There are many surgical techniques for the treatment of TOS described in literature. Surgery of the thoracic outlet can be performed trans-axillary, supraclaviculary, a combined supra- and infraclavicular approach, or with video-assisted or robotic-assisted thoracoscopic surgery.(1-6) The approach is mostly based on the preference of the surgeon. The extend of the surgery in the thoracic outlet varies enormously from performing a minimally invasive partial rib resection to a complete resection of the first rib, anterior and medial scalene muscles and complete neurolysis. Reasonable to good outcomes have been reported with all these different techniques. (1, 5, 7-21)

In our center we always perform a complete resection of every possible compressive structure in the thoracic outlet. Therefore, we do not report about performing a ‘first rib resection’ but prefer ‘decompression of the thoracic outlet’. This is based on both our experience in primary cases as well as recurrent or persistent, mainly neurogenic TOS cases. In primary surgery we see various sites of compression of the neurovascular bundle (e.g. scalenus minimus variants, fibrotic bands, aberrant insertion of the scalene muscles onto first rib and/or Sibson’ fascia). Sites that had not been visible by performing only a (partial) first rib resection. In recurrent or persistent TOS cases (recurrent complaints after initial success or persisting complaints after previous primary surgery) we often see scar tissue from a partially resected dorsal rib remnant or outgrowth, scalene muscle remnants or fibrous bands that were untouched in the primary surgery and now the cause of compression of the neurovascular bundle.(22) Other authors have also warned about recurrence due to regrowth of first ribs after partial resection or a pectoral minor syndrome amongst other compressive structures in the thoracic outlet.(19, 20, 23-26) To reduce the risk of recurrent or persistent TOS, we always perform a complete decompression of the thoracic outlet. A thoracic outlet decompression (TOD) in our hospital is performed through a trans-axillary approach (TA-TOD) under general anesthesia. A TA-TOD consists of a complete resection of the first rib (from sterno-costal cartilage into the costo-vertebral joint, surpassing and dividing the ligament of the tubercle and anterior costotransverse ligament). Any (in-) complete cervical rib is also completely removed. A partial resection (1.5–3 cm's of the lowest part) of the anterior and medial scalene muscle is performed as high as possible, including resection of possible anatomical variations (e.g. minimus scalene muscle or Sibson's muscle variant and/or fibrous bands to Sibson's fascia). A thorough neurolysis of the lower plexus (inferior trunk neurolysis, including C8 and T1 root and if possible C7) is performed routinely. By doing so aberrant fibrotic tissue is frequently encountered and removed.

Although there are multiple approaches to perform a thoracic outlet decompression, the authors favor the trans-axillary approach in primary cases for several reasons: Firstly, through the axilla, the exposure and removal of the complete first (and cervical) rib is more easily performed compared to other approaches. Secondly, the pectoral minor muscle has been frequently (up to 69%) identified as a cause of recurrence or residual TOS. (19, 23) The same trans-axillary incision can be used to perform a tenectomy of 3-5 cm's of the pectoral minor muscle. Thirdly, a trans-axillary approach leaves the supraclavicular region without any scar tissue, making possible future redo-surgery through a supraclavicular approach easier and prevents damage to the scalene fat pad. This scalene fat pad – if not scarred or otherwise damaged - can be divided at the level of the omohyoid muscle in a deep and a superficial layer and used as a wrap surrounding the completely freed brachial plexus in redo surgery.(27) The fourth and final argument in favor of a primary TA approach is the fact that it offers by far the best cosmetic result. This should not be underestimated in young patients. (13)

All patients are operated following a standardized surgical protocol. As stated before, we aim to remove all possible etiologies of compression in the thoracic outlet. However, since a minority of patients have the ‘standard’ thoracic outlet as depicted in anatomical drawings, there are many pitfalls during surgery. (28, 29) Therefore, we dissect layer by layer to adequately identify (and preserve) each essential structure (instead of focusing on the point of possible compression). Another advantage of this technique is that it creates predictability for the operating team that assists during surgery in providing exposure of vital structures and providing the required instruments in a step-by-step sequence. Following this standardized way of surgery, the team can predict the next move and anticipate to it. To facilitate this, we always use a camera during TOD-surgery.

In this article, we will describe the way a trans-axillary thoracic outlet decompression is performed in our center.

Patients are prepped by the anesthesiology team with one large bore IV catheter. We do not perform any additional locoregional anesthetic block anymore. (30) Patients are sedated and intubated. Only short acting muscle relaxants are given during intubation (succinylcholine) to make sure that there is no motor block during surgery which could obscure conduction damage of the electrocautery during surgical dissection close to the brachial plexus.

The patient is then placed on the contralateral side. But instead of turning the patient for the full 90 degrees, we limit the turn to 75 degrees. (Figure 1) A vacuum matrass is shaped to the patient's body form and activated. The head of the patient is then lifted onto a high supporting cushion to create a sharp angle of the neck towards the operated side. This relaxes the brachial plexus and facilitates manipulation of C8 and T1 branches during removal of the most dorsal part of the first rib. We fixate a small gel cushion on top of the patients head to prevent the Trimano® Fortis support arm (Arthrex, Naples, FL) accidentally damage the head of the patient during manipulation of the Trimano® arm.

The patient's arm is disinfected, draped with cotton gauze bandage, and then placed onto the arm rest that is already clicked onto the Trimano® arm system. We make sure that there is no excessive pressure of the arm rest in the crook of the arm. This system ensures a steady elevation of the arm and can be manipulated during surgery. The Trimano system is far more effective and steadier than a young trainee.

A Martin's Arm standard (Hayden Medical inc. ®, California, USA) is used to provide extra retraction during surgery. A Martin's Arm is placed on the operating table at the level of the patient's shoulder. A Deaver (Sklar corporation, Pennsylvania USA) retractor is placed onto the Martin's Arm. This retractor will provide posterior retraction of the tissue in the thoracic outlet and is exceedingly practical when resecting the posterior part of the rib. (Figure 1)

The assisting medical student or resident or surgeon provides retraction with the use of the Tebbets (Marina Medical, Florida, USA) fiber optic retractor. This retractor provides extra light in the axillary cavity during surgery but also provides suction to clear out the smoke that can be created during electrocautery. This retractor will retract the tissue in the anterior or cranial part of the thoracic outlet and is the most versatile assisting instrument during surgery.

The surgeon caries a led surgical headlight (Sunoptic®, Florida, United States) accompanied with a camera during surgery. This extra light in the thoracic outlet is very useful and the image from the camera is streamed to the screens in a hybrid operating room and provides the operating team a clear view on how the surgery is going on. The assisting surgeon gets visual feedback of the retraction that is provided and can learn the anatomy and tissue handling of the thoracic outlet. The assisting nurse can pro-actively react and provide the necessary tools for the next step in surgery.

The Ligasure™ (Maryland 23, Medtronic/Covidien IIc, Mansfield, MA, USA) device is routinely used in thoracic outlet decompressions. This bipolar device can seal and ligate blood vessels up to 7 mm and is very useful to dissect close to the brachial plexus without causing thermal damage. The resection of the scalene muscles can be performed without the risk of (minor) bleeding which might complicate recognition of the relevant anatomic structures.

An incision of 4-6 centimeters is made at the base of the axillary hair line in between pectoral major and latissimus dorsi muscles. The subcutaneous fat is split onto the less firm axillary fat. We identify the lateral border of the pectoral major muscle and create a tunnel medially towards the coracoid process of the scapula. Care is taken not to damage the medial and especially lateral pectoral nerves that run in between de major and minor pectoral muscles. The pectoral minor muscle attaches to the coracoid process; however, the fibers of the pectoral major can sometimes stick to the minor muscle. With gentle blunt dissection, the correct plane between the two muscles can be found. The pectoral major muscle fibers are placed behind the Tebbett® retractor, creating overview of the anatomy in this region. The pectoral minor muscle is isolated from all surrounding tissue with the use of peanut gauze. Beneath the pectoral minor muscle, the axillary vein, artery, and brachial plexus are in proximity. When the correct plane is dissected, a rubber loop is placed around the pectoral minor muscle and pulled caudally. An O'Shaughnessy clamp (Mahrsurgical, Bradford, United Kingdom) is then placed onto the more cranial fibers of the pectoral minor muscle and division of the muscle fibers is performed with the Ligasure™. The rubber band is released once the fibers are divided. With some traction caudally on the O'Shaughnessy clamp it is easy to transect the pectoral minor fibers in close proximity to the coracoid process. In this way we resect up to 4-5 centimeters of pectoral minor muscle fibers. After removal of the pectoral minor muscle, the retro pectoral area is inspected for any additional constraining fibers or bands. The index finger is used to control a floppy trajectory of the artery and thus brachial plexus.

We then focus onto the mobilization of the axillary fat. We create a tunnel onto the rib cage and dive below the axillary fat. This can be performed with blunt dissection with the aid of peanut gauze. During this dissection, great care is taken to identify and preserve the intercosto-brachial nerve 2 (ICB2). If this nerve is central in the incision and thus obscuring further exposition and surgery, we transect this nerve below the level of the intercostal muscles (aiming to reduce the chance of creating a neurinoma). If the nerve is more medial or lateral in the wound, the nerve is retracted behind either the Tebbett® or Deaver retractor. Fierce retraction of the ICB2 nerve almost always leads to allodynia and should therefore be avoided. (31)

The dissection of the thoracic outlet starts by identification of the subclavian vein (SV) that is completely freed from arm to costo-clavicular ligament over 4-6 cm's. The thoracic outlet is then dissected using a systematic approach going from medial to lateral. After the SV and following this technique, we dissect the subclavian artery (SA) and inferior trunc of the brachial plexus / T1 root before moving more upward into the thoracic outlet. This way, the anatomy can be identified ‘with relative ease’ and the risk of creating a tunnel in which the anatomy is unclear is averted.

After identification of the SV we dissect all the overlying tissue away towards the lateral side (SA side) of the arm. If a small vein or artery branch is present, the Ligasure™ is used to ligate these vessels. There are some side branches that are divided with the Ligasure™. The thyrocervical trunc is often identified and ligated. The mammarian artery mostly lies medially and is mostly not present in the thoracic outlet. If we do encounter this side branch, we always spare it. After the SA is freed from overlying/surrounding tissue, the nerves are freed, making absolutely sure T1 is identified. After obtaining a clear view on SV, SA en inferior trunc (C8 and T1), we again start at the medial side where the costo-clavicular ligament forms the medial boundery of the thoracic outlet. The costo-clavicular ligament can be transected now (using monopolar electrocautery). Great care should be taken for a phrenic nerve anterior to the SV and in that case in close proximity to the CCL during transection, which can be seen in 3-4% of all patients. (32) In most cases the phrenic nerve courses just posteriorly of the vein and costo-clavicular ligament and should be protected during transection. We use a forceps with a peanut on the top and place this just behind of the CCL to protect both SV and phrenic nerve.

In some patients, a subclavius muscle will be more developed and may sometimes have muscle fibers going down to the first rib. These fibers can be resected during this part of the surgery or whilst performing the first rib resection. It depends on size of the subclavius muscle whether to resect it or not (rare in NTOS cases, frequently in l VTOS cases this muscle is removed).

The SA runs lateral of the Anterior Scalene Muscle (ASM) and dissection of the SA is performed as much as possible from the ASM, carefully looking for both phrenic as well as a more caudally running accessory phrenic nerve present in 3-4% of patients. (32) We then focus on the inferior brachial plexus and identify the C8 nerve root. When we are certain that the T1 and C8 root join as inferior trunc we proceed to dissect behind the nerves towards the medial scalene muscle (MSM). In the majority of NTOS cases a fibrous or muscular scalene minimus variant in between SA and inferior trunc is found. The minimus variant is reported in literature between 7.8 and 71.7 per cent, however seems to be more present in patients with TOS. (33) Now, with clear and full vision on all structures that form the thoracic outlet, we use dissecting forceps to scoop the ASM as high as possible, preferably and in most cases after identification of the (accessory) phrenic nerve. This forceps is then opened approximately 2 centimeters and the Ligasure™ is used to transect the ASM as cranial as possible. The remaining caudal fibers of the ASM on the first rib can then be resected and removed. This creates a better overview of the thoracic outlet and its anatomy. The same maneuver is performed with the MSM fibers that were freed using peanut gauze. The long thoracic nerve is found in most cases and left laterally/cranially before the dissecting forceps is used to scoop the MSM. If we have problems scooping up the complete muscle, we transect the muscle in stages and press the dissecting forceps through the middle of the MSM fibers. Always take care not to scoop up any brachial plexus branches or the long thoracic nerve. Again, the remaining caudal fibers of the MSM are resected to obtain better overview. In case of a MSminius variant, we resect this muscle as high as possible, taking care of the small artery that often crosses from the SA towards the lateral thoracic outlet.

At this moment the first rib is identified and the inferior border is dissected free with electrocautery, taking care not to harm the pleura with this maneuver. Once a small opening is obtained, the parietal pleura on the backside of the first rib and apical pleura attached to Sibson's fascia is peeled away with blunt dissection. Using the backside of a forceps that is rotated 90 degrees to create enough space to be able to dissect underneath the first rib. A maneuver that facilitates this is a temporary period of apnea. This lowers the upper boarder of the lung and creates more space. During the resection of the first rib, we use apnea several times. Depending on the lung capacity of the patient, apnea can often be administered for multiple minutes. Once the pleura has been dissected free, the Ligasure™ is used to transect the intercostal muscles. Dorsally, towards the vertebrae, care should be taken not to harm the T1 branch that lies at the medial border underneath the first rib (or C8 especially in cervical rib cases). Once a major part of the first rib is freed, the overlying tissue (remnants of the ASM, MSM and intercostal muscle) is resected with electrocautery. A Liston (SURTEX®, Surrey, England) bone cutting forceps is used to transect the first rib medially and laterally. An eight mm bone nipper is used to resect all bony remnants. On the medial side care should be taken not to harm the SV or phrenic nerve. The phrenic nerve is posterior to the sternum and can be very close to the CSJ. At the upper side of the costosternal joint, the SV travels into the thoracic cavity. The rib is resected until the sternal cartilage is visible. The posterior aspect of the first rib can be a bit more difficult to dissect. Especially the most dorsal side of the first rib can lie very close to the T1 root. The combination of a tilted head (pillow) and a slight retraction reduction of the Trimano® enables mildly moving the T1 root away from the first rib, whilst performing bone resection with the 8mm bone nipper. Resection of the posterior part of the first rib is complete when the full cartilage of the vertebral body joint is visible. We then extensively rinse the thoracic outlet with saline fluid to remove all (microscopic) bony fragments/cells.

At the end of the procedure, the lower brachial plexus is inspected (C8-T1 forming the inferior trunc and in most cases C7) and any overlying tissue that might cause compression is resected. Orthosympatic fibers (Gray ramus communicans) that come from brachial plexus roots should not be mistaken for fibrous bands. Transaction will result in a Horner's syndrome.

In patients with VTOS, we perform a TA-TOD combined with a percutaneous transluminal angioplasty (PTA). During the TA-TOD we take care to release the SV circumferentially from the axilla well beyond the point of the costo-clavicular ligament. Great care should be taken not to damage the vein or its collaterals, since this might cause severe and difficult to control bleeding. The subclavius muscle is completely resected (far more than in neurogenic cases) and a 360° venolysis is performed. After the TA-TOD and external venolysis, the patient is placed supine and an ultrasound guided puncture of the basilic or brachial vein of the deep venous system is performed. We do not use the cephalic vein for PTA, since the post-thrombotic anomalies in the venous system often start more lateral then the confluence of the cephalic vein into the axillary/subclavian vein.

A guidewire is passed through the SV into the Superior Caval vein and PTA with a 6mm controlled-compliant PTA balloon is performed. Depending on the reaction of the SV, an ultra-non-compliant balloon can be used to treat the stenosis/occlusion of the SV. Ultimately, we aim to perform a PTA with at least a 10mm balloon in this first session, although in patients with smaller vessels we sometimes conclude with an 8mm balloon. Perforation of the vessel due to PTA directly after thoracic outlet surgery should always be avoided since this will cause major hemorrhage. In patients with residual stenosis directly after surgery, we perform a second PTA three months later. At that time, the risk of perforation and bleeding is reduced. A Wrapsody™ stent graft (Merit Vascular, Utah, United States) is available in case of SV perforation.

Patients are admitted to the surgical ward after TA-TOD. A chest X-ray is standard of care after surgery to check for pneumothorax, position of the drains and diaphragm level. Postoperative pain management is standardized and only paracetamol, non-steroidal anti-inflammatory drugs and low dose narcotic analgesic are administered. Patients with a normal postoperative course leave the hospital the day after surgery. Patients with VTOS receive oral anticoagulants up to 3 months after surgery or 3 months after 2nd (repeated) PTA.

Every patient referred from September 2016 until December 2021 was included into a prospective database. We compiled data on clinical presentation, diagnostic work-up, treatment, type of surgical procedures performed, postoperative care, outcomes and complications. Clinical outcomes were assessed by questionnaires including the DASH (Disability of Arm, Shoulder and Hand), CBSQ (Cervico-Brachial Symptoms Questionnaire), TOS- disability scale and Derkash classification.

Demographic data are summarized in table 1. In total 1438 patients were referred to our center from 2016 until 2021. Informed consent for data analysis was obtained in 1363 patients.

A diagnosis of TOS was made in 924 (72.8%) patients: 89.1% had NTOS, 10.5% had VTOS and 0.5% had ATOS. In total, 205 (15.1%) patients were diagnosed and treated for a different diagnosis and in 234 (17.17%) patients we did not find a diagnosis.

We performed surgery on 738 (79,8%) of all patients with a diagnosis of TOS: 636 patients with NTOS, 102 with VTOS and 18 with ATOS. A trans-axillary TOD was performed in 88.8% and a supraclavicular TOD was performed in 11.1% of patients. Patients with a diagnosis of TOS that did not undergo surgery either had a low burden of disease or were at high risk not to be compliant with the intensive rehabilitation program. Pectoral minor tenectomy was performed in 385 (60.6%) of all procedures for NTOS. Cervical ribs were seen and removed in 68 (10.7%) patients with NTOS, in 2 (1.9%) patients with VTOS and in 8 (44.4%) patients with ATOS. The mean length of hospital stay was 1.33 days (SD 0.59).

NTOS patients with a TA-TOD had a statistically significant decrease of the TOS-disability scale, CBSQ and DASH scores (P<0.001)(Figure 2) compared to preoperative values. This effect was seen up to 60 months of follow-up. There is no follow-up data for the conservatively managed group. In total, 46 (7.2%) patients with TA-TOD for NTOS had no improvement (persisting or recurrent NTOS symptoms) for which a supraclavicular TOD was performed.

In the surgically treated VTOS patients, a total of 17 (16.6%) patients presented with acute upper extremity deep venous thrombosis (UEDVT). These patients received thrombolysis followed by TA-TOD and PTA. All patients had complete lysis of thrombus before TA-TOD was performed. Every patient was treated with plain old balloon angioplasty (POBA) after TA-TOD with good angiographic result. Three patients (30%) had recurrent complaints within one month after surgery. No thrombosis was seen on angiography and recurrent stenosis was treated with POBA with good results. Primary patency was 70%, primary assisted patency was 100% for a follow-up period of 40 months. There were no recurrent thromboses. In the chronic VTOS patients (excluding the UEDVT cases where we did not have pre-lysis DASH/CBSQ scores), a statistically significant decrease in the TOS-disability scale, CBSQ and DASH scores was seen over time compared to preoperative values and lasted up to 60 months of follow up.

All patients with ATOS were selected for surgery. Derkash classification score was excellent in all patients treated for ATOS. We did not have information on DASH, CBSQ or SF 12 scores. (Article submitted.)

Overall, we had 40 (6.28%) complications: 8 patients with hematoma warranting exploration, 1 patient with a pneumothorax 1-week post-procedure, 2 patients with a superficial wound infection, 4 patients with permanent long thoracic nerve palsy, 3 patients with permanent phrenic nerve palsy and 13 patients with transient phrenic nerve palsy, 4 patients with transient Horner syndrome and 4 patients with permanent Horner syndrome and 1 patient with pulmonary embolism after surgery. There were no cases of brachial plexus palsy.