Development of disease-modifying treatments for rheumatoid arthritis
The treatment paradigm for rheumatic arthritis has now developed fromone of controlling pain and inflammation to one of preventing jointdestruction. Michael Seed discusses the development of novel drugs andthe role of animal models in their development
The treatment paradigm for rheumatic arthritis has now developed from one of controlling pain and inflammation to one of preventing joint destruction. Michael Seed discusses the development of novel drugs and the role of animal models in their development
Rheumatoid arthritis (RA) is an extremely painful, debilitating and destructive inflammatory disease that destroys diarthrodial joints. It affects between 0.5 and 1 per cent of the world’s population, and 420,000 in the UK alone, with women having a three-fold prevalence. Most will develop moderate disability at two years, with 40 per cent being unable to work at five years.
Traditionally treated with non-steroidal anti-inflammatory drugs, toxic disease-modifying drugs and anti-inflammatory cortico-steroids, there has been a step change in the discovery of the efficacy of low-dose methotrexate, and the discovery and exploitation of new therapeutic targets, such as the cytokines tumour necrosis factor, interleukin-1 and interleukin-6.
The treatment paradigm has now developed from one of controlling pain and inflammation to one of preventing joint destruction. However, anti-TNF therapies, such as the biologics etanercept and infliximab, have a 40 per cent failure rate.
The reasons for this are unclear but include population variations in TNF sequence to which anti-TNF antibodies have reduced affinity, the possibility of a TNF independent stage of the disease or the generation of patient immunity to the biologic.
Anti-B cell therapy has also been most successful, but the wholesale destruction of B cells for prolonged periods is worrying, although there are indications that tolerance may be induced afterwards.
These treatments are expensive (approximately £10,000 per annum), and requires outpatient administration and monitoring. Efficacy could be improved, and non-responders and costs could be overcome through refining existing therapeutics, identifying mechanisms that lead to null-responsiveness, the discovery and exploitation of new targets, and the development of orally active low molecular weight drugs acting through existing or novel targets.
Animal models of rheumatic disease play a significant role. Indeed, these have been instrumental in the proof of concept of the new therapies.
Model of drug discovery
The discovery and development of anti-TNF therapy is a model of modern drug discovery. The Kennedy Institute team1 discovered and developed the concept that TNF was a central mediator in the cytokine cascade in RA by merging human observations into testable hypotheses in animals.
TNF was known to be pro-inflammatory in animals and to be involved in joint destruction. It was identified in human RA synovium, in the synovial RA pannus and at the junction of the inflammatory tissue and degrading cartilage.
Assessing a wide variety of cytokines on RA synovium in culture showed that TNF and IL-1 were fundamental. Removal of TNF also reduced IL-1 synthesis and TNF replacement reversed these effects.
Testing this new hypothesis was not possible in the human. However, there are several rodent models of polyarthritic disease, one being collagen-induced arthritis.2
Using susceptible strains, immune sensitisation to collagen results in an aggressive arthritis characterised by joint inflammation and destruction of cartilage in bone. Proof of concept was obtained by inhibiting this disease using anti-mouse TNF.
The final confirmation was through the generation of human TNF hyperexpressing dba/1 mice, which developed a polyarthritis that was abrogated by anti-human TNF therapy. A pilot clinical trial with a monoclonal anti-human TNF antibody resulted in a dramatic improvement in swollen joint count at three weeks.
The hypothesis testing permitted by using this model was essential to the development of TNF as a valid drug target. The rest is history.
Animal models of immunity and arthritis have proven invaluable in developing other antirheumatic therapies, as well as revealing possible immune-mechanisms in the disease through hypothesis testing.
There are several important models, including collagen arthritis (in the rat, mouse or non-human primate), rat adjuvant arthritis (antigen or non-antigen), antigen arthritis (rat, mouse or rabbit), K/BxN arthritis and streptococcal cell wall arthritis.
Additionally, target knock-in or knock-out mice and immune system models through selective hypersensitivity reactions are used to establish the response of defined immune mechanisms to novel targets and therapies and give important indications of immune safety.
Cause and chronicity
Animal models and RA disease involve disorders of the immune system. Both show large variations in inflammatory cytokines, chemokines, adhesion molecules, cell activation and gene expression. Some overlap between the species, depending on the models, and others do not. Discoveries in animal models are translated into the human disease and then fed back.3
A good example is the recent interest in the T17-helper cell, a unique CD4(+) T-cell subset characterised by production of the cytokine IL-17 in RA. The first observations were by Wim van den Berg and his team,4 investigating the role of the then unknown cytokine IL-17 in antigen-induced arthritis.
Subsequent pioneering work illustrated that chronic joint disease evolved, initially being TNF-dependent, switching to IL-17 dependency in longer, chronic disease. The role of T17 cells in RA is now considered pivotal to our understanding of the immune cellular network of pannus, the rheumatoid granulation tissue that invades and destroys joints.
These models have played a key role in testing our understanding of the cause of the disease. RA is now recognised to derive from more than one inherited factor. For example, the best characterised are HLA DRB single epitope polymorphisms.
These are the major histocompatibility complex (MHC) class II molecules expressed on cells that present antigen to the CD4 T-helper cells. The mutations are related to the cleft in which the antigen binds and it is supposed that they present auto-antigens.
Finding a single antigen has been unsuccessful. Numerous factors have been proposed through the reactivity of T cells or antibodies of some patients to (among others) peptidoglycans, collagen, the Epstein Barr virus and, more recently, citrullinated peptides, where each is shown to be arthritogenic or to enhance arthritis through animal modelling.
These all point to the central role of not only the T cell, but also the B cell. Indeed, anti-B-cell therapy (rituximab) is highly successful, although drastic.
In addition to the sensitisation protocols, anticollagen antibodies will induce a poly-arthritis in a wide variety of otherwise unsusceptible strains, as will auto-antibodies to glucose-6-phosphate isomerase in K/BxN mice.
Interesting is the observation that non-antigenic stimuli, such as mineral oil pristane, can induce polyarthritis, which is MHC dependent and transferable by memory T cells. It appears to induce auto-reactivity to hnRNP-A2.
However, none of these systems has transferred en bloc to human disease. Possibly, at the initiation of RA, there is a restricted antigen and, as chronicity develops, epitope drift occurs with the disease becoming self-sustaining.
Indeed, rheumatoid pannus transplanted into severe combined immunodeficiency (SCID) mice is remarkably robust, retaining its architecture, cytokine synthetic profile and capacity to produce anticitrullinated peptide antibodies for several weeks.5
These polyarthritis models then, unlike RA, appear to be dependent on one antigen and are, therefore, criticised as having a single stimulus and being self-limiting. However, RA starts at a defined point and, if left to run, is indeed self-limiting, ending in ankylosis, as does rat adjuvant arthritis, for example.
The human disease takes decades to burn out while the rodent models take weeks. RA is also a cyclic disease, with periods of severe inflammation followed by less vigorous periods. This is mimicked by a relapsing model of autologous collagen-induced arthritis, which is a promising development. The question remains then, what of epitope spreading?
A major rethink in the epitope field in RA is the discovery that patients can be divided into those who have antibodies reactive to citrullinated epitopes and those who do not. This was based on findings using rat tissues that antibodies (termed antikeratin antibodies) expressed early in RA selectively bound to rat oesophagus.
These antibodies are expressed and have similar properties to anti-perinuclear factor, which binds profilaggrin and filaggrin. It was shown that the citrullination of profilaggrin by peptidylarginine deiminase (PAD) to filaggrin could enhance antigenicity. RA antibodies were then discovered to a variety of citrullinated filaggrin peptides.
This fundamental study, based on observation in animal tissues, resulted in a major rethink in our understanding of RA epitopes. Between 50 and 70 per cent of RA patients, compared with 2 per cent or less in healthy populations, express anticitrullinated protein antibodies (ACPA). PAD is expressed in RA synovium, and many citrullinated epitopes have been discovered.
However, these observations could be epiphenomena. PAD and citrullinated peptides are found in inflammatory tissues. Studies using murine collagen arthritis showed PAD and citrullinated epitopes in the synovium, and citrullination to increase the antigenicity of collagen to enhance disease.
Transgenic mice with the human HLA-DRB1 shared epitope have an exaggerated CD4 T-cell response when exposed to peptides that are citrullinated at the shared epitope specific recognition sites. Importantly, citrullination in collagen arthritic mice results in epitope spreading, paralleling the variety of ACPAs in RA.
So, a combination of animal theoretical and human observational research has transformed our understanding of the origins of RA. We now know that ACPA-positive individuals are linked to HLA DRB1 shared epitope, PTPN22, or CTLA4, as well as to the environment, for example, smoking.
Lars Klarescog6 surmised that citrullinated epitopes, derived in the lung from smoking, results in the development of ACPAs. The citrullinated epitopes formed during synovitis from an undefined cause, such as injury or infection, may initiate disease in susceptible individuals.
In ACPA-positive patients, PAD could, therefore, serve to be a novel drug target. The state of play for rheumatoid factor and ACPA-negative arthritides remains open.
Validation of novel molecular targets
For the appraisal of novel drug targets, the models of polyarthritis have been an essential, although sometimes misleading, part of the process.7
However, appraising successful and unsuccessful antirheumatic agents, and the models used in their development, shows that the three most successful of the modern therapies (methotrexate, etanercept and anakinra-an antagonist of IL-1) are all active in rat and mouse collagen and adjuvant arthritis.
The development of human species-selective therapeutics requires, for example, collagen-induced arthritis in non-human primates. B-cell depleting therapeutics, such as rituximab and belimumab, are tested for B-cell depletion in non-human primates without recourse to a disease model, since this is now well calibrated to the human. B-cell depletion in any event is effective in the rodent collagen arthritides, but not rat adjuvant arthritis.
Human-specific targets are verified in rodent models using parallel series of rodent specific molecules as well as target knock-out/knock-in technology.
There are, of course, examples where treatments are active in animal models and not in the clinic. The reasons for the failure of such clinical trials are many, often not the fault of preclinical modelling. Failure may reside in pharmaceutical or pharmacokinetic issues, incorrect dose ranging or something as simple as low affinity for the target.
Other reasons include the use of an inappropriate patient base in the clinical trial, non-responders or end-of-the-line RA. Certainly, the bar has been raised by the biologics, making it difficult for new agents to achieve therapeutic parity, for example, p38 inhibitors.
The risk of translational issues such as these can now be overcome using xenografts of human RA synovium implanted in immune-deficient SCID mice. These grafts function for several weeks and can be used to study cell adhesion and homing to the tissues, as well as assessing treatment effects on tissue organisation, gene expression, cytokine networks and erosive capacity.
For example, human specific anti-B cell therapy induces synovial disaggregation and dissolution of the graft. It is envisaged that these translational models replace the non-human primate, however, they are limited by the availability of human RA tissue.
Animal models for RA provide an essential part of an integrated approach to the discovery and verification process, with targets calibrated as far as it is possible to human disease. The blind acceptance of anti-rheumatic data in a model is rightfully long past, and can be misleading.
Our understanding of human rheumatoid disease has been fuelled by animal modelling, and recent advances in the models themselves and analytical technology are beginning to show a slow convergence between the two.
Michael Seed is William Harvey Inflammation Fellow and Sir Halley Stewart Trust Lecturer at the William Harvey Research Institute, Centre for Experimental Medicine and Rheumatology, Saint Barts, and the London Queen Mary’s School of Medicine and Dentistry, Queen Mary University of London
1. Feldmann M, Brennan F, Williams R, Woody J, Maini R. The transfer of a laboratory based hypothesis to a clinically useful therapy: the development of anti-TNF therapy of rheumatoid arthritis. Best Practice and Research Clinical Rheumatology 2004;18:59–80.
2. Koenders M, Joosten L, van den Berg WB. Potential new targets in arthritis therapy: interleukin (IL)-17 and its relation to tumour necrosis factor and IL-1 in experimental arthritis. Annals of Rheumatic Diseases 2006;65:Suppl 3:29–33.
3. Joosten L, van den Berg W. Murine collagen induced arthritis. In: Parnham MJ, editor. In vivo models of inflammation. 2nd ed. Progress in inflammation research. Basel: Birkhäuser, 2006:35–63.
4. McCann F, Williams R. Novel approaches to the therapy of rheumatoid arthritis based on an understanding of disease mechanisms. Immunology, Endocrine and Metabolic Agents in Medicinal Chemistry 2008;8:275–83.
5. Humby F, Bombardieri M, Manzo A, Kelly S, Blades M, Kirkham B, et al. Ectopic lymphoid structures support ongoing production of class-switched autoantibodies in rheumatoid synovium. PLoS Medicine 2009;6:e1.
6. Klareskog L, Padyukov L, Rönnelid J, Alfredsson L. Genes, environment and immunity in the development of rheumatoid arthritis. Current Opinion in Immunology 2006;18:650–5.
7. Schopf L, Anderson K, Jafee B. Rat models of arthritis: similarities, differences, advantages and disadvantages in the identification of novel therapeutics. In: Parnham MJ, editor. In Vivo Models of Inflammation. 2nd Ed. Progress in Inflammation Research. Basel: Birkhäuser, 2006: 1–63.
Citation: The Pharmaceutical Journal URI: 10062494
Recommended from Pharmaceutical Press