Soluble and Tissue Biomarkers in Ankylosing Spondylitis

The study of biomarkers has become a very important field of research in spondyloarthropathy. Biomarkers are useful for different aspects of the disease such as diagnosis, assessment of disease activity and outcome, including damage. The most commonly used biomarkers in spondyloarthropathies are HLA-B27 for diagnosis and erythrocyte sedimentation rate (ESR) and C-reactive protein (CRP) for disease activity. HLA-B27 is very sensitive but has a low specificity. ESR and CRP have both low sensitivity and specificity. The introduction of new and very expensive therapies is another reason for analysis of biomarkers. Clinicians need tools to predict more accurately disease activity, disease progression and response to therapy. This article focusses on the several known and new biomarkers of promise, including markers for cartilage and bone damage, and discusses some of the problems encountered during the search and development of new biomarkers.

Biomarkers, soluble and tissue-related, reflecting structural damage and disease activity, constitute a high priority for the drug discovery process and the understanding of the pathogenesis of a particular disease. The identification of relevant tools to evaluate the natural course, disease activity, treatment response and outcome of ankylosing spondylitis is of increasing relevance since the raised awareness and development of new therapeutic options. Until now these different aspects are monitored by artificial patient-centred or physician-centred constructs. Very often, their approach is indirect and is not free from disease-unrelated influences. The Outcome Measures in Rheumatology Soluble Biomarker Working Group has taken several major steps towards the development and implementation of such assessment methods. The major drawback is that these tools do not directly reflect biological and pathological processes. Serological biomarkers objectively measure different aspects of the biological and pathological process and may contribute to a major advance in the assessments of patients. The ultimate goal is the use of biomarkers in a personalised approach for disease management in clinical practice.

The National Institutes of Health (NIH) Biomarkers and Surrogate Endpoint Working Group defines a biological biomarker as a characteristic that is objectively measured and evaluated as an indicator of normal biological processes, pathogenic processes or pharmacological responses to therapeutic intervention . Biological biomarkers include markers for drug effect or response, as well as for diagnosis, prognosis or physiological status information, such as disease activity or damage not related to drug effect. The Food and Drug Administration (FDA) distinguishes three context-specific types of biomarkers: known valid biomarker, probable valid biomarker and possible valid biomarker based on the available scientific data for the biomarker ( http://www.fda.gov/cber/gdlns/pharmdtasub.htm ) .

Biomarkers allow quantitative assessment of diagnosis, disease processes and treatment response and therefore are important in clinical practice. They enable appropriate choice of therapy and drug dosage to maximise effect, minimise toxicity and monitor disease outcomes, and thus are the foundation of evidence-based medicine.

Conservative thinking, lack of quality control and methodological issues such as inappropriate statistical analysis and validation seriously hamper the development of new biomarkers. For this reason, the development of biomarkers lags significantly behind that of drug development. The absence of new and appropriate markers may slow down the evolution of patient-tailored targeted therapies. Within the rheumatology community under the umbrella of OMERACT (Outcome Measures in Rheumatology), a Working Group has developed clinical validation criteria for soluble biomarkers reflecting structural damage in rheumatoid arthritis (RA) and spondyloarthropathies (SpA) .( http://www.omeract.org/ ).

The techniques used to develop and measure biomarkers are diverse, for example, in vitro analyses such as protein expression, gene patterns or gene expression, and in vivo analyses such as in functional imaging that are still in early development in humans. Technologies available today may evaluate biochemical and chemical markers by proteomics and metabolomics; genetic markers using pharmacogenomics, gene expression profiles, systems biology and single nucleotide polymorphisms (SNPs) and structural markers using classical molecular imaging techniques.

Possible use of biomarkers in ankylosing spondylitis

Outcome or clinical endpoints in ankylosing spondylitis (AS) include diagnosis, inflammation, prognosis, disease activity, disease severity, damage, disability and quality of life (QoL). Recently, specific instruments have been developed to measure disease activity or damage but for other domains the instruments are still under development or not yet widely validated . Up till now there is a reasonable consensus about the definition of spondyloarthropathies and AS and about the use of bath ankylosing spondylitis activity index (BASDAI) as the gold standard for disease activity. Several sets of classification criteria for AS (modified New York criteria) and SpA (ESSG and Amor criteria) are available . Recently, new classification criteria were developed for axial SpA under the umbrella of Assessment of SpondyloArthritis International Society (ASAS) . Criteria for peripheral SpA are under development. Moreover, there is general agreement that SpA responds well to tumour necrosis factor (TNF) blockade. In general, erythrocyte sedimentation rate (ESR) and C-reactive protein (CRP) correlate poorly with disease activity evaluated by BASDAI or patient global assessment of disease activity, especially in patients with exclusive axial disease . ESR, CRP and even BASDAI cover the concept of disease activity only partly. Spinal inflammation detected by magnetic resonance imaging (MRI) cannot predict BASDAI scores in patients treated with anti-TNF. Different scoring systems to characterise inflammatory spinal lesions have been developed, including ASspiMRI (the AS spinal MRI score) and the spondyloarthritis research consortium of canada (SPARCC) MRI spinal inflammation index, but are recommended by ASAS only for study purposes .Moreover, it is not yet clear how MRI changes relate to changes in clinical variables. Therefore, MRI is not recommended for the evaluation of disease activity and treatment response in patients with AS .

It is agreed that HLA-B27 is a biomarker for diagnosis with high sensitivity but low specificity.

Instrument development for disease assessment must fulfil the OMERACT filter for validity (truth, feasibility and discrimination). For some domains, no gold-standard instrument has been identified. Additionally, different studies have used different instruments making cross-comparisons difficult. A better approach might be to define domains based on the biology of the disease and then to look for the outcome of different biological processes such as the inflammatory process or the tissue response including damage.

Biomarkers in AS can mainly be determined in blood (serum/cellular compartment). Synovial fluid and synovial tissue are useful only in patients with concomitant peripheral involvement. In patients with axial involvement or enthesial involvement, tissue from the spine or enthesis is difficult to obtain and subsequently not useful for regular evaluation.

Recent evolution in the understanding of AS and SpA emphasises the need for biomarkers that facilitate understanding of the underlying pathophysiological processes, diagnostic ascertainment, prediction of prognosis, assessment of burden of disease and, last but not least, prediction of treatment response. During the last decade, the development of TNF-blocking agents with beneficial symptomatic responses has dramatically changed the therapeutic approach in these patients. The cost and possible serious side effects are the major drawbacks of these therapeutic options. Biomarkers, which can predict the degree and duration of response and can detect possible side effects in an early stage, are needed.

Biomarkers for diagnosis

Diagnosing AS is a clinically driven process based on the observation of clinical signs and structural changes on X-rays. Classification criteria are useful for constituting uniform groups of patients but not for diagnosis. The modified New York criteria for AS are widespread and have good sensitivity and specificity but the presence of radiographic sacroiliitis is mandatory. The structural changes on X-rays are the consequence of the inflammatory process but do not reflect the inflammatory process itself. It takes several years before these structural changes are visible on conventional X-rays. These criteria only capture established forms and lack sensitivity for the earlier forms. Recently, a new set of criteria have been developed to facilitate the identification of earlier forms. A diagnostic algorithm was developed based on the available sensitivities, specificities and likelihood ratio for SpA-related features. This approach has been validated by the ASAS in an international study . These criteria were shown to be superior to the older criteria such as the ESSG and Amor criteria. The high association of HLA-B27 with AS, combined with a relatively high population prevalence, does not make it a good candidate for use as a screening marker alone but in a combined model, it can be very useful. Biomarkers for the diagnosis of AS must be evaluated in cohorts of patients with an established diagnosis based on the current gold standard, that is, the modified New York criteria for AS and radiographic sacroiliitis, in particular. An appropriate control population, including normal subjects or patients with mechanical low back or disc herniation, is essential. A statistically significant difference in frequency or level of biomarker between patients and controls must be combined with additional calculations for likelihood or specificity and sensitivity for selection of candidate biomarkers. For this purpose, the receiver operator characteristic (ROC) is often used. ROC curves combine sensitivity and specificity and are expressed by the area under the curve (AUC) and a score above 0.75 is acceptable.

The choice of the biomarker depends on the current understanding of disease susceptibility or pathogenic mechanisms. Since the genetic contribution to disease susceptibility is very high in AS, assessment of genetic markers (genotypes or SNPs) may be useful. The strong association with HLA-B27 has been known for a long time but is not very useful for diagnosis due to its low specificity. Recent advances in the genetics of AS have shown that it is a multigenic disease in which HLA-B27 contributes only about 20–30% of the genetic risk raising the possibility that additional genes such as IL1, ERAP-1 and IL23R may improve diagnostic accuracy in AS by combination approaches .

CRP failed as a diagnostic biomarker due to its low sensitivity that ranges from 38% to 75% . This is in contrast with the requirements for a diagnostic tool which requires maximal sensitivity without sacrificing specificity.

An alternative approach is to look for proteins or biological products as biomarkers reflecting distinct disease processes in AS as compared to other forms of chronic arthritis.

Since AS is characterised by bone remodelling, specific markers for bone formation may be promising candidates that help differentiate AS from other chronic arthritides . Osteopontin, a protein with an important role in bone remodelling, induces anchoring of osteoclasts to the bone matrix, and is also involved in inflammation. It shows increased levels in patients with AS compared to healthy controls. It also correlates with other proteins such as serum alkaline phosphatase (ALP), TNF-α and IL 6 but not with disease activity and reflects the bone remodelling process rather than disease activity in AS . Runx2 is a transcription factor related to bone formation and is involved in osteoblast differentiation. The increased expression of Runx2 in peripheral blood cells of patients with AS discriminates them from those with RA, who express lower levels of Runx2 .

Biomarkers for the stage of disease may also be of interest, especially since there is still a mean delay of 7 years between the first symptoms and diagnosis . A combination of markers for diagnosis and stage of disease may support the goals of early treatment. It is hypothesised but not yet proven that early treatment of AS may prevent the development of ankylosis. Levels of STREM-1, a soluble triggering receptor expressed on myeloid cells-1 (≥ 15 pg ml ⁻¹), are significantly higher in patients with AS compared to controls (31.3% vs. 10%; p < 0.027) but this level of sensitivity is insufficient for a diagnostic marker. Within the groups of patients with AS, STREM-1 correlates inversely with disease duration in AS and may be a marker for early disease in AS but is not related to disease activity .

Several groups have evaluated synovial tissue as a biomarker in spondyloarthritis or AS patients with peripheral involvement. Although peripheral arthritis in AS differs clinically, serologically and radiographically from other forms of chronic arthritis, some patients have an atypical presentation in clinical practice and may be difficult to label with a definite diagnosis .

In a diagnostic setting, no single histological feature of the synovium discriminates SpA from other forms of chronic arthritis. Baeten et al. developed a multi-parameter algorithm enabling the identification of the majority of patients with SpA. The characteristics that made the diagnosis of SpA likely were: (1) pronounced vascularisation, (2) moderate hyperplasia of the intimal lining layer, (3) increased presence of the scavenger receptor CD 163 in the lining and sublining layer and (4) the relative abundance of polymorphonuclear cells in the synovial tissue . Using a multi-parameter setting including synovial histopathology, staining with anti-citrulline, staining with MHC class II-HC gp39 peptide complex and detection for crystal deposition in patients with an atypical clinical presentation can be very useful. Only microscopic vascularity score >2 and tortuous vascular pattern has a high predictive value for SpA .