Biomechanical and clinical studies have shown that traumatic dislocation or
subluxation of the shoulder leads to labral detachment from the glenoid, a
so-called Bankart lesion, and to elongation of the capsular ligamentous
restraints1-5.
The open Bankart technique has been established as the socalled gold standard
for the treatment of anterior
instability6,7.
Arthroscopic stabilization for this problem has evolved from the use of
metallic staples to transglenoid sutures, to absorbable rivets, and to suture
anchors8-14.
Early arthroscopic stabilization techniques had failure rates that were higher
than those of open Bankart
stabilization15-17.
Our own experience with arthroscopic stabilization began fifteen years ago and
was disappointing. The results of our study, reported in 1995, yielded a 49%
recurrence rate with use of transglenoid
sutures18. These
disappointing results led to the abandonment of arthroscopic shoulder
stabilization and to our return to open Bankart repair.
In 1999, we reevaluated arthroscopic Bankart stabilization as we believed
that techniques and implants had evolved sufficiently to allow better
fixation. We adopted the technique with use of suture anchors described by
Wolf19, as the
failure rates described with the suture-anchor technique were
acceptable20-23.
Arthroscopic stabilization with suture anchors is, however, still developing,
and, as indications and contraindications are not well defined, several
questions still need to be answered.
The purpose of this study was to report our recent experience with
arthroscopic shoulder stabilization using suture anchors and to identify,
through a retrospective case series, specific factors related to the
postoperative recurrence of shoulder instability. We hypothesized that an
arthroscopic procedure that used modern techniques would treat instability
more effectively than previous techniques reported for patients with
traumatic, recurrent anterior instability.
Inclusion and Exclusion Criteria
I n order to evaluate the value of this arthroscopic Bankart procedure, we
decided to perform this operation only in a series of consecutive patients
with traumatic, recurrent anterior instability, regardless of the
preoperatively identified lesions. The criteria for inclusion were (1) the
presence of traumatic, recurrent anteroinferior shoulder instability, (2)
labral repair and capsule retensioning with use of a single arthroscopic
technique with suture anchors, (3) surgery performed by the senior surgeon
(P.B.) or under his direction, and (4) a clinical examination and interview
with the patient performed at least two years after surgery by independent
observers.
Exclusion criteria were (1) arthroscopic stabilization for acute (first)
anterior dislocation or subluxation; (2) arthroscopic stabilization after a
previous failed instability repair; (3) patient preference for open
stabilization; (4) other types of instability such as voluntary instability,
posterior instability, and multidirectional instability (defined as
instability in three
directions)24.
Between July 1999 and August 2001, 100 consecutive patients who had
arthroscopic Bankart repairs for traumatic, recurrent anterior shoulder
instability, with use of suture anchors, met the inclusion criteria.
Study Population
Nine patients were lost before two years of follow-up, leaving a cohort of
ninety-one patients available at the time of the final review. Seventy-one
patients (78%) were male. The mean age of the patients was 21.5 ± 3.5
years (range, twelve to forty-nine years) at the time of injury and 26.4
± 5.4 (range, seventeen to sixty-two years) at the time of surgery. The
dominant side was involved in fifty-three patients (58%). Bilateral anterior
instability was present in fifteen patients (16%). The diagnosis of anterior
instability was made when there was a history of subluxation with spontaneous
reduction or a history of dislocation requiring manual reduction, and all
patients had a positive anterior apprehension and relocation
test25,26.
Twenty-two patients had recurrent dislocations, thirty-nine had recurrent
subluxations, and thirty had both subluxations and dislocations. The average
number of instability episodes varied, with seven episodes (range, two to
forty) for the patients with dislocations, twenty-three episodes (range, two
to 150) for the patients with subluxations, and twenty episodes (range, two to
103) for the patients with dislocations and subluxations. Seventy-nine
patients (87%) were involved in sports, with forty (44%) who participated in
high-risk sports with contact and/or throwing. Twenty-four patients (26%)
played at a competitive, recreational level.
Preoperative Evaluation of Shoulder Laxity
Shoulder laxity was evaluated by clinical examination. Anterior shoulder
hyperlaxity was defined as external rotation of >90° with the arms at
the side (reaching the frontal plane). This sign is usually bilateral and is
an indicator of a congenitally weak anterior
capsule27. These
patients were also found to have general ligamentous hyperlaxity. Nine
patients demonstrated anterior hyperlaxity.
Inferior shoulder laxity was defined as a difference of >20° between
sides on hyperabduction (the Gagey
test)28,29.
This sign is usually unilateral and is an indicator of a stretched inferior
capsule because of plastic deformation of the inferior glenohumeral ligament
secondary to
instability29.
Twenty-six patients demonstrated such inferior laxity. Sixteen patients were
considered to have a stretched inferior glenohumeral ligament.
Evaluation of Osseous Lesions and Bone Loss
All patients underwent a preoperative radiographic evaluation, including
anteroposterior radiographs made with the arm in three different rotations
(neutral, external, and internal), a scapular lateral radiograph, and an
axillary radiograph. Assuming that bone loss was a risk factor for the
recurrence of instability, we asked our patients to have a preoperative
computed tomographic scan, unless they already had a magnetic resonance
imaging scan. Preoperative computerized tomography scans were available for
sixty-six patients. All patients had an arthroscopic examination, with the
arthroscope placed first in the posterior portal and then in the anterior
portal. Evaluation was initially performed with use of 20 mL of air
insufflation to gauge the humeral translation, and then it was done with
normal saline solution.
The glenoid surface was considered normal in forty-six patients, while
osseous lesions were seen in forty-five patients (49%): a glenoid fracture
with a detached bone fragment (shear or avulsion-fracture) was found in
thirty-four patients (37%) and a glenoid bone defect without any detached bone
fragment (compression-fracture), in twelve patients (13%)
(Fig. 1). We did not attempt to
quantify the bone fragment or defect with computed tomography scans because a
standardized radiographic protocol was not utilized. Amputation of the
anterior glenoid surface was estimated preoperatively by dividing the glenoid
rim into six sections of approximately 30° (A, B, C, D, E, and F)
(Fig. 2). The term
"glenoid bone defect" was applied if =25% (more than one
section) of the anterior glenoid rim were missing.
On the humeral side, a Hill-Sachs lesion was present in seventy-six
patients (84%) (Figs. 3-A and
3-B). This lesion was considered to be "small" in
sixty-four patients and "large" in twelve patients (13%). No
attempt was made to measure accurately the size of the bone defect in the
humeral head with computed tomography scans because a standardized protocol
was not used. The term "humeral bone defect" was arbitrarily
applied if a clinically "important" part of the humeral head
surface was missing on assessment during the arthroscopic procedure.
Evaluation of Labral and Ligamentous Lesions
Arthroscopic examination was used to evaluate the extent of the labral
detachment around the glenoid, the degree of capsular laxity, and the quality
of the tissue by direct visualization and palpation with a
probe30. The
glenoid rim was divided into six regions (A, B, C, D, E, and F) by two
imaginary oblique lines drawn from eleven o'clock to five o'clock and from one
o'clock to seven o'clock, and an imaginary transverse line drawn at the
equator of the glenoid (Fig.
2)31.
Labral detachment from the glenoid rim (i.e., a Bankart lesion) was present
in eighty-two patients (90%). The labrum was detached, together with the
capsule, in seventy-five patients, and it was torn in a bucket-handle
configuration in seven patients. According to the extent and location of the
labral detachment, seven types of lesions were found: C (two patients), BC
(eight), BCD (thirty-nine), ABC (eleven), ABCD (nineteen), and ABCDE (three).
In the remaining nine patients (10%), no Bankart lesion was found, but the
capsule was either stretched or torn.
Surgical Technique
All surgery was performed with use of a standardized technique by the
senior surgeon (P.B.). The patient was placed in the beach-chair position, and
the affected upper extremity was prepared and draped. A standard 30°
arthroscope was utilized. The posterior portal was created 1.5 cm inferior and
medial to the posterolateral corner of the acromion. The anterosuperior portal
was created adjacent to the anterior edge of the acromion through the rotator
interval in an "outsidein" fashion. A needle was used to locate
the desired position to allow both an appropriate angle of approach to the
glenoid (not too tangential to the articular surface) and to ensure easy
access to the anteroinferior capsule and labrum. A complete evaluation of the
capsular and osseous lesions was then performed. The surgeon evaluated the
amount of capsular shift needed, depending on the quality of the tissue and
the severity of the capsular stretch.
The labral detachment was completed, or created, with a Bankart rasp and
electrocautery. The rasp was used to detach and then elevate the soft tissue
from the glenoid rim with a lever arm maneuver. The goal was to mobilize the
inferior glenohumeral ligament and labrum such that it could be shifted
superiorly and laterally. Visualization of the fibers of the subscapularis
muscle beneath the labrum and a sensation of elasticity of the inferior
glenohumeral ligament when pulled with a grasper were evidence of a
satisfactory soft-tissue release. The glenoid neck was then decorticated with
use of a shaver and burr to create a cancellous bed to aid tissue-healing.
Depending on the size of the articular surface, three or four holes were
drilled in the glenoid rim at five, four, three, and two or one o'clock. The
holes were drilled at the margin of the articular surface to allow recreation
of the glenoid concavity (Figs. 4-A and
4-B).
A Wolf hooked needle (Linvatec, Largo, Florida) was used to pass a suture
(PDS II #1; Ethicon, Johnson and Johnson, Somerville, New Jersey) in the most
inferior part of the capsule, through both the labrum and the capsule. The
suture was placed on an absorbable anchor (Panalok; Mitek, Johnson and
Johnson) and inserted in the most inferior hole, at the five o'clock position.
The same maneuver was then performed to pass other sutures, which were
inserted with an anchor at the four, three, and two o'clock positions,
respectively. The average number of anchors used was 4.3 (range, two to
seven). Nine patients (10%) had no Bankart lesion; we performed a capsular
plication in four of them, and we created a Bankart lesion that was later
repaired, in addition to performing a capsular plication, in the remaining
five patients.
Postoperative Care
The patients were managed with the arm in a sling in internal rotation for
four weeks. Passive pendulum exercises were started on the day after surgery.
Patients were advised to perform these exercises five times a day for five
minutes at a time for the first four weeks. Rehabilitation with a
physiotherapist began at thirty days. External rotation was limited to 45°
until day 45. Strengthening was started between eight and twelve weeks, with
return to sports delayed until four to six months postoperatively.
Functional Evaluation
All patients returned for a follow-up evaluation at three, six, and twelve
months and then yearly thereafter. At the time of the last review, all
patients were examined and interviewed by two independent observers (M.V. and
J.Y.H.). The mean duration of follow-up was thirty-six months (range,
twenty-four to fifty-six months). Physical examination included the shoulder
range of motion and instability signs (apprehension test and relocation test).
The Rowe and Walch-Duplay scores (see Appendix) were performed at each
review32.
Definition of Failure
The recurrence of instability was considered a failure. This means that any
postoperative dislocation or any subjective complaint of occasional to
frequent subluxation was considered a failure.
Statistical Analysis
The chi-square test, Yates-corrected chi-square test, Student t test, and
Fisher variance analysis were used. The level of significance was set at p
< 0.05. Analysis was performed with use of StatView 5.0 (SAS Institute,
Cary, North Carolina).
Stability
Atrue recurrence (dislocation or subluxation) was present in fourteen (15%)
of the ninety-one patients who were available for review. Thirteen patients
who had a recurrence of instability were male. Seven of those patients
experienced a new, important traumatic event, while the other seven patients
did not. Six patients sustained a true dislocation, and eight reported a
subluxation. The mean delay to recurrence was 17.6 months (range, seven to
thirty-two months). Nine patients had a recurrence more than one year after
surgery; five of them had a recurrence between one and two years, and the
other four had a recurrence at more than two years postoperatively. The number
of recurrences was constant throughout the study period. Nine of the fourteen
patients underwent open revision surgery with a Latarjet procedure, which
resulted in a stable shoulder in all of them. The five remaining patients
declined additional surgery.
Moreover, nine patients had a persistent apprehension sign in the throwing
position (a positive apprehension test). The apprehension was severe enough to
interfere with sports or in some activities of daily living. Seven of these
nine patients, however, were satisfied or very satisfied with the results of
surgery.
Functional Results
The mean Rowe score was 77.8 points (range, 15 to 100 points), and the mean
Walch-Duplay score was 74.8 points (range, 15 to 100 points). Detailed scores
are summarized in Table I.
Sixty-two patients (68%) had good or excellent results according to both
scores. A return to sports was evaluated in the seventy-seven patients who had
no recurrence of instability. Fifty-eight patients (75%) went back to sports
at the same level, thirteen (17%) went back to sports but at a lower level,
and six (8%) stopped participating in any sports.
Subjective Results
Fifty-three patients (58%) were very satisfied, seventeen (19%) were
satisfied, eleven (12%) were disappointed, and ten (11%) were
dissatisfied.
Risk Factors Associated with Recurrence of Instability
We compared the fourteen patients who had a recurrence of instability with
the seventy-seven patients considered to be successes in order to determine
potential factors related to failure. We found several factors that were
associated with recurrent instability. First, glenoid bone loss (erosion or a
compression fracture involving >25% of the glenoid surface), without any
bone fragment detached, was found to be significantly associated with failure
(p = 0.01). Interestingly, with the numbers available, a glenoid fracture
(with detachment of a bone fragment or an avulsion fracture) was not
associated with a higher failure rate (p = 0.59). Other factors significantly
related to failure included humeral bone loss categorized as a large
Hill-Sachs lesion (p = 0.05) and a stretched inferior glenohumeral ligament (p
= 0.03) or anterior hyperlaxity (p = 0.01). Last, the number of sutures and
anchors was significantly related to failed arthroscopic stabilization;
specifically, patients who had three anchors or fewer had higher rates of
recurrent instability (p = 0.03).
On multivariate analysis, we found that the presence of a stretched
inferior glenohumeral ligament, anterior hyperlaxity, or a glenoid compression
fracture involving >25% of the glenoid surface led to a 75% recurrence rate
(p < 0.001).
With the numbers studied, other potential prognostic factors frequently
described in the literature were not found to be significant in the present
series. Specifically, the age of the patient at the time of the first episode,
gender, dominant arm, type of sport (contact or no contact), type of
instability (dislocation or subluxation), number of episodes, bilateral
instability, presence of a superior labrum anteroposterior lesion, or absence
of a Bankart lesion did not influence the recurrence
rate33-38.
Complications
No complications were related to the anchors or sutures. One patient had an
acute infection that resolved with arthroscopic débridement and
antibiotics; the shoulder was stable at the time of the last review. Three
patients were considered to have a stiff shoulder, lacking >20° in
passive elevation and external rotation. One patient had persistent pain, but
no explanation was found.
In the present study, we found that arthroscopic shoulder stabilization
with use of suture anchors led to a postoperative recurrence of instability in
15% of the patients at a mean follow-up of thirty-six months. Although the
recurrence rate may increase with time, these results are similar to those in
the few previous reports on arthroscopic stabilization with use of suture
anchors20,21,23,38-40.
In comparable series, Kandziora et
al.39 and Kim et
al.40 reported
failure rates of 16.5% and 10%, respectively. The failure rate in our study
compares favorably with those of Koss et
al.21 and
Lafosse41 who
reported rates of 30% and 18.5%, respectively. A lower recurrence rate of 8%
was reported by both Gartsman et
al.20 and
Tauro23 who added
shrinkage or inferior capsular tensioning to the capsulolabral repair. In our
patients, we did not use thermal shrinkage.
We found, as others have, that the number of fixation points is important
for the success of anterior arthroscopic
stabilization40. In
the present study, patients who had three or fewer anchors had a higher rate
of recurrent instability. These results suggest that four anchor points, at a
minimum, should be used to obtain secure shoulder stabilization, regardless of
the initial extent of the Bankart lesion.
Patients with bone defects (either on the glenoid or the humeral side) are
at risk of failure; they are, therefore, poor candidates for such a procedure.
In our study, the risk of postoperative recurrence was significantly related
to an osseous bone loss of the posterior aspect of the humeral head (a large
Hill-Sachs lesion) or of the anterior glenoid rim (a compression fracture with
a discrete bone fragment). By contrast, glenoid avulsion fractures (with
separation of a bone fragment) did not compromise the chance of success
(Fig. 5).
The fact that glenoid bone loss is associated with postoperative recurrence
is not surprising. It has been shown that an anterior glenoid bone defect can
affect glenohumeral stability in two ways. First, loss of part of the glenoid
surface reduces the concavity of the glenoid, and, second, the arc length of
the glenoid is
decreased42-46.
Recently, Burkhart et
al.47 and Lo et
al.48 suggested
that it is possible to quantify glenoid bone loss in shoulder instability
arthroscopically: the "inverted pear shape" of the glenoid is an
indicator of substantial glenoid bone loss. According to those authors, the
bare spot of the glenoid is the landmark that the arthroscopist should look
for since it is equidistant to the anterior, posterior, and inferior glenoid
rim in a normal shoulder. Those authors, in a cadaver study, measured the
distance from the center of the bare spot to the rim and observed an average
distance of 11
mm47. In a recent
paper, Kim et al. also quantified glenoid bone loss and found that
postoperative recurrence was related to an osseous defect of >30% of the
entire glenoid
circumference49. To
help with patient selection, preoperative radiographs like the Garth or
Bernageau radiographs and, more recently, preoperative computed tomography
scans with sagittal reconstruction have been recommended to detect and
quantify such glenoid bone
loss49-53.
The fact that humeral bone defects (i.e., large Hill-Sachs lesions) lead to
postoperative recurrent instability after arthroscopic Bankart repair is not
surprising either. The articular arc deficit of the humeral head allows
engagement of the bone defect on the anterior glenoid rim, the so-called
engaging Hill-Sachs
lesion44. In the
present study, we did not look for such potential engagement of the humeral
defect on the anterior glenoid rim during arthroscopy; the Hill-Sachs lesion
was just qualified as "large" on the basis of arthroscopic
observation. Sophisticated radiographic methods to measure humeral bone
defects in anterior shoulder instability have recently been proposed by
Kralinger et al.54
and Ito et al.55.
Hopefully, in the near future, it will be possible to accurately quantify
humeral bone defects with the help of preoperative computed tomography
scans.
Another important finding of the present study is that patients with
ligamentous hyperlaxity of the shoulder are at risk for a recurrence of
instability after arthroscopic stabilization. Hyperlaxity is difficult to
analyze because its definition is
variable27,56-59.
Excessive external rotation with the arm at the side, that is, rotation of
>90°, has been reported to be an indicator of shoulder hyperlaxity and
a risk factor for postoperative instability; this was confirmed in the present
series58,59.
This sign is almost always bilateral and is, for us, an indicator of anterior
hyperlaxity, resulting from a congenitally lax anterior capsule.
In addition, we found that a stretched inferior glenohumeral ligament,
defined as a preoperative asymmetrical hyperabduction test of >20°
compared with the contralateral side, was associated with failure of the
arthroscopic stabilization. We found the hyperabduction test of Gagey to be
valuable for the detection of inferior
laxity28,29.
It has been our experience that such inferior laxity is, most of the time,
acquired and related to the amount of plastic deformation of the inferior
axillary pouch after recurrent dislocations or subluxations.
Finally, we found that the association of a glenoid compression fracture
and a stretched inferior glenohumeral ligament leads to a 75% recurrence rate;
therefore, we believe that the combination of hyperlaxity (external rotation
of >90° and/or hyperabduction) and glenoid bone loss is a formal
contraindication to an arthroscopic Bankart repair
(Fig. 6). It is incumbent on
the surgeon to be aware of this pathologic association and to treat it
appropriately.
In summary, this retrospective study suggests that bone loss (on the
glenoid and/or on the humeral side) and/or ligamentous hyperlaxity represent
contraindications to arthroscopic shoulder stabilization. Those conditions
clearly compromise the chance of success of an arthroscopic Bankart repair. It
should be noted that nine of the fourteen patients who had failure of
arthroscopic treatment underwent revision with a Latarjet procedure, which
resulted in a stable shoulder in all of
them60. The results
of the present study have encouraged us to continue with arthroscopic
treatment for recurrent anterior instability with two caveats: patient
selection and rigorous technique. We reserve arthroscopic treatment for
patients with minimal to no humeral or glenoid bone loss. Moreover, we use a
technique that allows an optimal capsular shift and
tensioning61.