Defining tennis elbow characteristics – The assessment of magnetic resonance imaging defined tendon pathology in an asymptomatic population

Abstract

Background

This radiological study aims to assess the prevalence of lateral elbow pathology in an asymptomatic population using 3.0T magnetic resonance imaging (MRI).

Methods

Bilateral elbow MRI was undertaken in 30 asymptomatic volunteers. Exclusion criteria included elbow pain within 3 months, elbow trauma or previous diagnosis of lateral epicondylar tendinopathy. Baseline patient-reported outcome measures were recorded along with age and body mass index (BMI). Two musculoskeletal radiologists independently graded the degree of abnormality at the common extensor tendon.

Results

Thirty volunteers were categorised according to age; 35–44 (n = 10), 45–54 (n = 11), and 55–65 (n = 9) with a 1:1 male-to-female ratio. Radiological evidence of tendon abnormality was found in 37% of volunteers. The proportion with abnormal findings increased with age; 35–44 (10%), 45–54 (36%), 55–65 (67%) and BMI; 18–24.9 (23%), 25–29.9 (43%), > 30 (67%). Changes were generally ‘mild’ or ‘moderate’, with a single volunteer showing ‘severe’ pathology. Kappa for the radiographic agreement was 0.91 (0.83–0.98).

Discussion

This study has demonstrated MRI findings suggestive of pathology at the common extensor tendon to be prevalent in an asymptomatic population, increasing with age and BMI. This draws into question the diagnostic and prognostic value of MRI imaging in lateral epicondylar tendinopathy, especially in older patients.

Introduction

Lateral epicondylar tendinopathy (LET), known colloquially as ‘Tennis Elbow’, is a common condition that presents with pain on the outer aspect of the elbow. It has an estimated prevalence of 3% and most frequently afflicts those between the ages of 35 and 65 years. ¹ The condition is often self-limiting; however, 20% experience disabling pain with resultant restriction in their professional and social lives. The high prevalence of the condition within a working-age population imparts a significant socioeconomic burden. ²

LET is traditionally a clinical diagnosis based on history and examination. However, expensive modalities such as magnetic resonance imaging (MRI) are increasingly being utilised to gauge the severity of disease and justify decisions about treatment, including surgical intervention. ³

Changes in the common extensor tendon (CET) such as high signal intensity, oedema, thinning, thickening and tears in the tendon can be reliably identified on MRI.^4,5 These changes are identified in nearly all cases of symptomatic LET and have been shown to be associated with both surgical and histopathological findings.^6–8 However, defining a clear correlation between the degree of change on MRI and clinical symptoms has not been possible. The diagnostic and prognostic value of MRI findings in LET has been repeatedly challenged, with a major criticism being a lack of evidence on the prevalence of these imaging findings in an asymptomatic population.^9–12 To understand the true value of MRI, and create standards for the interpretation of imaging in LET, an evidence base of normative data is required.

In comparison, the prevalence of rotator cuff pathology within the shoulder in an asymptomatic population has been well described, with tears being reported in over 30% of patients aged over 60 and over 50% in those aged over 80.^13–15 This information has been invaluable to the shoulder community, and these three papers alone have been cited over 2000 times. A similar body of evidence is lacking for chronic soft tissue disorders of the elbow.

Several studies using ultrasonography to assess the common extensor tendon of asymptomatic volunteers have reported that abnormal findings are not uncommon.^16–19 Krogh et al. found positive colour Doppler activity, regarded as a sonographic characteristic of tendinopathy, in 9% of 264 asymptomatic adults aged between 20 and 96. The same study measured CET tendon thickness, another common finding in tendinopathy, but found no age-related differences. MRI, however, has been shown to be more accurate than ultrasound in diagnosing tendon thickening and tendinopathy. ²⁰ Despite the perceived superiority of MRI over ultrasound, there is a paucity of literature that features the use of MRI imaging in asymptomatic elbows.

The primary objective of this study was to obtain data on the prevalence of radiographic changes associated with tennis elbow in an asymptomatic population. A secondary objective was to explore associations with age, gender, body mass index (BMI) and hand dominance.

Methods

This prospective study was conducted from February to December 2021 after obtaining approval from the HRA and HRCW (REC 19/LO/1995, IRAS 245692). Funding was received from the British Elbow and Shoulder Society pump priming grant.

Minor amendments to the protocol were made due to the SARS-CoV-2 pandemic. The purpose-built MRI facility is located separately from the acute hospital. However, to limit public visitors to the site during the SARS-CoV-2 pandemic, the study population consisted entirely of NHS employees. Thirty healthy volunteers meeting the inclusion criteria, as shown in Figure 1, were recruited to participate in the study, with an equal split of males and females. Participants were purposively recruited by age and were categorised into three equal-sized groups chosen to represent the most common age of presentation of LET.

Figure 1.

Inclusion criteria.

Each participant made a single visit to undergo 3.0T MRI scanning of both elbows. On arrival, participants completed a questionnaire recording demographic information (age, gender, hand dominance, occupation, height, weight) and two standardised Patient Reported Outcome Measures, one region specific and the other a general health-related quality of life measure (Quick DASH and EQ5D).

The Quick Disabilities of the Arm Shoulder and Hand (Quick DASH) is an 11-item patient-reported questionnaire. ²¹ It has been designed as a region-specific measure of disability and symptoms in people with any or multiple musculoskeletal disorders of the upper limb and is an abbreviated version of the DASH score. It has three subscales: function, symptoms and psychosocial impact. Each item is scored from 1 (‘no difficulty’, ‘no symptoms’, ‘no impact’) to 5 (‘unable to do’, ‘very severe’, ‘high impact’). The composite score has been deemed unidimensional and therefore the items are summed to form a raw score and then converted to a 0 to 100 scale where a higher score reflects greater disability. Quick Dash has been shown to be the best-performing instrument in LET. ²²

The EQ-5D-5L is a standardised patient-reported questionnaire for the measurement of generic health status. ²³ It is one of the most commonly used generic health measures and can be converted to UK-specific value index scores using the EUROQOL Index Value Calculator.

BMI was calculated as height (metres) divided by weight (kilograms) squared and categorised as healthy (18–24.9), overweight (25–29.9) and obese (> 30).

MRI images were independently reviewed by two fellowship-trained musculoskeletal radiologists. The degree of change at the CET was graded using a previously published classification system described by Cha et al. ²⁴ as shown in Table 1. The weighted kappa for inter-rater agreement between the reporting radiologists was calculated as 0.91 (95% CI = 0.83–0.98) demonstrating “almost perfect agreement” when characterised by Landis and Koch’s interpretation. ²⁵ Where there was disagreement, images were later discussed between reviewers to reach a consensus. Data are presented descriptively as proportions and mean (± standard deviation), analysis was performed in SPSS (IBM Corp. Released 2021. IBM SPSS Statistics for Windows, Version 28.0. Armonk, NY: IBM Corp).

Table 1.

MR classifications of the common extensor tendon and ligament injuries (Cha et al. ²⁴ ).

‘Injury’ degree	Common extensor tendon
0 Normal	Complete homogenous low signal intensity without tendon thickening
1 Mild	Tendon thickening or thinning with increased internal signal intensity on fat-suppressed T2 image
2 Moderate	A fluid-filled gap affecting 20–80% of the thickness
3 Severe	A fluid-filled gap affecting more than 80% of the thickness

MR: magnetic resonance.

Results

Thirty volunteers were initially recruited to participate, of these two could only tolerate a single elbow MRI due to discomfort in the scanner. Another volunteer was unable to start the scan due to claustrophobia, and an additional volunteer from a different age category was recruited as a replacement. As a result, 30 volunteers and 58 elbows were scanned with an age distribution as follows: 35–44 (n = 10), 45–54 (n = 11) and 55–65 (n = 9), the age distribution was due to a purposive recruitment strategy to mirror the distribution of LET in a general population. Further demographic information is shown in Table 2. Volunteers had a range of occupations within the NHS, which are outlined in Table 3.

Table 2.

Participant demographics.

	Number of volunteers (by age group)
	Total	35–44 years	45–54 years	55–65 years
Sex
Male	15	5	6	4
Female	15	5	5	5
BMI
Healthy	13	6	4	3
Overweight	14	4	5	5
Obese	3	0	2	1
Mean BMI (±SD)	25.6 (±4.0)	23.4 (±3.2)	26.4 (±3.2)	26.9 (±4.9)
Handedness
Right	29	9	11	9
Left	1	1	0	0
Mean age, years (±SD)	49.0 (±8.6)	39.0 (±1.8)	49.5 (±3.4)	59.4 (±2.0)

BMI: body mass index.

Table 3.

Participant occupations.

Occupation	Number of volunteers
	Male	Female
Administration	3	5
Nursing staff	2	5
Surgeon	4	0
Physician	3	0
Housekeeper	0	1
Physiotherapist	0	1
Laboratory staff	1	3
Paramedic	1	0
Vaccination volunteer (retired post office worker)	1	0

Quick DASH score was zero in 28/30 concordant with excellent overall upper limb function and absence of pain. Two participants reported occasional, mild shoulder pain. The mean EQ-VAS score was 91 suggesting good overall health.

Eleven of 30 volunteers (37%) had abnormal MRI changes in one or both elbows. In six cases, changes were seen bilaterally meaning that a total of 17 elbows (29%) had abnormal changes on MRI. Most of the reported changes were mild or moderate according to the classification used, however, one volunteer had severe changes (Table 3). Examples of mild, moderate and severe changes are shown in Figure 2.

Figure 2.

Examples of mild (a), moderate (b) and severe (c) magnetic resonance imaging (MRI) findings at the common extensor tendon in asymptomatic volunteers.

The proportion of volunteers and elbows with changes on MRI increased with age, as shown in Table 4. A similar trend was seen with increasing BMI. There were seven male volunteers with changes on MRI compared to four females.

Table 4.

Abnormal MRI findings at the CET in asymptomatic volunteers.

	Number of elbows with an abnormal finding			Elbows with abnormal findings (%)	Volunteers with abnormal findings (%)
	Grade 1 (mild)	Grade 2 (moderate)	Grade 3 (severe)
Total	7	9	1	17/58 (29%)	11/30 (37%)
Age
35–45 (n = 10)	1	0	0	1/19 (5%)	1/10 (10%)
45–55 (n = 11)	2	3	0	5/21 (24%)	4/11 (36%)
55–65 (n = 9)	4	6	1	11/18 (61%)	6/9 (67%)
BMI
Healthy (n = 13)	2	3	0	5/25 (20%)	3/13 (25%)
Overweight (n = 14)	3	6	1	10/27 (37%)	6/14 (43%)
Obese (n = 3)	2	0	0	2/6 (33%)	2/3 (67%)

BMI: body mass index; CET: common extensor tendon; MRI: magnetic resonance imaging.

Discussion

This study reports the elbow MRI findings of 30 prospectively recruited participants with no history of elbow-related symptoms. We have found changes around the CET to be common in this cohort (37%) and prevalence increases with age, suggesting a degenerative process. We also observed a tendency for these findings to be bilateral, increase with BMI, and occur in a greater proportion of males than females.

In their systematic review in 2001, Pasternack et al. ¹² presented findings from seven studies that featured MRI of either asymptomatic contralateral elbows in LET sufferers or elbows of healthy control volunteers. Of 29 healthy elbows in total, 14% showed a high signal at the CET however the authors concluded that small sample sizes, study heterogeneity and methodological shortcomings resulted in uncertainty related to these findings. This review highlighted the need for well-designed studies on the topic, suggesting more carefully documented imaging parameters would be beneficial. Since then, there have been few papers that add to this discussion and no dedicated studies to prospectively determine the prevalence of lateral elbow pathology on MRI in the asymptomatic population.

In 2003, two small studies by Savnik et al. ²⁷ and Mackay et al. ²⁶ reported high signal intensity at the CET in 14% of healthy controls and signs of oedema in 35% of asymptomatic contralateral elbows of LET patients respectively. Both Van Kollenburg et al. ⁹ and Van Leeuwen et al. ²⁸ have since used retrospective data from elbow MRIs performed for other causes of elbow pain to estimate the rate of asymptomatic LET findings, suggesting that 19% and 11%, respectively of those without LET symptoms had signal change consistent with tendinopathy at the ECRB origin. In both these retrospective studies, the authors acknowledge that their control group was not truly asymptomatic and therefore the results should be interpreted with caution, the latter also included patients as young as 18 years of age.

The existing literature reports abnormal changes are seen on elbow MRI in the asymptomatic population. However, the overarching inference from previous studies is that further research on the clinical use of MRI in LET is required. Our study aimed to overcome the limitations described in previously published work, through the use of a well-designed methodology and prospective recruitment of patients. Our results found a higher prevalence of abnormal changes in the LET than in the majority of previous studies. This may be due to the systematic interpretation of radiological changes using clear imaging parameters. Furthermore, the mean age of patients in our study is higher than that of previous studies and targeted at an age associated with the highest prevalence of LET symptoms.

The term angiofibroblastic dysplasia has been used to describe the histological changes seen in the CET of symptomatic patients, which include granulation tissue, vascular hyperplasia, micro-rupture, abundant fibroblasts, and unstructured collagen.^24,29 Given that these findings have been shown to correlate with changes on MRI, ⁶ our results suggest that they are not just relevant in LET pathology but may also be found in asymptomatic individuals.

Knowledge of the effects of ageing on the microscopic appearance of the CET is limited, however, it may be reasonable to assume that the processes occurring in the CET over time are similar to those seen in other tendons. ³⁰ In vitro data suggest that cell proliferation and turnover within tendons are reduced with ageing while there is an accumulation of advanced glycation end-products (AGEs) resulting in reduced strength and modulus. ³⁰ Chard et al. ³¹ showed in their comparative histological study that the microscopic features of the CET of patients with symptomatic LET mirror those in a normal aged rotator cuff tendon. As mentioned previously, it has been shown that rotator-cuff lesions are a natural correlate of ageing and are often present without clinical symptoms. ¹³ As we might therefore expect, our results show that radiographic changes at the lateral elbow, including moderate and severe defects in the CET, are more common with increasing age and will often exist without symptoms.

Obesity has been consistently shown to increase the risk of tendinopathy and tendon injury. ³² This study suggests a possible association between MRI changes in the CET and increasing BMI, however, this should be interpreted with caution in the setting of low numbers in the obese category (n = 3) and because BMI tended to increase with age. This is an area that warrants further study and an adequately powered, prospectively designed study is recommended to facilitate statistical analysis of the association between BMI and asymptomatic MRI changes in the CET.

The relationship between handedness and symptomatic LET is unclear, although more than one study has shown dominant side involvement to be a risk factor. ³³ Our study found CET changes were most commonly bilateral (as shown in Table 5) and those with MRI findings in one elbow were more than 50% likely to show a change in the contralateral side. This is comparable with the rotator cuff, for which data suggests bilateral tears are prevalent in more than 50% of the population aged over 66. ¹⁵ In both joints, these findings are consistent with the notion that radiographic tendon changes are associated with natural ageing.

Table 5.

Distribution of patients with abnormal MRI findings at CET by handedness and sex.

	Handedness			Sex
	Dominant	Non-dominant	Bilateral	Male	Female
n =	2/11	3/11	6/11	7/11	4/11
%	18	27	55	64	36

BMI: body mass index; CET: common extensor tendon.

It is important to note that the type and frequency of physical activity, both in the workplace and recreationally, may have a role in the incidence of positive findings in the asymptomatic patient. There is strong evidence to support an association between occupational exposure to the certain wrist and elbow biomechanics and the incidence of symptomatic LET. ³⁴ The majority of the occupations listed in Table 3 could be considered ‘low risk’ in this regard with the exception of a surgeon, for which frequent movements of the wrist and elbow are inherent.

Among our asymptomatic volunteers, 4/15 were practicing orthopaedic surgeons, all of whom were male. Two of four (50%) had abnormal MRI findings; one mild, one moderate. Increased rates of LET and other musculoskeletal conditions among orthopaedic surgeons have been described previously in the literature.^35–37

In our study male volunteers appeared almost twice as likely to have asymptomatic MRI findings than females (Table 5). This is contrary to several population studies which have shown a higher prevalence of symptomatic LET in females than males,^38–40 despite others suggesting there is no significant association between LET and sex.^41,42 In general, epidemiological and clinical findings demonstrate a greater prevalence of musculoskeletal pain conditions in women than men, and that females will often report more severe symptoms; with various biological and psychosocial theories supporting this.^43–46 There is evidence from different populations to suggest females are also more likely to engage with and use healthcare services.^47–49 A combination of these factors, and differences in occupation related to gender as described above, could explain why our study of asymptomatic volunteers shows a greater proportion of positive findings in men, compared with previous studies into the prevalence of symptomatic LET in the population.

There are several limitations to this study including its small sample size and a relative lack of diversity within the study group, all of whom are NHS employees. The variety of occupations among participants and a lack of information about recreational activities affect the validity of our conclusions. Concurrent assessment of the asymptomatic volunteers with an ultrasound scan, which was initially planned but not achievable due to the SARS-CoV-2 pandemic, may have increased the utility of our observations by facilitating a useful comparison between the two imaging modalities. Another limitation was a lack of blinding; both reviewing radiologists were involved in the study design and were aware that the volunteers were asymptomatic. A controlled study comparing the incidence of MRI findings in symptomatic LET patients with healthy matched controls would further enhance an exploration of the true prevalence of imaging-defined tendon abnormalities.

Despite limitations, our findings are relevant, and having this information in the literature will have implications for clinical practice, not least by helping clinicians better interpret MR images. Attempts have previously been made to categorise the severity of LET using histopathological changes, clinical symptoms, and functional assessment, but no classification has been successful in predicting treatment outcomes.^50,51 Establishing an accurate estimation of the prevalence of asymptomatic CET pathology will help clinicians and allied professionals in the holistic assessment and management of lateral elbow pain. Whilst mild and moderate changes were common in our study only one volunteer had severe changes and so when interpreting an MRI of the elbow, a higher cut-off for defining meaningful radiographic pathology might be a more specific tool for predicting response to therapies.

This paper represents the largest study to date to prospectively determine the prevalence of MRI changes associated with LET in an asymptomatic population. The results show that MRI changes at the common extensor origin are common in an asymptomatic population. Our findings challenge the role of MRI in the assessment of LET. It is recommended that MRI should be used cautiously in this context and only as part of a comprehensive clinical assessment.