Annals of the Rheumatic Diseases 1990; 49: 7-11

SCIENTIFIC PAPERS

Magnetic resonance imaging of the shoulder inpatients with rheumatoid arthritis

G J Kieft, B A C Dijkmans, J L Bloem, H M Kroon

AbstractTo evaluate the ability of magnetic resonanceimaging (MRI) to detect shoulder abnor-malities 18 patients (36 shoulders) with rheu-matoid arthritis (RA) and shoulder complaintswere studied. Osseous abnormalities of theglenoid and humeral head were readilydetected with MRI. The imaging planes usedwere not suitable for the evaluation ofacromioclavicular joint involvement. Mag-netic resonance imaging depicted soft tissueabnormalities that were not clearly visualisedby plain film radiography, such as involve-ment of rotator cuff tendons and subacromialbursae, joint effusion, and muscular atrophy.Magnetic resonance imaging appears to be asensitive method for evaluation of gleno-humeral joint changes in patients with RA.

Department ofDiagnostic Radiology,University Hospital,Leiden, The NetherlandsG J KieftJ L BloemH M KroonDepartment ofRheumatology,University Hospital,Leiden, The NetherlandsB A C DijkmansCorrespondence to:Department of DiagnosticRadiology, Building 1:C2-S, PO Box 9600,2300 RC Leiden,The Netherlands.Accepted for publication3 March 1989

The shoulder is one of the more common sitesof involvement of rheumatoid arthritis (RA).Rheumatoid arthritis not only affects the syno-vium within the glenohumeral joint but can alsoinvolve the distal third of the clavicle, bursae,rotator cuff tendons, and surroundingmuscles.' 2

Clinically, shoulder arthritis can often gounrecognised. There are several reasons forthis: the onset of the disease in the shoulder isoften insidious, the shoulder joint is deeplyseated, and synovial tissue swelling often escapesnotice as it is difficult to detect unless suspected.It is important to recognise shoulder involve-ment early and if possible, to prevent furtherdamage by conservative treatment-forexample, by careful physiotherapy with activeand passive movements, before irreversiblechanges have occurred.

Radiography can be helpful in determiningthe extent of shoulder involvement in RA.Radiographic assessment of the stage of thedisease is based on the presence of boneerosions, joint space narrowing, and osteo-porosis. Joint erosions, not specifically theshoulder, usually occur within the first twoyears of the disease.3 Joint space loss andosteoporosis are not specific for RA.3 4

Magnetic resonance imaging(MRI) has provedits usefulness in the evaluation of the musculo-skeletal system.5 6 Owing to its superior softtissue contrast, tomographic nature, and multi-planar imaging, MRI is particularly valuable forassessment of the synovium, cartilage, muscles,tendons, and ligaments.7 8

This study was undertaken to determinewhether MRI can provide more informationabout shoulder involvement in patients with RA

than physical examination combined with plainradiographs.

Patients and methodsPATIENTSEighteen consecutive patients with definite orclassical RA, according to the American Rheu-matism Association criteria,9 who had visitedthe outpatient clinic of the department ofrheumatology and had shoulder complaintswere included in the study. All patients gavetheir informed consent to participation in thestudy. The table gives the clinical details ofthese patients. This group included sevenfemale and 11 male patients with median age 63years (range 42-79) and median duration ofdisease five years (range 1-23); one patient wasseronegative and 17 were seropositive. Accord-ing to x rays of the wrists, hands, and forefeetfive patients had non-erosive and 13 had erosivedisease. Fifteen of the 18 patients had been orwere being treated with disease modifyingdrugs. The motion of 30 shoulders was restrictedfor anteflexion or abduction or both (table).

METHODSPlain radiographs were taken in the antero-posterior (arm in exorotation) and axial views.The radiographs were evaluated in concert bytwo of the authors (BACD, HMK), one radio-logist, and one rheumatologist without know-ledge of the results of the MRI studies. Involve-ment of the glenohumeral joint, major tubero-sity, and acromioclavicular joint was graded bythe scoring method described by Kellgren'0:grade 0 no abnormalities, grade 1 dubiouserosions, grade 2 definite but mild erosions,grade 3 more destructive changes, and grade 4severe erosions.

Magnetic resonance imaging studies of theshoulders were performed with a 0 5 teslasuperconductive magnet MRI system (GyroscanS5 Philips, Best, The Netherlands). A flexible'wrap around' surface coil was used. Multipleimages in the transverse and oblique planes-that is, perpendicular to the glenoid surface,were obtained. The oblique plane has beendescribed in detail.8 Spin echo pulse sequenceswith relative TI weighting (600, 30) (repetitiontime ms, echo time ms) and relative T2 weight-ing by the dual echo technique (2000, 50-100)were used in all examinations. To keep thescanning time within acceptable limits twoexcitations were used in all cases. The totalscanning time for both shoulders was less thanone hour. A 250 mm field of view and 5 mm

7

on August 26, 2019 by guest. P

rotected by copyright.http://ard.bm

j.com/

Ann R

heum D

is: first published as 10.1136/ard.49.1.7 on 1 January 1990. Dow

nloaded from

Kieit, Dijkmans, Bloem, Kroon

Clinical, radiographic and magnetic resonance imaging (MRI) findings in 18 patients with rheumatoid arthritis

Patient numberSexAge (years)Duration of RA (years)Seropositive/ negativeErosive/non-erosiveAnteflexion (degrees)Abduction (degrees)'Glenohumeral involvement

RadiographyMRI

Acromioclavicular joint involvement -

RadiographsMRI

Minor and major tubercle involvement tRadiographvMRI

Cartilage destructionRadiographvMRI

Soft tissue abnormality detected with MRIR.Rotator cuff tendon destructionRotator cuff muscle atrophyEffusion or pannusLvmph nodes

F77

1

E170 17090/90

2M6319+

160 17090 90

3M7512

NE160, 1708090

4M532

E90 11070 40

M.Ni79

l

NE180 16090 80

6M5712

E170 17080 80

7

14180 1809() 90

8.Ni76

I

NE180 18090 90

9

'N422

1'17(0 17080 80

10

7-

E170 16090 90

00 00 00 33 00 00 (() 00 20 02 2I02 02 00 42 0(0 20 0(0 12 2 0 22

1 2 00 00 2 3 00 1 0 00 0 0 0 00 (0220 20 00 10 30 00 (10 3 1( 0(1 (10

00 1(0 10 4A4 0(0 1 1 2 0 I I 0 2 232 20 20 44 22 22 22 12 32 222

_+ + t--_~~~~ ~ ~~~~~~~_ _ _ _ ++-_ _-_ + +

*Values given for right shoulder,left shoulder.tGrade 0=no abnormalities; grade 1=dubious erosions; grade 2=definite but mild erosions; grade 3=more destructive changes; grade 4 severe erosiolls.TOwing to osteosvnthetic material this area could not be evaluated in this patient.%-absent; +=present; + + severe.

thick contiguous sections were chosen incombination with a 154 x 256 acquisition matrixand a 512x 512 display matrix.

Figure 1 shows the magnetic resonanceimages of the shoulder of a normal volunteer.Patient positioning proved simple. A comfort-able neutral humeral position with the patientsupine was chosen. The images were interpretedby two of the authors (GJK, JLB), bothradiologists, in concert without knowledge ofthe clinical history or radiographic findings.The findings for 110 subjects without RA, whohad undergone MRI of the shoulders for avariety of other reasons, as well as the findingsfor 10 normal volunteers and two cadavericshoulders served as controls.8 " Magneticresonance images were evaluated by the scoringsystem used for the radiographs. Bone erosionswere defined as marginal osseous defects with alow or non-homogeneous signal intensity on T1weighted images. 12 Joint effusion was diag-nosed when the joint capsule was distended bymaterial that was identified by its low signalintensity on T1I weighted images and high signalintensity on T2 weighted images. 4 Normal

synovial fluid is not visible on magnetic reson-ance images; when it is abnormal (pannusformation) it has a low signal intensity on T1and low or high signal intensity on T2 weightedimages."" Normal hyaline articular cartilagehas an intermediate signal intensity on TI andT2 weighted images. Cartilage disease wasdiagnosed when the articular cartilage hadthinned focally or diffusely.

In a final session the radiographs and mag-netic resonance images were reviewed togetherby two of the authors (GJK, BACD) to establishwhether or not the magnetic resonance imagesprovided additional information.

ResultsMagnetic resonance images of adequate tech-nical quality were obtained in all cases. Ametallic artefact was present in one shoulder.and thus the region of the tuberosities and thelateral part of the rotator cuff tendons in thisshoulder could not be evaluated. Other artefactswere minimal; if present they were attributableto motion.

Figure 1: Magnetic resonance images ofa normal left shoulder (37year old healthy man). (a) Ti weighted (repetition time (TR) 600 ms, echo time (TE) 30 ms)image ofthe supraspinatus tendon and muscle taken in the oblique plane; (b) T2 weighted (TR 2000, TE 100) image in the same plane; the relative intensitiesofthe normal musculoskeletal tissues remain the same; note the normal subdeltoid bursa (arrows); (c) TI weighted (TR 600, TE 30) transverse image at thelevel ofthe coracoid process. I =humeral head; 2= acromion; 3=glenoid cavity; 4=coracoid process; 5=supraspinatus muscle; 6=supraspinatus tendon;7=infraspinatus muscle; 8=subscapular muscle; 9= deltoid muscle; 10= trapezius muscle; II= acromioclavicular joint; 12= tendon of the long head of bicepsmuscle.

8

-t -t

+ -rt- t

on August 26, 2019 by guest. P

rotected by copyright.http://ard.bm

j.com/

Ann R

heum D

is: first published as 10.1136/ard.49.1.7 on 1 January 1990. Dow

nloaded from

Magnetic resonance imaging ofthe shoulder in RA

In five cases MRI was unable to detect erosive12 13 14 15 16 17 18 changes in the acromioclavicular joint that wereF M F F M F M clearly visible on the plain radiographs. On the73 52 69 66 66 73 S5 other hand, MRI showed abnormalities in six23 12 2 13 4 1 9+ + + + + + + shoulders which were not seen with plain film160/160 170/170 160/150 160/180 45/50 50/150 90J90 radiography.90/90 70/70 80/80 80/90 45/30 20/80 70/90 Joint space narrowing at the glenohumeral0/0 4/2 0/0 0/0 3/2 3/0 1/1 joint was diagnosed in eight joints by plain film0/0 4/2 0/0 4/0 4/3 2/2 0/2 radiography. The magnetic resonance images0/0 0/0 2/2 1/2 1/1 2/0 0/0 indicated that cartilage destruction had probably0/0 0/0 210 2/0 410 2/2 0/1 occurred in seven of these eight joints and in2/2 34t 1/1 2/2 4/3 3/2 210 three additional joints. The hyaline articular2/2 44t 0/2 212 4/3 212 2/2 cartilage showed no signal intensity changes

with disease.-/_ +/+ _,_ _X_ +/+ +1 +/ The soft tissue abnormalities detected by

MRI included rotator cuff tendon involvement_/++ +ft +/+ /, ++/++ +i+ +1+++/±+ +/ +/+ ++/++ ++/++ ++/+ +/+ in 11 and complete tendon destruction in five+/+ - +/+ +++/-/++ ++ ++/++ shoulders (table). Mild atrophic changes in the

rotator cuff muscles were detected in fiveshoulders; in nine other cases the cuff muscles

The table lists the osseous abnormalities were severely atrophied. Figures 3 and 4 showdetected with plain film radiography and MRI. examples of soft tissue abnormalities.All erosions visualised by plain film radiography, Magnetic resonance images showed jointboth at the tuberosities and in the glenohumeral effusion or inflammatory synovial tissue in thejoint, were easily seen on the magnetic resonance glenohumeral joint or the subacromial bursa, orimages. Glenohumeral joint involvement of both, in 22 joints.grade 2 or more was shown in 10 shoulders by Signal intensity of inflammatory synovialplain film radiography and in 19 shoulders by tissue in or adjacent to osseous erosions variedMRI. Abnormalities of grade 3 or more at the on T2 weighted images: eight times it producedtuberosities were detected in 18 shoulders with a low signal, 15 times a high signal, and threeradiography and in 31 shoulders with MRI. times a mixed signal intensity (fig 4). EnlargedThere were no significant abnormalities on the axillary lymph nodes (>1 cm) were detected inplain film radiographs that were not observed 14 shoulders.with MRI. Figure 2 shows an example of these No systemic correlation between limitation ofosseous abnormalities. joint motion and radiological findings could be

Figure 2: Small erosionsand joint effuston in theright shoulder of patientNo 9. (a and b)Anteroposterior and axialradiographs show a smallerosion at the greatertuberosity; (c) oblique TIweighted (repetition time(TR) 600 ms, echo time(TE) 30 ms) magneticresonance image also showsthe erosion (arrows) and inaddition, an intactsupraspinatus muscle; 40W(d) transverse Ti wieightedmagnetic resonance image(TR 600, TE 30) visuali'sestwo erosions in the humeral _____,_head (arrows); (e) transverse (C)T2 weighted image(TR 2000, TE 100). Ajoint effusion appears as anarea of high signal intensity rIt_in both the joint space andthe sheath of the bicepstendon (arrows).

9

on August 26, 2019 by guest. P

rotected by copyright.http://ard.bm

j.com/

Ann R

heum D

is: first published as 10.1136/ard.49.1.7 on 1 January 1990. Dow

nloaded from

Kieft, Dijkmans, Bloem, Kroon

Figure 3: Destructive changes in the right shouldet ofpatient No 4. (a,) Anteroposterorroladiographshows severe erosions and elevation of the humeral head; (b) transverse Ti weighted (repetition time(TR) 600 ms, echo time (TE) 30 ms) image shows complete destruction of the humeral head by theinflammatory process (arrows); infraspinatus and subcapsular muscles are severely atrophic (curvedarrows); (c) oblique TI weighted (TR 600, TE 30) image shows degeneration of the supraspinatusmuscle (black arrows), absence of its tendon, and a subdeltoid bursa filled with pannus (white arrows).

Figure 4: Changes in the signal intensity of inflammatory tissue within osseous erosions and hypertrophied synovium on Ti andT2 weighted images of the right shoulder ofpatient No 16. (a) TI weighted (repetition time (TR) 600 ms, echo time (TE)30 ms) image shows pannus in the glenohumeral joint and the subacromial bursa as well as the erosion which appear as areas oflow signal intensity (arrows). Note also severe atrophy of the supraspinatus muscle. (b) T2 weighted (TR 2000, TE 100)image: pannus in the joint and the erosion are now seen as areas of high signal intensity (arrows).

found. In general, the radiographs and MRI ofseverely impaired shoulders showed moresevere abnormalities-for example, patientsNos 4 and 16; shoulders with little or noimpairment proved to be severely involvedradiologically-for example, patients Nos 11,12, and 13.

DiscussionThe main conclusion of our study is that MRI ofthe shoulder can provide information aboutosseous and soft tissue involvement of the

glenohumeral joint in patients with RA; more-over, MRI gives more information than plainfilm radiography. It has to be noted, however,that the results of two radiological procedureswere compared without having a gold standard.

Soft tissue swelling was easily detected withMRI. A joint effusion has a low signal intensityon TI and a high signal intensity on T2weighted images. Osteoporosis itself cannot bevisualised with MRI, but the processes that leadto osteoporosis-synovial inflammation andpannus formation-can be detected. Pannusformation appeared as areas of low signal

10

on August 26, 2019 by guest. P

rotected by copyright.http://ard.bm

j.com/

Ann R

heum D

is: first published as 10.1136/ard.49.1.7 on 1 January 1990. Dow

nloaded from

Magnetic resonance imaging ofthe shoulder in RA

intensity on TI weighted images and as areas oflow or high signal intensity on T2 weightedimages. This variation in signal intensity on T2weighted images can be related to the differencebetween acute and chronic inflammation" or tohaemosiderin deposits within the hypertrophiedsynovium.12 16 There are similarities betweenpannus and fluid on T2 weighted images. Whenboth fluid and pannus had a high signalintensity on T2 weighted images we found itpossible to differentiate these two tissues withextreme T2 weighting. In such cases the fluidhas a higher signal than the synovial tissue.

In our series bone erosions could be bettershown with MRI than with plain radiography,presumably owing to the tomographic natureand multiplanar ability of MRI. Hyaline articu-lar cartilage can be visualised directly with MRIand diffuse destruction as well as focal damagecan be depicted. In one case, however, MRIdid not show cartilage abnormalities, whereasthe plain films showed an obvious loss of jointspace. This fault was probably due to a thinlayer of fluid within the joint space. In addition,MRI provided information about the integrityof the rotator cuff tendons.

Atrophic changes in the rotator cuff musclescould also be detected. Involvement of theacromioclavicular joint was not visualised inseveral cases. Because our image plane isperpendicular to the glenoid it passes throughthe acromioclavicular joint at an angle, resultingin suboptimal visualisation of the joint.8 Forthis reason other planes should be chosen forevaluation of the acromioclavicular joint.As a result of this study we have become

aware that damage to the tissues of the shoulderjoint is already present even when the radio-graphs are still quite normal. The correlationbetween clinical findings and tissue damage asdetected by MRI and radiography is quite poor.Other conditions, such as osteoarthritis andchondrocalcinosis, may produce cysts or defectsin trabecular bone similar to those in RA.Future investigations will be necessary to see ifwe are able to differentiate between thesedisease processes.

In conclusion, MRI is able to assess softtissue and bone involvement of the gleno-humeral joint in RA that is difficult to evaluateby other means. Other joints of the appendicularskeleton can also be evaluated; moreover, MRIis non-invasive and painless and there are nobiological risks involved. In view of its repro-ducibility MRI could become an objectivemethod for observing the stages of the diseaseand eventually for monitoring response to treat-ment.

1 Resnick D, Niwayama G, eds. Diagnosis of bone and jointdisorders. Vol 2. Philadelphia: Saunders, 1981: 945-9.

2 Harris E D. Rheurmatoid arthritis: the clinical spectrum. In:Kelley W N, Harris E D, Ruddy S, Sledge C B, eds.Textbook of rheumatology. Philadelphia: Saunders, 1981:928-63.

3 Brook A, Corbett M. Radiographic changes in early rheuma-toid disease. Ann Rheum Dis 1977; 36: 71-3.

4 Berens D L, Lockie L M, Lin R, Norcross B M. Rontgenchanges in early rheumatoid arthritis. Radiology 1964; 82:645-54.

5 Berquist T H, Ehman R L, Richardson M L, eds. Magneticresonance imaging of the musculoskeletal system. New York:Raven Press, 1987.

6 Burt C T, Koutcher J, Roberts J T, London R E, Chance B.Magnetic resonance imaging of the musculoskeletal system.Radiol Clin North Am 1986; 24: 321-31.

7 Beltran J, Noto A M, Herman L J, Lubbers L M. Tendons:high field-strength, surface coil MR imaging. Radiology1987; 162: 735-40.

8 Kieft G J, Bloem J L, ObermannW R, Verbout A J, RozingP M, Doornbos J. Normal shoulder: MR imaging. Radio-logy 1986; 159: 741-5.

9 Ropes M W, Bennett G A, Cobb S, Jacox R, Jessar R A.Revision of diagnostic criteria for rheumatoid arthritis. BullRheum Dis 1958; 9: 175-6.

10 Kellgren J H, Jeffrey M R, Ball Y, eds. The epidemiology ofchronic rheunatism. Oxford: Blackwell Scientific, 1963.

11 Kieft G J, Sartoris D J, Bloem J L, et al. Magnetic resonanceimaging of glenohumeral joint diseases. Skeletal Radiol1987;16: 285-90.

12 Beltran J, Caudill L, Herman L A, et al. Rheumatoidarthritis: MR imaging manifestations. Radiology 1987; 165:153-7.

13 Yulish B S, Lieberman J M, Newman A J, Bryan P J,Mulopulos G P, Modic M T. Juvenile rheumatoid arthritis:assessment with MR imaging. Radiology 1987; 165: 149-52.

14 Beltran J, Noto A M, Herman L J, Mosure J C, Burk J M,Christoforidis A J. Joint effusions: MR imaging. Radiology1986; 158: 133-7.

15 Terrier F, Hricack H, Revel D, et al. Magnetic resonanceimaging and spectroscopy of the periarticular inflammatorysoft-tissue changes in experimental arthritis of the rat. InvestRadiol 1985; 20: 813-22.

16 Muirden K D, Senator G B. Iron in the synovial membrane inrheumatoid arthritis and other joint diseases. Ann RheumDis 1968; 27: 3848.

11

on August 26, 2019 by guest. P

rotected by copyright.http://ard.bm

j.com/

Ann R

heum D

is: first published as 10.1136/ard.49.1.7 on 1 January 1990. Dow

nloaded from