A systematic review of inflammatory cells and markers in human tendinopathy

The inclusion criteria for studies in this systematic review consisted of the following:

Clinical studies investigating the presence of inflammation in tendinopathic tendons such as cross-sectional studies, case-control studies, prospective observational studies, randomized controlled trials. In vitro studies using tendon tissue, or cells derived from tendinopathic tendons were also included. Included studies were restricted to studies with an evidence level of 3 or better. In vitro studies using tissue, or cells that have been treated with cytokines or other agents or modified mechanically were excluded.

Studies with participants of any age presenting with tendinopathy were included. Specimens from spontaneous tendon ruptures were included considering the assumption that only tendinopathic tendons are prone to spontaneous ruptures. Diagnostic criteria of the above diseases include a clinical presentation of chronic pain or loss of function, confirmed with imaging modalities such as magnetic resonance imaging or ultrasound. Studies investigating the presence of inflammatory cells, immune cells, inflammatory markers in tendinopathic tendons, and tendon ruptures were included. Tendinopathy was defined as pain, diffuse or localized swelling, and impaired performance of the tendon. Tendon rupture was defined as a tear, visible with medical imaging, such as MRI or ultrasound, or macroscopically visible. Inflammatory, and immune cells were defined as leukocytes, neutrophils, eosinophils, basophils, mast cells, macrophages, monocytes, T-lymphocytes, B-lymphocytes, NK-cells, and dendritic cells. Inflammation markers were defined as fibroblast growth factors (FGF), platelet-derived growth factor (PDGF), transforming growth factor beta superfamily proteins (TGF-beta superfamily proteins), eicosanoids, COX-1, COX-2, and cytokines. Studies investigating the presence of possible initiators of inflammation, such as bacteria, trauma, mechanical stresses, inflammatory diseases, or other proposed factors were also included.

Studies from the search were merged in EndNote, and duplicates were removed. Application of exclusion, and then inclusion criteria were made by screening the titles and then the abstracts. The full texts were then obtained for the identified articles in order for data extraction. A PRISMA-flowchart of the study selection process is shown in Fig. 1. Studies were also identified by screening the reference lists.

A data extraction table was created. The data extraction for half of the studies was performed by one review author (GJ). The data extraction for the other half of the studies was performed by the second review author (CK). Control of the data extracted were performed in the same way.

Critical Appraisal Skills Programme (CASP) [12] appraisal form was used to assess the quality of included studies. The assessment of study quality was performed in an unblinded standardized manner independently by one reviewer (GJ) and control of the assessment was then assessed by a second reviewer (CK).

Due to the heterogeneity of the studies, e.g. study type, and outcome measures a meta-analysis could not be performed.

Using the search method mentioned above, 53 studies were included in this review. A total of 2306 tendinopathic tendon specimens were assessed. Tendon specimens were heterogenous in terms of location and presentation of tendinopathy. The details of the included studies will be included in the Table 1.

The results of the quality assessment are shown in the Table 1. The evidence level of included studies was assessed according to the Oxford Centre of Evidence based Medicine (OCEBM) [66]. All included studies included level 3 evidence (n = 53) regarding the presence of inflammatory cells in tendinopathic tendons. 50/53 studies scored 6 or higher out of 11 in CASP. It is notable that included studies have varying objectives, hence the results from quality assessment may not directly reflect how strictly inflammation in tendinopathic specimens were assessed.

Among the 53 included studies, 39 studies showed signs of inflammation in tendinopathic tendons, including the presence of inflammatory cells or an increase in inflammatory markers. The expression of inflammation does not seem to be correlated with any obvious confounders including site, presence of rupture, chronicity, or previous treatment of corticosteroid injections.

Twenty-five studies reported the presence of inflammation in the tendon specimens, and 14 suggested the absence of inflammatory cells. By using H&E to assess for inflammatory cells, tendinopathic tendons presenting with inflammatory cells were reported to range from 0 to 100% (mean 16%) between studies. For studies adapting IHC to stain for inflammatory cells, tendinopathic tendons showing inflammatory cells were reported to be 23 to 100% between studies (mean 88%).

For the 25 studies which supported the presence of inflammatory cells, 7 of the studies did not specify which types of cells were observed. The most common cell type was macrophages, found in 16 studies. Other cells include lymphocytes in 2 studies, mast cells in 8 studies, and granulocytes in 3 studies.

Twenty-two studies performed assessment on related inflammatory markers including at least one of the markers mentioned above. All studies indicated increased levels of some inflammatory markers measured. Inflammatory markers reported include IL-1. IL-6, IL-8, IL-10, IL-17, IL-33, COX-1, COX-2, TGF-b, TNF-a, FGF and more. A full list of inflammatory mediators detected can be found in the Table 1.

A variety of tendinopathic cases were included from the search, in terms of different locations and presentations.

It is also worth mentioning that data such as previous conservative treatments and activity levels were often non-recorded, or is inconsistent between the patients in the included studies. 19/53 studies included such additional description to the population sub-group that may influence the inflammatory status. In 12 studies, patients have undergone corticosteroid injection. In 5 studies, tendinopathic patients were reported to have experienced high activity level and overuse. In 2 studies, inflammation to the neighbouring structures such as glenohumeral arthritis or joint inflammation were identified as a source of inflammation affecting the tendon. In 1 study, bacteria was identified as a possible source of inflammation. In 1 study, there was a reported use of NSAIDs the recruited tendinopathic patients. Nevertheless, inflammation may be observed in tendinopathic tendons of any size of tear, location, or tendons previously treated with corticosteroid injection.

Heterogenicity of tendinopathic cases is a major limitation of the current review. In this review we included cases with a range of presentations from chronic pain to ruptured cases. Chronicity of the tendon disorder also varied greatly, which when combined may have a great impact on the presentation of inflammation in these specimens included.

Heterogenicity of the detection methods in both the presence of inflammatory cells and inflammatory markers were also significant in this review, making included studies less comparable. However, by comparing the results from different detection methods, we were able to identify a possible explanation in the emergence of the debate on whether chronic inflammation is present in tendinopathic tendons. This will be further discussed in the following section.

Heterogeneity of control specimens is also a limitation of the consistency between studies. In some cases, specimens from macroscopically healthy areas of tendinopathic tendons have been used as controls [60]. This practice is not recommended since it is possible that the whole tendon is affected by tendinopathy [67]. In other studies, the contralateral tendon has been used as control, even though some evidence suggests that unilateral rupture is preceded by bilateral damage [14]. A seemingly healthy tendon could be subclinical and asymptomatic. Tendons from a different anatomical location have also been used as controls [67]. There is a possibility that tendons found on different anatomical localization have a divergent biomechanical construction, due to the fact that the tendons are involved in different movements. Lastly, for several studies, specimens from cadavers have been used as controls. Tendon specimens from this group most likely represents a true healthy tendon. The inconsistency in control groups may have a particularly high impact when comparing inflammatory markers presented in tendinopathic tendons.

The sampling of tendinopathic specimens presenting in different stages was also difficult. Specimens from the debridement of ruptured tendons is common, but these specimens could only represent cases of chronic tendinopathy with an acute insult of tendon rupture. Some studies identified earlier stages of tendinopathy with various methods. There were articles which sampled the macroscopically intact sub-scapularis tendon beside a ruptured supraspinatus [21, 22, 59], and there was another study which defined early pathology as an impinged tendon sampled during acromial decompression [15]. Tendon specimens obtained from the listed protocols mostly represent cases which have likely received and failed other modalities of conservative treatment. Tendinopathy that would respond to conservative treatment also comprise a large part of the total cases. However, tendon specimens were never obtained for these cases due to ethical reasons.

Another limitation is that the demographic data of tendinopathic patients were not reported consistently between studies. For example, it is suspected that the usage of NSAIDs or corticosteroid injections could have a direct effect on the presentation of local inflammation within the tendon. Activity level also have potential effects on the presentation. However, these information is often neglected.

These concerns however reflect the current standpoint in this area of research. With the lack of studies with higher evidence level, a systematic review of highest quality is currently not feasible.

With a previous understanding that tendinopathy is a degenerative disease with an absence of inflammation, it is possible that studies showing an inconsistent result will more likely be published. However, due to the great heterogenicity of the studies included in this review, plotting a funnel plot could not effectively identify the presence or absence of existing publication bias. Only publications in databases have been covered by our study, therefore the coverage of “grey literature” has not been evaluated [68].

According to our search results, signs of inflammation, including either the presence of inflammatory cells or an increase in inflammatory markers, were observed in 39 out of 53 studies. One common feature of the studies which reported an absence of inflammation was that the only method used was to identify inflammatory cells with H&E staining.

Considering the comparable clinical presentation, sampling method and staining method shared across the studies, we highly suspect the diagnosing accuracy of using H&E staining in the identification of inflammatory cells. As mentioned in a previous study [13], though there was an increase in macrophages present in chronic tendinopathic tendons, tenocytes still make up the majority of cells within the tendon. As macrophages make up less than 10% of the total population, it can be difficult to identify inflammatory cells in a crowded background of tenocytes. The previous thought that inflammation is absent in chronic tendinopathy could be a result of sub-optimal detection with an inappropriate method.

Inflammatory cells observed in chronic tendinopathy were macrophages, lymphocytes, mast cells, and in some rare occasions, granulocytes. Except for granulocytes, the cell types indicate the state of inflammation as a chronic inflammation.

Macrophages are well known for its role in phagocytosis of infectious organisms [69]. However, the contribution of this cell type extends to many other systems including bone remodelling, erythropoiesis, brain and lung development [69]. It was also reported that macrophages play an important role in the regulation of inflammation [70]. Resident macrophages initiate the inflammatory response towards injury by recognizing damage associated molecular patterns (DAMPs) [69]. The process is followed by secretion of cytokines and eicosanoids, resulting in recruitment of inflammatory cells, with neutrophils being the first to enter the site [69]. The resolution of inflammation is also closely related to macrophage activity. The change in phenotype from M1 to a M2-like phenotype macrophage leads to a phenomenon known as the lipid mediator class switch [70].

Mast cells could have a significant role in tissue remodelling too. As reported in a previous review on mast cell physiology, it was reported that mast cell deficient mice initially had intact hair growth and bone density. However, defects were observed in case of injury, and tissue remodelling of the hair follicles and bone tissues were not comparable to that of healthy samples [71]. In another animal model on tendon injury, it was found that there were an increased expression of mast cells and myofibroblasts during the tendon healing process [72].

Lymphocytes are also known to be present in many autoimmune, inflammatory diseases such as Hashimoto thyroiditis and psoriasis [73]. It was hypothesized that the exaggerated recruitment of these cell types lead to un-controlled activation of macrophages, leading to excessive damage in cells and architecture [73]. It is possible that the same mechanism applies to tendons, where excessive lymphocyte activity could damage the extracellular matrices.

Granulocytes including neutrophils and eosinophils were reported in rare occasions. As mentioned above, neutrophils are the first cells to be recruited upon macrophage activation [69]. The presence of this cell type is indicative of an acute inflammatory state, reported only in cases of rupture.

Unlike the information provided by the presence of inflammatory cells, current evidence on the inflammation markers shown in tendinopathic tendons do not show a consistent picture. As summarized in a review in 1997 [74], inflammatory mediators contribute to inflammation via complex pathways, but common mediators can be categorized to be present in acute inflammation, chronic inflammation, or both. In this review, the inflammatory mediators reported were a mixture of mediators in the acute, chronic and common group. It is also notable that inflammatory markers assessed vary greatly between studies. It is therefore very difficult to deduce the character of inflammation in chronic tendinopathy with the current information on detectable inflammatory markers.

However, study of the inflammatory markers in tendinopathic patients could potentially be of high importance. As mentioned in the limitations, the sampling of tendinopathic tendons in different stages could be challenging due to the invasive nature of tendon sampling. However, with increasing evidence that tendinopathy is associated with chronic inflammation, there may be systemic changes detectable as inflammatory markers in more assessible specimens, such as blood cells. Investigation on the systemic changes in tendinopathic patients could be a rewarding field of study.

Causes of inflammation in tendinopathic tendons is currently unclear, however, 19/53 studies mentioned additional description to the recruited tendinopathic population, which may provide insights on whether the presentation of inflammation maybe limited to certain sub-groups. From our observations, inflammation is a character that often occur in chronic tendinopathy, regardless of location and chronicity. The presence of inflammation was also independent with previous treatments including steroid injection and NSAIDs. It was also directly mentioned in one included study that there were no association between steroid injections and the presentation of inflammation [20]. The previous understanding that chronic tendinopathy is an inflammation free disease may be due to low sensitivity of classical staining techniques using H&E. However, attention is drawn to two studies which reported a mere 20% rate of the presence of inflammatory cells in tendinopathic tendons despite adapting IHC for detection of related CD markers [27, 28]. The pattern is yet to be explained regarding the varying presentation of inflammation between tendinopathic tendons with similar clinical presentation.

One cause of inconsistency could be the existence of different stages to tendon injury. In some reports of acute tendon rupture, neutrophils and other granulocytes were reported present, likely acting as an acute response to trauma. However, current literature could not facilitate an educated deduction of factors associated with the presentation of inflammation. The cause, stage, and conservative managements received to treat tendinopathy could be some factors associated with the presentation of inflammation. Future studies with high quality are necessary to identify these variables.

The involvement of bacteria was also mentioned in some cases of tendinopathy. Although rarely assessed, there is a possibility that bacterial infection may play an important role in the presentation of chronic inflammation. Three studies which identified the presence of bacteria were excluded due to evidence level lower than 3. The Bacterial species identified in these studies were Mycobacterium tuberculosis [75, 76] and Borrelia [77]. One of the included studies identified staphylococcus genus as a possible initiating factor of inflammation [41]. In this study, the blood samples from the patients were negative for bacteria contrary to the presence of bacteria in the tendons. This suggests that the presence of bacteria was local to the AT. However, it is important to emphasize that the tendinopathic changes could favour the presence of bacteria and consequently the presence of bacteria in tendinopathic tissue may be secondary, as a result of the favouring environment. Further investigation into this topic may be a rewarding frontier.

The involvement of metabolic diseases associated with tendinopathy was also mentioned in several excluded case reports [78, 79]. The significance of such underlying disorders may also contribute to the varying presentation of inflammation observed. This concept is supported by previous reports that metabolic disorders including diabetes [80], obesity [81], gout [82], and hypothyroidism [83] can increase the risk of the development of tendinopathy. A systematic review in 2016 showed that various hormone receptors in tenocytes can be affected by hormonal imbalances of insulin, estrogens, thyroid hormones and growth hormone [84]. Hormone profile greatly impacts the inflammatory pathway, and one of such impacts is illustrated in a review describing inflammatory changes in diabetic and obese patients. Migration of inflammatory cells to adipose tissues lead to decreased availability and less effective healing in the tendon. Glycation of collagen and impaired cross linkage also contribute to sub-optimal healing [85]. As metabolic disorders impact on the hormonal profile differently, it is expected that inflammation following the metabolic insults may also vary.

According to an editorial in 1998 [7], the most prominent lesion in chronic tendinopathies is a degenerative process with an absence of inflammatory cells [7]. It was acknowledged that histological findings were inconsistent, and that signs of inflammation were present in some of the tendon specimens. The explanation towards this phenomenon was that inflammation could be the “primum movens” of the development of tendinopathy. It was believed that the transient inflammatory state would ultimately lead to the typical presentation of tendinosis, with an absence of inflammation [7]. In simplified terms, it was suggested in this study that active inflammation, if any, should only occur at early stages of the development of tendinopathy. Search results from this systematic review do not fully agree with the classical study. Presence of inflammatory cells were reported in chronic tendinopathy in the included studies.

It is acknowledged that the current review is not the first to discuss on the presence of inflammation in chronic tendinopathy. A recent systematic review published in 2016 [86] also reached a similar conclusion. Quoting from the review, “the absence of inflammation in tendinopathy were more based on belief rather than scientific data”. The current review agrees with the idea that inflammation may be present, but the findings suggest that its presentation may not be as straightforward. By performing a more sophisticated search with more included, relevant studies, we were able to more subjectively assess the inconsistent presentation of inflammation in chronic tendinopathy between studies. Chronic inflammation is present in the majority of chronic tendinopathy using specific staining techniques, but there are exceptions which do not show positive staining of any inflammatory cells.

Tendinopathy is previously understood as a degenerative disorder with an absence of inflammation [37]. According to this review, it is likely that a chronic inflammation can be present in tendinopathic tendons. Therefore, the anti-inflammatory approach may well be supported in the conservative management of tendinopathy.

However, this does not mean that the current management strategy is flawless. Publications on the efficacy of current management strategies on tendinopathy have also shown that anti-inflammatory treatments like NSAIDs or corticosteroids only provide a short term relief of symptoms, and may have a negative impact on its structural healing [87, 88]. This result is consistent to the current finding since anti-inflammatory drugs suppress the inflammatory state to achieve pain relief. However, the effects do not last since chronic inflammation can be caused by the presence of a persistent stimulus, or a deranged cellular function to resolve inflammation.

The inhibition of the inflammatory process may also negatively impact the natural healing process of tendon healing, leading to further degenerative changes. This concept is supported by previous literature on the balance of metalloproteinases (MMPs) and its significance in maintaining healthy tendon homeostasis. MMP levels are regulated by inflammation [89], and it was mentioned that balanced MMP activity play an important role in maintaining structural integrity of tendons through constant extra-cellular matrix (ECM) remodelling [90, 91]. Inhibition to inflammation may in turns inhibit tendon homeostasis, leading to poor integrity in long term.

The cause leading to chronic inflammation could thus be multifactorial and varying among tendinopathic patients. However, rather than straightforward inhibition to the inflammatory process with NSAIDs or glucocorticoids, identification of factors leading to chronic inflammation, and corresponding targeted treatment could be the key in solving the burden caused by this common but chronic disease.

Current literature is insufficient to deduce a common pattern in regard to the presence of inflammation in chronic tendinopathy. To improve our understanding on this issue, more high quality studies with a large sample size and comparable detection methods is required. Known confounding factors such as activity level, previous history of anti-inflammatory treatment, and the chronicity of the tendon disorder must also be consistently documented in future for valuable comparison.

Also, it is acknowledged that the sampling of tendon samples with different stages of tendinopathy and confounding factors such as metabolic disease could be challenging. A variety of potentially triggering factors of inflammation in tendinopathy have been shown in this study, such as overuse, associated inflammatory and metabolic diseases, and bacterial involvement.