Fasciculations demonstrate daytime consistency in amyotrophic lateral sclerosis

Fasciculations represent early neuronal hyperexcitability in amyotrophic lateral sclerosis (ALS). To aid calibration as a disease biomarker, we set out to characterize the daytime variability of fasciculation firing.

experience that ALS patients are often unaware of fasciculations until a clinician highlights them. From that time, patients report being most aware of this symptom when relaxed and the muscle is still. It is unknown whether this is due to true variability of fasciculation firing during the course of the day or variable levels of attention paid to consistent fasciculation patterns. This is important for two main reasons. First, the Awaji criteria emphasize the diagnostic value of fasciculation potentials in the neurophysiological assessment of ALS patients. 6 It has been proposed that a 70-to 90-second period is required to confidently exclude the presence of fasciculation potentials using needle electromyography (NEMG) 7 or high-density surface electromyography (HDSEMG). 8 However, an understanding of how diurnal fluctuations may influence these estimates is lacking. Second, this issue will have a bearing on the optimal design of future drug trials that may rely on biomarkers of hyperexcitability or fasciculations to assess outcome. 9 If there is a particular diurnal pattern to fasciculations, it would be essential to assess patients at a similar time of day.
Only one study has addressed this question, albeit in healthy adults only, showing that there was significant daytime variability in fasciculation potential frequency in the abductor hallucis longus. 10 Some variability may be attributed to specific daily triggers, such as caffeine consumption, sleep deprivation, and recent physical exertion, although the impact of these factors on neuronal hyperexcitability in ALS is poorly understood. 11,12 Other factors, such as age and stress levels, are likely to influence fasciculation occurrence over longer time-frames. 13 In this study, we set out to understand how both the subjective awareness and objective measurement of fasciculations vary during the course of the day in ALS patients.  complete a diary of hourly fasciculation awareness for 1 week leading up to each assessment, which took place every 2 months for up to 14 months.

| HDSEMG recordings
At the 9:00 AM assessment, baseline demographic data and a neurological examination were documented. Patients were asked to report the month of symptom onset (persistent focal weakness). We performed the ALS Functional Rating Scale-Revised (ALSFRS-R) 16  have been reported elsewhere. 18 In brief, an initial screen for motor unit potentials was applied to each recording channel. This involved detection of the most extreme amplitudes (positive and negative), representing the peaks and troughs of motor unit potentials. For each of these potentials, the channel with the greatest peak-trough amplitude difference was transferred into a "super-channel." Based on manual counts, we found a linear relationship between average noise levels and the optimal amplitude threshold for inclusion of fasciculation potentials. We confirmed that the optimal automated model was a noise-responsive algorithm, capable of adjusting its amplitude inclusion threshold according to the local noise level (referred to as "noise band"). In addition, areas of the recording with excessive noise were automatically identified and excluded from further analysis. This pipeline achieved a classification accuracy of 88% when applied to 5318 fasciculation potentials that had been identified manually. Unrelaxed motor unit activity was identified by Active Voluntary IDentification, a semiautomated, flexible system built to exclude regular trains of motor unit potentials. 19 Finally, fasciculation potential parameters for each recording (frequency, amplitude median, and amplitude interquartile range [IQR]) were computed by SPiQE. These calculations were performed using Prism version 7.0a.

| Analysis of parameter consistency
Using R version 3.3.1 (R Foundation for Statistical Computing, Vienna, Austria), we employed a two-way, random, absolute agreement intraclass correlation coefficient (ICC) model to assess the variability of fasciculation potential and data quality parameters between timepoints, with 100% reflecting perfect consistency. 20 The two-way model was considered the most appropriate, as it treated both the time-points and the assessed muscles as random effects. Estimates of the ICC, alongside 95% confidence intervals, were generated in R using the icc function as part of the irr package.

| Diaries of fasciculation awareness
Diaries from 19 patients were compiled (mean age: 63 years; 17 males, 2 females; mean duration of symptoms: 30 months; site of onset: 37% upper limb/32% lower limb/31% bulbar). A total of 574 patient-days of diary data were collected in blocks of 1 week. The mean daytime fasciculation awareness score was 0.28 muscle groups. There was significant diurnal variability with a coefficient of variation of 303% (Figure 1).

| Data quality
Ten patients with ALS each underwent three assessments at 9:00 AM, 12:00 PM, and 3:00 PM on the same day (patients' characteristics are displayed in Table 1) (ICC did not differ from zero) across the three time-points for both biceps and gastrocnemius (Figure 2d).

| Fasciculation potential frequency
Half of the patients achieved at least two fasciculation potential frequency (FF) measurements above 50/min in biceps (Figure 2a). The overall ICC across the three time-points for FF was 88% (95% CI: 76% to 95%). When each muscle type was analyzed separately, biceps achieved remarkable consistency, with an ICC of 95% (95% CI: 85% to 99%), whereas gastrocnemius was more variable, with an ICC of 64% (95% CI: 25% to 89%). The very high consistency in biceps recordings was particularly noteworthy in the context of extremely variable noise levels.

| Fasciculation potential amplitude (median and IQR)
When considering biceps recordings, the ICC did not differ from zero for amplitude median, but it did achieve high consistency for amplitude IQR (ICC: 95%; 95% CI: 87% to 99%; Figure 2b and c). Similarly, in gastrocnemius, the ICC did not differ from zero for amplitude median, but achieved a modest degree of consistency for amplitude IQR (ICC: 46%; 95% CI: 4% to 82%). Therefore, amplitude IQR was the most robust measure of fasciculation potential amplitude across the three time-points, excelling in biceps muscles in a parallel manner to fasciculation potential frequency.

| DISCUSSION
In this study we have highlighted the discord between the subjective awareness and objective quantification of fasciculations in ALS.
Although caution is advised when extrapolating beyond a 9:00 AM to led to parallel changes in amplitude median. It was therefore reassuring to see that fasciculation potential amplitude IQR was a much more robust measure, particularly in biceps. This is intuitive, as a rightward shift in amplitude would be unlikely to alter the spread of amplitudes significantly.
There are several limitations to this study. Only 10 patients were included, restricting the generalizability of these results among all ALS patients. We did not control for common modifiers of fasciculation potential frequency, such as caffeine consumption. Ideally, we would have liked to take recordings beyond a 9:00 AM to 3.30 PM time window, but the practical limitations of hospital-based assessments made this difficult. We only assessed two muscles, limiting the generalizability of these conclusions to other muscles. It would be particularly relevant to extend this work to muscles that are involved at a relatively early stage of ALS, such as the first dorsal interosseous and tibialis anterior.
In conclusion, in this study we have shown that fasciculation patterns in ALS patients were consistent throughout the day, particularly in biceps, and often at fasciculation potential frequencies above