Inclusion and Exclusion Criteria
Between 1999 and 2003, 197 all-arthroscopic rotator cuff repairs were
performed by the senior surgeons (L.L. and B.T.). All shoulders in the present
study had either an isolated supraspinatus tear or a supraspinatus tear with
an infraspinatus extension. Seventeen shoulders in the present study had
fraying of the subscapularis identified at the time of arthroscopy. However,
no shoulder that required operative repair of the subscapularis tendon was
included in the study. Shoulders with a full-thickness tear of at least one
tendon as documented on a preoperative computed tomography arthrogram that had
undergone repair of the rotator cuff lesion with use of the double-row suture
anchor technique and had been followed for a minimum of two years after
surgery were included in the present study. All shoulders had postoperative
arthrography, a computed tomography arthrogram, or a magnetic resonance
imaging arthrogram at a minimum of six months after surgery in order to
evaluate the integrity of the rotator cuff repair. Approximately 750 rotator
cuff repairs (including approximately 500 arthroscopic repairs and 250 open
repairs) were performed during the period of the study. However, the exclusion
criteria for entry into the study included single-row repair, open repair, a
concomitant subscapularis tear, the refusal of the patient to have a
postoperative arthrogram, and a duration of clinical follow-up of less than
two years. One hundred and five shoulders met the inclusion criteria. The
present study received institutional review board approval, and all patients
were enrolled in compliance with this protocol.
The indication for surgery was the failure of conservative treatment,
defined as a trial of physical therapy with the goal of strengthening of the
rotator cuff, deltoid, and scapular stabilizers. The study cohort included
ninety-five patients (forty-eight women and forty-seven men) who had had a
mean age of fifty-two years (range, thirty-six to seventy-nine years) at the
time of surgery. Ten patients had a bilateral procedure. The mean duration of
follow-up was thirty-six months (range, twenty-four to fifty-eight months).
Seventy-six patients had surgery on the dominant shoulder.
Classification of Rotator Cuff Tears
Each rotator cuff lesion was also evaluated in both the coronal and
sagittal planes at the time of arthroscopy. In the coronal plane, the lesion
was evaluated, according to the classification system of Patte, as distal
(thirty-six tears; 34.3%), intermediate (forty-seven tears; 44.8%), or
retracted (twenty-two tears;
21%)13. In our
experience, we have found that the degree of coronal plane retraction, and not
the absolute size of the lesion, has the greatest impact on the technical
difficulty of repair of rotator cuff tears. Small tears were defined as
supraspinatus ruptures with retraction to the articular margin on the humerus
(Patte type 1). Large tears were defined as either intermediate or retracted
supraspinatus lesions according to the Patte classification system
(Fig. 1). All of the retracted
supraspinatus tears were associated with at least an intermediate-level
infraspinatus tear and were therefore classified as massive rotator cuff
tears. Therefore, none of the retracted supraspinatus tears in our series were
classified as large rotator cuff tears. In order to incorporate the sagittal
plane of these tears into our classification scheme, we also considered
supraspinatus tears that extended into the infraspinatus to be large tears as
long as the degree of infraspinatus retraction was not greater than distal
retraction as defined by Patte. Massive tears were defined as those that were
characterized by a supraspinatus tear with retraction to the level of the
glenoid and an infraspinatus tear that was at least of intermediate grade
(that is, one with retraction medial to the articular margin of the
humerus).
Patient Evaluation and Determination of the Structural Integrity of
Repair
All patients underwent a standard history and physical examination and
completed a standardized questionnaire that included pain scales and
subjective functional assessments preoperatively and at least two years
postoperatively. The visual analog scale score for pain (range, 0 to 15
points, with 0 points representing maximum pain), the Constant score, and the
active range of motion were recorded for each
shoulder14. All
patients had computed tomography arthrography (103 shoulders) or magnetic
resonance arthrography (two shoulders) preoperatively and at a mean of
twenty-three months (range, six to forty-one months) postoperatively.
The structural integrity of the rotator cuff repair was evaluated by
examining the computed tomography or magnetic resonance imaging arthrography
images of the footprint reconstruction. An intact repair was defined as a
complete anatomic reconstruction of the footprint. Intratendinous leakage,
although stratified separately, was considered to be consistent with an intact
repair. In cases of small tears, resection or release of the rotator interval
was never necessary; thus, structural failure of the rotator cuff repair was
considered to have occurred when there was any extravasation of contrast
medium into the subacromial space. In cases of large and massive tears, an
intact repair was defined as a complete anatomic reconstruction of the
footprint. As the operative technique that was used to mobilize the torn
tendon edges in cases of large and massive tears always required resection or
release of the rotator interval to achieve an anatomic reduction, leakage of
contrast medium into the subacromial space after rotator cuff repair could not
be used as a method to evaluate the structural integrity of these repairs
(Fig. 2). Rather, the
arthrograms demonstrating extravasation of contrast medium through the
footprint reconstruction were classified as demonstrating either small
transtendinous leaks or large extravasations, although both were considered to
be failures of repair. This subclassification was made in order to determine
if the extent of leakage of contrast medium, and, therefore, the failure of
repair, was correlated with the functional outcome. As described by Boileau et
al., a transtendinous leak was considered to be evidence of partial
healing1.
Strength-Testing
Manual strength-testing was performed for each shoulder preoperatively and
at a minimum of two years postoperatively with use of a portable isometric
dynamometer (Isobex 2.0; Cursor, Bern, Switzerland). Strength-testing was
performed with the arm in 90° of abduction in the scapular plane and
neutral rotation while the patient was standing with the dynamometer at
shoulder level. The patient was instructed to hold this position with maximum
force for three seconds during the measurements.
Arthroscopic Rotator Cuff Repair
All patients in the present study underwent regional anesthesia with an
interscalene block before entrance into the operating room. The patients were
placed in the beach-chair position with the arm forward flexed with 3 kg of
traction. Three to five arthroscopic portals were used to perform the rotator
cuff repair. Typically, these portals were placed posteriorly,
posterolaterally, laterally, anterolaterally, anteriorly, and
anteroinferiorly. The subacromial space was cleared of bursa, reactive
synovitis, and subdeltoid adhesions, and acromioplasty was performed prior to
inspection of the rotator cuff in order to classify the tear. The
coracohumeral ligament, the superior capsule, and/or the rotator interval were
released as needed in order to maximize the mobility of the rotator cuff prior
to repair. Adequate release of the cuff was achieved when the tissue edges
could be easily reduced over the greater tuberosity with use of a grasper
instrument in order to avoid the so-called tension overload
phenomenon15.
After the greater tuberosity had been gently decorticated with a burr or
shaver, the first anchor was placed at the junction of the articular cartilage
and the medial aspect of the footprint on the greater tuberosity
(Fig. 3). The sutures were
passed through the tendon medially with use of a curved suture-passing device
(Spectrum; Linvatec, Largo, Florida) as a shuttle relay device in a horizontal
mattress pattern. As described by Lo and
Burkhart3, a medial
row and a lateral row of anchors were placed on the border of the anatomic
footprint of the superior rotator cuff
(Fig. 4-A). The mean number of
suture anchors (G2 anchor; Mitek, Raynham, Massachusetts) used for footprint
reconstruction in the present series was 3.7. The sutures (#2 Ethibond;
Ethicon, Somerville, New Jersey) from the lateral anchor were passed through
the tendon edges either as U stitches, a lasso-loop, or simple stitches
(Fig. 4-B). We believe that the
lasso-loop technique allows for superior fixation in the tissues by
approximating a Mason-Allen configuration through the rotator cuff (Figs.
4-C and 4-D). Fifty-nine
shoulders with fraying, tearing, or instability of the biceps tendon were
managed with biceps tenotomy (nine shoulders) or tenodesis (fifty shoulders).
A subacromial decompression with acromioplasty was performed in all
shoulders.
Rehabilitation
Our postoperative rehabilitation protocol restricted patients to pendulum
exercises starting on the first postoperative day and continuing for three
weeks, with the extremity resting in a 30° abduction pillow made from a
generic large stockinette and foam padding or pillow when the exercises were
not being performed. After three weeks, the patients were instructed to
commence passive range-of-motion exercises in the plane of the scapula with
the assistance of a physical therapist. Active motion exercises were not
permitted until six weeks after surgery, and hydrotherapy was strongly
encouraged.
Statistical Analysis
Measurements are expressed as the mean and the standard deviation. The
means were compared with use of the Student t test for continuous variables.
The level of significance was set at p < 0.05. Single-variable regression
analysis was used to determine if relationships between the etiology of the
rotator cuff tear, the age of the patient, the duration of symptoms, and the
Workers' Compensation status had a significant effect on the clinical outcome
parameters evaluated in the present study.
Pain Score, Strength, and Motion
Overall, the patients experienced marked pain relief after rotator cuff
repair. The mean pain score on the visual analog scale improved from 4.7
± 4.2 (range, 0 to 15) preoperatively to 12.8 ± 3.0 (range, 5 to
15) postoperatively (p < 0.0001) (Table
I). Shoulders with intact rotator cuff repairs had significantly
more pain relief than did shoulders with failed repairs, with mean
postoperative pain scores of 13.3 ± 2.6 (range, 5 to 15) and 11.2
± 3.6 (range, 5 to 15), respectively (p = 0.02).
The active range of motion significantly improved after rotator cuff
repair. The mean forward flexion improved from 108° ± 39°
(range, 30° to 150°) preoperatively to 147° ± 12°
(range, 90° to 150°) postoperatively (p < 0.0001). The mean
abduction increased from 94° ± 40.5° (range, 20° to
150°) preoperatively to 142° ± 18° (range, 60° to
150°) postoperatively (p < 0.001). Strength also improved significantly
after rotator cuff reconstruction, from a mean of 2.9 ± 1.4 kg (range,
0 to 8 kg) preoperatively to a mean of 6.3 ± 2.7 kg (range, 2 to 12.5
kg) postoperatively (p < 0.001).
In an attempt to determine the impact of failure of rotator cuff repair on
the clinical outcome, we divided the shoulders into two cohorts (those with an
intact rotator cuff repair and those with a failed repair) and compared the
clinical outcomes between the groups. The mean forward flexion was 151°
for the shoulders with an intact repair, compared with 142° for those with
a failed repair (p = 0.74). The mean Constant score was 80.8 (p = 0.17) for
the shoulders with an intact repair, compared with 76.4 for those with a
failed repair. The mean strength was 12.9 kg for the shoulders with an intact
repair, compared with 11.4 kg for those with a failed repair (p = 0.32).
Interestingly, the mean postoperative pain score was 12.9 for shoulders with
an intact rotator cuff repair as compared with only 11.4 for those with a
failed repair (p = 0.014). A trend toward superior clinical outcomes was
observed in shoulders with an intact rotator cuff repair, although pain was
the only category in which a significant difference was achieved.
Overall, with the numbers studied, single-variable regression analysis of
our data points did not reveal a significant relationship between age at the
time of surgery, the degree of fatty infiltration, the duration of symptoms
prior to surgery, or the duration of the follow-up period and the ultimate
clinical outcome. Biceps tenotomy or tenodesis, when analyzed as an
independent variable, did not produce a significant difference in the outcome
parameters studied. In addition, Workers' Compensation status, the etiology of
the tear (degenerative or traumatic), and the preoperative status of the
biceps tendon did not have a significant influence on the ultimate outcome in
our study population, with the number of patients studied.
Constant Scores
The mean Constant score was 43.2 ± 15.1 points (range, 8 to 83
points) preoperatively and 80.1 ± 11.1 points (range, 46 to 100 points)
at a minimum of twenty-four months postoperatively (p < 0.001)
(Table I).
Analysis of Structural Integrity of Repair with Computed Tomography
or Magnetic Resonance Imaging Arthrography
Only twelve of the 105 shoulders in the present study had structural
failure of the double-row suture anchor repair as assessed with computed
tomography or magnetic resonance imaging arthrography after a mean of
twenty-three months of follow-up (Fig.
5). This result reflects the total number of repairs with
structural failure as defined above, with eight shoulders having a small
transtendinous extravasation and four having a large leak of contrast
material6. There
were no failures of repair in shoulders with small rotator cuff tears. In
addition, the outcomes for shoulders that had undergone postoperative computed
tomography or magnetic resonance imaging arthrography after less than
twenty-four months of follow-up were compared with those for shoulders that
had undergone the imaging study at a minimum of twenty-four months of
follow-up in order to determine if this variation significantly impacted the
clinical or radiographic result. The postoperative Constant score was 79.7 for
the forty-four shoulders that had computed tomography or magnetic resonance
imaging arthrography after less than twenty-four months of follow-up, compared
with 81.1 for the sixty-one shoulders that had the imaging study after at
least twenty-four months of follow-up (p = 0.5). We also compared the failure
rates between these two groups of shoulders. Failure of the rotator cuff
repair was documented in four of the forty-four shoulders that had the imaging
study after less than twenty-four months follow-up, compared with eight of the
sixty-one shoulders that had the imaging study after at least twenty-four
months of follow-up; however, with these small numbers, this difference was
not significant (p = 0.27). The numbers of shoulders that had failure of the
rotator cuff repair in the large and massive rotator cuff tear cohorts were
too small to analyze.
Comparison of Clinical Outcome Measures According to Tear Size
A comparative analysis of the clinical outcome measures in the present
study between shoulders with small, large, and massive rotator cuff tears was
performed. We could not identify any significant differences in the outcome
measures between shoulders with large rotator cuff tears and those with
massive rotator cuff tears. However, when the clinical outcome measures from
the group of shoulders with small rotator cuff tears were compared with those
for shoulders with either large or massive rotator cuff tears, there were
several significant differences. The group of shoulders with small rotator
cuff tears achieved a mean strength of 7.19 ± 3.0 kg (range, 2 to 12.5
kg) after rotator cuff repair, compared with only 5.4 ± 1.92 kg (range,
3 to 9 kg) for the group of shoulders with massive tears and 6.11 ±
2.53 kg (range, 2.0 to 12.0 kg) for the group of shoulders with large tears (p
< 0.05 for both comparisons). Interestingly, the amount of preoperative
active abduction and forward flexion was significantly lower in the group of
shoulders with large and massive rotator cuff tears than in the group of
shoulders with small rotator cuff tears (p < 0.05), although the
postoperative values for these variables did not differ significantly between
the groups. Finally, we could not identify any difference between the groups
with regard to the preoperative and postoperative values for pain, the
Constant score, or active external or internal rotation (p > 0.05).
Complications
There were no surgical complications.
The technique for double-row suture anchor fixation for arthroscopic
rotator cuff repair was first described by Lo and
Burkhart3. Those
authors proposed that by placing two rows of suture anchors, one on the medial
side of the footprint and the other on the lateral side, a more anatomic
repair configuration could be achieved. The result, they hypothesized, would
be a stronger repair construct and a larger contact area for healing, yielding
superior clinical outcomes and a more durable rotator cuff repair.
To our knowledge, the present report describes the first study to
prospectively evaluate the structural integrity of arthroscopic rotator cuff
repairs performed with use of the double-row suture anchor technique and to
correlate the integrity of these repairs with clinical outcomes. The rate of
structural failure after double-row fixation was only 11% and, to our
knowledge, this value represents the lowest rate of structural failure after
either open or arthroscopic repair as reported in the literature. Galatz et
al., in a study on the results of all-arthroscopic reconstruction of large or
massive rotator cuff tears with use of single-row suture anchors and simple
sutures, reported recurrence of the tear in seventeen of eighteen patients as
assessed with
ultrasonography6.
Boileau et al., in a study of sixty-five consecutive patients who had been
managed with arthroscopic repair of an isolated supraspinatus tear with the
tension band suture technique, reported that forty-six of the sixty-five
repairs remained structurally intact as demonstrated with computed tomography
arthrography or magnetic resonance imaging at a minimum of six months after
surgery1.
Interestingly, Boileau and colleagues used a coronal plane classification
system proposed by Thomazeau in order to assess the size of the supraspinatus
tears so that thirty-two of the sixty-five patients, despite having only one
tendon tear, were classified as having a large tear. Sugaya et al. conducted a
retrospective study in order to compare the clinical and structural outcomes
of arthroscopic rotator cuff repairs performed with use of double-row and
single-row suture anchor
fixation2.
Thirty-nine patients managed with single-row suture anchor repair and
forty-one patients managed with double-row suture anchor repair were evaluated
clinically and with magnetic resonance imaging before and after surgery. The
investigators reported that double-row suture anchor fixation resulted in a
significantly stronger repair with a lower failure rate, although the use of
magnetic resonance imaging without intra-articular contrast medium made it
difficult to draw definitive conclusions about the integrity of these repairs.
Gleyze et al. evaluated the structural integrity of arthroscopic repairs of
isolated supraspinatus tendons in a multicenter study and found that
fifty-three of eighty-seven repairs remained
intact4. Wilson et
al. performed second-look arthroscopy for thirty-three patients who had had
arthroscopic staple fixation of tears varying in size from small to large and
reported that twenty-two of the thirty-three repairs remained intact at the
time of arthroscopic staple removal, approximately three months after
surgery5. Our
clinical results compare favorably to those published in the literature for
rotator cuff repair with use of open, mini-open, or arthroscopic
techniques8,14,16-24.
In the present study, the structural integrity of rotator cuff repairs was
analyzed with use of computed tomography or magnetic resonance imaging
arthrography. Charousset et al., in a recent study that analyzed the accuracy
of computed tomography or magnetic resonance imaging arthrography for the
detection of both partial and full-thickness rotator cuff tears, reported that
the method demonstrated 99% sensitivity and 100% specificity for the analysis
of lesions involving the
supraspinatus25.
Other authors also have advocated computed tomography arthrography as an
excellent imaging modality for the evaluation of rotator cuff
pathology26-30.
One important consideration in the present study was related to the analysis
of rotator cuff integrity with use of computed tomography arthrography after
the repair of large and massive tears. We believe that it is critically
important to assess the thickness and integrity of the footprint
reconstruction rather than the presence or absence of leaking contrast medium
after rotator cuff reconstruction in patients with larger tears because our
operative technique for releasing retracted cuff tears often includes excision
of part or all of the rotator interval. This technique is sometimes necessary
in order to obtain adequate mobilization of the retracted tendons and is an
excellent technique for preventing excessive tension at the site of a rotator
cuff repair. Furthermore, we suspect that at least some of the intratendinous
leakage identified during the analysis of postoperative images may, in fact,
have been the result of contrast medium traveling along the suture line
through the tendon, although it is difficult to determine the importance of
this observation.
At the present time, the optimal technique and anchorsuture configuration
have not been established for arthroscopic rotator cuff
repair1.
Nevertheless, why might double-row suture anchor fixation explain the superior
results of this study with regard to repair integrity? Within the last few
years, several investigators have analyzed the anatomy of the rotator cuff
footprint as well as the strength, contact area, and contact pressure of
various repair configurations in order to determine the optimal construct for
tendon healing after rotator cuff repair. Apreleva et al. evaluated the
three-dimensional structure of the rotator cuff footprint and determined that
single-row suture anchor repairs provide only point fixation in the area
immediately surrounding the anchor and, therefore, cannot restore the anatomic
footprint12.
Tuoheti et al. performed a cadaver study to compare the contact pressure and
contact area associated with three different repair methods: single-row suture
anchor fixation, transosseous repair, and double-row suture anchor
fixation11. The
authors reported that the contact pressures for double and single-row suture
anchor fixation were not significantly different and that both generated
significantly higher contact pressure than did the transosseous repair.
However, their study also demonstrated that the contact area for single-row
fixation was limited to the area immediately surrounding the anchor whereas
the double-row suture anchor fixation had the largest and most anatomic
contact area of the three fixation methods studied. Therefore, they concluded
that the double-row suture anchor method was likely to be superior to the
single-row technique for optimizing tendon healing after rotator cuff repair.
Waltrip et al. compared the strength of double-row, single-row, and
transosseous repairs in a cadaver
study10. They found
that the mean number of cycles to failure for the double-row suture anchor
repairs was significantly greater than those for the single-row simple suture
repairs or the transosseous repairs performed with a mattress stitch. Since
88.6% of the repairs in our study healed completely without structural failure
after a minimum of twenty-four months follow-up, the double-row suture anchor
technique may be the optimal construct for arthroscopic rotator cuff
repair.
Several authors have identified a strong correlation between improved
functional results and intact rotator cuff
repairs1,6,9,31.
Indeed, our study supports this trend even if significance was not achieved.
In the twelve shoulders from the present series that had either partial
healing or complete failure of the repair, the strength of the rotator cuff
was diminished relative to that in shoulders with an intact repair of the same
size tear. Interestingly, pain relief in shoulders with intact rotator cuff
repairs significantly improved as has been observed in other
studies1,6,31,32.
One potential weakness of the present study is that it represents an
evaluation of structural integrity and clinical outcomes after a mean of
thirty-six months. Other authors have reported on the long-term clinical
outcomes of rotator cuff
repair33,34,
but we suspect, as Boileau et al. suggested, that patients with an intact
rotator cuff repair will continue to do well in the long term and may even
have improvement1. A
second weakness of the present study is that we did not track the number of
rotator interval releases performed for mobilization of the torn tendons in
large and massive rotator cuff tears and, therefore, we cannot assess the
impact of rotator interval release as an independent variable on the outcome
parameters studied. Finally, although our results related to structural
integrity support arthroscopic double-row suture anchor fixation as the
optimal repair technique, we did not perform a randomized direct comparison of
single-row, arthroscopic double-row, and open double-row repair, so we are
unable to definitively conclude that this technique for rotator cuff repair is
superior to all others.
In conclusion, the double-row suture anchor technique for arthroscopic
rotator cuff repair resulted in superior tendon healing as compared with
previously studied open and other arthroscopic methods of repair. Intact
rotator cuff repairs resulted in markedly improved pain relief in comparison
with repairs that failed or only partially healed. Shoulders with repaired
large and massive rotator cuff tears had less strength than those with smaller
tears. These findings suggest that the double-row suture anchor configuration
may be the optimal repair construct for arthroscopic rotator cuff repair,
although long-term studies will be needed to validate this concept. ?