- Email: [email protected]
True anteroposterior (Grashey) view as a screening radiograph for further imaging study in rotator cuff tear
J Shoulder Elbow Surg (2013) 22, 901-907
www.elsevier.com/locate/ymse
True anteroposterior (Grashey) view as a screening radiograph for further imaging study in rotator cuff tear Kyoung Hwan Koh, MDa, Kye Young Han, MDb, Young Cheol Yoon, MDc, Seung Won Lee, MDd, Jae Chul Yoo, MDd,* a
Department of Orthopaedic Surgery, Seoul Medical Center, Seoul, South Korea Department of Orthopaedic Surgery, College of Medicine, Kangwon National University School of Medicine, Chuncheon, South Korea c Department of Radiology, Samsung Medical Center, Sungkyunkwan University School of Medicine, Seoul, South Korea d Department of Orthopaedic Surgery, Samsung Medical Center, Sungkyunkwan University School of Medicine, Seoul, South Korea b
Background: Although findings of conventional radiography seem nonspecific, it is still the first imaging modality used to evaluate patients with rotator cuff tears. The purpose of this study is to determine whether the true anteroposterior (AP) view of the glenohumeral (GH) joint (the thorax is rotated to the affected shoulder for 35 -45 ) is more sensitive than the conventional shoulder AP view (the beam and cassette are perpendicular to the torso but oblique to the glenohumeral joint) in terms of detecting rotator cuff tears. Materials and methods: Intermixed GH AP and conventional AP views of 160 consecutive shoulders, which were repaired by arthroscopic surgery, were reviewed in a blinded fashion. The detection rate of 5 pathognomonic signs for rotator cuff tear were determined by use of both radiographs: greater tuberosity (GT) sclerosis, GT osteophyte, subacromial (SA) osteophyte, GT cyst, and humeral head osteophyte. An additional comparison according to the tear size was performed. Results: The detection of all radiographic findings was significantly greater on the GH AP view than on the conventional AP view (P < .001 for GT sclerosis, P ¼ .003 for GT osteophyte, P ¼ .013 for GT cyst, P < .001 for SA osteophyte, and P ¼ .002 for humeral head osteophyte). In subgroup analysis by tear size, GT sclerosis was identified significantly more on the GH AP view for all tear sizes, GT osteophytes showed a higher detection rate, especially in medium-sized tears, and SA osteophytes showed a higher detection rate in medium and large to massive tears. Conclusion: The GH view is more sensitive than the conventional AP view for detecting pathognomonic findings of rotator cuff tear. In particular, the superiority of the GH AP view in detecting abnormal radiographic findings seemed prominent in medium-sized tears. Level of evidence: Level III, Study of Nonconsecutive Patients, Diagnostic Study. Ó 2013 Journal of Shoulder and Elbow Surgery Board of Trustees. Keywords: Shoulder; plain radiographs; rotator cuff tears; true anteroposterior view; conventional shoulder anteroposterior view
This study was approved by the Institutional Review Board of Samsung Medical Center (study No. 2008-04-012-001). *Reprint requests: Jae Chul Yoo, MD, Department of Orthopaedic Surgery, Samsung Medical Center, Sungkyunkwan University School
of Medicine, 50 Ilwon-Dong, Kangnam-Ku, Seoul, 135-710, South Korea. E-mail address: [email protected] (J.C. Yoo).
1058-2746/$ - see front matter Ó 2013 Journal of Shoulder and Elbow Surgery Board of Trustees. http://dx.doi.org/10.1016/j.jse.2012.09.015
902
K.H. Koh et al.
Figure 1 Conventional shoulder AP view (A), true AP view of GH joint (B), and coronal MRI view (C) of same patient showing fullthickness rotator cuff tear. GT sclerosis and SA osteophytes seem more evident on the GH AP view than the conventional AP view.
Although ultrasonography and magnetic resonance imaging (MRI) have become more popular for diagnosing shoulder pathologies, especially for rotator cuff diseases, conventional radiography remains the first-line imaging modality in patients with shoulder pain or trauma. Calcific tendinitis, fractures, advanced rotator cuff tears, or severely arthritic shoulders can be easily detected by use of plain radiographs. For the evaluation of rotator cuff pathologies, the role of plain radiographs has been limited; however, several radiographic techniques and findings have been reported to help us detect rotator cuff pathology. The narrowing of the acromiohumeral distance is a well-known predictor of prognosis in patients undergoing rotator cuff repair. Other findings such as acromial morphology, greater tuberosity (GT) sclerosis, GT osteophyte, GT cyst, humeral osteophyte, subacromial (SA) osteophyte, and acromioclavicular joint arthrosis can also be helpful for detecting rotator cuff tears.3,4,6,9,12,16,20,22,23 While reviewing radiographs of patients who had undergone arthroscopic rotator cuff repair, we found that some plain radiographs were more sensitive to these pathognomonic findings of rotator cuff tears. Specifically, the true anteroposterior (AP) view of the glenohumeral (GH) joint (also referred to as the Grashey view or true AP view of the shoulder [GH AP view]) showed more of these abnormal radiographic findings than the conventional shoulder AP view (Fig. 1). The purposes of this study were to determine whether the GH AP view is superior to the conventional AP view in terms of detecting abnormal radiographic findings indicative of the presence of a rotator cuff tear and, second, to see whether these findings are dependent on rotator cuff tear size. Thus, the null hypothesis of this study was that the GH AP view and conventional AP view did not differ in the detection of pathognomonic findings of rotator cuff tears.
enrolled in this study. Arthroscopically confirmed full-thickness rotator cuff tears were included. The exclusion criteria were those circumstances where simple radiographs might be distorted: revision surgery, rheumatoid arthritis, infection, a combined history of instability, GH joint degenerative arthritis of grade II or more according to the Hamada classification,9 and a history of fractures in the shoulder girdle including the proximal humerus, scapula, acromion, and clavicle. Of the 290 shoulders, 75 were confirmed to have partialthickness rotator cuff tears during arthroscopic surgery. In addition, 5 had rheumatoid arthritis, 24 showed arthritic changes of grade II or more by the Hamada classification, and 11 were revision cases. One patient had a history of infection around the shoulder, and seven patients had a history of fracture (six clavicular and one humeral surgical neck fracture). Seven rotator cuff tears with a history of traumatic dislocation were also excluded. After the exclusion criteria were applied, 160 shoulders were available. The shoulders were divided into groups according to tear size: small, medium, or large to massive.2 The tear size was classified according to the tear diameter, by use of the classification system of Post et al18 as follows: small, less than 1 cm; medium, 1 to 3 cm; large, 3 to 5 cm; and massive, greater than 5 cm. The small tear group consisted of 43 patients (14 men and 29 women with a mean age of 57.9 years). There were 69 patients (28 men and 41 women with a mean age of 59.6 years) in the medium tear group and 48 patients (15 men and 33 women with a mean age of 64.6 years) in the large to massive tear group. We included 45 individuals (19 men and 26 women with a mean age of 59.2 years) who were diagnosed as having idiopathic adhesive capsulitis,5 after magnetic resonance arthrography to exclude any combined pathology, as a control group.
Arthroscopic evaluation All arthroscopic surgeries were performed by a single surgeon, and tear sizes were measured from the bursal side after complete debridement and preparation for repair.2
Imaging protocol
Materials and methods Patient selection From September 2005 to January 2008, 290 consecutive shoulders in which arthroscopic rotator cuff repairs were performed were
All radiographs were taken at our institution with a beam-to-film distance of 1.1 m, at 70 kVp and 63 mAs (DHF-158HII; Hitachi, Tokyo, Japan). Radiographs were prepared by several technicians under the supervision of a senior technician. GH AP views were obtained with the patient rotated posteriorly approximately by 35 to 45 such that the plane of the scapula
Grashey view for rotator cuff tear
903
Figure 2 Schematic drawing of GH AP view (A) and conventional shoulder AP view (B).
Figure 4
Figure 3
GT osteophyte (arrow) on true AP view of GH joint.
GT sclerosis (arrow) on true AP view of GH joint.
(rather than the body) was parallel to the cassette. The beam was directed tangentially to the GH joint, and upper arm rotation was neutral (Fig. 2, A).1,19 Conventional AP views were obtained with the patient erect and in the coronal plane of the body parallel to the cassette. The beam was directed in an AP direction relative to the body with the upper arm in neutral rotation (Fig. 2, B).
Measurements All radiographs (patient and control groups) were serially numbered and randomly assigned to 2 reviewers, unaware of tear sizes, for evaluation. All measurements were performed by the reviewers independently (one author [Y.C.Y.] and one non-author) on 2 occasions with an interval of 1 week. One reviewer was a board-certified musculoskeletal radiologist with 5 years of experience, and the second reviewer was an orthopaedic surgeon not involved in the surgery (a fellow).
Figure 5
GT cyst (arrow) on true AP view of GH joint.
The reviewers used a standardized form containing the 5 positive radiologic signs of a rotator cuff tear, namely, the presence of GT sclerosis (Fig. 3), a GT osteophyte (Fig. 4), a GT cyst (Fig. 5), an SA osteophyte (Fig. 6), and a humeral head (HH) osteophyte (located at the inferior articular side of the HH)
904
K.H. Koh et al.
Figure 6
SA osteophyte (arrow) on true AP view of GH joint (A) and conventional shoulder AP view (B).
groups). k Values were calculated for intraobserver and interobserver reliabilities to assess levels of agreement. In terms of interobserver and intraobserver coefficients, k values were interpreted as follows: poor, below 0.40; fair, between 0.41 and 0.59; good, between 0.60 and 0.74; and excellent, 0.75 or higher.8 We then tested the hypothesis that there is a difference in the detection rate of each radiographic finding (GT sclerosis, GT osteophyte, GT cyst, SA osteophyte, and HH osteophyte) between the GH AP and conventional AP views with the use of a generalized estimating equation.10,11 Differences between the 2 views in terms of the detection rate of abnormal radiographic findings according to the tear sizes (small, medium, and large to massive groups) were evaluated with the McNemar test. Statistical analysis was undertaken with SAS software (version 9.1.3; SAS Institute, Cary, NC, USA).
Results
Figure 7
HH osteophyte (arrow) on true AP view of GH joint.
(Fig. 7).3,4,9,12,15,16,21 Because our primary objective was to determine early signs of a rotator cuff tear by plain radiography, the acromiohumeral distance was not measured because it is a sign of an advanced rotator cuff tear.7,12-14,17,20 Findings were classified as normal or abnormal. A picture archiving and communication system (Centricity; GE, Chicago, IL, USA) was used for all plain radiographic evaluations.
Statistical analysis GH AP and conventional AP view results of the 2 reviewers were compared. First, we compared sex and age distributions in the 4 study groups (control and small, medium, and large to massive tear
No sex distribution differences were found among the 4 study groups (P ¼ .58). In terms of age distribution, patients in the large to massive tear group were older than those in the other 3 groups: 58 years (range, 41-73 years) in small tear group, 60 years (range, 39-76 years) in medium tear group, 65 years (range, 44-77 years) in large to massive tear group, and 59 years (range, 39-71 years) in control group (P ¼ .0028). Demographic data are presented in Table I. Analysis of intraobserver reliability showed good to excellent reliability for all radiographic findings on both views (Table II). Interobserver reliability on the GH AP view was fair to good for all measurements, whereas k values for the conventional AP view ranged from poor to good (Table III). For all the measured variables, the percentages of abnormal radiographic findings were significantly greater on the GH AP view than the conventional AP view by use of the generalized estimating equation method. This was performed by controlling the effects of 2 reviewers and repetition of measurements (P < .001 for GT sclerosis,
Grashey view for rotator cuff tear Table I
Demographic data for all subgroups
Sex Male Female Dominant arm Right Left Involved arm Right Left Mean age (range) (y)
Table II
905
Small tear
Medium tear
Large to massive tear
Control
14 29
28 41
15 33
19 26
39 4
62 7
43 5
41 4
34 9 58 (41-73)
53 16 60 (39-76)
34 14 65 (44-77)
21 24 59 (39-71)
Intraobserver reliability (k value)
Reviewer 1 GH AP view Conventional AP view Reviewer 2 GH AP view Conventional AP view
GT sclerosis
GT osteophyte
GT cyst
SA osteophyte
HH osteophyte
0.77 0.72
0.87 0.80
0.79 0.84
0.81 0.76
0.72 0.90
0.73 0.70
0.90 0.86
0.85 0.81
0.67 0.52
0.61 0.68
A value of 0.40 or less was interpreted as poor; 0.41 to 0.59, fair; 0.60 to 0.74, good; and 0.75 or greater, excellent.
Table III
Interobserver reliability
GH AP view Conventional AP view
GT sclerosis
GT osteophyte
GT cyst
SA osteophyte
HH osteophyte
0.44 (fair) 0.38 (poor)
0.66 (good) 0.63 (good)
0.60 (good) 0.62 (good)
0.46 (fair) 0.29 (poor)
0.55 (fair) 0.59 (fair)
A value of 0.40 or less was interpreted as poor; 0.41 to 0.59, fair; 0.60 to 0.74, good; and 0.75 or greater, excellent.
P ¼ .003 for GT osteophyte, P ¼ .013 for GT cyst, P < .001 for SA osteophyte, and P ¼ .002 for HH osteophyte). In subgroup analysis according to tear size, GT sclerosis was identified significantly more on the GH AP view for all tear sizes, that is, 62.79% versus 37.21% in the small tear group (P ¼ .005), 71.01% versus 52.9% in the medium tear group (P ¼ .0007), and 69.79% versus 44.79% in the large to massive tear group (P ¼ .0002). The GH AP view showed a higher detection rate for GT osteophytes especially in the medium-sized tear group (P < .001). The GH AP view detected more GT cysts in the medium-sized tear group. However, it seemed not to show clinical relevance because of its lower detection rate of 16% to 25%. SA osteophytes showed a significantly higher detection rate in the medium and large to massive tear groups (P ¼ .021 and P < .001, respectively). The detection rate for HH osteophytes was not so high, and there was no significant difference in all tear sizes. Table IV shows the subgroup analysis for the proportions of
cases interpreted as abnormal for each radiographic finding on each view.
Discussion The GH AP view in our study showed a higher detection rate of abnormal radiographic findings in association with rotator cuff tears than the conventional AP view. The difference between the GH AP view and conventional AP view was especially prominent in medium-sized rotator cuff tears. The main reason for this is that the GH AP view gives a more externally rotated HH view, with greater exposure of the GT. This makes plain radiography more sensitive to evaluate rotator cuff pathologies. The figure showed that image of GT is more prominent when we take the GH AP whereas the GT in conventional AP is internally rotated to the front of the radiograph, making less visible of the sclerosis and osteophyte.
906
K.H. Koh et al.
Table IV
Proportions of cases interpreted as abnormal for each radiographic finding in each view
Small tear
Medium tear
Large to massive tear
Control
)
GH AP view Conventional P value GH AP view Conventional P value GH AP view Conventional P value GH AP Conventional P value
AP view
AP view
AP view
AP
GT sclerosis
GT osteophyte
GT cyst
SA osteophyte
HH osteophyte
54 (62.8%) 32 (37.2%) .0005) 98 (71.0%) 73 (52.9%) .0007) 67 (69.8%) 43 (44.8%) .0002) 32 (32. 6%) 21 (24.4%) .8321
38 (44.2%) 40 (46.5%) .7237 99 (71.7%) 64 (46.4%) <.0001) 71 (74.0%) 65 (61.7%) .2733 17 (18. 9%) 13 (14.4%) .6542
10 (11.6%) 5 (5.8%) .1655 34 (24.6%) 22 (15.9%) .0143) 25 (26.0%) 20 (20.8%) .2513 3 (3.3%) 1 (1.1%) .1179
45 (52.3%) 36 (41.9%) .0495) 82 (59.4%) 68 (49.3%) .0269) 74 (77.1%) 48 (50.0%) <.0001) 27 (30.0%) 15 (16.7%) .1235
15 (17.4%) 7 (8.1%) .0209) 38 (27.5%) 31 (22.5%) .1266 41 (42.7%) 31 (32.3%) .0412) 18 (20.0%) 6 (6.7%) .0938
Statistically significant.
In our study, we evaluated 5 radiographic findings but did not include the acromiohumeral distance because it is known to represent the advanced stage of rotator cuff tears. Among the radiographic findings, the detection rate for GT sclerosis, GT osteophytes, and SA osteophytes was approximately 20% higher on the GH AP view than the conventional AP view in our study. MRI or ultrasonography would be used for definite diagnosis of rotator cuff tears in most cases, and plain radiographs are taken for an initial evaluation. In fact, the detection rate found in our study would not be high enough to justify screening by radiography alone. In addition, ‘‘screening’’ might not be appropriate to use for rotator cuff tear patients because screening, by definition, means the identification of real patients in an asymptomatic population. Nevertheless, the detection rates of the 2 radiographic views found in this study justify the use of the GH AP view for deciding on future evaluation modalities. If a patient has a rotator cuff tear, there is a greater likelihood that it will be detected on the GH AP view. However, if we can find those abnormal radiographic findings in more than 70% of rotator cuff tears, plain radiographs can be used as a screening tool for further imaging study. Furthermore, a combination of suitable radiographs and history taking with physical examination would improve decision making regarding whether MRI is required. On the other hand, GT cyst and HH osteophyte were detected rarely on both radiographs. The prime purpose of our study was not to find the radiographic findings of rotator cuff tear; rather, it was to show the superiority of the GH AP view over the conventional shoulder AP view in detecting those abnormal radiographic findings. However, if the detection of HH osteophytes or GT cysts was not easy on any plain radiographs, comparison of the 2 radiographic views in identifying rotator cuff tears would have less clinical relevance. There was an interesting result when the patients were subdivided according to tear size. In smaller tears, the bony change would not be obvious enough to be detected. Thus, it
would be difficult to detect those abnormalities on both the GH AP and conventional AP views. In large to massive tears, bony changes such as GT sclerosis or GT osteophytes might have advanced too much to be found on both radiographic views. In medium-sized tears, however, the superiority of the GH AP view over the conventional AP view in detecting rotator cuff tear was significantly prominent. Thus, the GH AP view could be useful especially in medium-sized tear as a screening tool for further imaging study. k Values on the shoulder AP view showed poor intraobserver and interobserver reliabilities for GT sclerosis and SA osteophytes, whereas k values on the GH AP view produced fair to good reliabilities for all radiographic findings, indicating that the GH AP view better facilitates interpretation consistency. It might be argued that the shoulder AP view with external rotation is likely to produce the same results as the GH AP view, because of similar rotations of the humerus and GT. However, some rotator cuff patients have combined stiffness that makes external rotation difficult; in addition, different degrees of external rotation can introduce considerable variance, especially between centers. Accordingly, we suggest that more uniform radiographs can be obtained if the GH AP view with neutral rotation is adopted. Another possible criticism of this study is that the true GH AP radiograph is better than the conventional AP view (which is oblique to the GH joint) as in any joint of the body without scientific backing. On the GH AP view, we might be able to detect arthritic changes such as joint space narrowing or osteophytes more easily.19 In addition, it is possible that some traction spurs or osteophytes under the acromion can be more easily detected on the conventional AP view (Fig. 6) depending on the direction of the spurs or osteophytes. However, we found that the pathognomonic findings of rotator cuff tears such GT sclerosis or GT osteophytes were observed more frequently on the GH AP view than on the conventional AP view. In our daily clinical practice and in many studies mentioning
Grashey view for rotator cuff tear radiographs in rotator cuff tears, those 2 radiographs were used without clear distinction.3,4,6,9,12,16,20,22,23 We recommend that both radiographs be obtained as a routine set in patients who are suspected of having rotator cuff pathologies to increase the sensitivity of detecting abnormal findings. This study has several weaknesses that should be considered. First, although the radiographs were assessed by the 2 reviewers twice, some measurement or assessment bias was inevitable. Second, despite the relatively good interobserver coefficients obtained, it is likely that rotational variances were introduced during imaging. Stratification of age, which is one of the most important prognostic variables, was not performed.
Conclusion The true AP view of the GH joint is more sensitive than the conventional shoulder AP view in detecting abnormal findings (especially GT sclerosis and osteophytes) suggesting rotator cuff tears. In addition, the superiority of the GH AP view over the conventional AP view is more evident in medium-sized tears.
Acknowledgments The authors give special thanks to Sun Woo Kim, PhD, and Min-Ji Kim for statistical assistance.
Disclaimer The authors, their immediate families, and any research foundations with which they are affiliated have not received any financial payments or other benefits from any commercial entity related to the subject of this article.
References 1. Ballinger PW. Merrill’s atlas of radiographic positions and radiologic procedures. St Louis: Mosby; 1986. 2. Bayne O, Bateman JE. Long term results of surgical repair of fullthickness rotator cuff tears. In: Bateman JE, Welsh P, editors. Surgery of the shoulder. St Louis: Mosby; 1984. p. 167-71. 3. Bernageau J. Roentgenographic assessment of the rotator cuff. Clin Orthop Relat Res 1990:87-91.
907 4. Bonsell S, Pearsall AWIV, Heitman RJ, Helms CA, Major NM, Speer KP. The relationship of age, gender, and degenerative changes observed on radiographs of the shoulder in asymptomatic individuals. J Bone Joint Surg Br 2000;82:1135-9. 5. Buchbinder R, Green S, Youd JM, Johnston RV, Cumpston M. Arthrographic distension for adhesive capsulitis (frozen shoulder). Cochrane Database Syst Rev 2008:CD007005. http://dx.doi.org/10. 1002/14651858.CD007005 6. Cohen RB, Williams GR Jr. Impingement syndrome and rotator cuff disease as repetitive motion disorders. Clin Orthop Relat Res 1998:95-101. 7. Ellman H, Hanker G, Bayer M. Repair of the rotator cuff. End-result study of factors influencing reconstruction. J Bone Joint Surg Am 1986;68:1136-44. 8. Fleiss JL. Statistical methods for rates and proportions. 2nd ed. New York: Wiley; 1981. p. 211-236. 9. Hamada K, Fukuda H, Mikasa M, Kobayashi Y. Roentgenographic findings in massive rotator cuff tears. A long-term observation. Clin Orthop Relat Res 1990:92-6. 10. Hardin JW. Generalized estimating equations (GEE). Wiley Online Library; 2005. http://dx.doi.org/10.1002/0470013192.bsa250. 11. Lipsitz SR, Kim K, Zhao L. Analysis of repeated categorical data using generalized estimating equations. Stat Med 1994;13:1149-63. 12. Norwood LA, Barrack R, Jacobson KE. Clinical presentation of complete tears of the rotator cuff. J Bone Joint Surg Am 1989;71: 499-505. 13. Nove-Josserand L, Boulahia A, Levigne C, Noel E, Walch G. Coracohumeral space and rotator cuff tears [in French]. Rev Chir Orthop Reparatrice Appar Mot 1999;85:677-83. 14. Nove-Josserand L, Levigne C, Noel E, Walch G. The acromio-humeral interval. A study of the factors influencing its height [in French]. Rev Chir Orthop Reparatrice Appar Mot 1996;82:379-85. 15. Ozaki J, Fujimoto S, Nakagawa Y, Masuhara K, Tamai S. Tears of the rotator cuff of the shoulder associated with pathological changes in the acromion. A study in cadavera. J Bone Joint Surg Am 1988;70: 1224-30. 16. Pearsall AW IV, Bonsell S, Heitman RJ, Helms CA, Osbahr D, Speer KP. Radiographic findings associated with symptomatic rotator cuff tears. J Shoulder Elbow Surg 2003;12:122-7. http://dx.doi.org/10. 1067/mse.2003.19 17. Petersson CJ, Redlund-Johnell I. The subacromial space in normal shoulder radiographs. Acta Orthop Scand 1984;55:57-8. 18. Post M, Silver R, Singh M. Rotator cuff tear. Diagnosis and treatment. Clin Orthop Relat Res 1983:78-91. 19. Sanders TG, Jersey SL. Conventional radiography of the shoulder. Semin Roentgenol 2005;40:207-22. 20. Saupe N, Pfirrmann CW, Schmid MR, Jost B, Werner CM, Zanetti M. Association between rotator cuff abnormalities and reduced acromiohumeral distance. AJR Am J Roentgenol 2006;187:376-82. http:// dx.doi.org/10.2214/AJR.05.0435 21. Umans HR, Pavlov H, Berkowitz M, Warren RF. Correlation of radiographic and arthroscopic findings with rotator cuff tears and degenerative joint disease. J Shoulder Elbow Surg 2001;10:428-33. http://dx.doi.org/10.1067/mse.2001.117123 22. Werner CM, Conrad SJ, Meyer DC, Keller A, Hodler J, Gerber C. Intermethod agreement and interobserver correlation of radiologic acromiohumeral distance measurements. J Shoulder Elbow Surg 2008; 17:237-40. http://dx.doi.org/10.1016/j.jse.2007.06.002 23. Wohlwend JR, van Holsbeeck M, Craig J, Shirazi K, Habra G, Jacobsen G, et al. The association between irregular greater tuberosities and rotator cuff tears: a sonographic study. AJR Am J Roentgenol 1998;171:229-33.
www.elsevier.com/locate/ymse
True anteroposterior (Grashey) view as a screening radiograph for further imaging study in rotator cuff tear Kyoung Hwan Koh, MDa, Kye Young Han, MDb, Young Cheol Yoon, MDc, Seung Won Lee, MDd, Jae Chul Yoo, MDd,* a
Department of Orthopaedic Surgery, Seoul Medical Center, Seoul, South Korea Department of Orthopaedic Surgery, College of Medicine, Kangwon National University School of Medicine, Chuncheon, South Korea c Department of Radiology, Samsung Medical Center, Sungkyunkwan University School of Medicine, Seoul, South Korea d Department of Orthopaedic Surgery, Samsung Medical Center, Sungkyunkwan University School of Medicine, Seoul, South Korea b
Background: Although findings of conventional radiography seem nonspecific, it is still the first imaging modality used to evaluate patients with rotator cuff tears. The purpose of this study is to determine whether the true anteroposterior (AP) view of the glenohumeral (GH) joint (the thorax is rotated to the affected shoulder for 35 -45 ) is more sensitive than the conventional shoulder AP view (the beam and cassette are perpendicular to the torso but oblique to the glenohumeral joint) in terms of detecting rotator cuff tears. Materials and methods: Intermixed GH AP and conventional AP views of 160 consecutive shoulders, which were repaired by arthroscopic surgery, were reviewed in a blinded fashion. The detection rate of 5 pathognomonic signs for rotator cuff tear were determined by use of both radiographs: greater tuberosity (GT) sclerosis, GT osteophyte, subacromial (SA) osteophyte, GT cyst, and humeral head osteophyte. An additional comparison according to the tear size was performed. Results: The detection of all radiographic findings was significantly greater on the GH AP view than on the conventional AP view (P < .001 for GT sclerosis, P ¼ .003 for GT osteophyte, P ¼ .013 for GT cyst, P < .001 for SA osteophyte, and P ¼ .002 for humeral head osteophyte). In subgroup analysis by tear size, GT sclerosis was identified significantly more on the GH AP view for all tear sizes, GT osteophytes showed a higher detection rate, especially in medium-sized tears, and SA osteophytes showed a higher detection rate in medium and large to massive tears. Conclusion: The GH view is more sensitive than the conventional AP view for detecting pathognomonic findings of rotator cuff tear. In particular, the superiority of the GH AP view in detecting abnormal radiographic findings seemed prominent in medium-sized tears. Level of evidence: Level III, Study of Nonconsecutive Patients, Diagnostic Study. Ó 2013 Journal of Shoulder and Elbow Surgery Board of Trustees. Keywords: Shoulder; plain radiographs; rotator cuff tears; true anteroposterior view; conventional shoulder anteroposterior view
This study was approved by the Institutional Review Board of Samsung Medical Center (study No. 2008-04-012-001). *Reprint requests: Jae Chul Yoo, MD, Department of Orthopaedic Surgery, Samsung Medical Center, Sungkyunkwan University School
of Medicine, 50 Ilwon-Dong, Kangnam-Ku, Seoul, 135-710, South Korea. E-mail address: [email protected] (J.C. Yoo).
1058-2746/$ - see front matter Ó 2013 Journal of Shoulder and Elbow Surgery Board of Trustees. http://dx.doi.org/10.1016/j.jse.2012.09.015
902
K.H. Koh et al.
Figure 1 Conventional shoulder AP view (A), true AP view of GH joint (B), and coronal MRI view (C) of same patient showing fullthickness rotator cuff tear. GT sclerosis and SA osteophytes seem more evident on the GH AP view than the conventional AP view.
Although ultrasonography and magnetic resonance imaging (MRI) have become more popular for diagnosing shoulder pathologies, especially for rotator cuff diseases, conventional radiography remains the first-line imaging modality in patients with shoulder pain or trauma. Calcific tendinitis, fractures, advanced rotator cuff tears, or severely arthritic shoulders can be easily detected by use of plain radiographs. For the evaluation of rotator cuff pathologies, the role of plain radiographs has been limited; however, several radiographic techniques and findings have been reported to help us detect rotator cuff pathology. The narrowing of the acromiohumeral distance is a well-known predictor of prognosis in patients undergoing rotator cuff repair. Other findings such as acromial morphology, greater tuberosity (GT) sclerosis, GT osteophyte, GT cyst, humeral osteophyte, subacromial (SA) osteophyte, and acromioclavicular joint arthrosis can also be helpful for detecting rotator cuff tears.3,4,6,9,12,16,20,22,23 While reviewing radiographs of patients who had undergone arthroscopic rotator cuff repair, we found that some plain radiographs were more sensitive to these pathognomonic findings of rotator cuff tears. Specifically, the true anteroposterior (AP) view of the glenohumeral (GH) joint (also referred to as the Grashey view or true AP view of the shoulder [GH AP view]) showed more of these abnormal radiographic findings than the conventional shoulder AP view (Fig. 1). The purposes of this study were to determine whether the GH AP view is superior to the conventional AP view in terms of detecting abnormal radiographic findings indicative of the presence of a rotator cuff tear and, second, to see whether these findings are dependent on rotator cuff tear size. Thus, the null hypothesis of this study was that the GH AP view and conventional AP view did not differ in the detection of pathognomonic findings of rotator cuff tears.
enrolled in this study. Arthroscopically confirmed full-thickness rotator cuff tears were included. The exclusion criteria were those circumstances where simple radiographs might be distorted: revision surgery, rheumatoid arthritis, infection, a combined history of instability, GH joint degenerative arthritis of grade II or more according to the Hamada classification,9 and a history of fractures in the shoulder girdle including the proximal humerus, scapula, acromion, and clavicle. Of the 290 shoulders, 75 were confirmed to have partialthickness rotator cuff tears during arthroscopic surgery. In addition, 5 had rheumatoid arthritis, 24 showed arthritic changes of grade II or more by the Hamada classification, and 11 were revision cases. One patient had a history of infection around the shoulder, and seven patients had a history of fracture (six clavicular and one humeral surgical neck fracture). Seven rotator cuff tears with a history of traumatic dislocation were also excluded. After the exclusion criteria were applied, 160 shoulders were available. The shoulders were divided into groups according to tear size: small, medium, or large to massive.2 The tear size was classified according to the tear diameter, by use of the classification system of Post et al18 as follows: small, less than 1 cm; medium, 1 to 3 cm; large, 3 to 5 cm; and massive, greater than 5 cm. The small tear group consisted of 43 patients (14 men and 29 women with a mean age of 57.9 years). There were 69 patients (28 men and 41 women with a mean age of 59.6 years) in the medium tear group and 48 patients (15 men and 33 women with a mean age of 64.6 years) in the large to massive tear group. We included 45 individuals (19 men and 26 women with a mean age of 59.2 years) who were diagnosed as having idiopathic adhesive capsulitis,5 after magnetic resonance arthrography to exclude any combined pathology, as a control group.
Arthroscopic evaluation All arthroscopic surgeries were performed by a single surgeon, and tear sizes were measured from the bursal side after complete debridement and preparation for repair.2
Imaging protocol
Materials and methods Patient selection From September 2005 to January 2008, 290 consecutive shoulders in which arthroscopic rotator cuff repairs were performed were
All radiographs were taken at our institution with a beam-to-film distance of 1.1 m, at 70 kVp and 63 mAs (DHF-158HII; Hitachi, Tokyo, Japan). Radiographs were prepared by several technicians under the supervision of a senior technician. GH AP views were obtained with the patient rotated posteriorly approximately by 35 to 45 such that the plane of the scapula
Grashey view for rotator cuff tear
903
Figure 2 Schematic drawing of GH AP view (A) and conventional shoulder AP view (B).
Figure 4
Figure 3
GT osteophyte (arrow) on true AP view of GH joint.
GT sclerosis (arrow) on true AP view of GH joint.
(rather than the body) was parallel to the cassette. The beam was directed tangentially to the GH joint, and upper arm rotation was neutral (Fig. 2, A).1,19 Conventional AP views were obtained with the patient erect and in the coronal plane of the body parallel to the cassette. The beam was directed in an AP direction relative to the body with the upper arm in neutral rotation (Fig. 2, B).
Measurements All radiographs (patient and control groups) were serially numbered and randomly assigned to 2 reviewers, unaware of tear sizes, for evaluation. All measurements were performed by the reviewers independently (one author [Y.C.Y.] and one non-author) on 2 occasions with an interval of 1 week. One reviewer was a board-certified musculoskeletal radiologist with 5 years of experience, and the second reviewer was an orthopaedic surgeon not involved in the surgery (a fellow).
Figure 5
GT cyst (arrow) on true AP view of GH joint.
The reviewers used a standardized form containing the 5 positive radiologic signs of a rotator cuff tear, namely, the presence of GT sclerosis (Fig. 3), a GT osteophyte (Fig. 4), a GT cyst (Fig. 5), an SA osteophyte (Fig. 6), and a humeral head (HH) osteophyte (located at the inferior articular side of the HH)
904
K.H. Koh et al.
Figure 6
SA osteophyte (arrow) on true AP view of GH joint (A) and conventional shoulder AP view (B).
groups). k Values were calculated for intraobserver and interobserver reliabilities to assess levels of agreement. In terms of interobserver and intraobserver coefficients, k values were interpreted as follows: poor, below 0.40; fair, between 0.41 and 0.59; good, between 0.60 and 0.74; and excellent, 0.75 or higher.8 We then tested the hypothesis that there is a difference in the detection rate of each radiographic finding (GT sclerosis, GT osteophyte, GT cyst, SA osteophyte, and HH osteophyte) between the GH AP and conventional AP views with the use of a generalized estimating equation.10,11 Differences between the 2 views in terms of the detection rate of abnormal radiographic findings according to the tear sizes (small, medium, and large to massive groups) were evaluated with the McNemar test. Statistical analysis was undertaken with SAS software (version 9.1.3; SAS Institute, Cary, NC, USA).
Results
Figure 7
HH osteophyte (arrow) on true AP view of GH joint.
(Fig. 7).3,4,9,12,15,16,21 Because our primary objective was to determine early signs of a rotator cuff tear by plain radiography, the acromiohumeral distance was not measured because it is a sign of an advanced rotator cuff tear.7,12-14,17,20 Findings were classified as normal or abnormal. A picture archiving and communication system (Centricity; GE, Chicago, IL, USA) was used for all plain radiographic evaluations.
Statistical analysis GH AP and conventional AP view results of the 2 reviewers were compared. First, we compared sex and age distributions in the 4 study groups (control and small, medium, and large to massive tear
No sex distribution differences were found among the 4 study groups (P ¼ .58). In terms of age distribution, patients in the large to massive tear group were older than those in the other 3 groups: 58 years (range, 41-73 years) in small tear group, 60 years (range, 39-76 years) in medium tear group, 65 years (range, 44-77 years) in large to massive tear group, and 59 years (range, 39-71 years) in control group (P ¼ .0028). Demographic data are presented in Table I. Analysis of intraobserver reliability showed good to excellent reliability for all radiographic findings on both views (Table II). Interobserver reliability on the GH AP view was fair to good for all measurements, whereas k values for the conventional AP view ranged from poor to good (Table III). For all the measured variables, the percentages of abnormal radiographic findings were significantly greater on the GH AP view than the conventional AP view by use of the generalized estimating equation method. This was performed by controlling the effects of 2 reviewers and repetition of measurements (P < .001 for GT sclerosis,
Grashey view for rotator cuff tear Table I
Demographic data for all subgroups
Sex Male Female Dominant arm Right Left Involved arm Right Left Mean age (range) (y)
Table II
905
Small tear
Medium tear
Large to massive tear
Control
14 29
28 41
15 33
19 26
39 4
62 7
43 5
41 4
34 9 58 (41-73)
53 16 60 (39-76)
34 14 65 (44-77)
21 24 59 (39-71)
Intraobserver reliability (k value)
Reviewer 1 GH AP view Conventional AP view Reviewer 2 GH AP view Conventional AP view
GT sclerosis
GT osteophyte
GT cyst
SA osteophyte
HH osteophyte
0.77 0.72
0.87 0.80
0.79 0.84
0.81 0.76
0.72 0.90
0.73 0.70
0.90 0.86
0.85 0.81
0.67 0.52
0.61 0.68
A value of 0.40 or less was interpreted as poor; 0.41 to 0.59, fair; 0.60 to 0.74, good; and 0.75 or greater, excellent.
Table III
Interobserver reliability
GH AP view Conventional AP view
GT sclerosis
GT osteophyte
GT cyst
SA osteophyte
HH osteophyte
0.44 (fair) 0.38 (poor)
0.66 (good) 0.63 (good)
0.60 (good) 0.62 (good)
0.46 (fair) 0.29 (poor)
0.55 (fair) 0.59 (fair)
A value of 0.40 or less was interpreted as poor; 0.41 to 0.59, fair; 0.60 to 0.74, good; and 0.75 or greater, excellent.
P ¼ .003 for GT osteophyte, P ¼ .013 for GT cyst, P < .001 for SA osteophyte, and P ¼ .002 for HH osteophyte). In subgroup analysis according to tear size, GT sclerosis was identified significantly more on the GH AP view for all tear sizes, that is, 62.79% versus 37.21% in the small tear group (P ¼ .005), 71.01% versus 52.9% in the medium tear group (P ¼ .0007), and 69.79% versus 44.79% in the large to massive tear group (P ¼ .0002). The GH AP view showed a higher detection rate for GT osteophytes especially in the medium-sized tear group (P < .001). The GH AP view detected more GT cysts in the medium-sized tear group. However, it seemed not to show clinical relevance because of its lower detection rate of 16% to 25%. SA osteophytes showed a significantly higher detection rate in the medium and large to massive tear groups (P ¼ .021 and P < .001, respectively). The detection rate for HH osteophytes was not so high, and there was no significant difference in all tear sizes. Table IV shows the subgroup analysis for the proportions of
cases interpreted as abnormal for each radiographic finding on each view.
Discussion The GH AP view in our study showed a higher detection rate of abnormal radiographic findings in association with rotator cuff tears than the conventional AP view. The difference between the GH AP view and conventional AP view was especially prominent in medium-sized rotator cuff tears. The main reason for this is that the GH AP view gives a more externally rotated HH view, with greater exposure of the GT. This makes plain radiography more sensitive to evaluate rotator cuff pathologies. The figure showed that image of GT is more prominent when we take the GH AP whereas the GT in conventional AP is internally rotated to the front of the radiograph, making less visible of the sclerosis and osteophyte.
906
K.H. Koh et al.
Table IV
Proportions of cases interpreted as abnormal for each radiographic finding in each view
Small tear
Medium tear
Large to massive tear
Control
)
GH AP view Conventional P value GH AP view Conventional P value GH AP view Conventional P value GH AP Conventional P value
AP view
AP view
AP view
AP
GT sclerosis
GT osteophyte
GT cyst
SA osteophyte
HH osteophyte
54 (62.8%) 32 (37.2%) .0005) 98 (71.0%) 73 (52.9%) .0007) 67 (69.8%) 43 (44.8%) .0002) 32 (32. 6%) 21 (24.4%) .8321
38 (44.2%) 40 (46.5%) .7237 99 (71.7%) 64 (46.4%) <.0001) 71 (74.0%) 65 (61.7%) .2733 17 (18. 9%) 13 (14.4%) .6542
10 (11.6%) 5 (5.8%) .1655 34 (24.6%) 22 (15.9%) .0143) 25 (26.0%) 20 (20.8%) .2513 3 (3.3%) 1 (1.1%) .1179
45 (52.3%) 36 (41.9%) .0495) 82 (59.4%) 68 (49.3%) .0269) 74 (77.1%) 48 (50.0%) <.0001) 27 (30.0%) 15 (16.7%) .1235
15 (17.4%) 7 (8.1%) .0209) 38 (27.5%) 31 (22.5%) .1266 41 (42.7%) 31 (32.3%) .0412) 18 (20.0%) 6 (6.7%) .0938
Statistically significant.
In our study, we evaluated 5 radiographic findings but did not include the acromiohumeral distance because it is known to represent the advanced stage of rotator cuff tears. Among the radiographic findings, the detection rate for GT sclerosis, GT osteophytes, and SA osteophytes was approximately 20% higher on the GH AP view than the conventional AP view in our study. MRI or ultrasonography would be used for definite diagnosis of rotator cuff tears in most cases, and plain radiographs are taken for an initial evaluation. In fact, the detection rate found in our study would not be high enough to justify screening by radiography alone. In addition, ‘‘screening’’ might not be appropriate to use for rotator cuff tear patients because screening, by definition, means the identification of real patients in an asymptomatic population. Nevertheless, the detection rates of the 2 radiographic views found in this study justify the use of the GH AP view for deciding on future evaluation modalities. If a patient has a rotator cuff tear, there is a greater likelihood that it will be detected on the GH AP view. However, if we can find those abnormal radiographic findings in more than 70% of rotator cuff tears, plain radiographs can be used as a screening tool for further imaging study. Furthermore, a combination of suitable radiographs and history taking with physical examination would improve decision making regarding whether MRI is required. On the other hand, GT cyst and HH osteophyte were detected rarely on both radiographs. The prime purpose of our study was not to find the radiographic findings of rotator cuff tear; rather, it was to show the superiority of the GH AP view over the conventional shoulder AP view in detecting those abnormal radiographic findings. However, if the detection of HH osteophytes or GT cysts was not easy on any plain radiographs, comparison of the 2 radiographic views in identifying rotator cuff tears would have less clinical relevance. There was an interesting result when the patients were subdivided according to tear size. In smaller tears, the bony change would not be obvious enough to be detected. Thus, it
would be difficult to detect those abnormalities on both the GH AP and conventional AP views. In large to massive tears, bony changes such as GT sclerosis or GT osteophytes might have advanced too much to be found on both radiographic views. In medium-sized tears, however, the superiority of the GH AP view over the conventional AP view in detecting rotator cuff tear was significantly prominent. Thus, the GH AP view could be useful especially in medium-sized tear as a screening tool for further imaging study. k Values on the shoulder AP view showed poor intraobserver and interobserver reliabilities for GT sclerosis and SA osteophytes, whereas k values on the GH AP view produced fair to good reliabilities for all radiographic findings, indicating that the GH AP view better facilitates interpretation consistency. It might be argued that the shoulder AP view with external rotation is likely to produce the same results as the GH AP view, because of similar rotations of the humerus and GT. However, some rotator cuff patients have combined stiffness that makes external rotation difficult; in addition, different degrees of external rotation can introduce considerable variance, especially between centers. Accordingly, we suggest that more uniform radiographs can be obtained if the GH AP view with neutral rotation is adopted. Another possible criticism of this study is that the true GH AP radiograph is better than the conventional AP view (which is oblique to the GH joint) as in any joint of the body without scientific backing. On the GH AP view, we might be able to detect arthritic changes such as joint space narrowing or osteophytes more easily.19 In addition, it is possible that some traction spurs or osteophytes under the acromion can be more easily detected on the conventional AP view (Fig. 6) depending on the direction of the spurs or osteophytes. However, we found that the pathognomonic findings of rotator cuff tears such GT sclerosis or GT osteophytes were observed more frequently on the GH AP view than on the conventional AP view. In our daily clinical practice and in many studies mentioning
Grashey view for rotator cuff tear radiographs in rotator cuff tears, those 2 radiographs were used without clear distinction.3,4,6,9,12,16,20,22,23 We recommend that both radiographs be obtained as a routine set in patients who are suspected of having rotator cuff pathologies to increase the sensitivity of detecting abnormal findings. This study has several weaknesses that should be considered. First, although the radiographs were assessed by the 2 reviewers twice, some measurement or assessment bias was inevitable. Second, despite the relatively good interobserver coefficients obtained, it is likely that rotational variances were introduced during imaging. Stratification of age, which is one of the most important prognostic variables, was not performed.
Conclusion The true AP view of the GH joint is more sensitive than the conventional shoulder AP view in detecting abnormal findings (especially GT sclerosis and osteophytes) suggesting rotator cuff tears. In addition, the superiority of the GH AP view over the conventional AP view is more evident in medium-sized tears.
Acknowledgments The authors give special thanks to Sun Woo Kim, PhD, and Min-Ji Kim for statistical assistance.
Disclaimer The authors, their immediate families, and any research foundations with which they are affiliated have not received any financial payments or other benefits from any commercial entity related to the subject of this article.
References 1. Ballinger PW. Merrill’s atlas of radiographic positions and radiologic procedures. St Louis: Mosby; 1986. 2. Bayne O, Bateman JE. Long term results of surgical repair of fullthickness rotator cuff tears. In: Bateman JE, Welsh P, editors. Surgery of the shoulder. St Louis: Mosby; 1984. p. 167-71. 3. Bernageau J. Roentgenographic assessment of the rotator cuff. Clin Orthop Relat Res 1990:87-91.
907 4. Bonsell S, Pearsall AWIV, Heitman RJ, Helms CA, Major NM, Speer KP. The relationship of age, gender, and degenerative changes observed on radiographs of the shoulder in asymptomatic individuals. J Bone Joint Surg Br 2000;82:1135-9. 5. Buchbinder R, Green S, Youd JM, Johnston RV, Cumpston M. Arthrographic distension for adhesive capsulitis (frozen shoulder). Cochrane Database Syst Rev 2008:CD007005. http://dx.doi.org/10. 1002/14651858.CD007005 6. Cohen RB, Williams GR Jr. Impingement syndrome and rotator cuff disease as repetitive motion disorders. Clin Orthop Relat Res 1998:95-101. 7. Ellman H, Hanker G, Bayer M. Repair of the rotator cuff. End-result study of factors influencing reconstruction. J Bone Joint Surg Am 1986;68:1136-44. 8. Fleiss JL. Statistical methods for rates and proportions. 2nd ed. New York: Wiley; 1981. p. 211-236. 9. Hamada K, Fukuda H, Mikasa M, Kobayashi Y. Roentgenographic findings in massive rotator cuff tears. A long-term observation. Clin Orthop Relat Res 1990:92-6. 10. Hardin JW. Generalized estimating equations (GEE). Wiley Online Library; 2005. http://dx.doi.org/10.1002/0470013192.bsa250. 11. Lipsitz SR, Kim K, Zhao L. Analysis of repeated categorical data using generalized estimating equations. Stat Med 1994;13:1149-63. 12. Norwood LA, Barrack R, Jacobson KE. Clinical presentation of complete tears of the rotator cuff. J Bone Joint Surg Am 1989;71: 499-505. 13. Nove-Josserand L, Boulahia A, Levigne C, Noel E, Walch G. Coracohumeral space and rotator cuff tears [in French]. Rev Chir Orthop Reparatrice Appar Mot 1999;85:677-83. 14. Nove-Josserand L, Levigne C, Noel E, Walch G. The acromio-humeral interval. A study of the factors influencing its height [in French]. Rev Chir Orthop Reparatrice Appar Mot 1996;82:379-85. 15. Ozaki J, Fujimoto S, Nakagawa Y, Masuhara K, Tamai S. Tears of the rotator cuff of the shoulder associated with pathological changes in the acromion. A study in cadavera. J Bone Joint Surg Am 1988;70: 1224-30. 16. Pearsall AW IV, Bonsell S, Heitman RJ, Helms CA, Osbahr D, Speer KP. Radiographic findings associated with symptomatic rotator cuff tears. J Shoulder Elbow Surg 2003;12:122-7. http://dx.doi.org/10. 1067/mse.2003.19 17. Petersson CJ, Redlund-Johnell I. The subacromial space in normal shoulder radiographs. Acta Orthop Scand 1984;55:57-8. 18. Post M, Silver R, Singh M. Rotator cuff tear. Diagnosis and treatment. Clin Orthop Relat Res 1983:78-91. 19. Sanders TG, Jersey SL. Conventional radiography of the shoulder. Semin Roentgenol 2005;40:207-22. 20. Saupe N, Pfirrmann CW, Schmid MR, Jost B, Werner CM, Zanetti M. Association between rotator cuff abnormalities and reduced acromiohumeral distance. AJR Am J Roentgenol 2006;187:376-82. http:// dx.doi.org/10.2214/AJR.05.0435 21. Umans HR, Pavlov H, Berkowitz M, Warren RF. Correlation of radiographic and arthroscopic findings with rotator cuff tears and degenerative joint disease. J Shoulder Elbow Surg 2001;10:428-33. http://dx.doi.org/10.1067/mse.2001.117123 22. Werner CM, Conrad SJ, Meyer DC, Keller A, Hodler J, Gerber C. Intermethod agreement and interobserver correlation of radiologic acromiohumeral distance measurements. J Shoulder Elbow Surg 2008; 17:237-40. http://dx.doi.org/10.1016/j.jse.2007.06.002 23. Wohlwend JR, van Holsbeeck M, Craig J, Shirazi K, Habra G, Jacobsen G, et al. The association between irregular greater tuberosities and rotator cuff tears: a sonographic study. AJR Am J Roentgenol 1998;171:229-33.