (range 1 year - 53 years) in males.
In the orbital region, several pathologies can be found that are diagnostic of NF1 and can be relevant to orbital dimensions, e.g. dysplasia of the sphenoid wing and PNF. Therefore, the two groups were sub-divided according to the presence of sphenoid wing dysplasia (Table I).
X-ray technique. Patients were seated and the front and nose were positioned in contact with the lead-backed cassette. The central beam was focused in the posterior anterior direction to the nasal tip and ran parallel to the Frankfurt horizontal. This position allows for the projection of the orbital walls and the relationship of the orbital walls to adjacent skeletal structures, e.g. the clinoid processes, ethmoid cells and frontal sinus (Figure 1).
Data registration and morphometry. Patients' data were registered in Ortho Express® (Computer Forum, Elmshorn, Germany). All radiographs were scanned and processed in Dental Vision® (Computer Forum). Dental Vision® is a software that enables the processing of radiographs of different sources and files the data in a database with graphical user interface. In order to allow for the identification of landmarks relevant for this study, the program was adapted to the study design. Several landmarks were defined that allow the calculation of lines and angles related to the orbital ring (Table II). A distinctive feature during the preparation of radiographs for registration in Dental Vision® is the attachment onto the radiograph of a transparent foil measuring 5×5 cm2, imprinted with a metric scale. This measure allows the calibration of the length measurements on the scanned radiographs. Angles were measured in degrees and distances were recorded in millimetres. The orbital horizontal index was obtained by dividing the outer horizontal diameter by the inner horizontal diameter (inner orbital distance) and thus has no unit. The inner orbital diameter was separately measured (in mm) in order to define differences between groups relevant for hypertelorism in NF1.
The recording of distances on plain orbital radiographs followed a standardized protocol in each single case allowing the comparison of data. However, it has to be kept in mind that the values are not equivalent to quantified distances measured directly on human skulls (Tables V, VI, VII, VIII, IX, X, XI and XII).
Definition of landmarks. Laterality of bilateral reference points in the horizontal plane is indicated by plus and minus signs during the calculations (+, right hand side, −, left hand side) and in the Tables and Figures by using ‘R’ and ‘L’ as a part of the acronym designating the point. Line segments are designated by the acronyms of their anatomically defined end-points. Angles are designated by the identification of the constitutive line segments (Table III). The suffix ‘1’ marks the right, and ‘2’ the left side of the body in angle definitions (Table IV). The median line (M) is constructed perependicular to the line connecting the Z-points and is passing through the crista galli (Figure 2). Illustrations of landmarks, lines and angles are provided in Figures 1, 2, 3 and 4.
Parameters. Radiographs were investigated for asymmetries of the orbital ring with special reference to the described pathologies. In order to obtain comparable data, several metric parameters and angles were measured in a standardized fashion (Tables III and IV). It was assumed that orbits on plain radiographs of the DNF group do not show substantial intra-individual asymmetries (16). In order to use the radiographs of this subgroup as a control group, the detailed parameters were first measured in patients with DNF to exclude side-specific differences.
Statistics. Data were exported from Dental Vision to Excel® (Microsoft Corp., Redmond, WA, USA) and then transferred to SPSS® (SPSS Inc., Chicago, IL, USA). Paired and unpaired t-tests were used to distinguish for morphological parameters according to different diagnoses. P-Values lower than 0.05 were defined as being significant.
Results
Orbital height. Differences concerning orbital height in the DNF group were calculated to be non-significant (t-test) (Figure 3). The DNF group was used as a reference to study the height of the orbit in patients with facial PNF (Table V). The height of orbits in all 53 patients with PNF (106 orbits) was compared to the height of all patients with DNF (2×20=40 orbits). The comparison of the orbital height between affected orbits of the PNF group and the orbits of the DNF group revealed highly significant differences (Table V).
Laterality of PNF was determined and intra-individual comparison of orbital heights was performed in this group. Left-sided PNF (N=26) revealed elevated orbital height compared to the non-affected side. This difference was statistically significant (t-test). The same calculation was performed for patients with PNF of the right side (N=27, t-test). Intra-individual comparison revealed highly significant statistical differences concerning this parameter (Table V). Both tests showed significant intra-individual differences of orbital radiomorphology between the affected and non-affected side in patients with PNF of the orbital region concerning the parameter ‘orbital height’. The mean orbital height of the PNF-affected orbit was higher in this group than in patients with left-sided PNF.
Patients with orbital PNF who also had developed sphenoid wing dysplasia (also analyzed on available magnetic resonance images and/or computed tomograms) constituted a subgroup and were compared to the DNF group. It should be noted that four patients of the DNF group were also affected by sphenoid wing dysplasia (Table I). The distribution of this finding in both groups corroborates the definition of sphenoid wing dysplasia as an independent and defining feature of this entity (5). The orbital heights of these four patients did not differ from the values obtained in other patients of the DNF group.
Orbital width. Maximum orbital width in patients of the DNF group did not differ significantly between the right or left sides (Table VI).
The overall comparison of orbital widths between the DNF and PNF groups confirmed the hypothesis of altered orbital width being a consequence of orbital PNF in NF1. All PNF-affected orbits irrespective of side (N=2×53) were selected. The data of 20 DNF orbits (N=40) were collected. The t-test showed a tendency toward a smaller orbit in the DNF group, but revealed no statistically significant difference of orbital width dependent on orbital PNF.
In all patients with PNF the maximum orbital widths were intra-individually compared. In order to reveal side-specific variants, left-sided and right-sided orbital PNF constituted separately calculated subgroups. The t-test for the left-sided PNF subgroup revealed no statistically significant difference of orbital width compared to the unaffected right side. The same comparison of orbital widths in patients with PNF of the right orbital region revealed statistically significant differences in comparison to the unaffected left side: the orbital width of this PNF subgroup was greater than on the other side. A further analysis compared the differences of left vs. right PNF-affected orbits. This comparison revealed no statistical significant difference of orbital width in PNF-affected orbits.
The orbital widths of all patients with PNF and recognized sphenoid wing dysplasia (N=26, mean=42.51 mm) were compared to the widths of orbits of the DNF group (N=20, mean=39.58 mm). These differences between orbital widths proved to be statistically significant (Table VI). The results support the interpretation of an impact of sphenoid wing dysplasia plus PNF on orbital width measurable on plain orbital radiographs.
Alpha-angle of the orbit. The arms of alpha-angle of the orbit are defined by the cranial extension of the two line segment named orbital height and M-line (Figure 4). The measurement of this angle was intended to show the direction of lateral orbital extension. The hypothesis of this investigation was to reveal a more latero-caudal extension of the orbit in patients with PNF. This finding would fit into the well-known egg-shaped appearance on plain orbital radiographs repeatedly reported in the literature and possibly related to orbital PNF.
The first test intended to exclude distinctions between right and left orbits of patients with DNF concerning the oblique cranio-caudal extension (t-test): alpha-angles of right (mean 14.52°) and left orbits (15.73°) did not differ significantly (Table VII). All PNF-affected orbits were then compared to the orbits of patients with DNF. The alpha-angle of PNF-affected orbits was on average 10.96° and in DNF it was 15.13°. This difference proved to be highly significant, indicating a caudal extension of PNF-affected orbits. The next test was performed taking into account the side of orbital affection by PNF. The alpha-angle of right-sided PNF (N=27) was on average 12.16° on the affected side and 13.95° on the non-affected side. Mean values for patients with left-sided PNF-affected orbits (N=26) were lower, both on the affected side (10.92°) and the apparently normal right side (11.50°). Differences between the alpha-angle of the affected and non-affected side were not significant, irrespective of the side of PNF localization (t-test). All affected PNF orbits (N=53) had a mean alpha-angle of 10.81°. All non-affected orbits of patients with PNF (N=53) had a mean alpha-angle of 12.75°. These values did not differ significantly. These findings show a tendency of PNF-affected orbits to exhibit caudal extension and a narrowing of the orbital walls. Finally, the subgroup of orbital PNF-affected individuals who also had developed a sphenoid wing dysplasia was tested for alterations of the alpha-angle compared to patients of the DNF group. The alpha-angle of orbits in patients with PNF/sphenoid wing dysplasia (N=26, mean 11.74°) and DNF (N=20, mean 15.13°) revealed no statistically significant difference (Table VII): sphenoid wing dysplasia had no distinguishable effect on the oblique cranio-caudal expansion of PNF-affected orbits compared to patients with DNF.
Beta-angle of the orbit. This investigation was based on the calculation of the angle formed by the horizontal lines connecting both the most cranial points and the most caudal points of both the orbits. Direction of angular opening was registered and compared to the laterality of PNF. The starting hypothesis of this investigation was that tumor-associated sides will have a larger cranio-caudal distance, resulting in an opening of the angle on the affected side. However, this hypothesis was not proven in all cases (Table VIII). In five cases with PNF of the left side, the cranio-caudal extension of the orbit appeared to be smaller than that on the contralateral side, resulting in an angular opening to the right side (angles varying between 0.04° and 2.68°). Two cases with right-sided PNF also exhibited angular opening to the non-affected side (2.13° and 3.96°). All other beta-angles demonstrated the expected angular opening to the affected side.
Beta-angles of patients with PNF and DNF were compared. This calculation intended to reveal differences of the orbital heights and in cranio-caudal orbital topography in DNF (N=20) and PNF (N=53). The hypothesis of this investigation is that small differences of the cranio-caudal extension in DNF may be present despite the symmetrical appearance of the orbits in the frontal plane. Mean values of beta-angles in PNF of 4.51° and DNF of 1.43° differed significantly. This result reveals an enlarged beta-angle in the majority of patients with PNF. The comparison of left- and right-sided orbital PNF revealed no differences of beta-angle (Table VIII).
Finally, the beta-angles of patients with PNF and with sphenoid wing dysplasia (N=26) and those with DNF (N=20) patients were compared. The patients with orbital PNF plus ipsilateral sphenoid wing dysplasia had a mean beta-angle of 6.7° and those with DNF of 1.43°. These differences proved to be highly significant (Table VIII) suggesting an impact of PNF plus sphenoid wing dysplasia on the investigated parameter.
Gamma-angle of the orbit. The measurement of this angle indicates the direction of the orbit to the lateral side. The arms of the gamma-angle can become extended resulting in an open angle or be narrowed to an angle of about 90°. The hypothesis of this measurement is a narrowing of the gamma-angle in orbital PNF indicating the extension to a latero-caudal direction. Values of gamma-angles about 90° would determine a roundish orbital rim (Figure 4). However, all mean values of gamma-angle were greater than 100°, irrespective of diagnostic group (Table IX).
The first group studied was the DNF group (N=20). Left and right orbits showed no statistically significant differences of gamma-angles (Table IX).
The gamma-angles of all PNF-affected orbits were then compared with all DNF orbits. This test was carried to study the impact of PNF on the relation of the orbital quadrants. The differences of the gamma-angle did not differ significantly between groups (t-test).
The intra-individual comparison of the gamma-angles revealed no statistically significant differences between orbits affected by PNF and contralateral unaffected orbits. Moreover, the presence of sphenoid wing dysplasia had no effect on this parameter.
Horizontal orbital distances. A PNF of the first and/or second trigeminal branch is not associated with an altered index of the horizontal orbital distances, compared to orbits of patients with DNF (Table X). However, patients with a PNF of the first and/or second trigeminal branch plus ipsilateral sphenoid wing dysplasia exhibited an altered index of the horizontal orbital distances, compared to orbits of patients with DNF (Table XI).
The absolute values (mm) of the inner orbital distances were determined dependent on the phenotype of patients with NF1. In plain orbital radiographs, no statistically significant difference of the distances of most of the inner orbital margins was found between the PNF and DNF group (Table XII).
Discussion
This study describes some radiological alterations of the orbit associated with NF1. These findings were derived from plain radiographs and are predominantly found in orbits affected by further pathognomonic stigmata of NF1, in particular PNF. Our results suggest a unilateral phenomenon. The frequency of orbital alterations in NF1 is not known. The studies conducted on orbitofacial affections in NF1 are usually confined to discrete findings and do not differentiate the complex pathology often detectable in this condition. The number of patients with neurofibromatosis and facial PNF appears to be low [3/223, 1.3%, (28)]. This finding was confirmed in later studies [1/77, 1.3% (29), 8/257, 3.1% (30)].
This study does not intend to define the prevalence of orbital alterations in NF1 but to investigate whether plain radiographs allow for some preliminary assumptions about the affection of the orbit in this condition, and possible associations with other clinically relevant findings.
Osseous alterations. Alterations of the orbit in the neurofibromatoses were described soon after the introduction of roentgenology in the diagnosis of bone diseases. Tauber (31) and Kren (32) described a sphenoid wing dysplasia and associated findings of the upper jaw's alveolar process. Nowadays, the dysplastic sphenoid wing is regarded as a pathognomonic finding in NF1 (5), but with varying degrees of manifestation (18). Later, hypertelorism was added to the head and neck findings in neurofibromatosis, detected in about 25% of patients (33). A further finding in NF1 is optic pathway glioma (34). The possible impact of optic pathway glioma on orbital growth is presently not known (14). The pathogenesis of orbital dysplasia in NF1 is also not known and has been a matter of debate for a long time, some authors arguing in favour (35) and others against (18) a theory of local tumour pressure resulting in osseous modification of the orbit. Furthermore, microphthalmos or even loss of the eye in the course of invasive PNF growth may contribute to the alteration of the orbital volume (16). Our results point to an effect of PNF on the extension of the orbital ring in latero-caudal direction in a considerable number of patients.
Sphenoid wing dysplasia. Sphenoid wing dysplasia is a defining feature of NF1 and independent of a PNF of the orbital region. Jacquemin et al. (15) propose an early closure of sutures as the cause of sphenoid wing dysplasia in NF1. However, this pathomechanism does not explain the highly variable alterations of brain development associated with this finding and the associated severe ipsilateral skull base malformations beyond sphenoid wing dysplasia. These authors speculate that PNF of the orbital region interferes with the orbital bones and thus contributes to the dysplasia. On the other hand, clear evidence was established that sphenoid wing dysplasia can develop without any apparent neurofibroma at this site (16, 18).
Orbital height. Orbital height did not differ between the orbits of patients with DNF. However, differences in orbital height became apparent in patients with PNF affecting the orbit. Our investigation reveals an increase of orbital height on the side of the orbital affected by a PNF in patients with NF1. The side of PNF had no impact on this phenomenon. Increase of the parameter ‘height’ determined on plain orbital radiographs should alert the clinician to look for associated manifestations of NF1 in the orbital region, in particular a PNF.
Orbital width. Orbital PNF appears to be associated with a caudal and lateral extension of the orbit in the horizontal plane, resulting in an ‘egg-shaped’ deformity of the orbital ring (14, 15). Our results support this simplifying radiographic description of PNF-associated orbital malformations, as far as findings derived from plain radiographs are concerned. This deformity is not present in patients with DNF. The widening of the orbit was a frequent phenomenon of the affected side in orbital PNF-affected patients, irrespective of the side affected, but statistically significant differences of orbital width were only found in patients affected by a PNF of the right orbit. Sphenoid wing dysplasia affected ‘width’ only in association with a PNF. Both parameters ‘height’ and ‘width’ illustrate the intimate relationship between the development of a very slowly growing tumour, but which is developing early in life and the skeleton.
Alpha-angle. The alpha-angle was determined in order to define the direction of the shift of the orbital vertical axis in PNF-affected orbits. It was assumed that the differences in orbital height and width associated with PNF may cause a measurable shift of the vertical axis. The alpha-angle differed significantly between PNF-affected and non-affected orbits of patients with NF1, resulting in an up-righting of the axis on the affected side. This indicates that the orbital alterations in PNF-affected orbits are more pronounced in the caudal direction (height) than in the lateral side (width). However, direct measurement of height and width gave clearer results than did calculation of the axial shift represented by the alpha-angle.
Beta-angle. The orbital deformities in NF1 are manifold and this was intended to show a distinctive pattern of orbital dysmorphology at the orbital ring. The beta-angle was introduced into the study design in order to reveal whether there are regularities in the cranio-caudal orbital extension associated with PNF. The determination of angles allows the comparison of radiographs irrespective of age (36). Indeed, many patients with PNF exhibited divergence of the arms of the angle at the orbital PNF side. However, this finding was not obtained in all patients with PNF. On the other hand, the beta-angle was extremely low and showed no side specificity in patients with DNF, pointing to the positioning of patients almost parallel to the film plane. The difference between the beta-angles of patients with PNF (4.1°) and those with DNF (1.43°) was highly significant. Therefore, an enlarged beta-angle can be used as an indicator of orbital PNF, but lack of beta-angle enlargement does not exclude the presence of PNF in the orbitotemporal region.
Gamma-angle. Side-specific deviations of the angle determined by the crossing of orbital diagonals could indicate orbital dysplasia. A narrowing of the diagonals and therefore a smaller gamma-angle could indicate an erection of the vertical axis. Patients with DNF exhibited a greater gamma-angle than did those with PNF. However, this difference was statistically not significant. Determining the gamma-angle is not sensitive enough to identify orbital alterations in NF1 on plain orbital radiographs, probably due to concordant alterations of the intersections of the lines constituting the gamma-angle.
Interorbital distance. The interorbital distance was calculated as the ratio between inner to outer orbital borders. An orbital PNF did not affect this ratio. The inner interorbital distances showed no significant differences between the groups (DNF vs. PNF). However, the interorbital distances appeared to be slightly narrowed in patients of the PNF group. These findings do not discount the presence of hypertelorism in NF1 in general, but do not show any difference between both subgroups of NF1 concerning horizontal measuring points relevant to determine horizontal orbital distances in this radiological projection. Hypertelorism was found in about 25% of patients with NF (33) and appears to be consistently associated with those who harbour large deletions of the NF1 gene (37).
Conclusion
Plain radiographs of the orbit reveal some characteristics that can be used to substantiate the tentative diagnosis of NF1. Sphenoid wing dysplasia can be differentiated on plain radiographs and is very likely to be an independent skeletal finding in NF1. However, the alterations of the orbital rim can also be indicative of NF1. These alterations are closely associated with the extension of a facial PNF affecting the orbito-temporal region and appear to be unilateral findings. The characterized alterations of the orbit detected in an individual with NF1 should alert the clinician to search for an associated PNF. These alterations of the orbital rim show some variations of these parameters, as can be expected in an inheritable disease with a wide spectrum of signs and symptoms.
Acknowledgements
The Authors appreciate the cooperation with Andrea Rusche, Computerforum, Elmshorn, Germany, who programmed the software allowing the cephalometric analysis with Dental Vision®.
Footnotes
-
This article was presented in part in oral form on the occasion of the 15th Hamburger Symposium on Tumor Markers, Hamburg, Germany, May 29-31, 2011.
-
Conflicts of Interest
None.
- Received January 10, 2013.
- Revision received February 7, 2013.
- Accepted February 7, 2013.
- Copyright© 2013 International Institute of Anticancer Research (Dr. John G. Delinassios), All rights reserved