The role of potassium in muscle membrane dysfunction in end-stage renal disease

Objective: Uremic myopathy is a condition seen in end-stage renal disease (ESRD), characterized by mus- cle weakness and muscle fatigue, in which the pathophysiology is uncertain. The aim of this study was to assess the role of abnormal serum constituents in ESRD patients by relating them to the excitability prop- erties of the tibialis anterior muscle, at rest and during electrically induced muscle activation, by recording muscle velocity recovery cycles (MVRC) and frequency ramp responses. Methods: Eighteen ESRD patients undergoing hemodialysis were evaluated by blood sample, MVRC, and frequency ramp (before and near the end of dialysis treatment), quantitative electromyography, and nerve conduction studies. Patients were compared to 24 control subjects. Results: In patients, muscle relative refractory period, early supernormality, late supernormality after 5 conditioning stimuli, and latency of the last of 15 and 30 frequency ramp pulses were strongly associated with potassium levels (p < 0.01), showing depolarization before and normalization in the end of hemodialysis. Conclusions: In ESRD patients, the muscle membrane is depolarized, mainly due to hyperkalemia. Signiﬁcance: Since normal muscle fatigue has been attributed to potassium-induced depolarization, it seems likely that this mechanism is also a major cause of the exaggerated muscle fatigue and weakness in ESRD patients. Federation Neurophysiology. Published B.V. an (http://creativecommons.org/licenses/by/4.0/).


Introduction
Uremic myopathy is a term used in common deficiency of muscle function in patients with end-stage renal disease (ESRD) (Floyd et al., 1974;Serratrice et al., 1967). Uremic myopathy refers to the constellation of functional and occasionally structural muscle abnormalities, in patients with ESRD caused by the uremic state itself (Campistol, 2002). The primary symptom of uremic myopathy is weakness of proximal muscles (Floyd et al., 1974;Quintanilla and Sahgal, 1984), however, the relatively nonspecific findings of limited endurance, fatigue, and exercise  (Campistol, 2002).
These non-specific clinical signs are primarily functional in origin, since physical examination, electromyography (EMG), and muscle enzymes are most often normal in ESRD patients diagnosed with uremic myopathy (Floyd et al., 1974), and muscle biopsies only occasionally reveal structural alterations such as muscle fiber atrophy. Most patients are therefore diagnosed clinically (Diesel et al., 1993, Lazaro andKirshner, 1980).
The exact etiology of uremic myopathy remains uncertain, but it is likely that it has a multifactorial origin, and several interrelated pathogenic mechanisms have been proposed (Al-Hayk and Bertorini, 2007;Campistol, 2002;Kunis et al., 2001;Thompson et al., 1993;Liu et al., 2020;Nishi et al., 2020). Hyperkalemia and acidosis, which are often seen in ESRD patients (Moranne et al., 2009), are two such likely contributing factors, which could account for the functional abnormalities without generating structural changes (Riggs, 1989;Sinha and Agarwal, 2013;Z'Graggen et al., 2010).
By giving ESRD patients 3 hours of 'potassium clamp' dialysis Arnold et al. confirmed that hyperkalemia was responsible for nerve depolarization, and they proposed this chronic depolarization as a cause of uremic neuropathy (Arnold et al., 2014). They also demonstrated a significant increase in walking speed with dietary restriction of potassium in such patients and attributed this to nerve excitability changes (Arnold et al., 2017). It could, however, at least as well be attributed to the (unmeasured) parallel changes in muscle excitability, since hyperkalemia affects muscle and nerve similarly.
In 2009, Z'Graggen and Bostock developed a method to record muscle velocity recovery cycles (MVRC) at rest, offering a clinically attractive approach for the assessment of muscle membrane properties (Z'Graggen and Bostock, 2009). The MVRC method is now an established technique that has been used to determine the excitability properties of skeletal muscle in vivo in several clinical conditions (Kristensen et al., 2020;Kristensen et al., 2019;Lee et al., 2019;Tan et al., 2020;Tankisi et al., 2021;Witt et al., 2020b;Witt et al., 2019). This method uses direct muscle stimulation and multi-fiber recordings to assess the effect of multiple conditioning stimuli on the duration of refractoriness (relative refractory period) and the magnitude and characteristics of periods of transient increased velocity (supernormal periods) following muscle action potentials (Bergmans, 1971;Z'Graggen and Bostock, 2009;Z'Graggen et al., 2011).
In 2010, Z'Graggen et al. conducted a study on MVRCs in ESRD patients undergoing hemodialysis (HD), which revealed chronic depolarization of the muscle membrane mainly caused by hyperkalemia (Z'Graggen et al., 2010). Since then, the original MVRC protocol with 1 and 2 conditioning stimuli has been extended by the addition of an option with 5 conditioning stimuli, and a frequency ramp protocol to detect abnormal responses to repetitive stimulation (Boerio et al., 2012;Tan et al., 2016;Tan et al., 2014;Tan et al., 2018). The frequency ramp protocol imitates the conventional short exercise test by simulating the effect of progressive muscle activation to assess its effect on muscle excitability (Boerio et al., 2012).
This study was undertaken to test if MVRC using additional conditioning stimuli and frequency ramp recordings, and the dependence on serum constituents associated with uremic myopathy on this, could further elucidate the pathophysiology of uremic myopathy in patients with ESRD undergoing HD therapy.

Subjects
Eighteen patients with ESRD undergoing HD treatment were enrolled in the study. Patients were recruited from the Dialysis Clinic, Aarhus University Hospital (AUH), Denmark from February 2019 to July 2019. Patients were treated by a regimen of HD or HDF (hemodiafiltration) 2-4 times a week, delivered through an arteriovenous fistula (for clinical characteristics see Table 1). Dialysate composition: potassium 2 mM, sodium 138 mM, calcium 1.25 mM, chloride 105 mM, magnesium 0.75 mM, glucose 5 mM, bicarbonate 36 mM, and acetate 3 mM. Subjects were excluded if: they were less than 18 years old, had a medical history of peripheral nervous system disease or entrapment neuropathy, had any coagulopathies, or received anticoagulation therapy (except dalteparin, administered during the HD procedure). Patients suffering from severe physical or cognitive impairment, were also excluded. Furthermore, 24 healthy control subjects similar in age and sex ratio were included. Controls were recruited by advertisements at www.forsoegsperson.dk and AUH concomitantly. The above exclusion criteria also applied to controls and controls were further excluded, if they had one of the following conditions: a medical history of malignancy, diabetes mellitus, alcoholism, medication, or other causes of polyneuropathy (PNP) or myopathy. This study was approved by the Regional Committee on Health Research Ethics (1-10-72-53-17). Written informed consent was obtained from all patients and controls according to the Declaration of Helsinki.

Examinations
Using same equipment, patients were examined at the Dialysis Clinic, AUH, and controls at the Department of Clinical Neurophysiology, AUH. In patients, clinical evaluation, electrophysiological examination, and blood tests were performed just before HD. Muscle excitability recordings and blood tests were repeated within the last 15 minutes of HD treatment. Controls were examined at one occasion and did not undergo a blood test.

Clinical evaluation
Patients and controls were evaluated clinically by neurological examination including assessment of muscle strength, deep tendon reflexes, and sensory modalities (touch, pinprick, and vibration). Muscle strength in the lower extremities was evaluated using the Medical Research Council (MRC) scale ranging from 0 to 5. Patients were questioned about their subjective experience of generalized muscle fatigue, limited endurance, and exercise limitation after being diagnosed with ESRD.

Laboratory examinations
The blood samples were collected immediately after muscle excitability recordings i.e. right after the patient was connected to the dialysis generator and again within the last minutes of HD treatment. Blood was taken from the arterial needle in the arteriovenous fistula and collected in a blood tube (BD Vacutainer lithium heparin), and a blood gas syringe (Radiometer PICO). ABL835 FLEX was used for blood gas analysis within five minutes after collection of blood and Chemistry XPT for remaining blood analysis. Laboratory examinations included serum levels of potassium, sodium, ionized calcium, phosphorus, chloride, creatinine, carbamide, and blood gas analysis (pH, bicarbonate, and base excess).

Electrophysiological studies
Electrophysiological examinations were performed on the lower leg. The dominant side was chosen for examination. In case of damage to the leg, e.g. scar tissue, the non-dominant side was examined. Skin was cleansed using skin prepping gel and alcohol swipes. The skin temperature of the examined area was retained between 32-35°C, using a heating lamp when necessary. Skin temperature was measured with an infrared skin thermometer and controlled in the beginning of the examination. Nerve conduction studies (NCS) and EMG were carried out using Keypoint EMG equipment 2.11 (Dantec, Skovlunde, Denmark), while MVRCs and frequency ramp were recorded using a separate set-up as described below.
2.2.3.1. Conventional electrophysiological methods. Routine motor NCS of the peroneal and tibial nerves, and sensory NCS of the sural nerve were performed in all healthy controls and patients to exclude entrapment neuropathy and evaluate the presence of PNP using conventional methods and surface electrodes (Stålberg et al., 2019. Motor NCS of the peroneal nerve was performed by supramaximal stimulation using a handheld bipolar stimulator at the ankle proximally, and distally to the fibular head. Compound muscle action potentials (CMAPs) were recorded from a surface electrode at the extensor digitorum brevis muscle. Motor NCS of the tibial nerve was performed in the same manner, by stimulating the nerve behind and proximal to the medial malleolus, and in the popliteal fossa. CMAPs were recorded from a surface electrode at the abductor hallucis muscle.
Sensory NCS of the sural nerve was performed by stimulating the nerve 13 cm proximal to the active surface electrode at sura. Sensory nerve action potentials (SNAPs) were recorded from a recording electrode between the lateral malleolus and the Achilles tendon. The evaluated parameters for motor NCS were distal motor latency, conduction velocity, CMAP amplitude, and minimum Fwave latency. For sensory NCS, the conduction velocity and SNAP amplitude were evaluated.
A standard EMG with quantitative motor unit potential (MUP) analysis was performed in the anterior tibial (TA) muscle using a concentric needle electrode for recording (25 mm Â 30G, Dantec). Standard filter settings (20 Hz-10 kHz) were used at the department: gain (100 lV/division) and sweep speed (10 ms/division).
Quantitative MUP analysis was done by sampling at least 20 different MUPs from 10 different sites in the muscle during weak voluntary contraction. Mean duration, amplitude of all potentials and percentage of polyphasic potentials were evaluated. > 15% polyphasic MUPs are interpreted as abnormal (Stålberg et al., 2019). The presence of spontaneous activity, i.e. fibrillation potentials, positive sharp waves, and fasciculations was also assessed.

Muscle excitability recordings.
Recordings were performed in the TA muscle in all patients and controls. Stimulation and recording were controlled by the QtracS component of QtracW software (written by H. Bostock, copyright Institute of Neurology, University College London, UK) using the M3REC3 protocol. The recording setup included a DS5 isolated bipolar constant current stimulator, a D440 isolated amplifier, and a HumBug 50 Hz noise eliminator (Digitimer Ltd, UK). Stimulation currents were delivered through an isolated monopolar needle electrode (25mmx26G, TECA), serving as cathode. The needle was inserted perpendicularly into the distal third of the TA muscle. A non-polarizable surface electrode served as anode and was placed on the skin 10 mm distal to the cathode. Muscle recordings were obtained using a concentric needle electrode (25 mm Â 30G, Dantec) inserted perpendicularly, 20 mm proximal to the stimulating needle electrode along the muscle fiber direction. A ground electrode was placed distal to  (Charlson et al., 1987). b PNP = Polyneuropathy based on nerve conduction studies (=abnormality in at least two nerves ). c PTH = Parathyroid hormone levels (measured within 3 weeks before examination). d spKt/V = single pool Kt/V, dialysis dose. e UF = Ultrafiltration.
the stimulating needle electrode. The leads were taped to the skin to prevent needle movement during the examination. The needle electrodes were adjusted to obtain a stable single negative peak response with a low stimulation intensity (2.5-7.5 mA). Once set, the stimulation current was not changed during a recording. MVRCs at rest were recorded following 1, 2, and 5 conditioning stimuli, all separated by 10 ms. The inter-stimulus interval (ISI) between the last conditioning stimulus and the test stimulus varied from 1000 to 1.4 ms in 34 steps in approximately geometric series. Test stimuli were delivered every 2 s. The option to proceed with the frequency ramp follows the MVRCs at rest recording in the M3CR3 protocol. To characterize the effects of progressive muscle activation, excitability measures were derived during repetitive stimulations. In the initial part, 10 cycles of test stimuli alone are delivered at 0.5 Hz. After this period, test stimuli continue to be administered every 2 s. However, they are preceded by 1 s trains of conditioning stimuli that increase linearly in frequency (in 1 Hz steps) to a maximum of 30 Hz over 1 minute. Thereby, over a 1-minute period, the average rate of stimulation is increased from 0.5 to 15.5 Hz. Finally, an additional 30 s period of test stimulation was delivered at 0.5 Hz. The MVRC and frequency ramp stimulation protocols are described elsewhere (Boerio et al., 2012;Z'Graggen et al., 2010) and the examination is presented as a video recording (Witt et al., 2020a).

Muscle velocity recovery cycles
Latencies were measured from start of the test stimulus to the negative peak of the muscle action potential. The effects of 1, 2, and 5 conditioning pulses on the latency of the test response were calculated as percentage differences compared with the response to the test stimulus alone. Excitability measures derived from MVRC recordings: (1) Muscle relative refractory period (MRRP), defined as the shortest interpolated ISI at which the latencies of the unconditioned and conditioned test response were identical, (2) early supernormality (ESN) measured as the largest percentage latency reduction of conditioned muscle action potential at ISIs < 15 ms, (3) ESN after 5 conditioning stimuli (5ESN), (4) late supernormality (LSN) measured as the average percentage latency reduction between ISIs of 50 and 150 ms, (5) extra-late supernormality following 2 conditioning stimuli (2XLSN) calculated as LSN (2)-LSN, and (6) extra-late supernormality following 5 conditioning stimuli (5XLSN) calculated as LSN(5)-LSN. MRRP and ESN are very sensitive to changes in membrane potential, while LSN is considered to explore changes in transverse tubules (Z'Graggen and Bostock, 2009).

Frequency ramp
Excitability measures made during frequency ramp: The latency of the negative peak of the muscle action potential, expressed as a percentage of baseline latency recorded at 15 Hz (Lat(15 Hz)) and 30 Hz (Lat(30 Hz)) during the ramp and 30 s after the ramp has stopped (Lat(30 Hz + 30 s)). Latency changes were different for the first and last response in each 1 s train of action potentials, and these changes are indicated by the subscripts First and Last , respectively, so that Lat(15 Hz) First was the latency to the first of 15 conditioning pulses, expressed as a percentage of the baseline value.

Statistics
Data were exported from QtracP, component of QtracW software, and statistical computations were performed using STATA (16.0, Sweden). QtracP was used to generate figures.
The assumption of normal distribution was checked by plotting QQ-plots and histograms, while SD test (unpaired data) and Bland-Altmann plots (paired data) were used to check the assumption of equal variance.
For parametric data, the student's unpaired t-test with equal variance (5XLSN unequal) was used to compare patients and controls. For non-parametric data, the Wilcoxon rank-sum test was used. Differences between measurements before and after HD were tested using paired sample t-test. Few parameters were log-transformed to obtain normality. If this was not feasible, non-parametric Wilcoxon signed-rank test were performed. Differences are presented as mean difference with 95% confidence interval (95% CI). Log-transformed data were back-transformed and are presented as geometric mean x/Ä SD. Non-parametric comparisons were presented as median difference with interquartile range (IQR). Correlation between changes in excitability measures and serum constituents was assessed using linear regression. Mixed linear regression was applied when looking at all recordings (before and after HD), considering the random effect of subjects. This was relevant since each subject was represented twice in the analysis. When assessing relationships between muscle excitability measures and serum constituents, values of p < 0.05 were considered significant.

Clinical and conventional NCS and EMG findings
Five patients showed signs of PNP based on NCS and clinical findings ( Table 1) while none of the controls had abnormal NCS or clinical findings. Three patients could not tolerate the full extent of NCS because of discomfort arising during supramaximal nerve stimulation. Instead, the occurrence of PNP was evaluated based on clinical findings. No subjects showed signs of peroneal nerve entrapment neuropathy according to criteria developed by Danish Consensus Group (Fuglsang-Frederiksen and Pugdahl, 2011). All subjects showed normal muscle strength (MRC = 5) in lower limbs and hip girdle at the clinical evaluation. However, all patients described a feeling of increased fatigue, limited endurance, and exercise limitation arising after being diagnosed with ESRD. EMG was only available in 17 patients and 18 controls, due to late implementation of EMG on the control subjects (6 controls) and discomfort arising during examination (1 patient). The patient group did not differ significantly from controls with regards to MUP duration (patients mean = 14.91 ms (95% CI 14.04; 15.78) vs controls mean = 14.01 ms (95% CI 13.59; 14.42), p = 0.0598) or MUP amplitude (patients mean = 334 lV (95% CI 294; 373) vs controls mean = 327 lV (95% CI 297; 357), p = 0.7842). Four patients showed signs of chronic neurogenic changes, with prolonged duration compared to controls (outside the 95% CI). Patients with prolonged MUP duration also had > 15% polyphasic potentials. One patient showed myopathic changes with short duration and small amplitude compared to controls (outside the 95% CI) (Fig. 1). Neither patients nor controls showed signs of spontaneous activity.

MVRC recordings
MVRC measures are compared in Table 2. There was a shift of the curves to the right and upwards in patients before HD compared to controls (Fig. 2). This finding is reflected in a marked increase in MRRP and a decrease in the remaining parameters. After HD all the MVRC parameters were shifted towards control values, such that there was no significant difference in MVRC measures between controls and after HD (Table 2). Comparing before HD to after HD and controls, there was a significant difference (p < 0.05) in the mean of all MVRC measures (Table 2). Particularly, MRRP, ESN, 5ESN, 2XLSN, and 5XLSN, were different (p < 0.001). Adding more conditioning stimuli (2 or 5) enhanced the early and late supernormality in all subjects (Fig. 2).

Frequency ramp recordings
Changes in mean latencies are compared in Table 2. Two patients and one control had technically inadequate frequency ramp measurements. This was likely because of needle displacement due to muscle contractions at high frequency. In all cases, the latency measurements during the frequency ramp showed a U-shaped latency curve, with initial progressive reduction in latency, reaching a plateau before gradually increasing again (Fig. 3). Patients before HD went ''round the U" more rapidly compared to both after HD and controls (Fig. 3). There was a significant reduction (p < 0.05) in all mean latencies before HD compared to both after HD and controls (Table 2). Especially, Lat(15 Hz) First , Lat(15) Last, Lat(30 Hz) First , and Lat(30 Hz) Last were reduced (p < 0.001). There was no significant difference in mean latencies between controls and after HD recordings.

Correlation with serum constituents
Mean blood test results before and after HD are shown and compared in Table 3. All serum constituents were significantly changed (p < 0.001) after HD, except for sodium and ionized calcium. Potassium, phosphorus, chloride, creatinine, and carbamide were decreased, while pH, base excess, and bicarbonate were increased.
To explore the cause of changes in muscle excitability, excitability measures were compared with serum constituents. Correlations were first sought for measurements before HD, when there should be a stable distribution of the electrolytes between extra-and  To test further whether the changes in excitability measures caused by HD could be accounted for by changes in specific serum factors, we tested for correlation between the pooled excitability measures (before + after HD) and the corresponding serum measures (Table 4A). Changes in all excitability measures were associated with potassium (p < 0.01), but there were also noticeably strong associations between most excitability measures and phosphorus, creatinine, carbamide, pH, base excess, and bicarbonate (Table 4A). However, in each case this association was reduced to insignificance when correlation between potassium and serum constituents were taken into account (Table 4B).
We did not perform a multiple linear regression with all serum constituents included, because of the small sample size. There were no significant differences in excitability measures and serum constituents when comparing patients with PNP (n = 5) and without PNP (n = 14).

Discussion
This study assessed MVRCs of the TA muscle and characterized the effects of electrically induced muscle activation (frequency ramp), providing information about muscle membrane properties in ESRD patients. In summary, we demonstrated, that selected measures of muscle excitability are associated with potassium levels in ESRD patients on HD comparing measures before and after HD. Particularly, we found that hyperkalemia led to increased MRRP, Lat(15 Hz) Last , and Lat(30 Hz) Last and decreased ESN, 5ESN, and 5XLSN (Fig. 4).  It is assumed that ESN is attributed to the passive decay of membrane charge, whereas LSN is attributed to potassium accumulation in transverse tubules (Z'Graggen and Bostock, 2009). Membrane depolarization is indicated by a prolonged MRRP and reduced early supernormality (Z'Graggen and Bostock, 2009;Z'Graggen et al., 2011), which were associated with hyperkalemia in ESRD patients. These results are consistent with Z'Graggen et al.'s findings (Z'Graggen et al., 2010). We found that HD was able to normalize all MVRC parameters, whereas Z'Graggen et al. described that HD was not adequate to normalize the late phase of supernormality (Z'Graggen et al., 2010). These discrepancies are probably due to the difference in time for the second testing. Z'Graggen et al. performed the second testing after terminating HD, whereas this study did the testing within the last 15 minutes of HD to avoid the influence of rapid redistribution of electrolytes, especially increase in plasma potassium, after terminating HD (Abuelo, 2018).
Since Z'Graggen et. al performed their study in 2010, the original MVRC protocol has been extended with 5 conditioning stimuli and a frequency ramp protocol (Boerio et al., 2012), which provides us with parameters expected to have a more pronounced association to changes in potassium level. We recorded MRVCs with trains of 5 conditioning stimuli to increase the early and late component of supernormality. ESN was strongly related to hyperkalemia both with 1 and 5 conditioning stimuli (ESN and 5ESN). On the other hand, LSN showed increasing association with hyperkalemia, when adding more conditioning stimuli, with a decrease in 5XLSN significantly associated to hyperkalemia. This is probably due to a more pronounced accumulation of potassium in the transverse tubules at 5 conditioning stimuli compared to only 1 or 2.
By adding the frequency ramp protocol, we could investigate the effects of progressive muscle activation. We found, that measures before HD went more rapidly ''around the U" (Fig. 3), and an increase in Lat(15 Hz) Last and Lat(30 Hz) Last were significantly related to hyperkalemia. This is likely due to the increasing depolarizing effect of potassium accumulating intracellularly during the trains of action potentials, which, when mild, results in a reduction of latency, but with increasing frequency trains the depolarization reaches a degree, where it starts to cause sodium channel inactivation, and consequently latencies start to increase back towards baseline (Tan et al., 2014). Patients with hyperkalemia may reach the point where sodium channels are inactivated more quickly because of high potassium level intracellularly prior to examination.
It is well established that the most frequent neurological manifestation of hyperkalemia is muscle weakness (Riggs, 2002). Mild hyperkalemia can also be responsible for fatigue as a presenting symptom (Sezgin, 2020). A detailed analysis of the many factors implicated in the development of normal muscle fatigue has concluded that the dominant factor is an increase in extracellular potassium (Cairns and Lindinger, 2008). Potassium builds up during exercise, causing membrane depolarization, and animal experiments show that peak tetanic force drops precipitously when membrane potential depolarizes beyond À60 mV. Hyperkalemia accelerates fatigue by giving the rundown of the potassium gradient a head start. The multiple mechanisms by which potassiuminduced membrane depolarization reduces force, inducing weakness and fatigue, mostly involving inactivation of voltage-gated sodium channels in sarcolemmal and transverse tubule membranes, are discussed further in a recent review (Lindinger and Cairns, 2021).
Our frequency ramp recordings, with latency changes going ''round the U" (Fig. 3) suggestive of increased sodium channel inactivation, are consistent with this potassium hypothesis of muscle fatigue (Cairns and Lindinger, 2008;Sejersted and Sjogaard, 2000) and our new patient recordings show how this mechanism of fatigue is exacerbated by the high potassium levels in ESRD patients.
The association between muscle excitability measures and other serum factors than potassium, seen when pooling before and after HD measures (Table 4A), may indicate an additional association with acid-base balance, but all relationships turned out insignificant after adjusting for potassium. Conversely, when we adjusted the significant associations between muscle excitability measures and potassium for any other serum constituent, phosphorus, carbamide, and creatinine were able to remove the significant relationship between potassium and 5ESN, and Lat(30 Hz) Last , respectively. This could indicate that these serum constituents also have an importance for changes in excitability measures, however, no associations were seen when pooling before and after dialysis. Only phosphorus was significantly associated to 5ESN before dialysis. We cannot exclude the possibility that other, unknown uremic toxins may also affect muscle membrane potential, e.g. by interfering with energy supply as seen in ESRD patients showing elevated resting skeletal muscle oxygen consumption and lower mitochondrial coupling ratio indicating disrupted muscle mitochondrial metabolism and uncoupling of oxidative phosphorylation (Popkov et al., 2019;Rao et al., 2018). However, the strength of the relationship with potassium indicates that any other factors Comparisons were made using paired t-test, presented as mean difference (95% CI) ( a geometric mean x/Ä SD ratio) or b Wilcoxon signed-rank test presented as median difference (IQR). IQR = interquartile range (1st quartile; 3rd quartile). *p < 0.05. ** p < 0.01. *** p < 0.001. **** p < 0.0001.
Lotte Hardbo Larsen, W.J. Z'Graggen, H. Bostock et al. Clinical Neurophysiology 132 (2021) 3125-3135 are probably of minor importance. In addition, HD is known to ''over-correct" the composition of electrolytes resulting in more alkalotic and lower potassium concentration in the end of HD than normal people. We did not find significant changes in EMG parameters; only one patient showed signs of myopathic changes, and no patients showed clinical muscle weakness. However, the described feeling of increased fatigue, limited endurance, and exercise limitation could probably be due to the changes in muscle excitability measures due to high potassium levels in the blood. These reversible functional changes (Z'Graggen et al., 2010), may be early signs of uremic myopathy which suggests that MVRC and frequency ramp could be used to detect early changes of disease, before detectable with conventional electrophysiological methods and occurrence of structural changes in biopsies. Table 4 Association between serum constituents and excitability measures in end-stage renal disease patients.    We have seen that there is no simple clinical test for uremic myopathy: the primary symptom is muscle weakness, but in most cases the only clinical symptom is the relatively non-specific one of muscle fatigue. Here we consider whether that the membrane depolarization by hyperkalemia, for which our muscle recordings provide good evidence, can help explain these symptoms.
The idea that muscle membrane depolarization is the crucial factor in uremic toxicity was proposed by Cotton et al. (Cotton et al., 1979), who recorded membrane potential directly with glass microelectrodes (Cunningham et al., 1971). In 9 ESRD patients they found that a mean muscle fiber resting potential of À78.5 mV was improved to À87.8 mV after 7 weeks of HD, and in another 7 patients on regular HD they found that lowering the frequency from 6 hours thrice weekly to 6 hours twice weekly caused muscle membranes to depolarize from À90.2 to À80.1 mV with consequent increase in symptoms of general malaise and fatigue. They proposed serial membrane potential measurements as a powerful tool to assess adequacy of dialysis, but their method was inconvenient and was not followed up. The MVRC method has amply confirmed the findings on muscle membrane potential changes by more convenient, though indirect, means, and also provided good evidence that much the most important cause of membrane depolarization is hyperkalemia.

Limitations
We were only able to include a relatively small sample size, making it impossible to include all parameters in a multiple regression, which would be preferable in assessing associations between excitability measures and all serum constituents. Muscle excitability parameters in particular MRRP and ESN are temperature-dependent , and the significant difference in temperature between patients and controls may account for some of the difference seen in excitability measurements. However, there was no temperature difference before and after HD for which we found the most prominent findings. We examined the anterior tibial muscle, since its well-defined end-plate zone makes it particularly suitable for MVRC recordings. However, to clarify the relationship between the changes revealed by this study and uremic neuropathy it will be necessary to extend the recordings to proximal muscles, which are more prone to the disease. It will also be important to test for a relationship between hyperkalemia and the structural, histochemical and immunohistological changes revealed by muscle biopsy. Another main limitation of our study is that we could only discuss the role of serum constituents in uremic myopathy. Future studies should address additional possible mechanisms such as uremic sarcopenia/myopathy, possibly resulting from imbalance between muscle protein synthesis and catabolism (Nishi et al., 2020) and mitochondrial dysfunction (Liu et al., 2020).

Conclusion
This study suggests that muscle membrane in ESRD patients is depolarized, mainly due to hyperkalemia. This has been supported with pronounced changes in MVRC and frequency ramp parameters as well as the relationships with serum constituents. It is argued that this potassium-induced depolarization resembles that in fatigued muscle and is likely to be a major cause of uremic myopathy in ESRD patients on HD therapy. Fig. 4. Unadjusted significant relationship between muscle excitability measures and potassium levels in combined patient recordings (before + after dialysis). Regression lines and slope coefficients (b) are for all 36(32) recordings, comprising 18(16) before dialysis (filled circles) and 18(16) after dialysis (empty circles). See Figs. 2 and 3 for excitability measure abbreviations, ***=p < 0.001.

Declaration of Competing Interest
Professor Hugh Bostock receives royalties from UCL for sales of his Qtrac software used in this study. The other authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.