Introduction
Myasthenia gravis (MG) is a rare autoimmune disorder of the neuromuscular junction that is characterized by fluctuating fatigable weakness of voluntary muscles, including common symptoms like weakness of the ocular muscles, bulbar muscles, and generalized muscle weakness (limb, neck, and respiratory muscles). MG-related muscle weakness tends to increase during periods of activity and to improve after rest.
Clinical monitoring of patients with MG and demonstration of the benefit of new treatments in clinical trials both require that MG severity is measured accurately. This can involve different measurement approaches, such as the use of biomarkers or clinical outcome assessments [
1]. The latter uses physician, trained evaluator, or patient assessments of the severity of typical clinical manifestations of MG: muscle weakness and fatigability in various muscle groups (ocular, bulbar, respiratory, axial, and limb muscles). The most widely accepted outcome measures of MG severity [
2,
3] are the Quantitative MG score (QMG) [
4], MG Composite score (MGC) [
5], and MG-Activities of Daily Living (MG-ADL) [
6]. These three outcome measures provide a summary of the severity of MG, combining the clinical manifestations of MG in a single number (“multi-component indices”), with the underlying assumption that the clinical manifestations for each muscle group reflect a common concept of overall MG severity.
While these instruments were all developed to assess the same concept, the reported correlation of the widely used outcome measures in MG is highly variable [
5,
7‐
10], suggesting that they capture different facets of MG severity. The lack of consistently strong associations between these measures suggests both the need for a better understanding of what constitutes MG severity and for further research to develop better measures for use in clinical trials and clinical practice.
In addition to these widely used outcome measures, the MG Symptoms Patient-Reported Outcome (PRO) is a novel PRO measure specific to MG which was developed using the current best standards, which build on extensive direct input from patients [
11,
12], and which has demonstrated promising psychometric performances [
13]. Integrating the patient voice in clinical decision-making, through PRO measures, has been identified as critical in many contexts, from clinical research [
14‐
16] to routine clinical practice [
17,
18], and even more specifically in rare diseases [
19]. Thus, the MG Symptoms PRO has been developed to exclusively capture the symptoms of MG as perceived by the patients (and not as evaluated by clinicians). A patient-reported measure also allows the assessment of the severity of MG symptoms experienced by the patients over a longer period (and not only at the time of the examination). Contrary to the other outcome measures of MG severity, the MG Symptoms PRO uses independent measures of severity for each symptom group, with a separate scale for muscle weakness in the sentinel muscle groups (ocular, bulbar, respiratory), as well as a scale for muscle fatigability of all muscle groups [
13]. It also includes a standalone scale for physical fatigue, including concepts not only of general energy and stamina but also physical manifestations of fatigue in the form of heaviness and weakness in the limbs and body in general, which has recently been flagged as an important symptom for patients with MG [
2,
20]. This modular approach, focusing individually on each different symptom, enables the characterization of clinical manifestations of MG, which may provide a more versatile—and sensitive—measurement system for the severity of MG.
Our objective was to use MG-specific outcome measures data collected in a Phase 2 clinical trial to gain a better understanding of how the typical clinical manifestations of MG may inform the overall severity of the disease, which will eventually allow better measurement of MG severity. Moreover, in this research, we explored how the modular approach of the MG Symptoms PRO complements the most widely accepted outcome measures to assess MG severity.
Discussion
The newly developed MG Symptoms PRO complements the set of currently widely used measures of disease activity and severity of MG. It provides greater granularity for the assessment of the cardinal symptoms of MG (weakness of the ocular muscles, bulbar muscles, and generalized muscle weakness in limb, neck, and respiratory muscles), providing more extensive coverage of the severity of MG, capturing even for mild symptoms that may be imperfectly captured by other measures. It also fills an important gap in the measurement of fatigue. Finally, its modular nature, in which each symptom is assessed independently, allows the possibility of a more meaningful interpretation by focusing on the symptoms that are particularly relevant for a patient at a specific level of MG severity.
Our analyses identified a continuum underpinned by all non-ocular items of the outcome measure data collected in the MG0002 study, which we propose constitute a reasonable representation of the overall severity of MG. The application of the Rasch model assumed a hypothesized continuum of overall severity of MG that is reflected by each of the manifestations of MG captured by the outcome measures included in the analysis. Overall, the clinical and conceptual underpinnings that prompted this hypothesis were substantiated by our quantitative findings. The continuum exposed by our analysis reflects the signs and symptoms captured by all the commonly used outcome measures in MG (MG-ADL, QMG, and MGC), as well as the newly developed MG Symptoms PRO in a clinically meaningful way. In this continuum of MG severity, the lowest levels of severity were characterized by muscle weakness in limbs and physical fatigue, while bulbar muscle weakness manifestations were estimated to be typical of most severe MG. Previous research applying the Rasch model independently to the MG-ADL, QMG, and MGC items also indicated that they could be mapped on a continuum of MG severity [
6,
29,
30]. Our analysis with the Rasch model concluded that the severity of ocular muscle weakness signs and symptoms (ptosis, diplopia, blurry vision) was not accurately reflecting the severity of MG characterized according to the other muscle groups. Previous applications of the Rasch model to individual MG-specific outcome measures had already identified that the ocular items did not fit well with the others [
29], and that their contribution to the characterization of severity was unstable across the different analyses, as they were specific to either very mild [
6], very mild and very severe MG [
30], or very mild and moderate MG [
29]. This singularity of ocular muscle weakness in the assessment of overall MG severity can probably be related to the differential susceptibility of ocular muscles to the autoimmune process of MG, and to neuromuscular transmission disorders generally [
31,
32].
Our analyses also provided useful insight into both the commonly used measures in MG and the newly developed MG Symptoms PRO. The QMG and MG Symptoms PRO had the widest coverage of the overall MG symptom severity, and were able to accurately capture mild, moderate, and severe MG. In contrast, the MG-ADL and MGC did not effectively capture mild MG, which confirms previous findings documenting the floor effect seen with MG-ADL [
3,
9].
The correlations observed between the MG-ADL, QMG, and MGC scores were moderate (ranging from 0.5 to 0.8), which show that, while there is common variability (i.e., common information) captured by these scores, they also have substantial differences in the concept they measure. The levels of correlation estimated here are in line with the wide range of correlations previously observed for these instruments, ranging from as low as 0.33 (between QMG and MG-ADL) to 0.88 (between QMG and MGC, which share some very similar items) [
5,
7‐
10]. In parallel, the MG Symptoms PRO scores assessing weakness in the different muscle groups (Ocular, Bulbar, and Respiratory Muscle Weakness) and Physical Fatigue had lower correlations with the commonly used measures of overall MG severity, which was expected as these scores only capture a specific facet of MG. This finding consolidates the modular approach of MG Symptoms PRO as a complementary way to measure MG symptom severity. Our analysis showed that the MG Symptoms PRO scales cover a wide range of overall MG severity (with scales including all core symptoms of MG, especially a comprehensive assessment of physical fatigue) and, at the same time, provide independent assessments of the severity of each symptom relevant to patients with MG.
The downside of this finer conceptual granularity is the length of the instrument. A shorter version could be developed by deleting some redundant items (as flagged by the conceptual mapping), but the modular nature of the instrument also enables another and perhaps more promising solution: it is possible to use only the items relevant to a specific targeted population, or to the individual patient. For example, patients with severe MG will probably find the Bulbar Muscle Weakness domain the most relevant to consider, while for patients with a milder presentation of MG, the Physical Fatigue domain is most likely better suited.
Another noteworthy finding of our analyses was that physical fatigue, which has been identified as an important symptom in MG [
13,
20], was instrumental in the characterization of a wide range of severity from milder to moderate forms of MG. As the MG Symptoms PRO is the instrument measuring physical fatigue the most comprehensively, its coverage of the milder end of the continuum was improved compared to all other instruments, including the QMG. This feature of the MG Symptoms PRO can be especially beneficial in the context of clinical trials, where the objective is to detect the improvement of MG symptoms. Fatigue in MG includes more than physical fatigue: for example, general fatigue or mental fatigue have been reported [
33,
34], including in the development of the MG Symptoms PRO [
13]. Further research would be needed to explore how these other facets of fatigue fit into the overall severity of MG.
While our findings are an important addition to the understanding of MG severity measurements, there are some limitations to be considered. The first limitations of our analyses are related to sampling. Our analyses were conducted using data from a single study that included only 43 patients; however, multiple data collected for each patient were included in the analysis, leading to more than 500 unique assessments of each instrument. Additionally, the sample was from a clinical trial, so it was a selected sample of patients and may not be representative of a general MG population (e.g., participants in the clinical trial may have more severe MG). In particular, the sample excluded patients with ocular symptoms only; consequently, the findings may not apply to patients with ocular MG. Further research with larger samples, maybe with milder MG and with ocular symptoms only, would be warranted to confirm our results. Another caveat is that some deviations from the Rasch model were observed. While these deviations may be important in most applications of the Rasch model (e.g., for creating a new measure), they are less important in our context: our interpretation was on trends for groups of items (not individual items) to characterize the expression of the underlying continuum. Nonetheless, the estimates from the Rasch model may be marginally impacted by these deviations, so the invariance of the item parameter estimates in other samples might be further explored. Another limitation of our analysis is that we did not conduct any longitudinal analysis to explore how the items from the different outcome measures change over time relative to one another. Hence, we did not explore the invariance of MG severity measurement over time when using MG-specific outcome measures. Finally, our analyses only included the MG-specific outcome measures available in the MG0002 study. Further research should investigate how other existing instruments fit in terms of coverage of the MG symptom severity continuum that we uncovered in our analysis. Specifically, the recently developed MG Impairment Index [
35] was not included in our analyses. It was collected in a subgroup of patients in the MG0002 study, but this sample was so small that it could not reasonably be added to our analyses.