The superior, middle, and inferior ganglia comprise the cervical portion of the sympathetic system. As the most cranial part of the paravertebral sympathetic trunk, the superior cervical sympathetic ganglion (SCSG) is the largest of the three and serves to distribute sympathetic fibers to the head and neck region (12). It is composed of multipolar ganglion cells with intervening nerve fibers, and receives two or three branches from the ascending pharyngeal artery (3). Previous cadaveric studies described the SCSG as fusiform or flattened in shape and embedded within connective tissue anterior to the transverse process of the second, third, and rarely, the fourth cervical vertebrae (245). Its location has been confirmed by several reports (246789).

Clinically, pathologies of the SCSG, including iatrogenic injuries during spine surgery, can result in Horner's syndrome; ipsilateral blepharoptosis, miosis, and facial anhydrosis (610). For patients with neuropathic facial pain, such as trigeminal neuralgia or postherpetic neuralgia, percutaneous nerve block of the SCSGs can provide pain relief and is an alternative to stellate ganglion block (11112). Therefore, it is important to determine the precise anatomical location of the SCSG on pre-procedural imaging.

Recently, Yuen et al. (13) reported a case of enlarged SCSG after radiation therapy for nasopharyngeal carcinoma-mimicking metastatic retropharyngeal lymph node, which presented as a well-defined mass with avid enhancement at the level of the C2 vertebra on 3T-MRI. Normal imaging appearance of SCSGs and characteristics that differentiate SCSGs from retropharyngeal lymphadenopathy remain to be determined. To the best of our knowledge, there are no reports of the normal imaging appearance of SCSG on neck MRI, including anatomical variations. We intended to assess the imaging anatomy of normal SCSGs on 3T-MRI using predefined anatomical criteria and evaluate MRI features of SCSGs to differentiate them from pathologies such as retropharyngeal lymphadenopathy. In our clinical experience, we frequently observe intraganglionic hypointensity near the center of SCSGs on axial MR images, hence, we also determined whether intraganglionic hypointensity might be a characteristic finding of SCSGs.

Fifteen cases were excluded during consensus reading of 106 sites of necks for the following reasons: retropharyngeal lymphadenopathy (n = 12), parapharyngeal tumor (n = 2), and surgical excision of vagal schwannoma (n = 1). Of the remaining 91 neck sites, 66 (73%) in 38 patients were identified as definite SCSGs and 25 (27%) in 15 patients were identified as probable SCSGs. The imaging characteristics of definite and probable SCSGs were homogeneous high signal intensity on T2WI, intermediate signal intensity on T1WI, and homogeneous and marked enhancement on CE FS T1WI.

Independent reading by the two radiologists revealed that 16 of the 25 probable SCSGs (64%; 18% of all SCGCs) were located anterior to the LCM and medial to the ICA; 6 (24%; 7% of all SCGCs) were located lateral to both the ICA and the LCM, and 3 (12%; 3% of all SCGCs) were located posterior to the ICA and lateral to the LCM (Fig. 2). The interrater and intrarater agreements for the relative locations of probable SCSGs were excellent (interrater agreement, κ = 0.874–0.918; intrarater agreement, κ = 0.829–0.927).

Intraganglionic hypointensity was present in 59 of the 66 definite SCGCs (89%) on CE FS T1WI, 58 of the 66 definite SCGCs (88%) on T2WI, and 21 of the 66 definite SCGCs (32%) on T1WI by reader 1; and in 60 (91%) on CE FS T1WI, 58 (88%) on T2WI, and 27 (41%) on T1WI by reader 2. Interrater and intrarater agreements for visibility of intraganglionic hypointensity were good-to-excellent, ranging from 0.627 to 0.779 for interrater agreement and 0.779 to 1.000 for intrarater agreement. Among probable SCSGs, intraganglionic hypointensity was present in 23 of the of 25 probable SCSGs (92%) on CE FS T1WI, 23 of the of 25 probable SCSGs (92%) on T2WI, and 8 (32%) of the of 25 probable SCSGs on T1WI by reader 1; and 22 (88%) on CE FS T1WI, 21 (84%) on T2WI, and 8 (32%) on T1WI by reader 2. Interrater and intrarater agreements for visibility of intraganglionic hypointensity were also excellent for each pulse sequence, ranging from 0.740 to 0.915 for interrater agreement and 0.839 to 1.000 for intrarater agreement. Occurrence of intraganglionic hypointensity showed no significant difference between definite and probable SCSGs for any pulse sequence (all p > 0.05).

The SCSGs were most commonly found at the level of the C2 vertebra (n = 61, 67%), followed by between the C2 and C3 vertebra (n = 17, 19%), and at the level of the C3 vertebra (n = 13, 14%), in order. The 91 SCSGs had maximum axial diameter (mean ± SD), 7.7 ± 1.8 mm (range, 4.8–13.2 mm), minimum axial diameter, 3.7 ± 0.9 mm (range, 2.0–6.5 mm), and longitudinal diameter, 27.9 ± 5.9 mm (range, 15.0–45.0 mm).

The relative location of the SCSG to the ICA, the relative location of the SCSG to the LCM, the size of the SCSG, or the presence of the intraganglionic hypointensity showed no significant differences between the right and the left sides of necks (all p > 0.05).

Susceptibility weighted imaging showed the intraganglionic hypointensity as a marked hypointense focus on the magnitude image that became hyperintense to the same degree as venous structures on the phase image, suggesting an intraganglionic vessel (Fig. 3) (15). Histologic examination of a harvested SCSG from an adult cadaver showed an SCSG with an axial diameter of 4.5 mm that was encapsulated by thick connective tissue, evenly distributed ganglionic cells, and a central vascular structure with a diameter of 0.6 mm, which was confirmed as a venule. There was no dense fibrosis or dense deposit of lipofuscin granules in the ganglion cells near the center of the SCSG (Fig. 4).

In this study, 3T-MRI was successful in localizing 66 of 91 neck sites (73%) as definite SCSGs based on all four criteria. In the remaining 25 neck sites, probable SCSGs were anterior to the LCM and medial to the ICA (n = 16), lateral to both the ICA and the LCM (n = 6), or posterior to the ICA and lateral to the LCM (n = 3). Intraganglionic hypointensity was most conspicuous on CE FS T1WI, suggestive of potential as a distinguishing feature of SCSGs. Several cadaveric studies have focused on the normal anatomy of the SCSG with regards to surgical approach of the cervical region or cervical sympathetic block (616).

Interestingly, we found that 27% of the SCSGs were not located at the position described in the literature. The probable SCSGs in this study were located more medial (anterior to the LCM) or more lateral (lateral to the ICA), and even posterior to the ICA. There could be several reasons for this variation. First, the LCM converges medially in a cranial direction (56) and the cervical sympathetic trunk converges medially in a caudal direction. As the SCSG is the most superior part of the sympathetic trunk, the SCSG could be more medially located depending on the vertebral level of the SCSG. Second, the neck position for MR imaging might differ from the neck position in previous cadaveric studies (614), because some of the studies were performed with hyperextension (814) and rotation of the neck to the contralateral side (6) to simulate the anterior approach to the cervical spine, which might alter the relative position of SCSGs to the LCM. Third, lateral or posterior location to the ICA (10% of the probable SCSGs) has never been described in the literature and this finding could be secondary to tortuosity of the ICA (17). Because the SCSG is embedded within the deep layer of the deep cervical fascia (6), the ICA invested within the carotid sheath might change its position relative to the SCSG (18). Regardless of the causes of these anatomic variations, SCSG can vary in location relative to the ICA and may diverge from the LCM in the caudocranial direction.

Occurrence of intraganglionic hypointensity was another notable finding of this study. It was likely that the intraganglionic hypointensity was associated with a vessel or fatty connective tissue engulfed within the ganglion by traversing nerve fibers. However, microscopic examination only revealed a venule at the center with no intervening connective tissue, fibrosis, or dense deposit of lipofuscin granules in the ganglion cells. However, a final conclusion should be made with caution because the structure also appeared hypointense on CE FS T1WIs. We were unable to correlate the intraganglionic hypointensity with any histologic feature of a ganglion due to lack of reports in the literature. In addition, we have only indirect evidence in support of the structure as a venule. Therefore, further studies are required to reveal the true nature of the intraganglionic hypointensity in the near future. Despite the unclear anatomical nature of the structure, the radiological importance of the intraganglionic hypointensity is highlighted in patients with head and neck cancer with retropharyngeal lymph node metastases. Based on the occurrence of typical intraganglionic hypointensity, radiologists can confirm the paired, prevertebral, ovoid, enhanced structures at the expected location as the bilateral SCSGs rather than metastatic retropharyngeal lymphadenopathy.

The major limitation of our study is the absence of surgical confirmation of the nature of the imaging findings. Surgical confirmation was not performed, because the identified structure was strongly presumed as normal, and there are potential detrimental consequences of surgery or biopsy. Verification bias was controlled by the strict application of evidence-based criteria based on a literature review and excellent interrater and intrarater agreement. In addition, we excluded any case with confusing abnormalities in retropharyngeal, parapharyngeal, or prevertebral space or with anatomical distortion secondary to previous surgery. Another limitation is that although our additional studies suggested that the intraganglionic hypointensity might be an intraganglionic venule, this should be interpreted with caution due to the limited number of cases.

In conclusion, 3T-MRI visualized 73% of SCSGs as vertically-oriented fusiform or ovoid-shaped T2-hyperintense strongly enhanced structures located medial to the ICA and lateral to the LCM between the C2 and C4 vertebrae. The location of the remaining 27% of SCSGs was anterior to the LCM, lateral to the ICA, or posterior to the ICA. Intraganglionic hypointensity was seen in 90% of cases and could be helpful to differentiate normal SCSGs from retropharyngeal lymphadenopathy.