Classification in Idiopathic Inflammatory Myopathies: Where Are We Now?

Abstract

The classification of idiopathic inflammatory myopathies (IIMs) remains a significant challenge in the field of rheumatology and neuromuscular medicine. Despite advances in our understanding of these rare and heterogeneous diseases, the establishment of universally accepted classification criteria remains elusive. We recently highlighted substantial heterogeneity in the application of classification criteria, underscoring the need for unified frameworks [1].

This editorial examines the current landscape of IIM classification, highlighting the proliferation of competing frameworks, the evolving role of biomarkers, and the broader implications for clinical practice and research. We explore the key barriers to achieving consensus and highlight opportunities to balance scientific rigor with clinical utility, ultimately enhancing patient care and accelerating therapeutic innovation.

Idiopathic inflammatory myopathies (IIMs) represent a heterogeneous group of systemic autoimmune disorders and include polymyositis (PM), dermatomyositis (DM), clinically amyopathic dermatomyositis (CADM), immune-mediated necrotising myopathy (IMNM), anti-synthetase syndrome (ASyS), overlap myositis (OM), and inclusion body myositis (IBM). These conditions are characterized by a predilection for skeletal muscle inflammation, although amyopathic forms exist. Additionally, these conditions often present with a spectrum of extra-muscular manifestations such as distinctive cutaneous lesions, interstitial lung disease, myocarditis, and arthritis. Variations in the clinical spectrum, prognosis, and therapeutic responses suggest distinct underlying pathophysiological mechanisms.

The rarity of IIM adds another layer of complexity to the classification dilemma. With limited patient populations available for study, generating robust data to support detailed subtyping is challenging, and the statistical power of many studies remains constrained. The scarcity of cases often necessitates lumping diverse clinical entities into broader categories to ensure sufficient cohort sizes for research, even if this oversimplifies the nuanced differences between subtypes. Historically, DM and PM have been studied together; however, contemporary literature indicates DM itself to be a highly heterogeneous condition characterized by distinct antibody-defined subtypes. Additionally, many conditions traditionally classified as PM have now been redefined into more precise subtypes such as IMNM and ASyS, rendering true PM extremely rare.

These issues of rarity and heterogeneity are reflected in the classification criteria landscape. A fundamental dilemma is whether classification criteria should adopt a broad framework which incorporates subtyping, as in the EULAR-ACR 2017 classification model [2] or whether IIM subtypes merit separate, detailed classification efforts, as exist for IMNM, IBM, CADM, and ASyS (Table 1). Over-arching IIM criteria promote standardization for research and reduce complexity. Distinct subtype classifications arguably enhance precision but risk fragmenting already limited research efforts.

The evolution of classification frameworks for IIM and the plethora of definitions employed in the contemporary literature highlights the complexities of classification. The pioneering framework established in 1975 by Bohan and Peter [3] focused on clinical features such as symmetrical proximal muscle weakness, elevated muscle enzyme levels, electromyographic abnormalities, and characteristic muscle biopsy findings. Despite its foundational importance, this early schema had notable limitations, including reliance on nonspecific electromyographic findings, imprecise definitions of DM skin lesions, and vague exclusion criteria. Over subsequent decades, the identification of new clinical subsets and an expanding spectrum of myositis-specific autoantibodies have rendered these early criteria increasingly inaccurate. As diagnostic techniques evolved, newer classification criteria were developed (Table 1). The EULAR/ACR classification criteria, introduced in 2017, represented a significant advancement by integrating clinical features, laboratory findings, muscle biopsy characteristics, and a defined autoantibody (anti-Jo1) [2]. Importantly, this represented the first data-driven classification system for IIMs which were validated with a robust methodology, a marked advance over previous criteria sets. However, the 2017 criteria have notable limitations: they do not include ASyS or OM as distinct subtypes and omit key clinical features such as mechanic's hands, shawl sign, and ILD, potentially leading to subtype misclassification and inflating the prevalence of polymyositis [4]. Additionally, IMNM was recognized as a distinct entity partway through the assembly of the derivation cohort, limiting its distinction from PM in the subclassification tree [2]. More broadly, another challenge across all IIM classification efforts is the substantial clinical heterogeneity within recognized subgroups, with important disease-specific features—such as anti-MDA-5 associated rapidly progressive ILD and the higher prevalence of malignancy linked to anti-TIF1-gamma and anti-NXP2 auto-antibodies [5]—often unaccounted for.

Efforts are currently underway to reassess and update the EULAR/ACR criteria [6]. Recent literature has highlighted the limitations of the ACR/EULAR classification criteria, noting that the inclusion of myositis-specific antibodies (MSAs), myositis-associated antibodies (MAAs), and neuromuscular imaging could enhance classification sensitivity [7, 8]. In parallel with the broader revision of EULAR/ACR criteria, several initiatives are focusing on refining the classification of specific IIM subtypes. For example, the absence of ASyS from the EULAR/ACR criteria is being addressed through a collaborative project aimed at developing dedicated criteria for ASyS as a distinct entity within the IIM framework: the CLASS project [9]. Beyond EULAR/ACR working group initiatives, other centres and collaborations have launched independent initiatives to refine IIM classification. These include efforts to classify amyopathic DM [10] and initiatives by the European Neuromuscular Centre (ENMC) targeting IBM [11], IMNM [12] and ASyS [13]. This growing landscape highlights the complexity of reaching a global consensus on the classification of IIM, with multiple consortia, not to mention numerous single-centre and discipline-specific initiatives, undertaking parallel efforts. It remains uncertain how these various efforts will be integrated with the proposed revision of EULAR/ACR IIM. We note that there have been at least 17 different classifications criteria proposed and used since the publication of the Bohan and Peter criteria (Table 1), suggesting that in real-world practice there is a lack of consensus on an accepted gold-standard criteria. This diversity underscores concerns highlighted in Section 1, that capturing the full heterogeneity of IIM within a single set of classification criteria may be inherently challenging.

Autoantibody testing presents its own constellation of challenges in IIM classification. Despite their central role in contemporary practice, significant issues persist with test standardization, interpretation, and accessibility. Commercial assays vary substantially in sensitivity and specificity, while immunoprecipitation—considered the gold standard—remains prohibitively labor-intensive, expensive, and available only at specialized centers, creating substantial delays in diagnosis. Additionally, there is no unified model for interpreting these tests across laboratories and regions, with inconsistent reference ranges and reporting practices leading to potential misclassification. The validity of many newer antibody tests remains incompletely characterized, especially for antibodies like anti-HMGCR, anti-NXP2, and anti-MDA5, where clinical correlations continue to evolve. Further complicating matters, novel autoantibodies continue to emerge, suggesting our current panel captures only a fraction of the immunological diversity within IIMs. The ultimate solution may lie in precision phenotyping that integrates genetic profiling, comprehensive autoantibody screening, and molecular characterization to enable accurate classification and personalized therapeutic approaches that target specific pathophysiological pathways rather than broad clinical syndromes.

The evolution of classification criteria poses challenges in interpreting historical data. Researchers must navigate the divergence between past and current standards, integrating registry data and advancements in biomarkers and imaging technologies to refine classification systems [1]. For example, much of the existing evidence based on current myositis treatment strategies was derived from studies that collectively analyzed heterogeneous IIM populations without detailed consideration of distinct clinical subtypes. Similarly, our understanding of IIM malignancy risk predominantly stems from historical cohort per registry-based studies that may not have delineated subtypes as understood today.

The field of IIM has experienced major advances that have transformed our understanding of these complex diseases. The discovery of novel autoantibodies has significantly refined our ability to delineate disease subtypes, while advances in imaging techniques, such as MRI and PET scans, have enhanced diagnostic precision by identifying subclinical muscle inflammation [14], patterns of muscle involvement, extra-muscular involvement [15] and targeting muscle biopsy [16]. Additionally, molecular studies have provided deeper insights into IIM pathophysiology, uncovering pathways and novel biomarkers [17]. Collectively, these advances have expanded the scope and complexity of classification, emphasizing the need for frameworks that accommodate rapidly evolving scientific knowledge.

However, the pace of discovery presents challenges for the field of classification. Developing robust, internationally accepted criteria requires extensive multidisciplinary collaboration, large-scale data collection, and careful validation, all of which take considerable time. Creation of the 2017 EULAR/ACR criteria, for example, began in 2004 [2], taking over a decade to publish, with clinical data collection taking place between 2008 and 2011. By the time of publication, commercially available autoantibody assays had evolved significantly and become more widely available, leaving a gap between the cohort data used to generate the criteria and the clinical phenotyping tools available to clinicians in practice. The prolonged timelines for classification development can lead to frameworks that, while rigorous, may already be outdated by the time of their implementation. This creates a paradox where classification criteria, designed to standardize and advance the field, may lag behind current clinical and scientific knowledge. While data-driven classification criteria are regarded as the gold standard [18] because they provide evidence-based frameworks that promote reproducibility and standardization, they are inherently constrained by the quality and scope of the underlying dataset.

Traditionally, IIM classification systems have been led by clinicians with IIM-specific expertise and hence emphasize clinical features and organ-specific manifestations. This clinical approach has been instrumental in guiding diagnosis and treatment, but it risks oversimplifying the underlying biology and may fail to capture the shared molecular mechanisms that transcend organ-specific boundaries.

The advent of multi-omic profiling offers a transformative opportunity to rethink IIM classification from a molecularly driven, organ-agnostic perspective. Precision modeling may offer the most promising path forward for IIM classification. Liquid biopsies, which can capture circulating biomarkers including cell-free DNA, exosomes, and cytokines, represent a less invasive alternative to traditional muscle biopsies while potentially offering greater insights into disease activity and progression. Multi-omic studies integrating genomic, transcriptomic, proteomic, and metabolomic data are revealing molecular signatures that transcend traditional clinical boundaries, potentially redefining how we conceptualize these disorders [19]. Critically, emerging research on gene–environment interactions suggests that IIM development reflects complex interplays between genetic susceptibility and environmental triggers, including infections, medications, and ultraviolet exposure. Until we develop accurate classification systems incorporating these molecular insights, our field will likely remain decades behind other rheumatological conditions [20, 21] where newer pathophysiological drivers are established. The current literature continues to expand rapidly, often derived from cohorts using ambiguous or outdated classification criteria, creating a fragmented evidence base that impedes meaningful progress in understanding disease mechanisms and developing targeted therapies.

For example, DM is characterized by a prominent type 1 interferon signature [22]. As such, at a molecular level, DM shares greater similarity with other type 1 interferon-driven diseases, such as SLE, than with molecularly distinct myopathies like IMNM and IBM [23]. This observation raises critical questions about the appropriateness of classifying DM alongside other forms of myositis with fundamentally different underlying mechanisms.

This is an exciting time in IIM research, with the field experiencing unprecedented expansion in clinical trials and therapeutic innovations. This rapidly evolving therapeutic landscape highlights the pressing need for harmonized classification criteria to guide clinical research effectively. Beyond clinical trials, accurate disease definitions are also essential for developing clinical guidelines. Accurate and consistent classification also serves to attract research funding, direct research interest, and attract commercial investment by clearly delineating disease burden.

In spite of these challenges and limitations, there are clear paths forward. Future progress is contingent upon the synthesis of clinical data with contemporary molecular and imaging techniques, the establishment of dynamic, adaptable classification frameworks, and the strengthening of international collaborative networks. Open communication, transparency in data sharing, and sustained multidisciplinary engagement will be important to foster consensus and drive iterative refinement of classification systems. The ultimate solution may lie in precision phenotyping that integrates genetic profiling, comprehensive autoantibody screening, and molecular characterization to enable accurate classification and personalized therapeutic approaches that target specific pathophysiological pathways rather than broad clinical syndromes.

Conceptualization: J.D. Data curation: J.D., K.S., I.T. Formal analysis: all authors Funding acquisition: J.D. Investigation: J.D., K.S., I.T. Methodology: all authors Project administration: J.D. Resources: J.D. Supervision: J.D., L.G. Validation: J.D., L.G. Visualization: all authors Writing – original draft preparation: K.S., I.T. Writing – review and editing: all authors.

L.G. and J.D.: The views and opinions expressed are solely those of the author and do not represent or reflect those of any affiliated institution.

The authors declare no conflicts of interest. J.D. has received consulting fees/research funding from NKARTA and CSL PTY LTD.