DOI:10.2214/AJR.04.1818
AJR 2006; 186:237-241
© American Roentgen Ray Society
Diagnostic Performance of MR Arthrography After Rotator Cuff Repair
Sylvain R. Duc1,
Bernard Mengiardi1,
Christian W. A. Pfirrmann1,
Bernhard Jost2,
Juerg Hodler1 and
Marco Zanetti1
1 Department of Radiology, University Hospital Balgrist, Forchstrasse 340,
Zurich CH-8008, Switzerland.
2 Department of Orthopedic Surgery, University Hospital Balgrist, Zurich
CH-8008, Switzerland.
Received November 24, 2004;
accepted after revision January 12, 2005.
Address correspondence to S. R. Duc
(sylvain.duc{at}balgrist.ch).
Abstract
OBJECTIVE. The objective of this study was to investigate the
diagnostic performance of MR arthrography after rotator cuff repair.
MATERIALS AND METHODS. MR arthrographic examinations of the shoulder
performed after rotator cuff repair and before revision surgery were
retrospectively analyzed in 48 patients (31 males, 17 females; mean age, 50.3
years; age range, 17-69 years). Full-thickness and partial-thickness defects
of the supraspinatus, infraspinatus, and subscapularis tendons were diagnosed
independently by two radiologists. Revision surgery served as the standard of
reference.
RESULTS. Observer 1 correctly recognized five of eight intact
supraspinatus tendons, 10 of 19 partial-thickness defects, and 19 of 21
full-thickness defects. For observer 2, the numbers were three of eight, eight
of 19, and 18 of 21. The corresponding numbers for the infraspinatus tendon
for observer 1 were 28 of 31, 0 of three, and 14 of 14 tendons. For observer
2, they were 28 of 31, two of three, and 11 of 14. For the subscapularis
tendon, observer 1 made the correct diagnosis in 18 of 31, five of six, and
nine of 11 tendons. The results for observer 2 were 26 of 31, one of six, and
10 of 11 tendons. Interobserver agreement (weighted 
) was 0.47 for the
supraspinatus, 0.64 for the infraspinatus, and 0.20 for the subscapularis
tendons, respectively.
CONCLUSION. Postoperative full-thickness defects of the rotator cuff
are reliably diagnosed with MR arthrography. The diagnostic performance for
partial-thickness defects is only moderate.
Keywords: joint • MR arthrography • MRI • musculoskeletal imaging • orthopedic surgery • shoulder
Introduction
Full-thickness defects of previously repaired rotator cuff tears can be
diagnosed with an accuracy of 83-90% on standard MR images
[1,
2]. Differentiation of
partial-thickness defects of the rotator cuff tendons from clinically
irrelevant postoperative distortion and thinning of tendons is more difficult.
The reported diagnostic performance of MRI has been variable
[1,
2]. In shoulders that have not
undergone surgery, MR arthrography improves the accuracy for detection of
supraspinatus [3] and
subscapularis [4] abnormalities
when compared with standard MRI. This method may also be worthwhile in the
evaluation of the postoperative shoulder.
The purpose of this study was to investigate the diagnostic performance of
MR arthrography after rotator cuff repair.
Materials and Methods
Patients
Forty-eight patients (31 males, 17 females; mean age, 50.3 years; age
range, 17-69 years) were included in this retrospective study in a consecutive
fashion if the following criteria were met: recurrent symptoms after rotator
cuff repair, MR arthrography performed according to our institution's standard
protocol no longer than 6 months before revision surgery, and a surgical
report including a precise description of rotator cuff morphology. The mean
interval between MR arthrography and revision surgery was 53 days (range,
1-169 days). The initial repair was performed during an open procedure in 40
patients and during arthroscopy in eight patients. Transosseous refixation was
performed in 15 patients, a tendon suture was used in three patients, and
suture anchors were used in five patients; for 25 patients, no detailed
surgical data were available.
Before revision surgery the following symptoms were encountered, with more
than one symptom possible in any patient: Forty patients had shoulder pain, 14
complained about shoulder weakness, and 14 reported limited range of motion
for the shoulder. Revision surgery was performed during an open procedure in
28 patients and during arthroscopy in 20 patients. The following diagnoses
were made during revision surgery: Twenty-one supraspinatus tendons had a
full-thickness defect, 19 had a partial-thickness defect, and eight were
intact. Fourteen infraspinatus tendons had a full-thickness defect and three,
a partial-thickness defect; 31 were intact. Eleven subscapularis tendons had a
full-thickness defect, six had a partial-thickness defect (superior border of
the tendon involved), and 31 were intact. In 19 patients, one tendon was
involved (supraspinatus tendon, n = 16; subscapularis tendon,
n = 3). In 17 patients, two tendons were involved (supraspinatus and
infraspinatus tendons, n = 10; supraspinatus and subscapularis
tendons, n = 7). In seven patients, all three tendons were
involved.
The institutional review board responsible for our institution does not
require approval for the review of patients' records or images for each single
study. Patient rights are protected by a law that requires that patients are
informed about the possibility of scientific review of their anonymous data
and that they have the opportunity to reject such use of their data.
Imaging Protocol
For MR arthrography, the joint was injected under fluoroscopic guidance.
The intraarticular needle position was verified by injecting 1 mL of iopamidol
(Iopamiro 200, Bracco Diagnostics). Subsequently, 10 mL of gadopentetate
dimeglumine (2 mmol/L [Magnevist, Schering]) and 1 mL of mepivacaine
hydrochloride (Scandicain 2%, AstraZeneca) were injected. The Swiss drug
administration (Swissmedic) has approved the use of gadolinium for
intraarticular injection. The MR examination was performed within 15 min or
less after injection either on a 1.5-T scanner (Magnetom Symphony, Siemens
Medical Solutions) or a 1.0-T scanner (Magnetom Expert, Siemens). A dedicated
shoulder coil was used.
The protocol on the 1.5-T scanner included a coronal oblique proton
density-weighted fat-saturated turbo spin-echo (SE) sequence (TR/TE, 2,350/15;
field of view, 160 x 100 mm; matrix, 512 x 512; section thickness,
4 mm; number of signals averaged, 1; turbo factor, 7), a coronal oblique
T2-weighted fat-saturated turbo SE sequence (3,000/91; field of view, 160
x 100 mm; matrix, 512 x 512; section thickness, 4 mm; number of
signals averaged, 1; turbo factor, 7), a coronal oblique T1-weighted
fat-saturated SE sequence (792/12; field of view, 160 x 100 mm; matrix,
512 x 512; section thickness, 3 mm), a sagittal oblique T1-weighted SE
sequence (500/12; field of view, 160 x 100 mm; matrix, 512 x 512;
section thickness, 4 mm), and an axial T1-weighted SE sequence (500/12; field
of view, 160 x 100 mm; matrix, 512 x 512; section thickness, 3
mm).
The imaging protocol on the 1.0-T scanner included a coronal oblique
dual-echo turbo SE sequence (TR/TE first echo, second echo, 3,500/15, 105;
field of view, 180 mm; matrix, 512 x 512; section thickness, 4 mm;
number of signals averaged, 3; turbo factor, 7), a coronal oblique T1-weighted
fat-saturated SE sequence (TR/TE, 800/20; field of view, 160 mm; matrix, 256
x 256; section thickness, 4 mm), a sagittal oblique T1-weighted turbo SE
sequence (700/12; field of view, 160 mm; matrix, 256 x 256; section
thickness, 5 mm; number of signals averaged, 4; turbo factor, 2), and an axial
T1-weighted turbo SE sequence (580/20; field of view, 160 mm; matrix, 512
x 512; section thickness, 4 mm; number of signals averaged, 1; turbo
factor, 2).
View larger version (75K):
[in this window]
[in a new window]
[as a PowerPoint slide]
|
Fig. 1B —69-year-old man with full-thickness defect of rotator cuff.
Coronal oblique T1-weighted fat-saturated turbo SE image (800/20) shows
contrast material within full-thickness defect (arrows) and within
subacromial bursa.
|
|
View larger version (110K):
[in this window]
[in a new window]
[as a PowerPoint slide]
|
Fig. 2A —67-year-old woman with partial-thickness defect of rotator
cuff. Coronal oblique T2-weighted turbo spin-echo (SE) image (TR/TE,
3,500/105) shows subtle articular-sided partial-thickness substance defect of
supraspinatus tendon (arrow).
|
|
View larger version (96K):
[in this window]
[in a new window]
[as a PowerPoint slide]
|
Fig. 2B —67-year-old woman with partial-thickness defect of rotator
cuff. Coronal oblique T1-weighted fat-saturated turbo SE image (800/20) shows
contrast material within defect (arrow). Susceptibility artifacts are
visible around major tuberosity.
|
|
Image Analysis
MR examinations were retrospectively analyzed independently by two
radiologists. At the time of the study, observer 1 had specialized in
musculoskeletal radiology for 10 years and observer 2, a fellowship-trained
radiologist, had 2 years' experience in musculoskeletal radiology. The
observers were blinded with regard to the surgical results. The supraspinatus,
infraspinatus, and subscapularis tendons were evaluated separately. A
full-thickness defect of the supraspinatus or infraspinatus tendon was
diagnosed when contrast material was present throughout the tendon substance
or when the tendon was not visualized. The presence of contrast material
within the subacromial-subdeltoid bursa was noted on T1-weighted images. An
articular or bursa-sided partial-thickness defect of the supraspinatus or
infraspinatus tendons was diagnosed in the presence of contrast material
partially entering the articular-sided tendon substance (mainly diagnosed on
T1-weighted sequences) and preexisting bursal fluid entering the bursa-sided
tendon substance. Subscapularis tendon defects were classified according to a
previously published classification
[4] in agreement with the
criteria used intraoperatively by our orthopedic surgeons. A defect involving
the cranial third or less of the tendon was described as a partial-thickness
defect. Such abnormalities are commonly associated with a lesion of the
reflection pulley of the biceps tendon
[5]. A defect involving more
than the cranial third was considered to represent a full-thickness
defect.
An error analysis was performed after completion of the study by a panel
composed of the two original observers, an additional musculoskeletal
radiologist with 7 years' experience in musculoskeletal MRI, and an orthopedic
surgeon specializing in shoulder surgery and experience with MRI of the
shoulder. They reviewed the cases that had a discrepancy between MRI and
intraoperative findings in the diagnosis of a full-thickness defect. The
errors were categorized as follows: difficulty in attributing the lesion to
the correct tendon, presence of subacromial contrast material (representing a
potential bias toward a full-thickness defect), blooming artifacts, and
presence of a severely thinned tendon.
Statistics
Sensitivity, specificity, diagnostic accuracy, and positive and negative
predictive values of MR arthrography were calculated. Interobserver agreement
was analyzed with weighted kappa statistics. SPSS software (version 11,
Statistical Package for the Social Sciences) was used for statistical
analyses.
Results
For detection of full-thickness defects of the supraspinatus tendon (Figs.
1A and
1B), the sensitivity was
comparable for both observers (observer 1, 19 of 21 defects detected
[sensitivity, 90%]; observer 2, 18 of 21 defects detected [sensitivity, 86%]).
Specificity was more observer-dependent. Observer 1 detected 24 of 27 tendons
without full-thickness defect (specificity, 89%), and observer 2 detected 16
of 27 such tendons (specificity, 59%). Accuracy was 90% (43 of 48 tendons
correctly diagnosed) for observer 1 and 71% (34 of 48 tendons correctly
diagnosed) for observer 2.
For detection of full-thickness infraspinatus defects, sensitivity,
specificity, and accuracy for observer 1 were 100% (14 of 14 tendons, 34 of 34
tendons, 48 of 48 tendons), respectively. For observer 2, they were 79%, 94%,
and 90% (11 of 14 tendons, 32 of 34 tendons, 43 of 48 tendons),
respectively.
For detection of full-thickness subscapularis defects, sensitivity,
specificity, and accuracy for observer 1 were 82%, 100%, and 96% (nine of 11
tendons, 37 of 37 tendons, 46 of 48 tendons), respectively. For observer 2,
they were 91%, 92%, and 92% (10 of 11 tendons, 34 of 37 tendons, 44 of 48
tendons), respectively.
Interobserver agreement for the diagnosis of defects of the supraspinatus
tendon was moderate ( = 0.47), good for the infraspinatus tendon
( = 0.64), and fair for the subscapularis tendon ( = 0.20).
When all three possible diagnoses were taken into consideration (intact
tendons, partial-thickness defects, full-thickness defects), observer 1
correctly recognized five of eight intact supraspinatus tendons, 10 of 19
partial-thickness defects (Figs.
2A and
2B), and 19 of 21
full-thickness defects (Figs.
1A and
1B), respectively, resulting in
an agreement of 71% (34/48). For observer 2, the numbers were three of eight,
eight of 19, and 18 of 21 (agreement, 60% [29 of 48 tendons]), respectively.
The corresponding numbers (intact tendons, partial-thickness defects,
full-thickness defects) for the infraspinatus tendon for observer 1 were 28 of
31, 0 of three, and 14 of 14 tendons (agreement, 88%), respectively. For
observer 2, they were 28 of 31, two of three, and 11 of 14 (agreement, 85%).
For the subscapularis tendon, observer 1 made the correct diagnosis in 18 of
31, five of six, and nine of 11 tendons (agreement, 67%), respectively. The
results for observer 2 were 26 of 31, one of six, and 10 of 11 tendons
(agreement, 77%). Additional results are shown in Tables
1,
2,
3.
Contrast material was found in the subacromial-subdeltoid bursa in 24 of
the 26 shoulder joints with a full-thickness defect of at least one tendon but
also in 13 of 22 joints with intact tendons or with partial-thickness defects
(Figs. 2A,
2B,
3A, and
3B). Correspondingly, contrast
material did not leak from the glenohumeral joint space into the
subacromial-subdeltoid bursa in two of 26 with and in nine of the 22 joints
without a full-thickness defect (Figs.
4A and
4B).
View larger version (122K):
[in this window]
[in a new window]
[as a PowerPoint slide]
|
Fig. 3A —40-year-old man with intact rotator cuff. Coronal oblique
T2-weighted fat-saturated turbo spin-echo (SE) image (TR/TE, 3,000/91) shows
intact rotator cuff. Postoperative hyperintensity can be seen within tendon
substance (arrow). One observer incorrectly diagnosed case as
bursa-sided partial-thickness defect.
|
|
View larger version (130K):
[in this window]
[in a new window]
[as a PowerPoint slide]
|
Fig. 4A —37-year-old woman with full-thickness defect not diagnosed on
MR arthrography. Coronal oblique T2-weighted turbo spin-echo (SE) image
(TR/TE, 3,500/105) with typical postoperative signal irregularity of
supraspinatus tendon (arrow). No defect is visible.
|
|
View larger version (106K):
[in this window]
[in a new window]
[as a PowerPoint slide]
|
Fig. 4B —37-year-old woman with full-thickness defect not diagnosed on
MR arthrography. Coronal oblique T1-weighted fat-saturated turbo SE image
(800/20) shows susceptibility artifact (arrows) overlapping
supraspinatus tendon. No contrast material is visible within subacromial
bursa.
|
|
The results of the error analysis are summarized in
Table 4. More than one error
could occur in a specific case. The most common error was incorrect
localization of a defect to either the supraspinatus or the infraspinatus
tendon on the MR images (five of seven shoulders).
Postoperative susceptibility artifacts contributed to the incorrect
diagnosis in one of seven cases (Figs.
4A and
4B). Contrast material leaking
into the subacromial bursa may have misled the observers to diagnose a defect
at MRI in three of seven cases. In one subscapularis tendon, a false-positive
diagnosis of a full-thickness defect was made in the presence of a thin
fibrous plate.
Discussion
After surgical repair of the rotator cuff, approximately 25% of patients
remain symptomatic [6]. Many
may have recurrent or residual rotator cuff defects. The published prevalence
of such abnormalities varies between 20% and 47%
[7-10].
They are more common in patients with a short interval between the onset of
the symptoms and surgery [11].
The prevalence of postoperative defects is higher in symptomatic than in
asymptomatic patients (47% vs 21%, respectively)
[10]. Symptoms are more common
with increasing size of postoperative defects
[8,
9,
12].
In our study, MR arthrography showed a diagnostic performance in the
diagnosis of full-thickness defects similar to those reported in earlier
publications using unenhanced MR examinations
[1,
2]. Previously reported
sensitivities and specificities of unenhanced MRI vary between 86-100% and
25-91%, respectively [1,
2,
13]. For the detection of
partial-thickness rotator cuff defects, the reported results are variable.
Although Magee et al. [2]
presented a sensitivity and specificity of more than 80%, Owen et al.
[1] were unable to distinguish
partial rotator cuff defects from distorted but intact tendons. The
classification of defects used for the subscapularis tendon differed from the
one used for the supra- and infraspinatus tendons because it was based on the
craniocaudal extent of the lesion
[4]. This classification is the
same as the one used by orthopedic surgeons and reflects the typical
development of subscapularis defects—that is, in a craniocaudal
direction.
Postoperative distortion and scarring may lead to tendon surface
irregularities mimicking partial-thickness defects
[10,
14]. In addition,
fluid-sensitive sequences may show intratendinous signal abnormalities
mimicking a tendon defect. Such abnormalities may persist for a long time
postoperatively [10].
Contrast material leaking into the subacromial-subdeltoid bursa is known to
be of little relevance postoperatively, contrary to the preoperative situation
[15,
16]. This has been confirmed
by our results. Rather unexpectedly, we found that contrast material does not
necessarily leak into the subacromial bursa in the presence of a
full-thickness rotator cuff defect. Presumably, such defects are sealed by
biomechanically nonfunctional scar tissue. The differentiation of scar tissue
from functional but thinned tendon is difficult and can easily lead to
diagnostic errors. MR arthrography apparently is not superior to standard MRI
in diagnosing partial defects because distortion and signal abnormality of the
intact tendon cause the same problems for both types of examinations.
The retrospective design of our study represents an important limitation.
The surgical data of the initial operations were commonly incomplete because
most of the patients were initially treated outside our institution. Surgical
reports for the second operation were incomplete with regard to abnormalities
not relating to the rotator cuff. Therefore, potential advantages of MR
arthrography in diagnosing abnormalities of articular cartilage and the labrum
were not evaluated.
In conclusion, postoperative full-thickness defects of the rotator cuff are
reliably diagnosed with MR arthrography. The performance of MR arthrography in
the diagnosis of partial-thickness defects is only moderate. MR arthrography
does not improve the diagnostic performance compared with the results of
standard MRI reported in the literature.
References
- Owen RS, Iannotti JP, Kneeland JB, Dalinka MK, Deren JA, Oleaga L.
Shoulder after surgery: MR imaging with surgical validation.
Radiology 1993;186
: 443-447[Abstract/Free Full Text]
- Magee TH, Gaenslen ES, Seitz R, Hinson GA, Wetzel LH. MR imaging of
the shoulder after surgery. AJR 1997;168
: 925-928[Abstract/Free Full Text]
- Hodler J, Kursunoglu-Brahme S, Snyder SJ, et al. Rotator cuff
disease: assessment with MR arthrography versus standard MR imaging in 36
patients with arthroscopic confirmation. Radiology1992; 182:431
-436[Abstract/Free Full Text]
- Pfirrmann CW, Zanetti M, Weishaupt D, Gerber C, Hodler J.
Subscapularis tendon tears: detection and grading at MR arthrography.
Radiology 1999;213
: 709-714[Abstract/Free Full Text]
- Weishaupt D, Zanetti M, Tanner A, Gerber C, Hodler J. Lesions of
the reflection pulley of the long biceps tendon: MR arthrographic findings.
Invest Radiol 1999;34
: 463-469[CrossRef][Medline]
- Wolfgang GL. Surgical repair of tears of the rotator cuff of the
shoulder: factors influencing the result. J Bone Joint Surg
Am 1974; 56:14
-26[Abstract/Free Full Text]
- Knudsen HB, Gelineck J, Sojbjerg JO, Olsen BS, Johannsen HV,
Sneppen O. Functional and magnetic resonance imaging evaluation after
single-tendon rotator cuff reconstruction. J Shoulder Elbow
Surg 1999; 8:242
-246[CrossRef][Medline]
- Gazielly DF, Gleyze P, Montagnon C. Functional and anatomical
results after rotator cuff repair. Clin Orthop Relat
Res 1994; 304:43
-53
- Harryman DT 2nd, Mack LA, Wang KY, Jackins SE, Richardson ML,
Matsen FA 3rd. Repairs of the rotator cuff: correlation of functional results
with integrity of the cuff. J Bone Joint Surg Am1991; 73:982
-989[Abstract/Free Full Text]
- Zanetti M, Jost B, Hodler J, Gerber C. MR imaging after rotator
cuff repair: full-thickness defects and bursitis-like subacromial
abnormalities in asymptomatic subjects. Skeletal
Radiol 2000; 29:314
-319[CrossRef][Medline]
- Gerber C, Fuchs B, Hodler J. The results of repair of massive tears
of the rotator cuff. J Bone Joint Surg Am2000; 82:505
-515[Abstract/Free Full Text]
- Galatz LM, Ball CM, Teefey SA, Middleton WD, Yamaguchi K. The
outcome and repair integrity of completely arthroscopically repaired large and
massive rotator cuff tears. J Bone Joint Surg Am2004; 86:219
-224[Abstract/Free Full Text]
- Motamedi AR, Urrea LH, Hancock RE, Hawkins RJ, Ho C. Accuracy of
magnetic resonance imaging in determining the presence and size of recurrent
rotator cuff tears. J Shoulder Elbow Surg2002; 11:6
-10[CrossRef][Medline]
- Spielmann AL, Forster BB, Kokan P, Hawkins RH, Janzen DL. Shoulder
after rotator cuff repair: MR imaging findings in asymptomatic
individuals— initial experience. Radiology1999; 213:705
-708[Abstract/Free Full Text]
- Calvert PT, Packer NP, Stoker DJ, Bayley JI, Kessel L. Arthrography
of the shoulder after operative repair of the torn rotator cuff. J
Bone Joint Surg Br 1986; 68:147
-150
- DeOrio JK, Cofield RH. Results of a second attempt at surgical
repair of a failed initial rotator-cuff repair. J Bone Joint Surg
Am 1984; 66:563
-567[Abstract/Free Full Text]
CiteULike 
Complore 
Connotea 
Del.icio.us 
Digg 
Reddit 
Technorati What's this?
Top
Abstract
Introduction
