“Tennis elbow” (TE) is a common tendinopathy affecting the arm. The aetiology and pathogenesis are unknown, but the condition is considered to be overuse injury and is characterised by pain in the muscle origin at the lateral humeral epicondyle (extensor origin). The symptoms of TE occur after overuse of the wrist extensors, whereas overuse of the wrist flexors results in “golfer’s elbow” (GE), a condition characterised by pain in the muscle origin at the medial humeral epicondyle. Diagnostic criteria for TE include pain at the extensor muscle origin, elicited by forced extension, and tenderness upon palpation of the tendinous origin of the muscles at the lateral humeral epicondyle.1 The condition is known to be difficult to treat.2

Innervation of the muscle origin of the extensor carpi radialis brevis (ECRB) muscle has been delineated in a series of studies.3^–5 There is a distribution of sensory nerve fibres in close association with a subpopulation of small blood vessels and sensory nerve fibres also occur as constituents of nerve bundles.4 Furthermore, sympathetic innervation, as evidenced in sections stained for tyrosine hydroxylase (TH) and neuropeptide Y, has also been shown in the muscle origin at the lateral epicondyle, specifically in association with some of the arteries, whereas almost no evidence for parasympathetic innervation of the blood vessels of this tissue has been found.5 The conclusion is that the innervation is heterogeneously distributed, with sensory innervation being confined to smaller vessels while sympathetic innervation is found in arterial walls, and also that vasoconstrictor innervation of the blood vessels (sympathetic) seems to dominate over vasodilator innervation.5

It has previously been shown by ultrasonography (US) and colour Doppler (CD) imaging that there is a high blood flow in the muscle origin at the lateral epicondyle in patients with TE, which was seen in the extensor origin of 21/22 chronically painful elbows, but only in 2/22 pain-free elbows.6 This seems to correspond to another tendinopathic condition, “jumper’s knee” (patellar tendinopathy), in which an association has been noted between the degree of pathological vascularity and the level of pain.7 As with jumper’s knee,8 9 interventional injection therapy targeting regions with high blood flow has been proven to accomplish pain relief in patients with TE.10 11 However, the mechanisms of the treatment and thus the reason for its success are unclear, as is the reason for the high blood flow.

In recent years much research has been devoted to the expression in non-neuronal cells of substances that are traditionally confined to neurons. One such substance that has been closely studied is acetylcholine (ACh), a classic neurotransmitter, which has been found to be produced by various non-neuronal cells and its possible role in the pathogenesis of various diseases has been discussed.12 Other classic neurotransmitters have also been found outside the nervous system, including catecholamines,13 the group of neurotransmitters to which epinephrine, norepinephrine and dopamine belong. In addition, it has been show in human tendon tissue that the tenocytes (tendon tissue equivalents of fibroblasts) contain synthesising enzymes for both ACh and catecholamines and, most interestingly, it has been shown that this is a particularly prominent feature in tendons of patients with chronically painful tendinopathies/tendinosis, this being the case for both patellar14^–17 and Achilles18 19 tendinopathy. This is interesting not least because stimulation of receptors for both catecholamines and ACh can lead to cell proliferation,20 21 interfere with collagen production and accumulation,22 23 modulate pain sensation,24 25 and control vasoregulation, and in the case of adrenergic receptors, can lead to cell degeneration and apotosis.26 27 Such processes are markedly influenced in tendinopathy.28^–31 Whether there is also local production of catecholamines and/or ACh in the connective (tendinous) tissue of the muscle origins in the arm is yet not known, which is a disadvantage when trying to understand the environment in which TE (and GE) occurs. Therefore, the aim of the present study was to investigate whether catecholamines and/or ACh are produced by the cells (fibroblasts) of the tissue at the muscle origin at the lateral humeral epicondyle in patients with TE and in healthy controls. As a reference, the tissue at the muscle origin at the medial humeral epicondyle in patients with GE was also studied to see if the pattern of signal substance production at this muscle origin differs from that on the lateral side of the elbow, as these sites harbour comparable tissues. Immunohistochemistry was used to detect the synthesising enzymes for catecholamines (TH) and ACh (choline acetyltransferase; ChAT).

METHODS

The investigation was approved by the ethics committee of the Medical Faculty, Umeå University, Sweden and the Local Committee on the Use of Human Subjects at Huddinge Hospital, Stockholm, Sweden, and the experiments were conducted according to the principles of the Declaration of Helsinki. All participants gave informed consent.

Participants

Tissue samples

Seven patients who had undergone surgery for TE (four men, three women; mean age 43 years, range 35 to 52) and four for GE (two men, two women; mean age 31 years, range 24 to 40), giving a total of 12 elbows (one woman had bilateral TE) for investigation. The mean duration of pain symptoms was 34 months (range 14 to 84) for TE and 19 months (range 3 to 27) for GE (table 1). The control samples (n = 6) comprised tissue samples from the lateral epicondyle in six healthy asymptomatic people (five men, one woman; mean age 31 years, range 24 to 40), which were investigated in parallel.

Diagnostic criteria

Diagnostic criteria for the patients were pain on palpation of the extensor origin at the lateral epicondyle (TE) or of the flexor origin at the medial epicondyle (GE), respectively and pain elicited from these areas by forced extension of the wrist (TE) or flexion of the wrist (GE). Other causes for lateral or medial elbow pain, such as arthritis, synovitis, generalised pain syndrome, and radiculopathy from cervical spine or nerve entrapment were excluded by clinical examination. The clinical examinations were performed by an experienced hand surgeon with a special interest in tennis and golfer’s elbow (B-OL) or in a few cases (three patients with TE) by an orthopaedic surgeon specialising in sports medicine (EZ). In the latter cases, US and CD were also used to confirm diagnosis (structural changes and high blood flow).

Tissue preparation and evaluation

Biopsies

Biopsies were taken during surgical treatment under regional intravenous anaesthesia (prilocaine, 5 mg/ml) (n = 9), general anaesthesia (n = 2) or local anaesthesia (lidocaine 10 mg/ml with epinephrine 5 μg/ml) (n = 1), using a tourniquet applied to the upper arm and the biopsy specimens were taken through a skin incision 5–6 cm long. During surgery, the extensor carpi radialis brevis (ECRB) muscle origin at the lateral epicondyle (patients with TE) or the origin of the flexor muscles at the medial epicondyle (patients with GE), was identified. The biopsies, measuring approximately 5×8 mm, were taken by sharp detachments as close to the bone insertion as possible. The control biopsy specimens were taken from the most proximal part of the muscle origin, but not the junction with the bone. Under local anaesthesia (prilocaine 10 mg/ml), using a tourniquet applied to the upper arm, the biopsy specimens were taken through a skin incision 2 cm long.

Tissue preparation

Specimens from all three groups were treated by immersion overnight at 4°C in a solution of 4% formaldehyde in 0.1 mol/l phosphate buffer pH 7.0. The specimens were thoroughly washed in Tyrode solution with 10% sucrose at 4°C overnight, then mounted on thin cardboard in OCT embedding medium (Miles Laboratories, Naperville, Ilinois, USA), frozen in propane chilled with liquid nitrogen and thereafter stored at −80°C until sectioning.

Serial sections 7 μm thick were cut using a cryostat from specimens from all groups. The sections were mounted on slides pre-coated with chrome–alum gelatin. They were dried and then processed for immunohistochemistry or stained with H&E for examination of tissue morphology.

Primary antibodies

The ChAT antibody (code AB144Pl; Chemicon, Temecula, California, USA) was raised in goat against human placental ChAT. The antibody is reported by the supplier to be reactive with ChAT in a number of species. It was used at a concentration of 1:50.

The antigen for the TH antibody (code P40101; Pel-Freez, Rogers, Arkansas, USA) is a denatured (by sodium dodecyl sulphate) purified recombinant rat and bovine TH antigen. It is reported by the supplier to cross-react with all mammalian forms of TH tested to date. The antibody was raised in rabbit and used in these studies at a concentration of 1:50.

Immunofluorescence

The sections were first pre-treated with acid potassium permanganate for 2 minutes to enhance specific immunofluorescence reactions,32 and then rinsed three times for 5 minutes each in phosphate-buffered saline (PBS). After incubation for 20 minutes in a 1% solution of Triton X-100 (Kebo Lab, Stockholm, Sweden) in 0.01 mol/l PBS pH 7.4, the sections were rinsed three times for 5 minutes each in PBS and then incubated for 15 minutes at room temperature in either 5% normal swine serum (code X0901; Dako, Glostrup, Denmark) in PBS supplemented with 0.1% bovine serum albumin (BSA) (for TH) or in 5% normal donkey serum (code 017-000-121; Jackson Immuno-Research, West Grove, Philadelphia, USA) in PBS (for ChAT). The sections were then incubated with the primary antibody diluted in PBS with BSA (TH) or in PBS without supplementary BSA (ChAT), in a humid environment. Incubation was performed for 60 minutes at 37°C. After another three washes for 5 minutes each in PBS and another incubation in normal serum as described above, the sections were incubated with secondary antibody for 30 minutes at 37°C. The secondary antibodies used were tetramethylrhodamine isothiocyanate (TRITC)-conjugated swine antirabbit IgG (code R0156; Dako) diluted 1:40 (TH) and fluorescein isothiocyanate (FITC)-conjugated donkey antigoat IgG (code: 705-095-147; AffiniPure; Jackson ImmunoResearch) diluted 1:100 (ChAT). The sections were finally mounted in microscopy mounting medium (Vectashield; Vector Laboratories, Burlingame, California, USA) after a final set of washes (3×5 minutes) in PBS.

Comparative stains were performed on sections of rat adrenal medulla (TH) and fixed human colonic tissue (ChAT). In addition, fixed patellar tendon tissue (patients with tendinopathy), stained for both ChAT and TH, was examined simultaneously as a control for intracellular reactions in tenocytes.

All sections were examined under a light microscope (Axioskop 2 Plus; Zeiss, Oberkochen, Germany) equipped with epifluorescence optics and photographed with a digital camera (DP70; Olympus, Tokyo, Japan), by two of the researchers (PD, EZ).

RESULTS

General morphology

The muscle origin tissue from both the medial and lateral epicondyles was mainly made up of collagen forming well-organised, parallel fibre bundles and of spindle-shaped cells, often numerous, interpreted as being fibroblasts. In some areas, the collagen fibre orientation was occasionally observed to be more irregularly arranged. Some specimens showed an abundance of small blood vessels, most of which were confined to zones of loose connective tissue interspersed between the collagen bundles. No evident differences were seen between the different groups in their general tissue architecture (cellularity, collagen formation) but the specimens that showed an abundance of small blood vessels were found only in the patient groups.

Sympathetic innervation patterns

Several blood vessels, mostly arterioles, were seen to be innervated by plentiful TH-positive nerve fibres, which were mainly located just outside the media adventitia junction, but some were also located in the outer parts of the adventitia (fig 1A,C); however, most vessels were not innervated at all. In addition, freely coursing TH-immunoreactive nerve fibres were seen between the collagen bundles, and a few nerve fascicles expressing TH-immunoreactivity were also found (fig 1B). No obvious difference in sympathetic innervation patterns were noticed between the different groups studied.

• Download figure
• Open in new tab
• Download powerpoint

Figure 1

Sections of tissue from the muscle origins at (A,C) medial and (B) lateral epicondyles in patients with golfer’s and tennis elbow respectively stained with tetramethylrhodamine isothiocyanate (TRITC) immunofluorescence stains for TH. Positive reactions were visible as perivascular innervation (A,C, arrowed) close to the media–adventitia junction (asterisks) of arterioles, but some nerve reactions were also located in the outer part of the adventitia (C, arrowhead). (B) A freely coursing nerve fibre (arrow) and a small nerve fascicle (N) were positive for TH. Bars, 25 μm.

Tyrosine hydroxylase-like immunohistochemical reactions in local cells

TH-like immunoreactions (LI) were detected in a subpopulation of the fibroblasts (table 1). These positive reactions were seen in the form of fine intracellular punctuate reactions in the tissue of 4/7 patients with TE (2/4 men; 2/3 women), 2/4 patients with GE (men 1/2; 1/2 women) and 0/6 controls. The most common reactions were seen in the parts of the samples showing very large number of cells, and the clearest reactions were found in fibroblasts of the muscle origin tissue from the medial epicondyle in patients with GE (fig 2A).

• Download figure
• Open in new tab
• Download powerpoint

Figure 2

Sections of tissue from the muscle origin at the medial epicondyle in a patient with golfer’s elbow, stained with (A) tetramethylrhodamine isothiocyanate (TRITC) and (B) fluorescein isothiocyanate (FITC) stains for (A) tyrosine hydroxylase (TH) and (B) choline acetyltransferase stain (ChAT). Bars, 10 μm. (A) Intracellular punctuate reactions for TH were visible inside several fibroblasts (arrows); (B) there was no positive staining for ChAT found, thus only background fluorescence was visible in the fibroblasts (arrows).

Choline acetyltransferase stains

No reactions for ChAT were detected, neither in nerve structures, nor in local cells of the tissue (fig 2B), in any of the groups.

Control stains

Control tissues, known to contain cells expressing TH, were processed in parallel. Medullar cells of the rat adrenal gland were stained with specific immunohistochemical reactions using the TH antiserum (not shown). ChAT control stains were performed on fixed human colonic tissue, which showed the presence of specific immunohistochemical reactions in the nerve fibres of the myenteric and submucous plexa (not shown).

Simultaneously processed patellar tendon tissue from patients with patellar tendinopathy showed intracellular reactions in tenocytes concerning both TH and ChAT stains (not shown), confirming previous findings.14 16

DISCUSSION

The present study gives the first evidence that local, non-neuronal production of catecholamines occurs in human tissue at the muscle origin at both the lateral and the medial humeral epicondyle in patients with TE and GE, respectively. This was shown by immunohistochemistry showing the presence of the catecholamine-synthesising enzyme TH in the fibroblasts of the tissue samples from 4/7 patients with TE and 2/4 patients with GE, and no detectable levels of this enzyme were found in fibroblasts of control tissue from the lateral epicondyle (0/6). Although the patient material is limited, there does not seem to be any gender difference in the occurrence of TH-LI in local cells. In addition, no evident association between the presence or absence of such immunohistochemical reactions and the duration of symptoms was seen.

The finding of the biosynthetic enzyme for catecholamines in fibroblasts of human muscle origin tissue in the arm corresponds to recent findings in human patellar and Achilles tendon tissue. In these tissue types also, the tissue cells themselves seem capable of producing catecholamines,15 16 18 and the findings were most pronounced in patients with tendinopathy. However, in contrast to patellar14 17 and Achilles19 tendinopathy, in which the tissue cells also seem capable of producing acetylcholine (ACh), no evidence of such production in patients with TE or GE was found in the present study using staining for the ACh-synthesising enzyme ChAT. There are therefore differences in the cholinergic system between tendons in the leg and the tissue at the muscle origin in the arm. As we performed extensive control stains, the method we used for ChAT detection was confirmed to be working correctly. Hence, sections of both human colonic tissue and patellar tendon (tendinopathy) tissue, processed in parallel with the study tissues, gave positive reactions for ChAT. In our previous studies, we showed using preabsorption stains that these are specific reactions.14 33 It is possible that the amount of cellular ChAT present in the muscle origin tissue is very low, making it undetectable by the method used in this study. It is known from other investigations that the presence of an intracellular substance in tendinous tissue, as confirmed by detection of mRNA, cannot always be confirmed at the protein level by immunohistochemistry.34 In the future, studies at the genetic level, such as in situ hybridisation, should be performed on the muscle origin tissue to shed further light on this question concerning the cholinergic system. It is also important to keep in mind that the muscle origins of the elbow and the Achilles/patellar tendons represent tissue types with different functions. The muscle origin tissue represents the connective tissue of a muscle origin, whereas the Achilles tendon is a tendinous continuation of the muscle to the insertion site and the patellar tendon is a distal tendon insertion. Hence, the local cells (fibroblasts/tenocytes) at these sites are likely to have certain functional differences. In addition, differences in cholinergic activity or response for the blood vessels in the arms versus the legs have previously been shown, with one study showing greater vasodilator response to ACh in the arm compared with the leg.35

Concerning innervation patterns, we found no evidence of nerves positive for ChAT, whereas several nerve structures displaying TH-immunohistochemical reactions were detected. These latter were mainly seen in association with some of the blood vessels, which corresponds to previous findings of sympathetic innervation in human muscle origin tissue of the arm,5 but some reactions were also observed in the form of freely coursing nerve fibres and in fibres of a few nerve fascicles.

Generally, a large number of cells (fibroblasts) was found in the muscle origin tissue of the present study. In principle, this corresponds to previous descriptions indicating that TE tissue is hypercellular,28 but no evident increase in cellularity was seen in the patient groups compared with the control group. This was also the finding of a previous study, in which the general morphology of the muscle origin tissue in healthy individuals was found to grossly correspond to that seen in patients with TE.4 In fact, based on experience of examinations of tendon morphology, the cells of the control samples of the present study were found to be numerous compared with cells of normal distal tendons.36 We believe this reflects the fact that muscle origin tissue has a different function than distal tendon tissue. However, we did find areas of numerous small blood vessels in several of the patients in the present study, which were not seen in the controls. This might represent evidence of a neovascularisation, previously described in histological studies of TE tissue.37

Tyrosine hydroxylase in non-neuronal tissue

TH is the enzyme that catalyses the first and rate-limiting step in the biosynthesis of catecholamines (dopamine, norepinephrine, adrenaline) and is found in the brain and the sympathetic nervous system, including the adrenal medulla.38 It has also been found to exist outside the nervous system, indicating that catecholamines can also be produced by non-neuronal cells. Thus, in line with our findings here and those of studies on patellar and Achilles tendinopathy,15 16 18 are findings in the mouse that show that hepatic stellate cells (ie, the principal fibrogenic cells of the liver) contain catecholamine biosynthetic enzymes and release norepinephrine.13 Furthermore, these fibrogenic cells of the liver are suggested to respond to and produce catecholamines to promote hepatic fibrosis.39 In addition, in vitro studies have shown that human cells from a fibroblast-derived cell line can be induced to express TH.40

The regulation of TH gene expression is known to be complex and several regulatory mechanisms have been presented. Transcription factors such as Nurr1 and Pitx3 have received particular attention and it has been shown that overexpression of these factors separately induces endogenous TH expression, including in non-neuronal cells.41 However, responses to transcriptional factor expression is thought to be cell-dependent,42 and many studies suggest that other, so far unknown, factors are required for the induction of TH expression.41^–43 In light of this, one might ask whether the TH-positive cells of the painful muscle origins in the present study represent a specific subpopulation of cells in this tissue. This seems to be a logical question, as we found that within the same tissue sample, some fibroblast-like cells were positive and others were negative. Are the TH-positive cells in fact a form of stem cells recruited for the purpose of tissue repair? In the mouse, it has been shown that some cells of tendinous tissue display certain stem cell-like properties.44 It is also known that catecholamine stimulation can lead to fibroblast proliferation, an effect seen to be increased after injury.45 This makes it tempting to speculate whether the presence or absence of catecholamine-producing cells in a painful muscle origin has implications on the self-healing capacity of the tissue. However, this question cannot be answered by the present study, as all patients were treated surgically with removal of the affected tissue.

Locally produced catecholamines and vasoregulation

US and CD investigations of the muscle origin at the lateral epicondyle have shown that patients with TE have high blood flow.6 It has also been shown that intratendinous injections, guided by US and CD and directed at this high blood flow, can give pain relief in most patients with TE.10 Furthermore, in a 2-year follow-up study, it was shown that most of the successfully treated patients with TE displayed normal blood flow at follow-up.46 However, many questions are left unanswered about why this high blood flow exists and why injections within the area of such vascularity seem to work. Are the vessels merely a ‘visual marker’ for sensory nerves in association with blood vessels4 or does the pathological blood flow in itself contribute to the condition? Furthermore, is the increased blood flow seen by CD a sign of a pathological neovascularisation or simply a visualisation of normal vessels behaving ‘abnormally’, possibly under the influence of pathological autonomic regulation? There are results suggesting that local dysfunction of the sympathetic nervous system may be associated also with the pain in TE.47 Because CD registers blood flow, the increased vascularity seen in TE corresponds to an increase in blood flow. This could represent pre-existing vessels with an increased flow rate, which in turn could be caused by a vasoconstriction. In light of that, it is important to keep in mind that the vasoconstrictor innervation of these tissues heavily outweighs the vasodilatory counterparts.5 However, no increase in presence of sympathetic nerves has been observed in patients with TE, in comparison to healthy individuals. Of course, the nerves could fire up their signals, but it is also possible, that a vasoconstriction in part is caused by locally produced catecholamines in the damaged/painful muscle origins.

Nevertheless, only 4/7 patients with TE showed signs of local production of catecholamines in the present study, whereas there was high blood flow, as seen with US and CD, in 21/22 patients with TE in a previous study.6 This difference might have several possible explanations. Firstly, it is unlikely that even pathological blood flow is regulated solely by local catecholamine production; the perivascular sympathetic innervation is most likely involved in the vasoregulation of the tissue. Secondly, as discussed above for ChAT, it is possible that the methods we used in the present study cannot detect very low levels of TH. Finally, it should be noted that in the reported study,6 there were interindividual differences in blood flow between different patients.

Locally produced catecholamines and pain; concluding remarks

As described earlier, the connective tissue at the muscle origin at the lateral epicondyle is supplied with sensory nerves,4 and pain can thus theoretically elicit from within this tissue. Keeping this in mind, it is interesting to note that catecholamines can affect adrenergic receptors on the membranes of sensory nerves, thereby activating or potentiating pain signals in these.48 Furthermore, it has previously been shown that several neuropathic pain states may be associated with pathological sympathetic activity.49 In light of this, it is interesting to note that our findings of both sympathetic innervation and local production of catecholamines in human muscle origin tissue give morphological evidence for sympathetic effects to occur in this tissue. To what extent these two supplies of catecholamines play a role in the pain mechanisms of TE and GE remains to be elucidated.