SEARCH

SEARCH BY CITATION

Presentation of the debate, Dr. Maksymowych: Introduction to the motion

  1. Top of page
  2. Presentation of the debate, Dr. Maksymowych: Introduction to the motion
  3. In support, Dr. Elewaut: Inflammation and new bone formation in ankylosing spondylitis are tightly linked
  4. In opposition, Dr. Schett: Inflammation and new bone formation in ankylosing spondylitis are independent processes
  5. In support, Dr. Elewaut: Rebuttal
  6. In opposition, Dr. Schett: Rebuttal
  7. AUTHOR CONTRIBUTIONS
  8. REFERENCES: In support:
  9. In opposition:

A hallmark pathologic feature of spondylarthritis (SpA) is the development of ankylosis in axial joints. This process is the primary reason for the development of major disability. Its prevention therefore constitutes a key goal of management in SpA. There has been a longstanding assumption that inflammation is the process that sets in motion the chain of events that leads to ankylosis, that the two processes continue to be inextricably linked as the disease progresses over time, and that new bone formation occurs only at sites where inflammation has been activated. However, direct testing of this hypothesis has been difficult because of several factors, such as the lack of availability of objective clinical and laboratory measures of inflammation, delay in diagnosis, slow progression of the disease, and lack of access to tissue for direct histopathologic analysis.

Investigators have turned to indirect approaches, such as magnetic resonance imaging (MRI), animal models of SpA, and evaluation of the effects of anti–tumor necrosis factor α (anti-TNFα) therapies on bone formation. The reports of these studies assessing the relationship between inflammation and structural changes have generated substantial controversy, culminating in numerous symposia at international meetings. An alternative hypothesis has proposed that an as-yet-unknown pathogenic trigger(s) simultaneously induces both inflammation and activation of stromal progenitor cells, which ultimately lead to endochondral new bone formation, but these processes proceed along essentially disconnected, or “uncoupled,” pathways as the disease progresses. Once disease is established, therefore, inflammation and new bone formation can occur at distinct locations.

Discerning the true relationship between inflammation and ankylosis is pivotal to future advances in disease modification because if these processes are largely uncoupled, disease-modifying therapies will need to directly target bone-formation pathways. If, on the other hand, inflammation and ankylosis are indeed “coupled,” this would predict that early and effective suppression of inflammation would prevent the development of ankylosis. The full scope of this passionate debate is now brought to life in the pages of Arthritis & Rheumatism for this inaugural Unresolved Questions in Rheumatology: Current Debates, as two leading authorities on this subject argue their respective positions on the motion.

In support, Dr. Elewaut: Inflammation and new bone formation in ankylosing spondylitis are tightly linked

  1. Top of page
  2. Presentation of the debate, Dr. Maksymowych: Introduction to the motion
  3. In support, Dr. Elewaut: Inflammation and new bone formation in ankylosing spondylitis are tightly linked
  4. In opposition, Dr. Schett: Inflammation and new bone formation in ankylosing spondylitis are independent processes
  5. In support, Dr. Elewaut: Rebuttal
  6. In opposition, Dr. Schett: Rebuttal
  7. AUTHOR CONTRIBUTIONS
  8. REFERENCES: In support:
  9. In opposition:

An overwhelming amount of evidence supporting the link between inflammation and new bone formation in ankylosing spondylitis (AS) has accumulated over recent years, and the counterevidence has been hampered by a variety of limitations, including the patient cohorts studied and the mouse models investigated. It is therefore surprising that controversy has continued in this area.

The primary evidence in favor of the tight relationship between inflammation and ankylosis is derived from clinical observations, particularly from MRI studies. These studies have prospectively assessed the relationship between inflammation and new bone formation in patients with AS, particularly at stages preceding the onset of structural damage (preradiographic phase). Studies performed in the last decade have enabled us to directly evaluate over time the relationship between prior inflammation and the subsequent development of new bone in anatomically identical regions. In addition to their important diagnostic utility, these studies have, through noninvasive prospective examination, led to tremendous progress in understanding the sequence of events leading to AS (1).

Several stages have thus been defined, the earliest being the appearance of inflammatory lesions on fat-suppressed MRI sequences. The severity of these lesions has been shown to correlate with histopathologic quantification of inflammation (2, 3). Furthermore, these lesions are promptly reduced during the weeks following therapy with anti-TNF agents (4–6). Most strikingly, vertebral inflammation identified on MRI could predict the formation of new syndesmophytes seen on conventional radiographs 2 years later (7). It has also long been known that the use of nonsteroidal antiinflammatory drugs (NSAIDs) not only affects the signs and symptoms of AS, but also retards the structural progression of the disease (8).

So why has there been so much debate on this matter? The primary reason for this controversy is the apparent lack of efficacy of TNF-blocking strategies in retarding structural progression in AS, despite their dramatic beneficial effects on the signs and symptoms of the disease and the resulting improvements in functional outcomes (9–11). In addition, MRI followup studies showed that new syndesmophytes may also occur in areas where the baseline MRI did not show signs of inflammation (12). In a mouse model of ankylosing enthesitis, it was demonstrated that etanercept treatment was not able to attenuate the formation of new bone (13), and conversely, in studies of mice overexpressing TNF, no signs of bone remodeling were apparent (14). This has led to the suggestion that both cardinal features of AS are actually uncoupled to a large extent. There are, however, a number of limitations with all of the above-mentioned studies questioning the linkage between inflammation and new bone formation.

First, the apparent lack of efficacy of TNF blockade on structural progression was deduced from comparisons between the outcome data after 2 years of TNF blockade and the data on “natural” progression seen in the historic Outcome Assessments in Ankylosing Spondylitis International Study (OASIS) cohort, a cohort of AS patients followed up prospectively by an international team more than a decade ago, prior to the introduction of TNF-blocking agents (15). The use of a historical cohort, the short duration of followup, the relatively low sensitivity of the outcome parameters of radiographic progression (as assessed by the modified Stoke Ankylosing Spondylitis Spine Score [SASSS]) are important biases to be taken into consideration.

The recently reported results of an 8-year followup assessment of AS patients treated with infliximab supported this view, as they indicated significant inhibition of structural progression (16). Curiously, in the initial 4 years following initiation of TNF inhibition therapy, no difference from controls was noted, but the effect observed during the following 4 years was marked. This was reflected in the number of new syndesmophytes that were found, which were also notably reduced in patients who received anti-TNF therapy (16). These results suggest that the effect of TNF inhibition on structural progression in AS is related to the duration of therapy. Furthermore, initiation of TNF blockade earlier in the disease process may lead to a much more pronounced impact on structural progression. Evidence has accumulated in support of the view that inflammatory lesions can progress to so-called fat lesions, which respond poorly to anti-TNF therapy and are considered to be an intermediate stage in the development of new syndesmophytes (17). Such studies aimed at assessing the impact of TNF inhibition on long-term structural progression in early axial SpA are currently under way.

Second, the appearance of new syndesmophytes in areas where baseline MRI did not show inflammation occurs only in a limited proportion of individuals and may simply reflect the limitations of a single MRI examination in the context of a chronically evolving lesion. Furthermore, several of these lesions may have already transitioned into fat lesions, an important risk factor for progression to new bone formation (17).

Third, while studies of mouse models have generated interesting mechanistic results, there are also profound limitations that could jeopardize the application of such findings to the disease in humans. For example, in the ankylosing enthesitis that occurs in male DBA/1 mice, the level of inflammation is relatively mild, and the TNF neutralization achieved in vivo may only be partial (13). Genetic backcrosses using TNF receptor–deficient mice in these models remain to be performed. In contrast, in mouse strains that overexpress TNF, no obvious signs of new bone formation were noted, despite the presence of inflammation (14). This has been elegantly demonstrated to be linked mechanistically to the TNF-driven induction of an inhibitor of the Wnt pathway, Dkk-1, which prohibits new bone formation (14).

Mouse models in which TNF is overexpressed, however, are generated in mice of an entirely distinct genetic background (C57BL/6), which may profoundly affect the intrinsic bone-remodeling capacity. This is of critical importance, as several studies have highlighted that many aspects of bone biology are largely dependent upon the genetic background (18, 19). Hence, studies assessing the remodeling capacity of TNF inhibition versus overexpression in relation to different genetic backgrounds need to be performed. This may have clinical implications as well, since AS is a largely genetically determined disease. Consistent with this, recent studies have identified subsets of patients with more advanced structural progression as well as genetic polymorphism(s) for such genes as large multifunctional peptidase 2 (20). Furthermore, the levels of TNF (and likewise, the resulting secondarily induced cytokines) achieved in these mouse models may profoundly exceed the levels that occur in humans with AS or SpA. Measurement of serum levels of proinflammatory cytokines in humans has, on the whole, led to variable results, with marked heterogeneity that may reflect the overall fluctuating nature of inflammation in human AS and SpA (21, 22). In contrast, results obtained in TNF-driven mouse models display gradually increasing or consistently elevated serum levels of TNF, depending on the mouse model used (23, 24). The impact of these “supraphysiologic” levels should not be underestimated, as they may lead to a profound shift toward the well-described catabolic features attributed to TNF.

In summary, there is now substantial evidence that the inflammatory process in ankylosing spondylitis is tightly linked to the formation of new bone and that early and prolonged inhibition of proinflammatory cytokines, notably TNF, effectively modulates structural progression. The arguments used to counteract these observations are weakened by the wealth of experimental limitations.

In opposition, Dr. Schett: Inflammation and new bone formation in ankylosing spondylitis are independent processes

  1. Top of page
  2. Presentation of the debate, Dr. Maksymowych: Introduction to the motion
  3. In support, Dr. Elewaut: Inflammation and new bone formation in ankylosing spondylitis are tightly linked
  4. In opposition, Dr. Schett: Inflammation and new bone formation in ankylosing spondylitis are independent processes
  5. In support, Dr. Elewaut: Rebuttal
  6. In opposition, Dr. Schett: Rebuttal
  7. AUTHOR CONTRIBUTIONS
  8. REFERENCES: In support:
  9. In opposition:

Inflammation and new bone formation are the key pathogenic processes in AS. Despite the concurrent presence of inflammation and new bone formation in individuals with AS, their causal relationship has never been proven, and it is intriguing to speculate that new bone formation occurs independently of inflammation. Both processes may thus reflect two sides of the same coin, sharing one common pathogenic ancestor (e.g., an infectious trigger) but developing mutually independently of each other. Several lines of evidence support such a concept.

The consistent failure of TNFα-blocking agents, which are considered the best antiinflammatory agents for AS thus far, to affect new bone formation in AS is perhaps the most striking proof of such a concept (1–3). This finding is surprising, since in the treatment of rheumatoid arthritis (RA), the antiinflammatory activity of a drug is virtually unequivocally linked to its effect on the maintenance of the joint structure. Such a tight link as seen in RA is not seen in AS with any of the TNF-blocking agents, all of which fail to retard syndesmophyte formation despite their excellent antiinflammatory activity. Thus, uncoupling of the antiinflammatory effect from the structure-modifying effect of antiinflammatory drug therapy in AS suggests that these two processes are not well linked to each other.

One could argue, however, that TNF acts as an antianabolic agent and that blockade of TNF can therefore hardly be expected to inhibit new bone formation (4, 5). Nonetheless, the fact that the use of TNF-blocking agents in AS usually inhibits osteitis very well but is not followed by retardation or arrest of syndesmophyte formation is striking and suggests that these processes are not linked. It is an appealing concept that TNF blockade may act substantially differently in early disease, before syndesmophyte growth has emerged; however, this hypothesis has so far been proven neither experimentally nor clinically. Prostaglandin E2 (PGE2) inhibition may act differently from TNF blockade, and PGE2 inhibition has been shown to retard new bone formation in AS (6). One remains doubtful, however, whether this effect of PGE2 inhibition truly proves the link between inflammation and new bone formation, since PGE2 is an important mediator of osteoblast differentiation (7) and function and since inhibition of new bone formation may merely reflect the antianabolic effect of PGE2 inhibition, rather than its antiinflammatory effect.

Data from magnetic resonance tomography of patients with AS have also not supported a tight link between inflammation and new bone formation (8–10). Of importance, inflammation and local bone changes occur at distant sites, which contrasts with the findings obtained in RA. Inflammation in AS is reflected by bone marrow changes (osteitis), which are most likely responsible for the substantial axial bone loss observed in AS that leads to premature osteoporosis and increased fracture risk. New bone formation in AS, however, is confined to the periosteal bone compartment, outside the cortical bone lining, and does not affect the trabecular bone. Moreover, the presence of osteitis as visualized by MRI is only poorly linked to new bone formation in AS. In particular, skeletal sites that show osteitis are not identical to the sites where syndesmophytes later develop. Sites with more chronic lesions, in contrast, where the low water content reflects the absence of inflammation, show some preponderance for new bone formation (11). These lesions with low water content are also termed “fatty” lesions, although the presence of adipose tissue has never been formally proven, and the low water content may also be based on poorly vascularized mesenchymal tissue other than adipose tissue. Thus, imaging data obtained in AS patients also do not support a tight link between inflammation and new bone formation.

Finally, data from basic science studies do not really support a direct link between inflammation and new bone formation. Many of the proinflammatory cytokines are potent inducers of bone resorption, and some of them, such as TNFα, are at the same time potent inhibitors of new bone formation (12). Thus, inflammation favors a negative balance of bone turnover, which is reflected by enhanced bone loss during inflammatory disease (13). Direct induction of new bone formation by inflammation is therefore highly unlikely. Consistent with this, animal models of arthritis, which are dependent on continuous overproduction of proinflammatory cytokines such as TNFα, are highly bone destructive but do not show evidence of new bone formation (5). Animal models with only minimal and/or spurious inflammatory phases, such as the arthritis model in male DBA/1 mice, can, in contrast, lead to massive bony overgrowth (14), which suggests a poor, if not inverse, link between inflammation and new bone formation. In fact, instead of inflammation, mechanical factors that lead to localized stromal remodeling at sites with a specific predilection, such as the insertions of the tendons (entheses), may have a far more important impact on new bone formation (15).

In summary, evidence of a tight link between inflammation and new bone formation in AS is scarce. Inflammation and its related cytokine milieu represent a bone catabolic, rather than anabolic, factor. Accordingly, bone erosion associated with inflammation does not represent a prerequisite for new bone formation (16). It thus seems that inflammation and new bone formation in AS develop independently. They may be seen as reflecting two different reaction patterns to a common, probably infectious, insult. The reaction of soft tissue corresponds to osteitis, whereas the reaction of the hard tissue is reflected by syndesmophyte formation.

In support, Dr. Elewaut: Rebuttal

  1. Top of page
  2. Presentation of the debate, Dr. Maksymowych: Introduction to the motion
  3. In support, Dr. Elewaut: Inflammation and new bone formation in ankylosing spondylitis are tightly linked
  4. In opposition, Dr. Schett: Inflammation and new bone formation in ankylosing spondylitis are independent processes
  5. In support, Dr. Elewaut: Rebuttal
  6. In opposition, Dr. Schett: Rebuttal
  7. AUTHOR CONTRIBUTIONS
  8. REFERENCES: In support:
  9. In opposition:

Dr. Schett discusses at great length the motion that inflammation and new bone formation are largely independent in AS. While it is indeed curious that in the treatment of RA, the antiinflammatory activity of TNF-blocking agents is strictly linked to the maintenance of joint structure, a comparable link between antiinflammatory activity and retardation of syndesmophyte formation in AS has been less convincing overall. Thus, the argument that TNF acts as antianabolic agent and that TNF inhibition could therefore hardly be expected to inhibit new bone formation is potentially important. As I outlined earlier, the clinical studies supporting an uncoupling of inflammation and new bone formation are weakened by a number of experimental flaws. Furthermore, the results of an 8-year followup of infliximab-treated AS patients that were recently reported highlighted significant inhibition of structural progression (16). This was reflected in the number of new syndesmophytes that were found, which were also notably reduced in anti-TNF–treated patients (16). These results support the view that the effect of TNF inhibition on structural progression in AS may be related simply to the duration of therapy. Along the same lines, earlier initiation of treatment regimens with TNF blockade may lead to a much more pronounced impact on structural progression. Therefore, the results of these early-intervention trials that are currently ongoing will be of great importance in resolving the issue.

The impact of PGE2 inhibition by NSAIDs on inflammation and structural progression is well established. While it is true that the inhibition of new bone formation by PGE2 inhibition may reflect its antianabolic, rather than its antiinflammatory, effect, this has not been formally proven. In any case, PGE2 is at least a clear illustration of a common pathway that links inflammation and new bone formation in AS. An intriguing question is whether inflammation and new bone formation occur at identical sites or more distantly from each other.

While MRI studies of the spine have highlighted the clearly common sites of new bone formation preceded by the presence of inflammatory lesions, this is not always the case. The simplest explanation for this observation may be the limitations of a single MRI examination in the context of a chronically evolving lesion. Alternatively, lesions may have transitioned into so-called fatty lesions, showing a preponderance of new bone formation. Even if it should be demonstrated that the presence of inflammatory cells and the formation of new syndesmophytes occur at distinct anatomic sites, as shown in Achilles tendon enthesitis (25), this would still not exclude an interconnection between the two processes. Hence, an array of inflammatory mediators produced at sites of inflammation will have important paracrine and even systemic effects. The formation of new bone is, overall, likely to be the result of a number of combining factors, including inflammatory mediators, physical (biomechanical) stress, regional microanatomy, or even local differences in the distribution of progenitor cells.

In any case, the controversy surrounding this topic highlights the fact that the balance between new bone formation and inflammation is far from completely understood. Both the studies in AS patients and the studies in mouse models are hindered by the intrinsic limitations outlined earlier. While the clinical evidence supporting the link between inflammation and new bone formation is most solid, additional evidence is needed in mouse models of preclinical disease. Specific challenges in this regard include incorporating the impact of the genetic background, achieving full inhibition of proinflammatory pathways, and the use of mouse models of inflammation characterized by cytokine levels that are comparable to those observed in humans with AS. These studies are of great importance for a better understanding of the underlying pathways and, ultimately, therapeutic interventions.

In opposition, Dr. Schett: Rebuttal

  1. Top of page
  2. Presentation of the debate, Dr. Maksymowych: Introduction to the motion
  3. In support, Dr. Elewaut: Inflammation and new bone formation in ankylosing spondylitis are tightly linked
  4. In opposition, Dr. Schett: Inflammation and new bone formation in ankylosing spondylitis are independent processes
  5. In support, Dr. Elewaut: Rebuttal
  6. In opposition, Dr. Schett: Rebuttal
  7. AUTHOR CONTRIBUTIONS
  8. REFERENCES: In support:
  9. In opposition:

Results from MRI examinations were used by Dr. Elewaut as major support in his argument for a causal relationship between inflammation and bone formation in AS. It is, however, difficult to conclude causality from cross-sectional and even from longitudinal imaging data. The very limited histopathologic data in AS and additional analyses of osteitis in other diseases indeed suggest that bone marrow signal enhancements in T2-weighted scans correspond to inflammatory lesions. However, whether osteitis causes new bone formation or whether inflammatory lesions and syndesmophyte formation instead represent independent aspects of the same underlying process is yet unknown.

A close link between osteitis and new bone formation is not really supported by data for the following reasons. First, the association between osteitis seen on MRI and syndesmophyte formation seen on radiography is rather poor, with only a minor proportion of inflammatory spots being associated with new bone formation, which suggests that other factors are involved in the formation of syndesmophytes (8–10). Second, the intervals between MRI scans and radiographs are usually long, with months to years between investigations, making it difficult, if not impossible, to prove whether these two processes are indeed causally linked. Third, inflammatory lesions (high T2 signal) in AS can “resolve” and turn into more chronic lesions with low water content (low T2 signal), which may indicate mesenchymal tissue responses. Recent data suggest that these types of lesions are better linked to syndesmophyte formation, as this process is also based on mesenchymal tissue remodeling, than to inflammatory lesions (11). Further sequential MRI studies are therefore necessary to shed more light on this concept.

The idea that TNF inhibition may indeed block syndesmophyte formation and that we have just missed such effects is highly unlikely for the following reasons. First, as mentioned, the OASIS cohort, which was used as a comparator for structural progression in AS, comes from the era before TNF inhibition. This fact, however, suggests a tendency for undertreatment in these patients associated with incomplete control of inflammation and rather high, but not low, structural progression. If we hypothesize that inflammation indeed drives syndesmophyte formation, such a cohort with a rather high chance for progression would represent an ideal comparator. Second, the absence of structural effects has been consistently documented after several years of TNF inhibition therapy, whereas the structural benefits of TNF inhibition in RA become evident as early as a few months after the initiation of treatment. Whether small structural benefits after an even longer exposure to TNF inhibition may exist has yet to be confirmed, but may be of limited clinical relevance anyway. Also, such long-term associations may be confounded by other environmental and disease-specific variables. Third, outcome measures for syndesmophyte formation in AS, such as scores on the SASSS, have their limitations; however, one should also consider that the same measures have allowed us to document the effects of NSAIDs on syndesmophyte formation. These latter data further support the concept that control of inflammation is not sufficient to affect syndesmophyte formation and that antianabolic effects of NSAIDs on the bone rather than their antiinflammatory effects may explain the inhibition of syndesmophyte formation.

There is wide agreement that animal models will never represent a one-to-one model for a human disease. However, animal models teach us common concepts of disease. It is thus striking that all arthritis models with an initial acute inflammatory phase followed by a gradual decrease in the intensity of inflammation show clear evidence of new bone formation. This finding is highly consistent among different arthritis models and different genetic backgrounds, such as collagen-induced, adjuvant-induced, or serum transfer–induced arthritis. An additional, very appealing, example is the male DBA/1 mouse model, which has a very short and spurious inflammatory phase that is followed by excessive new bone formation. In contrast, animal models based on permanent inflammatory burden, such as TNF-transgenic mice, which never reach the resolution phase of inflammation, do not show signs of new bone formation. These observations suggest that resolution of inflammation, rather than inflammation itself, is of key importance for new bone formation in AS.

In summary, current data suggest that syndesmophyte formation is driven by mesenchymal tissue responses associated with the deposition of bone. The cause for this sometimes excessive mesenchymal tissue response is as yet unknown, but it appears that resolution of inflammation, rather than inflammation itself, is required to start this process. Other factors, such as mechanical stress, however, may be important additional triggers for the mesenchymal tissue response in AS. It is thus of seminal importance to define the molecular switches that combine resolution of inflammation, mechanical stress, and new bone formation.

REFERENCES: In support:

  1. Top of page
  2. Presentation of the debate, Dr. Maksymowych: Introduction to the motion
  3. In support, Dr. Elewaut: Inflammation and new bone formation in ankylosing spondylitis are tightly linked
  4. In opposition, Dr. Schett: Inflammation and new bone formation in ankylosing spondylitis are independent processes
  5. In support, Dr. Elewaut: Rebuttal
  6. In opposition, Dr. Schett: Rebuttal
  7. AUTHOR CONTRIBUTIONS
  8. REFERENCES: In support:
  9. In opposition:

In opposition:

  1. Top of page
  2. Presentation of the debate, Dr. Maksymowych: Introduction to the motion
  3. In support, Dr. Elewaut: Inflammation and new bone formation in ankylosing spondylitis are tightly linked
  4. In opposition, Dr. Schett: Inflammation and new bone formation in ankylosing spondylitis are independent processes
  5. In support, Dr. Elewaut: Rebuttal
  6. In opposition, Dr. Schett: Rebuttal
  7. AUTHOR CONTRIBUTIONS
  8. REFERENCES: In support:
  9. In opposition: