“A hypothesis is… the obligatory starting point of all experimental reasoning. Without it no investigation would be possible and one would learn nothing...”. These are the words of the 19th-century French scientist Claude Bernard, widely considered as the father of modern experimental physiology.

Many hypotheses have been put forward over the years in attempts to explain the still incompletely clarified aetiology and pathogenesis of chronic tendon pain (tendinopathy). At the turn of the millennium, Khan et al made new intriguing suggestions,1 partly contradicting previously presented theories on the cause of tendinopathy. They speculated that biochemical mediators in the tendon tissue might influence or irritate nociceptors in or around the tendon. Half a decade later, novel findings of non-neuronal production of signal substances in human tendon cells (tenocytes) in tendinopathy seem to give new support to such a “biochemical” hypothesis.

Old “tendinitis” theories, assuming an inflammatory process as the cause for chronic tendon pain, have already been discarded, as microdialysis studies showed no increase in prostaglandin E₂ levels in chronically painful tendons compared with asymptomatic ones,2 3 and because no inflammatory cells are detected in histological examination of tendon tissue from patients with tendinopathy.4 The histological findings instead clearly point in another direction, implying that the condition is better characterised as a degenerative-like process, sometimes defined as “tendinosis”.4 A traditional theory regarding the cause of the pain is that pain derives from the separation of collagen fibres in severe cases of tendinopathy, but this suggestion has been heavily contradicted with convincing arguments by Khan et al,1 one such strong argument being that collagen excision from patellar tendons in autograft harvesting causes minimal pain to the donor site. Nevertheless, the physical strain to which the tendon is exposed is generally thought to be one of the most important factors behind the development of tendinopathy, a conclusion influenced not least by the fact that the tendons that are subjected to high mechanical demands (eg, patellar and Achilles tendons) are the ones that are most often affected.5 Thus, tendon overload/overuse still seems to be the most commonly accepted explanation model for the aetiology of tendinopathy.6^–9 However, because Achilles tendinopathy is also known to affect patients with a sedentary lifestyle,10 overload or overuse alone does not seem to explain the condition and there is still no explanation linking overuse-induced microtrauma to pain.

It is in light of this lack of sufficient evidence for previously existing theories that Khan et al presented their “biochemical” hypothesis in 2000.1 Since then, interesting new studies have followed, some suggesting that there is an increase in sensory nerve fibres in tendinopathy.11 12 However, the results were not significant or observed in comparisons with ruptured tendons, thus making any conclusions difficult to draw. In addition, other studies investigating the nerve patterns in both asymptomatic and painful tendons have not been able to show any evident difference in the occurrence of sensory nerves.13^–15 Nevertheless, all these studies contribute to one very important fact: there are sensory afferents present in the tendon tissue, a tissue previously thought to be hyponeural, although the presence of such afferents seems very sparse in deep parts of the actual tendon tissue proper (most nerves instead being found in the loose paratendinous connective tissue surrounding the tendon).13^–15 This important information tells us that pain sensation actually can originate in and be transmitted from the tendon. It does not explain where the pain comes from in tendinopathy, but as the afferents are there it means that these sensory nerves can be affected, whether by nerve transmitters from other components of the peripheral nervous system or speculatively by substances produced in the tendon tissue itself, as the biochemical hypothesis theorises. In view of this, it is interesting to note that tendon nerve fascicles that contain nerve fibres positive for the sensory marker substance P (SP) also contain fibres positive for sympathetic nerve markers15 and receptors for several signal substances such as SP,16 acetylcholine (ACh),17 and catecholamines,15 substances all traditionally confined to the nervous system. In addition, receptors for the excitatory nerve transmitter glutamate have been found on nerves in tendons.18 Hence, there is a morphological basis for the suggestion that all these signal substances theoretically are able to influence the nerves of the tendon.

So, where do these signal substances come from?

As sensory afferent nerve fibres and sympathetic efferent nerve fibres apparently co-exist within the same fasicles,15 it does not seem far-fetched to hypothesise that the adrenergic receptors found in these fascicles15 are more specifically located on the membranes of sensory neurons and targeted by catecholamines released from sympathetic nerve endings. In support of the speculation that catecholamines may modulate sensory information, it has been previously suggested that the sympathetic nervous system can generate or enhance pain19 and that this is accomplished via functional interactions with sensory afferents under certain pathological conditions.20 Such interactions are proposed to occur in the form of release of norepinephrine from sympathetic terminals targeting adrenergic receptors on membranes of afferent neurons.21 However, in the case of tendinopathy, the source of the catecholamines may also, rather surprisingly, be the tenocytes themselves, as these tendon cells in humans are found to contain tyrosine hydroxylase, the rate-limiting enzyme in catecholamine synthesis, this being most clearly shown for tendons from patients with tendinopathy.15 22 23

Furthermore, studies have shown that human tenocytes not only contain biosynthetic enzymes for catecholamines, but also for ACh. Thus, the ACh synthesising enzyme choline acetyltransferase and vesicular ACh transporter have been found in tenocytes of tendons from patients with tendinopathy, but not (or only sparsely) in control tendons, suggesting that these cells are capable of production and release, of ACh in chronically painful tendons.17 24 In addition, these findings were found to be particularly prominent in patients with severe tendinopathy—that is, in patients resistant to various treatments.25 This local, non-neuronal production of ACh might theoretically stimulate the M₂ muscarinic ACh receptors present in the tendons, including those in the nerves.17 24 Thus, it is possible that the effects of non-neuronal ACh may include effects in relation to pain. It support of such an assumption, it has been shown that ACh, when applied to human skin, can induce pain,26 and that an ACh analogue, carbachol, has been shown to stimulate excitation of nociceptive C-fibres (sensory afferents) in vitro in rat skin.27 It has also been shown previously that sensory nerve fibres in the rat skin express M₂ muscarinic ACh receptors,28 implying that the excitatory effects on sensory afferents seen by stimulation with ACh or its analogues could be mediated, at least partly, by activation of these M₂ receptors. However, contradictory to this, other studies have shown that stimulation of M₂ receptors on sensory neurons may inhibit nociception.29

The findings of locally produced catecholamines and ACh in tenocytes have been shown on both the protein (via immunohistochemistry) and mRNA (in situ hybridisation) levels.15 17 22^–25 Very recently, tenocytes of human tendons have also been shown to express SP mRNA30 and vesicular glutamate transporter 2 (VGluT2) at both the protein and mRNA levels,31 VGluT2 being an indirect marker for glutamate release. In addition, via microdialysis, it has been shown that the level of glutamate in chronically painful tendons is significantly higher than in normal tendons.2 3 All these findings are of interest, taking into consideration that nerves in tendon tissue express receptors for both SP (neurokinin-1 receptor)16 and glutamate (NMDAR1),2 and that both SP32 and glutamate33 are known to be mediators of pain.

Other observations during recent years that are also interesting, in view of the findings of locally produced signal substances in chronically painful tendons, are the observations that the blood vessel walls of human tendons express receptors for catecholamines (α₁, α_2A and β₁ adrenoreceptors)15 22 and receptors for ACh17 24 and SP.16 Studies on tendons using ultrasound and colour or power Doppler imaging have found increased vascularity (measured as augmented blood flow), in close relation to the area with structural changes in tendinopathy.34^–36 Furthermore, very interestingly, an association has been noted between the degree of such pathological vascularity and the level of pain in patients with tendinopathy.37 One could thus speculate whether the locally produced signal substances might play a role in the differences seen in vasoregulation in tendinopathy. It is known that ACh is a vasodilator38 and that stimulation of α₁39 40 and α_2A41 adrenoreceptors mediate constriction of blood vessels, whereas stimulation of β₁ adrenoreceptors mediates relaxation of blood vessels.42 In addition, SP has been shown to exert vasoregulatory effects.43 44 However, the changes in vascularity might not only be a result of changed flow in pre-existing vessels, but rather that of an angiogenesis process. Interesting new findings by Scott et al of vascular endothelial growth factor-A (VEGF) in painful patellar tendons, but not in controls, give evidence of involvement of this important angiogenic peptide in the increased vascularity seen in tendinopathy and thus suggest that neovascularisation is at least part of the process.45 In addition, in support of the latter theory, the proliferation marker Ki67 has been seen to label tendinopathy vessels.46

A final possible target, perhaps even the most plausible target, for the locally produced signal substances is the tenocytes themselves, as they have been shown, particularly in tendinopathy, to express receptors for catecholamines,15 22 Ach,17 24 25 and SP.16 30 It is thus possible that the locally produced signal substances act in an autocrine or paracrine fashion, to exert effects on the tendon tissue itself. We know that there are marked degenerative-like tissue changes in chronically painful tendons (tendinosis changes), such as hypercellularity, vascular proliferation and discontinuity of collagen,4 but tendinopathy has also been associated with apoptotic events.47^–50 With this in mind, it is tempting to speculate on an association between the tissue changes that take place in tendinopathy and a possible autocrine or paracrine regulation of this via the tenocytes. In favour of such a suggestion are the findings that ACh51 and norepinephrine52 can stimulate the proliferation of myofibroblastic hepatic stellate cells in mice and induce collagen gene expression in these cells. It is interesting that cells in tendinopathy tendons are reported to have a myofibroblastic appearance.53 Furthermore, prolonged stimulation of receptors for catecholamines on rat fibroblasts may induce proliferation of these cells, an effect that is increased after injury.54 In addition, stimulation of ACh receptors on pulmonary fibroblasts may augment collagen accumulation,55 and agonists of ACh receptors can also stimulate angiogenesis and increase collagen deposition during wound healing of the skin.56 In addition, SP has been shown to stimulate angiogenesis.57 Furthermore, in vivo stimulation of certain receptors for catecholamines may induce apoptosis of cardiac muscle cells in the rat heart.58

In summary, studies of recent years on human tendinopathies have provided us with evidence of a local, non-neuronal production in tenocytes of signal substances traditionally confined to neurons. These findings reinforce the previously presented “biochemical” hypothesis, as the locally produced signal substances might have effects on tissue changes, vascular regulation and/or pain signalling. The potential clinical implications are considerable; substances proven to inflict or enhance pain might be blocked for therapeutic purposes and substances (or analogues of such substances) found to promote tissue healing might be added. However, it should also be emphasised that if there is a true upregulation of this local production of signal substances in tendinopathy, the cross-sectional nature of the studies published to date makes it impossible to determine whether such an increase is causative for the processes of tendinopathy or merely a byproduct of the disease. Hence, further investigations on animal models and/or in vitro cultures of tendon cells are needed to develop the “biochemical” hypothesis further. But to end as we started, in the words of Claude Bernard: “A fact in itself is nothing. It is valuable only for the idea attached to it or for the proof which it furnishes”.