Abstract
Objective. Systemic inflammationˆ is assessed through measurement of acute-phase reactants such as C-reactive protein (CRP) and erythrocyte sedimentation rate (ESR). With few exceptions, most randomized controlled trials (RCT) have assessed acute-phase reactants (CRP and ESR) as part of the American College of Rheumatology (ACR) 20 response criteria. As part of the Group for Research and Assessment of Psoriasis and Psoriatic Arthritis (GRAPPA)–Outcome Measures in Rheumatology (OMERACT) working group, we performed a systematic review of the literature to assess the performance of inflammatory biomarkers in psoriatic arthritis (PsA).
Methods. A systematic search of PubMed and Embase was performed. The search included peer-reviewed articles and scientific meeting abstracts about RCT and longitudinal observational studies that assessed systemic inflammation using acute-phase reactants in PsA. Studies were assessed following the components of the OMERACT filter including construct validity, responsiveness, and predictive validity.
Results. There were 2764 articles retrieved, and 71 articles were included for this systematic review. Twenty-eight articles reported CRP and/or ESR separately, and the remaining articles reported CRP and/or ESR as part of the ACR response criteria. Studies assessing OMERACT responsiveness provided conflicting reports. Inflammatory biomarkers had construct validity for more active disease. Evidence suggests that an elevation of ESR predicts cardiovascular outcomes.
Conclusion. Data regarding assessment of systemic inflammation using acute-phase reactants (CRP and ESR) are limited. There is only weak evidence to support normalization of these biomarkers in predicting good clinical outcomes/remission criteria. The predictive value for cardiovascular outcomes was generally good. Further studies to assess systemic inflammation in PsA using acute-phase reactants and other laboratory biomarkers are needed.
Psoriatic arthritis (PsA) is a chronic immune-mediated inflammatory arthritis1 that occurs in 14–30% of patients with psoriasis and can lead to significant joint damage and disability2,3,4,5. PsA is a multifaceted, heterogeneous disease that manifests with the following clinical domains: peripheral arthritis, axial disease, enthesitis, dactylitis, and skin and nail disease. Measuring disease activity accurately can be difficult in PsA, because of the heterogeneity of the clinical features of the disease. Longitudinal studies have consistently demonstrated increased cardiovascular (CV) risk and related metabolic abnormalities in patients with psoriatic disease6,7,8,9. Several studies found an association between the extent of inflammation (both joint inflammation and acute-phase reactants) and poor disease outcomes, including the development of joint damage, cardiometabolic outcomes, and mortality10–17.
An updated PsA Core Domain Set was endorsed at Outcome Measures in Rheumatology (OMERACT) 2016 and includes musculoskeletal disease activity, skin disease activity, fatigue, pain, patient’s global assessment (PtGA), physical function, health-related quality of life, and systemic inflammation18.
Systemic inflammation may be assessed through measurement of acute-phase reactants such as C-reactive protein (CRP) and erythrocyte sedimentation rate (ESR). With few exceptions, most randomized controlled trials (RCT) have assessed acute-phase reactants (CRP/ESR) as part of the American College of Rheumatology (ACR) 20 response criteria primary efficacy endpoints.
The purpose of this paper is to review and summarize the data available for assessment of systemic inflammation using CRP and ESR in PsA RCT, as well as longitudinal observational studies (LOS) and cross-sectional studies. This systematic review will inform selection and/or development of an outcome measurement instrument for the assessment of systemic inflammation in PsA clinical trials.
MATERIALS AND METHODS
The original search strategy, up to May 2015, has been previously described to identify PsA RCT and observational studies19. The search included 2 databases, PubMed and Embase, and had 2 parts: the term for the disease concept “Psoriatic Arthritis” combined with operator “AND” with the validated Cochrane RCT sensitivity filter for each database20,21. The literature search was updated from June 2015 to September 2016 using the same methodology; hand searches representing recent interleukin (IL)-17 and IL-6 inhibitors were included from September 2016. We excluded pediatric studies (children age 0–18 yrs) and used the following limits: human studies, English language (Figure 1).
The screening of all titles and abstracts for potential inclusion was performed by 2 investigators (ME and LE). Selected publications were retrieved in full, and 2 reviewers (ME and LE) independently assessed them for eligibility. To be included in the systematic review, original studies in the English language needed to fulfill the following criteria: (1) human studies; (2) adult patients with PsA (> 18 yrs); and (3) RCT, LOS, or cross-sectional studies. Exclusion criteria included animal studies, pediatrics (age 0–18), studies other than RCT, observational studies, and review articles.
Two authors (ME and YYL) independently extracted the data according to a standardized form. Discrepancies were resolved by consensus and involvement of a third author if needed (LE). For each study included, we recorded the following data: population, study design, study drug (if applicable), duration of study, sex, age (weighted mean), PsA disease duration (weighted mean), acute-phase reactant assessed separately, type of acute-phase reactant (ESR and/or CRP), and the extent of change of the acute-phase reactant.
The following components of the OMERACT filter22 were independently assessed for each study: responsiveness, construct validity, and predictive validity. Responsiveness was evaluated by the ability of the tool to demonstrate change in response to an intervention (e.g., study drug). Construct validity was achieved when the level of acute-phase reactants was concurrently compared with a theoretical concept of inflammation (e.g., clinical disease activity/disease state). Predictive validity was considered when acute-phase reactant predicted a theoretical concept related to systemic inflammation (e.g., clinical outcomes, radiographic joint damage, CV outcomes). We illustrated the process as recommended in the Preferred Reporting Items for the Systematic reviews and Meta-Analyses (PRISMA) statement23.
RESULTS
An initial literature search for PsA domains and instruments in PsA clinical trials19 retrieved 2079 entries, of which 60 full-text articles met inclusion criteria. We updated the search in PubMed and Embase and retrieved 685 additional entries. Eleven full-text articles were included, bringing the total number of full-text articles to 71, representing RCT, LOS, and cross-sectional studies (Figure 2).
Following review of full articles and supplementary data, only 28 articles were found to have reported ESR and/or CRP separately. The rest assessed CRP and/or ESR as part of ACR response without separate reporting of their results. The 28 articles represented 12 RCT [tumor necrosis factor inhibitor (2), methotrexate (1), IL-6 inhibitor (1), phosphodiesterase-4 inhibitor (3), IL-17 inhibitor (3), bisphosphonate (1), a panel of 57 protein biomarkers (1)] and 16 observational studies. The total number of patients with PsA included in the studies reviewed was 4761; 40% were female, their mean age was 48.2 years, and mean PsA duration was 8.2 years. The total number of patients in each study varied from 18 to 596.
Eighteen articles assessed OMERACT responsiveness, the ability of CRP/ESR to demonstrate change in response to treatment (11 RCT, 7 observational). Nine studies24–32 have shown reduction of acute-phase reactants with different treatments. These studies have shown significant reduction of CRP/high-sensitivity CRP (hsCRP), and to a lesser extent, of ESR (Table 1).
Nine other studies33–41 did report a small change in CRP/ESR; however, they either did not report a p value or the p value was not significant (Table 1).
Ten studies (Table 2) have assessed OMERACT construct validity (level of biomarker correlated with disease state/disease activity measure) representing 3 RCT, 5 LOS, and 1 cross-sectional study. Acute-phase reactants were associated with clinical measures of disease activity. Shen, et al42 demonstrated that patients with PsA had higher CRP/ESR compared to controls, while Sterry, et al39 demonstrated that PsA patients with enthesitis had higher CRP than those without enthesitis.
The association between levels of inflammation at baseline and response to treatment has been assessed in 3 studies with conflicting results. Schett, et al32 reported that elevated CRP (CRP > 8 mg/l) predicted ACR20 response to apremilast. In contrast, Saad, et al43 reported that patients with high ESR (ESR > 28 mm/h) or CRP (CRP > 20 mg/l) are less likely to achieve European League Against Rheumatism response (OR 0.54, 95% CI 0.30–0.96) and remission (OR 0.54, 95% CI 0.31–0.97) at 6 months. Further, Coates, et al44 demonstrated that low ESR predicted minimal disease activity (MDA) in multivariate model with RR 0.62 (p < 0.02).
One other study assessed other potential inflammatory biomarkers. Ademowo, et al45, using a proteomic approach in patients with PsA, identified a panel of 57 protein biomarkers in synovial tissue samples of patients with PsA prior to therapy that predicted treatment response to biologic therapy with an area under the curve of 0.76. This novel biomarker panel was developed to be measured at baseline to predict patient response to biologics (adalimumab and abatacept). Some of these proteins on the candidate biomarker panel [fibrinogen, type 11 collagen, serum amyloid A, haptoglobin, and the S100 family proteins (A8, A9, A11, and A12)] have been found previously to play significant roles in inflammation45. This study suggests that novel biomarkers may be used in the future to predict PsA response to treatment.
Two LOS30,46 and a cross-sectional study47 reported that ESR and CRP do not predict MDA, 28-joint Disease Activity Score remission, PtGA, physician’s global assessment, and Assessment of Spondyloarthritis international Society partial remission. No study has evaluated the value of ESR/CRP in predicting radiographic joint damage. Castaneda, et al48 assessed CV morbidity and associated risk factors for CV disease in Spanish patients with chronic inflammatory rheumatic diseases and unexposed individuals attending rheumatology clinics in a cross-sectional study. They did not establish a statistically significant association between inflammatory markers and CV risk in the PsA group (721 out of 2234 patients); however, most patients generally had mild disease activity at recruitment, were followed periodically by rheumatologists, and over 40% were treated with biologics. Most of the patients had low levels of disability and the acute-phase reactants were within normal range.
Regarding OMERACT predictive validity (predictive validity was considered when acute-phase reactant predicted radiographic damage or clinical outcome), 4 LOS evaluated the predictive validity of ESR and CRP (Table 3). Shen, et al42 demonstrated that cumulative ESR (defined as cumulative averages of ESR over time) was associated with increased arterial stiffness, assessed by pulse wave velocity, independently of traditional CV risk factors (OR 9.455, 95% CI 1.939–46.093; p = 0.005). Eder, et al10,15 showed in 2 different studies that elevated levels of ESR were associated with increased CV risk. In the age- and sex-adjusted models, increased adjusted mean ESR (OR 1.41, 95% CI 1.09–1.82) was associated with a more severe degree of carotid atherosclerotic plaques in patients with PsA. In addition, in a longitudinal cohort study of 1091 patients with PsA, Eder, et al10 demonstrated that ESR was an independent predictor of clinical CV events in women after controlling for CV risk factors (RR 1.83, p = 0.02). Geijer, et al49 have reported in univariate analysis that baseline ESR correlated with baseline radiographic joint damage as assessed by Wassenberg scores (p = 0.027).
DISCUSSION
Systemic inflammation characterizes psoriatic disease and the assessment of the extent of inflammation using laboratory biomarkers has a potential for improved evaluation of disease activity in patients with PsA. To our knowledge, this is the first study to assess systemic inflammation in PsA using the OMERACT filter. This systematic literature review represents a critical examination of the published data regarding the state of validation of the most commonly used inflammatory biomarkers in PsA.
In this systematic review, all studies have assessed an acute-phase reactant: CRP, hsCRP, or ESR. However, only 40% reported the results of these biomarkers separately, even though all RCT have assessed these biomarkers as part of the primary endpoint, the ACR20 response. Data available for evidence synthesis were therefore limited.
Studies assessing OMERACT responsiveness provided conflicting reports. Nine of these studies24–32 demonstrated responsiveness with treatment, while 9 other studies33–41 did not report significant results. Clearly there is a need for more homogeneous studies to assess this component further with larger RCT or novel biomarkers.
Inflammatory biomarkers had good construct validity, with CRP/ESR being higher in patients with PsA, and patients with enthesitis versus those without enthesitis39,42. While the association between ESR/CRP and clinical disease outcomes was conflicting, there was more evidence to suggest that an elevation of ESR predicts CV outcomes. CRP, which was associated with increased CV risk in chronic inflammatory arthritis50, was not associated with CV risks in PsA based on a single cross-sectional study. This finding may be explained by the relatively low levels of chronic inflammation found in PsA using only traditional methods (CRP, ESR).
One of the limitations of our review is that not all papers reported the results of CRP/ESR separately and therefore only a few studies inform construct and predictive validity. Moreover, some of the studies were observational, raising the issue of selection bias and the uncertainty of the comparability of groups.
Inflammatory biomarkers have good construct validity based on their association with active disease. Inflammatory biomarkers have been incorporated into composite measures to assess clinical response in clinical trials. However, when used alone they provide inconsistent evidence of responsiveness in clinical trials. There is only weak evidence to support these biomarkers in predicting good clinical outcomes/remission criteria. Derived from 4 studies, the predictive value for CV outcomes was generally good.
Further large RCT to assess systemic inflammation in PsA using acute-phase reactants and high-sensitivity assays are needed. We recommend that all future RCT should assess and report the results of acute-phase reactants/systemic inflammation separately. Additional studies are also needed to identify novel laboratory biomarkers for the assessment of systemic inflammation in PsA.
- Accepted for publication August 3, 2018.