MR findings in humeral epicondylitis. . A systematic review

Copyright C Acta Radiologica 2001

Acta Radiologica 42 (2001) 434–440 Printed in Denmark ¡ All rights reserved

AC TA R A D I O L O G I C A ISSN 0284-1851

MR FINDINGS IN HUMERAL EPICONDYLITIS A systematic review I. P1, E.-M. T2, M. L3, T. V4 and A. M1 1

Finnish Institute of Occupational Health, 2Medivire, Occupational Health Services, 3Department of Radiology, Helsinki University Central Hospital and 4Radiology Unit, Finnish Institute of Occupational Health, Helsinki, Finland.

Abstract

Purpose: To highlight the importance of meta-analysis in diagnostic imaging by presenting a systematic search of the literature on the accuracy of MR imaging in epicondylitis. Material and Methods: The literature was comprehensively reviewed to identify studies on MR findings in epicondylitis. Reviewers blind to the clinical diagnoses screened the data according to predetermined inclusion criteria. Data were collected and validity and relevance were assessed on structured forms. Results: Seven studies including 148 patients with epicondylitis were accepted for the analysis. Eleven asymptomatic contralateral elbows and 29 elbows of healthy volunteers served as controls. The volunteers were distinctly younger than the patients. The MR technique was divergent, and the observed pathological changes also varied. The most frequent alteration was a change in the common extensor tendon signal (90%, 95% confidence interval 84–94%); 14% of the healthy volunteers and 50% of the contralateral elbows displayed the similar alteration. Conclusion: Small sample size and methodological shortcomings in the original studies make the assessment of MR findings in epicondylitis questionable. There is a need for well-designed studies in which clinical features and occupational backgrounds as well as imaging parameters are carefully documented.

Considering the growing demands and limited resources of health care systems reliable data are essential both in clinical and policy decisions (1). The number of medical publications is increasing exponentially and different studies on the same issue may produce conflicting results. It is difficult to keep abreast of the development even in one’s own field on the basis of clinical research data. Review articles provide the opportunity to cope with this demand (19). Traditional reviews are usually conducted by experts on the field in question. Most reviews are unsystematic, narrative summaries in which informal and subjective methods are used to collect and interpret the information. The current interest in evi434

Key words: Tennis elbow, tendinitis; diagnostic imaging, MR imaging; meta-analysis. Correspondence: Iris Pasternack, Finnish Institute of Occupational Health, Topeliuksenkatu 41a A, FIN-00250 Helsinki, Finland. FAX π358 9 77 50 87 99. Accepted for publication 30 March 2001.

dence-based medicine has led to an increase in the publication of systematic reviews. In a systematic review, the study subjects are original studies rather than patients. A meta-analysis is a systematic review where a statistical combination of the results of separate studies is used to produce a single estimate of the considered question (17). Humeral epicondylitis is a common, painful condition of the elbow and forearm, with an incidence of 4–7 per 1,000 per year in general practice (2, 10). Affected tissues are the tendons and muscles of the forearm. The exact pathogenesis of the disorder is debated (7, 16). Forceful and repetitive twisting, squeezing or flexion-extension movements of the fingers or the wrist are assumed to be

HUMERAL EPICONDYLITIS

risk factors (9). Still it is a matter of dispute to which extent the disorders are caused by physical workload (3, 13). It has been postulated that tendon and muscle-related diseases of the upper extremities account for more than half of the recorded work-related musculoskeletal ailments in a working population (13). The diagnosis of epicondylitis is based on patient history and physical examination. Further evaluation is needed only when conservative treatment is unsuccessful. MR imaging has been considered to be a useful diagnostic modality both in acute and chronic elbow disorders due to its excellent soft tissue resolution and sensitivity in detecting oedema and haemorrhage (11). On the other hand, asymptomatic structural abnormalities are also detected on MR (16). We undertook this study to promote evidencebased thinking in clinicians and researchers in the field of radiology. We chose to conduct a systematic review to evaluate the accuracy of MR imaging in epicondylitis, which we consider a relatively clear-cut clinical entity. Material and Methods

The method of this systematic review was recommended by the C M W G  S R  S  D T (5). In the protocol we defined the objective of our study and the means of literature retrieval as well as inclusion and exclusion criteria for the studies. Structured forms for data extraction, clinical relevance and validity assessment were prepared. We searched the Medline database from 1966, Embase from 1990, Nioshtic from 1990 to December 1998 and Current Contents until 12/99. The search strategy combined the text word ‘‘epicondylitis’’ and Medical Subject Headings (MeSH) terms ‘‘tennis elbow’’ or ‘‘tendinitis’’ and ‘‘elbow’’ with ‘‘magnetic resonance imaging’’. We screened the reference lists of the pertinent studies and reviews and contacted specialists active in the field to supplement our search with additional original articles and unpublished data. We searched manually the following journals from 1990 to December 1998: Radiology, American Journal of Roentgenology and the RSNA Index to Imaging Literature. We included in our systematic review all studies in which MR imaging was used to evaluate the elbow anatomy and pathology in patients with epicondylitis. We excluded review articles, case reports, and studies on normal anatomy. Two reviewers (I.P., E.T.) screened the studies indepen-

dently, first by titles and abstract and finally by the whole text. Disagreements were solved by consensus. Three reviewers (I.P., E.T., M.L.) extracted the data independently using structured forms for clinical and demographic data, technical factors for MR imaging and the imaging findings (Tables 1– 3). The following items were noted for clinical relevance: The number of participants and drop-outs, their age and occupation, co-morbid conditions, the severity of the disease, way of referral and invasive treatments of elbow disorders prior to MR investigation. For internal validity of the studies, we considered the study type, the use of a valid reference standard (in this case a proper clinical diagnosis), blinding of assessments, comparability of controls and inter- and intra-observer agreement of assessments. Sensitivity, specificity and accuracy were calculated whenever possible. The 95% confidence intervals (CI) for proportions were computed using the binomial distribution. Results

Our searches in electronic literature databases disclosed 25 articles in Medline, an additional 11 in Embase, and 1 more in Nioshtic. Of these 37 articles, we selected 24 possibly relevant studies as complete text articles. From the references herein we acquired 8 more articles. Hand searches and personal communication did not bring any further studies. Only 5 out of the 32 studies met our inclusion criteria (6, 12, 14, 15, 20). Later on, we included 2 more studies from the 1999 update of the Medline database (4, 18). The 7 included studies were published in 1989– 1999. There were 1 retrospective and 2 prospective case-control studies (6, 12, 18). In another 2 prospective studies, the MR findings were set against the histopathological findings of the operated cases (4, 14). Two studies were retrospective case series (15, 20). There were a total of 148 patients, 5–35 in each study, with 144 cases of lateral and 4 of medial epicondylitis (Table 1). Twenty-nine healthy elbows in 26 volunteers and 11 symptomfree contralateral elbows of the epicondylitis patients served as controls. The age range of the patients was 22–71 years, and that of the controls was 22–46 years. The weighted mean of the given mean ages were 44 years for patients and 29 years for controls. All cases were subacute or chronic and the symptoms had persisted at least for 1 month. Except for 20 patients playing racket sports (15) there was no description of the patients’ occupation or leisure-time activities. Local cortisone injections given within 2–3 months prior to the MR 435

436 ª

2 mª10 y

6 had pain on the lat. epicondyle and 1 at the tip of the olecranon

ª

ª

ª

1 operated on

Occupation/exercise

Duration of disorder: years/months

Severity: symptoms, clinical findings

Invasive treatment prior to MR

ª not stated.

ª

36ª51

ª

Age, years

Other disorders

5/2

ª

Sex M/F

ª

ª

ª

3

controls: healthy volunteers

none had no history of gout or Caelbow injury pyrophosphate arthropathy

no steroid injection within 2 months

7 lat.

patients

Coel et al. 1993 (ref. 6)

3 lat. 2 med.

Participants with epicondylitis, n

patients

Zwaan et al. 1989 (ref. 20)

Table 1

ª

no steroid injection within 3 months

ª

6 mª4 y

most played racquet sports, no professional athletes

35ª71, mean 44

17/3

20 lat.

patients

Potter et al. 1995 (ref. 15)

ª

ª

no local injections

ª

1 mª7 y, mean 17.1 m

⬎3 m

ª

no steroid or local anaesthesia injections within 2 months

ª

ª

29ª58, mean 47

8/15

23 lat.

patients

ª

ª

22ª29, mean 25

2/5

7

controls: healthy volunteers

ª

ª

29ª58, mean 47

ª

11

controls: contralateral elbow

Steinborn et al. 1999 (ref. 18)

ª

33ª48

11/23

34 lat.

patients

burning pain over the epicondyle

no history of elbow disorders

2 professional athletes

22ª46, mean 31

ª

16 (19 elbows)

controls: healthy volunteers

Pfahler et al. 1998 (ref. 14)

ª

1 mª2 y

ª

29ª62, mean 38

ª

22 lat. 2 med.

patients

Martin & Schweitzer 1998 (ref. 12)

Patient characteristics in the included studies

ª

no steroid injections within 3 months

ª

3 mª3 y

ª

22ª63, mean 45

17/18

35 lat.

patients

Bredella et al. 1999 (ref. 4)

I. PASTERNACK ET AL.

HUMERAL EPICONDYLITIS

Table 2 Technical factors affecting MR imaging in the included studies Zwaan et al. 1989 (ref. 20)

Coel et al. 1993 (ref. 6)

Potter et al. 1995 (ref. 15)

Martin & Schweitzer 1998 (ref. 12)

Field strength, T

1.5

1.5

1.5

1.5

0.2

0.2

Patient position

elbow above head

elbow above head

elbow at side

elbow above head

ª

ª

ª

MR equipment

Siemens Magnetom

ª

GE Signa

GE Signa

ª

Arthroscan Esaote

GE Signa (nΩ30); Prima Elscint (nΩ3); Arthroscan Esaote (nΩ2)

shoulder coil

ª

unpolarized solenoidal

commercial elbow coil

Coil type

extemity coil, transmitterπ receiver

dedicated 5-inch receiver extremity coil only surface coil

Pfahler et al. Steinborn et al. 1998 (ref. 14) 1999 (ref. 18)

Bredella et al. 1999 (ref. 4) 1.5 (nΩ30); 1.0 (nΩ3); 0.2 (nΩ2)

Slice thickness, mm

4

3

1.5ª4

4

ª

ª

4; 3.5; 4

FOV, cm

ª

ª

12ª13

12ª15

ª

ª

12; 14; 14

Pulse sequences STIR

no

yes

no

yes

no

no

T1/FSE T2/FSE proton-weighted FSE

yes yes yes

no no no

no yes yes

yes yes no

yes yes no

yes yes no

frequency-selective fat-suppression technique GRE

no

no

yes

yes

no

no

no

no

yes

no

no

no

no

no

no

no

no

yes

yes

no

3

1

3

2

2

2

3; 3; 3

I.v. Gd Directions, n

18 no and 12 yes; yes; yes yes; yes; no yes; yes; yes 18 no and 12 yes; yes; yes yes; no; no

ª not stated.

examination was an exclusion criterion in 5 studies (4, 6, 14, 15, 18). Both the coils and sequences and positioning of patients for MR varied (Table 2). The standard position of the patient was supine with the arm extended above the patient’s head. In 1 study the arm was at the side, and in 3 of the studies the position was not stated. High-field MR examination at 1.5 T was used in 4 studies (6, 12, 15, 20), low-field MR at 0.2 T in 2 studies (14, 18) and in 1 study three different field strengths, 1.5 T, 1.0 T and 0.2 T, were used (4). Slice thicknesses from 1.5 to 4 mm were applied. The field-of-view (FOV) reported in 3 studies (4, 12, 15) was 12–15 cm. In all except 1 of the studies, either T1-, T2- or proton-weighted spin-echo sequences were used. Fat-suppression images, either by means of the frequency-selective fat-suppression technique or by STIR were obtained in 4 studies (4, 6, 12, 15). The most frequent MR finding was a signal increase in the common extensor tendon. This sign was reported in 90% of the patients (CI 84–94%),

in 14% (CI 4–32%) of the healthy controls and in 50% (CI 23–83%) of the contralateralal elbows. Thickening or thinning of the common extensor tendon was reported in 55% (CI 46–63%) of the patients as compared to 3% (CI 0–18%) of the healthy controls and 27% (CI 6–61%) of the contralateral elbows. Abnormalities in the radial collateral ligament were seen in almost half of the patients in 2 of the studies (47%, CI 34–61%) (4, 15). Various other signs were seen sporadically (Table 3). Clinical relevance of the included studies: There was no description of severity of the disease and the only information available was the apparent chronicity of the condition. Only limited demographic data, mainly age and sex distribution, were presented in the studies. The characteristics of the control groups remained unclear as well. The description of the MR method was adequate for reproducing the imaging in 3 of the studies (4, 12, 15). Validity of the included studies: In all of the 4 437

438

ª not stated. ? data not available.

Tendon signal alteration Thickening of tendon Thinning of tendon Tendon enhancement after i.v. Gd M. anconeous oedema Bone marrow oedema Chondral degeneration/defects Periosteal reaction Fluid in/about bursa capituli radii Fluid adjacent to lateral epicondyle Abnormal lig. collat. radii Radial separation (separation between extensor carpi radialis brevis and radial collat. ligament) Non-specific alterations in artic. cubiti Degenerative bone alterations Joint effusion

Findings searched for/found in

ª ª ª ? 0/3 ª ª ª ª ª ª ª ª ª ª

ª ª ª ? 7/7 0/7 ª ª 4/7 2/7 ª ª ª ª ª

ª ª ª ? ª ª ª ª ª ª ª ª 0/5 ª ª

healthy volunteers

patients

Coel et al. 1993 (ref. 6)

patients

Zwaan et al. 1989 (ref. 20)

ª ª ª

20/20 ª 14/20 ? ª ª 3/20 ª ª ª 4/20 10/20

patients

Potter et al. 1995 (ref. 15)

ª ª ª

24/24 19/24 ª ? 2/24 3/24 ª 0/24 0/24 ª ª ª

patients

ª ª ª

3/19 1/19 ª ? ª ª ª ª ª ª ª ª

healthy volunteers

Martin & Schweitzer 1998 (ref. 12)

ª ª ª

32/34 ª ª 21/34 ª ª ª ª ª ª ª ª

patients

Pfahler et al. 1998 (ref. 14)

MR findings in patients and controls in the included studies

Table 3

ª ª ª

22/23 13/23 ª ? 0/23 ª ª ª ª ª ª ª

patients

ª ª ª

1/7 0/7 ª ? ª ª ª ª ª ª ª ª

ª ª ª

6/11 3/11 ª ? ª ª ª ª ª ª ª ª

healthy contralat. volunteers symptom-free elbows

Steinborn et al. 1999 (ref. 18)

ª 12/35 9/35

35/35 35/35 ª ? ª 1/35 4/35 ª ª ª 22/35 ª

Bredella et al. 1999 (ref. 4)

I. PASTERNACK ET AL.

HUMERAL EPICONDYLITIS

Table 4 Validity of included studies Zwaan et al. 1989 (ref. 20)

Coel et al. 1993 (ref. 6)

Potter et al. 1995 (ref. 15)

Martin & Schweitzer 1998 (ref. 12)

prospect. yes

retrospect. no

prospect. yes

prospect. yes

retrospect. yes

prospect. yes

yes

no

yes

unclear

yes

no

ª

?

?

no yes

ª no

yes no

poor (mismatch in age) no no

?

ª no

poor (mismatch in age) yes no

no no

no

no

no

yes

no

no

no

no

no

yes

yes

no

no

yes

no

no

no

no

no

no

no

Study design retrospect. Was the choice of patients independent of ª the MR findings (consecutive, random sample)? Were MR results compared with a valid no reference standard (clinical diagnosis)? Comparability of controls? ?

Blinding of MR assessment? Was the observer agreement of MR assessment determined? Was the reproducibility of the clinical diagnosis determined? Was the MR method described well enough to reproduce the test? Were sensitivity and specificity calculated?

Pfahler et al. Steinborn et al. Bredella et al. 1998 1999 1999 (ref. 14) (ref. 18) (ref. 4)

ª not stated. ? data not available.

prospective studies, consecutive patients with epicondylitis entered the study without any selection (4, 6, 12, 14) (Table 4). Three studies provided an adequate description of the clinical diagnosis as the reference standard (6, 12, 18). In 2 studies, the MR findings were evaluated with the reviewers blind to the clinical diagnosis (12, 14). Kappa values for observer agreement were determined in 1 study (6). Sensitivities or specificities were not calculated in the original articles. Discussion

The increased T2-weighted signal intensity of the common extensor tendon, which reflects oedema and/or fatty degeneration of the tendon, was the most common finding in epicondylitis. Only the 2 oldest studies (from 1989 and 1993) failed to report this sign. Calculated sensitivity for tendon signal increase was 0.90. Specificity was 0.86 when healthy subjects served as controls and 0.65 when contralateral symptom-free elbows served as controls. Thus, the specificity was lowered by the frequent occurrence of signal changes in the contralateral elbow. This may suggest age-related degenerative changes in the common extensor tendon. Due to potential confounding factors, a cautious attitude to these numerical values should be taken. Oedema of the anconeus muscle was found extensively in 1 study (6) but rarely or not

at all in the other studies. The diversity of this sign cannot be explained. Local injections of steroids, which may alter the muscle signal for up to 1 month (8), had not been given within 2 months prior to MR investigation in the above-mentioned study (6). Despite an intensive search, only 7 studies could be included into this systematic review. Common methodological problems were small sample size and poor comparability between patients and healthy control subjects in terms of age and other possible confounders. Insufficient documentation of the patient characteristics weakened the applicability of the results. Detailed description of the MR method was often missing; in 1 recent study only pulse sequences and field strength were stated while all other technical facts were not (14). There is evidence that inadequate detail presentation may increase the estimates of test accuracy. Inconsistent MR methods and variable reporting of findings render combination of the results difficult. Two studies were performed with a dedicated extremity coil, allowing the use of a smaller FOV yielding better image quality. The use of contrast agent in 2 studies was motivated by the assumption that there is enhanced blood flow in an acute inflammatory condition. It is not possible to obtain frequency-selective fat suppression with low-field MR, a fact that may make the interpretation of contrast enhancement more complicated, espe439

I. PASTERNACK ET AL.

cially in fatty surroundings. The commercial name of the MR equipment, not mentioned in 2 studies, might have been of relevance as the results obtained with certain sequences may vary with different equipment. New treatment modalities, especially pharmaceutical, need to go through extensive analysis on efficiency and effectiveness before implementation. Similar assessments of the clinical or patient-centered outcomes or cost-effectiveness should be conducted also for diagnostic technologies. Most patients with epicondylitis show rapid improvement, so that expensive imaging studies might be a waste of resources. However, there are some important reasons for the use of diagnostic modalities that are less observer-dependent than clinical examination alone. First, the follow-up of patients is more objective with documented images. Second, MR imaging may be important for the patients as far as medical insurance is concerned by detecting objectively the specific changes related to epicondylitis. Third, a noteworthy challenge is to recognize at an early stage the cases that are refractory to conservative treatment. Fourth, detailed and repeated anatomical imaging helps us to understand the pathogenesis of this syndrome better. We conclude that there is limited evidence for the usefulness of MR imaging in diagnosing epicondylitis. Instead there is an obvious need for well-designed diagnostic research. The crucial points in the study design are to define the purpose of the study and to state the inclusion and exclusion criteria of the patients, preferably using some consensus classifications for diagnosis. Description of the occupation and other activities beside information on age and sex would make the results more reliable. Blinded assessment is required to avoid verification bias. Careful documentation of the technical parameters is equally important for the credibility of the results. REFERENCES 1. A L., R R., C F. et al.: A scale of methodological quality for clinical studies of radiologic examinations. Radiology 217 (2000), 69. 2. A W. J., H E. M., A R. & B L. M.: Corticosteroid injections for lateral epicondylitis. A systematic overview. Br. J. Gen. Pract. 46 (1996), 209.

440

3. B N. J., H G., N J. & S W. M.: Occupational causes of disorders in the upper limb [see comments]. BMJ 304 (1992), 309. 4. B M. A., T P. F., F R. C., F J. F., W T. K. & G H. K.: MR imaging findings of lateral ulnar collateral ligament abnormalities in patients with lateral epicondylitis. AJR 173 (1999), 1379. 5. C M G  S R  S  D T: Recommended methods, updated 6 June 1996. http://som.flinders.edu.au/ fusa/cochrane/. 6. C M., Y C. Y. & K J.: MR imaging of patients with lateral epicondylitis of the elbow (tennis elbow). Importance of increased signal of the anconeus muscle. AJR 161 (1993), 1019. 7. D A., G G. A., R N. & W C.: Tennis elbow. A clinicopathologic study of 22 cases followed for 2 years. Acta Orthop. Scand. 61 (1990), 535. 8. F R. & S D. (editors): Magnetic resonance imaging in orthopaedics & sports medicine. LippincottRaven, Philadelphia 1997. 9. G A., M G., D V G. & S N.: Epidemiology of musculoskeletal disorders due to biomechanical overload. Ergonomics 41 (1998), 1253. 10. H E. M., P S. M., L M., H G. & C P.: Pragmatic randomised controlled trial of local corticosteroid injection and naproxen for treatment of lateral epicondylitis of elbow in primary care. BMJ 319 (1999), 964. 11. H R. J.: Magnetic resonance imaging of the elbow. Magn. Reson. Q. 9 (1993), 188. 12. M C. E. & S M. E.: MR imaging of epicondylitis. Skeletal Radiol. 27 (1998), 133. 13. N J., J S. & K J. H.: A narrative review on classification of pain conditions of the upper extremities. Scand. J. Rehabil. Med. 31 (1999), 153. 14. P M., J C., S M. & R H. J.: Magnetic resonance imaging in lateral epicondylitis of the elbow. Arch. Orthop. Trauma Surg. 118 (1998), 121. 15. P H. G., H J. A., M R. M., DC E. F., O’B S. J. & A D. W.: Lateral epicondylitis. Correlation of MR imaging, surgical and histopathologic findings. Radiology 196 (1995), 43. 16. R W., W L. E., C R. & M B. F.: Microscopic histopathology of chronic refractory lateral epicondylitis. Am. J. Sports Med. 20 (1992), 746. 17. S H. S., B J., R D., A-B V. A. & C T. C.: Meta-analyses of randomized controlled trials. N. Engl. J. Med. 316 (1987), 450. 18. S M., H A., J C., B H. & R M.: Magnetic resonance imaging of lateral epicondylitis of the elbow with a 0.2-T dedicated system. Eur. Radiol. 9 (1999), 1376. 19.  T M. W., A W. J., K B. W. & B L. M.: Method guidelines for systematic reviews in the Cochrane Collaboration Back Review Group for Spinal Disorders. [Editorial.] Spine 22 (1997), 2323. 20. Z M., R E., W H. D., P M. & V H.: MR-Tomographie des Ellenbogengelenk. Digitale Bilddiagn. 9 (1989), 12.