Skeletal Radiol (2006) 35: 909–914DOI 10.1007/s00256-006-0157-6 SCIENTIFIC ARTICLE

Martin WilliamsRobert G. W. LambertGian S. JhangriMichael GraceJay ZelasoBen WongSukhvinder S. Dhillon

Received: 16 August 2005Revised: 24 April 2006Accepted: 28 April 2006Published online: 2 June 2006# ISS 2006

Humeral head cysts and rotator cuff tears:an MR arthrographic study

Abstract Objective: Humeral tuber-osity cysts are a common finding,with previous reports suggesting theyare related to rotator cuff tear oraging. The aim of this study was toinvestigate the characteristics of cystsin the tuberosities of the humeral headand their relationship with rotator cufftear and age. Design and patients:Shoulder MR arthrograms were re-viewed in 120 consecutive patients—83 males (mean age 38.0, range 19–59 years) and 37 females (mean age41.2, range 15–59 years). Patientswere referred for investigation of avariety of conditions, and instabilitywas suspected in only a minority ofcases. MR was performed before andafter direct arthrography with 0.01%solution of gadolinium. Cysts weredefined as well-demarcated circular/ovoid foci in two planes that demon-strated high signal on pre-arthro-graphic T2W sequences. Location,size and numbers of cysts and post-arthrographic enhancement weredocumented, along with the locationof rotator cuff tears, if present.Results: Cysts in the tuberosities ofthe humerus were identified in 84patients (70%), and were seen seven

times more frequently in the posterioraspect of the greater tuberosity thananteriorly. Most cysts (94%) demon-strated communication with the jointpost-arthrogram. Rotator cuff tearswere present in 36 patients, and 79%of all tears occurred in supraspinatustendon. There was no significantdifference in the occurrence of cystsbetween patients older or youngerthan age 40 or between genders, butrotator cuff tears were seen signifi-cantly more often in the older agegroup (p<0.01). Tuberosity cysts androtator cuff tears did not appear to berelated (p=0.55). However, whilst thislack of association was quite obviousposteriorly (p=0.84), the trend in theanterior aspect of the greater tuberos-ity is not as clear (p=0.14).Conclusions: Humeral cysts are mostoften located in the posterior aspect ofthe greater tuberosity, communicatewith the joint space and, in thislocation, are not related to aging orrotator cuff tear.

Keywords Magnetic resonanceimaging . Arthrography . Cyst .Humeral head . Rotator cuff

Introduction

Radiographic findings from patients with a documentedrotator cuff tear have been found to include a decreasedacromiohumeral distance, a hooked acromion and degen-erative changes of the greater tuberosity and acromiocla-

vicular joint [1–4]. These changes also include the presenceof cysts within the vicinity of the rotator cuff insertion onthe greater tuberosity.

It is unclear whether these cysts are age-related, develop-mental, occur in asymptomatic patients or are specific torotator cuff pathology. A radiographic study of patients with

M. Williams (*)Department of Radiology,Southmead Hospital,Westbury-on-Trym, Bristol,BS10 5NB, UKe-mail: martwilliams@f2s.comTel.: +44-0117-9505050Fax: +44-0117-9506376

R. G. W. Lambert . G. S. Jhangri .M. Grace . J. Zelaso . B. Wong .S. S. DhillonDepartment of Radiologyand Diagnostic Imaging,University of Alberta Hospital,8840-112th,Edmonton, T6G 2B7, Canada

symptomatic and surgically documented rotator cuff tearshas shown a positive correlation with greater tuberositycysts. These radiographic abnormalities were not found inasymptomatic subjects without a rotator cuff tear [5].However, others have disagreed on the significance ofhumeral cystic changes on subsequent MR studies [6, 7].

Etiology of these cysts is not fully understood. Micro-avulsive cuff tears may incite an inflammatory reaction,resulting in synovial fluid entering the cysts through smallcommunicating pores on their surface [7]. They may besecondary to contact of the bone underlying the cuff tearwith the coracoacromial arch. Alternatively, they could bedevelopmental and therefore unrelated to aging or under-lying shoulder pathology.

This latter theory is supported by the finding that theincidence of humeral head cysts in asymptomatic patientshas been noted as ranging from 15–45% [6–8]. Few studieshave examined the distribution of these cysts on thehumeral head. It has been suggested that the presence ofcysts within the lesser tuberosity and anterior aspect of thegreater tuberosity has a high specificity for subscapularisand supraspinatus tendon tears respectively [6]. This hasbeen supported by the relatively high prevalence of cystswithin the posterior aspect of the greater tuberosity inasymptomatic shoulders [8]. The purpose of this study wasto examine the relationship of shoulder pathology bypresence and location of cysts with patient age and rotatorcuff tears as seen on MR arthrograms.

Materials and methods

A retrospective search of MR arthrograms of the shouldercovering an 18-month period in a single MR center wasconducted. Patient inclusion criteria were: a) a prearthro-gram MR examination, b) a standard anterior fluoroscopi-cally guided injection of the shoulder using a volume of 12–

14 cc of injection mixture comprising 10 cc contrast, 10 ccsterile water and 0.1 cc gadolinium (Magnevist, BerlexLaboratories, Wayne, NJ, USA), and c) a post-arthrogramMR examination. Patient exclusion criteria were: unavail-ability of studies for review, previous shoulder surgery, andpre and post MR examinations more than 7 days apart.Although cysts develop slowly, this cut-off was chosen tominimize the chance of pathology occurring in the interimwhich could complicate the interpretation of the results.

A standard positioning and imaging protocol was uti-lized, and patients had a minimum of the followingsequences: pre-arthrogram (FOV 180 mm, slice thickness4 mm, matrix 512×256) consisting of oblique coronal andsagittal T2TSE (TR3000/TE99), oblique coronal andsagittal T2TSE with fat suppression (TR3000/TE90);post-arthrogram (FOV 160 mm, slice thickness 3 mm,matrix 256×256) consisting of oblique coronal and sagittal(TR700/TE13) and axial (TR400/TE10) T1SE with fatsuppression.

All scans were performed at a single center, with a strictpositioning protocol requiring the humerus to be in full (notforced) external rotation. The greater tuberosity was dividedinto anterior and posterior by a virtual line drawn through the‘12 o’clock’ position with respect to the shaft of the humeruson the sagittal images (Fig. 1). Cysts identified as anteriorand posterior to this line were defined as located in the‘anterior’ and ‘posterior’ greater tuberosity respectively.This represented the exact anatomical transition betweeninsertion of supraspinatus and infraspinatus tendons in manycases. On occasion, it was not possible to clearly separatethese two tendons at their insertion or identify separateanterior and middle facets of the greater tuberosity, and sothe 12 o’clock position was chosen as a reproduciblelandmark. (Preliminary assessment of non-study cases hadconfirmed interobserver disagreement in dividing the tuber-osity into anatomical facets in some cases). In a similarmanner, insertional rotator cuff tears located anterior and

Fig. 1 Diagrammatic represen-tation (a) and correspondingT2W sagittal image (b) illus-trating the division of humeralgreater tuberosity into anteriorand posterior by the use of avertical line drawn through the12 o’clock position

910

posterior to this line were assigned to the supraspinatus andinfraspinatus tendons respectively. The lesser tuberosity wasalso studied for the presence of cysts and subscapularistendon pathology.

Cysts were defined as well-demarcated rounded or ovoidfoci visualized on two planes and demonstrating homo-genous hyperintensity on all pre-gadolinium T2W se-quences. The arthrographic study was used to determinewhether cysts communicated with the joint space. Acommunicating cyst was defined as a cyst seen on the pre-arthrographic sequences that contained contrast material onthe post-arthrographic sequences (Figs. 2 and 3). Cysts thatdid not demonstrate any filling with contrast were recordedas non-communicating. Demographic information andclinical history were collected from clinical records. BothMR examinations were evaluated for the presence andlocation of rotator cuff tears, tuberosity cysts and anycoexistent glenohumeral or labral pathology.

After establishing our definitions and review protocol,final training consisted of five cases randomly selected andevaluated independently by four observers with subsequentconsensus review. There was good agreement on the pre-sence or absence of cysts and whether they filled withcontrast material or not, and the presence or absence ofrotator cuff tears. An equal number of cases were read in-dependently by one of the four observers. Any equivocalcases were selected by the individual observer and reviewedjointly by all four observers to achieve a consensus opinion.

Data were stored and verified, with descriptive statisticsbeing determined on a group basis. Comparisons betweensub-groups were done by chi-square and t-tests. Allstatistical tests were at the 0.05 probability level.

Results

One hundred and twenty eight imaging files of consecutivepatients were retrieved. Eight patients did not meet in-

clusion criteria because pre and post MR arthrograms weregreater than 7 days apart. Two patients had undergonebilateral examinations. One shoulder examination fromeach of these two patients was randomly selected for review.The final study group was comprised of 120 MRarthrograms in 120 patients. The mean age of the groupwas 39 (SD±11) years and 83 (69%) were male (Table 1).Patients were referred for a range of indications includingtrauma 68, pain 43 and instability 9. The majority had theright side imaged, and subsequent analysis demonstrated nodifference between the left and right side, nor any differencerelative to referral history. Rotator cuff tears (supraspinatus,infraspinatus and subscapularis) were found in 36 shoulders(30%). Cysts occurred in 84 shoulders (70%), and the vastmajority of cysts (94%) were classified as communicating.Glenolabral abnormality of any kind was seen in only 20%of shoulders.

There were no significant differences (p=0.30) whencomparing between patients with and without cysts for thoseless than age 40 versus those age 40 or greater. There wasalso no significant difference (p=0.64) between males andfemales when comparing cysts versus no cysts (Table 2).However, there was a significant difference (p<0.01) be-tween those with or without a tear when comparing betweenthose aged less than 40 versus those aged 40 or greater with

Fig. 2 Arthrogram of a 44-year-old male with no rotator cuff orlabral pathology. (a) Pre-arthro-gram sagittal T2FSE denom-stating collection of intraosseouscysts within the posterior greatertuberosity/bare area(arrows).(b) Corresponding post-arthro-gram sagittal T1W with fatsaturation (FS) with intra-articular gadolinium, showingextension of gadolinium into thecysts, comfirming communica-tion with the joint space

Table 1 Demographic and pathology characteristics

N=120 %

Age in years (mean ± SD: 39±11)<40 58 48≥40 62 52

GenderM 83 69F 37 31

Patients with rotator cuff tears 36 30Patients with cysts 84 70

911

tears being more likely to occur in the older category. Therewas no significant difference between genders (Table 3).

Tuberosity cysts were seen in the majority of shoulders,and occurrence in more than one location of a shoulder wasnot unusual, with observed distribution as follows: 17(20%) anterior, 13(15%) lesser and 76 (90%) posterior(Table 4). The size of the cysts did not seem to vary relativeto the location but multiple cysts were much more commonposteriorly.

There were no significant associations found for theoccurrence of rotator cuff tendon tears in conjunction withcysts for either the shoulder overall (p=0.55), or theanterior (p=0.14), lesser (p=0.34), and posterior (p=0.84)tuberosities (Table 5).

Discussion

In this study, communicating and non-communicatingcysts were found in 70% of shoulders examined. Themajority were seen in the posterior aspect of the greatertuberosity, a finding in keeping with previous studies [6].Almost all of these cysts were of the communicating type,and appeared unrelated to age, gender, or presence ofrotator cuff tear or other joint pathology.

These results are somewhat different from previousreports. Post-mortem studies demonstrated cysts in theanatomical neck of the humerus with rotator cuff tears [7].However, this was an older population (mean age 70 years)with radiographic evidence of cuff tears, which includedtuberosity changes of sclerosis and cyst formation in theinclusion criteria. Most cysts communicated with the jointspace, but some were partly filled with fibrous or fibrinousmaterial.

AMR-based study has suggested an association betweenage and presence of cuff tears, and presence of humeraltuberosity cysts [10], as well as impingement from bonyspurs on the acromion, as a coexisting factor in develop-ment of cuff tears. However, it was unclear from this studywhether a direct association existed between bony spurs orimpingement and humeral tuberosity cystic changes. In ourstudy, the older age group (40–59 years) were significantlymore likely to have a rotator cuff tear; however, there wasno age relationship relative to the occurrence of a cyst. Anassociation of rotator cuff lesions and cyst-like lesions onthe humeral tuberosity has been documented on a radio-graphic study, but this observation is of limitedclinical usefulness due to poor high inter-observer vari-ability [11].

Fig. 3 Sagittal T2FS pre-arthrogram (a) and axial and sagittal T1WFS post-arthrogram images (b, c) demonstating a communicating cyst inthe lesser tuberosity in a 50-year-old female patient with no evidence of rotator cuff pathology

Table 2 Cyst status by demographics

N=120 Cysts No cysts p

Age in years 0.30<40 58 38 20≥40 62 46 16Gender 0.64Male 83 57 26Female 37 27 10

Table 3 Tear status by demographics

N=120 Tears No tears p

Age in years <0.01<40 58 9 49≥40 62 27 35Gender 0.59Male 83 24 59Female 37 12 25

912

An MR evaluation of 140 painful shoulders concludedthere were two distinct types of humeral cysts: one seen at theattachment of the supraspinatus and subscapularis tendonswhich is closely related to cuff tears, and the other in the barearea of the anatomical neckwhich is related to increasing age[6]. In this report, cyst formation was found in the superior(anterior) facet, middle facet and lesser tuberosity only insubjects with supraspinatus and subscapularis tears respec-tively. These relationships were not found in our study, inwhich cysts were seen in the anterior facet and lesser tuber-osity in subjects both with and without cuff tears. Ourfindings are similar with respect to the lack of correlation ofthe occurrence of cuff tears near the posterior aspect of thegreater tuberosity and incidence of cysts at that site.

It is clear that cysts found in the posterior aspect of thegreater tuberosity differ significantly from cysts foundelsewhere in the humerus. They are more numerous andoccasionally larger. Given that they appear unrelated to anyunderlying pathology, one could theorize that these cystsare developmental. This would be supported by recent

studies showing that cystic changes in the humeral head canbe found in asymptomatic populations. In aMR study of theasymptomatic shoulders of athletes, nine of 40 shoulders(22.5%) had sclerotic or cystic changes in the greatertuberosity [9]; and in another report, cysts limited to theposterior aspect of the greater tuberosity were identified in15% to 45% of asymptomatic shoulders [8].

No relationship was found between presence or numberof cysts with age in our population, which one wouldexpect if these were primarily degenerative in etiology. Ourpatient group was relatively young (under 60 with averageage of 39), and younger than in previous studies [7, 10, 11].However, whilst increasing age might explain the risingincidence of humeral head cysts in other populations [10],age cannot explain the high prevalence of cysts seen in ourstudy group.

There are several limitations to this study. Surgical orpathological correlation of the MR findings is lacking. Thiswould have been helpful for confirmation of the rotator cuffinjuries and assessment of other intra-articular pathology.However, surgical evaluation of the bone surface wouldonly provide a limited estimate of the number and distri-bution of cysts, and in this respect our results are likely to bemore accurate than a surgical estimate, especially as 94% ofthe cysts demonstrated contrast enhancement with intraar-ticular gadolinium. It is possible we may have missed somecommunicating cysts due to cyst contents being viscous,resulting in lack of contrast filling. We feel this would be asmall number, given that 94% were designated as being ofthe communicating type. In addition, it was not possible toperform arthroscopy correlation for most of the patients, asthe majority were managed conservatively. If our resultsoverestimated the number of cuff tears, this would not affectthe conclusion. Furthermore, if the number of cuff tearswere underestimated, the “missed” tears would most likelyhave been in the supraspinatus tendon. How much of aneffect this would have on the outcome is debatable. Thehigh prevalence of cysts in the posterior aspect of the greatertuberosity numerically overwhelms the statistical analysis,and it seems unlikely that changing the number of “anterior”cuff tears would significantly change the correlation. Ourstudy did not include a detailed characterisation of the rota-tor cuff tears, and further study would require largernumbers and involve investigating any relationship be-tween the type of rotator cuff tear (full/partial/bursal/articular/insertional) and presence of tuberosity cysts, withanalysis by location.

The decision to divide the humeral tuberosity into ante-rior and posterior on the sagittal image at the “12 o’clock”position was taken early during the research design. Evenafter initial training, we found that in many cases it was notpossible to identify and separate superior and middle facetsof the greater tuberosity. In the end, we believe that this isnot a source of significant error in our results, as cysts wererarely seen close to this line of demarcation, and almost all

Table 5 Association between cysts and rotator cuff tears bylocation

Humerus – presence of any tuberosity cyst

n=120 Cysts No cysts p

* Rotator cuff tears:Shoulder 0.5584 No 57 2736 Yes 27 9Supraspinatus 0.1486 No 10 7634 Yes 7 27Subscapularis 0.34114 No 10 1046 Yes 1 5Infraspinatus/teres minor 0.84117 No 74 433 Yes 2 1

*There may be more than one location of rotator cuff tear in ashoulder

Table 4 Cyst location and size

Location by tuberosity

n=120 Anterior Lesser Posterior

Subjects with cysts* 17 13 76Size of largest cyst≤2 mm 10 5 32>2 mm 7 8 44Number of cysts1 7 6 172 or more 10 7 59

*Cysts may occur in more than one location

913

cysts in the greater tuberosity were obviously either“anterior” or “posterior”.

In conclusion, the majority of patients, regardless of age,demonstrate the presence of small intraosseous cysts in theposterior aspect of the greater tuberosity that communicatewith the joint. These are not related to rotator cuff tear, and

in the majority of cases would appear to be incidentalfindings. Cysts in the anterior aspect of the greatertuberosity and the lesser tuberosity may be seen in thepresence or absence of rotator cuff tear, and the etiologyand significance of these cysts is not clear.

References

1. Hamada K, Fukuda H, Mikasa M,Kobayashi Y. Roentgenographic find-ings in massive rotator cuff tears. Along-term observation. Clin Orthop1990;254:92–96

2. Neer CS 2nd. Impingement lesions.Clin Orthop 1983;173:70–77

3. Tuite MJ, Toivonen DA, Orwin JF,Wright DH. Acromial angle on radio-graphs of the shoulder. correlation withthe impingement syndrome and rotatorcuff tears. AJR Am J Roentgenol1995;165:609–613

4. Norwood LA, Barrack R, Jacobson KE.Clinical presentation of complete tearsof the rotator cuff. J Bone Joint SurgAm 1989;71:499–505

5. Pearsall AW, Bonsell S, Heitman RJ,Helms CA, Osbahr D, Speer KP.Radiographic findings associated withsymptomatic rotator cuff tears.J Shoulder Elbow Surg 2003;12:122–127

6. Sano A, Itoi E, Konno N, Kido T,Urayama M, Sato K. Cystic changes ofthe humeral head on MR imaging.Relation to age and cuff-tears. ActaOrthop Scand 1998;69:397–400

7. Cotton RE, Rideout DF. Tears of thehumeral rotator cuff; a radiological andpathological necropsy survey. J BoneJoint Surg Br 1964;46:314–328

8. Wright T, Yoon C, Schmit BP. ShoulderMRI refinements. differentiation ofrotator cuff tear from artifacts andtendonosis, and reassessment of normalfindings. Semin Ultrasound CT MR2001;22:383–395

9. Connor PM, Banks DM, Tyson AB,Coumas JS, D’Alessandro DF. Mag-netic resonance imaging of the asymp-tomatic shoulder of overhead athletes. a5-year follow-up study. Am J SportsMed 2003;31:724–727

10. Needell SD, Zlatkin MB, Sher JS,Murphy BJ, Uribe JW. MR imaging ofthe rotator cuff. peritendinous and boneabnormalities in an asymptomatic pop-ulation. AJR Am J Roentgenol1996;166:863–867

11. Huang LF, Rubin DA, Britton CA.Greater tuberosity changes as revealedby radiography. lack of clinical useful-ness in patients with rotator cuff dis-ease. AJR Am J Roentgenol1999;172:1381–1388

914