- Email: [email protected]
Magnetic resonance imaging of rotator cuff tearing and degenerative tendon changes: Correlation with histologic pathology
Magnetic resonance imaging of rotator cuff tearing and degenerative tendon changes: Corr.lion with histologic pathology Kimio Nakagaki, MD, Jlro Ozaki, MD, Yasuharu Tomita, MD, and Susumu Tamai MD, Kashihara, Nara, Japan
We studied degeneration of the torn rotator cuff by comparing the signal intensity of the torn tendon edge on magnetic resonance imaging with cellularity of the tendon. Fifty-one shoulders with rotator cuff tears requiring surgical repairs were scanned before surgery in a O.5T magnetic resonance imaging system. T7 -, T2-, T2*-, and proton-density-weighted sequences were obtained. Three specimens from the torn tendon edge that were obtained at operation were examined microscopically, and the number of tendon cells was calculated and averaged. Degeneration of the torn rotator cuff edge was associated with an increase in signal intensity of the torn tendon edge on T7 -, T2-, T2*, and proton-density-weighted images and a decrease in tendon cellularity. The T2-, T2*-, and proton-density-weighted images were more useful, because on the T7 -weighted image it was difficult to identify the torn cuff edge. (J SHOULDER ELBOW SURG
7993;2: 756-64.)
It
is important to evaluate the degree of degeneration of a torn rotator cuff before surgery to accurately assess the prognosis after operative treatment.' Also, if we can evaluate the degree of degeneration of the torn rotator cuff and the size of the defect before surgery, we can estimate how much torn cuff edge we should excise and whether or not we will be able to repair the tear with simple sutures to the greater tuberosity. In the literature no reports have been made on the quantitative analysis of the correlation between degeneration of the torn rotator cuff edge and signal intensity of this remaining tendon on magnetic resonance imaging (MRI). In advanced degeneration of the rotator cuff, it has been shown that cellularity of the tendon is diminished. 1 , 7 In this study we compare the signal intensity of the torn rotator From the Department of Orthopaedic Surgery, Nara Medical University Reprint requests KlmlO Nakagakl, MD, Department of Orthopaedic Surgery, Nora Medical University, Kashihara, Nara, 634 Japan Copyright © 1993 by Journal of Shoulder and Elbow Surgery Board of Trustees 1058-2746/93/$100 + 10 32/1147339
156
cuff edge on various MR images with cellularity of the tendon and thus provide a noninvasive method to evaluate the degree of degeneration present in the torn rotator cuff.
MATERIAL AND METHODS Fifty-one shoulders in 50 patients with rotator cuff tears requiring surgical repair were scanned before surgery, and the tendon tear edge was examined histologically. In this study group, 11 women and 39 men had an age range from 35 to 84 years, with a mean age of 58.2 years. The mean preoperative period of nonoperative treatment was 9.6 months. These shoulders were scanned before surgery with a 20 cm circular surface coil with a 0.5 T unit (MRT-50A/2, Toshiba Medical System, Tokyo, Japan). Patients were examined in the supine position with the arm held in neutral rotation. Imaging was performed in the oblique coronal plane (scapular plane) parallel to the long axis of the supraspinatus muscle. The T1-weighted sequence (spin echo method TRITE = 450/20) and the T2*-weighted sequence (field echo method TRITE = 500/22, flip angle = 30°) were obtained in all shoulders (Figure 1, C).
J. Shoulder Elbow Surg. Volume 2, Number 3
Nakagaki et al.
157
Figure 1 MR imaging of full-thickness tear of supraspinatus tendon. 0,
Deltoid muscle; 5, suprasupinatus muscle. A, T1-weighted image (SE 450/20). identification of torn rotator cuff edge IS difficult. Signal intensity: T = 84.0, C = 106.0, D = 202.1, S = 223.4. B, T2-weighted image (SE 2000/80). Identification of torn rotator cuff edge is easier. (Signal intensity: T = 75.1, C = 536.0, D = 116.8, S = 296.8). C, T2*-weighted image (FE 500122 FA30). Identification of torn rotator cuff edge is even easier. (Signal intensity: T = 208.8, C = 685.0, D = 348.6, S = 529.9). D, Proton-densityweighted image (SE 2000/30). Identification of torn rotator cuff edge is much easier. (Signal intensity: T = 572.9, C = 1159.0, D = 857.6, S = 1099.6).
The T2-weighted sequence (spin echo method TRITE = 2000/80) was obtained in 36 shoulders (Figure 1, B), and the proton-densityweighted sequence (spin echo method TRITE = 2000/30) was obtained in 42 shoulders (Figure 1, 0). The thickness of the imaging section was 5 mm, with a 1 mm space between sections. The field of view was 250 mm, and the average number of signals was two. Matrix size was 256 x 512 in T1- and T2*-weighted sequences and was 160 x 512 in T2- and proton-densityweighted sequences. We were able to clearly identify the torn cuff
edge of all the patients on MRI and to diagnose the extent of rotator cuff tearing before surgery. Thirty-seven patients had a full-thickness tear of the rotator cuff, and 13 patients had a partialthickness tear. Of those with full-thickness tearing, three patients had a small tear (tear size <1 cm), 21 had a medium tear (1 cm ::5 tear size <3 em), seven had a large tear (3 cm ::5 tear size <5 cm), and seven had a massive tear (tear size :2:5 cm). One patient had a medium tear in the right shoulder and a massive tear in the left. Of the partial-thickness tears, seven patients had a bursal surface tear, and six had an ar-
158
J. Shoulder Elbow Surg.
Nakagaki et al.
May / June 1993
Calculation of signal intensity in MRI
Figure 2 Mean signal Intensity was calculated for each of the nine areas (0.86 mm x 1 0 mm) at torn rotator cuff edge on MRI.
Spots of measurement of signal intensity an MRI torn cuff edge cuff defect,,
'vr/~~~
5iiiiiiii
,.----0.----- supraspinatus muscle
deltoid ---H-lK],tI muscle
Figure 3 Locations far measurement of signal Intensity on MRI
ticular side tear. At surgery, we found that the preoperative diagnoses of the types of rotator cuff tearing were correct. We analyzed the mean signal intensity of
nine areas (0.86 mm x 1.0 mm) of the torn rotator cuff edge using MR imaging (Figure 2), and we regarded the value as the signal intensity of the torn cuff edge. When a torn edge
J. Shoulder Elbow Surg. Volume 2, Number 3
Nokogoki et 01.
159
Figure 4 A, Histologic specimen of mildly degenerative torn cuff. Many tenocytes and distinct wavy appearance are visible (arrow). (Original magnification x 100) B, Histologic specimen of severely degenerative torn cuff. It shows loss of cellularity and absence of wavy appearance. (Original magnification x 100) was thin, we obtained the mean signal intensity in the slightly more proximal thick cuff area. In the same manner, we calculated the signal intensities of the middle points of the deltoid muscle, the supraspinatus muscle, and the area of the cuff defect (Figure 3). We determined the ratia of the signal intensity of the torn cuff (C) to that of the deltoid muscle (D) and similarly determined the ratio of the signal intensity of the torn cuff to that of the suprasupinatus muscle
(5).
Histologic specimens were obtained at surgery from the torn tendon edge. When the torn cuff edge was thin, we obtained the spp:.imen slightly more proximal to have a sample of about 3 mm in thickness. The three specimens from each shoulder were examined microscopically with a magnification of x 400, and the number of tendon cells in each part was calculated and averaged. The average was regarded as the cellularity of the torn tendon edge.
160
Nokogoki et 01.
1. Shoulder Elbow Surg. May / June 7993
Signal Intensity on T1-weighted Image
Signal Inlensity on T2-weighted Image
Xl00
7
Ttest 0.1
6
4 T
2
~~~~:::=~T
I I
I I
0
9
.
8 7
~=.,---=:::::::;=e:
I
I
~
I
I
.L
.L
~,
4
I
3 2
I
1
0 Tom Cuff Edge
Cuff Defect Portion
0 ........- Torn Cuff Edge
B
_1
,, ,,
T
6 5
I
f
•I
A
Tlest
11 10
5
3
Xl00 12
Cuff Defect Portion
Signal Intensity on Proton-density-weighted Image X1000
Ttest
3
Ttes!
0
O
2
Cuff Defect Portion
Tom Cuff Edge
Cuff Defect Portion
Figure 5 Statistical testing assessing difference between Signal Intensity of torn rotator cuff edge and that of tendon defect on various types of MR imaging. A, Statistical testing assessing difference between signal intenSity of torn rotator cuff edge and that of defect on Tl-weighted image. 8, Statistical testing assessing difference between signal intensity of torn rotator cuff edge and that of defect on T2-weighted image. C, Statistical testing assessing difference between signal intensity of torn rotator cuff edge and that of defect on T2*-weighted (field echo) image. D, Statistical testing assessing difference between signal intensity of torn rotator cuff edge and that of defect on proton-density-weighted image
Histologic examination of the torn rotator cuff with mild degeneration showed many tenocytes and a distinct wavy appearance. The tendon bundles stained deep pink with hematoxylineosin (Figure 4, A). However, more severely degenerated tendon edges showed loss of cellularity and absence of the wavy appearance. The tendon bundles become homogeneous in texture and stained weakly (Figure 4, 8).3· 4. 8 The following studies were performed for interpretation of the MRI examination. We ex-
amined the difference between the signal intensity of the torn cuff edge and that of the cuff defect in each MRI sequence with the paired ttest. The level of significance was set at 0.05. We then assessed which sequence most clearly confirmed the presence of the torn cuff edge. We calculated the Pearson's correlation coefficient between C/D, CIS, and cellularity of the torn cuff. When the absolute value of the correlation coefficient was larger than 0.5, we recognized a significant difference. Using this information, we then determined the relationship
J. Shoulder Elbow Surg. Volume 2, Number 3
Nokogoki et
X,OO
,
CelIu1anty and
00 R--o
celluilriy and CIS
X'O
630 (il
R--o 455 (0 002<1'<0 02)
"""
20
Cellu1l1rl1y and CID Ao=-o.543 (il
,
'9
'8
;00 ••
17 '6
\ 0
~~
0
13 ... a...... 12 11
........ \
... "'
t~~ - • •
\,.
6
\
S
0
-'
•
,
3
'
0
':'
,
2
o
..
~o
0° 0
4
------
"""'"",,--,
\
'
o
.'001
B
A
CelIu1anty of Torn Cuff Edge
CelIu1anty lWld CIO : R--o.614 (0
,
\
\
Cellulanty ard CIS . A--O 441 (O.OO2<:p
,,
\
2
'
0:00
3
2
4
on T2·welghled Image
Cellularrty of Torn Cuff Edge
Xl00 \
CelluilVty and CIS : R=-o 289 (0 1<:p
,
\°0
O' I·
0
~
••
i
0
0
'~
3 \
161
CeIJlJanty of Tom Cuff Edge
CellUanty of Tom Cuff Edge
3
01.
X'O 20 19 18 17 18 15
\
CelIu1arty lWld CIO . Rz-o.638 (0
't
eea.Any lWld CIS A--o.345 (O.05
14
'3
12
"'0 9
8
7 6
o.
5
4 3 2 1
1
c
• ; CIO 0;00
I
2 on FI8ld Echo Image
0-4------_----"---....--------';:..,.
•:00 I
0 1 2
o
0;00
on
Proton~...".;ghIed
Image
Figure 6 Correlation between the signal intensity of the torn rotator cuff divided by the signal intensity of the deltoid (C/O) or the signal intensity of the torn cuff divided by the signal intensity of the supraspinatus muscle (CIS) and the cellularity of the torn rotator cuff. Solid lines are regression lines between CIO and cellularity of torn cuff. Dotted lines are regression lines between CIS and cellularity of torn cuff. A, Correlation between CIO, CIS on Tl-weighted image and cellularity of torn rotator cuff. B, Correlation between CIO, CIS on T2-weighted image and cellularity of torn rotator cuff. C, Correlation between CIO, CIS on T2*-weighted (field echo) image and cellularity of torn rotator cuff. D, Correlation between CIO, CIS on proton-density-weighted image and cellularity of torn rotator cuff. between the signal intensity of the torn cuff on each sequence of MRI and the pathologic degeneration of the torn cuff.
RESULTS On the T1-weighted sequence no clear difference was found between the signal intensity of the torn cuff edge and that of the tendon defect (0.1 < p < 0.2) (Figure 5, A), demonstrating that identification of the torn cuff edge was very difficult on the T1-weighted image. The
correlation coefficient between CI D on the T1weighted image and cellularity of the torn cuff was -0.630 (0 < p < 0.002) (Figure 6, A). Thus there was a negative relationship between CI D and cellularity of the torn cuff. In the shoulders in which C/D on the Tl-weighted image was increased, cellularity of the torn cuff was diminished. With more advanced degeneration of the torn cuff, the signal intensity of the torn rotator cuff edge increased. The correlation coefficient between CIS on the T1-weighted im-
162
Nokogoki et 01.
J. ShoulderElbow Surg. May./June 7993
Figure 7 Histologic sections showing degenerative changes In rotator cutt tendon. A, There IS marked disorganization of tendon fibers representing Interstitial teanng (arrow). (Original magnification xl 00) B, There are many chondrocyte-like cells (arrow) in degenerative tendon. (Onginal magnification x 200) C, lipomatosIs (arrow) in degenerative tendon (OrigInal magnification x 100)
age and cellularity of the torn cuff was - 0.455 (0.002 < P < 0.02). CIS was not correlated with cellularity of the torn tendon. On the T2-weighted sequence, a distinct difference was seen between the signal intensity of the torn tendon edge and that of the tendon defect (0 < P < 0.002) (Figure 5, B), demonstrating that identification of the torn tendon edge was easy on the T2-weighted images. The correlation coefficient between C/D on the T2weighted images and cellularity of the torn cuff was -0.543 (0 < p < 0.002) (Figure 6, B). Thus a negative relationship was found between C/D and cellularity of the torn cuff. In the shoulders in which C/D on the T2-weighted image was increased, cellularity of the torn cuff was
diminished. With more advanced degeneration of the torn cuff, the signal intensity of the rotator cuff edge increased. The correlation coefficient between CIS on the T2-weighted image and cellularity of the torn cuff was - 0.289 (0.1 < p < 0.2). CIS was not correlated with cellularity of the torn tendon. On the T2*-weighted sequence a distinct difference was found between the signal intensity of the torn cuff edge and that of the defect (0 < P < 0.002) (Figure 5, q, demonstrating that identification of the torn cuff edge was easy on the T2*-weighted images. The correlation coefficient between C I D on the T2* -weighted image and cellularity of the torn cuff was -0.614 (0 < p < 0.002) (Figure 6, q. Thus a
J. Shoulder Elbow Surg. Volume 2, Number 3
negative relationship was found between C/O and cellularity of the torn cuff. In the shoulders in which C/O on the T2*-weighted image was increased, cellularity of the torn cuff was diminished. With advancing degeneration of the torn cuff, the signal intensity of the torn rotator cuff edge increased. The correlation coefficient between CIS on the T2*-weighted image and cellularity of the torn cuff was - 0.441 (0.002 < P < 0.02). CIS was not correlated with cellularity of the torn tendon. On the proton-density-weighted sequence, a distinct difference was found between the signal intensity of the torn cuff edge and that defect (0 < p < 0.002) (Figure 5, D), demonstrating that identification of the torn cuff edge was easy on the proton-density-weighted images. The correlation coefficient between C/O on the proton-density-weighted image and cellularity of the torn cuff tendon was - 0.638 (0 < p < 0.002) (Figure 6, D). Thus negative relationship was found between C/O and cellularity of the torn cuff. In the shoulders in which C/O on the proton-density-weighted image was increased, cellularity of the torn cuff was diminished. With advancing degeneration in the torn cuff, the signal intensity of the torn rotator cuff edge increased. The correlation coefficient between CIS on the proton-density-weighted image and cellularity of the torn tendon edge was - 0.345 (0.05 < P < 0.1 ). CIS was not correlated with cellularity of the torn tendon.
DISCUSSION Among previous studies on the relationships between MRI and degeneration of the rotator cuff, Kieft et al,2 reported that the signal intensity of the rotator cuff in patients with impingement syndrome was increased on both T1- and T2-weighted images because of tendon degeneration and inflammation. Ziatkin et al. 9 stated that the signal intensity of the rotator cuff in patients with rotator cuff tendinitis was increased in T1- and proton-density-weighted images, probably reflecting the presence of increased free water within the tendon secondary to edema and inflammation. Raffi et al. 5 reported that rotator cuff tendinitis or tendon degeneration could be diagnosed when the tendon showed an increased signal intensity on proton-density and T2-weighted images, and histologic evaluation of these tendons demon-
Nokogoki et 01.
163
strated disorganization of tendon fibers with fatty infiltration. Some researchers have studied degeneration of the rotator cuff histologically. Brewer! reported that histologic examination of degenerated rotator cuff demonstrated loss of cellularity, loss of staining, fragmentation of tendon fibers, and reduction of vascularity. Yamanaka 7 described the histologic changes in aged, degenerated rotator cuff tendons, citing a loss of the wavy appearance of the tendon fibers, loss of cellularity, hyperplasia of the intima of arterioles, and an increase of chondryocyte-like cells. In our study, C/O in the T1-, T2-, T2*-, and proton-density-weighted images showed a negative correlation with cellularity of the torn tendon edge. Thus an increase in signal intensity of the torn rotator cuff edge on the T1-, T2-, T2*-, and proton-density-weighted images was associated with an advance in degeneration. Histologic sections of these degenerated, torn tendons showed many interstitial tears (Figure 7, A}.5 Furthermore, so-called "fibrocartilaginous" transformation was found in those sections (Figure 7, B)/ and fatty infiltration was commonly seen (Figure 7, q.5 Thus it appears that the high-signal intensity of the degenerative, torn cuff on T2- and T2*-weighted sequences could be due to the synovial fluid in the interstitial cuff tear and an increase of chondrocyte-like cells. The high-signal intensity on T1- and proton-density-weighted sequences seems to be due to fatty infiltration. CIS on the T1-, T2-, T2*-, and proton-density-weighted images was not correlated with cellularity of the torn cuff. Thus it was not associated with degeneration. We believe this is true, because the signal intensity of the suprasupinatus muscle with a chronic rotator cuff tear is variegated by fatty infiltration because of the muscle atrophy. It has sometimes been possible to recognize high-signal linear bands within the supraspinatus muscle belly, indicative of fatty replacement, on MR images. 9 In such cases it is difficult to calculate the signal intensity of the supraspinatus muscle. So the signal intensity of the torn cuff edge should be compared with that of the deltoid muscle rather than with the supraspinatus muscle. However, identification of the torn cuff edge on the T1-weighted images was difficult, but it
164
Nokogoki et 01.
J. Shoulder Elbow Surg.
May / June 1993
was easy on the T2-, T2* - and proton-densityweighted images. Although the synovial fluid in the cuff defect was identified by a high-signal intensity on the T2-, T2* - and proton-densityweighted images, it did not show a high-signal intensity on the T1-weighted image. Thus, to determine the presence or absence of a rotator cuff, the signal intensities on the T2-, T2*-, and proton-density-weighted images, rather than on the T1-weighted image, should be studied. Once the torn tendon edge is identified, all types of images will exhibit an increase in the image intensity when there are degenerative changes in the tendon.
2 Kieft GJ, Bloem JL, Rozlng PM, Obermann WR Rotator cuff Impingement syndrome MR Imaging Radiology
1988;166'211-4
3 Lindblom K On pathogeneSIs of rupture ot the tendon aponeurosis of the shoulder 10lnt Acto Radlol
1939;20:563-77
4 Macnab I Rotator cuff tendinitis Ann Roy Coli ~uIY
1973;53:271-87
5 Raffi M, Flrooznla H, Sherman 0, Minkoff J, Weinreb :, Gollmbu C, et 01 Rotator cuff lesion' Signal patterns at MR Imaging. Radiology 1990;177 817-23 6. Reflor HJ, Krodel A, Melzer C Examination of the po thology of the rotator cuff Arch Orthop Trauma Surg
1987; 106;301-8
7. Yamanaka K Pathological study of the suprasprnatus tendon The Journal of the Japanese OrthopaediC AssoCiation 1988;62 1121-38 8 Wrlson CL Lesion of the supraspinatus tendon Archives of Surgery 1943;46'307-25
REFERENCES 1. Brewer BJ. Aging of the rotator cuff The American Journal of Sports Medicine 1979;7 102-10
9 Zlatkrn MB, Dahnka MK, Kressel HY Magnetic resonance Imaging of the shoulder. Magnetic Resonance Quarterly
1989,53-22
Bound Volumes Available to Subscribers Bound volumes of the 1993 issues of JOURNAL OF SHOULDER AND ELBOW SURGERY are available to subscribers (only) from the publisher, at a cost of $46.50 for domestic, $55.76 for Canadian, and $52.50 for international subscribers, for Vol. 1 (January-December); shipping charges are included. Each bound volume contains subject and author indexes, and all advertising is removed. Copies are shipped within 60 days after publication of the last issue in the volume. The binding is durable buckram, with the journal name, volume number, and year stamped in gold on the spine. Payment must accompany all orders. Contact Mosby, Subscription Services, 11830 Westline Industrial Dr., St. Louis, MO 63146-3318, USA; phone (314)453-4351 or (800)453-4124.
Subscriptions must be in force to qual Ny. Bound volumes are not available in place of a regular subscription.
We studied degeneration of the torn rotator cuff by comparing the signal intensity of the torn tendon edge on magnetic resonance imaging with cellularity of the tendon. Fifty-one shoulders with rotator cuff tears requiring surgical repairs were scanned before surgery in a O.5T magnetic resonance imaging system. T7 -, T2-, T2*-, and proton-density-weighted sequences were obtained. Three specimens from the torn tendon edge that were obtained at operation were examined microscopically, and the number of tendon cells was calculated and averaged. Degeneration of the torn rotator cuff edge was associated with an increase in signal intensity of the torn tendon edge on T7 -, T2-, T2*, and proton-density-weighted images and a decrease in tendon cellularity. The T2-, T2*-, and proton-density-weighted images were more useful, because on the T7 -weighted image it was difficult to identify the torn cuff edge. (J SHOULDER ELBOW SURG
7993;2: 756-64.)
It
is important to evaluate the degree of degeneration of a torn rotator cuff before surgery to accurately assess the prognosis after operative treatment.' Also, if we can evaluate the degree of degeneration of the torn rotator cuff and the size of the defect before surgery, we can estimate how much torn cuff edge we should excise and whether or not we will be able to repair the tear with simple sutures to the greater tuberosity. In the literature no reports have been made on the quantitative analysis of the correlation between degeneration of the torn rotator cuff edge and signal intensity of this remaining tendon on magnetic resonance imaging (MRI). In advanced degeneration of the rotator cuff, it has been shown that cellularity of the tendon is diminished. 1 , 7 In this study we compare the signal intensity of the torn rotator From the Department of Orthopaedic Surgery, Nara Medical University Reprint requests KlmlO Nakagakl, MD, Department of Orthopaedic Surgery, Nora Medical University, Kashihara, Nara, 634 Japan Copyright © 1993 by Journal of Shoulder and Elbow Surgery Board of Trustees 1058-2746/93/$100 + 10 32/1147339
156
cuff edge on various MR images with cellularity of the tendon and thus provide a noninvasive method to evaluate the degree of degeneration present in the torn rotator cuff.
MATERIAL AND METHODS Fifty-one shoulders in 50 patients with rotator cuff tears requiring surgical repair were scanned before surgery, and the tendon tear edge was examined histologically. In this study group, 11 women and 39 men had an age range from 35 to 84 years, with a mean age of 58.2 years. The mean preoperative period of nonoperative treatment was 9.6 months. These shoulders were scanned before surgery with a 20 cm circular surface coil with a 0.5 T unit (MRT-50A/2, Toshiba Medical System, Tokyo, Japan). Patients were examined in the supine position with the arm held in neutral rotation. Imaging was performed in the oblique coronal plane (scapular plane) parallel to the long axis of the supraspinatus muscle. The T1-weighted sequence (spin echo method TRITE = 450/20) and the T2*-weighted sequence (field echo method TRITE = 500/22, flip angle = 30°) were obtained in all shoulders (Figure 1, C).
J. Shoulder Elbow Surg. Volume 2, Number 3
Nakagaki et al.
157
Figure 1 MR imaging of full-thickness tear of supraspinatus tendon. 0,
Deltoid muscle; 5, suprasupinatus muscle. A, T1-weighted image (SE 450/20). identification of torn rotator cuff edge IS difficult. Signal intensity: T = 84.0, C = 106.0, D = 202.1, S = 223.4. B, T2-weighted image (SE 2000/80). Identification of torn rotator cuff edge is easier. (Signal intensity: T = 75.1, C = 536.0, D = 116.8, S = 296.8). C, T2*-weighted image (FE 500122 FA30). Identification of torn rotator cuff edge is even easier. (Signal intensity: T = 208.8, C = 685.0, D = 348.6, S = 529.9). D, Proton-densityweighted image (SE 2000/30). Identification of torn rotator cuff edge is much easier. (Signal intensity: T = 572.9, C = 1159.0, D = 857.6, S = 1099.6).
The T2-weighted sequence (spin echo method TRITE = 2000/80) was obtained in 36 shoulders (Figure 1, B), and the proton-densityweighted sequence (spin echo method TRITE = 2000/30) was obtained in 42 shoulders (Figure 1, 0). The thickness of the imaging section was 5 mm, with a 1 mm space between sections. The field of view was 250 mm, and the average number of signals was two. Matrix size was 256 x 512 in T1- and T2*-weighted sequences and was 160 x 512 in T2- and proton-densityweighted sequences. We were able to clearly identify the torn cuff
edge of all the patients on MRI and to diagnose the extent of rotator cuff tearing before surgery. Thirty-seven patients had a full-thickness tear of the rotator cuff, and 13 patients had a partialthickness tear. Of those with full-thickness tearing, three patients had a small tear (tear size <1 cm), 21 had a medium tear (1 cm ::5 tear size <3 em), seven had a large tear (3 cm ::5 tear size <5 cm), and seven had a massive tear (tear size :2:5 cm). One patient had a medium tear in the right shoulder and a massive tear in the left. Of the partial-thickness tears, seven patients had a bursal surface tear, and six had an ar-
158
J. Shoulder Elbow Surg.
Nakagaki et al.
May / June 1993
Calculation of signal intensity in MRI
Figure 2 Mean signal Intensity was calculated for each of the nine areas (0.86 mm x 1 0 mm) at torn rotator cuff edge on MRI.
Spots of measurement of signal intensity an MRI torn cuff edge cuff defect,,
'vr/~~~
5iiiiiiii
,.----0.----- supraspinatus muscle
deltoid ---H-lK],tI muscle
Figure 3 Locations far measurement of signal Intensity on MRI
ticular side tear. At surgery, we found that the preoperative diagnoses of the types of rotator cuff tearing were correct. We analyzed the mean signal intensity of
nine areas (0.86 mm x 1.0 mm) of the torn rotator cuff edge using MR imaging (Figure 2), and we regarded the value as the signal intensity of the torn cuff edge. When a torn edge
J. Shoulder Elbow Surg. Volume 2, Number 3
Nokogoki et 01.
159
Figure 4 A, Histologic specimen of mildly degenerative torn cuff. Many tenocytes and distinct wavy appearance are visible (arrow). (Original magnification x 100) B, Histologic specimen of severely degenerative torn cuff. It shows loss of cellularity and absence of wavy appearance. (Original magnification x 100) was thin, we obtained the mean signal intensity in the slightly more proximal thick cuff area. In the same manner, we calculated the signal intensities of the middle points of the deltoid muscle, the supraspinatus muscle, and the area of the cuff defect (Figure 3). We determined the ratia of the signal intensity of the torn cuff (C) to that of the deltoid muscle (D) and similarly determined the ratio of the signal intensity of the torn cuff to that of the suprasupinatus muscle
(5).
Histologic specimens were obtained at surgery from the torn tendon edge. When the torn cuff edge was thin, we obtained the spp:.imen slightly more proximal to have a sample of about 3 mm in thickness. The three specimens from each shoulder were examined microscopically with a magnification of x 400, and the number of tendon cells in each part was calculated and averaged. The average was regarded as the cellularity of the torn tendon edge.
160
Nokogoki et 01.
1. Shoulder Elbow Surg. May / June 7993
Signal Intensity on T1-weighted Image
Signal Inlensity on T2-weighted Image
Xl00
7
Ttest 0.1
6
4 T
2
~~~~:::=~T
I I
I I
0
9
.
8 7
~=.,---=:::::::;=e:
I
I
~
I
I
.L
.L
~,
4
I
3 2
I
1
0 Tom Cuff Edge
Cuff Defect Portion
0 ........- Torn Cuff Edge
B
_1
,, ,,
T
6 5
I
f
•I
A
Tlest
11 10
5
3
Xl00 12
Cuff Defect Portion
Signal Intensity on Proton-density-weighted Image X1000
Ttest
3
Ttes!
0
O
2
Cuff Defect Portion
Tom Cuff Edge
Cuff Defect Portion
Figure 5 Statistical testing assessing difference between Signal Intensity of torn rotator cuff edge and that of tendon defect on various types of MR imaging. A, Statistical testing assessing difference between signal intenSity of torn rotator cuff edge and that of defect on Tl-weighted image. 8, Statistical testing assessing difference between signal intensity of torn rotator cuff edge and that of defect on T2-weighted image. C, Statistical testing assessing difference between signal intensity of torn rotator cuff edge and that of defect on T2*-weighted (field echo) image. D, Statistical testing assessing difference between signal intensity of torn rotator cuff edge and that of defect on proton-density-weighted image
Histologic examination of the torn rotator cuff with mild degeneration showed many tenocytes and a distinct wavy appearance. The tendon bundles stained deep pink with hematoxylineosin (Figure 4, A). However, more severely degenerated tendon edges showed loss of cellularity and absence of the wavy appearance. The tendon bundles become homogeneous in texture and stained weakly (Figure 4, 8).3· 4. 8 The following studies were performed for interpretation of the MRI examination. We ex-
amined the difference between the signal intensity of the torn cuff edge and that of the cuff defect in each MRI sequence with the paired ttest. The level of significance was set at 0.05. We then assessed which sequence most clearly confirmed the presence of the torn cuff edge. We calculated the Pearson's correlation coefficient between C/D, CIS, and cellularity of the torn cuff. When the absolute value of the correlation coefficient was larger than 0.5, we recognized a significant difference. Using this information, we then determined the relationship
J. Shoulder Elbow Surg. Volume 2, Number 3
Nokogoki et
X,OO
,
CelIu1anty and
00 R--o
celluilriy and CIS
X'O
630 (il
R--o 455 (0 002<1'<0 02)
"""
20
Cellu1l1rl1y and CID Ao=-o.543 (il
,
'9
'8
;00 ••
17 '6
\ 0
~~
0
13 ... a...... 12 11
........ \
... "'
t~~ - • •
\,.
6
\
S
0
-'
•
,
3
'
0
':'
,
2
o
..
~o
0° 0
4
------
"""'"",,--,
\
'
o
.'001
B
A
CelIu1anty of Torn Cuff Edge
CelIu1anty lWld CIO : R--o.614 (0
,
\
\
Cellulanty ard CIS . A--O 441 (O.OO2<:p
,,
\
2
'
0:00
3
2
4
on T2·welghled Image
Cellularrty of Torn Cuff Edge
Xl00 \
CelluilVty and CIS : R=-o 289 (0 1<:p
,
\°0
O' I·
0
~
••
i
0
0
'~
3 \
161
CeIJlJanty of Tom Cuff Edge
CellUanty of Tom Cuff Edge
3
01.
X'O 20 19 18 17 18 15
\
CelIu1arty lWld CIO . Rz-o.638 (0
't
eea.Any lWld CIS A--o.345 (O.05
14
'3
12
"'0 9
8
7 6
o.
5
4 3 2 1
1
c
• ; CIO 0;00
I
2 on FI8ld Echo Image
0-4------_----"---....--------';:..,.
•:00 I
0 1 2
o
0;00
on
Proton~...".;ghIed
Image
Figure 6 Correlation between the signal intensity of the torn rotator cuff divided by the signal intensity of the deltoid (C/O) or the signal intensity of the torn cuff divided by the signal intensity of the supraspinatus muscle (CIS) and the cellularity of the torn rotator cuff. Solid lines are regression lines between CIO and cellularity of torn cuff. Dotted lines are regression lines between CIS and cellularity of torn cuff. A, Correlation between CIO, CIS on Tl-weighted image and cellularity of torn rotator cuff. B, Correlation between CIO, CIS on T2-weighted image and cellularity of torn rotator cuff. C, Correlation between CIO, CIS on T2*-weighted (field echo) image and cellularity of torn rotator cuff. D, Correlation between CIO, CIS on proton-density-weighted image and cellularity of torn rotator cuff. between the signal intensity of the torn cuff on each sequence of MRI and the pathologic degeneration of the torn cuff.
RESULTS On the T1-weighted sequence no clear difference was found between the signal intensity of the torn cuff edge and that of the tendon defect (0.1 < p < 0.2) (Figure 5, A), demonstrating that identification of the torn cuff edge was very difficult on the T1-weighted image. The
correlation coefficient between CI D on the T1weighted image and cellularity of the torn cuff was -0.630 (0 < p < 0.002) (Figure 6, A). Thus there was a negative relationship between CI D and cellularity of the torn cuff. In the shoulders in which C/D on the Tl-weighted image was increased, cellularity of the torn cuff was diminished. With more advanced degeneration of the torn cuff, the signal intensity of the torn rotator cuff edge increased. The correlation coefficient between CIS on the T1-weighted im-
162
Nokogoki et 01.
J. ShoulderElbow Surg. May./June 7993
Figure 7 Histologic sections showing degenerative changes In rotator cutt tendon. A, There IS marked disorganization of tendon fibers representing Interstitial teanng (arrow). (Original magnification xl 00) B, There are many chondrocyte-like cells (arrow) in degenerative tendon. (Onginal magnification x 200) C, lipomatosIs (arrow) in degenerative tendon (OrigInal magnification x 100)
age and cellularity of the torn cuff was - 0.455 (0.002 < P < 0.02). CIS was not correlated with cellularity of the torn tendon. On the T2-weighted sequence, a distinct difference was seen between the signal intensity of the torn tendon edge and that of the tendon defect (0 < P < 0.002) (Figure 5, B), demonstrating that identification of the torn tendon edge was easy on the T2-weighted images. The correlation coefficient between C/D on the T2weighted images and cellularity of the torn cuff was -0.543 (0 < p < 0.002) (Figure 6, B). Thus a negative relationship was found between C/D and cellularity of the torn cuff. In the shoulders in which C/D on the T2-weighted image was increased, cellularity of the torn cuff was
diminished. With more advanced degeneration of the torn cuff, the signal intensity of the rotator cuff edge increased. The correlation coefficient between CIS on the T2-weighted image and cellularity of the torn cuff was - 0.289 (0.1 < p < 0.2). CIS was not correlated with cellularity of the torn tendon. On the T2*-weighted sequence a distinct difference was found between the signal intensity of the torn cuff edge and that of the defect (0 < P < 0.002) (Figure 5, q, demonstrating that identification of the torn cuff edge was easy on the T2*-weighted images. The correlation coefficient between C I D on the T2* -weighted image and cellularity of the torn cuff was -0.614 (0 < p < 0.002) (Figure 6, q. Thus a
J. Shoulder Elbow Surg. Volume 2, Number 3
negative relationship was found between C/O and cellularity of the torn cuff. In the shoulders in which C/O on the T2*-weighted image was increased, cellularity of the torn cuff was diminished. With advancing degeneration of the torn cuff, the signal intensity of the torn rotator cuff edge increased. The correlation coefficient between CIS on the T2*-weighted image and cellularity of the torn cuff was - 0.441 (0.002 < P < 0.02). CIS was not correlated with cellularity of the torn tendon. On the proton-density-weighted sequence, a distinct difference was found between the signal intensity of the torn cuff edge and that defect (0 < p < 0.002) (Figure 5, D), demonstrating that identification of the torn cuff edge was easy on the proton-density-weighted images. The correlation coefficient between C/O on the proton-density-weighted image and cellularity of the torn cuff tendon was - 0.638 (0 < p < 0.002) (Figure 6, D). Thus negative relationship was found between C/O and cellularity of the torn cuff. In the shoulders in which C/O on the proton-density-weighted image was increased, cellularity of the torn cuff was diminished. With advancing degeneration in the torn cuff, the signal intensity of the torn rotator cuff edge increased. The correlation coefficient between CIS on the proton-density-weighted image and cellularity of the torn tendon edge was - 0.345 (0.05 < P < 0.1 ). CIS was not correlated with cellularity of the torn tendon.
DISCUSSION Among previous studies on the relationships between MRI and degeneration of the rotator cuff, Kieft et al,2 reported that the signal intensity of the rotator cuff in patients with impingement syndrome was increased on both T1- and T2-weighted images because of tendon degeneration and inflammation. Ziatkin et al. 9 stated that the signal intensity of the rotator cuff in patients with rotator cuff tendinitis was increased in T1- and proton-density-weighted images, probably reflecting the presence of increased free water within the tendon secondary to edema and inflammation. Raffi et al. 5 reported that rotator cuff tendinitis or tendon degeneration could be diagnosed when the tendon showed an increased signal intensity on proton-density and T2-weighted images, and histologic evaluation of these tendons demon-
Nokogoki et 01.
163
strated disorganization of tendon fibers with fatty infiltration. Some researchers have studied degeneration of the rotator cuff histologically. Brewer! reported that histologic examination of degenerated rotator cuff demonstrated loss of cellularity, loss of staining, fragmentation of tendon fibers, and reduction of vascularity. Yamanaka 7 described the histologic changes in aged, degenerated rotator cuff tendons, citing a loss of the wavy appearance of the tendon fibers, loss of cellularity, hyperplasia of the intima of arterioles, and an increase of chondryocyte-like cells. In our study, C/O in the T1-, T2-, T2*-, and proton-density-weighted images showed a negative correlation with cellularity of the torn tendon edge. Thus an increase in signal intensity of the torn rotator cuff edge on the T1-, T2-, T2*-, and proton-density-weighted images was associated with an advance in degeneration. Histologic sections of these degenerated, torn tendons showed many interstitial tears (Figure 7, A}.5 Furthermore, so-called "fibrocartilaginous" transformation was found in those sections (Figure 7, B)/ and fatty infiltration was commonly seen (Figure 7, q.5 Thus it appears that the high-signal intensity of the degenerative, torn cuff on T2- and T2*-weighted sequences could be due to the synovial fluid in the interstitial cuff tear and an increase of chondrocyte-like cells. The high-signal intensity on T1- and proton-density-weighted sequences seems to be due to fatty infiltration. CIS on the T1-, T2-, T2*-, and proton-density-weighted images was not correlated with cellularity of the torn cuff. Thus it was not associated with degeneration. We believe this is true, because the signal intensity of the suprasupinatus muscle with a chronic rotator cuff tear is variegated by fatty infiltration because of the muscle atrophy. It has sometimes been possible to recognize high-signal linear bands within the supraspinatus muscle belly, indicative of fatty replacement, on MR images. 9 In such cases it is difficult to calculate the signal intensity of the supraspinatus muscle. So the signal intensity of the torn cuff edge should be compared with that of the deltoid muscle rather than with the supraspinatus muscle. However, identification of the torn cuff edge on the T1-weighted images was difficult, but it
164
Nokogoki et 01.
J. Shoulder Elbow Surg.
May / June 1993
was easy on the T2-, T2* - and proton-densityweighted images. Although the synovial fluid in the cuff defect was identified by a high-signal intensity on the T2-, T2* - and proton-densityweighted images, it did not show a high-signal intensity on the T1-weighted image. Thus, to determine the presence or absence of a rotator cuff, the signal intensities on the T2-, T2*-, and proton-density-weighted images, rather than on the T1-weighted image, should be studied. Once the torn tendon edge is identified, all types of images will exhibit an increase in the image intensity when there are degenerative changes in the tendon.
2 Kieft GJ, Bloem JL, Rozlng PM, Obermann WR Rotator cuff Impingement syndrome MR Imaging Radiology
1988;166'211-4
3 Lindblom K On pathogeneSIs of rupture ot the tendon aponeurosis of the shoulder 10lnt Acto Radlol
1939;20:563-77
4 Macnab I Rotator cuff tendinitis Ann Roy Coli ~uIY
1973;53:271-87
5 Raffi M, Flrooznla H, Sherman 0, Minkoff J, Weinreb :, Gollmbu C, et 01 Rotator cuff lesion' Signal patterns at MR Imaging. Radiology 1990;177 817-23 6. Reflor HJ, Krodel A, Melzer C Examination of the po thology of the rotator cuff Arch Orthop Trauma Surg
1987; 106;301-8
7. Yamanaka K Pathological study of the suprasprnatus tendon The Journal of the Japanese OrthopaediC AssoCiation 1988;62 1121-38 8 Wrlson CL Lesion of the supraspinatus tendon Archives of Surgery 1943;46'307-25
REFERENCES 1. Brewer BJ. Aging of the rotator cuff The American Journal of Sports Medicine 1979;7 102-10
9 Zlatkrn MB, Dahnka MK, Kressel HY Magnetic resonance Imaging of the shoulder. Magnetic Resonance Quarterly
1989,53-22
Bound Volumes Available to Subscribers Bound volumes of the 1993 issues of JOURNAL OF SHOULDER AND ELBOW SURGERY are available to subscribers (only) from the publisher, at a cost of $46.50 for domestic, $55.76 for Canadian, and $52.50 for international subscribers, for Vol. 1 (January-December); shipping charges are included. Each bound volume contains subject and author indexes, and all advertising is removed. Copies are shipped within 60 days after publication of the last issue in the volume. The binding is durable buckram, with the journal name, volume number, and year stamped in gold on the spine. Payment must accompany all orders. Contact Mosby, Subscription Services, 11830 Westline Industrial Dr., St. Louis, MO 63146-3318, USA; phone (314)453-4351 or (800)453-4124.
Subscriptions must be in force to qual Ny. Bound volumes are not available in place of a regular subscription.