The term spondylarthritis (SpA) encompasses the closely related diseases ankylosing spondylitis (AS), reactive arthritis (ReA), psoriatic arthritis, arthritis associated with bowel diseases, and undifferentiated arthritis. Several decades ago, the discovery of a strong association of these entities with the expression of HLA–B27 led to different hypotheses regarding HLA–B27–based pathology (1).

HLA–B27 is a class I major histocompatibility complex (MHC) molecule consisting of a heavy chain bound to β2-microglobulin. To date, 24 different HLA–B27 subtypes are known; HLA–B2705 is the dominant subtype, from which the other allelic variants were generated by mutation. However, despite the vast number of published articles on this topic, there is no widely established pathogenic concept of the association of HLA–B27 in SpA.

The “arthritogenic peptide hypothesis” suggests an antigen-specific immunopathology based on the presentation of specific autologous or microbial arthritogenic peptides by disease-associated HLA–B27 molecules. Given the inflammatory reaction that is most prominent at the cartilage–bone interface, cartilage-derived autoantigens are interesting candidates in SpA-related autoimmunity. In particular, the G1 domain of the proteoglycan aggrecan was of high interest since immunization of mice led to an experimental disease resembling SpA (2). However, this immunoreaction was neither HLA–B27 restricted nor CD8+ T cell driven. Furthermore, other rheumatic diseases were also associated with an immunoreaction against the G1 domain of aggrecan.

Collagen as an integral part of cartilage structures and collagen-derived peptides are also of interest. Derivatives of type II collagen and type XI collagen bind to HLA–B27, but no consistent reaction of cytotoxic T lymphocytes toward these HLA–B27–bound peptides occurs (3). Atagunduz et al (4) performed a screening analysis in order to identify autoreactive CD8+ T cells specific for cartilage-derived peptides in patients with AS and identified nonamer peptides derived from type VI collagen and type II collagen as being stimulatory for CD8+ T cells.

Previous studies described HLA–B27–restricted T cell clones with specificity for microbial antigens and the potential for autoreactivity in patients with SpA (5). Furthermore, activation of HLA–B27–specific cytotoxic T lymphocytes occurred after exposure to Chlamydia trachomatis in HLA–B27–transgenic animals (6), pointing to a molecular mimicry mechanism with the breakdown of the peripheral immune tolerance, and highlighting HLA–B27 as an autoantigen itself. Several investigators observed a strong similarity between peptides derived from the autologous HLA–B27 molecule and arthritogenic peptides derived from chlamydial or enterobacterial antigens (7). Another mimicry mechanism is deducible from the sequence homology between the pLMP2 antigen derived from the latent membrane protein 2 of Epstein-Barr virus and a self peptide generated from the vasoactive intestinal peptide type 1 receptor presented by distinct HLA–B27 subtypes (8).

The “autodisplay hypothesis” offers an additional link between HLA–B27 and immunopathology. Luthra-Guptasarma and coworkers described an autocatalyzed conformational change of HLA–B27 heavy chains leading to occupation of the molecule's own peptide-binding groove (9).

Thus, HLA–B27 might be a gatekeeper for autoimmunity by presenting arthritogenic peptides. Furthermore, HLA–B27–derived peptides could act as autoantigens as well. However, the classic CD8+ T cell–mediated immunoreactivity expected after presentation by class I MHC molecules is not required for SpA-related immunopathology, since depletion of CD8+ T cells in HLA–B27–transgenic rats had no impact on the onset and severity of experimental arthritis (10).

As a result, non–antigen-presenting effects of HLA–B27 are now in focus. This approach is additionally based on biochemical distinctions between HLA–B27 and other HLA class I molecules. The intracellular 3-dimensional folding process of the HLA–B27 heavy chain is slower compared with that of other HLA class I molecules. This carries the risk of misfolding within the endoplasmic reticulum (ER). Furthermore, the HLA–B27 heavy chain has a propensity for heavy chain dimerization, which is based on the existence of thiol groups and the formation of disulfide bridges. This dimerization particularly appears in the absence of β2-microglobulin (11). The consequences of these biochemical distinctions have been studied mostly in animal models.

The accumulation of misfolded and dimerized heavy chain is followed by ER-associated degradation and formation of complexes with the ER chaperone BiP (12). The engagement of BiP by misfolded HLA–B27 heavy chains leads to displacement and release of factors bound to this chaperone (e.g., activating transcription factor 6, inositol-requiring enzyme 1, and PERK). The net effect of these changes is induction of an ER stress response with consecutive cellular activation (13). HLA–B27 heavy chain homodimers are also expressed on the cell surface and have potential to induce deviant immune responses. They can mimic HLA class I molecules, which might explain the involvement of CD4+ T cells in SpA-related immunopathology. Furthermore, these homodimeric structures can interact with killer immunoglobulin receptors or leukocyte immunoglobulinlike receptors (14). Other candidate genes potentially involved in the HLA–B27 misfolding processes are interesting. However, the results of studies that have been performed thus far appear to be somewhat inconclusive.

In contrast to the other forms of SpA, in ReA the association with infection is evident, and persistence of antigens derived from the arthritogenic bacteria and of elevated serum antibodies can be shown (15), pointing to a prolonged immunoreaction attributable to impaired bacterial elimination or antigen clearance in patients with ReA.

Many investigators have attempted to identify a link between these observations and the expression of HLA–B27. Based on in vitro findings, results of several studies indicate that HLA–B27 might influence the behavior of host cells and impair their response to persistent intracellular bacteria. In particular, the replication of Salmonella in HLA–B27–positive monocytic cells is enhanced (16). Thus, many potential links between HLA–B27 and the deviant immunoreaction are described, but their pathogenic relevance remains to be proven.

A different approach concerns enthesial inflammation, which is a characteristic finding in all forms of SpA. The correlation between the immunopathology described above and the manifestation of enthesitis still must be defined. In this issue of Arthritis &Rheumatism, the article by Benjamin et al (17) aims to elicit the histoanatomic background of SpA-related enthesial inflammation. We recently showed that approximately half of patients undergoing surgical straightening of the spine because of kyphosis display active foci of destruction, characterized by osteoclasts and fibroblasts producing matrix metalloproteinases and cathepsin K at both facet and spinal processes and demonstrating that inflammatory–destructive mechanisms persist far beyond the development of ankylosis. Elucidating the molecular and cellular basis of the pathways mediated by enthesial inflammation and these sites of destruction will be a challenging task for future research (18).

The term “enthesis” comes from ancient Greek, meaning inserting or interposing something (e.g., inserting a letter in a word). Physicians in the 19th century used the term enthetic to describe diseases that were inoculated from sources outside of the body, such as infections. Enthesis, in contrast, was synonymous with prosthesis or implant. In the 20th century, the term was used to describe the attachment of a tendon, ligament, or joint capsule into the adjacent bone (19). An enthesis comprises an osteotendinous or osteoligamentous junction, forming an insertion site that becomes part of the adjacent bone but still can be distinguished from mature bone tissue.

According to Benjamin and colleagues, entheses comprise 2 different groups, depending on the tissue that is present at the insertion site. Fibrous entheses comprise the direct insertion of dense fibrous connective tissue (tendon or ligament) into bone or periosteum. Fibrocartilaginous entheses are insertion sites where an intermediate tissue between dense connective tissue and cartilage is found (20).

Entheses form biologic interfaces that link soft tissue and hard tissue boundaries in order to dissipate mechanical stress and to provide optimal myofascial stability. Due to their locations at sites of mechanical challenges, entheses are prone to overuse injuries. Insertional tendinopathies or enthesopathies of the forearm (tennis and golfer's elbow) or the leg (jumper's knee and achillopathies) are well-documented problems in sports medicine. Because entheses are also the main focus in SpA, they are of great clinical interest beyond these conditions.

The European League Against Rheumatism recommended that “any pathological changes of an enthesis” should be described by the term enthesopathy, whereas the term enthesitis has been designated only for inflammatory alterations (21). Enthesitis is a typical clinical feature of SpA and has been seen traditionally as a focal insertional disorder. Several imaging studies performed with magnetic resonance and ultrasound, however, have challenged this view. Soft tissue swelling, osteopenia of the subentheseal bone, bone cortex irregularity at the insertion site with adjacent periostitis, and soft tissue calcification within the enthesis are the typical radiographic findings in enthesitis. The swelling of ligaments as well as low echogenicity caused by inflammation and edema can be seen using high-frequency real-time ultrasonography, whereas magnetic resonance imaging shows gadolinium enhancement within the soft tissue (22). Interestingly, these signs of inflammation can be observed not only at the insertion site itself but also in adjacent ligaments, in the peritendinous soft tissue including fat pads and bursae, and bone structures.

In concert with the diffuse clinical pattern of symptoms that accompany enthesitides, the concept of an “enthesis organ” evolved, applying to many joint-related and extraarticular sites (23). The enthesis organ concept stands for more than just the enthesis itself and includes adjacent structures that share considerable dissipation of stress and mechanical load.

In patients with SpA, not all entheses are similarly affected. As hypothesized previously (24), clinical differences may be explained by the observations that inflammatory changes of entheses occur mostly at sites of repeated exposure to mechanical trauma. The importance of mechanobiology in the relationship between exercise and entheses has been recently postulated (25). Several studies, however, have failed to establish correlations between the anatomic concept of entheses and the pathogenesis of SpA. Those studies lacked cases of early disease, samples from young patients, or suitable controls.

Does the latest study by Benjamin et al (17) shed new light on this topic? In their objectives, the authors propose to explain the basis for entheseal-associated bone disease in SpA by analyzing microanatomic and histopathologic relationships within the enthesis organ. By examining serial sections of entheses from 60 cadavers, the authors observed a thinning of the deep cortical boundary at virtually all fibrocartilaginous entheses and, in addition, small holes in the cortical shell, preferentially at entheses derived from the lower limb. Further evaluations addressed fissures (defined as microdamage) and vascular or reparative changes. Vascularization was found to be increased around holes in the cortical shell but was also present on the soft-tissue side of the enthesis. Reactive bone formation and a highly orientated trabecular network were observed, but these were more common in the lower extremity than in the upper limb. Based on their observations, Benjamin et al conclude that the functional integration of the enthesis is associated with microdamage and repair at the hard tissue–soft tissue interface.

This is an interesting finding. Entheses are designed as stabilizing anchors between muscles, tendons, and the skeleton in order to transmit mechanical force during movement. As mentioned previously, entheses do not attach and act in an isolated manner. The microanatomic alterations that were seen in the trabecular network of the bone further help to strengthen the concept of an enthesis organ. Moreover, fibrocartilaginous entheses appear to be dynamic tissue that can respond to different activity levels. In this regard, the amount of uncalcified entheses is increased in the ligamentum patellae of patients with jumper's knee and, in turn, is reduced in the digital extensor tendons of patients with rheumatoid arthritis; in contrast, histologic assessments in SpA have detected local contact sites of blood vessels with the zone of uncalcified fibrocartilage tissue of the enthesis, described as “enthesial discontinuities” (for review, see ref.24). With respect to the study by Benjamin and coworkers, it remains elusive whether these alterations are pathologic changes or physiologic adaptations. Within fibrocartilaginous entheses, moreover, no blood vessels can be found, and it is unexplained so far how enthesial inflammation should emerge within avascular areas. Taken together, the main emphasis of the present study of Benjamin et al is shifting our blinkered medical view of enthesopathies associated with SpA and biomechanical factors from a focal to a multifocal or systemic disorder of the enthesis organ.

Still, when good research is performed, more questions are raised than answered. Although the study by Benjamin et al was performed with a large cohort of serial sections, the study design remains descriptive. The samples used in the study were derived from healthy subjects, i.e., cadavers lacking a distinct pathology of the musculoskeletal system. Moreover, the age range of the cadavers used was 49–101 years, with a mean age of 84 years. With respect to SpA, this large control group is age biased. “Nothing in biology makes sense except in the light of evolution” (Theodosius Dobzhansky, Geneticist, 1900–1975). If we try to see through the eyes of evolution, our skeleton is designed to last for 40–50 years. By using bone samples derived from people older than age 80 years, we probably reach the normal borders of human physiology. Thus, cortical holes and other entheseal-associated changes might be degenerative or osteoporotic alterations of the aging bone.

Starting from these interesting morphologic changes, we find it rather difficult to make a conclusive link between anatomic findings and the pathogenesis of rheumatologic disorders. Mechanobiologic factors acting on the enthesis organ might indeed be responsible for the bone involvement of enthesopathies in general and SpA in particular. Still, it remains unclear how anatomy and immunobiology should be connected within the concept of the present study. Hypervascular cortical holes and microdamage of the bone adjacent to entheses (as seen in a population of healthy elderly individuals) are interesting pathomorphologic findings with respect to biomechanics and degenerative changes. However, even within the clinical context that enthesitides are preferentially observed at mechanically traumatized insertion sites, the question of how the onset and perpetuation of an immunopathologic, inflammatory disorder of young persons can be linked to microanatomic changes in bone regions of the elderly is far from being answered.

Mechanobiology has emerged as an intriguing concept, from which we probably have not yet learned all consequences in terms of health and disease of the musculoskeletal system. Before leaving the field of hypotheses behind, however, further studies must be awaited.


  1. Top of page