Multibiomarker disease activity score: an objective tool for monitoring rheumatoid arthritis? A systematic review and meta-analysis

Abstract

Objectives

The multibiomarker disease activity (MBDA) score is an objective tool for monitoring disease activity in RA. Here we report a systematic review and meta-analysis of the clinical value of the MBDA score in RA.

Methods

We performed a systematic literature search in five medical databases—MEDLINE (via PubMed), Cochrane Library (CENTRAL), Embase, Scopus and Web of Science—from inception to 13 October 2021. Original articles reporting on the performance of the MBDA score’s correlation with conventional disease activity measures or the predictive and discriminative values of the MBDA score for radiographic progression, therapy response, remission and relapse were included.

Results

Our systematic search provided a total of 1190 records. After selection and citation searches, we identified 32 eligible studies. We recorded moderate correlations between MBDA score and conventional disease activity measures at baseline [correlation (COR) 0.45 (CI 0.28, 0.59), I² = 71.0% for the 28-joint DAS with CRP (DAS28-CRP) and COR 0.55 (CI 0.19, 0.78), I² = 0.0% for DAS28 with ESR] and at follow-up [COR 0.44 (CI 0.28, 0.57, I² = 70.0% for DAS28-CRP) and found that the odds of radiographic progression were significantly higher for patients with a high baseline MBDA score (>44) than for patients with a low baseline MBDA score (<30) [OR 1.03 (CI 1.02–1.05), I² = 10.0%].

Conclusion

The MBDA score might be used as an objective disease activity marker. In addition, it is also a reliable prognostic marker of radiographic progression.

Introduction

RA is a systemic autoimmune disease affecting ≈0.5–1% of the population [1]. According to the EULAR recommendations, the aim of the therapy in RA is to achieve remission, or at least low disease activity [2]. Early treatment with DMARDs and a treat-to-target treatment strategy are recommended by current guidelines and are considered to be the optimal way to prevent long-term functional decline by minimizing cartilage and bone damage [3–5].

Given that the treat-to-target therapeutic approach requires close monitoring of disease activity, the need for reliable, objective disease activity measures (DAMs) is undeniable. The currently available, widely used options for monitoring disease activity and progression are either subjective or non-specific: the 28-joint DAS (DAS28), Simplified Disease Activity Index (SDAI), and Clinical Disease Activity Index (CDAI) all include subjective assessments of disease activity by the patient and/or the provider [6–8]. Although non-specific inflammatory markers such as CRP and ESR are used to calculate the DAS28 and SDAI, the incorporation of a scoring system based on the combination of inflammatory markers and additional biomarkers could further objectify the measurement of disease activity. Structural damage, a major factor defining the course of the disease, can be assessed by radiography and quantified with the Sharp–van der Heijde (SvdH) scoring system [9]. There are several known risk factors for radiographic progression, including high disease activity monitored by non-specific inflammatory markers such as CRP, RF and ACPA seropositivity [10]. However, RF and ACPA are not suitable for monitoring disease activity [11].

The multibiomarker disease activity (MBDA) score system is an algorithm based on the serum level of 12 biomarkers [IL-6, TNF receptor type 1 (TNFR1), vascular cell adhesion molecule 1 (VCAM-1), epidermal growth factor (EGF), vascular EGF A (VEGF-A), YKL-40, matrix metalloproteinase-1 (MMP-1), MMP-3, CRP, serum amyloid A (SAA), leptin and resistin], resulting in a scale from 0 to 100 [12]. The MBDA score presents an objective disease monitoring system and thus may contribute to personalized therapeutic plans conforming to modern medical views. In addition to monitoring disease activity, the MBDA score may also predict radiographic progression [13–16].

Several studies have evaluated the utility of the MBDA score and a meta-analysis has been conducted on the correlation of the MBDA score with conventional DAMs; however, the predictive and discriminative values of the MBDA score has yet to be analysed in a comprehensive manner [17]. Here we report a systematic review and meta-analysis of the clinical value and utility of the MBDA score for monitoring RA by determining the correlation of the MBDA score with conventional DAMs and the predictive and the discriminative values of the MBDA score for radiographic progression, therapy response, remission and relapse.

Materials and methods

Our systematic review and meta-analysis are reported according to the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) 2020 Statement [18]. The recommendations of the Cochrane Prognosis Methods Group [19] were followed and the review protocol was registered on PROSPERO.

We performed a systematic literature search of five medical databases—MEDLINE (via PubMed), Cochrane Library (CENTRAL), Embase, Scopus and Web of Science—from inception to 13 October 2021. Original articles reporting on the performance of the unadjusted MBDA score’s correlation with conventional DAMs or the predictive and discriminative values of the MBDA score for radiographic progression, therapy response, remission and relapse were included in the systematic review and meta-analysis. Single case reports were excluded.

RA was defined by the ACR 1987 [20] and ACR/EULAR 2010 [21] classification criteria. Radiographic progression was measured by the change in the SvdH score per time unit, therapy response was defined by the EULAR criteria for therapy response and remission and relapse were defined by the different cut-off values of conventional DAMs.

Study selection and data extraction were carried out by two independent reviewers and disagreements were resolved by a third reviewer. The quality assessment of the outcomes was carried out separately by two reviewers using the Quality In Prognosis Studies (QUIPS) tool for assessing the risk of bias [22].

Further details regarding the search and selection strategy, data extraction, data synthesis and analysis are detailed in the supplementary methods available at Rheumatology online.

Results

Search and selection and characteristics of the included studies

Our systematic search provided 1190 records; after duplicate removal we screened 708 duplicate-free records. Thirty eligible studies [13–16, 23–48] were identified after title, abstract and full-text selection and two additional studies [49, 50] during citation search. Of these studies, we included 24 in the quantitative analysis [13–15, 23–25, 27–29, 31, 32, 34–38, 40, 41, 43, 44, 46, 47, 49, 50] and 8 in the qualitative analysis [16, 26, 30, 33, 39, 42, 45, 48]. The summary of the selection process is shown in Fig. 1. We conducted a meta-analysis assessing the correlation of MBDA scores with conventional DAMs and the predictive value of the MBDA score for radiographic progression. Studies that could not be included in the meta-analysis and reports of other outcomes are detailed in the systematic review.

The characteristics of the identified studies for the systematic review and meta-analysis and the patient characteristics of included studies are detailed in Table 1 and Supplementary Table S1, available at Rheumatology online.

MBDA score for the assessment of disease activity

Studies assessing the utility of the MBDA score for disease activity monitoring calculated the correlation of MBDA scores with conventional DAMs. Studies using Pearson’s correlations could not be included in the meta-analysis due to a lack of statistical power, but are displayed in forest plots for visualization (see Supplementary Figs S1–S3, available at Rheumatology online). The results of studies using Spearman’s correlation are detailed below.

Six study groups in five publications [27, 29, 36, 46, 47] with a total of 667 subjects showed a moderate correlation between baseline MBDA score and baseline DAS28-CRP [correlation (COR) 0.45 (CI 0.28, 0.59), I² = 71.0%] (see Fig. 2A). Excluding conference abstracts from the analysis, similar results were observed; four publications [27, 36, 46, 47] with a total of 324 subjects demonstrated a moderate correlation between baseline MBDA score and baseline DAS28-CRP [COR 0.46 (CI 0.10, 0.72), I² = 81.0%].

Assessing the correlations of baseline MBDA scores with baseline DAS28-ESR, a moderate correlation was found based on the results of two publications with a total of 127 subjects [COR 0.55 (CI 0.19, 0.78), I² = 0.0%] (see Fig. 2A).

Further metrics associated with disease activity [CRP, ESR, 28-joint swollen joint count,28-joint tender joint count, patient global assessment (PtGA), CDAI, power Doppler ultrasound (PDUS)] showed low and moderate correlations and are detailed in Supplementary Fig. S4, available at Rheumatology online.

Six study groups from four publications [29, 36, 46, 47] with a total of 287 subjects revealed a moderate correlation between follow-up MBDA score and follow-up DAS28-CRP [COR 0.44 (CI 0.28, 0.57), I² = 70.0%] (see Fig. 2B). After the exclusion of conference abstracts from the analysis, three articles [36, 46, 47] with a total of 137 subjects showed a moderate correlation between baseline MBDA score and baseline DAS28-CRP [COR 0.38 (CI −0.02, 0.68), I² = 18.0%].

The only study investigating the correlations of follow-up MBDA scores with follow-up DAS28-ESR found a moderate correlation [COR 0.49 (CI 0.22, 0.69)] between MBDA score and DAS28-ESR (Fig. 2B) [47].

Other parameters associated with disease activity (ESR, SJC28, TJC28, PtGA, PDUS) showed low–moderate correlations and are detailed in Supplementary Fig. S5, available at Rheumatology online.

Ten study groups from six articles [14, 28, 29, 36, 47, 50] with a total of 698 subjects demonstrated a moderate correlation between the change in MBDA score and the change in DAS28-CRP [COR 0.40 (CI 0.32, 0.48), I² = 19.0%]. Seven study groups from six articles [25, 35, 38, 47, 49, 50] with a total of 543 subjects exhibited a moderate correlation between the change in MBDA score and the change in DAS28-ESR [COR 0.56 (CI 0.51, 0.60), I² = 71.0%] (see Fig. 2C). Excluding conference abstracts from the analysis, similar results were recorded. The change in MBDA score moderately correlates with the change in DAS28-CRP [COR 0.43 (CI 0.25, 0.59), I² = 47.0%] based on the results of six study groups of four publications [14, 36, 47, 50] with a total of 418 subjects, and with DAS28-ESR [COR 0.52 (CI 0.43, 0.60), I² = 0.0%] based on the results of four publications [35, 47, 49, 50] with a total of 298 subjects.

Further parameters linked to disease activity (CRP, CDAI, SDAI, HAQ) showed low–moderate correlations and are shown in Supplementary Fig. S6, available at Rheumatology online.

The results of the subgroup analysis based on the length of the follow-up showed similar results and are displayed in Supplementary Figs S7 and S8, available at Rheumatology online.

MBDA score for the assessment of radiographic progression

Three study groups of three articles with a total of 22 subjects showed a low correlation between baseline MBDA score and baseline SvdH score [COR 0.13 (CI −0.25–0.47), I² = 79.0%] and five study groups of four articles with a total of 307 subjects demonstrated a low correlation between the change in MBDA score and the change in SvdH score [COR 0.08 (CI −0.06–0.21), I² = 79.0%] as well (see Fig. 3).

When evaluating the predictive value of the MBDA score for radiographic progression, three studies [13–15] with a total of 481 subjects showed that the odds of radiographic progression are significantly higher for patients with a high baseline MBDA score (>44) than for patients with a low baseline MBDA score (<30) [OR 1.03 (CI 1.02, 1.05), I² = 10.0%] (see Fig. 4). In contrast, the odds of progression for patients with a high baseline DAS28-CRP were not significantly higher than for patients with a low baseline DAS28-CRP [OR 1.12 (CI 0.91–1.37), I² = 0.0%] (see Supplementary Fig. S9, available at Rheumatology online). The characteristics of the studies evaluating the predictive value of the MBDA score and DAS28-CRP for radiographic progression are detailed in Table 2.

Five additional studies evaluating the utility of the MBDA score for the assessment of radiographic progression could not be included in our quantitative synthesis [16, 27, 34, 36, 39]. Markusse et al. [16] found that higher MBDA scores at baseline were associated with an increased risk of radiographic progression in the subsequent year, therefore the MBDA score can be considered an independent predictor for radiographic progression. The discriminative value of the MBDA score was also assessed and the results showed that the MBDA score discriminated more between radiographic progression and no radiographic progression than the DAS at baseline and 1 year. Hirata et al. [34] reported that patients with moderate or high MBDA scores had a greater risk of radiographic progression than patients with low or moderate MBDA scores. Li et al. [39] also found that radiographic progression was not frequent when MBDA scores were low; univariate and multivariate analyses showed that high MBDA scores were strongly associated with radiographic progression. In a study by Krabbe et al. [36], none of the patients with radiographic progression had low MBDA scores. In contrast, Bouman et al. [27] found no association between baseline MBDA score and radiographic progression.

MBDA score for the assessment of therapy response, remission and relapse

We identified four studies [30, 34, 38, 47] investigating the utility of the MBDA score for the assessment of therapy response, six studies [13, 14, 33, 40, 42, 49] for remission and five studies [24, 26, 27, 45, 48] for relapse. However, these studies were not eligible for quantitative synthesis due to the widely varying outcome measures.

The change of MBDA score from baseline to 6 months was significantly associated with good or moderate EULAR response vs non-response at 6 months by Roodenrijs et al. [47]; however, the baseline MBDA score was not associated with EULAR response vs non-response. Similar results were recorded by Li et al. [38]. Although the baseline MBDA score was not associated with EULAR response at 3 months, changes in MBDA scores differentiated responders from non-responders. Hambardzumyan et al. [30] also reported that the MBDA score was significantly associated with treatment outcomes at 3 months. In the study of Hirata et al. [49], EULAR good responders were found to have significantly greater reductions in the MBDA score from baseline than EULAR moderate responders and EULAR moderate responders had significantly greater reductions than EULAR non-responders.

The MBDA score was found to be an appropriate discriminator of remission/low disease activity and moderate/high disease activity, according to two studies [13, 42]. Ma et al. [42] reported that the baseline MBDA score and the time-integrated MBDA score discriminated between remission and non-remission at 1 year as well. Two studies found no significant association between baseline MBDA score and remission, although, according to Brahe et al. [14, 40], the change in MBDA score was associated with clinical remission. Hirata et al. [33, 49] recorded the association of MBDA remission with clinical remission.

High baseline MBDA scores were associated with significantly greater proportions of patients experiencing relapse based on the results of Ghiti Moghadam et al. [48] and significantly higher MBDA scores were recorded in relapsed patients by Rech et al. [45]. Boeters et al. [26] found that high MBDA scores during DMARD treatment and before treatment reduction were associated with an increased risk of relapses in patients who reduced or stopped DMARD treatments. Bouman et al. [27] reported the borderline positive predictive value of baseline MBDA score for flare of patients with low disease activity at baseline. According to Bechman et al. [24], baseline MBDA scores were not predictive of flare. However, a sensitivity analysis limited to flares with an increase in high disease activity determined by MBDA score (> 44) did show an association between baseline MBDA value and flare risk.

Funnel plots and leave-one-out analysis

No evidence of publication bias was observed in the funnel plots for the correlations of MBDA scores with conventional DAMs (see Supplementary Figs S10–S12, available at Rheumatology online). The results of the leave-one-out analysis are detailed in Supplementary Tables S2–S4, available at Rheumatology online, showing no outlier article.

Risk of bias assessment

The majority of the outcomes of the studies included in the meta-analysis (n = 79) and the systematic review (n = 37) were rated as having a low or moderate risk of bias. The risk of bias was low in 35 outcomes of the studies included in the meta-analysis and 29 outcomes of the studies included in the systematic review, moderate in 32 outcomes of the studies included in the meta-analysis and 5 outcomes of the studies included in the systematic review and high in 12 outcomes of the studies included in the meta-analysis and 3 outcomes of the studies included in the systematic review. Common methodological limitations across studies were attrition rates, study confounding and statistical analysis and reporting. The quality assessment scores for all outcomes are shown in Supplementary Tables S5 and S6, available at Rheumatology online.

Discussion

Since the recommendation for the treatment of RA—the treat-to-target therapeutic approach—requires close monitoring of disease activity, the importance of objective scoring systems is indisputable. By conducting a systematic review and meta-analysis on the utility of the MBDA score to assess disease activity, radiographic progression, remission and relapse, we aim to promote decision making on the applicability of the MBDA score in clinical practice.

When analysing the correlations of MBDA score with conventional DAMs by a random-effects model, moderate correlations were recorded, similar to the meta-analysis conducted by Johnson et al. [17]. DAS28-CRP and DAS28-ESR, which are considered the gold standard DAMs in RA, both showed moderate correlations with MBDA at baseline and follow-up, as well as in the change in DAS28-CRP and DAS28-ESR with the change in MBDA. Other DAMs detailed in the supplement showed weaker correlations with MBDA score, except for CRP. The correlation of the MBDA score with CRP individually was stronger than with DAS28-CRP. As the MBDA score does not contain the results of clinical assessment, its deviation from the conventional DAMs is not surprising. However, the MBDA score was designed to complement, not replace conventional DAMs, therefore its deviation from conventional DAMs can even be advantageous [51].

Since the MBDA score contains markers of cartilage and bone damage, such as MMP-3, in addition to the inflammatory markers implemented in currently used DAMs, such as CRP, it is a realistic possibility that it can outperform conventional DAMs in predicting radiological progression [52]. The results of our meta-analysis suggest that the MBDA score can be an independent predictor of radiological progression, as the odds of radiographic progression were significantly higher for patients with a high baseline MBDA score than for patients with a low baseline MBDA score, while there was no significant difference between low- and high-baseline DAS28-CRP. However, while the cut-off values for high and low MBDA scores were the same in the included studies, different cut-off values were used to define DAS28-CRP subgroups, which may influence these results and highlight the need for further investigations (see Table 2). Furthermore, the SvdH score showed a low correlation with the MBDA score at baseline and at follow-up, which suggests that these data should be interpreted with caution. These results are in line with the results of the studies included in our systematic review and also with the results of the previous meta-analysis by Curtis et al. [53] and the systematic review by Abdelhafiz et al. [54]. The limitation of both our study and the study by Curtis et al. [53] is the lack of included studies investigating the efficacy of DAS28-CRP for predicting radiographic progression independent of the MBDA score, potentially leading to biased results.

Based on the studies included in the systematic review, the change in MBDA score is associated with therapeutic response and seems to discriminate between therapy responders and non-responders [30, 38, 47, 49]. However, baseline MBDA scores were not predictive of therapy response [38, 47]. Similarly, while the change in MBDA score was found to be associated with remission and MBDA score discriminated remission/low disease activity and moderate/high disease activity [13, 14, 33, 42, 49], no significant associations were found between baseline MBDA scores and remission [14, 40]. In contrast, in the case of relapse, the baseline MBDA score was reported to be a predictor, although no clear conclusions can be drawn due to the heterogeneity of study designs and the potential for false positivity due to multiple testing [24, 27, 48].

There are several strengths of our study. We implemented a rigorous methodology to achieve the highest quality of evidence and provide a structured analysis of the outcomes discussed in the literature. We provide a comprehensive summary on the utility of the MBDA score for monitoring RA disease activity and also the predictive and discriminative value of the MBDA score for radiographic progression, therapy response, remission and relapse, presenting the results of quantitative analysis for both the correlation of the MBDA score with conventional DAMs and the predictive value of the MBDA score for radiographic progression.

Our main limitation is the heterogeneity of the populations. A wide range of anti-rheumatic drugs was used in the included studies, with potentially varying effects on the MBDA score: the IL-6 receptor-blocker tocilizumab may increase the serum level of IL-6 by preventing receptor binding, therefore influencing the change in MBDA score via one of the 12 included biomarkers [46]. TNF inhibitors may also influence MBDA score indirectly by decreasing the serum level of TNF-α. Hirata et al. [34] compared anti-TNF-α and anti-TNF-α-receptor drugs and found no significant difference between the two groups; however, further studies are needed to assess the effect of targeted therapies on the serum level of the biomarkers included in the MBDA score and therefore their effect on the change of MBDA score [34]. Furthermore, the different follow-up times used for the assessment of disease activity may also increase the heterogeneity.

By including a higher number of patients and uniformizing the follow-up time for evaluation and the cut-off values of DAS28-CRP for remission, future studies would enable further comprehensive analysis to urge implementation of the MBDA score in daily clinical practice.

Conclusion

The MBDA score can be highly valuable in RA patient care, both for monitoring disease activity and for predicting radiological progression. However, further studies are needed to better assess the utility of the MBDA score and also the potential role of individual biomarkers in disease activity monitoring.

Supplementary material

Supplementary material is available at Rheumatology online.

Data availability

The datasets used in this study can be found in the full-text articles included in the systematic review and meta-analysis. The data underlying this article will be shared upon reasonable request to the corresponding author.

Authors’ contributions

L.V.K. and F.A.M. were responsible for conceptualization, project administration, data curation, visualization and writing the original draft. E.G. was responsible for conceptualization, project administration, data curation and visualization. E.B. was responsible for conceptualization and data curation. F.D. was responsible for conceptualization, data curation and methodology. B.S. was responsible for methodology, formal analysis, validation and visualization. A.A. and E.L. were responsible for data curation and visualization. D.C. and P.H. were responsible for conceptualization, methodology and supervision. A.B. and G.N. were responsible for conceptualization, methodology, supervision and writing the original draft. All authors provided critical conceptual input and approved the final version of the article.

Funding

This work was supported by grants from National Scientific Research Programs (OTKA K-131479), Rheuma Tolerance for Cure (H2020 777357) and the Competitive Central Hungary Operational Programme (VEKOP-2.3.2-16-2016-00002). Project TKP-76-8/PALY-2021 has been implemented with support provided by the Ministry of Innovation and Technology of Hungary from the National Research, Development, and Innovation Fund financed under the TKP2021-EGA/TKP2021-NVA/TKP2021-NKTA funding scheme.

Disclosure statement: The authors have declared no conflicts of interest.

Acknowledgments

No ethical approval was required for this review as all data were already published in peer-reviewed journals. No patients were involved in the design, conduct or interpretation of our study.

References

Author notes