This study enrolled patients diagnosed as ALS based on the Awaji criteria (de Carvalho et al., 2008), and who had undergone LF-RNS tests on the accessory nerve at our hospital between June 2019 to June 2023. Exclusion criteria were as follows: (1) patients with a history of poliomyelitis, (2) patients with a spinal cord tumor, (3) patients with other diseases affecting the peripheral nerves, neuromuscular junction (NMJ), or muscles, and (4) patients with incomplete clinical information. This study was approved by the ethics committee of our hospital and written informed consent was obtained from each participant.

Clinical information was collected at the time of diagnosis, including sex, age at onset, onset site, disease duration, diagnostic level, ALSFRS-R score (Cedarbaum et al., 1999), body mass index (BMI), forced vital capacity (FVC), and the results of RNS tests. The rate of disease progression (ΔFS) was calculated through the following formula: ΔFS = [48–ALSFRS-R score]/duration (month) (Labra et al., 2016).

Experienced neurologists followed up on the participants regularly every 6 months with a telephone call or a face-to-face interview until June 30, 2024. For a deceased patient, survival time was defined as the interval time between date of onset and date of death or tracheotomy, which was taken as the equivalent to death. For the surviving patients, survival time was calculated from date of onset to the date of the last follow-up visit (June 30, 2024). Patients who made no contact on two successive occasions were assigned to loss in the follow-up group.

The electrodiagnostic studies were performed on a Keypoint workstation (31A06, Alpine Biomed ApS, Copenhagen, Denmark). Skin temperature over the examined muscle was maintained at 32°C or above throughout the entire period of measurement. Spontaneous activity was considered as present when reproducible trains were observed for at least 300 ms following needle insertion at >2 sites within the same muscle. Ten MUPs were measured in each muscle.

LF-RNS was performed on the side with the more severe symptom involvement, and its results were recorded from the trapezius muscle with stimulation of the accessory nerve, abductor digiti minimi (ADM) muscle for the ulnar nerve, and tibialis anterior muscle (TAM) for the peroneal nerve. For patients asymptomatic on both upper limbs, the RNS test was performed on the right side. A train of 10 3 Hz stimulus was delivered to the nerves, which was repeated three times at an interval of at least 30 s between each stimulus train. The peak-to-peak amplitude decrement and the difference in negative peak area from the first CMAP to the fourth/fifth one were measured and expressed as percentages. A decrement of 10% or more was considered to be abnormal (Kimura, 1989).

Venous blood was collected from the cubital vein of the patient early in the morning after an overnight fast. Biochemical indicators were measured in the department of biochemistry in our hospital. The serum interleukin-6 (IL-6) level was measured using a chemiluminescence immunoassay and with the Immulite 2000 chemiluminescence immunoassay analyzer (Siemens), according to the instructions of the manufacturer. The functional sensitivity (FS) of IL-6 was 2 pg./mL. The detection values of IL-6 lower than FS were recorded as <2 pg./mL. Referring to the methods reported in the previous literature, values lower than FS were calculated as 0.5 × FS without affecting the statistical results. Serum triglyceride (TG), total cholesterol (TC), high-density lipoprotein cholesterol (HDL-C), low-density lipoprotein cholesterol (LDL-C), serum uric acid (UA), serum creatinine, and total serum bilirubin (TSB) were measured using an autoanalyzer (Cobas 8,000 modular analyzer series; Roche Diagnostics, Basel, Switzerland).

The pulmonary function test (PFT) was performed by certified technicians in the pulmonary function laboratory of our hospital. Pulmonary ventilation was measured with a spirometer (MasterScreen Body; Jaeger, Wurzburg, Germany) and the measurement results were analyzed with computer software (Master Lab Manager V5.31.0 software; Jaeger). Each test was repeated three times, and the best result was chosen as the final result. All spirometric values were expressed as a percentage of the predicted values.

MS-Excel 2016 (Microsoft, Redmond, Washington) and MATLAB R2019b (MathWorks, Natick, Massachusetts) were employed for statistical analysis. Measurements that were normally distributed were expressed as mean ± standard deviation (SD), and those non-normally distributed as quartile M (Q1, Q3) values. Medians were compared with the Mann–Whitney U test between two groups. Enumeration data were presented as the number of patients and percentages. The Chi-square test was used for comparison between different enumerative groups. Bonferroni correction was made to p values when several statistical tests were being performed simultaneously. The survival analysis was performed by Kaplan Meier method with a log-rank test. The significance level was set at p < 0.05.

The extent of variability both in amplitude decrement and area decrement between the first train and the second or third one was investigated by means of coefficient of variation (CV). As to the area decrement from the first CMAP to the fourth, the values of the differences between the first train and the second, as well as between the third one and the first, were used to derive the mean difference value. Standard deviation and CV were derived from the data set that included the mean difference values of all patients. The lower CV indicates greater stability.

Stepwise regression was used for model building. In model validation, the predicted number of patients with abnormal decrement was derived by calculation of the model based on data from patients with abnormal decrements. The predicted number of patients with normal decrements was derived by calculation based on data from patients with normal decrements. Consistency in the same results was calculated as (predicted number of patients with abnormal decrement + predicted number of patients with normal decrement)/(actual number of patients with abnormal decrement + actual number of patients with normal decrement) × 100%.

A total of 421 patients with ALS were enrolled in this study, all having undergone LF-RNS on the trapezius with stimulation on the accessory nerve. Details on the demographics of ALS patients are presented in Table 1. Of the 421 patients, 191 had also undergone LF-RNS on the ADM, and 54 on the TAM.

Two hundred and eight patients attended follow-up visits between June 2019 and June 2024. One hundred and five patients had died or undergone tracheostomy, and 103 were still alive at the last follow-up visit. A total of 225 patients have complete records of biochemical tests. For the multivariate model building and validation, their clinical data were divided into a training set and a test set (Figure 1).

The rate of abnormal decrement in trapezius is higher than that in ADM and TAM (Table 2). No significant difference was found in the abnormal rate of either amplitude or area decrement when the first CMAP is compared with either the fourth or fifth one.

The abnormal rate in area decrement is higher than that in amplitude decrement, and 98.8% of the patients with an abnormal amplitude decrement in the accessory nerve also had abnormal decrement measured by the area. However, of the 287 patients with an abnormal area decrement between the first CMAP and the fourth at 3 Hz on the trapezius, 169 (58.9%) had an abnormal amplitude decrement. Of the 248 patients with a normal amplitude decrement, 118 (47.6%) had abnormal area decrement.

Overall, the CV of the area decrement is lower than that of the amplitude decrement (Table 3). Specifically, as far as the decrement response from the first CMAP to fourth CMAP on the trapezius during 3 Hz RNS is concerned, the mean value of CV for area decrement (0.995) is lower than that for amplitude decrement (1.073). The mean absolute differences between stimulus trains in amplitude decrement is significantly different from that in area decrement (Supplementary Table 4). Similar results can be found on ADM and TAM. As such, CMAP area decrement is more stable than amplitude decrement.

A total of 345 patients underwent 3 Hz RNS of the accessory nerve and needle EMG of the ipsilateral trapezius. A total of 48 patients underwent 3 Hz RNS of the peroneal nerve and needle EMG of the ipsilateral TAM. CMAP decrement is correlated with MUP duration and amplitude (Table 4). Patients with fibrillation potentials and/or positive sharp waves (fib-psw) in trapezius have a larger decrement range than those without fib-psw (Table 5).

Since no significant difference was found in the abnormal rate of amplitude or area decrement when the first CMAP is compared with either the fourth or fifth one and the number of patients having abnormal decrement on ADM and TAM was limited, we took the decrement from the first stimuli to the fourth on the trapezius as an example to further investigate the correlations between clinical indicators and the rates of abnormal decrement.

Seven clinical indicators (onset age, sex, onset site, diagnostic level, diagnostic delay, BMI, FVC), and nine biochemical indicators (HDL-C, LDL-C, LDL/HDL, IL-6, TC, TG, TSB, serum creatinine, and serum UA) were considered as components that might contribute to model building. The final results of the stepwise regression (Supplementary Table 1) showed that the following model provided the best fit for the data set:

Where Ys is the standardization of area decrement (Y), X₁ is FVC, X₂ is HDL-C, X₃ is serum UA, D₁ = 0 if female, D₁ = 1 if male. (D₂, D₃) represents onset site: upper limb onset = (0, 0), lower limb onset = (1, 0), and bulbar onset = (0, 1). The F-value is 5.713 (p < 0.001, n = 172), Durbin Watson test (DW test) statistic value is 2.029 and the corresponding p-value is 0.5729. T-values for each parameter reach significant level (p < 0.05).

The total consistency between the actual value and the prediction value is 75.581%, where the discrepancy mainly occurs when the decrement is <10% (Table 6). Leave-one-out cross-validation and external validation (Supplementary Table 2) suggest a good and stable prediction performance of model (1).

FVC, HDL-C, serum UA, sex, and onset site are independent factors contributing to the area decrement (Figure 2). The model (1) suggests that with all other variables being held constant, a one unit increase in FVC (% predicted), HDL-C, or serum UA, were, respectively, associated with a 0.066, −4.4351, or 0.147% decreases in the range of area decrement (Supplementary Table 3). The expected range of area decrement in males was 2.9% lower than that in females. For patients with lower limb onset and bulbar onset, the expected ranges of area decrement decreased by 6.565 and 3.988%, respectively, in comparison with patients with upper limb onset. Therefore, HDL-C had the greatest impact on decremental response as measured by CMAP area, followed by serum UA and FVC; females had a larger range of area decrement than males; upper limb onset contributed more to the decremental response in accessory nerve than lower limb onset and bulbar onset.

Notably, in our cohort, there are 118 patients having abnormal decrement in CMAP area but normal decrement in amplitude. To find out their clinical characteristics, we compared them with patients having abnormal decrement in both measurements (Table 7). The results suggested that the patients with abnormal decrement only in area have shorter disease duration, higher ALSFRS-R scores, and a larger percentage of non-definite ALS and upper limb onset. No statistical difference in sex, onset age, BMI, FVC (%), and disease progression rate was observed.

Finally, we analyzed the correlation between decremental response and survival. The mean survival time was 56.2 months (median: 53; 95%CI: 51.6–60.8) in our cohort. The median and mean survival time in patients with abnormal area decrement during 3 Hz RNS on the trapezius or ADM is shorter than that in patients with normal area decrement, but it did not reach statistical significance (Figure 3).