Introduction

The thoracic outlet syndrome (TOS) complex comprises three related disorders with affection of the neurovascular bundle in its pass through the costoclavicular space: classic TOS, vascular TOS and non-specific TOS [1]. Two of these, classic TOS and non-specific TOS, lead to neurological symptoms. Classic TOS (true neurological TOS or neurogenic TOS) is caused by compression of spinal nerves C8 and Th1 or the inferior trunk of the brachial plexus, often by a taut fibrous band which classically extends from a rudimentary cervical rib or elongated transverse process to the first thoracic rib [2, 3]. Compression of these lower parts of the brachial plexus leads to weakness and wasting of the thenar and hypothenar eminence, the ulnar hand intrinsic muscles and in severe cases the medial forearm muscles [4]. Sensory abnormalities including pain and paraesthesiae of the medial arm and forearm may be less pronounced in classic TOS. Therapy of classic TOS entails surgical sectioning of the fibrous band and in some approaches also of the cervical rib or the first rib [4, 5]. Contrary to classical TOS, non-specific TOS (or disputed TOS) has been a controversial subject and no consensus exists regarding the underlying pathology of non-specific TOS. Sensory complaints are generally reported as the predominant manifestation but motor weakness is also commonly described [5, 6]. Proposed therapies range from conservative treatment with specific physical therapy to surgical removal of the first thoracic rib and adjacent muscles [5–7].

Definitive diagnosis is notoriously difficult in both classic TOS and non-specific TOS. Differential diagnoses include motor neuron diseases, ulnar or median neuropathy, C8 radiculopathy, other brachial plexus lesions from neoplasms or inflammatory lesions. Electrophysiological studies play an important role in differentiating these conditions. Imaging studies have so far not offered significant aid in the diagnosis of TOS. Radiography is usually performed to confirm or exclude the presence of a cervical rib [8] but the presence of a cervical rib is not specific for a diagnosis of TOS since it has a prevalence of 0.7-1.6 % in the general population [9, 10]. Therefore, detection of more specific pathomorphological correlates for the confirmation of this entity is desirable.

Technical advances in magnetic resonance imaging (MRI) pulse sequences and higher field strength now allow for detailed imaging of the peripheral nervous system, also termed MR neurography (MRN) [11–13]. The typical MRN criterion for a nerve lesion is increased signal on T2 weighted (T2W) images confined to the site of the nerve lesion [14, 15]. The brachial plexus is an especially important target for MRN [16] since it is only with difficulties accessible to neurophysiological examination. In this investigation, we tested whether an optimised MRN protocol for the brachial plexus using custom-made surface coils was able to detect pathomorphological correlates of TOS, and whether these imaging correlates could then be confirmed by surgery.

Materials and methods

This prospective study was approved by the Ethics committee of the Medical Faculty of Heidelberg University (S-057/2009) and written informed consent was obtained from all participants. The inclusion criterion was a clinical diagnosis or at least high suspicion of classic TOS or non-specific TOS, with ENG (electroneurography)/EMG (electromyography) ruling out radiculopathy or any peripheral neuropathy located further distal than the plexus level. Exclusion criteria were previous surgery at the brachial plexus, including previous surgery for TOS or known tumour or any other mass lesion in or near the brachial plexus. Out of 36 patients examined, 4 were excluded because of previous TOS surgery, 1 because of post-surgical pressure palsy with pseudoneuroma formation and 1 because of subsequently diagnosed inflammatory neuropathy, leaving 30 patients (19 women, 11 men; age 40.7 ± 12.9 years) in the study group.

Patients were examined at the Department of Neuroradiology of Heidelberg University Hospital, between 08/2010 and 10/2012. MRN examinations were carried out on a 3-T unit (Magnetom VERIO; Siemens, Erlangen, Germany). All patients received the same examination protocol, which was specifically developed for imaging of the brachial plexus from the spinal nerve root level to the most distal levels in the axilla (Tables 1).

Table 1 MRI sequences
Full size table

The protocol was designed in the following way: the SPACE (3D turbo spin echo with variable flip angle) STIR (short tau inversion recovery) sequence provided a bilateral overview of the brachial plexus and the cervical and upper thoracic spine. The T2W SPAIR (spectral adiabatic inversion recovery) was best for displaying nerve lesions, while the PDW (proton density weighted) SPACE gave a 3D view of plexus continuity and surrounding tissue. The latter two sequences were acquired using a custom-designed coil dedicated for high-resolution imaging of the supra-, retro- and infra-clavicular brachial plexus (NORAS, Würzburg, Germany). This is a two-element 2 x 4-channel surface coil (clothes-pin-coil) positioned on the corresponding body surface on the symptomatic side. Multi-planar image reconstruction was possible on the T2W SPACE STIR sequence as well as the PDW SPACE sequence with isotropic voxel size. Contrast agent was not administered unless clinical examination indicated additional vascular compression. Reliable discrimination of vessels against nerves was possible by their hyperintense signal and tubular winded course which was assessed slice-by-slice on the T2W SPAIR sequence and, if necessary, on PDW SPACE.

The evaluation of MR images was performed by two neuroradiologists (P.B., M.P.) with more than 4 and 7 years of training in MRN, respectively. Additional three-dimensional image evaluation and reconstruction was undertaken by these two readers since standard reconstruction in axial and sagittal orientation does not reliably allow identification of subtle abnormalities of the brachial plexus and its surrounding tissues. Images were evaluated for anatomical anomalies such as bony extensions of cervical vertebrae or cervical ribs and hypointense fibrous bands extending from these, which are known to predispose for TOS. The brachial plexus was assessed at every level from spinal nerves to plexus cords for potential compression by adjacent structures and for signal changes on T2W indicative of nerve lesions. Only patients with positive MRN findings were subsequently surgically explored.

Results

Out of 30 patients included in this study, seven (23 %) displayed unambiguous findings in MRN as pathomorphological correlates of TOS (Table 2). A T2W lesion selective to the compressed plexus elements was detected in all of these patients (Figs. 1, 2 and 3). Four of them were cases of classic TOS (patients 1, 2, 4 and 5) and three had symptoms of non-specific TOS (patients 3, 6 and 7). In five (patients 1, 2, 3, 4 and 7), a fibrous band was detected which originated from a cervical rib in three cases. One patient had a pseudarthrosis between a cervical rib and the first rib, another had a synostosis between the first and second rib.

Table 2 Clinical information of patients with positive MRN findings
Full size table
Fig. 1
figure 1

a Radiography proves the presence of bilateral cervical ribs. b Reconstructed sagittal PDW SPACE image displays the entire course of a taut fibrous band (arrows) from the cervical rib (arrowhead) to the first thoracic rib, directly situated along the inferior trunk. c Sagittal T2W SPAIR displays increased T2W signal extending from the inferior trunk to the medial cord displayed here

Full size image
Fig. 2
figure 2

a Elongated transverse process (arrowheads) of vertebra C7 in the coronal plane. b, c Sagittal T2W SPAIR shows compression of the inferior trunk between an elongated transverse process (arrowhead) and the subclavian artery (sa) as well as a fibrous band (arrows). The exact insertion site of the fibrous band into another rib or the pleura could not be ascertained by imaging in this case. d Selective increased T2W signal (*) in those plexus fascicles that become the medial cord. All other plexus structures are normointense

Full size image
Fig. 3
figure 3

a Coronal reconstructed view in PDW SPACE of the fibrous band (arrow) in close proximity to the inferior trunk. b Sagittal reconstructed view in PDW SPACE of the fibrous band extending from the cervical rib (arrowhead) downwards. c Sagittal reconstruction of the 3D SPACE STIR shows increased T2W signal in the C8 spinal nerve (*) and inferior trunk (not on display) directly proximally to the contact with the fibrous band, along with slight calibre increase

Full size image

Ultrasound was not part of the standard diagnostic work-up of the study but was able to show a compression of the relevant plexus elements in four patients (patients 2, 4, 5 and 7). Fibrous bands were suspected by ultrasound in three of these cases but were depicted clearly only on MRN.

Surgery subsequently confirmed positive MRN findings (Figs. 4 and 5). All operated patients showed significant improvement in their symptoms.

Fig. 4
figure 4

a Sagittal slice of T2W SPAIR shows a thin fibrous band (arrows) which originates from an elongated transverse process and compresses the C8 spinal nerve and inferior trunk dorsally. The inferior trunk is swollen (*) with slightly increased T2W signal. b Intraoperative preparation of the fibrous band via a supraclavicular approach. Left is mediocaudal, up is mediocranial. The red vessel loop is on the subclavian artery. c View of the swollen inferior trunk after sectioning of the fibrous band and the hypertrophied scalene muscles

Full size image
Fig. 5
figure 5

a coronal SPC STIR of the brachial plexus. A T2W lesion of the inferior trunk (*) is observed distal to a hypointense longitudinal structure (arrow). b Sagittal reconstruction shows this structure to be a fibrous band. Directly ventral to this band is the inferior trunk, which yields high T2W signal, whereas superior and medial trunks appear normal. c Surgical exploration confirms a complex compression of the inferior trunk (*), mainly by a thickened fascia of the middle scalene muscle (arrow), corresponding to the fibrous band suggested by MRN

Full size image

In the remaining 23 patients with clinical diagnosis or suspicion of TOS, MRN did not detect abnormalities, and surgical exploration was to our knowledge not performed.

Discussion

The results reported here demonstrate that MRN can indeed detect specific findings to confirm a diagnosis of TOS. Two diagnostic criteria are helpful in the imaging diagnosis: (1) the detection of an anatomical structure compressing the lower part of the brachial plexus and (2) the presence of an associated abnormal focus on T2W within the compressed plexus portions but sparing the upper plexus. Both criteria were positive in seven of 30 patients described here. Imaging findings were subsequently confirmed by the diagnostic “gold-standard” of surgical exploration of the lower brachial plexus and followed by resection of the anatomical structure responsible for entrapment.

In five out of seven cases, a fibrous band was found compressing the lower parts of the proximal brachial plexus. The occurrence of fibrous bands as compressive structures was reported by Thomas and Cushing in 1903 [2] and has since been observed regularly in situ by anatomists and surgeons [3, 9]. The reliable confirmation of the presence of a compressive fibrous band has to our knowledge not been achieved previously. The detection by MRN in our patients was made possible by specially designed surface coils and an optimised protocol devised for imaging of the brachial plexus.

In addition to the described anatomical structures compressing the brachial plexus, the presence of a T2W abnormality within the affected spinal nerves or inferior trunk—but sparing other plexus elements—was observed in all patients. Peripheral nerves are normally isointense to slightly hyperintense to muscle. Hyperintensity on T2W indicates some pathological alteration [13] which is not specific for any certain aetiology but highly sensitive to nerve lesions and very precise for their spatial localisation, provided that artefacts such as the magic angle have been excluded [17]. In the case of entrapment neuropathies, an abnormally high T2W signal within a peripheral nerve should be confined to the site of the lesion [14, 15]. In our patients, the presence of a selective T2 lesion in C8 or the inferior trunk substantiated the presumed entrapment as well as the site of the nerve lesion.

Interestingly, morphological correlates were found not only for classic TOS but also for three cases with symptoms of non-specific TOS. These were due to a fibrous band from an elongated transverse process in two cases and due to a synostosis between first and second thoracic rib in another. These morphological substrates, in conjunction with selective T2 lesions in the compromised plexus structures, left no doubt of a morphologically identifiable underlying pathology, which is occasionally disputed for non-specific TOS, as the name “disputed” TOS suggests [6]. MRN identified a clear surgical target in these patients to decompress the brachial plexus, and surgery resulted in pain relief for them. However, the number of patients with negative findings also suggests that there are cases of non-specific TOS without detectable correlate on MRI.

A number of limitations apply to this study. Definitive surgical exploration for those patients with negative MRN findings was to our knowledge not performed. We, therefore, cannot determine the number of false-negative findings in MRN. This means that while the method had a positive predictive value of 100 % in our patient group, the negative predictive value remains undetermined. We believe the sensitivity of the method to be high for classic TOS, which also has an electrophysiological correlate of structural nerve damage, and lower for the heterogeneous non-specific TOS. While we did find morphological correlates of TOS in three patients with non-specific TOS, these may not be representative for most patients with non-specific TOS, which may be highly diverse in symptoms, their aetiology and adequate treatment. In the interpretation of images, care must be taken not to misinterpret vascular structures as fibrous bands since they can yield similarly hypointense signals. This differentiation is possible by following the course of vascular structures and their eventual entry into larger arteries or veins.

In conclusion, this is the first study of thoracic outlet syndrome in which subtle anatomical anomalies in the course of the brachial plexus, together with T2W abnormalities of the compromised plexus portions, have confirmed the clinical diagnosis. The proposed technique should be of value in the assessment of this diagnostically challenging neurological disease.