Reducing Diagnostic Delay in Axial Spondyloarthritis: Could Lipocalin 2 Biomarkers Help?

Abstract

Early diagnosis and therapy in axial spondyloarthritis, axSpA, is known to reduce long-term morbidity. However, the time from symptom onset to diagnosis is typically delayed by several years, and this situation has not improved in recent years despite greater clinical awareness and better imaging. This narrative review discusses the underlying causes for axSpA diagnostic delay. It is proposed that to reduce axSpA diagnostic delay, a better understanding of the axSpA subclinical inflammatory process is required, together with machine learning-enabled inflammation/repair biomarkers such as lipocalin 2 and lipocalin 2-matrix metalloprotease 9, developed through extensive clinical domain knowledge.

1. Introduction

Axial spondyloarthritis, axSpA, is a debilitating chronic inflammatory disease usually characterized initially by lower back pain and the associated imaging changes in the spine and sacroiliac joints. Diagnosis is usually delayed for an average of about 7 years from the onset of lower back pain, a delay which, despite intense clinical study, has not decreased for several decades [1,2,3,4,5,6,7,8,9,10]. With each passing year of diagnostic delay, the quality of life for affected patients decreases [11]. This narrative commentary will review the factors proposed for axSpA diagnostic delay, consider the underlying factors contributing to lack of progress in delay reduction, and offer an approach that likely has promise for identifying axSpA earlier and with additional precision.

2. Current Approaches to axSpA Diagnosis

Historically, the diagnostic criteria for axSpA were defined by rheumatologist committee consensus [12,13,14,15]. This reflects the observations that established that “ankylosing spondylitis” displays features of radiologic sacroiliitis, spinal neo-ossification, and genetic HLA-B27 predominance. However, at least in the early stages, axSpA does not have a pathognomonic feature distinguishing the disease from chronic and nonspecific lower back pain. As the study of the disease progressed, the committee consensus defined an additional diagnostic category, the non-radiologic variant without x-ray features, nr-axSpA [16,17]. Although originally considered as an early stage of ax-SpA, this diagnostic category is distinctive as only about 10% of nr-axSpA cases evolve to radiologic-axSpA [18]. While recognizing that nr-axSpA has expanded the population with ax-SpA, particularly for women and HLA-B27-negative individuals, this advance did not result in earlier disease diagnosis. Similarly, some patients with other conditions such as inflammatory bowel disease [19] or psoriasis [20,21] were found to have axSpA as well [22]. Again, while this expanded the known axSpA population, the time to diagnosis was not reduced. The time from first symptoms to axSpA diagnosis varies greatly in different populations [3,4,7,8,9]. Recently, this has led patient support organizations to initiate a global effort to raise axSpA awareness and to provide clear, standardized pathways for specialist referrals for patients experiencing chronic lower back pain [1].

3. What Are the Underlying Causes for axSpA Diagnostic Delay?

To reduce axSpA diagnostic delay, rheumatologists have been focused on trying to define early symptoms and signs [23,24], deploy more sensitive imaging [22], develop novel biomarkers [25,26,27], as well as address health system deficiencies [3,5] which limit disease recognition and patient access to care. As laudable as the progress that has been made is, axSpA diagnostic delay has not been reduced. Some underlying reasons are obvious. Clinical signs and symptoms of early disease are not specific and are often evanescent. Except for spinal neossification, present in only about 50% of established cases, there are no pathognmonnic features of axSpA. Hence, axSpA categorization and classification continues to be set by committee criteria. Because of overlapping features, the imaging of “early disease” is often problematic [28]. Furthermore, what is considered early by imaging may become evident only after years of clinical symptoms or other disease features. The standard serum biomarker C-reactive protein (CRP), now known to be a marker of systemic inflammation, exceeds the 11 mg/L threshold for clinical inflammation in only about 30% of axSpA cases [29]. Similar to studies of atherosclerosis and cardiovascular morbidity, CRP in the range of >3 mg/L and <11 mg/L, indicative of subclinical systemic inflammation, has been found in chronic lower back pain patients who evolve to axSpA [30]. However, serum CRP > 3 and <11 is currently thought to be an inadequate diagnostic and weak prognostic indicator.

Since HLA-B27 was discovered to be associated with axSpA over 50 years ago [31], serum biomarkers related to axSpA susceptibility, diagnosis, and inflammation activity have been extensively studied. In recent years, with the advent of TNF inhibitors and other biologic therapeutics, the pace of biomarker investigations has since accelerated [32]. Figure 1.

Serum biomarker studies have been focused on individual immunologic, protein, or genomic markers, or multiple markers analyzed individually or as signatures, usually by conventional statistics. The general lack of success for biomarkers as a diagnostic feature of axSpA can be attributed to a lack of deep understanding of the axSpA inflammation process and, in some cases, insufficient knowledge of the role and metabolism of the biomarkers studied.

Notable features of inflammation in axSpA include the following:

• Subclinical inflammation manifested as chronic back pain and fatigue can persist for long periods without changes in physical signs or imaging.
• Inflammation/repair phases in chronic diseases can be incomplete and can have different patterns in different diseases.
• Therapies can modify inflammation/repair phases in both amplitude and duration.
• Analgesics and anti-inflammatory drugs used nonspecifically to treat chronic back pain can mask ongoing subclinical inflammation.

4. Biomarkers to Reduce axSpA Diagnostic Delay

A promising avenue for biomarker diagnostics for axSpA is to exploit the disease’s specific features within the axSpA inflammation/repair process that are gradually becoming recognized [33].

Enablers of this approach include the selection of biomarkers that are generated locally at inflammation sites, biomarkers that are cleared via the kidneys but not catabolized or phagocytized, and biomarkers that are related to each other metabolically but have different concentrations or other features in different inflammation phases. To analyze these biomarkers, it is necessary to demonstrate their serum concentration relationships in individual patients.

This personalized medicine approach for biomarkers can be facilitated by the machine learning analysis of serial samples coupled to existing clinical domain knowledge related to the inflammation processes specific to axSpA [34,35]. A candidate biomarker system is lipocalin 2 (LCN2) and lipocalin 2-matrix metalloprotease 9 (LCN2-MMP9). Lipocalin 2, formerly known as neutrophil gelatinase-associated lipocalin (NGAL) B [36], is elevated in a wide variety of acute and chronic inflammatory states [37,38] including axial spondyloarthritis [39,40,41,42]. In active inflammation, LCN2, a component of cell membranes, shows increased synthesis and a subsequent release into the extracellular space from cells at the inflammation site. Locally, LCN2 acts to block the cellular uptake of extracellular protein fragments [43]. Similarly, inflammation stimulates MMP9 synthesis and its release from fibroblasts into the extracellular spaces [44,45]. Activated MMP9 breaks down extracellular proteins, facilitating tissue resorption early in inflammation, and subsequently enabling collagen polymerization (fibrosis) during repair. Also, in inflammation, protein disulfide isomerase in the oxidative environment [46] facilitates disulfide bond breakdown and re-isomerization in both LCN2 and MMP9, resulting in LCN2–MMP9 formation [47,48]. With LCN2 bound to MMP9 as LCN2–MMP9, MMP9 prevents MMP auto-degradation catalytic activity, thereby enhancing MMP9 proteolytic activity in inflammation [49].

Moreover, LCN2–MMP9 and its functional effects reflect the very special molecular relationship between LCN2 and MMP9 as there is convincing evidence that LCN2 and MMP9, alone amongst the metalloproteases, have co-evolved and that this coevolution is restricted to primates [50].

Both LCN2 and LCN2–MMP9 are resistant to catabolism and are cleared by the kidneys. Hence, these molecules in serum comprise a useful biomarker system, with LCN2 reflecting inflammation, LCN2–MMP9 reflecting repair activity. Individually, both serum LCN2 and LCN2–MMP9 have been shown to be elevated in many clinical inflammatory states, with biomarker elevations observed as early as 24 h after inflammation onset. Development of a test suite utilizing serum LCN2, LCN2–MMP9, and machine learning to elucidate the biomarker relationships with each other and with the clinical outcome data, shows promise for an inflammation biomarker that is more sensitive and more specific for the early diagnosis of axSpA [51,52]. Moreover, the LCN2 and LCN2-MMP9 system can quantitatively reflect inflammation and repair activity [51,52,53]. With therapy, changes in these markers can potentially reflect axSpA treatment response with increased accuracy.

5. Conclusions and Future Directions

Advances in the understanding of the pathophysiology of inflammation and repair processes offer the prospect that serum biomarkers which are indicative of the phase, amplitude, and balance of inflammation and repair will enable us to detect axSpA-specific inflammation patterns early in the disease where physical signs and imaging of inflammation are still subclinical. The clinical utility of these markers will be enhanced by machine learning coupled with axSpA-specific clinical domain knowledge.