Detection of the Stellate and Thoracic Sympathetic Chain Ganglia with High-Resolution 3D-CISS MR Imaging

BACKGROUND AND PURPOSE: Despite the importance of the sympathetic nervous system in homeostasis and its putative role in various disease states, little is known regarding our ability to image the sympathetic chain and sympathetic chain ganglia, perhaps owing to their small size. In this retrospective study, we sought to evaluate the normal anatomy of the sympathetic chain ganglia and assess the detectability of the sympathetic chain and sympathetic chain ganglia on high-resolution 3D-CISS images. MATERIALS AND METHODS: This study included 29 patients who underwent 3D-CISS MR imaging of the thoracic spine for reasons unrelated to abnormalities of the sympathetic nervous system. Patients with a prior spinal operation or visible spinal pathology were excluded. The sympathetic chain ganglia were evaluated using noncontrast 3D-CISS MR imaging. Statistical analyses included t tests and measures of central tendency. The Cohen κ statistic was calculated to evaluate interrater reliability. RESULTS: The stellate ganglion and thoracic chain ganglia were identified in all subjects except at the T10–T11 and T11–T12 levels. The stellate ganglion was found inferomedial to the subclavian artery and anterior and inferior to the transverse process of C7 in all subjects. Thoracic sympathetic chain ganglia were identified ventral to the costovertebral junction in all subjects from T2 to T10. There was strong interobserver agreement for the detection of the sympathetic chain ganglia with κ > 0.80. The size, shape, and location of these structures corresponded with gross anatomic and surgical observations. CONCLUSIONS: The thoracic sympathetic chain ganglia can be identified on precontrast 3D-CISS MR imaging. This technique may aid in the initial evaluation of stellate ganglion and/or sympathetic chain ganglia size and signal change for comparison in future studies.

T he autonomic nervous system is composed of a sympathetic component, responsible for the "fight or flight" response, and a parasympathetic portion, known for "rest and digest" functions. The sympathetic component differs in structure from the remainder of the peripheral nervous system. The sympathetic chain is generally divided into 2 nerve trunks that extend the entire length of the vertebral column along its lateral aspect. [1][2][3] The stellate ganglion provides sympathetic supply to the head, neck, and up-per limbs, while the sympathetic chain ganglia (SCG) provide key innervations to the thoracic viscera, including the heart. [1][2][3][4][5] The stellate ganglion receives nerve fibers from the C6, C7, and T1 nerve roots with occasional contributions from T2, T3, and T4. The thoracic sympathetic chain receives nerve fibers from nuclei located within the lateral gray horn of the T2-T8 spinal segment that leave the spinal cord in the anterior rootlets of thoracic nerves, then traverse the white communicating rami, to enter the sympathetic trunk, synapsing onto their target postganglionic neurons in the paravertebral thoracic sympathetic ganglia. The postganglionic fibers traveling to the upper limb originate from the middle cervical ganglion, the stellate ganglion, and the upper thoracic ganglia; join the thoracic nerves through the gray rami communicantes; and then enter the brachial plexus and travel along different nerves to their end targets. 6,7 Stellate ganglion and SCG blocks are performed for various conditions, including complex regional pain syndrome, postherpetic neuralgia, herpes zoster, causalgia, or intractable angina. 1 An SCG block can also be performed to treat patients with vaso-spasm (eg, primary or secondary Raynaud phenomenon), phantom pain, cardiac arrhythmias (including long QT syndromes), and primary hyperhidrosis. [1][2][3][4][5] Patients are generally referred for SCG block due to persistence of symptoms (pain, vasospasm, hyperhidrosis, and so forth) despite medical management. 4,[8][9][10] An SCG block is routinely performed in the clinical setting with sonography, fluoroscopy, or CT used to provide anatomic landmark guidance. However, the SCG itself is never directly visualized before the procedure. 10 Although the efficacy of interventions is well-established, a considerable proportion of patients do not see improvement and/or resolution of symptoms (eg, continued pain in up to 27.2%-33% in patients treated with SCG radiofrequency ablation for complex regional pain syndrome and recurrence of symptoms within 7 weeks in Ͼ66% of cases). 11 The reasons for the lack of short-term and long-term treatment benefit are not well-understood. Authors have suggested several possibilities, including incomplete SCG block secondary to anatomic variation. 2,[11][12][13][14] In other words, it is possible that in some failed cases, the SCG was not directly and/or incompletely targeted because the structure itself was not actually visualized and was in a location variant from the commonly described one.
The peripheral components of the sympathetic nervous system have not been previously well-characterized on cross-sectional imaging, to our knowledge. There are only 2 small case series that identified the sympathetic chain ganglia on T1weighted MR imaging, 2,12 primarily identifying the location of only the stellate ganglion.
The 3D-CISS refocused gradient-echo MR imaging sequence is a high-spatial-resolution technique performed to evaluate intracranial and spinal nerves. 15-20 3D-CISS MR imaging generates images with contrast ideally proportional to the ratio of T2/T1 relaxation time, generating what appear to be heavily T2weighted images. 18 In our experience, 3D-CISS MR imaging before intravenous contrast administration provides visualization in vivo of the sympathetic chain ganglia within the paravertebral fat. In this retrospective observational study, we aimed to evaluate the normal anatomy of the stellate ganglion and thoracic sympathetic chain ganglia and assess the detectability of each of the thoracic sympathetic chain ganglia on noncontrast CISS images.

Study Group
This observational study was approved by the Johns Hopkins University institutional review board, and informed consent was waived in this Health Insurance Portability and Accountability Act-compliant retrospective study. Initially, 29 subjects underwent high-resolution CISS clinical MR imaging for the evaluation of either back pain or CSF leak. Subjects with other spine pathology, including 2 spinal tumors and 3 cases of prior spinal operations, were excluded from the study. Subjects were included in the study if the entire thoracic spine, including the thoracic inlet, was imaged. Studies were evaluated for image quality, and 1 scan with images degraded by motion artifacts was excluded. On the initial review of images, the SCG was noted to enhance uniformly on postcontrast images and became indistinct from surrounding tissue. Therefore, only precontrast CISS images were used to collect data on the SCG; 6 scans with only postcontrast images were ex-cluded from the study for this reason. Only the first valid examination was included for each subject when multiple examinations had been performed. Ultimately, 24 examinations of 24 subjects were included for evaluation (Fig 1).

Study Design
From the cohort of 24 patients who met the inclusion criteria, 480 normal thoracic sympathetic chain ganglia and 48 stellate ganglia were evaluated. Quantitative image analysis was performed by 2 independent readers blinded to each other's interpretation. The first reader, reader A, is a fellowship-trained radiologist with Ͼ6 years of experience. The second reader, reader B, is a first-year neuroradiology fellow. The detectability of the sympathetic chain ganglia was evaluated using precontrast enhanced 3D-CISS MR imaging. The expected location of the SCG was extrapolated from anatomy and surgical literature. 6,7 Both readers were credentialed neuroradiologists and had reviewed the anatomic literature before this study, to reach a consensus regarding the expected location of the sympathetic chain ganglia. A training or washout period was not attempted due to the limited number of subjects who met the inclusion criteria. The stellate ganglion is most commonly located anterior to the C7 vertebra transverse process, below the subclavian artery, superior to the first rib. 6,7 The thoracic SCG are located at the costovertebral junction anterior to the head of the ribs. 6,7 Confirmation of stellate ganglion and the SCG was made, ensuring connection of the SCG with spinal nerves and interconnection within the SCG chain on multiplanar (sagittal and coronal) reconstructions. The stellate ganglion is not entirely spheric or ellipsoid. Therefore, a calculation of volume using the largest diameter in each 3D plane would be misleading. Instead, the authors opted to quantify size consistent with reports in the anatomic literature. 6,7,[21][22][23][24][25][26] The largest 2 axial dimensions of the ganglion were recorded (in millimeters), and surface area was subsequently obtained. Both readers were aware of the study objectives, both reviewed the studies at different locations and times, and both were blinded to subjects' clinical data.

Statistical Analysis
Statistical analysis was performed using MedCalc for Windows, Version 12.5 (MedCalc Software, Mariakerke, Belgium). After controlling for subject sex, we performed measures of central tendency (mean, median) and a paired t test to assess differences between the mean sizes of the left and right SCG and to discern whether there was a difference in the observed size of the SCG between the right and left sympathetic chains. Additionally, a test was performed to evaluate interrater reliability.

RESULTS
In our cohort of 10 men and 14 women (mean age, 51.7 years), the stellate ganglion and thoracic chain ganglia were successfully identified in all subjects except at the T10 -T11 and T11-T12 levels. The subjects enrolled in the study tolerated the MR imaging procedure well, and there were no acute complications during or immediately after this study.
Precontrast CISS demonstrated the stellate ganglion and SCG to be isointense relative to gray matter noted in the spinal cord. The stellate ganglion (Fig 2) and thoracic SCG (Figs 3 and 4) were present in all subjects in the expected location. The stellate ganglion (also referred to as T1 in anatomic texts 6 ) was the largest of the sympathetic chain ganglia, with mean sizes of 36.62 Ϯ 9.63 mm 2 on the right and 36.90 Ϯ 8.9 mm 2 on the left. These size estimates were concordant with an anatomic study performed by Zhang et al. 7 The next 2 largest SCG were at T2 (13.70 Ϯ 5.94 mm 2 on the right and 13.56 Ϯ 5.25 mm 2 on left) and T3 (10.23 Ϯ 2.85 mm 2 on the right and 10.21 Ϯ 2.42 mm 2 on the left, Fig 3). The size of the SCG decreased in a craniocaudal direction from the stellate ganglia to the thoracolumbar junction (Fig 5). The T4to T10-thoracic chain ganglia ranged from 4.1 to 19.35 mm 2 . Table 1 summarizes the mean size and SD of the thoracic SCG. The 2-tailed P value demonstrated no significant difference between the mean values of the right and left sympathetic chain ganglia with a P ϭ .82 (95% CI, Ϫ0.108 to 0.088). There was strong interobserver agreement for detection of the SCG with Ͼ 0.8.

DISCUSSION
Isotropic or near-isotropic high-resolution 3D-CISS imaging provides the capability of visualizing structures not typically seen with standard spine MR imaging techniques. Our study shows that stellate and thoracic sympathetic chain ganglia can be readily characterized on precontrast 3D-CISS MR imaging. Our results demonstrate the stellate ganglia to be the largest of the SCG with a   gradual decrease in size of the SCG at subsequent levels. Additionally, the stellate ganglia were noted to be the most lateral in location, with gradual medialization of the sympathetic chain trunk, with lower thoracic levels situated most medially. These findings are consistent with gross anatomic studies. 6,7,[21][22][23][24][25][26] Prior studies have reported the size of the stellate ganglia, which vary ranging from 10 to 250 mm 2 ( Table 2). The SCG size observed in our cohort, 35 mm 2 , falls within the range reported by Hogan and Erickson, 3 Moore, 24 Kiray et al, 25 Zhang et al, 7 and Marcer et al. 26 No prior systematic study identifies and characterizes the sympathetic chain ganglia using 3D-CISS technique. Two small case series previously reported by Hogan and Erickson 3 and Slappendel et al 12 have attempted to identify the stellate ganglion using MR imaging (with cohorts of 10 and 8 subjects, respectively). Both studies used T1-weighted imaging to demonstrate the stellate ganglion; however, these studies did not demonstrate the remainder of the thoracic sympathetic chain ganglia.
There are several limitations of our study. This is a retrospective study and thus has disadvantages that are inherent to the design, including selection bias. Our cohort included only those patients who underwent high-resolution imaging for back pain or CSF leak, a population that may not be representative of the normally distributed population. Sample size is another limitation of the study, which may result in under-or overestimation of the mean size of the SCG at various levels. However, because our data appear to be congruent with the anatomic literature, the degree of variance between our results and the overall population may not be large. "Experimenter effect" is another potential limitation of this study. Because both readers were aware of the aim and design of this study, the measurement and final consensus of imaging characteristics could have been biased. Thus, our findings should be confirmed with a larger normally distributed patient cohort free of the aforementioned limitations.

CONCLUSIONS
The thoracic sympathetic chain ganglia can be readily seen and well-characterized on precontrast 3D-CISS MR imaging. This technique can aid in the initial evaluation of potential stellate and/or SCG pathology and allow posttreatment follow-up.     7 19.3 6.3 Marcer et al (2012) 26 18.5 8.17 4.5 MRI-based measurement Hogan and Erickson (1992) 3 10 (in largest dimension)