SpineNovember2005 10/14/05 2:08 PM Page 355

J Neurosurg Spine 3:355–363, 2005

Surgical treatment of thoracic outlet syndrome:

a randomized trial comparing two operations

RISHI N. SHETH, M.D., AND JAMES N. CAMPBELL, M.D.

Department of Neurosurgery, School of Medicine, The Johns Hopkins University,
Baltimore, Maryland

Object. Various surgical approaches have been proposed for the treatment of thoracic outlet syndrome (TOS). The
authors of this study focused on the differences in outcome after supraclavicular neuroplasty of brachial plexus (SNBP
[no rib resection]) and transaxillary first rib resection (TFRR) in patients in whom the dominant clinical problem was
pain.
Methods. Fifty-five patients were randomized to undergo TFRR or SNBP. Patients with an anomalous cervical rib,
intrinsic weakness, and primarily vascular findings were excluded from the study. Preoperatively, the following find-
ings were typically observed: provocation of symptoms by certain postures (the so-called spear-throwing position as
well as downward tugging of the shoulder) and marked tenderness in the supraclavicular fossa.
The intergroup severity of the symptoms was comparable. Eight patients were lost to follow up. There were 24
TFRRs (in two cases the procedure was bilateral) and 25 SNBPs. The mean follow-up interval was 37 months. In both
groups pain decreased significantly after surgery. By all measures the TFRR operation conferred superior results.
Patients reported significantly less pain (39 6 7 compared with 61 6 7; score range 0–100 on a visual analog scale),
greater percentage of pain relief (52 6 8% compared with 30 6 7%), and less pain (3.7 6 0.4 compared with 5.1 6
0.5) on an affective scale (all p , 0.05) in the TFRR and SNBP groups, respectively). In the TFRR group, 75% of
patients reported good or excellent outcomes compared with 48% in the SNBP group (p , 0.05).
Conclusions. Transaxillary first rib resection provided better relief of symptoms than SNBP. The major compres-
sive element in patients with TOS-associated pain appeared to be the first rib.

KEY WORDS • thoracic outlet syndrome • brachial plexus • peripheral nerve • pain •
nerve entrapment

T HORACIC outlet syndrome refers to disorders involv-

ing the artery, vein, and/or nerves at the thoracic in-
let.5,19,30,33,39 Vascular obstruction is relatively easily
diagnosed based on clinical and radiological studies. Ede-
ma, swelling, and cyanosis of the arm may result from
compression of the subclavian vein.14,28,30,42 Arm claudica-
tion, and fatigue associated with arm abduction, is a fea-
ture of arterial obstruction.11,23,28,36 Patients may also pre-
sent with clear neurological deficits secondary to wasting
of the muscles innervated by the lower plexus.4,12,13,40,41,43 In
such cases, the need for surgical intervention is clear.
There is a fourth group of patients whose presentation is
characterized by pain but no clear neurological deficits or
clear vascular symptoms. These patients typically com-
plain that their symptoms are aggravated by external rota-
tion/elevation of the shoulder (the so-called spear-throwing
position). These symptoms include pain in the shoulder re-
Abbreviations used in this paper: SNBP = supraclavicular neu-
roplasty of the brachial plexus; TFRR = transaxillary first rib resec-
tion; TOS = thoracic outlet syndrome; VAS = visual analog scale.
gion and numbness/tingling that both extend to the hand.
The diagnostic criteria are not well established, and wheth-
er patients with these signs and symptoms should undergo
surgery is disputed.5,47 In addition no clear radiological
abnormalities have been validated.3,27 The diagnosis is
based on history, physical examination, and exclusion of
other diagnoses such as cervical spondylosis. Physical ex-
amination-detected abnormalities include provocation of
symptoms by the spear-throwing position and by downward
tugging of the arm.18,48 Patients typically feel marked ten-
derness in the supraclavicular region and a Tinel sign (pares-
thesias in the hand and fingers with digital percussion in the
supraclavicular fossa) may be present. Spondylotic disease
has been deemed unlikely to be the cause of such symp-
toms, particularly when neck range of motion is normal.
In addition to the question of whether to operate, the
choice of surgical approach is disputed as well.1,7–9,16 Some
authors have advocated SNBP, a procedure in which the bra-
chial plexus is dissected free of constricting bands and mus-
cle,20,21,24,28,32,34,38 whereas others have advocated TFRR.6,15,22,
25,35,44
The main goal of the present study was to determine
which of these operations is more effective, considering

J. Neurosurg: Spine / Volume 3 / November, 2005 355

SpineNovember2005 10/14/05 2:08 PM Page 356
both risk and symptomatic relief. Comparison of the out-
comes will provide an opportunity to determine whether
surgery has any role in the treatment of this condition. If
surgery has no role and if prior reports of improvements
are attributable to the disease’s natural history, then the
outcomes of the two operations should be similar.
Clinical Material and Methods
General Procedures
Approval for the study was obtained through the insti-
tutional review board. Patients were entered into the study
based on inclusion/exclusion criteria described in Patient
Selection. Patients were randomly assigned to undergo
SNBP or TFRR. Outcomes were derived from an evalua-
tion of questionnaire and interview results.
Patient Selection
Patients selected in our study were adults with pain as
the predominant presenting symptom and in whom TOS
had been diagnosed by the senior author (J.N.C.). Patients
with neurological deficits, as evidenced by muscle atro-
phy or sensory loss (at rest), and those who presented with
symptoms of vascular occlusion were excluded from the
study. Other exclusion criteria included prior TOS sur-
gery, findings of cervical spondylosis, and cervical rib-in-
duced plexus compression. Typically, there was a triad of
findings: symptoms (increased shoulder pain and tingling/
numbness extending into the hand) provoked by the spear-
throwing position and downward tugging of the shoulder,
as well as tenderness in the supraclavicular fossa. In all
cases, neurological examination ruled out the possibility
of a neurological deficit. Because prior experience indi-
cated that electromyography and nerve conduction studies
typically revealed no abnormal findings in this group of
patients, these studies were not typically performed. Like-
wise imaging studies, when conducted, of the brachial
plexus did not reveal any abnormalities. All patients had
previously been referred for physical therapy that involved
stretching and shoulder girdle strengthening, but no im-
provement had been shown.2,18,26
Patient Randomization
Once patients were selected as surgical candidates, the
operative choices were explained to them. They were told
that they could select one or the other procedure or that the
operation would be randomly selected for them; if patients
selected one procedure, they were excluded from the study.
Patients were told that the comparative risks and benefits of
the two operations were unknown. All patients in this study
agreed to the randomization procedure and none was ex-
cluded. Once randomized, no patient dropped out of the
study to undergo the alternative operation.
Collection of Outcome Data
A standard questionnaire was mailed to the patients. All
patients were then interviewed over the telephone or in per-
son by an individual (R.N.S.) not involved in patient care.
The outcome measures for pain were percentages of
pain relief, pain level status represented by a score on a
VAS (100-mm line), and categorical assessment of pain
356
R. N. Sheth and J. N. Campbell
relief (excellent, good, fair, or poor). Patients were asked
to indicate their average pain level as well as the pain level
at its worst and best. Patients also identified where they
felt pain, numbness, and tingling. Patients were also asked
if their symptoms were better or same during the arm raise
maneuver after surgery.
Surgical Approach
Supraclavicular Neuroplasty of the Brachial Plexus. Sur-
gery is performed after induction of general endotracheal
anesthesia with the patient in the supine position. A small
bump (2-cm-thick sheet) is placed under the shoulder, and
the head is turned in the opposite direction. A 7-cm trans-
verse incision is made one finger’s breadth above the clav-
icle, starting medially at about the middle of the belly of
the sternocleidomastoid muscle. Care is taken to preserve
the supraclavicular cutaneous nerves in the lateral part of
the incision. After mobilization of the scalene fat pad and
omohyoid muscle, the phrenic nerve is isolated along the
anterior scalene muscle. The anterior scalene muscle is
then divided completely and the upper, middle, and lower
trunks of the plexus are freed circumferentially (complete
neuroplasty). The neuroplasty is then advanced to the lev-
el of the C-8 and T-1 nerve roots (Fig. 1). Vessel loops are
placed around the nerve roots, and special care is taken to
remove any bands or other soft tissue that might be com-
pressing the roots or lower trunk. The first rib is left intact.
The middle scalene muscle is taken down to make more
room for the lower trunk. Connective tissue arising from
the apex of the pleura is removed from the area around the
lower plexus, and on occasion the pleura is opened. No at-
tempt is made to close the pleura if this happens, and no
patient in this series received a chest tube. A small drain
is placed in the wound prior to closure.
Transaxillary First Rib Resection. Surgery is performed
after induction of general endotracheal anesthesia with the
patient in the supine position. A bump is placed under the
shoulder, tilting the plane of the patient upward by ap-
proximately 20 to 30˚. The table is elevated sufficiently to
bring the operative field to a point just below the shoulder
level of the surgeon. The arm is elevated and secured to an
anesthesia screen (Fig. 2 left). The surgeon is positioned to
allow complete access to the axilla. The scrub nurse is
positioned on the opposite side and is in a position to hold
a retractor when necessary. A 6- to 7-cm curvilinear inci-
sion is made at the base of the anterior part of the axilla,
at a point just inferior to where the skin breaks away from
the chest wall. This anterior incision is designed to avoid
injury to the long thoracic nerve. Dissection proceeds
directly to the chest wall and then up to the area of the first
rib. The intercostobrachial nerve can be visualized or at
least palpated as it exits the second intercostal space and is
carefully preserved. If it interferes with the operation, the
nerve is mobilized until it can be retracted safely. Some-
times it is easier to retract the intercostobrachial nerve up-
ward, whereas at other times it is better to allow the nerve
to fall away below the retractor. The lower edge of the first
rib is cleared of muscle, and the pleura is mobilized away
from the rib. A tear in the pleura is of no consequence as
long as a drain is used postoperatively to prevent a hemo-
thorax. The anterior scalene muscle is severed at its inser-
J. Neurosurg: Spine / Volume 3 / November, 2005
SpineNovember2005 10/14/05 2:08 PM Page 357
Two surgical approaches in treating TOS
FIG. 1. Operative approach used in the SNBP procedure. The inset shows the location of the skin incision, and the
main figure shows the anatomy of the important structures. Dissection proceeds to the C-8 and T-1 nerve roots and the
lower trunk. A = artery; m = muscle; n = nerve; v = vein.
tion point between the subclavian vein and subclavian
artery, and the middle scalene is cleared away from the rib
posteriorly. The rib is removed in a piecemeal fashion
with care to prevent traction injury to the plexus (Fig. 2
right). To decompress the plexus, it is necessary to resect
the rib so that it is a short distance (, 1.5 cm) from the
transverse process. Anteriorly the rib is resected antero-
medial to the subclavian vein. The lower trunk of the
plexus, subclavian artery, and subclavian vein are under
direct view during the resection, and special care is taken
to ensure that the posterior extent of the resection extends
beyond the crossing point of the brachial plexus.
Statistical Analysis
Except where indicated, intergroup differences and pre-
and postoperative differences were analyzed using t-tests.
Confidence limits are presented as the mean 6 standard
error of the mean unless otherwise indicated.
Results
A total of 117 patients with TOS were evaluated and
were considered potential candidates for the study. Of
these, 62 patients were excluded according to our selec-
J. Neurosurg: Spine / Volume 3 / November, 2005
tion criteria. After the initial screening, 55 patients were
randomized to undergo one of the two surgeries. Eight pa-
tients were lost to follow up (four in each group). Forty-
seven completed a follow-up questionnaire and responded
to questions in a follow-up interview.
Of the 47 patients, 40 (85%) were women. The mean
age of the population at the time of surgery was 37 6 7
years (6 standard deviation; range 18–58 years). Exclud-
ing patients lost to follow up, there were 24 TFRR and 25
SNBP procedures. Two patients underwent TFRR involv-
ing both limbs. Twenty-six procedures were left sided.
The mean duration of preoperative symptoms was 47 6
5 months (range 3–150 months). The mean follow-up du-
ration was 37 6 5 months (6 standard error of the mean;
range 3–138 months). The preoperative pain severity was
rated high in most patients. More than 65% of patients
described their pain as distressing or intolerable (VAS
score . 75).
Symptoms developed in 38 patients (81%) after a trau-
matic event (16 were involved in a motor vehicle accident;
19 in a work-related incident; three in a nonwork-related
event); in the other 19% the origin of symptoms was spon-
taneous. The mean duration between the onset of symp-
toms and surgery was greater in the nine patients with
357
358
SpineNovember2005
10/14/05
2:08 PM
Page 358

FIG. 2. Left: The operative setup used in the TFRR procedure. The incision is made just at the anterior lower margin of the axillary hairline. The arm is suspend-
ed on an L-shaped apparatus. A roll is placed under the scapula, and the table is positioned so that the operative site is just below the surgeon’s shoulder level. The
scrub nurse is positioned on the opposite side to help with retraction. Right: The diagram shows the surgeon’s view for the transaxillary approach. The subclavian
vein, anterior scalene muscle, subclavian artery, and brachial plexus all appear draped over the rib in an anteroposterior position. The rib is resected in a piecemeal
fashion by using a bone rongeur, after the muscle and pleura have been cleared away (see inset). The intercostobrachial nerve is noted and preserved. The posterior
extent of the rib resection is within 1.5 cm of the transverse process so that the plexus can be fully decompressed. The piecemeal resection of the rib helps minimize
the need for retraction and allows the operation to be conducted safely via a small incision.

J. Neurosurg: Spine / Volume 3 / November, 2005

R. N. Sheth and J. N. Campbell
SpineNovember2005 10/14/05 2:08 PM Page 359

Two surgical approaches in treating TOS

TABLE 1
Summary of demographic data obtained in patients undergoing
different surgeries for TOS-related symptoms
Treament Group

Factor TFRR SNBP

male/female ratio 4:20 4:21

postop period (mos) 42 6 7 31 6 7
age (yrs) at op 35 6 2 38 6 2
VAS pain severity 77 6 3 82 6 3
increased symptoms w/ arm abducted (%) 92 92
onset attributable to trauma (%) 71 71

spontaneous-onset symptoms (60 6 18 months) compared

with the 16 patients involved in a motor vehicle accident
(33 6 4.7 months; p = 0.03) but not in the 19 patients sus-
taining a work-related injury (47.7 6 8.8 months; p = 0.25).
Comparison of Preoperative Data
Characteristics in patients who underwent TFRR were
similar to those who underwent SNBP with regard to male/
female ratio, age at onset, follow-up duration, pain severity,
and incidence of inciting trauma (Table 1). The way in
which pain was distributed among different areas was also
similar between the groups, and the majority of patients re-
ported neck, shoulder, arm, forearm, and hand/finger pain
(Fig. 3A). Complaints of numbness and tingling were also
common and these symptoms were felt in the hand and fin-
gers in more than 80% of the patients (Fig. 3B).
Outcome After Surgery
Pain. Four measurements were assessed to determine
the change in pain after the two operations: change in VAS
score, percentage of pain relief, change in the affective rat-
ings of pain, and categorical assessment of pain relief. An-
alysis of these findings demonstrated that TFRR was asso-
ciated with greater benefits than SNBP. As a measure of
internal validity of test questions, we compared postoper-
ative VASs and affective scale scores (Fig. 4). These
results were highly correlated (R2 = 0.83).
Patients used the VAS to rate their average pain, least
amount of pain, and greatest amount of pain. The mean in-
tergroup preoperative VAS score (mean pain level) was
similar (TFRR 77 6 3 and SNBP 82 6 3; p = 0.28). The
mean rating of average pain after surgery was lower for
the TFRR-treated patients (39 6 7) compared with the
SNBP-treated patients (61 6 7; p = 0.03) (Fig. 5). Simi-
larly, the mean VAS score for the “least pain experienced”
postoperatively was significantly lower in patients who un-
derwent TFRR (30 6 6) than in those treated with SNBP
(49 6 7; p = 0.04). Even though the “greatest amount of
pain” felt during the day in the postoperative period was
lower in the TFRR-treated group (47 6 8), this difference
was not significantly different compared with the SNBP-
treated group (67 6 8; p = 0.09).
Patients were asked to indicate the percentage of pain
relief, and this value was significantly higher in patients
treated with TFRR (52 6 8%) than in those who under-
went SNBP (30 6 8%; p , 0.005).
Patients were asked to rate their pre- and postoperative
pain by using a nine-point affective scale. The pain-related
FIG. 3. Bar graphs demonstrating the percentage of patients who
described pain (upper) or numbness/tingling (lower) in different ar-
eas of the body preoperatively. The intergroup distribution of symp-
toms was similar.
labels on this scale were as follows: none, not bothersome,
slightly annoying, annoying, very annoying, distressing,
very distressing, intolerable, and very intolerable. For pur-
poses of statistical analysis numbers were assigned to each
category from 0 (indicating no pain) to 8 (representing very
intolerable pain). Changes in these affective ratings of pain
paralleled changes in VAS scores. Before the surgery
patients rated their pain as very distressing to intolerable
(6.5 6 0.2). After SNBP there was a significant improve-
ment from the baseline level (distressing, 5.1 6 0.5; p =
0.005 [Fig. 6A]). The TFRR-treated patients also improved
(very annoying, 3.7 6 0.4; p , 0.0001 [Fig. 6B]), and this
improvement was significantly better than that following
SNBP (p = 0.03).
The patients who underwent TFRR were significantly
more likely to report good or excellent pain relief (75%
compared with 48%, respectively; p = 0.02).
Numbness. Preoperatively the overall number of areas
in which numbness was felt were comparable between
groups (TFRR, 5.2 6 0.61 and SNBP, 5.5 6 0.5; p = 0.77
[Fig. 3B]). The number of numb areas was significantly
reduced after both surgeries (TFRR, 2.8 6 0.6; p , 0.001
and SNBP, 4.6 6 0.6; p , 0.05) but the difference be-
tween the two operations was significant (p = 0.03).
Seventy-two percent of the patients whose preoperative
symptoms were aggravated by arm raising felt improve-

J. Neurosurg: Spine / Volume 3 / November, 2005 359

SpineNovember2005 10/14/05 2:08 PM Page 360
FIG. 4. Comparison of postoperative pain ratings based on the
VAS compared with the affective scale for each patient (R2 = 0.83)
(affective scale: 0, no pain; 1, not bothersome; 2, slightly annoying;
3, annoying; 4, very annoying; 5, distressing; 6, very distressing; 7,
intolerable; and 8, very intolerable).
ment after TFRR, whereas only 48% of the patients with
this symptomatic trigger felt an improvement after SNBP.
When asked if they would undergo this surgery again for
the same result, 19 (79%) of the 24 TFRR-treated patients
reported that they would, whereas 14 (56%) of the 25
SNBP-treated patients said they would.
Patients who fared well after TFRR in general were
working preoperatively. Patients who did not work before
surgery were unlikely to return to work. Patients not work-
ing because of pain symptoms preoperatively, therefore,
are at significant risk for not faring well after surgery.
No patients in this series suffered from a brachial plexus
injury, pneumothorax, infection, hematoma, or other seri-
ous complications as a result of surgery.
Discussion
Thoracic outlet syndrome represents a group of im-
pingement disorders, involving the subclavian vein, the
subclavian artery, and/or the brachial plexus. Different
surgical approaches have been advocated for its treatment,
including posterior thoracotomy for removal of the first
rib, TFRR, and SNBP both with and without first rib re-
section.6,10,15,17,21,22,24,25,28,31,35,44 We report here the first ran-
domized study to compare two surgical approaches,
TFRR and SNBP without rib resection.
Patient Selection
Our patients presented with pain and neurological symp-
toms, but without bona fide neurological deficits (clear
weakness, muscle atrophy, or ongoing sensory deficit). We
chose these patients because they represented the most
common group with TOS that presented to the senior au-
thor (J.N.C.). We excluded other patients to ensure a popu-
lation of those included in the series. These patients clearly
had a neurological disorder as nearly all complained of
prominent sensory symptoms (described as pins and nee-
dles), and numbness, as well as pain that extended from the
shoulder to the hand. The sensory distribution varied but
360
R. N. Sheth and J. N. Campbell
FIG. 5. Bar graph showing pre- and postoperative VAS pain
scores (*p = 0.01 [SNBP group]; **p , 0.00001 [TFRR group]).
The score decreased significantly for both groups (1p = 0.03).
tended to be more prominent on the ulnar aspect of the
hand. No patient had symptoms restricted to the thumb and
index finger.
Most patients had distinctive tenderness in the supra-
clavicular fossa, increased pain/hand numbness with arm
abduction and external rotation of the shoulder (spear-
throwing position), and increased pain/hand numbness with
downward movement of the shoulder.
Transaxillary Resection, the Optimal Operation
By all measures the TFRR operation proved the better
procedure for relief of symptoms. The outcomes were
measured after a substantial postoperative duration (mean
36 months). Thus, these results represent long-term out-
comes.
Although the literature is replete with series involving
the surgical treatment of TOS, one may argue that im-
provements seen reflect biased reporting, placebo effects,
and/or those derived due to passage of time independent of
surgical effects. That one operation in this study yielded
decidedly better outcomes than the other provides evidence
that surgery may confer benefits over natural history.
Our patients reported severe symptoms that had been
present typically for many years. Nearly all patients had
undergone extended trials of physical therapy that did not
prove successful.29 Arguments that these symptoms
should be treated conservatively must be considered in
terms of the substantial disability that these patients have
sustained. Clearly the decision to recommend surgery
should be influenced by a risk–benefit analysis. Substan-
tial complication rates have been reported for TOS.7,8,10,37
In this series no patient suffered any morbidity, but the se-
nior author (J.N.C.) has observed transient brachial plex-
us deficits in other surgically treated patients. Thus, pa-
tients who undergo surgery must be mindful that even in
experienced hands risks are not insubstantial.
Some authors have advocated that none of these pa-
tients should be treated surgically.9,45,46 Because the sur-
gery-associated morbidity rate is low and because approx-
J. Neurosurg: Spine / Volume 3 / November, 2005
SpineNovember2005 10/14/05 2:08 PM Page 361
Two surgical approaches in treating TOS
FIG. 6. Bar graphs depicting pre- and postoperative pain deter-
mined using a category scale (SNBP, p = 0.005; TFRR, p =
0.0001). Upper: Preoperatively, almost all patients complained
of their pain as distressing to intolerable. Lower: Postoperatively,
however, the number of cases in which patients complained of dis-
tressing and intolerable symptoms was 33% in the TFRR group
and 68% in the SNBP group. Patients fared better after the TFRR
in their perception of pain (p = 0.03). For presentation purposes the
original scale is reduced from nine to four categories by grouping
the descriptors in groups of two (See Fig. 4; the “no pain” desig-
nation actually included three categories: no pain, not bothersome,
and slightly annoying).
imately 75% of the TFRR-treated patients reported good
or excellent results, surgery appears a justifiable option.
This analysis is predicated on the estimation of the mor-
bidity rate being low. The TFRR procedure entails risks
likely related to one’s surgical experience. Reports of sub-
stantial morbidity have been advanced as reasons to be
cautious when advocating TOS surgery.7,8,10,37
The results of this study implicate the first rib as the ma-
jor factor in causing TOS. The implication is that the rib
occupies a high position such that it pushes up on the lower
aspect of the plexus. At the time of surgery one observes
the lower trunk draped over the rib. Whether that consti-
tutes an impingement, however, is a subjective judgment.
High-resolution peripheral nerve magnetic resonance
imaging has been indicated as a means of identifying com-
pression. Continued experience with magnetic resonance
J. Neurosurg: Spine / Volume 3 / November, 2005
imaging will have to be obtained to determine how sensi-
tive it is. We have seen several cases in which the lower
trunk appears to be deviated superiorly in the region of the
first rib. A caveat is that many patients only experience
neurological symptoms when their arm is in certain po-
sitions (for example, the spear-throwing position). Ideal-
ly neuroimaging should be performed with the patient’s
arm in the symptom-provoking position, but physical con-
straints may make this impractical.
Of course in conducting a first rib resection other struc-
tures are involved. These include the anterior and middle
scalene muscle and other fibrous bands that may attach to
the first rib. Because the supraclavicular approach involved
manipulation of these structures, these putative compres-
sion points likely do not account for the major compression
in most patients. It has yet to be determined in a control trial
whether the addition of neuroplasty would augment the im-
provements derived from TRFF. Indeed, many patients in
this series did appear to benefit from the supraclavicular
approach (48% of patients reported a good or excellent
outcome). There may well be a subclass of patients in
which the supraclavicular approach yields an improved
outcome. Indeed it has been policy since this study was
completed to offer a supraclavicular approach to patients
in whom TFRR produced an unsatisfactory outcome. Our
impression anecdotally is that offering this surgical option
is helpful, at least in some cases.
Another type of surgery that has been advocated is the
first rib resection via the supraclavicular approach.15,24,37
This surgery offers the advantages of the transaxillary
approach (that is, rib resection) with the additional possi-
bility of retracting soft-tissue structures compressing the
plexus. The reason for not undertaking this approach is
based on the senior author’s impression that it puts the
plexus at risk of injury due to retraction. This issue is
deserving of further consideration, however.
McCarthy, et al.,24 found that supraclavicular decom-
pression for TOS led to benefit in 76% of their patients. In
their study, however, approximately one third of the pa-
tients harbored a cervical rib (compared with none in the
present series). The median follow-up duration was 12
months (whereas it was . 3 years in the present study),
and outcome was determined by the surgeon’s review of
patient charts (compared with independent systematic re-
view in the present study). Cheng and associates7 com-
pared supraclavicular rib resection and supraclavicular
neurolysis and found similar short- and long-term results.
They advocated rib-sparing surgery based on its associat-
ed lower morbidity rate.
Wenz and Husfeldt44 reported on 80 patients who under-
went TFRR and were followed for 6.6 years. They ob-
served complete resolution or improvement of symptoms
in 85% of their cases. Pain relief was achieved in 75% of
our TFRR-treated patients. Wenz and Husfeldt44 had treat-
ed all the forms of TOS, and in some of their patients cer-
vical ribs were present. In a retrospective study Cuypers
and coworkers9 evaluated the findings in 92 patients who
underwent TFRR and found that the surgery was success-
ful in only 52%; 37% of the patients had arterial and ve-
nous compression. Lacking a randomized series, the rele-
vance of these data is unclear, in particular because of the
more heterogeneous patient population.
361
SpineNovember2005 10/14/05 2:08 PM Page 362
Analysis of Treatment Failures
In many patients surgery failed to yield a benefit and in
the majority only partial symptomatic relief was achieved.
Several possibilities could account for this. Some patients
in this study likely had some other occult cause of their
symptoms not addressed by surgery. A second possibility
is that the surgery was inadequate. In the SNBP operation,
however, we undertook a thorough circumferential neuro-
plasty of each brachial plexus element and retraction of
the anterior and middle scalene muscles, as well as other
bands, including the fibrous tissue arising from the apical
pleura and, in some cases, arteries and veins that might
conceivably obstruct the plexus. The failure to resolve all
pain symptoms in the TFRR-treated patients could relate
to an inadequate rib resection.25 Our goal was to resect the
rib to within 1.5 cm of the transverse process, and this was
uniformly achieved in all cases. We favor the notion that
relief of symptoms is incomplete because compression
alone is not necessarily the sole problem in TOS. The
structures removed during TOS surgery have a suspen-
sion-related function. The brachial plexus may be tethered
by the weight of the arm tugging downward on the plexus.
Indeed in some cases symptoms were palliated when the
patients wore a sling that helped to hold up the arm, pre-
venting the weight of the arm from pulling down on the
plexus. This tethering effect on the plexus will not be
helped by further efforts at decompression. Indeed further
removal of supporting structures could, in fact, make the
tethering worse.
Conclusions
This is the first randomized trial of two different opera-
tions in the treatment of TOS. The TFRR provided better
relief of symptoms, in particular pain, than the SNBP with-
out rib resection. It is likely that the first rib is the major
cause of compression in the majority of patients.
Acknowledgments
Drs. Lee Dellon, Lawrence Empting, and Neal Naff all played
important roles in helping with this study at various stages.
References
1. Adson AW: surgical treatment for symptoms produced by cer-
vical ribs and the scalenus anticus muscle. Surg Gynecol Ob-
stet 85:685–700, 1947
2. Aligne C, Barral X: Rehabilitation of patients with thoracic out-
let syndrome. Ann Vasc Surg 6:381–389, 1992
3. Bilbey JH, Muller NL, Connell DG, Luoma AA, Nelems B:
Thoracic outlet syndrome: evaluation with CT. Radiology 171:
381–384, 1989
4. Campbell JN: Thoracic outlet syndrome, in Frymoyer JW, Duc-
ker TB, Hadler NM, et al (eds): The Adult Spine: Principles
and Practice, ed 2. Philadelphia, Lippincott-Raven Publishers,
1997, pp 1349–1355
5. Campbell JN, Naff NJ, Dellon AL: Thoracic outlet syndrome.
Neurosurgical perspective. Neurosurg Clin North Am 2:
227–233, 1991
6. Carty NJ, Carpenter R, Webster JH: Continuing experience
with transaxillary excision of the first rib for thoracic outlet syn-
drome. Br J Surg 79:761–762, 1992
7. Cheng SW, Reilly LM, Nelken NA, Ellis WV, Stoney RJ: Neu-
362
R. N. Sheth and J. N. Campbell
rogenic thoracic outlet decompression: rationale for sparing the
first rib. Cardiovasc Surg 3:617–624, 1995
8. Cikrit DF, Haefner R, Nichols WK, Silver D: Transaxillary or
supraclavicular decompression for the thoracic outlet syn-
drome. A comparison of the risks and benefits. Am Surg 55:
347–352, 1989
9. Cuypers PW, Bollen EC, van Houtte HP: Transaxillary first rib
resection for thoracic outlet syndrome. Acta Chir Belg 95:
119–122, 1995
10. Dubuisson AS, Kline DG, Weinshel SS: Posterior subscapular
approach to the brachial plexus. Report of 102 patients. J Neu-
rosurg 79:319–330, 1993
11. Gergoudis R, Barnes RW: Thoracic outlet arterial compression:
prevalence in normal persons. Angiology 31:538–541, 1980
12. Gilliatt RW: The classical neurological syndrome associated
with a cervical rib and band, in Greep JM, Lemmens HA, Roos
DB, et al (eds): Pain in Shoulder and Arm. The Hague: Mar-
tinus Nijhoff, 1979, pp 173–183
13. Gilliatt RW, Le Quesne PW, Logue V, Sumner AJ: Wasting of
the hand associated with a cervical rib or band. J Neurol Neu-
rosurg Psychiatry 33:615–626, 1970
14. Glass BA: The relationship of axillary vein thrombosis to the
thoracic outlet compression syndrome. Ann Thoracic Surg 19:
613–621, 1975
15. Hempel GK, Shutze WP, Anderson JF, Bukhari HI: 770 consec-
utive supraclavicular first rib resections for thoracic outlet syn-
drome. Ann Vasc Surg 10:456–463, 1996
16. Kashyap VS, Ahn SS, Machleder HI: Thoracic outlet neurovas-
cular compression: approaches to anatomic decompression and
their limitations. Semin Vasc Surg 11:116–122, 1998
17. Kim DH, Cho YJ, Tiel RL, Kline DG: Outcomes of surgery in
1019 brachial plexus lesions treated at Louisiana State University
Health Sciences Center. J Neurosurg 98:1005–1016, 2003
18. Lindgren KA: Conservative treatment of thoracic outlet syn-
drome: a 2-year follow-up. Arch Phys Med Rehabil 78:
373–378, 1997
19. Lord JW JR, Rosati LM: Thoracic outlet syndromes. Clin
Symp 23:3–32, 1971
20. Luoma A, Nelems B: Thoracic outlet syndrome. Thoracic sur-
gery perspective. Neurosurg Clin North Am 2:187–226, 1991
21. Makhoul RG, Machleder HI: Developmental anomalies at the
thoracic outlet: an analysis of 200 consecutive cases. J Vasc
Surg 16:534–545, 1992
22. Martin GT: First rib resection for the thoracic outlet syndrome.
Br J Neurosurg 7:35–38, 1993
23. Martin J, Gaspard DJ, Johnston PW, Kohl RD Jr, Dietrick W:
Vascular manifestations of the thoracic outlet syndrome. A sur-
gical urgency. Arch Surg 111:779–782, 1976
24. McCarthy MJ, Varty K, London NJ, Bell PR: Experience of su-
praclavicular exploration and decompression for treatment of
thoracic outlet syndrome. Ann Vasc Surg 13:268–274, 1999
25. Mingoli A, Feldhaus RJ, Farina C, Cavallari N, Sapienza P,
diMarzo L, et al: Long-term outcome after transaxillary approach
for thoracic outlet syndrome. Surgery 118:840–844, 1995
26. Novak CB, Collins ED, Mackinnon SE: Outcome following
conservative management of thoracic outlet syndrome. J Hand
Surg Am 20:542–548, 1995
27. Panegyres PK, Moore N, Gibson R, Rushworth G, Donaghy M:
Thoracic outlet syndromes and magnetic resonance imaging.
Brain 116:823–841, 1993
28. Pang D, Wessel HB: Thoracic outlet syndrome. Neurosurgery
22:105–121, 1988
29. Peet RM, Henriksen JD, Anderson TP, Martin GM: Thoracic-
outlet syndrome: evaluation of a therapeutic exercise program.
Mayo Clin Proc 31:281–287, 1956
30. Pollak EW (ed): Clinical presentation, in Thoracic Outlet Syn-
drome: Diagnosis and Treatment. New York: Futura, 1986,
pp 37–70
31. Pollak EW: Surgical anatomy, in Thoracic Outlet Syndrome:
J. Neurosurg: Spine / Volume 3 / November, 2005
SpineNovember2005 10/14/05 2:08 PM Page 363
Two surgical approaches in treating TOS
Diagnosis and Treatment. New York: Futura, 1986, pp 3–36
32. Roos DB: Congenital anomalies associated wit thoracic outlet
syndrome. Anatomy, symptoms, diagnosis and treatment. Am
J Surg 132:771–778, 1976
33. Roos DB: Edgar J. Poth Lecture. Thoracic outlet syndromes:
update 1987. Am J Surg 154:568–573, 1987
34. Roos DB: New concepts of thoracic outlet syndrome that ex-
plain etiology, symptoms, diagnosis and treatment. Vasc Surg
13:313–321, 1979
35. Roos DB: Transaxillary approach for first rib resection to re-
lieve thoracic outlet syndrome. Ann Surg 163:354–358, 1966
36. Ross JP: The vascular complications of cervical rib. Ann Surg
150:340–345, 1959
37. Sanders RJ, Monsour JW, Gerber WF, Adams WR, Thompson
N: Scalenectomy versus first rib resection for treatment of tho-
racic outlet syndrome. Surgery 85:109–121, 1979
38. Sanders RJ, Pearce WH: The treatment of thoracic outlet syn-
drome: a comparison of different operations. J Vasc Surg 10:
626–634, 1989
39. Sheth RN, Belzberg AJ: Diagnosis and treatment of thoracic
outlet syndrome. Neurosurg Clin N Am 12:295–309, 2001
40. Sunderland S (ed): Disturbances of brachial plexus origin asso-
ciated with unusual anatomical arrangements in the cervico-
brachial region: the thoracic outlet syndrome, in Nerves and
Nerve Injuries. New York: Churchill Livingstone, 1978, pp
901–919
41. Tender GC, Thomas AJ, Thomas N, Kline DG: Gilliatt–Sum-
ner hand revisited: a 25-year experience. Neurosurgery 55:
883–890, 2004
42. Urschel HC Jr, Razzuk MA: Improved management of the
J. Neurosurg: Spine / Volume 3 / November, 2005
Paget–Schroetter syndrome secondary to thoracic outlet com-
pression. Ann Thorac Surg 52:1217–1221, 1991
43. Urschel HC Jr, Razzuk MA: Management of the thoracic outlet
syndrome. N Engl J Med 286:1140–1143, 1972
44. Wenz W, Husfeldt KJ: [Thoracic outlet syndrome—an interdis-
ciplinary topic. Experience with diagnosis and therapy in a 15-
year patient cohort (80 trans-axillary resections of the 1st rib in
67 patients) and a literature review.] Z Orthop Ihre Grenzgeb
135:84–90, 1997 (Ger)
45. Wilbourn AJ: Thoracic outlet syndromes: a plea for conserva-
tism. Neurosurg Clin N Am 2:235–245, 1991
46. Wilbourn AJ, Porter JM: Neurogenic thoracic outlet syndrome:
surgical versus conservative therapy. J Vasc Surg 15:880–882,
1992
47. Wilbourn AJ, Porter JM: Thoracic outlet syndromes, in Weiner
MA (ed): Spine: State of the Art Reviews. Philadelphia: Han-
ley & Belfus, 1988, pp 598–626
48. Wright IS: The neurovascular syndrome produced by hyperab-
duction of the arms. Am Heart J 29:1–19, 1945
Manuscript received January 3, 2005.
Accepted in final form August 26, 2005.
Current address for Dr. Sheth: Department of Neurological Sur-
gery, University of Miami, Lois Pope Life Center, 1095 Northwest
14th Terrace (D4-6), Miami, FL 33136.
Address reprint requests to: James N. Campbell, M.D., Depart-
ment of Neurosurgery, School of Medicine, The Johns Hopkins
University, 600 North Wolfe Street, Meyer 5-109, Baltimore, Mary-
land 21287. email: jcampbel@jhmi.edu.
363