Diagnostic modelling and therapeutic monitoring of immune-mediated necrotizing myopathy: role of electrical myotonia

Abstract Delayed diagnosis of immune-mediated necrotizing myopathy leads to increased morbidity. Patients with the chronic course without 3-hydroxy-3-methylglutaryl-coenzyme-A reductase-IgG or signal recognition particle-IgG are often challenging to diagnose. Immunotherapy response can also be difficult to assess. We created a statistical model to assist immune-mediated necrotizing myopathy diagnosis. Electrical myotonia versus fibrillations were reviewed as biomarkers for immunotherapy treatment response. Identified were 119 immune-mediated necrotizing myopathy cases and 938 other myopathy patients. Inclusion criteria included all having electrophysiological evaluations, muscle biopsies showing inflammatory/necrotizing myopathies, comprehensively recorded neurological examinations, and creatine kinase values. Electrical myotonia was recorded in 56% (67/119) of retrospective and 67% (20/30) of our validation immune-mediated necrotizing myopathy cohorts, and significantly (P < 0.001) favoured immune-mediated necrotizing myopathy over other myopathies: sporadic inclusion body myositis (odds ratio = 4.78); dermatomyositis (odds ratio = 10.61); non-specific inflammatory myopathies (odds ratio = 8.46); limb-girdle muscular dystrophies (odds ratio = 5.34) or mitochondrial myopathies (odds ratio = 14.17). Electrical myotonia occurred in immune-mediated necrotizing myopathy seropositive (3-hydroxy-3-methylglutaryl-coenzyme-A reductase-IgG 70%, 37/53; signal recognition particle-IgG 29%, 5/17) and seronegative (51%, 25/49). Multivariate regression analysis of 20 variables identified 8 (including electrical myotonia) in combination accurately predicted immune-mediated necrotizing myopathy (97.1% area-under-curve). The model was validated in a separate cohort of 30 immune-mediated necrotizing myopathy cases. Delayed diagnosis of cases with electrical myotonia occurred in 24% (16/67, mean 8 months; range 0–194). Half (8/19) had a chronic course and were seronegative, with high model prediction (>86%) at the first visit. Inherited myopathies were commonly first suspected in them. Follow-up evaluation in patients with electrical myotonia on immunotherapy was available in 19 (median 21 months, range 2–124) which reduced from 36% (58/162) of muscles to 7% (8/121; P < 0.001). Reduced myotonia correlated with immunotherapy response in 64% (9/14) as well as with median creatine kinase reduction of 1779 U/l (range 401–9238, P < 0.001). Modelling clinical features with electrical myotonia is especially helpful in immune-mediated necrotizing myopathy diagnostic suspicion among chronic indolent and seronegative cases. Electrical myotonia favours immune-mediated necrotizing myopathy diagnosis and can serve as an adjuvant immunotherapy biomarker.

Needle electromyography (EMG) is an important tool in myopathy evaluations, with myopathic motor units and fibrillation potentials assisting in the determination of and optimal muscle biopsy sites (Naddaf et al., 2018;Sener et al., 2019). In one large series of IMNM (Kassardjian et al., 2015b), electrical myotonia has been reported to occur in 51% (32/51) but details of the muscles affected, and outcomes with immunotherapy were not provided. Although not specific for any disorder, electrical myotonia is known to occur in myotonic dystrophies (DM1&2), non-dystrophic myotonia and less commonly in sporadic inclusion body myositis (sIBM) and toxic-metabolic myopathies (Paganoni and Amato, 2013). In the metabolic myopathy 'Pompe', electrical myotonia was reported in the paraspinal or tensor fascia latae muscles selectively versus generalized involvements in the other conditions (Kassardjian et al., 2015a). The pathophysiology of electrical myotonia is not well understood, but subthreshold ion channel dysfunction in the muscle membrane is considered the primary mechanism (Metzger et al., 2020).
Herein, we study whether electrical myotonia combined with other clinical features can be modelled to assist in IMNM diagnosis and serve as an adjuvant biomarker for the following immunotherapy.

Materials and methods
The study was performed with patient written consent and institutional research board approval. This study followed the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) reporting guideline. We searched our electronic medical record (1 January 2004 to 31 December 2018) to identify all retrospective cases. We then validated the IMNM cohort with subsequent consecutive unique cases diagnosed after 1st January 2019. Inclusion criteria were as follows: age greater than 18 years, EMG, CK and comprehensive scored neuromuscular examinations all within 3 months of pathological and serological confirmatory IMNM testing. IMNM consensus definition was utilized in all patients including muscle weakness in the presence of elevated CK and muscle pathology confirming scattered necrotic myofibres without significant endomysial inflammatory exudates with or without HMGCR, SRP54 IgG positivity (Allenbach et al., 2018). Other acquired inflammatory myopathy controls were also identified that included s-IBM, DM and other non-specific inflammatory myopathies (IM) (De Bleecker et al., 2015;Milone, 2017;Allenbach et al., 2018). The IM category consisted of cases with muscle pathology showing necrosis and regeneration with endomysial or perimysial inflammation with or without auto aggressive inflammation and clinical features not consistent with any other diagnosis including those historically labelled polymyositis. Hereditary myopathy cases with confirmatory genetic testing were also identified including LGMD, mitochondrial myopathies (MtM) and myotonic dystrophies (DM1&2). The immune-mediated and genetic disorders chosen for comparison were based on earlier reports of confusion of those disorders with IMNM and disorders where clinical and electrophysiological overlap might occur including with electrical myotonia (Lotz et al., 1989;Horvath et al., 2003;Kollberg et al., 2005;Allenbach et al., 2014;Hanisch et al., 2014;Kassardjian et al., 2015b;Kazamel et al., 2016;Nojszewska et al., 2018;Mohassel et al., 2019).
EMG studies were performed with a disposable concentric 25-37 mm needle using standard stimulation and recording techniques (Nicolet Viking and Cadwell machines). EMG reports were reviewed for the presence of myotonic discharges and fibrillation potentials utilizing electronic data extraction from EMG reports with chart review confirmation. Myotonic discharges were defined as 20-80 Hz repetitive discharges, waxing and/or waning in amplitude and frequency. For each muscle examined during needle EMG, the presence of myotonic discharges was documented in a comment box associated with the muscle, and the presence of fibrillation potentials documented for each muscle.
All identified IMNM and a subset of the other myopathy designations were randomly selected for chart extraction of 20 clinical variables. The variables selected for review were chosen from our own experience and earlier reports of potential distinguishers between IMNM and other myopathies including statin exposure, CK elevations greater than 1000 U/l, hamstring weakness, hip flexor weakness, tibialis anterior involvement, neck weakness, absent finger flexor weakness and others (Grable-Esposito et al., 2010;Barohn et al., 2014;Kassardjian et al., 2015b).
Among IMNM patients clinical examinations, CK values, HMGCR-IgG (Inova ELISA) and SRP54-IgG (EuroImmune immunoblot) autoantibody status, and EMG findings were reviewed in correlation with the initiation of immunotherapy. The examination findings of patients at diagnosis and at the time of last repeated EMG were used to measure clinical changes. Muscle strength was determined by converting to Medical Research Council (MRC) grading scale summing eight sites bilaterally (neck flexors, deltoids, biceps, wrist extensors, gluteus maximus, gluteus medius, quadriceps and ankle dorsiflexors). A normal examination corresponds to a score of 80 versus the lowest score of 0 corresponding to all 4 limbs being flail and inability to flex the neck. Clinical improvements were defined by having a reduced CK level and improved summated MRC score.

Statistical analysis
Wilcoxon's rank-sum test was used to assess continuous variables; Fisher's exact test was used for binomial variables. Two-sided and P-values less than 0.05 were considered statistically significant for the entire cohort. Univariate and multivariate regression analysis was performed comparing the clinical variables associated with IMNM versus other myopathies.

Data availability
The data file used to generate predicted and actual probability of IMNM prediction is provided in Supplementary  Table 1 and on our online IMNM calculator (http:// imnm.info/).
Clinical myotonia was not reported in any IMNM patients. The median time from muscle weakness onset to our EMG among IMNM patients was 4 months (range 1 week-72 months). The median age was not significantly different among IMNM patients with electrical myotonia, 64 years (range 27-84 years) versus 57 years (range 18-87 years) in those without electrical myotonia. Approximately half (48%, 32/67) with electrical myotonia were female. The CK at initial presentation among IMNM patients was not significantly different, with an average of 7737 U/l (range 393-29 000) in patients with electrical myotonia versus 6663 U/l (537-22 080) in patients without. At the time of our first EMG, 32% (39/119) were on some type of immunotherapy, most commonly prednisone monotherapy 56% (22/39) with 17 of these patients also on a steroid-sparing immunotherapy, 8 receiving IVIG, methotrexate or rituximab. Among them, myotonic discharges were present in 55% (21/39) of those on prednisone alone and 30% (5/17) with combination immunotherapy.
The multivariate regression modelling provided very high diagnostic accuracy, with 97.1% area-under-curve receiver operating characteristics distinguishing IMNM from LGMD, DM, sIBM, IM, MtM and DM1&2. Electrical myotonia and seven clinical variables in combination performed best in the distinction of other myopathies. The other variables in the best-predictive model were statin exposure, CK > 1000 U/l, deltoid weakness, gluteus maximus weakness, finger extensor greater than finger flexor weakness, absence of finger flexor greater than finger extensor weakness and absence of ankle dorsi-flexor weakness. The probability of having IMNM   varied depending on the scoring of these eight variables (Supplementary Table 1). Using 75% probability as a cut-off value, the majority (76%, 90/119) of IMNM patients could be identified using our predictive algorithm, with 1.3% (3/238) of alternative myopathy (2 IM, 1 sIBM) identified using the same 75% cut-off.

Delayed IMNM diagnosis requiring repeat muscle biopsies
Twenty-four percent (16/67) of IMNM patients with electrical myotonia underwent multiple biopsies due to diagnostic uncertainty (15 had 2 biopsies and 1 had 4 biopsies). All but three patients were seen prior to the availability of serologic testing, which ultimately was performed and positive in only 50% (8/16: HMGCR n ¼ 6; SRP54 n ¼ 2; seronegative n ¼ 9; not available n ¼ 1). Insidious chronic course occurred in 50% (8/16). Among these chronic onset cases, DM1&2 (n ¼ 1), LGMD (n ¼ 5), RYR1 (n ¼ 1) and MtM (n ¼ 1) myopathies were thought to be the initial diagnosis but genetic testing was negative or created uncertainty with a variant of unclear significance in large gene panel testing. Eight cases had sub-acute onsets and the initial pathology suggested IM (n ¼ 2), sIBM (n ¼ 1) or non-diagnostic biopsy (n ¼ 5) for which repeat biopsy was revisited with clinical declines. The median delay from symptom onset to IMNM diagnosis was 8.5 months (range 0-132 months). All but one patient had experienced clinical declines before IMNM diagnosis was made and combination immunotherapy commenced. In 81% (13/16) initiating recommended combination immunotherapy (Allenbach et al., 2018) stabilized or improved patient course. Utilizing our multivariate prediction algorithm and data from our first clinical visit, the probability of IMNM over other myopathy diagnosis ranged from 93% to 99% (Table 4 and  Supplementary Table 1).
Of the 30 prospective patients, we identified two that had no myotonic discharges on initial EMG but myotonic discharges on subsequent studies. In the first case, there were clear clinical declines over 1 month on steroid monotherapy with the rise of CK by 2500 U/l. Addition of IVIG leads to stabilization. In the second case, there was equivocal worse thigh weakness which raised the possibility of type 2 fibre atrophy from high dose steroids when the CK had elevated in a borderline range, 300 U/l. However, new myotonic discharges were seen in multiple muscles (vastus, paraspinals and tensor fascia lata) previously normal and IVIG was initiated. Within 3 months clear improvements were seen (MRC 50-66 and CK value reduction of 1200 U/l).

Discussion
The earliest clinical reports of IMNM recognized statin exposure, marked CK elevations, rapid declines, proximal muscle weakness and need for early aggressive long-term immunotherapy (Needham et al., 2007;Grable-Esposito et al., 2010). Our study quantifies the very high odds ratio of having CK values >1000 U/l and statin exposure in the distinction of other myopathies. Our investigation also demonstrates that combining eight clinical features including electrical myotonia can lead to a higher probability to distinguish IMNM from other myopathies having overlapping clinical and/or electrophysiological features (Lotz et al., 1989;Horvath et al., 2003;Kollberg et al., 2005;Allenbach et al., 2014;Hanisch et al., 2014;Kassardjian et al., 2015b;Kazamel et al., 2016;Nojszewska et al., 2018;Mohassel et al., 2019). Specifically, a model utilizing electrical myotonia, statin exposure, CK > 1000 U/l, presence of deltoid, gluteus maximus, finger extensor weaknesses and absence of finger flexor and ankle dorsi-flexor weaknesses can predict IMNM at greatest accuracy by area under the curve statistical modelling.
Electrical myotonia occurred in 56% and 67% of our initial and validation IMNM cohorts respectfully, most commonly with HMGCR-IgG, but also in seronegative and SRP54-IgG positive cases. Electrical myotonia occurred in a wide variety of muscles, most frequently in the proximal upper extremity and paraspinal muscles. Paraspinal muscles should be routinely examined as these muscles oftentimes are the only muscle affected. The identification of electrical myotonia was especially helpful for those who were seronegative or had an insidious chronic course where LGMD, MtM, sIBM and other IM conditions were initially suspected and diagnosis delayed. Patients scoring >75% probability on the model should strongly be considered for an IMNM diagnosis and caution not to over-interpret inflammatory infiltrates on Probability that clinical features favour IMNM (immune-mediated necrotizing myopathy) versus dermatomyositis, sporadic inclusion body myositis, non-specific inflammatory myopathy, mitochondrial myopathies, limb-girdle muscular dystrophy and myotonic dystrophy 1&2. (þ) ¼ variable present; (À) ¼ variable absent; CL ¼ confidence limit; IMNM ¼ immune-mediated necrotizing myopathy. muscle biopsies or variants of unclear significance on genetic testing emphasized. Using the >75% cut-off, only one sIBM and two IM cases were scored of the 238 other myopathy cohort.
Utilizing the data, we have created an online calculator of the probability of IMNM (http://imnm.info/). We emphasize that a calculator based on a statistical model cannot replace sound clinical judgement; thorough history and examination and laboratory testing are still needed for a clinic-sero-pathological diagnosis of IMNM. As an example, multiple drugs and rare metabolic myopathies might produce transient rhabdomyolysis with high probability scores at one specific time (Nance and Mammen, 2015). However, in any patient presenting with sub-acute onset proximal muscle weakness, CK > 1000 U/l without a clear cause, HMGCR-IgG and SRP-IgG test should be sought and/or muscle biopsy should be considered. We hope the recognition of myotonic discharges and our calculator can reduce the delay in diagnosis we observed especially in chronic and seronegative cases as has been reported previously (Grable-Esposito et al., 2010;Suzuki et al., 2012;Garcia-Rosell et al., 2013;Kassardjian et al., 2015b;Ramanathan et al., 2015). We also emphasize that electrical myotonia need not be present to provide accuracy in the model predictions as shown in Supplementary Table 1 where of those with 75% or greater predictions of IMNM, 44% (11/25) do not have electrical myotonia.
The exact mechanism of electrical myotonia in IMNM is unknown. Because we found statin exposure was most commonly linked to electrical myotonia, this is likely a pathological clue with membrane channel dysfunction, secondary to membrane over-expression of HMGCR or channel dysfunction following an immune response. immunotherapy and at last treatment follow-up. Myotonic discharges and fibrillations reduce in frequency while on immunotherapy. On initial pre-treatment EMG Case 3 had electrical myotonia identified in both lumbar and thoracic paraspinal muscles and in Case 13 electrical myotonia was also identified in the infraspinatus muscle. On follow-up post treatment EMG Case 4 also had electrical myotonia identified in the infraspinatus muscle. Electrical myotonia resolution correlated better to treatment response than fibrillation resolution. IMNM ¼ immunemediated necrotizing myopathy; FDI ¼ first dorsal interossei; Hip girdle ¼ gluteus maximus, gluteus medius, iliopsoas or tensor fascia lata.
Previous smaller studies have also provided similar inference (Meriggioli et al., 2001;de Almeida et al., 2008;Kassardjian et al., 2015b). The administration of simvastatin to rabbits has been shown to produce myotonic discharges in association with elevated CK and necrotic, and/or degenerative muscle fibres on biopsy (Nakahara et al., 1992). Further analysis using intracellular microelectrodes in mice showed that perfusion of normal muscles with a solution containing simvastatin or pravastatin resulted in (i) a decrease in threshold currents, (ii) prolongation of spike latency, (iii) repetitive firing and (iv) after depolarization (Sonoda et al., 1994). These abnormalities were thought to represent a dysfunction of the muscle membrane, but ultimately a subthreshold channel dysfunction may explain the high observed frequency of myotonic discharges in IMNM (Metzger et al., 2020). In this cohort, the resolution of electrical myotonia is typically followed by administration of combination immunotherapy and associate with better clinical outcomes than in patients with persistent electrical myotonia. Persistent myotonic discharges on subsequent EMGs may suggest ongoing disease activity and distinguish steroid myopathy from ongoing disease activity. Routine follow-up EMG is not required, but in those where therapeutic response to immunotherapy is uncertain, examining for electrical myotonia may be helpful. We saw this in one patient where steroid myopathy versus active inflammatory disease was distinguished by new myotonic discharges helping prompt escalation of treatment leading to clinical improvements. We note that electrical myotonia resolution correlated better to treatment response than resolution of fibrillations which may persist due to the presence of regenerating muscle fibres.
The limitation of this study is its retrospective nature and lack of available serological testing on all patients, with only a fraction having longitudinal follow-up. Our multivariate model is however, validated in a subset that was not used to generate the statistical model. Our multivariate model will require prospective validation in a larger cohort, but currently, this algorithm is shown helpful in considering the probability of IMNM over other myopathies. Our findings can enhance current clinicosero-pathologic guidelines, expediting diagnosis and earlier recommended immunotherapies.

Supplementary material
Supplementary material is available at Brain Communications online.

Competing interests
Dr. M.M. receiving research funding from a Mayo Clinic benefactor and through discretionary funding from the Department of Neurology for unrelated research projects, and an honorarium to serve as associated editor of Neurology Genetics. Dr. C.D.K. has participated in medical advisory boards for Akcea, Alexion and Takeda, and received teaching honoraria from Alexion and Sanofi Genzyme. Dr. D.D. receiving research support from the Center of Multiple Sclerosis and Autoimmune Neurology, Translational Research Innovation and Test Development Office, and Grifols Pharmaceuticals; has consulted for UCB Pharmaceuticals (all compensation for consulting activities is paid directly to Mayo Clinic); has a patent pending for KLHL11 as marker of neurologic autoimmunity; and is on the editorial board of Journal of Clinical Medicine. Dr. J.R.M. has received research support from Translational Research Innovation and Test Development Office, has patents on the use of mass spectrometry to measure monoclonal immunoglobulins, and receives royalties related to these patents from The Binding Site. Dr. C.J.K. is on the therapeutic CMTA advisory board. No authors have any conflicts of interests as related to this work.

Funding
This work was supported by the Mayo Clinic Foundation, the Center of Individualized Medicine and the Center for MS and Autoimmune Neurology. The funders had no role in the design and conduct of the study; Collection, management, analysis or interpretation of the data; preparation, review or approval of the manuscript; or decision to submit the manuscript for publication.