A comparative review of multidetector CT angiography and MRI in the diagnosis of jugular foramen lesions

Introduction 

The jugular foramen is a triangular canal on the infero-medial surface of the petrous pyramid formed by the temporal and occipital bones.1 A variety of disease entities arise from the structures normally found within the foramen, i.e., the cranial nerves 9, 10, and 11, the internal jugular vein and inferior petrosal sinus, and very rarely from the posterior meningeal artery. Primary bone (giant cell tumour, chondrosarcoma, and chondroblastoma) and meningeal disease, e.g., meningioma, can also affect the jugular region as well as masses within the subarachnoid space, e.g., epidermoid cyst.

The most frequent lesions that require an accurate differential diagnosis are glomus jugulare tumours (60–80%), schwannoma/neuromas and meningiomas (both approximately 10%).2, 3 A glomus tumour is a paraganglioma arising from the chemoreceptor tissue in the adventitia of the jugular vein, or in the parasympathetic fibres carried along the Arnold or Jacabsons nerves of the vagus and glossopharyngeal nerves, respectively.2, 4 Neuromas arise from the covering of any of the nerves, but most commonly originate from the ninth cranial nerve.5 Tumours at the foramen can have a variable extension above or below the skull base, with larger lesions extending from the neck to the intracranial cavity. The typical clinical presentations for all underlying disease is similar: hearing loss, pulsatile tinnitus (classical, but by no means specific, for a glomus tumour), and more vague symptoms related to nerve palsies, such as difficulties swallowing and phonating, neck and shoulder weakness, syncope, and hypotension.

The three commonest masses are benign and produce expansion of the jugular bulb. Irregular erosive enlargement of the jugular foramen is typical of a glomus tumour, but often neither the clinical signs nor the bone abnormality can distinguish the underlying disease, and so detailed computed tomography (CT) and magnetic resonance imaging (MRI) are regularly used as imaging techniques.

In addition to providing a probable pathological diagnosis, imaging can offer information regarding the primary vascular nature of the lesion, any enlarged arteries supplying the tumour, and any secondary venous compressions or occlusions. Both aid surgical and interventional preoperative planning. In many instances both methods are performed because they are complementary,3, 4 but also partly because there is conflicting evidence as to which test is diagnostically superior 4, 6, 7, 8 and could stand alone. There are currently no studies directly comparing the efficacy of each modality on the same cohort of patients. This study compares the imaging characteristics of multidetector CT angiography (MDCTA) against standard axial and gadolinium-enhanced MRI, the techniques used most commonly at this, and other, institutions.

Materials and methods 

Patients, with tumours predominantly occurring at the jugular foramen, were identified through a search of the neuroradiology database, operating log books, and lists from the surgeons handling such cases, from 2001 to 2008. This revealed 15 cases, including 10 females with an average age of 59 years at the time of diagnosis. The MDCTA and MRI reports were reviewed and the imaging was re-evaluated retrospectively (by E.T.). A histological diagnosis was available in 11 patients.

MR protocol 

All studies were performed on a variety of 1.5T systems, but with a standard technique including sagittal T1 and axial T2-weighted scans with axial and coronal T1 sequences before and after gadolinium contrast enhancement, at a section thickness of 5mm with a 1mm intersection gap. Short-tau inversion recovery (STIR) and fluid-attenuated inversion recovery (FLAIR) were also performed in selected cases at the initial supervising radiologists' discretion.

Ten patients underwent MRI first, with the average interval between MRI and MDCTA being 4 months (range 1 month to 2 years, with a mode of 1 month). The hard copies of the 30 images were randomly arranged, and re-reported by an experienced neuroradiologist who was unaware of the pathological diagnosis. The original reports from all the images were also recorded. In all cases, the conclusions of the re-report agreed with those of the original report. In addition to the primary diagnosis, the review considered the presence or absence of abnormal arteries, as well as patency, compression, or normality of the jugular vein (see listings in Table 1).

Table 1. Summary data for computed tomography angiography (CTA) and magnetic resonance imaging (MRI)

		CTA				MR			Pathology
		Radiodensity (HU)	Jugular vein	Abnormal vasculature	Diagnosis	Enhancement	Jugular vein	Diagnosis
	1	150	Occ	As Ph+veins	G	Y- S&P	?	G	G
	2	117	Occ	Occipital artery+veins	G	Y- S&P	Occ	G	G
	3	371	Comp	No	G	Y	Patent	G or N	G
	4	192	Comp	As Ph+veins	G	Y-S&P	Comp	G	G
	5	160	Patent	As Ph+veins	G	Y-S&P	Comp	G	G
	6	202	Occ	As Ph+veins	G	Y-S&P	Comp	G	G
	7	70	Occ	No	N	No	?	N	N
	8	73	Occ	No	N	Y	Occ	G or N	N
	9	98	Occ	No	N	Y	Occ	G or N	N
	10	58	Occ	No	N	Y	Occ	G or N	N
	11	62	Occ	No	N	Y	Occ	G or N	N
	12	251	Occ	As Ph+veins	G	Y-S&P	?	G	-
	13	221	Comp	As Ph+veins	G	Y-S&P	?	G	-
	14	55	Occ	No	N	Y	Patent	G or N	-
	15	215	Comp	As Ph+veins	G	Y	?	G or N	-

G, glomus; N, neuroma; As Ph is ascending pharyngeal artery; Y-S&P, yes—“salt and pepper”; Occ, occluded; Comp, compressed; ?, can't tell.

The MDCTA images were further assessed on a dedicated Philips MDCT workstation. Multiplanar reformatted images were generated by a neuroradiologist (E.T.). Hounsfield unit measurements were recorded for the tumour, and an assessment was made on the presence of abnormal arteries and draining veins, as well as the involvement of the jugular vein, which were not specifically recorded by the initial reporting radiologist.

Results 

Details of the imaging and pathological findings are given in Table 1. The postoperative histological results showed six glomus jugulare tumours and five neuromas. MDCTA made the correct diagnosis in all 11 cases based primarily on whether there was significant arterial phase enhancement—a finding specific to the glomus tumours (Fig. 1a). The mean radiodensity of the glomus tumours was 210 HU (range 117–371 HU), compared with 69 HU (range 58–98 HU) for the neuromas (Fig. 1b).

View Large Image Download to PowerPoint

Figure 1 (a) Coronal CT angiogram showing avid enhancement of a right-sided glomus tumour (arrow) and (b) coronal CT angiogram showing lack of enhancement of a left-sided neuroma (arrow).

All 15 MRI images showed contrast enhancement except in one patient. In this case a confident diagnosis of neuroma was made, which was confirmed histologically. In the remaining four proven neuromas an assured diagnosis could not be made based on the imaging findings of a uniformly enhancing tumour. In the six proven glomus tumours, a confident diagnosis was made using MRI in five cases based on the presence of tumour flow voids (“salt and pepper” appearance), which is characteristic of the condition.4 The MRI reports and review on the remaining proven glomus tumours failed to differentiate between a glomus and neuroma (Fig. 2). The present results are similar to another study that showed strong enhancement in eight out of nine neuromas.5

View Large Image Download to PowerPoint

Figure 2 (a) Axial gadolinium-enhanced MRI images of a left-sided glomus tumour (arrow) and (b) left-sided neuroma (arrow). The enhancing characteristics of the two images are similar, with the glomus tumour not displaying the characteristic “salt and pepper” flow voids. The unenhanced T1 and T2-weighted images (not shown) were also unable to confidently differentiate the tumours.

A pathology report was not available in four patients as they continue to be managed conservatively. A confident diagnosis was made in all cases using CTA (three glomus; one neuroma), but in only two with MRI, and both showed the “salt and pepper” pattern typical of a glomus. The other two cases demonstrated homogeneous enhancement and no confident diagnosis could be made.

Jugular vein analysis on MDCTA reported nine cases as occluded, five as compressed but patent, and one patent and normal. In five cases, an accurate judgement was unable to be made using MRI, and disagreed with the MDCTA results in one patient. There was no correlation between jugular vein status and pathological diagnosis.

Further vascular review was made using multiplanar reformatted CT images. A diagnosis of glomus tumour was given using MDCTA in nine patients and neuroma in six. In eight of the nine glomus tumours there was a demonstratable enlarged arterial supply, seven involving the ascending pharyngeal and one the occipital artery (Fig. 3). All nine cases demonstrated enlarged and early filling of draining veins, and all nine confirmed compression or occlusion of the jugular vein. There were no enlarged vessels detected in the neuroma cases.

View Large Image Download to PowerPoint

Figure 3 Coronal MIP (maximum intensity projection) reconstruction showing the enlarged ascending pharyngeal feeding artery entering the glomus tumour (arrows). The normal contralateral ascending pharyngeal is also shown (arrowhead).

One further advantage of the MDCTA being a vascular study is the ability to process images as venograms demonstrating the specific anatomy in the individual. This can influence surgical planning and determine whether an affected jugular vein can be sacrificed to ensure adequate tumour removal (Fig. 4).

View Large Image Download to PowerPoint

Figure 4 Volume-rendered acquisition of the venous dataset showing a right-sided glomus tumour (G) viewed from behind. The image clearly shows that the ipsilateral tranverse sinus (arrows) is the dominant draining vein (contralateral sinus shown by arrowheads). This alerts the surgeon that an attempt should be made to preserve the jugular vein at tumour resection.

Discussion 

Both CT and MRI are widely used in the diagnosis of lesions at the jugular foramen. The present study has highlighted that many patients have both tests performed, irrespective of which examination was performed first. This is partly because the information that they provide is deemed complementary, but also because there is a lack of studies comparing the diagnostic efficiency of the two methods.9 Although only two disease entities have been included in this study, they account for over 90% of all such masses at this site, despite the extensive list of possible lesions.3 Meningioma, statistically the next commonest pathology at the jugular fossa, not uncommonly arises from the skull base, but very rarely cause expansion of the fossa. On first principles it would be expected that the enhancement of a meningioma would be less than a glomus tumour and more than a neuroma, but there are no confirmatory observations. This study supports the view that either imaging modality can diagnose lesions involving the jugular foramen, but that MDCTA is more accurate. Previous studies have stated that MRI is the examination of choice because it offers superior soft-tissue contrast and multiplanar imaging.5, 7, 8, 9 However, the expanding capabilities of MDCT, including isotropic multiplanar reconstruction, improved temporal and spatial resolution, and angiographic analysis 10 has thrown into doubt the previously accepted case for MRI superiority.

CT investigation of jugular foramen masses is rarely performed at the arterial phase. A vascular timed CT technique has been applied to a series of patients with tinnitus and showed excellent differentiation of arterial, venous, and bone anatomy, but unfortunately no tumours were found in their 16 consecutive patients.11 Unlike dedicated arterial-phased imaging, the timing of any standard contrast-enhanced CT performed with the usual non-timed minutes delay will show no differential enhancement between the two disease entities (Fig. 5). MDCT angiography using a radiodensity assessment only, accurately differentiated all the glomus tumours from the neuromas.

View Large Image Download to PowerPoint

Figure 5 CT venogram of a right-sided neuroma (arrow) showing contrast enhancement even at 60s delayed timing.

Bone destruction is a well-documented CT characteristic of skull-base tumours. Glomus tumours classically cause an erosive or permeative pattern at a late stage,12 with neuromas resulting in smooth enlargement often with a sclerotic margin. The findings of the present study on bone detail were not included in the results because interpretation was biased by the characteristic enhancement of the adjacent tumour. However, in two of the neuromas, the bone changes were permeative rather than smooth, and so would have indicated a glomus. Furthermore, small tumours of any type do not necessarily affect the bone.12

Appreciation of the vascular nature of these lesions aids the pathological diagnosis and provides the surgeon with valuable preoperative information 9 and selects those who may benefit from preoperative embolization. MDCTA can avoid the need to perform invasive catheter angiography purely for diagnostic purposes, as it has been used in the past.9, 10 MDCTA identified arterial feeders in eight of the nine glomus tumours, and in none of the neuromas. In addition to abnormal arteries, high-flow glomus tumours are associated with enlarged and early filling draining veins, present in all cases in this study.

MRI employing the standard sequences, as in the present study, does not offer additional vascular information. The present results show that MDCTA allowed more accurate assessment of the jugular vein, informing the surgeon as to whether or not the vein could be sacrificed, needs to be retained, or is invaded by tumour. MR arteriography and venography have been shown to be effective in enabling visualization of the blood vessels of glomus tumours. A limitation of the present study is that we retrospectively compared MDCTA against the restricted capabilities of standard axial MRI. A study comparing the angiographic potential of both methods may be useful, although a comparative study determining the efficacy of three different MR angiographic (MRA) techniques involving paragangliomas has already shown MRA to be limited by demonstrating only some of the tumour feeding arteries.13

The results of the present study using MDCT in assessing skull-base masses has changed our practice from using MRI to CT, but many patients are referred to us having undergone MRI at their local hospital. Hopefully this article will alter the practice in the referring hospitals also.

In conclusion, MDCTA was used to accurately differentiate between glomus tumours and neuromas in all patients by demonstrating a significant difference in the contrast enhancement intensity between the two tumours, enabling the correct diagnosis to be readily made. MRI can identify glomus tumours when the characteristic flow voids are present, but this is an inconsistent finding. MRI is less reliable in the diagnosis of neuromas, where the intensity of contrast enhancement is similar to that of a glomus tumour. MDCTA also offers vessel analysis, giving invaluable arterial and venous information to the surgeon and advising on the possible benefits of preoperative embolization. The findings of the present study indicate that MDCTA can be the sole examination in the assessment of possible masses at the jugular fossa, offering accurate pathological identification and vessel assessment, thus surpassing MR as the supposed “examination of choice” in such clinical situations.