J Shoulder Elbow Surg (2014) 23, 1682-1690

www.elsevier.com/locate/ymse

Long-term outcome of segmental reconstruction

of the humeral head for the treatment of locked
posterior dislocation of the shoulder
Christian Gerber, MD, FRCSEd(Hon)*, Sabrina Catanzaro, RN,
Michele Jundt-Ecker, MD, Mazda Farshad, MD, MPH

Department of Orthopaedics, University of Z€

urich, Balgrist University Hospital, Z€urich, Switzerland

Background: Locked posterior glenohumeral dislocations with impaction fractures involving less than
30% to 35% of the humeral head are most frequently treated with lesser tuberosity transfer into the defect,
whereas those involving more than 35% to 40% are treated with humeral head arthroplasty. As an alterna-
tive, reconstruction of the defect with segmental femoral or humeral head allograft has been proposed, but
the long-term outcome of this joint-preserving procedure is unknown.
Methods: Twenty-two shoulders in 21 patients with a locked posterior shoulder dislocation and an impaction
of at least 30% (mean, 43%) of the humeral head were treated with segmental reconstruction of the humeral
head defect. They were reviewed clinically and radiographically at a minimum follow-up of 5 years.
Results: Of the 22 shoulders, 19 could be followed up at 128 months (range, 60-294 months) postopera-
tively. Only 2 of the 19 patients needed a prosthesis more than 180 months after the index operation. Of the
other 17, 4 had radiographically advanced osteoarthritis (OA), 4 had mild OA, and 9 had no or minimal
OA. Eighteen shoulders were rated as subjectively excellent, none were rated as good, and one was
rated as fair. The final Constant-Murley score averaged 77 points (range, 52-98 points), the Subjective
Shoulder Value averaged 88% (range, 75%-100%), and only 2 patients had mild to moderate pain.
Mean active anterior elevation was 145
, and mean external rotation with the arm at the side was 42
.
Conclusion: Segmental reconstruction of humeral head defects for large anteromedial impaction fractures
caused by locked posterior dislocations durably restores stability and freedom from pain with an excellent
subjective long-term outcome.
Level of evidence: Level IV, Case Series, Treatment Study.
Ó 2014 Journal of Shoulder and Elbow Surgery Board of Trustees.
Keywords: Segmental reconstruction; humeral head; locked posterior dislocation; long-term

Traumatic, posterior dislocation of the shoulder is usually
associated with an anteromedialdso-called reverse Hill-
Sachsor McLaughlindimpression fracture of the humeral
This study was approved by the responsible ethical committee (Ref. Nr.
EK: KEK-ZH-Nr. 2011-0292).
*Reprint requests: Christian Gerber, MD, FRCSEd(Hon), University of
Z€
urich, Balgrist, Forchstrasse 340, CH-8008 Z€urich, Switzerland.
E-mail address: christian.gerber@balgrist.ch (C. Gerber).
head. This lesion can be so large that closed reduction becomes
impossible or is followed by immediate redislocation on slight
internal rotation of the reduced arm. Locked dislocations may
be seen acutely. Often, however, they are initially missed and
referred to specialized institutions with considerable delay.
Chronic dislocations, defined as those with a diagnostic delay
of at least 1 month, are likely to be associated with osteopenia
of the part of the humeral head that is no longer in contact with

1058-2746/$ - see front matter Ó 2014 Journal of Shoulder and Elbow Surgery Board of Trustees.
http://dx.doi.org/10.1016/j.jse.2014.03.017
Head reconstruction for locked posterior dislocation
the glenoid, as well as with degenerative changes of the
incongruous glenohumeral joint.
The pathomechanical role of the humeral head defect was
probably first recognized by McLaughlin,14 who recom-
mended transferring the subscapularis tendon into the defect
to prevent recurrence of dislocation. Neer and colleagues13
modified this procedure by recommending transfer of the
lesser tuberosity with its attached subscapularis tendon into
the defect. For humeral defects involving more than
approximately 35% to 40% of the humeral head, however,
such reconstruction may compromise joint mobility and
further reconstructive procedures, so humeral head arthro-
plasty or total shoulder replacement (total shoulder arthro-
plasty [TSA]) is usually preferred.2,15,17 The mean age of the
affected patients, however, is around 45 years,13,16 so joint-
preserving procedures would be desirable. Reconstruction
of the humeral head defect with a segmental allograft has
been proposed as an alternative to restore function and sta-
bility without compromising the original anatomy of the
glenohumeral joint.10 The long-term outcome of this joint-
preserving procedure, however, is unknown.
The purpose of this study was to review a consecutive
series of patients with locked posterior dislocations treated
with segmental humeral head reconstruction to determine the
clinical value of segmental humeral head reconstruction for
locked posterior shoulder dislocation.
Methods
A total of 21 consecutive patients with 22 locked posterior dis-
locations (bilateral in 1 patient) with humeral head defects of at
least 30% (mean, 43%; range, 30%-55%) were treated with
segmental reconstruction of the humeral head and were studied.
Four of these patients were the basis of a previous preliminary
report, and their long-term results are included in this study.10
There were 4 women (4 shoulders) and 17 men (18 shoulders).
The mean age of the patients at the time of dislocation was
44 years (range, 25-75 years). Four shoulders were treated acutely
within 7 days of injury, 5 shoulders were treated subacutely be-
tween 1 week and 1 month after injury, and 13 chronic cases were
treated with a mean delay of 6.3 months (range, 1-15 months).
Eight dislocations were incurred during an epileptic seizure and
two were incurred during a hypoglycemic convulsion; the
remaining 12 dislocations resulted from sports-related injuries
(biking in 5, soccer in 1, skiing in 2, horseback riding in 1, and
walking/jogging in 3). Two patients were referred after failed open
reduction and internal fixation of a proximal humeral fracture in
which the concomitant posterior glenohumeral joint dislocation
had not been recognized. All subacute and chronic patients were
seen for ‘‘painful stiffness’’ and had severe loss of passive and
active external rotation. Except for chronic cases, no functional
clinical assessment was carried out because of suspicion of pos-
terior shoulder dislocation.
After confirmation of the diagnosis with conventional radiog-
raphy, computed tomography (CT) was carried out in all patients.
CT documented an impression fracture of the head involving at
least 30% of the cartilaginous circumference but absence of
relevant posterior glenoid rim fractures or avulsions of the
1683
Figure 1 The defect size is calculated on a CT slice taken at or
immediately below the coracoid tip. The percentage of the carti-
lage angle (185
in this case) that the defect angle (70
in this
case) represents is the estimated size of the defect (38% in this
case).
infraspinatus or supraspinatus tendons. The size of the humeral
head defect was measured on the preoperative CT scan with the
head dislocated posteriorly. On a CT scan obtained at or imme-
diately below the level of the coracoid, a circle was laid over the
humeral head. A line from the center of the circle to the cartilage
immediately adjacent to the lesser tuberosity and another line
from the center of the circle to the posterior end of the cartilage
adjacent to the infraspinatus insertion determined the cartilage
angle. A second angle, the defect angle, was the angle formed by
the lines connecting, first, the anterior limit and, second, the
posterior limit of the defect with the center of the humeral head.
The percentage of the cartilage angle that the defect angle rep-
resented was the estimated size of the humeral head defect
(Fig. 1). A posterior glenoid rim defect of more than half of the
maximal anteroposterior diameter was considered relevant.11
Operative treatment
A previously described operative technique10 was performed, with
slight modification. Patients were placed in the beach-chair posi-
tion and operated on under general anesthesia for optimal relax-
ation. A deltopectoral approach was used. We divided the
subscapularis tendon approximately 1 cm from its insertion from
the superior border of the tendon to its inferior insertion, carefully
avoiding injury to the anterior circumflex vessels and protecting
the arcuate artery lateral to the bicipital groove. The capsule was
then divided vertically close to its humeral attachment. The su-
perior glenohumeral ligament and the coracohumeral ligament
were divided because failure to do so had resulted in difficulties in
reducing the humerus in the first cases. The capsular release was
carried out inferiorly beyond the subscapularis insertion. A first
attempt was then made to fully internally rotate the arm and pull
1684 C. Gerber et al.

the head laterally using a bone hook inserted into the anteromedial
head defect. If reduction could not be obtained, a special laminar
spreader with a smooth surface was applied to the glenoid and the
defect in the humeral head and was opened so that the head could
be pushed laterally in between 10
and 40
of abduction. After the
capsule was distended in this fashion, reduction was carried out
starting in maximal internal rotation, with the assistant pushing the
head anteriorly with his hand applied to the posterior aspect of the
shoulder and the surgeon pulling the head laterally, using a bone
hook, in the humeral head defect. After reduction, the humerus
was stabilized in neutral rotation. If, on intraoperative testing,
slight internal rotation led to immediate posterior redislocation,
segmental humeral head reconstruction was performed. In 17
shoulders, a segmental fresh-frozen femoral or humeral head
allograft from an institutional bone bank was then shaped like an
orange segment and adapted to fill the defect and restore the
spherical shape of the humeral head (Fig. 2). The contoured graft
was press fitted into the defect and usually fixed with two 3.5-mm
cancellous lag screws (Synthes, Paoli, PA, USA), which were
countersunk into the graft and oriented toward the greater tuber-
osity to definitely exit extra-articularly behind the cartilaginous
head (Figs. 2 and 3). Segmental reconstruction was carried out
with structural autografts of the iliac crest in 5 patients: In 2 acute
and 2 subacute cases, the impressed cartilage was elevated and
large iliac crest grafts were used to buttress the cartilage, which
was repositioned to near its original position; an unstable covering
cartilaginous flap was fixed with a screw to the remaining head as
an allograft. Autograft was only used in defects no larger than
40%. There were 2 acute and 2 subacute cases in which residual
cartilage could be elevated and used as restored joint surface, as
well as 1 chronic dislocation because the patient refused im-
plantation of an allograft. Anatomic restoration of the humeral
shape prevented intraoperative redislocation in all cases. The
anterior capsule was loosely repaired with absorbable sutures. The
posterior capsule was not addressed surgically. The subscapularis
was mobilized from the anterior glenoid neck and repaired end to
end without shortening with 3 or 4 No. 3 Ethibond sutures
(Ethicon, Somerville, NJ, USA) by use of modified Mason-Allen
stitches.12 The rotator interval was not closed.

Postoperative care

The arm was positioned in a brace with the neutrally rotated arm in
10
to 20
of abduction. The arm was taken out of the brace twice a
day for passive range-of-motion exercises with about 20
to 30
of
external and internal rotation. Use of the brace was discontinued at 6
weeks, and active exercises were started. Reaching behind the body
was not allowed for a further 2 weeks (total of 8 weeks).
Results
Of the 22 shoulders, 19 could be reviewed clinically and
radiographically for the purpose of this study by examiners
not involved in the primary treatment of the patients. One case
was excluded because the shoulder underwent revision to
TSA 10 months postoperatively because of persistent poste-
rior subluxation with subsequent collapse of the allograft and
painful glenoid erosion. A second patient refused further
follow-up after a routine clinical and radiographic follow-up
Figure 2 Images in a 27-year-old, right-handed patient who was
involved in a sports accident. (A) Posterior dislocation and hu-
meral head impaction involving 40% of humeral head circum-
ference. (B) Reconstruction with segmental femoral head
allograft. (C-F) Anteroposterior and axillary lateral radiographs
obtained postoperatively (C, D) and at 210 months postoperatively
(E, F). (G, H) Normal, pain-free shoulder function at 210 months
postoperatively.
visit at 3 months had shown an excellent early outcome. The
patient who underwent bilateral surgery died after a follow-up
examination more than 5 years after treatment of the first
shoulder but less than 5 years after treatment of the second
shoulder, so the second shoulder was not included in this
study. He and the remaining 18 patients were reviewed at a
mean of 10.7 years (range, 60-294 months) postoperatively.
Head reconstruction for locked posterior dislocation
Figure 3 Images in a 33-year-old man who fell. (A, B) Posterior
fracture dislocation with humeral head impaction estimated to be
50% of humeral head circumference. (C, D) Classic loss of function
caused by chronic, locked posterior shoulder dislocation. (E-H)
Radiographs obtained postoperatively (E, F) and at 148 months (G,
H). (I, K) Pain-free, full mobility at 148 months postoperatively.
Examination consisted of a structured interview and physical
examination, including assessment of the Constant-Murley
score (CS),3,4 calculation of the age- and gender-adjusted
1685
relative CS,8 and assessment of the Subjective Shoulder
Value (SSV).9 True anteroposterior and axillary lateral ra-
diographs were obtained for all patients. Particular attention
was paid to static subluxation,11,18 which was assessed on the
axillary lateral radiographs, and to osteoarthritis (OA). OA
was graded as minimal if there was mild deformity of the
humeral head but no narrowing of the glenohumeral joint line;
as moderate if, in addition, there was partial loss of the gle-
nohumeral joint line on at least 1 of the 2 radiographic pro-
jections; and as severe if there was complete loss of the
glenohumeral joint line on at least 1 of the 2 radiographic
projections.
Perioperative complications
There were no systemic, neurologic, or local perioperative
wound complications.
Revision surgical procedures
Of the original 22 shoulders, 3 (13%) ultimately underwent
revision using arthroplasty (the 2 patients lost to follow-up
did not undergo revision to arthroplasty). The first case is the
early failure described previously, which underwent revision
to TSA at 10 months. The patient obtained an excellent result
but was excluded from further study as an early failure. The
other 2 prosthetic revisions took place more than 15 years
after the index allograft reconstruction: In 1 patient, the hu-
meral head reconstruction had not recentered the humeral
head; this resulted in static posterior subluxation with
collapse of the head and active external rotation of less than
0
. A rotational osteotomy was performed to regain external
rotation at 14 months. Although both the allograft and, more
so, the residual head progressively collapsed and the shoulder
was relatively stiff, the patient continued to work at full ca-
pacity in a manual profession until 181 months after the index
operation when his advancing OA had become painful
enough to warrant hemiarthroplasty. A third, female patient,
who had rheumatoid arthritis, had an excellent result for
15.5 years, and then diffuse osteochondromatosis suddenly
developed. At age 74 years, she was treated with a reverse
total shoulder replacement and obtained an excellent sub-
jective and objective result. During head resection, the allo-
graft was shown to have healed to the adjacent native bone
without collapse. The latter 2 cases were considered long-
term outcomes of segmental allografts and are included in
this analysis with their final outcomes. Secondary avascular
necrosis (AVN) of the humeral head developed, presumably
due to alcohol abuse, in 1 further patient, and she died before
further treatment. Her last clinical and radiographic result is
included in this study.10 All 3 replacement surgical pro-
cedures were performed through the original deltopectoral
interval without problems. In 3 further patients, removal of
the screws was carried out at 8, 10, and 18 months after the
index operation to avoid glenoid cartilage damage.
1686 C. Gerber et al.

Table I Effect of time between injury and operative repair on development of OA and on clinical outcome
Acute (n ¼ 4) Subacute (n ¼ 5) Chronic (n ¼ 10) Significance: acute plus
Mean SD Mean or No. SD Mean or No. SD subacute (n ¼ 9)
of patients of patients vs chronic (n ¼ 10)

Follow-up period (mos) 161.75 73.08 103.20 33.91 127.80 88.33 NS)
Defect size (%) 38.75 11.09 41.00 5.48 45.50 7.25 NS)
Age (y) 48.40 7.77 48.76 14.05 43.16 15.32 NS)
New OA or increase in OA 1 of 4 1 of 5 5 of 10 NSy
CS
Total points 84.50 10.66 88.00 9.90 73.89 10.36 P < .05)
Relative % 96.00 5.66 96.25 7.50 83.86 7.40 P < .05)
SSV 87.50 6.45 98.75 2.50 79.86 12.65 P < .05)
NS, Not significant.
) Two-tailed Student t test.
y
Fisher exact test.

Table II Effect of allograft versus autograft for segmental reconstruction

Allograft (n ¼ 14) Autograft (n ¼ 5) Statistical significance
Mean or No. of patients SD Mean or No. of patients SD
Follow-up period (mos) 143.29 77.55 87.00 46.46 NS
Defect size (%) 45.71 7.03 35.00 3.54 P < .0001)
Age (y) 46.96 14.18 42.30 11.53 NS)
OA 7 of 14 0 of 5 NS (P ¼ .11)y
CS
Total points 77.50 11.33 85.00 11.79 NS)
Relative % 89.50 9.40 92.20 9.18 NS)
SSV 83.90 12.73 93.00 8.37 NS)
NS, Not significant.
) Two-tailed Student t test.
y
Fisher exact test.

Outcome
At a mean of 128 months (range, 60-294 months) post-
operatively, the mean SSV had improved from 33% (range,
0%-50%) preoperatively to 88% (range, 60%-100%). The
follow-up period in this study extended to a time before the
introduction of the CS. Because of this or because an acute
fracture dislocation was not scored, only 12 patients had a
preoperative CS and 15 had a preoperative SSV. All shoulders
were scored at final follow-up. The mean preoperative CS
was 37 points (range, 18-85 points), corresponding to a mean
relative CS of 41% (range, 20%-91%). At final follow-up, the
mean CS was 77 points (range, 52-98 points) and the mean
relative CS was 84% (range, 75%-100%). The mean subscore
for pain in the CSdout of 15 points (where 0 points indicates
maximal pain and 15 points indicates no pain)dwas 8 points
(range, 3-14 points) preoperatively and 14 points (range, 9-15
points) at final follow-up. An interesting and surprising
finding was that 15 of 19 patients indicated that they had no
pain. Mean active elevation of 145
(range, 110
-170
) and
mean active external rotation with the arm at the side of 42

(range, 0
-60
) were achieved.
Despite the excellent pain scores, 4 of the 17 patients
without prostheses had radiographically advanced OA, 4 had
mild OA, and 9 had no or minimal OA. Eighteen patients
(including 2 who underwent replacement surgery) rated their
affected shoulders as excellent, none as good, and 1 as fair.
None of the results were related to the age or gender of
the patients. Conversely, the clinical outcome in the 10
chronic cases was worse than that in the 9 acute and sub-
acute cases (mean relative CS of 84% vs 96% and mean
SSV of 93% vs 80%). The higher prevalence of OA in the
chronic cases did not reach statistical significance. The
follow-up period for the chronic cases, however, was
slightly shorter, and a longer follow-up period could only
have increased the observed difference. OA did not develop
in roughly half of the patients with a chronic dislocation,
however, and we could not detect a time span from injury to
operation that would predict the development or absence of
development of OA. The 5 chronic patients with OA had
Head reconstruction for locked posterior dislocation 1687

Table III Effect of defect size on development of OA and on clinical outcome

Defect <45% (n ¼ 10) Defect 45% (n ¼ 9) Statistical significance
Mean or No. of patients SD Mean or No. of patients SD
Follow-up period (mos) 112.20 57.42 146.56 88.89 NS)
Defect size (%) 36.50 4.12 50.00 3.54 P < .001)
Age (y) 48.90 16.42 42.22 8.57 NS)
OA 5 of 10 8 of 9 P < .05y
CS
Total points 83.67 9.08 75.25 13.12 NS)
Relative % 94.13 7.83 86.14 9.03 NS)
SSV 91.25 7.91 82.00 14.48 NS)
NS, Not significant.
) Two-tailed Student t test.
y
Fisher exact test.

Table IV Significance of presence or absence of OA

No new OA or increase in OA New OA or increase in OA (n ¼ 7) Statistical significance
(n ¼ 12)
Mean or No. of patients SD Mean or No. of patients SD
Follow-up period (mos) 107.00 55.19 165.29 91.24 P ¼ .161)
Defect size (%) 39.58 7.22 48.57 5.56 P ¼ .008)
Allograft 7 of 12 7 of 7 P ¼ .106y
Chronic 5 of 12 5 of 7 P ¼ .349y
Age (y) 45.63 13.93 45.91 13.47 P ¼ .966)
CS
Total points 83.36 9.92 73.00 12.36 P ¼ .113)
Relative % 92.56 8.71 87.17 9.43 P ¼ .289)
SSV 90.56 8.82 81.50 14.82 P ¼ .218)
) Two-tailed Student t test.
y
Fisher exact test.

only a slightly longer time span between injury and treat-
ment than the 5 chronic patients without OA (Table I).
Whether autografts or allografts are better could not be
determined in this study for several reasons. First, the follow-
up period for the allograft cases averaged 143 months,
whereas that of the autograft cases averaged only 87 months.
Second, the mean defect size reconstructed with allograft
was 46%, whereas that reconstructed with autograft was only
35% (P < .005). Third, only 1 of the 5 autograft re-
constructions, as opposed to 9 of the 14 allograft re-
constructions, was in the less favorable chronic group. We
could not determine whether the allograft or the aforemen-
tioned confounders caused the observed high prevalence of
OA in the cases treated with allograft (Table II). Conversely,
we could document that the clinical outcome with allograft is
not inferior to that of autograft, even though the allograft-
treated cases were more chronic, had larger defects, and
had longer follow-up. The clinical outcome therefore does
not suggest a clinical advantage of autograft (Table II) but
suggests that restoration of the sphericity of the humeral head
is the key factor for a favorable clinical outcome.
When the 10 defects of up to 40% were compared with
the 9 defects of at least 45%, the larger defects had a
longer mean follow-up period (147 vs 112 months). OA
was markedly more prevalent in the large–defect size
group but the clinical outcome was only slightly and
insignificantly inferior to that in the smaller–defect size
group (Table III).
In an attempt to determine the risk factors for the
development of OA, we compared the patients without new
OA or without progression of OA (n ¼ 12) with the pa-
tients in whom OA either had developed or had increased
by at least 1 degree (n ¼ 7): Only the defect size was
statistically significantly larger in the OA group. Use of
allograft was more frequent and the follow-up period was
longer in the OA group, both without reaching statistical
significance. It may be that the study is underpowered to
reach statistical significance for these parameters. The
same is true for the clinical assessment, which did show a
roughly 10% lower SSV and CS, but the most surprising
finding is the excellent subjective clinical outcome despite
OA (Table IV).
1688
Figure 4 Images in a 42-year-old patient who was involved in a
bicycle accident and had a delay in diagnosis of 12 months. (A, B)
Painful restriction of active range of motion, notably with –40
of
external rotation. (C) Moderate OA and head defect of 55% on
radiograph. (D) OA of residual head intraoperatively. (E, F)
C. Gerber et al.
Discussion
Although it is recognized that chronicity of the dislocation
with softening of the remaining humeral head and OA of the
glenohumeral joint are relevant factors, the anteromedial
impression fracture of the humeral head is the most consis-
tent and biomechanically most relevant feature of a chronic,
locked posterior dislocation of the glenohumeral joint. In the
3 cases on which he operated, McLaughlin14 observed that
‘‘even a small amount of internal rotation or flexion of the
adducted extremity resulted in immediate redislocation,
except when the humeral defect was obliterated by an in-
strument of appropriate bulk.’’ He used the end of the divided
subscapularis to fill this defect and restored stability. The
McLaughlin procedure has been used in several studies
dealing with defects usually involving less than 40% of the
humeral cartilage circumference; the reported results are
mixed, and it is not recommended for defects larger than 40%
to 45%, for long-standing dislocations, or for elderly
patients.13,16,19
Neer and colleagues13 later introduceddas a modification
of the McLaughlin proceduredtransfer of the lesser tuber-
osity with its attached subscapularis tendon into the humeral
defect. This procedure has yielded excellent short-term to
midterm results in a series of 7 acute cases1 and good to
excellent13 or fair to good results for more chronic cases.19 It
currently seems to be preferred over the original McLaughlin
procedure and is an established and successful procedure for
defects involving less than 33% to 50% of the articular sur-
face. In addition to limiting glenohumeral range of motion in
relatively large defects, lesser tuberosity transfer and sub-
scapularis transfer have been considered risk factors for po-
tential future revisions.16
Hemiarthroplasty and TSA have been advocated for de-
fects involving more than approximately 40% of the carti-
laginous circumference. In an early study, one-third of
hemiarthroplasties failed; however, the size of the lesion was
not specified in these cases.13 Of 6 total shoulder re-
placements, 1 failed early because of recurrent instability.
Though clinically satisfied, 2 of the remaining 5 patients had
pain at the 2- to 4-year follow-up assessment, with 1 patient
showing a circumferential radiolucent line around the gle-
noid component. The other 3 patients had a complete
radiolucent line around the glenoid component at 2 to 3 years
postoperatively. The only patient who had undergone TSA
=
Segmental femoral head allograft reconstruction of humeral head
defect. (G, H) Some flattening of residual head at 3 months
postoperatively. (I, K) Consolidation of graft and moderate OA at
5 years. At 25 years of follow-up, OA had increased (L, M) and
the patient was functionally moderately limited but pain free; he
still considered his result excellent (N, O). He had worked at full
capacity in a manual profession for the past 24 years, never took
any medication, and did not consider any further treatment.
Head reconstruction for locked posterior dislocation
after lesser tuberosity transfer had a fair result.13 Twelve
patients from the Mayo Clinic treated with shoulder arthro-
plasty for locked posterior dislocation were reviewed after a
mean of 9 years.17 Two patients needed early revision, and at
final follow-up, 1 patient achieved an excellent result, 6 had a
satisfactory result, and 5 had an unsatisfactory result.
Although the size of the treated defect was not specified, at
least the depicted patient with a recurrent dislocation had a
defect no larger than the defects treated with segmental
allograft reconstruction in this study. In another series of total
shoulder replacements, the mean pain score remained at 3.5
of 10, and 4 of 7 patients did not obtain active anterior
elevation of more than 100
.2 Good results with a mean CS of
60 points were reported in 12 shoulder arthroplasties
implanted for particularly large defects7 at 3 years post-
operatively, but 2 patients showed mild to severe migration of
the humeral head. Overall, the results of hemiarthroplasty
and TSA for locked posterior shoulder dislocations are fair,
and it appears that this form of treatment should be reserved
for extremely large defects or situation in which the
remaining humeral head is already (partially) collapsed
(ping-pong ball).16 Although the previously reported pros-
thetic reconstructions may not have perfectly restored head
size and glenohumeral geometry, it is unknown why the re-
ported results of arthroplasty are inferior to the results
observed with segmental head reconstruction.
Before the advent of lesser tuberosity transfer and hemi-
arthroplasty, Dubousset6 proposed a posterior approach to
reconstruct the posterior capsule combined with restoring the
shape of the humeral head using autograft. It cannot be
determined whether these good and excellent results were
due to reconstruction of the humeral head or due to the
posterior capsulolabral repair. Allograft restoration of the
shape of the humeral head through a deltopectoral approach
without addressing the posterior capsule was first reported
almost 30 years later.10 The results in 4 cases at a minimum
follow-up of 5 years were encouraging, although 1 patient
had delayed AVN of the remaining humeral head and 1 pa-
tient who had had moderate OA at the operation had a mild
functional limitation. Diklic et al5 reviewed 13 consecutive
allograft reconstructions of defects of 25% to 50% at a mean
follow-up of 4.5 years. They reported no recurrence of
instability, 1 patient with osteonecrosis of the humeral head,
and 12 excellent results, for an overall CS of 87%. Nine pa-
tients were pain free, 3 had occasional mild, and the patient
with AVN had moderate pain.
The results of our study confirm that the excellent and
good results reported at midterm follow-up5,10 are main-
tained at long-term follow-up (Figs. 2 and 3). The clinical
outcome is surprising especially in terms of subjective
evaluation and pain. Despite the prevalence of OA in roughly
half of the patients, pain was consistently absent or minimal
(Fig. 4). The findings of the subjective assessment, with 14
excellent results and 1 fair result, as well as a mean SSV of
88%, are unexpected. Objectively, there were 2 early failures:
1 graft collapse leading to early shoulder replacement and 1
1689
flattening of the head with incomplete joint reduction leading
sequentially to rotational osteotomy and shoulder replace-
ment. Nonetheless, 2 of the 3 prosthetic revisions became
necessary at more than 15 years after the index operation. We
consider a survival of 15 years for these allografts as a suc-
cess, especially because subsequent prosthetic surgery could
be carried out in a shoulder with surgically undistorted
anatomy.
With the excellent results observed in this long-term
study, the question of limitations of the study and optimal
indications arises. We did not have a control group; the pa-
tients were prospectively documented but retrospectively
reviewed. We only treated patients with defects larger than
30% and no larger than 55%. Although smaller defects can
certainly be treated with segmental reconstruction, this may
not be necessary. We cannot determine whether segmental
allograft or autograft reconstruction is superior or inferior to
lesser tuberosity transfer in small defects. For larger defects,
however, comparison with the literature suggests that if the
results of Diklic et al5 and those in our study can be repro-
duced, segmental head reconstruction may be the treatment
of choice for this pathology.
If segmental head reconstruction is selected, cases that
must be treated more than 1 month after the dislocation are
likely to fare well clinically but to fare less well than those
treated earlier. Conversely, age, gender, the size of the defect,
and the degree of OA did not significantly influence the
outcome. The development of OA is mostly related to the size
of the defect and probably does lead to somewhat inferior
results ultimately, although the study was underpowered to
reach significance for the inferior clinical results in the pa-
tients with OA in this series.
Conclusion
In a consecutive series of 22 locked posterior disloca-
tions of the shoulder, anatomic allograft or autograft
reconstruction of the anteromedial humeral head defect
yielded excellent clinical results at a mean follow-up of
10.7 years and a maximum follow-up of 25 years.
Comparable long-term follow-up is not available for
treatment alternatives. With an ultimate failure rate of
13% and a mean SSV of 88% at more than 10 years,
segmental humeral head reconstruction yields results for
the treatment of locked posterior shoulder dislocation
with a head defect of 30% to 55% that are at least as
good as the results reported for alternative forms of
treatment at approximately 5 years of follow-up.
Disclaimer
The authors, their immediate families, and any research
foundations with which they are affiliated have not
1690
received any financial payments or other benefits from
any commercial entity related to the subject of this article.
References
1. Banerjee M, Balke M, Bouillon B, Wafaisade A, Helm P, Akoto R,
et al. Excellent results of lesser tuberosity transfer in acute locked
posterior shoulder dislocation. Knee Surg Sports Traumatol Arthrosc
2013;21:2884-8. http://dx.doi.org/10.1007/s00167-012-2217-x
2. Cheng SL, Mackay MB, Richards RR. Treatment of locked posterior
fracture-dislocations of the shoulder by total shoulder arthroplasty.
J Shoulder Elbow Surg 1997;6:11-7.
3. Constant CR, Gerber C, Emery RJH, Søjbjerg JO, Gohlke F, Boileau P.
A review of the Constant score: modifications and guidelines for its
use. J Shoulder Elbow Surg 2008;17:355-61. http://dx.doi.org/10.
1016/j.jse.2007.06.022
4. Constant CR, Murley AH. A clinical method of functional assessment
of the shoulder. Clin Orthop Relat Res 1987;(214):160-4.
5. Diklic ID, Ganic ZD, Blagojevic ZB, Nho SJ, Romeo AA. Treatment of
locked chronic posterior dislocation of the shoulder by reconstruction of
the defect in the humeral head with an allograft. J Bone Joint Surg Br
2010;92:71-6. http://dx.doi.org/10.1302/0301-620X.92B1.22142
6. Dubousset J. Luxation posterieure de l’epaule. Rev Chir Orthop
Reparatrice Appar Mot 1967;53:65-85.
7. Gavriilidis I, Magosch P, Lichtenberg S, Habermeyer P, Kircher J.
Chronic locked posterior shoulder dislocation with severe head
involvement. Int Orthop 2010;34:79-84. http://dx.doi.org/10.1007/
s00264-009-0762-9
C. Gerber et al.
8. Gerber C. Latissimus dorsi transfer for the treatment of irreparable
tears of the rotator cuff. Clin Orthop Relat Res 1992;(275):152-60.
9. Gerber C, Fuchs B, Hodler J. The results of repair of massive tears of
the rotator cuff. J Bone Joint Surg Am 2000;82:505-15.
10. Gerber C, Lambert SM. Allograft reconstruction of segmental defects
of the humeral head for the treatment of chronic locked posterior
dislocation of the shoulder. J Bone Joint Surg Am 1996;78:376-82.
11. Gerber C, Nyffeler RW. Classification of glenohumeral joint insta-
bility. Clin Orthop Relat Res 2002;(400):65-76.
12. Gerber C, Schneeberger AG, Beck M, Schlegel U. Mechanical
strength of repairs of the rotator cuff. J Bone Joint Surg Br 1994;76:
371-80.
13. Hawkins RJ, Neer CS, Planta RM, Mendoza FX. Locked posterior
dislocation of the shoulder. J Bone Joint Surg Am 1987;69:9-18.
14. McLaughlin HL. Posterior dislocation of the shoulder. J Bone Joint
Surg Am 1952;34:584-90.
15. Mestdagh H, Maynou C, Delobelle JM, Urvoy P, Butin E. Traumatic
posterior dislocation of the shoulder in adults. Apropos of 25 cases [in
French]. Ann Chir 1994;48:355-63.
16. Robinson CM, Aderinto J. Posterior shoulder dislocations and frac-
ture-dislocations. J Bone Joint Surg Am 2005;87:639-50. http://dx.doi.
org/10.2106/JBJS.D.02371
17. Sperling JW, Pring M, Antuna SA, Cofield RH. Shoulder arthroplasty
for locked posterior dislocation of the shoulder. J Shoulder Elbow Surg
2004;13:522-7. http://dx.doi.org/10.1016/j.jse.2004.02.012
18. Walch G, Ascani C, Boulahia A, Nove-Josserand L, Edwards TB.
Static posterior subluxation of the humeral head: an unrecognized
entity responsible for glenohumeral osteoarthritis in the young adult.
J Shoulder Elbow Surg 2002;11:309-14. http://dx.doi.org/10.1067/
mse.2002.124547
19. Walch G, Boileau P, Martin B, Dejour H. Unreduced posterior luxa-
tions and fractures-luxations of the shoulder. Apropos of 30 cases [in
French]. Rev Chir Orthop Reparatrice Appar Mot 1990;76:546-58.