Skull base ligamentous mineralisation: evaluation using computed tomography and a review of the clinical relevance.

OBJECTIVES
To determine the frequency, morphologic and demographic characteristics, and clinical relevance of the mineralisation of six skull base ligaments (interclinoid, caroticoclinoid, petrosphenoid, posterior petroclinoid, pterygospinous, and pterygoalar).


METHODS
This is a retrospective review of 240 CT scans of the paranasal sinuses (ages 6-80 years). A limited systematic review was performed primarily using Embase and Medline databases.


RESULTS
Ligamentous mineralisation was well delineated on CT and occurred at ≥ 1 location in 58.3% of patients. There was a nonsignificant trend towards a greater incidence with advancing age. The interclinoid and posterior petroclinoid ligaments were most commonly mineralised (22.1% and 18.3%, respectively); the petrosphenoid and pterygoalar ligaments were least frequently mineralised (10.8% and 6.3%, respectively). The mean age of patients with posterior petroclinoid mineralisation was significantly greater than those with interclinoid and petrosphenoid mineralisation and was not seen in patients aged 6-20 years. The literature review highlighted the clinically relevant potential for mineralised ligaments to cause barriers to surgical access (e.g. to the foramen ovale), increase the risk of neurovascular injury during surgery at the skull base (e.g. during anterior clinoidectomy), and predispose to neural impingement.


CONCLUSIONS
Skull base ligamentous mineralisation is commonly encountered on CT imaging. Given the potentially significant clinical implications, an understanding of the morphological appearances is of importance to those planning interventions at the skull base. To the authors' knowledge, this study is the first to comprehensively evaluate such a wide range of skull base ligaments using CT. For some ligaments, the incidence on CT has not been previously described.


Introduction
Several ligaments exist at the skull base, but knowledge of their anatomy is limited amongst clinicians owing to the paucity of coverage in mainstream anatomical texts.
However, improvements in minimally invasive neurosurgical techniques have made accurate identification of these structures invaluable for surgical planning, particularly when they become mineralised [1][2][3]. Mineralised ligaments can present barriers to surgical access, alter the appearances of familiar anatomical landmarks, or prevent structural mobilisation during surgery, thereby increasing the risk of neurovascular injury [4][5][6]. Additionally, mineralised skull base ligaments have been implicated in neural impingement syndromes as a result of mechanical compression of nerves against ossified bars or within the foramina that mineralised ligaments may form [7][8][9][10][11]. Hence, skull base ligamentous ossification is relevant to radiologists, neurologists, and neurosurgeons managing patients with skull base pathology.
The available literature is predominantly derived from studies of dry skulls, with only a minority using imaging to evaluate these structures (see the tabulated summary of the subsequent systematic review). To the authors' knowledge, this represents the first comprehensive study to use computed tomography (CT) to systematically evaluate the frequency of incidental skull base ligamentous mineralisation in a modern ethnically diverse population.

Materials and methods
Institutional approval was obtained, and the requirement for informed consent waived. A retrospective review of high-resolution, non-contrast CT studies of the paranasal sinuses (scanned between April 2014 and January 2017) was carried out. Consecutive cases were selected until equal numbers were achieved for each of 15 age groups (range 6-80 years). Scanning took place on a variety of systems, including SOMATOM Definition Edge (Siemens Healthcare, Erlangen, Germany), iCT, and Brilliance 40 (Philips Medical Systems, Eindhoven, Netherlands) scanners using a kVp of 120 kV, mAs of 25-50, minimum collimation of 0.6-0.625 mm, and a pitch of 0.624-0.8. Each imaging study was evaluated by there independent observers PT, SH, and FC, and the presence of mineralisation (calcification or ossification) for the six skull base ligaments was recorded. Initial detection was carried out by analysing thin axial reconstructions, and detailed evaluation of morphology was carried out using multiplanar reconstructions. The ligaments examined, their anatomical courses, and planes used to evaluate them on CT are detailed in Table 1. Examples of the appearances of the ligaments on CT are demonstrated in Figs. 1, 2, and 3.
In each case, mineralisation was considered 'partial' if it extended from 50 to < 100% of the ligament's length and 'complete' if it extended to involve the entire length of the ligament. The so-called contact type of mineralisation, where a subtle suture line may be seen at the midpoint of an osseous bar, was considered complete for the purposes of this study [12]. If complete mineralisation resulted in the formation of a foramen, the thickness of the bony bar (at its midpoint) and the corresponding foraminal area were measured using double oblique sagittal reformats on a PACS workstation using syngo.via software (Siemens Healthcare, Erlangen, Germany). Ligaments with < 50% mineralisation, including small bony spurs, were excluded. The use of 50% was chosen as it was felt to be both clinically relevant and simpler to facilitate reproducibility; it has also been employed in prior studies of ligamentous mineralisation [18,23]. In the case of interobserver discordance, an agreement was achieved through consensus. Where available, demographic information was recorded.
Statistical testing of multiple correlated samples was carried out using a one-way ANOVA with post hoc analysis using the modified Tukey method and two-tailed t testing, and the chi-squared test was employed to analyse the distribution of categorical variables using Vassarstats [24] and Microsoft Excel® (Redmond, WA); a p value of < 0.05 was deemed to be significant. A systematic review of the English language literature was carried out as per PRISMA [25] guidelines using Embase and Medline databases primarily with additional studies identified through study references and a limited search using Google Scholar. The following search terms were utilised 'interclinoid' , 'caroticoclinoid' , 'sellar bridge' , 'petrosphenoid' , 'petroclinoid' , 'pterygospinous' 'Civinini + ligament' , 'pterygoalar' , 'Hyrtl + ligament' , and 'crotaphitico-buccinatorius'. Studies were excluded if they were deemed irrelevant (e.g. pertaining to other parts of the body). Selected case reports were included if a potentially clinically consequential observation was documented.
Partially or completely mineralised skull base ligaments in at least one location were found in 58.3% of patients (n = 140).
Mineralisation was observed in all age groups, but least frequently amongst the 16-20 years age group (31%) and most frequently in the 56-60 years age group (81%) Fig. 4. Dividing the population into 5 larger groups of 48 patients, each revealed lower mean proportions of mineralised ligaments amongst the 6-20 and 21-35 years groups compared with older patients. Although the difference was nonsignificant (p = 0.0795, using a one-way ANOVA test), there was a trend towards increasing mineralisation with age. Additionally, the rate of complete mineralisation (patients with ≥1 completely ossified bar on either side) showed increasing frequency with age    66-80 years = 13%. The difference between the means was significant (one-way ANOVA: F-ratio = 4.06; p = 0.0329); however, on breakdown of the differences between the means using the Tukey method, only the difference between the 6-20 year and 51-65 year groups was found to be statistically significant (p = < 0.05).
The proportion of patients with mineralised ligaments was highest amongst those of white British/European heritage, followed by black British, African, and Caribbean heritage (57%) and those of British Asian/Southeast Asian heritage (52%). The lowest proportion was seen amongst those of other heritages. However, the difference between the proportions of ossified ligaments amongst white and black and white and Southeast Asian patients was nonsignificant (p = 0.635 and p = 0.382, respectively, using a chi-squared test).
Overall, there was a very slight male preponderance for ligamentous mineralisation with 74 males and 66 females (M:F = 1.12:1).

Ligament type
The incidence of ligamentous ossification (both partial and complete) varied according to the ligament type, with the interclinoid ligament being most commonly identified and the pterygoalar ligament least commonly identified (the proportions for all ligaments are detailed in Table 2).
The majority (four of six) of mineralised ligaments were more commonly unilateral, but the caroticoclinoid and petroclinoid ligaments were more commonly bilateral. The proportions of bilaterally and unilaterally mineralised ligaments are detailed in Table 3.
Mineralised interclinoid and caroticoclinoid ligaments could be seen in all age groups. However, the remaining ligament types were not present in all age groups; for example, mineralised petrosphenoids were not encountered in the 6-10 and 16-20 year groups. The frequencies of each ligament type amongst the various age groups are depicted in Fig. 3.
Overall, there was no statistically significant difference between the mean ages of patients with mineralised ligaments (0.0777, using a one-way ANOVA test); however, breakdown analysis of the differences between the groups revealed a significantly higher mean age for patients with posterior petroclinoid ligamentous mineralisation compared to those with interclinoid and petrosphenoid mineralisation (p = 0.004 and p = 0.009, respectively).
The thickness of the mineralised ligaments varied slightly, with the thinnest being the pterygospinous ( Table 4). The smallest foramen was formed by the mineralised petrosphenoid ligament, and the largest foramen was formed by the mineralised interclinoid ligament (Table 4).

Multiple ligaments
Ossification of multiple (> 1) ligament types was observed in 26.7% (n = 64) patients. The majority (76.6%; n = 49) of these patients had a combination of two ossified ligaments, with the interclinoid and caroticoclinoid ligaments in combination (n = 20) and the petroclinoid and pterygospinous ligaments in combination (n = 11) being the most common. Ossification of > 2 ligament types was seen in 23.4% (n = 15) of patients and ossification of > 3 ligament types in 3.1% (n = 2) of cases.

Limited systematic review
Screening yielded 492 abstracts in the initial search; however, following the removal of duplicates and studies that did not meet the inclusion criteria, 61 records remained for inclusion (Table 5).

Discussion
Mineralisation of skull base ligaments can occur as a result of an interplay between a broad range of factors, including genetics, metabolic abnormalities, and mechanical stress [68]. Such factors may explain de novo mineralisation later in life. However, the presence of ligamentous skull base mineralisation in children without an obvious inductive stimulus [12] may reflect developmental variation, which some have termed atavistic (i.e. representing evolutionary remnants) owing to the presence of similar ossified structures in non-human species [69].
It is clear from this study that mineralisation of skull base ligaments is a common finding (58.3%). In keeping with a suspected predominantly developmental origin, mineralisation was present in all age groups, although there was a nonsignificant trend towards an increased incidence with age. The association was however stronger for complete ligamentous mineralisation and varied with ligament type. In particular, the mean age of patients with posterior petroclinoid ligamentous mineralisation was higher than those with interclinoid or petrosphenoid mineralisation and was not observed in individuals aged 6-15 and 31-35 years. This finding likely reflects the nature of the posterior petroclinoid ligament, which is in fact a fold of dura mater (rather than a true ligament) that arises from the fixed portions of the tentorial incisura, and calcification of the dura is generally rarely seen in children [18,19,70]. There was no significant difference in the rate of ligamentous mineralisation amongst the largest ethnic groups included within the study; however, variance exists in the literature with higher rates of observed mineralisation in some (particularly Greek) populations, suggesting a potential genetic predisposition [14,30,57].
The clinical significance of mineralised interclinoid and caroticoclinoid ligaments arises primarily from their close relationships with the paraclinoid internal carotid artery (with the caroticoclinoid ligament potentially forming a solid ring around it) and cavernous sinus. In particular, the presence of ossified bars in these locations can make the extradural removal of the anterior clinoid process during clipping of paraclinoid aneurysms extremely difficult, requiring increased drilling and manipulation, which is accompanied by an increased potential risk of carotid rupture [2,5,14,26,31,71]. Furthermore, these structures can complicate the excision of central skull base tumours where the internal carotid artery and cavernous sinus require exposure [2]. In addition, the presence of a completely mineralised caroticoclinoid ligament may alter the appearance of the middle clinoid process, which can be used as landmark for the anteromedial roof of the cavernous sinus and transition between the cavernous and clinoid segments of the internal carotid artery during endoscopic endonasal approaches to the pituitary gland [5,6]. Furthermore, the presence of high-density calcification in the parasellar region may cause confusion on CT angiography if the viewer is unfamiliar with skull base ligamentous mineralisation; indeed, mineralisation of the interclinoid has been confused with para-posterior communicating artery aneurysm [38]. Finally, 'sellar bridges' have been associated with dental and other developmental abnormalities, including Gorlin-Goltz syndrome [26,40,42,72,73].        -Meta-analysis.
*Includes elongation of the middle clinoid process **Complete includes contact type
The clinical significance of petrosphenoid ligamentous mineralisation principally arises from its close relationship to the abducens nerve, which passes below it within Dorello's canal [17]. For example, in the setting of raised intracranial pressure and uncal herniation, the mineralised ligament may protect the abducens nerve, but may present a noncompliant structure against which the oculomotor nerve may be compressed [16]. Furthermore, the passage of the abducens nerve beneath a densely mineralised ligament is postulated to have a role in abducens nerve palsy as it would create a noncompliant structure around the nerve, which would limit expansion in the setting of neural inflammation [11]. Finally, the petrosphenoid ligament is a helpful landmark during subtemporal-transtentorial-transpetrous approaches to the posterior and middle cranial fossae and its mineralisation may lead to the misidentification of anatomical localisation [16,74].

Posterior petroclinoid ligament (fold)
This structure was the second most commonly mineralised ligament (18.3%), which is higher than some studies of dry skulls (1.4-9%) [18,49], but comparable to prior radiographic and CT studies [16,23,46]. This likely reflects the superiority of imaging in detecting fine calcified structures that may not be preserved in dry skulls.
The clinical significance of posterior petroclinoid ligament (or dural fold) mineralisation derives from its proximity to neural structures. In particular, in its course between the anterior petrous ridge to the posterior clinoid process, it forms the roof of the porus trigeminus and medial border of the oculomotor trigone (with the oculomotor nerve running over the ligament) [18]. In cases of mineralisation, Wysiadecki et al. found greater fixation of the dural sheath of the oculomotor nerve, which may increase the risk of neural injury during intraoperative manipulation, and prior division with an appropriate instrument may be required [16,50]. It may also increase the risk of oculomotor neural injury following relatively insignificant head trauma, as a result of compression of the nerve against a noncompliant ligament [10,47]. Finally, there has been speculation that compression of the trigeminal nerve may occur in the setting of an extensively mineralised posterior petroclinoid ligament and may be considered for those in whom prior microvascular decompression has failed [18,75].
The clinical significance of pterygospinous and pterygoalar ligamentous mineralisation arises from their capacities to form barriers to surgical access as well as their close relationship to neural structures. Although both ligaments are in close proximity anatomically, they are distinct in their courses, most notably posteriorly, with the pterygospinous ligament (a thickening of the interpterygoid aponeurosis) attaching to the spine of the sphenoid and the pterygoalar ligament (a thickening of the lateral interpterygoid or pterygotemporomaxillary aponeurosis) attaching more laterally to the undersurface of the sphenoid [22]. Furthermore, whilst both ligaments attach to the lateral pterygoid plate anteriorly, the pterygoalar ligament attaches more superiorly, at the level of the root [20]. This is particularly relevant for access to the foramen ovale for percutaneous rhizotomy or cavernous sinus biopsy where a mineralised pterygoalar ligament can create a wall-like barrier lateral to the foramen ovale, making percutaneous access difficult or even impossible, particularly via a trans-zygomatic approach [1,21,51,64,66,76]. In addition, mineralisation of either ligament may impede trans-zygomatic exploration of the external skull base as well as the parapharyngeal or retropharyngeal spaces [21,60].
Following the descent of the mandibular division of the trigeminal nerve through the foramen ovale, it undergoes branching. Some of these pass through the foramina created by the mineralised pterygospinous and pterygoalar ligaments. In particular, branches to the tensors tympani and veli palatini and medial pterygoid can pass through the foramen of Civinini and motor branches to the temporal, buccinator lateral pterygoid, and sometimes masseter muscles may pass through the foramen created by the pterygoalar ligament [7,20,22]. However, the association with neural branches is variable; indeed, von Lüdinghausen et al. described four potential branching patterns (A-D) in relation to a mineralised pterygospinous ligament with lateral displacement of the branches to the temporalis, masseter, and pterygoid muscles being most common and medial displacement of the branches being least common [60]. Others have described further variations, such as division of the lingual nerve into an anterior and posterior division by a mineralised ligament, which can increase the risk of entrapment [77]. Entrapment may also arise when the lingual nerve passes between an ossified pterygospinous ligament and the medial pterygoid muscle [9,67,78]. In addition, Krmpotić-Nemanić et al. noted that various types of lateral pterygoid plate enlargement (including complete ossification of the pterygospinous ligament) resulted in the displacement of the lingual and inferior alveolar branches resulting in fixation and increased risk of compression [7]. It is also suggested that a mineralised pterygospinous ligament may potentially cause the compression of other branches of the mandibular nerve (auriculotemporal nerve in particular), leading to periauricular sensory or parotid glandular secretomotor symptoms [64,66,76].

Limitations
Whilst noncontrast CT provides excellent delineation of mineralised structures, it does not allow for the detailed visualisation of soft tissue anatomy such as nerves and blood vessels that may be affected by ligamentous mineralisation. In the future, MRI may be useful in determining the precise relationships between mineralised ligaments and local cranial nerves. In addition, given the retrospective nature of the study, only limited clinical data was available; therefore, it is not known whether any of the cases included suffered symptoms in relation to ligamentous mineralisation.

Conclusion
The presence of ligamentous skull base mineralisation is a relatively common phenomenon on CT. These structures can present barriers to minimally invasive surgical access to the infratemporal fossa and increase the risk of neurovascular injury at the central skull base. Furthermore, ligamentous mineralisation has been implicated in neural entrapment. Therefore, knowledge of these structures is of great importance to avoid undesirable complications.

Availability of data and materials
The datasets generated and/or analysed during the current study are not publicly available in accordance with local data protection policies, but anonymised data are available from the corresponding author on reasonable request.
Authors' contributions PT and SH were involved in the conception, design, acquisition, analysis of data, and drafting of the manuscript. AO was involved in the design, analysis, and drafting of the manuscript. FC was involved in the data acquisition and conception. SEJC was involved in the design, conception, analysis, and drafting of the manuscript. All authors read and approved the final manuscript.
Ethics approval and consent to participate Institutional approval was obtained and the need for consent waived.

Consent for publication
Not applicable.