Our cross-sectional multi-center surveys were conducted in 2010, 2012, 2015, and 2021 at 20 neurological centers in total [Japan MG registry (JAMG-R) study]. In each of these surveys, consecutive and established MG patients with various stages of illness over a short duration (4 months) were all enrolled to avoid potential bias. Our study was based on accurate reports by motivated neurologists (not on reports from patients), with many detailed data all quantified and entered on special case cards created in Claris FileMaker Pro^® (Claris International Inc., California, The United States), which were then directly expanded and integrated into a large Excel^® file (Microsoft Corporation, Washington, The United States) and used as the source of the database for the analyses. Input errors and missing values were corrected or re-surveyed through discussion between the secretariat (Y.N.) and the neurologist in charge. We confirm that we have read the Journal’s position on issues involved in ethical publication and affirm that this report is consistent with those guidelines. The ethics committees of each participating institution approved the study protocols. Written informed consent was obtained from all patients enrolled in the study.

Clinical information obtained routinely in our surveys included: sex, current age, age at onset, duration of disease, duration from onset to start of immunotherapy administration, presence of bulbar symptoms, history of MG crisis, presence of AChR-Ab or MuSK-Ab, history of thymectomy and thymic histology, current dose and maximum dose of PSL, use of non-steroid oral immunosuppressants such as calcineurin inhibitors (CNIs), plasmapheresis, and/or intravenous immunoglobulin (IVIg), MGFA classification at worst condition of the disease, current clinical status according to MGFA-PIS, severity scores at current and worst condition, body mass index (BMI), and QOL of patients (15-item myasthenia gravis quality of life scale [MG-QOL15]). Clinical severity was determined according to the MG-ADL, QMG, and MGC. Due to the coronavirus disease 2019 (COVID-19) pandemic, use of a spirometer was avoided, and QMG was not evaluated in the 2021 survey. Regarding QMG, only data obtained in 2010 (1st survey), 2012 (2nd survey) and 2015 (3rd survey) were available.

Data from 3800 patients in total were collected from a series of four surveys in 2010, 2012, 2015, and 2021 ( Figure 1 ). Of those, 267 were excluded due to missing data from the MGFA-PIS, MG-ADL, QMG, and/or MGC. In total, data from 3533 patients, representing a real number of 2486 patients (428 patients enrolled in two surveys, 143 patients in three surveys, and 111 patients in all four surveys), were analyzed (287 patients from the 2010 survey (3, 6), 640 patients from the 2012 survey (24), 923 patients from the 2015 survey (9), and 1683 from the 2021 survey). Of these 3533 patient data, 749 patients with ocular MG were excluded, given that levels of severity scores corresponding to MM-or-better naturally differ between patients with ocular MG and gMG. The remaining 2784 patients with gMG were divided into two groups: 1432 patients with MM-or-better status (achieving CSR, PR, or MM); and 1352 patients with I-or-worse status [not achieving MM or better status; status of improved (I), unchanged (U), worse (W), or exacerbation (E) in MGFA-PIS] (20).

Diagnosis of MG was performed according to the diagnostic criteria set forth in the 2014 Japanese clinical guideline for MG (8, 16): In brief, the diagnosis was based on clinical findings (fluctuating muscle symptoms with easy fatigability and recovery after rest) and the presence of antibodies against skeletal muscle acetylcholine receptor (AChR-Ab) or muscle-specific tyrosine kinase (MuSK-Ab), or, when neither AChR-Ab nor MuSK-Ab was detected, on clinical findings and clinical improvement after administration of anticholinesterase, decremental muscle responses to a 3-Hz train of repetitive nerve stimuli (25) and/or an eyelid ice pack test with thorough exclusion of other diseases. A therapeutic diagnosis by response to plasma exchange was considered to confirm the diagnosis. Single-fiber electromyography was not performed systematically. AChR-Ab were measured by radioimmunoassay, using ¹²⁵I-α-bungarotoxin ≥ 0.3 nmol/L considered positive (5) and MuSK-Ab were measured via radioimmunoprecipitation assay, with a value ≥ 0.02 nmol/L considered positive (26). In Japanese clinical guidelines, low-density lipoprotein receptor-related protein (LRP) 4 antibody-positivity using current assay systems are not considered diagnostic findings due to reasons including a lack of disease specificity (17), and therefore are not measured systematically but are only measured in antibody-negative cases as a reference finding.

The MG-QOL15 is a 15-item patient-administered questionnaire, with each item scored from 0 to 4 points, for a maximum total score of 60 points (27). The revised MG-QOL15 (MG-QOL15r) is a simplified version of the MG-QOL15, with very similar questions that have a 0–2 points distribution and a maximum total score of 30 points (28). As both the original and revised MG-QOL15 were taken as measures of QOL at different time points in our survey, a conversion value of total MG-QOL15 score/60 or total MG-QOL15r score/30 was used to create a corrected MG-QOL15 (cMG-QOL15) for analysis.

Clinically assigned MM-or-better patients (n = 1432) were classified into two groups using the individual cutoff values for three severity metrics (MG-ADL, QMG, and MGC): the strict MM-or-better group and the optimistic MM-or-better group, with severities below or above the cutoff level. Clinical characteristics were compared between strict and optimistic MM-or-better groups on each scale. Due to the COVID-19 pandemic, use of a spirometer was avoided, and QMG score was not evaluated in the 2021 survey. Therefore, regarding the strict/optimistic MM-or-better group on QMG score, data for a total of 745 patients from the 2010, 2012, and 2015 surveys were analyzed.

Comparisons between two groups were performed using the Mann–Whitney U test for continuous variables. Categorical variables were analyzed using Fisher’s exact test to compare the two groups, and the chi-square test to compare the three or more groups. Multiple comparisons for analyses of outcome measures were corrected using Bonferroni correction, in which the level of statistical significance is divided by the trial number of univariate analyses between the two groups, resulting in p-values < .002 on Table 1 as statistical significance, <.004 on Table 2 , and <.003 on Table 3 . Receiver operating characteristic (ROC) curves were used to determine the ideal cutoff values for the MG-ADL, QMG, and MGC. The cutoff point on the ROC curve statistically corresponded to the point at which sensitivity - (1 - specificity) is maximal. This point was the maximum Youden index, which was used as the optimality criterion in cutoff point selection. Sensitivity and specificity values according to the optimal cutoff attributed to the Youden index were identified. The area under the ROC curve (AUC), which measures the ability of a binary classifier to distinguish between groups, was determined to evaluate the discrimination ability of the cutoff points. AUC with 95% confidence interval (CI) was reported. JMP Pro 15 statistical software^® (SAS Institute Inc., North Carolina, The United States) was employed for the analysis, and a value of p <.05 was considered to indicate statistical significance. For CIs, the lower and upper limits were analyzed from bias-corrected confidence limits based on the bootstrap method at the level of p <.05.

Comparison of demographic data between MM-or-better and I-or-worse groups showed that the MM-or-better group had a significantly lower frequency of women, a shorter time to initiation of immunotherapy, and a higher rate of positivity for AChR-Ab ( Table 1 ). Naturally, current MG-ADL, QMG, MGC, and cMG-QOL15 were significantly lower in the MM-or-better group than in the I-or-worse group. Regarding treatment in the MM-or-better group, current PSL dose and frequency of CNI use were lower, maximum dose of PSL and frequency of plasmapheresis did not differ significantly, and frequency of IVIg use was lower when compared to the I-or-worse group.

To analyze cutoff points between MM-or-better and I-or-worse groups on MG scores, ROC curves were drawn for MG-ADL, QMG, MGC, and cMG-QOL15, respectively. ( Figures 2A–D ). The AUC for MG-ADL, QMG, MGC and cMG-QOL15 were 0.912, 0.888, 0.906 and 0.832 respectively ( Table 4 ). Based on the ROC curve, cutoff values were analyzed as follows: for MG-ADL, a cutoff of 2 points offered 82.0% sensitivity and 85.0% specificity; for QMG, a cutoff of 7 points offered 88.7% sensitivity and 70.0% specificity; for MGC, a cutoff of 4 points offered 87.4% sensitivity and 77.9% specificity; for cMG-QOL15, a cutoff of 0.2 points offered 74.6% sensitivity and 77.8% specificity ( Table 4 ).

To examine the impact of overlapping data, we created another dataset (n=1936) by deleting the older data from duplicate patients (first data for double duplicates, first and second data for triple duplicates, and first, second and third data for quadruple duplicates), and analyzed it. The analyzed data corresponding to Tables 1 , 4 are shown in Supplementary Tables 2, 3 , which demonstrated almost the identical results between 2784 ( Tables 1 , 4 ) and 1936 ( Supplementary Tables 2, 3 ) patients.

All three scale cutoffs were achieved in 622 (42.5%) of the 1462 patients with gMG from the 1st survey to the 3rd survey (patients within cutoffs of both MG-ADL and QMG, n = 695 [47.5%]; QMG and MGC, n = 725 [49.6%]; MG-ADL and MGC, n = 673 [46.0%]).

Common characteristics of the optimistic MM-or-better group across all cutoffs naturally included significantly higher current MG-ADL, current QMG, and current MGC compared with the strict MM-or-better group ( Table 2 ). Specifically, current cMG-QOL15, which reflects patient QOL, was significantly higher (worse QOL) in the optimistic MM-or-better group than in the strict MM-or-better group when differentiated with severity scores ( Figure 3 , p <.0001; with MG-ADL cutoff, 0.12 ± 0.15 vs. 0.27 ± 0.23; with QMG cutoff, 0.11 ± 0.13 vs. 0.18 ± 0.13; with MGC cutoff, 0.13 ± 0.16 vs. 0.28 ± 0.22). In addition, optimistic MM-or-better patients based on MG-ADL cutoff were less frequently treated with CNIs and plasmapheresis and included a higher proportion of elderly patients, compared to strict MM-or-better patients ( Table 2 ; Supplementary Table 4 ). Optimistic MM-or-better patients based on the QMG cutoff showed a significantly higher frequency of female patients, a higher proportion of MGFA III and V at worst condition, and higher QMG score at worst condition ( Supplementary Table 5 ). Optimistic MM-or-better patients based on the MGC cutoff showed a higher proportion of females, a higher proportion of elderly patients, and higher QMG at worst condition ( Table 2 ; Supplementary Table 6 ).

Among patients with gMG clinically assigned as strict MM-or-better (n = 1174), 914 patients achieved minimal symptom expression (MSE), defined as scoring 0–1 points on the MG-ADL scale (29). Differences between the group achieving MSE (MG-ADL 0–1) and the group achieving strict MM-or-better (MG-ADL 0–2) on the MG-ADL scale are shown in Table 3 . Mean MGC were significantly lower in the MSE group than in the strict MM-or-better group, whereas mean QMG did not differ significantly between groups and actual differences in cMG-QOL15 were small (0.10 ± 0.13 vs. 0.12 ± 0.15, respectively) regardless of significance, due to the large sample size.