Revision to the musculoskeletal domain of the BILAG-2004 index to incorporate ultrasound findings

Abstract

Objectives

To improve the definitions of inflammatory arthritis within the musculoskeletal (MSK) domain of the BILAG-2004 index by incorporating imaging findings and clinical features predictive of response to treatment.

Methods

The BILAG MSK Subcommittee proposed revisions to the BILAG-2004 index definitions of inflammatory arthritis, based on review of evidence in two recent studies. Data from these studies were pooled and analysed to determine the impact of the proposed changes on the severity grading of inflammatory arthritis.

Results

The revised definition for severe inflammatory arthritis includes definition of ‘basic activities of daily living’. For moderate inflammatory arthritis, it now includes synovitis, defined by either observed joint swelling or MSK US evidence of inflammation in joints and surrounding structures. For mild inflammatory arthritis, the definition now includes reference to symmetrical distribution of affected joints and guidance on how US may help re-classify patients as moderate or no inflammatory arthritis. Data from two recent SLE trials were analysed (219 patients). A total of 119 (54.3%) were graded as having mild inflammatory arthritis (BILAG-2004 Grade C). Of these, 53 (44.5%) had evidence of joint inflammation (synovitis or tenosynovitis) on US. Applying the new definition increased the number of patients classified as moderate inflammatory arthritis from 72 (32.9%) to 125 (57.1%), while patients with normal US (n = 66/119) could be recategorized as BILAG-2004 Grade D (inactive disease).

Conclusions

Proposed changes to the definitions of inflammatory arthritis in the BILAG-2004 index will result in more accurate classification of patients who are more or less likely to respond to treatment.

Introduction

SLE has heterogeneous manifestations in multiple organ systems. The BILAG, formed in 1984, developed the BILAG disease activity index, first published in 1988 [1]. This index is based on physicians’ intention to treat different severities of SLE manifestations across eight organ systems and is used in clinical practice and research. The BILAG index was last revised in 2004, considering advancements in the understanding and detection of SLE disease activity, and now covers nine organ systems [2]. Disease activity indices are central to clinical guidelines, research and access to high-cost drugs in SLE, and therefore should be periodically reviewed and revised in line with emerging knowledge [3–5].

Musculoskeletal (MSK) manifestations affect up to 95% of patients with SLE. ‘MSK distress’ is one of the most commonly experienced ‘difficult symptoms’ affecting quality of life, and MSK manifestations are the most common reason for a patient to be enrolled in a clinical trial [6, 7]. The BILAG-2004 index MSK domain includes items for inflammatory arthritis and myositis.

The current BILAG-2004 index will only allow a moderate (BILAG-2004 Grade B) or severe (BILAG-2004 Grade A) score (the usual minimums for trial entry or new therapy) for inflammatory arthritis if there is synovitis in at least one or two joint(s), respectively, in addition to impaired physical function for A [8]. While the observation of clinical joint swelling is often used to guide determination of synovitis, the BILAG-2004 definition of moderate inflammatory arthritis acknowledges this can be determined through history or examination. By a similar principle, the SLEDAI glossary includes several different features that may be used to identify synovitis as well as swelling on examination.

Patients with SLE demonstrate less clinical joint swelling than other inflammatory arthritides, when they have active synovitis [9]. As such, patients with active synovitis are often classified as BILAG-2004 Grade C within the current definitions, which is not usually considered to be an indication for glucocorticoid or immunosuppressive treatment. This scoring therefore could lead to under-recognition and under-treatment of inflammatory arthritis in SLE.

Furthermore, the assessment of SLE patients for synovitis can be complicated by co-existent non-inflammatory pain, such as from FM, which is reported in around a quarter of patients [10]. Since this is difficult to differentiate from arthralgia due to inflammation, it may be inappropriately attributed to SLE and scored as BILAG-2004 Grade C or even higher. Inclusion of patients with non-inflammatory pain, which is less likely to respond to immunosuppression, may have contributed to underestimation of the efficacy of novel therapies in trials.

Modern imaging, such as US, has emerged as an effective tool to detect subclinical synovitis [11, 12]. US is inexpensive, non-ionizing and non-invasive, and allows assessment of multiple joints and peri-articular structures at the point-of-care [13]. The OMERACT definitions for ultrasonographic synovitis, originally designed for RA, have been validated in SLE [14–16]. Recent studies have demonstrated that 27–28% of SLE patients with arthralgia but no clinically swollen joints have US evidence of synovitis, and there is general support for the fact that US has a role in detecting subclinical synovitis [16–18]. A systematic review of studies using MSK US in SLE identified significant methodological variation between studies, which could explain the wide range in the proportions of abnormalities detected [19].

In a recent multicentre study [US Evaluation For mUsculoskeletal Lupus (USEFUL)], 58.6% patients with SLE and physician-determined inflammatory joint symptoms had a ‘positive’ US for clinically significant synovitis [greyscale (GS) >1 or power Doppler >0] or tenosynovitis [9]. Of these, only 68% had one or more clinically swollen joint. Following exclusion of patients with concomitant FM, those with a positive US, irrespective of presence of clinically swollen joints, were significantly more likely to report an improvement in symptoms after glucocorticoid therapy [9]. Apart from swelling, of the other symptoms of inflammatory arthritis, symmetrical small joint distribution of symptoms, but not early morning stiffness, was found to show the best association with positive US.

The aim of the present work was to propose a data-driven revision to the definition of inflammatory arthritis in BILAG-2004 index that incorporates evidence of synovitis by US and other relevant clinical variables, and to assess the impact of the reclassification on a cohort of SLE patients.

Methods

A BILAG MSK Subcommittee (C.-S.Y., E.M.V., L.-S.T., A.P., C.R., C.J.E., K.M. and R.D.S.) was established to oversee the revision to Musculoskeletal System of the BILAG-2004 index. The glossary and scoring of all items in MSK domain of the BILAG-2004 index were reviewed. The committee identified that inflammatory arthritis items (mild, moderate and severe) required revision to achieve the following:

• Incorporate US findings of synovitis, tenosynovitis, tendonitis or enthesitis, into the definitions of inflammatory arthritis

• Incorporate clinical features that were found to be helpful in differentiating inflammatory arthritis from non-inflammatory arthropathy into the glossary

• Other changes to improve the glossary in the MSK domain

Data sources

Data from two studies on SLE patients with inflammatory joint pain, assessed using MSK US, were used to determine the impact of the revisions [9, 16]. Patient characteristics were extracted, along with BILAG-2004 grading for inflammatory arthritis and presence or absence of inflammation on MSK US. Only patients with BILAG-2004 Grade A, B or C inflammatory arthritis were included in the analysis. The effects of upgrading patients classified as BILAG-2004 Grade C to Grade B and downgrading to Grade D, based on US findings, was then evaluated.

Statistical analysis

Descriptive statistics were used to summarize demographic characteristics, BILAG-2004 MSK score and US abnormalities of patients from the two studies. To analyse the association of swollen joint thresholds, swollen joint count (28 joints for Zayat et al. [16] and 66 joints for USEFUL [9]) was grouped into zero, one or more, two or more, three or more, or four or more. Abnormal US was defined by one or more joint with GS hypertrophy >1 or power Doppler >0, or one tendon with GS >0 and/or power Doppler >0 in the bilateral hands and wrists. This definition was used to compare the swollen joint thresholds, and to calculate the numbers of BILAG-2004 Grade C patients who would be reclassified as BILAG-2004 Grade B using new proposed criteria. Lastly, we compared physician visual analogue scale (VAS) between the different clinical and US criteria that define the old and new BILAG grades using Kruskal–Wallis tests and pairwise comparisons using Wilcoxon rank sum test.

Results

Analysis of US findings for inflammatory arthritis definitions

Association between the swollen joint count thresholds and US abnormality in both studies are shown in Fig. 1. The threshold of one swollen joint was associated with the greatest increase in rate of US positivity with little difference seen when the threshold was increased further in both included studies [9, 16].

While research studies have used OMERACT scoring to define abnormal US, in routine clinical practice this is more difficult to standardize, training the scoring method is variable, and a sonographer will consider a range of factors to decide whether findings truly represent MSK inflammation. We therefore did not include the OMERACT grading in the revised definition.

Revision to definition of severe inflammatory arthritis

Consistent with the above data, the definition of severe arthritis that defines BILAG-2004 Grade A disease is based on severity of symptoms as well as joint count. We have provided clarification on the meaning of basic activities of daily living (ADL). Basic ADL are defined as skills required to manage basic physical needs including ambulating, feeding, dressing, personal hygiene, continence and toileting [20]. We have also made clear that the ADL impairment must be because of the synovitis and not due to other non-inflammatory MSK conditions such as OA, joint damage or FM. As most patients with BILAG-2004 Grade A had US evidence of inflammation, US findings did not have any additional value and was not incorporated into the definition. The other components of the definition are unchanged.

Revision to definition of moderate inflammatory arthritis

In line with this, imaging suggestive of inflammation (synovitis, tenosynovitis, tendonitis or enthesitis) in one or more joint was offered as an alternative to clinical synovitis, tenosynovitis or tendonitis in one or more joint in classifying a patient with SLE as having moderate inflammatory arthritis.

We have maintained that clinical observation of synovitis (including swelling) in one or more joint can allow classification as moderate inflammatory arthritis, based on the receiver operating characteristic curve analysis. The USEFUL study demonstrated that it was joint swelling, rather than joint pain/tenderness, which was predictive of US inflammation. Joint pain/tenderness is also commonly seen in non-inflammatory conditions, such as OA and FM and can be difficult to differentiate. We have therefore removed ‘synovitis can be observed through history’ from this definition.

We have also removed ‘loss of functional range of movement’ as a requirement for moderate inflammatory arthritis since joint swelling alone is a good predictor of US synovitis. Non-inflammatory joint issues can again confound this, and the committee recognized from adjudication in clinical trials that this term may cause confusion, especially for investigators whose first language is not English, as well as confusion with the definition of loss of function in severe arthritis.

Revision to definition of mild inflammatory arthritis

Mild arthritis is scored when there is no swelling, so relies on symptoms to attribute pain to SLE inflammation, which may be subjectively interpreted. The USEFUL study also determined that symmetrical, small joint distribution of symptomatic joints was associated with US positivity, whereas early morning stiffness, reports of prior response to therapy or presence of SLE in other organs were not. The term ‘inflammatory-type pain’ has been superseded with a definition to indicate symmetrical small joint distribution may be suggestive of synovitis, and as such clinical judgement is required to determine whether the patients’ symptoms are due to active inflammation or are non-inflammatory in aetiology. If a US is performed, a normal scan of the symptomatic joints makes clinically significant, treatment-responsive inflammation unlikely. We have added consideration of re-grading of patients with pain but no swelling as no inflammatory arthritis if the US is normal (in addition to the above allowance to reclassify patients as moderate arthritis if the US shows inflammation). Therefore, if US is performed, we would expect all patients to be classified as moderate or no arthritis, not mild arthritis.

We have removed the terminology of arthralgia and myalgia as they are non-specific and easily confounded by non-inflammatory MSK pain especially FM.

Other changes to the glossary

We have updated the information included in the BILAG-2004 glossary, to guide clinicians on how to use the revised definitions of inflammatory arthritis and to highlight the US features that are suggestive of synovitis, tenosynovitis, tendonitis orenthesitis:

• GS hypertrophy of synovium/tendon sheath

• Power Doppler signal in synovium/tendon sheath

• Joint effusion

These features are not specific for SLE synovitis and other conditions that can cause similar findings need to be excluded. Therefore, it is important to use clinical judgement to interpret findings on US. For the imaging findings to be valid, there must have been no significant improvement in symptoms of the suspected inflammatory arthritis from the time of US being performed to current clinical assessment. Once imaging has been used to classify a patient as having inflammatory arthritis, repeat imaging is not necessarily required to define improvement, persistence or worsening in the same symptoms previously attributed to SLE.

Impact of the revised definitions

Data from two SLE studies using MSK US (of hands and wrists) were pooled for the analysis [9, 16]. Twenty-five patients in these studies were classified as asymptomatic inflammatory arthritis (BILAG-2004 Grade D) and were excluded from this analysis. A total of 219 patients were included with demographics and distribution of BILAG-2004 grading for inflammatory arthritis summarized in Table 1.

In the dataset from Zayat et al., 100% of patients graded as BILAG-2004 Grade A were found to have inflammation on US, suggesting that there was no benefit in adding US findings to the revised definition. There were 3/14 (21.4%) of patients from the USEFUL study who were graded BILAG-2004 Grade A but did not have evidence of inflammation on US. This may have been due to clinically observed active synovitis outside the hands and wrists, as these were the only sites imaged.

We further examined the 133 participants in the USEFUL study who were classified as having BILAG-2004 Grade A or B inflammatory arthritis and stratified them by combinations of the presence or absence of any clinical joint swelling or evidence of inflammation on MSK US. The existing BILAG-2004 definitions would only have classified those patients with clinical joint swelling as severe (Grade A) or moderate (Grade B) inflammatory arthritis (n = 68). Participants with no clinical joint swelling but evidence of inflammation on MSK US would have been classified as mild (BILAG-2004 Grade C) inflammatory arthritis in the existing definitions (n = 18). Applying the revised BILAG-2004 definitions, we demonstrate this group have significantly higher physicians global MSK VAS than the group without joint swelling or US synovitis (P < 0.05) but comparable to those with clinical joint swelling [16]. These results are shown in Fig. 2.

A total of 119 patients from both studies [9, 16], with mild inflammatory arthritis (BILAG-2004 Grade C) underwent MSK US of the hands and wrists, with 53 (44.5%) demonstrating US evidence of inflammation (Table 2). With the revised definitions, these 53 patients were re-classified as moderate inflammatory arthritis (BILAG-2004 Grade B). Applying the revised definition would mean that the number of patients who could now be classified as moderate inflammatory arthritis would rise from 72 (32.9%) to 125 (57.1%). These results are shown in Table 3.

For patients classified as mild inflammatory arthritis, a US demonstrating no evidence of inflammation may prompt the clinician to reconsider whether inflammation is the cause of these symptoms. This was the case for 66 of the 119 patients in this analysis. The revised definitions do not mandate that these patients should be downgraded from mild to no inflammatory arthritis (BILAG-2004 Grade D or E), but we anticipate there will be a proportion of patients who are downgraded. A sensitivity analysis with downgrading at lower and upper levels of 20% and 50% showed that between 13 and 33 of patients previously classified as mild inflammatory arthritis within this cohort may be downgraded to no inflammatory arthritis (BILAG-2004 Grade D or E).

The BILAG agreed on the final revisions in a meeting on 18 February 2022 (Table 4).

Discussion

We present the revised BILAG-2004 definitions for inflammatory arthritis, which are both evidence-based and in keeping with current practice [9]. Zayat et al. and the USEFUL study identified the role of US in identifying SLE patients with inflammatory arthritis and also revealed clinical and US features which may indicate a patient is more likely to respond to an escalation in therapy, which are reflected in the revised definitions, summarized in Table 3 [9, 16].

The primary driver of this work was to incorporate US findings into the index; however, upon review of the existing definitions, the subcommittee identified other areas requiring improvement. Periodic review and re-evaluation of disease activity indices is commonplace in other autoimmune rheumatic diseases, such as RA [21].

We chose to take a pragmatic approach to the use of US findings in identifying patients with active inflammatory arthritis. The OMERACT definition of synovitis on US in a synovial joint is grade ≥2 GS synovial hypertrophy (GSH) with or without grade ≥1 power Doppler signal [15]. Both of these features can be graded using a semi-quantitative system from 0–3 at each examined joint [22]. While this grading system is reliable between users, US reports in clinical (rather than academic) practice often do not provide grading, tending to comment purely on presence or absence of GSH or power Doppler. We therefore chose not to include the grades of GSH/power Doppler defined by OMERACT in the revised glossary. We intend to direct clinicians towards the US features (effusion/GSH/power Doppler), which may suggest the presence of active synovitis in an appropriate clinical context. We expect clinical trials may specify timeframes between a US being performed and enrolment, which is reasonable, though for routine clinical practice we did not feel it appropriate or necessary to suggest such restriction. To optimize the likelihood that the imaging findings would still be reflective of the patient’s clinical condition, we have suggested in the glossary that there should be no significant improvement in the patient’s MSK symptoms from the time of US to the time of the current assessment. So as not to increase unnecessary burden on radiology services, repeat US is not required to demonstrate improvement in inflammatory arthritis, even if US findings were used to classify severity. The treating clinician can determine improvement clinically with or without US assessment.

Ultimately, changes in management of patients with SLE are based on clinical judgement. We were mindful that we did not want the definitions to become excessively prescriptive, and we intend for this to be used as an adjunct to clinical judgement. There may be concern that the revised definitions may disproportionately benefit better-resourced centres with greater access to MSK US. The revised definitions do not mandate that all patients undergo US assessment to meet any particular classification of inflammatory arthritis severity and these patients can be classified in the absence of US. Nevertheless, as it is a relatively inexpensive and widely used diagnostic modality, there should be ready access to MSK US.

Within the definition of severe inflammatory arthritis, we have chosen not to include US findings as there is no additional benefit and we expect clinical assessment to be sufficient. By providing more detailed explanation on ‘basic ADLs’, with examples, in the definition of severe inflammatory arthritis, we hope to guide clinicians’ assessments of the impact of disease on their patient [20]. From experiences in SLE clinical trial adjudication, we recognized that some patients were classified as having severe inflammatory arthritis on the grounds of joint swelling but ADL impairment because of concomitant non-inflammatory pathology, such as OA of the hips or knees. Attributing the ADL impairment to joints with active synovitis aims to ensure that each aspect of this definition is relating to inflammatory joint activity only.

The definition of moderate inflammatory arthritis now requires the presence of one joint with active synovitis, tenosynovitis or tendonitis, either identified clinically with swelling or on US. We have not included detection of enthesitis on clinical examination in the definition of moderate inflammatory arthritis as it is difficult to clinically differentiate inflammatory enthesitis from mechanical enthesopathy or conditions causing tenderness over entheseal areas, such as FM. US can identify inflammatory enthesitis in SLE and as such US evidence of enthesitis may be used to define a patient with moderate inflammatory arthritis within the revised definition [23, 24].

The prior definition stated that active synovitis ‘observed or through history’ in one or more joint was required to be classified as moderate. In the USEFUL study, it was the symmetrical distribution of joints and swelling, but not pain without swelling, which were predictive of US inflammation. Our analysis (Fig. 2) demonstrates how the inclusion of participants with US evidence of inflammation, but no clinical joint swelling, allows the classification of an additional cohort as moderate (BILAG-2004 Grade B) inflammatory arthritis. The mean physician MSK VAS for these patients was significantly higher than those without clinical joint swelling or evidence of inflammation on MSK US (P < 0.05), justifying our assertion that these participants should be regarded as having at least moderate (BILAG-2004 Grade B) inflammatory arthritis. There was a cohort of patients who had evidence of clinically swollen joints but no evidence of inflammation on MSK US (n = 13). This group would include patients with clinical joint swelling outside of the joints assessed by MSK US. The mean for this group is similar to those with evidence of inflammation on MSK US (with or without clinical joint swelling). However, it did not show significant difference in physician MSK VAS compared with those without either parameter, and this is most likely due to the small number of patients. We feel that it would still be appropriate to be classified as moderate inflammatory arthritis (BILAG-2004 Grade B) classification.

The subcommittee acknowledges that there is significant overlap between moderate and mild inflammatory arthritis when it comes to the patients’ description of symptoms and have focused on features with proven discriminatory value, with respect to responding to an escalation in treatment. We have included these predictive features within the definition of mild inflammatory arthritis, again to aid clinicians in distinguishing between inflammatory and non-inflammatory pain, with or without an MSK US. Our decision to remove arthralgia and myalgia reflect our focus on clinical features demonstrated to be predictive of response to treatment escalation. We recognized that arthralgia and myalgia are common symptoms in SLE patients and there is significant overlap with non-inflammatory MSK pain (such as FM, joint damage and OA) resulting in difficulty in attribution to disease activity. Furthermore, arthralgia/myalgia on its own (without inflammatory arthritis) would not result in treatment escalation.

This revision has some limitations. In the USEFUL study, US were only performed in the hands and wrist. As hands and wrists are the most common site of synovitis in SLE, we would expect the probability of US abnormalities in other joints to be low, if none was detected in the hands or wrists [25]. It is possible for patients to have synovitis outside the hands and wrists that could be the main driver for the severity scored. US is not 100% sensitive for inflammatory arthritis and false negatives are recognized [26]. Therefore, we have not mandated that a patient initially classified with mild inflammatory arthritis (BILAG-2004 Grade C) should be re-classified as no inflammatory arthritis (BILAG-2004 Grade D) if the US showed no evidence of inflammation. Furthermore, we only included data from two recent MSK US studies in SLE for this analysis. As discussed in the Introduction there are additional studies, though with wide methodological variation between the studies, therefore we elected to restrict the data used for this analysis to the two included studies [19].

In research, using US to identify patients with moderate inflammatory arthritis can enrich trial populations with patients likely to derive benefit from immunomodulatory therapy.

Conclusion

We have revised the BILAG-2004 disease activity index on inflammatory arthritis including incorporation of clinical and US findings suggestive of inflammation, which will improve the grading of the severity of the inflammatory arthritis.

Data availability

The data underlying this article will be shared on reasonable request to the corresponding author.

Funding

No specific funding was received from any bodies in the public, commercial or not-for-profit sectors to carry out the work described in this article.

Disclosure statement: The authors declare no other conflicts of interest relevant to this work.

References