Blood Biomarkers for Monitoring and Prognosis of Large Vessel Vasculitides

Acute phase reactants performed among routine laboratory investigations include erythrocyte sedimentation rate (ESR) and C-reactive protein (CRP). These examinations are inexpensive and widely used in clinical practice. CRP is directly produced by the liver in response to interleukin (IL)-6 [17, 18]. Conversely, a raised ESR is related to multifactorial mechanisms, including increased levels of fibrinogen, alpha-1 and gamma-globulins, as well as reduced plasma albumin and anaemia [19, 20]. ESR increases progressively with age, and this is believed to reflect the accrual of oxidative damage and the progressive establishment of a pro-inflammatory milieu in the elderly, collectively known as ‘inflammaging’ [21]. There is no consensus concerning the upper limits of normal for the ESR, although a simple rule identifies the cut-off at age/2 (+ 5, if female) [22]. However, no cause was identified in up to 47% of subjects with an ESR between 50 and 99 mm/h [23], limiting the specificity of such marker.

Elevation in the concentration of acute phase reactants is frequent during active TA and may help to uncover an inflammatory state and pave the way for further workup. The intensity of the systemic inflammatory response is highly variable and normal acute phase reactants were observed in 10–30% of patients in some studies [4, 24••]. A proportion of these patients [4, 24••], however, were diagnosed in an inactive phase of TA and it is unclear whether they might have experienced a prior systemic inflammatory response.

In GCA, a highly intense systemic inflammatory response is typical. Indeed, ESR values above 50 mm/h are part of the 1990 classification criteria of the American College of Rheumatology (ACR) for GCA [25]. However, in a cohort of 177 patients with biopsy-proven GCA [10], eighteen (10.2%) showed normal values of both ESR and CRP (defined as an ESR value of ≤ 22 mm/h in men and ≤ 29 mm/h in women, and a value of CRP ≤ 8 mg/L), at the time of diagnosis. It should be noted that only 7 of these 18 patients were GC-naïve at the time of the biopsy and this might have influenced the results. Visual disturbance despite normal ESR and CRP was present in 3 of the 7 patients, of whom one had visual loss. A systematic review showed the percentage of patients with GCA with normal inflammatory indices may range from 0 to 22% [26].

Another major acute phase parameter is serum amyloid A (SAA), synthetised and secreted by the liver in different isoforms and at inflamed sites including the temporal arteries [27]. It has been recently observed that this SAA participates in blood vessel modification, invasion and destruction [28]. SAA also induces the synthesis of IL-6 and has proangiogenic properties through upregulation of vascular endothelial growth factor receptor-2 (VEGFR-2) [29]. A cross-sectional study on a cohort of 99 TA patients (with a mean disease duration of 73 months) showed higher levels of serum SAA as compared to healthy controls [30]. Despite mean SAA levels significantly differed from TA subjects and controls, overlap was substantial and patients could not be accurately discriminated by using SAA serum levels alone [30].

In GCA, SAA serum concentration levels significantly differ between untreated GCA and healthy controls, with an average 83-fold increase in the former as compared to controls [31]. Interestingly, SAA concentrations were the strongest positive correlate of IL-6 levels [31]. No study has compared SAA in treatment-naïve GCA and treated active or inactive disease versus healthy controls. Being identified as a marker of disease activity, SAA may also be helpful in defining a relapse [32]. However, similar to ESR and CRP, as SAA levels may increase also in other conditions (such as microbial infections), its value for diagnosing GCA appears to be limited.

Considering the inflammatory milieu of LVV, biomarker research in the field has studied several other inflammatory molecules. IL-6 is one of the pivotal pro-inflammatory cytokines and may be locally generated in the inflamed arteries both in TA and GCA [33, 34]. High IL-6 release is considered to account for the prominent systemic inflammatory symptoms observed in several patients [12]. Despite its key role in LVV pathogenesis and the efficacy of IL-6-targeted therapies, no study has verified the diagnostic value of serum IL-6 for GCA and TA. Indeed, although IL-6 measurement is becoming increasingly available in many laboratories, it is likely to have similar limitations to the analysis of ESR and CRP despite much higher costs. Considering the insufficient evidence to support its use, IL-6 analysis is seldom requested in current clinical practice. Serum IL-18 is raised both in TA and GCA [12, 35‐37]. IL-18 is a pro-inflammatory cytokine believed to be produced predominantly by vascular antigen-presenting cells which activate Th1 and Th17 cells [12]. However, an important overlap in IL-18 levels was observed between LVV patients and controls, with poor discrimination of LVVs from other inflammatory conditions [12, 35‐37].

Pentraxin-3 (PTX-3) is an acute phase protein synthetised locally at sites of inflammation, where it is believed to exert important actions in the opsonisation of self and foreign antigens and to contribute to structural and functional activity of the extracellular matrix. Plasma levels of PTX-3 were elevated in TA and GCA [38, 39], as well as in other vascular conditions such as atherosclerosis, acute myocardial infarction and following acute vascular damage during stenting procedures [40‐43].

Considering that the pathogenesis of LVV involves multiple branches of the immune system (antigen-presenting cells, T and B cells), research has focused on autoantibodies. However, specific autoantibodies have never been reported in the LVVs and the role of B cells in these conditions remains elusive. However, several autoantibodies have been identified, including anti-endothelial (AECA), antiphospholipid and anti-ferritin antibodies. A specific pathogenic role for these autoantibodies remains to be demonstrated, since all conditions characterised by tissue injury may elicit autoantibody production. AECA are widely reported in the vasculitides including LVV. Although AECA have been associated with LVV disease pathogenesis, their precise role in disease development and progression is not defined. Antigenicity remains of interest and recent studies have identified endothelial protein C receptor and scavenger receptor class B type 1 as AECA antigenic targets in TA [44].

Antiphospholipid antibodies have been observed in 30–80% of patients with GCA [45‐48] and in up to 41% with TA [49]. The potential use of these antibodies as activity markers is discussed below. Anti-ferritin antibodies have been identified in 92% of untreated active GCA patients (with progressively lower prevalence in patients with flares during therapy and in inactive GCA) [50], and also in 62% of TA patients [51]. These antibodies were found to be less abundant in other diseases and healthy controls. Further studies are required to confirm these results and to define any potential pathogenic role and clinical utility [52].

An alternative strategy has been adopted in studies evaluating multiple potential biomarkers in parallel. In TA, screening based on a 440-cytokine protein array [53] identified higher plasma levels of tissue inhibitor of metalloproteinases (TIMP)-1, P-cadherin and MMP-9 and lower levels of WNT inhibitory factor (WIF)-1 in the plasma of TA patients as compared to healthy controls. A validation in two independent cohorts confirmed higher plasma levels of TIMP-1 and P-cadherin as compared to controls. P-cadherin is a cell adhesion molecule involved in multiple processes including cell migration and growth, while TIMP-1 is a metalloproteinase inhibitor with a proposed role in fibrotic conditions and it may be speculated that this protein contributes to vascular remodelling and arterial wall fibrosis. The involvement of metalloproteinases in vascular remodelling is also highlighted by findings from another study that found higher levels of MMP-9 in peripheral blood in GCA [54]. Increased MMP-9 mRNA was detected in the lamina media of the inflamed arteries, a finding consistent with the pathogenic model of local synthesis of matrix-remodelling factors in LVVs.

Recent studies have evaluated blood biomarkers that might help to identify occult GCA in the setting of PMR, by comparing patients with or without LV-GCA in FDG-PET. One study found that ESR and the concentrations of both the soluble tyrosine kinase receptor Tie-2 and angiopoietin-2, markers of angiogenesis, were significantly elevated in patients with PMR and LV-GCA as compared to those with PMR only [55••]. Angiopoietin-2 outperformed the discriminative potential of both CRP and ESR. Another study indicated that low levels of MMP-3 might indicate the presence of underlying GCA in patients with clinical PMR [56].

In addition to acute phase reactants included in the routine biochemistry panel, other inflammatory molecules have been evaluated during the follow-up of LVVs, including adhesion molecules, which are likely relevant in leukocyte recruitment to the arterial wall. Results are not always concordant between different studies and require further confirmation in independent analyses. Among inflammatory cytokines, higher levels of serum IL-6 and IL-18 have been reported in active TA patients as compared to inactive subjects [35, 36]. Serum concentrations of IL-6 are significantly more sensitive than ESR for active disease, in either untreated or treated GCA patients [82]. Studies comparing the sensitivity of IL-6 over that of CRP in detecting disease activity are lacking, probably due to the very high correlation of these two biomarkers. However, anti-IL-6 receptor therapies directly attenuate the levels of downstream molecules including CRP. In this specific setting, measuring IL-6 rather than CRP might be rationale, even though IL-6 levels usually rise after the first infusions of IL-6 receptor antagonists [83, 84]. A recent paper showed that persistently high levels of IL-6 despite treatment with tocilizumab may predict future relapses [85].

A preliminary small study on 29 patients with TA observed higher plasma levels of the complement protein iC3b (inactivated C3b) and lower levels of the soluble form of the platelet-endothelium adhesion molecule (PECAM-1, also known as CD31) and of iC5b-9 [86].

Rather than a marker of systemic inflammation, plasma PTX-3 may reflect vascular inflammation [68], as its levels in TA did not reflect NIH activity criteria nor ITAS but were rather associated with imaging findings including arterial wall enhancement at MRI and FDG uptake at PET [38, 87].

Goel at al. [88] evaluated the cytokine profile in 32 patients with TA and found that only interferon (IFN)-γ was associated with disease activity, with higher serum levels in active patients. However, longitudinal profiling of patients revealed that treatment efficacy correlated with lower levels of pro-inflammatory cytokines (IFN-γ, IL-6 and IL-23) and higher levels of anti-inflammatory factors such as IL-10 and transforming growth factor (TGF)-β.

Plasma levels of tumour necrosis factor alpha (TNF-α), IL-6, IL-12p40 and intercellular adhesion molecule (ICAM)-1 are higher during relapses of GCA [89, 90, 91•]. Further research identified increased levels of B cell–attracting chemokine-1/CXC motif ligand 13 (BCA1/CXCL13) and soluble IL-2 receptor α (sIL-2Rα) and lower levels of IFN-γ-induced protein 10/CXC motif chemokine 10 (IP10/CXCL10) in active as compared to inactive GCA [92].

Serum OPN is another biomarker whose levels have been correlated with GCA activity [71]. During remission, serum OPN levels were similar to those of healthy controls and did not differ between subjects with GC-driven and tocilizumab-driven remission. Considering that OPN synthesis is partially independent from IL-6, the authors concluded that it might be a promising biomarker to monitor patients on IL-6 antagonists.

Considering the importance of cell-mediated immunity, it is not surprising that studies have evaluated the correlation of disease activity and treatment effect on the immunophenotype of circulating lymphocytes. This research has proposed a role for Th1 and Th17 cells, although with relevant differences between TA and GCA. In active TA, increased circulating Th1 and Th17 cells were observed, together with raised levels of IL-2, IFN-γ, IL-17A, monocyte chemotactic protein (MCP)-1 and macrophage inflammatory protein (MIP)-1β [93]. Patients prescribed GC therapy showed lower Th1 cytokines such as IL-2 and IFN-γ but not Th17 cytokines. In GCA, treatment with GC therapy suppressed circulating Th17 but not Th1 cells and this paralleled a reduction in the concentration of cytokines promoting Th17 differentiation, such as IL-1β, IL-6 and IL-23, while levels of the Th1 cells producing IL-12 remained unaffected [94]. The authors concluded that Th17 responses were associated with acute manifestation of GCA while persistent Th1 responses might be responsible for relapses during GC tapering. A third, more recent, study assessed the immune phenotype of peripheral leukocytes (T cells, B cells, natural killer cells, dendritic cells, monocytes and granulocytes), observing that disease relapse was associated in TA and GCA with a higher number of circulating Th1, Th17 and T follicular helper cells. In TA, increased numbers of CD8+ T cells in relapsing patients were additionally observed [95]. Th1, Th17 and T follicular helper cells, but not CD8+ T cells, decreased after treatment with biologic agents in GCA and TA. Further studies are required to confirm the different responses of cellular subsets to GC observed in TA and GCA and to define their relevance for pathogenesis and biomarker identification.

The use autoantibodies as markers of LVV activity and relapse remains to be clarified. One study identified high titres of IgM anti-endothelial cell antibodies in active TA [96], although it is not clear if these represent pathogenic factors or consequences of arterial injury. Anticardiolipin antibodies have been correlated with fluctuation in disease activity, being present in 74% of patients with GCA flares and in 2% of patients with clinical remission on appropriate treatment [45]. Another study [50] focused on anti-ferritin antibodies, which were observed with highest prevalence in active untreated GCA/PMR (92%), intermediate prevalence in GCA/PMR flares during follow-up (69%) and with lowest prevalence during clinical remission (18%). Ferritin is released by injured cells and the pathogenic relevance of these promising observations is unclear. Beyond autoantibodies, studies regarding cytokine promoting B cell responses such as B cell survival factors activation factor (BAFF) and A proliferation–inducing ligand (APRIL) gave discordant results in TA and GCA [97‐99], and further research is required in this field.

The YKL-40 protein, also known as chitinase-3-like protein 1 (CHI3L1), is a heparin- and chitin-binding lectin that is released by neutrophils, macrophage and vascular smooth muscle cells upon activation. YKL-40 is involved in inflammation, extracellular tissue remodelling and fibrosis. In addition to PTX-3, YKL-40 is believed to be involved in inflammation and remodelling of the extracellular matrix after tissue injury and potentially represents a biomarker of arterial inflammation and remodelling. Serum levels of YKL-40 and IL-6, IL-8, IFN-γ, MMP-2, MMP-9, PTX-3 and OPN have been studied in 40 TA patients [100•]. Although all these parameters were increased in active versus inactive disease, none of them had adequate discriminatory capacity to distinguish between the two groups. A multivariate logistic regression model including ESR, CRP, IL-6, PTX-3 and MMP-9 improved discriminative accuracy between active and inactive TA according to NIH criteria by adding YKL-40 (sensitivity: 85.1%; specificity: 94.3%). However, a limitation of YLK-40 is represented by in vitro reports that synthesis is sensitive to GCs [101]. In contrast, a recent study found that serum levels of YLK-40 in GCA patients were persistently high despite GC treatment [72]. Interestingly, YKL-40 levels were clearly decreased in treatment-free remission. In the same study, strong expression of YKL-40 in temporal artery biopsies (TABs) was observed, mainly at the intima-media border region. This observation led to the hypothesis that YKL-40 is produced predominantly in fully developed GCA with transmural inflammation, and hence may require a longer duration of therapy to induce remission. It is now important to confirm this hypothesis and to verify whether YKL-40 levels may guide GC tapering and reflect arterial damage such as aneurysms or dissection.

The S100 proteins, a family of calcium-binding proteins with pro-inflammatory properties, are expressed and released at local sites of inflammation [102]. Accordingly, they are proposed to play a role in the pathogenesis of multiple rheumatic conditions, thus representing appealing biomarkers of disease activity and course.

The two best-known members of this family are S100A8 and S100A9, which form non-covalent heterodimers, known as calprotectin. This complex, which can act as a damage-associated molecular pattern, is released by monocytes and neutrophils during transendothelial migration [103]. The calprotectin complex is abundant in arteritic lesions and its plasma levels were shown to reflect activity in GCA: despite not being able to outperform traditional biomarkers, the addition of calprotectin and S100A12 levels to a predictive model composed of ESR and CRP improved the ability of the Birmingham Activity Score (BVAS) to discriminate between active and inactive TA and GCA [104•]. However, discordance exists concerning the results of studies in TA. One study found that ESR, CRP, calprotectin and S100A12 levels were similar in patients with active and inactive disease [104•], while another observed higher calprotectin levels in active patients and in those that experienced progressive arterial remodelling during follow-up [105]. Importantly, calprotectin levels did not correlate with IL-6 and the acute phase response [72], thus deserving further research as a marker of arterial inflammation or remodelling, particularly in patients on IL-6 pathway inhibitors.

Vasa vasorum neoangiogenesis represents a crucial step in LVVs [12, 13]. VEGF serum levels have been shown to reflect disease activity and arterial inflammation at imaging in GCA [87] but not in TA [106]. There is evidence suggesting that Ang-1/Ang-2 might be involved in GCA, as higher levels of Ang-1 have been reported in active GCA, while increased Ang-2 levels during remission have been observed in patients with imminent disease relapse [72]. Neoangiogenesis requires a tight modulation of pro- and anti-angiogenic factors which include, among others, full-length chromogranin A (CgA) and its fragment vasostatin-1. Interestingly, the anti-angiogenic potential of these two CgA peptides was found to be reduced in TA patients experiencing progressive arterial remodelling [107].

MMPs are proteases that degrade components of the extracellular matrix, thus playing a role in several key biological processes including leukocyte extravasation, cell migration and extracellular matrix turnover and remodelling. MMP-9 participates in migration of smooth muscle cells (SMCs) from media to intima, a process leading to intimal hyperplasia [108]. Higher blood levels of MMP-2 and MMP-9 were reported in active vs inactive TA [100], while increased levels of the MMP inhibitor TIMP-1 were observed in active versus inactive GCA [92].

Circulating endothelial precursor cells are markers of endothelial cell damage and repair/angiogenesis. A small number of studies address the number of endothelial precursor cells in the peripheral blood in TA, using different gating strategies to identify these cells (CD309-VEGFR2/CD133/CD34 [106, 109] or CD146/CD31/CD34+, CD45− cells [110]). Their number in peripheral blood has been reported to correlate with disease activity in TA [109, 110], although a more recent study did not confirm these results [106] and the exact mechanism underlying this occurrence is scarcely understood.