Quantitative whole-body muscle MRI in idiopathic inflammatory myopathies including polymyositis with mitochondrial pathology: indications for a disease spectrum

Idiopathic inflammatory myopathies (IIM) are rare and heterogeneous acquired autoimmune disorders. They present with a varying degree of muscular weakness, muscular inflammation, and a spectrum of extramuscular manifestations [1, 2]. The current classification includes dermatomyositis (DM), sporadic inclusion body myositis (sIBM), immune-mediated necrotizing myopathy (IMNM), and antisynthetase syndrome (ASyS)/overlap myositis (OLM) [3, 4]. There is a vivid debate as to polymyositis (PM) should be considered a distinct IIM entity, mainly based on clinicoseropathological studies showing that the majority of PM specimen can be classified into one of the other IIM categories after thorough re-evaluation [5, 6]. However, a small but relevant number of patients with histological presentation of IIM remain seronegative and thus may not be captured by the current classification [7].

This is exemplified by a rare IIM-variant termed polymyositis with mitochondrial pathology (PM-Mito). This disease was first described based on a cohort of 10 patients with clinical phenotype resembling sIBM, but without typical histopathological findings, such as vacuoles and protein aggregates. However, in addition to muscular inflammation, myohistology in all patients showed marked increase of cytochrome c oxidase (COX)-negative muscle fibers and multiple mtDNA deletions, suggesting a mitochondrial pathology [8]. Interestingly, COX-negative muscle fibers are also found in sIBM [9, 10]. Based on the clinical impression and histological overlap, the authors considered PM-Mito a variant of sIBM [8]. Since then, several studies have provided further evidence for a sIBM disease spectrum, including PM-Mito as precursor or slowly progressive variant [11‐15]. However, others reported a relevant response to immunosuppressive treatment in a variable portion of PM-Mito patients, atypical for sIBM. In addition, some of the PM-Mito-patients did not develop an sIBM phenotype, even after several years of disease duration [16, 17]. Therefore, PM-Mito may comprise a heterogeneous subset of disease entities with varying potential for progression to sIBM.

While immunosuppressive therapy is generally considered effective in other forms of IIM, patients with sIBM typically do not show a sustained improvement [18], and initiation of immunosuppression may be detrimental, increasing drug-related morbidity [19, 20]. In this regard, sufficient and early delineation of the different IIM entities is of particular importance. While most patients with sIBM present with a characteristic clinical and histological phenotype, a significant number of patients are initially misdiagnosed [21‐23]. One reason is the lack of specific features in early disease stages of sIBM [22]. Therefore, alternative diagnostic parameters for disease delineation and monitoring are mandatory. Muscle MRI has the potential to yield diagnostic specificity, and to provide an objective method to measure disease severity. While general information on MRI in IIM is limited, few studies have reported findings for specific IIM subtypes [24, 25].

In patients with PM/ASyS/OLM, the typical feature is a symmetrical diffuse edema, mainly affecting proximal muscle groups [26‐28]. Degenerative pathology is initially negligible in most cases, but may be encountered to varying degrees in later stages [27]. While reports on muscle MRI in IMNM suggest a similar pattern of muscular edema, there appears to be an earlier and more pronounced muscular damage with fatty infiltration [29‐31]. In sIBM, degenerative changes appear to be variable and potentially widespread. However, early and consistent degeneration of quadriceps femoris is frequently observed, with selective sparing of the rectus femoris and the adductor compartment [32‐36]. In lower legs, the medial head of the gastrocnemius is predominantly affected [33‐36]. Several authors reported an intramuscular gradient in sIBM, with distal parts of the muscle being more severely affected [32, 34, 35]. Furthermore, asymmetry is considered to be a common feature [32, 34]. Although muscular edema is common, it appears to be less prominent than in other IIMs [36]. Both semi-quantitative and quantitative evaluations of degenerative changes in sIBM were shown to correlate with disease severity [34, 36]. Moreover, a predefined sIBM-specific pattern of muscle involvement has been demonstrated to sufficiently differentiate sIBM from other IIM variants [37].

Although the value of the available MRI-studies in IIM is unequivocal, there are some obvious limitations and unsolved issues. Most studies do not rely on systematic semi-quantitative assessment of the individual muscles (by means of a grading system), impeding interpretation. Quantitative imaging strategies have only been used in very few studies. In most cases, exclusive imaging of the lower extremities instead of whole-body examination was used, resulting in a relevant lack of information. Of particular interest, no MRI findings in patients with PM-Mito have been reported so far, despite the ambiguity in terms of classification and delineation to other IIM variants. Therefore, the aim of the present study was a comprehensive semiquantitative and quantitative evaluation of whole-body muscle MRI findings in different IIM variants including PM-Mito.

The present study reports on one of the largest cohorts of patients with IIM studied with quantitative whole-body MRI to date. It is also the first study to directly address PM-Mito as a distinct IIM subtype.

In a first approach, muscular degeneration of 56 different muscles was semiquantitatively assessed to identify differences between subtypes of IIM, and to derive a disease-specific pattern of degeneration. While patients with sIBM showed a rather severe fatty replacement, modest changes were seen in most of the patients with PM-Mito, as well as in about half of those with PM/ASyS/OLM and three thirds of those with IMNM. This finding reflects the progressive, degenerative nature of sIBM and in contrast the treatable, potentially reversible muscular damage of other forms of IIM [45]. However, a relevant subset of patients with PM/ASyS/OLM and IMNM showed marked muscular damage. This phenomenon has previously been described in various forms of IIM, particularly in patients with IMNM [29]. The fact that fatty degeneration in IMNM was less common in the present study may partly rely on the disease duration at time of MRI, which was shorter in IMNM and PM/ASyS/OLM. However, the observed degenerative changes in both PM/ASyS/OLM and IMNM suggest a risk for irreversible muscular damage, emphasizing the need for early and sufficient immunosuppressive treatment.

Regarding the disease-specific patterns of muscular degeneration, Fig. 5 shows a schematic representation of the summarized results. A consistent subset of muscles was found to be typically involved in sIBM. Predominant involvement of the lower extremities with early and typical affection of quadriceps femoris (particularly vastus lateralis and intermedius), as well as sartorius, medial head of gastrocnemius, and flexor hallucis longus was found. Apart from this core subset, most of the lower limb muscles appeared to be potentially involved in sIBM, with exception of adductors (especially adductor longus et brevis), semitendinosus, and short head of the biceps femoris. This is in good agreement with previous reports, describing a similar pattern of muscular affection, including inter-individual variability in patients with sIBM [33‐36]. The lack of relevant upper extremity involvement is probably due to MRI acquisition procedures. The forearms, which are typically affected in sIBM and therefore relevant for differentiation from other IIM variants, are located at the edge of the field of view and therefore cannot be adequately assessed. In addition, the local resolution is relatively low and does not allow adequate discrimination of individual muscles. Other studies have shown variable MR morphological involvement of the upper extremities in sIBM [33].

In contrast to the consistent morphological changes observed in sIBM, no specific pattern of muscular degeneration was identified in PM-Mito and PM/ASyS/OLM. This may be partly due to the overall low prevalence of fatty involution in these subgroups, which may not allow the identification of congruent alterations. Furthermore, the individual response to immunosuppressive therapy with corresponding impact on degeneration and MR-morphology has to be considered. Similar influencing factors may also apply to IMNM. However, muscular degeneration appeared to be pronounced in some IMNM patients, with a comparable selectivity for gluteal muscles, hamstrings (long head of biceps femoris and semimembranosus), and posterior compartment of the lower legs (medial head of gastrocnemius, soleus). Similar changes have been reported previously [29]. However, Pinal-Fernandez et al. also observed a relevant involvement of the adductors, not observed in this study.

With regard to intramuscular edema, the pattern of distribution was mainly restricted to the muscles identified as being subject of muscular degeneration. Accordingly, gluteal and axial muscles as well as upper extremities were more severely affected in PM/ASyS/OLM and IMNM, while predominant degeneration of the lower extremities was observed in sIBM. Edema was generally less pronounced in PM-Mito. These results might argue for a direct impact of muscular inflammation in the pathophysiology of muscular degeneration in IIM, as suggested by previous reports [46, 47]. Again, there was a high inter-individual variation, both within specific IIM subgroups and between the subgroups. This may be related to the aforementioned clinical heterogeneity regarding disease duration and immunosuppressive treatment.

Several previous studies identified asymmetry as a typical feature of sIBM in distinction to other forms of IIM [32, 33, 36]. Asymmetry in both muscular degeneration and edema in sIBM was confirmed in this study. However, a similar degree of asymmetry was observed in other forms of IIM, questioning the specificity of this feature. This is in contrast to the clinical impression of a typically symmetrical involvement in PM-Mito, PM/ASyS/OLM and IMNM. The reason for this phenomenon remains elusive. Subclinical differences only detected by MRI may be discussed.

To further evaluate the role of muscular degeneration in disease delineation and to address the role of quantitative MRI as a biomarker of disease severity, mean fat fraction (MFF) of selected muscles was analyzed. Consistent with the semi-quantitative findings, MFF was higher in sIBM compared to all other IIMs. These differences were particularly evident in the distal parts of the muscles, while the MFF of the proximal parts was not significantly altered. Thus, a proximo-distal gradient of muscular degeneration appeared to be a typical feature of sIBM in nearly all lower limb muscles examined. This aligns with previous reports [34, 37]. In contrast, most of the PM-Mito patients did not display similar findings, even after several years of disease activity. Therefore, the results of the present study argue against PM-Mito as an sIBM precursor in general, as suggested by some authors [11‐15]. However, a small subset of PM-Mito and PM/ASyS/OLM patients showed MRI-features that were observed in sIBM, especially with regard to the proximo-distal gradient. These patients may indeed be in an oligosymptomatic stage of sIBM, lacking definite clinical or histological characteristics. In these cases, quantitative MRI may be a novel clinical tool to identify these patients at an early stage, facilitating the application of disease-specific therapies in clinical trials and routine practice. Furthermore, this interesting overlap may be indicative of a spectrum disorder ranging from PM/OLM and PM-Mito toward sIBM [16, 45, 48]. This is of particular importance, especially regarding immunosuppressive therapy. Based on the evidence gained, there are no arguments that should generally limit the use of immunosuppressive therapy in patients with PM-Mito. However, the approach presented in this study appears to be highly operational. The integrated use of clinical, serological and muscle biopsy information, together with MRI examination, is mandatory to further facilitate the delineation of IIM subgroups and to allow an informed decision regarding the immunosuppressive strategy.

In contrast to the findings in the lower extremities, there was a relevantly increased MFF in gluteal and paraspinal muscles in PM/ASyS/OLM and IMNM, whereas corresponding changes were not seen in sIBM and PM-Mito. This observation may reflect the predominant proximal and axial involvement in PM/ASyS/OLM and IMNM, as demonstrated for IMNM in a previous study [49].

To address the suitability of quantitative MRI-imaging as surrogate for clinical severity, MFF was correlated with several clinical and radiological parameters. For several muscles, there was a significant correlation between MFF and the total number of affected muscles in sIBM (quadriceps femoris, sartorius, semitendinosus, long head of biceps femoris, medial head of gastrocnemius), PM/ASyS/OLM (sacrospinalis, gluteus medius, rectus femoris, vastus medialis, vastus intermedius, semitendinosus, long head of biceps femoris, flexor hallucis longus, medial head of gastrocnemius), and IMNM (all muscles studied except gluteus medius). Therefore, MFF of these muscles may serve as a surrogate for the overall burden of degenerative changes. Focused quantitative assessment of these muscles may reduce resources, both in terms of acquisition and evaluation. No such correlation was found in PM-Mito.

An inverse correlation was observed when comparing the disease severity as measured by a modified MRC sum score and MFF for selected muscles for sIBM (medial head of gastrocnemius, Fig. 4B), and PM-Mito (members of quadriceps femoris and hamstrings). Thus, the results of this study suggest quantitative MRI measures as a potential clinical tool to monitor disease progression in a subset of IIM patients (e.g., sIBM and PM-Mito). In this context, it seems contradictory that no significant treatment effects were observed on the MRI parameter studied. However, there was a significant difference in the number of treated and untreated patients. In addition, treated patients had a significantly longer disease duration. Both may be relevant confounding factors. In addition, a relevant inter-individual variability in MR morphological response to treatment must be considered, which would be best captured in sequential examinations. Therefore, longitudinal analyses in larger cohorts are mandatory to further investigate the suitability of MFF as a potential biomarker in specific IIM subtypes. In this regard, a recent study in sIBM has shown promising results [50]. Notably, most of the muscles identified as typically affected in sIBM did not show a significant correlation with clinical severity. This may be explained by an early and exaggerated involvement of these muscles that does not reflect the overall clinical progression. There was no correlation between clinical affection and MFF in PM/ASyS/OLM and IMNM. Again, this could be seen as an argument for both the heterogeneity of these patients and the potentially reversible muscular damage not necessarily translating into persistent degeneration.

This study has several obvious limitations. Although the present study is one of the largest quantitative whole-body analyses in terms of the total number of studied patients, the size of the individual subgroups (especially PM-Mito) is comparatively small, impeding data interpretation. Another relevant issue is the retrospective nature of the analysis, resulting in a relevant heterogeneity in terms of disease duration, immunosuppressive treatment, and availability of clinical information (including serological status). This applies in particular to the PM/ASyS/OLM subgroup that—from a clinical standpoint—shows marked differences in clinical and histological appearance. However, patients with PM/ASyS/OLM in this study had similar MR morphological features, and separation would have resulted in even smaller group sizes. In this respect, the PM/ASyS/OLM subgroup in this study can be considered as a combined disease-control group distinct from sIBM. In addition, clinical scores for disease severity were not consequently captured. Hence, the MRC-SS was the only score suitable for retrospective derivation from the available clinical data. However, this score was not consequently evaluated in the context of IIM. Furthermore, some of the MRI parameters (Fischer grading and edema evaluation) used in this study are semiquantitative measures and therefore may be subject to an inter-observer bias. To our knowledge, there is no specific study on the inter-observer reproducibility of the Fischer score. However, based on findings for other semi-quantitative radiological parameters, a relevant bias must be considered [51]. The MFF as a quantitative measure has been shown to have high reproducibility when used with manual muscle segmentation based on anatomical landmarks, as used in this study [52].

In summary, the results of this study delineate sIBM patients from other forms of IIM, both in semiquantitative and quantitative approaches. Of particular interest, patients with PM-Mito did not consistently show alterations observed in sIBM, even in the sensitive quantitative measures applied. This argues against the hypothesis that PM-Mito is a slowly progressive variant of sIBM per se, but favors a spectrum disorder including PM/OLM, PM-Mito, and sIBM. Alternatively, the mitochondrial pathology in PM-Mito may also be an epiphenomenon, not pointing to a specific disease category. As techniques for early detection of sIBM patients have been identified as a major issue for targeted inclusion in clinical trials [53], quantitative MRI might sufficiently add to the diagnostic toolbox in IIM. This also applies with regard to objective measures of disease progression and severity.