Electrophysiological assessment of peripheral nerves and neuro-muscular junctions in patients following self-poisoning with paraquat


 Paraquat is neurotoxic. We aimed to study the electrophysiological effects of peripheral nerves and neuromuscular junction (NMJ) in the survivors of paraquat poisoning. A cohort study was conducted on patients following paraquat poisoning. Controls were recruited. The assessments were performed around one and six weeks after the exposure. Motor nerve conduction velocity (MNCV), amplitude of compound muscle action potential (CMAP), sensory nerve conduction velocity (SNCV), F-wave studies, cardiovascular response to different stimuli, sympathetic skin response (SSR) studies, and exercise modified supramaximal slow repetitive stimulation (RNS) and electromyography (EMG) were performed. There were 28 (21 males) patients and 56 controls. The mean (SD) age of the patients and the controls were 29 (12) and 31 (11) years. Significant impairment at the first assessment in the SNCV of ulnar nerve, amplitude of ulnar nerve CMAP on distal stimulation, and F-wave occurrence in median, ulnar and tibial nerves; change of systolic blood pressure three minutes after standing and SSR amplitude ( vs controls) was observed. All parameters reverted to normal at six weeks after the exposure. There was electrophysiological evidence for somatic nerve, autonomic, and NMJ dysfunction following acute paraquat poisoning which was not seen at six weeks after the exposure.

Although paraquat is classi ed as a Class 2 moderately toxic substance by the World Health Organization, the Pesticide Action Network Asia and Paci c considers paraquat to be a Class 1 toxic substance due to its lack of an antidote, high acute toxicity, and delayed effects. Paraquat poisoning has a very high case fatality (>50%) compared to most other pesticides, and is a leading cause for death from pesticide poisoning in many countries, including Sri Lanka before it was banned in 2008 3,4 .
Paraquat causes toxicity by generating Reactive Oxygen Species (ROS) and other apoptosis-related molecules via three distinct pathways: 1-one electron reduction by the avoenzyme NADPH-cytochrome P450 reductase and a subsequent redox cycle involving superoxide dismutase (SOD) and glutathione pools, 2-Inhibition of the mitochondrial electron transport chain, and 3-interaction with enzymes such as nitric oxide synthases, thioredoxin reductase, NADPH oxidase, NADH-ubiquinone oxidoreductase, and xanthine oxidase 1,5 .
Paraquat natively exists as a divalent cation (PQ 2+ ), which undergoes redox cycling to produce the monovalent cation PQ + in the presence of cellular diaphorases such as NADPH oxidase and nitric oxide synthase 6 .
The reduced paraquat radical is immediately re-oxidized to PQ 2+ in the presence of molecular oxygen, forming a superoxide anion radical (O 2 − ). Superoxide radical is then dismutated to molecular oxygen and hydrogen peroxide by SOD. Further, there is generation of the hydroxyl free radical (HO.) in the presence of iron through the Fenton reaction. The generation of such highly reactive oxygen species induces toxicity in multiple organs such as the lung, liver, kidney, and muscles 1,5 .
However, the most important target organ for paraquat toxicity is the lung, contributed to by the transport of paraquat against a concentration gradient into the lung. However, muscle is the largest reservoir and releases paraquat slowly over many days 7 . Hence, the neuromuscular junctions (NMJ) in somatic and autonomic nerves are potentially exposed for prolonged periods.
Many other organs are also affected: centrizonal hepatic necrosis, proximal renal tubular damage, myocardial damage, and skeletal muscle damage with focal necrosis are seen in post-mortem of patients with paraquat poisoning 8 .
Although the case fatality following deliberate ingestion of paraquat is around 65%, there are reports of paraquat survivors even with large doses of paraquat ingestion [9][10][11] . Hence the morbidity of survivors is an important area to be explored.
A study conducted on patients admitted after paraquat self-poisoning showed that they had signi cantly low nerve conduction velocity. The parameters returned to normal after 6 weeks 12 . However, it remains unknown whether paraquat-induced oxidative stress could lead to peripheral demyelinating neuropathy In humans, a peripheral burning sensation (perhaps indicating neuronal involvement) is a common clinical feature that indicates a poor prognosis 13 . A progressive severe toxic myopathy has also been reported after paraquat poisoning. Microstructural changes in the extrapyramidal ganglia and hippocampus have been observed in neuroimaging studies of paraquat poisoned victims 14 .
Paraquat intoxication triggers an elevation of oxidative stress in the nerve, provoking lipid peroxidation and myelin protein aggregation leading to demyelination 12 .
However, surprisingly, although paraquat is a potential human neurotoxicant, 15 published studies exploring neurotoxic effects of paraquat in humans are limited to XX. Therefore, we aimed to explore peripheral nerve (somatic and autonomic) and NMJ function in survivors of paraquat poisoning.

Materials And Methods
A cohort study was conducted with matched controls over two years in a tertiary care hospital in the Acute neurotoxicity studies are generally done in few days to 14 days after exposure to the substance 19,20 . Studies on sub-acute toxicity were carried out up to 90 days 20,21 . We selected the midpoints of these time frames to record the data on nerve function; that is around one week and six weeks after the ingestion. Sensory and motor nerve conduction studies, F-wave studies, and electromyography were performed to assess the function of somatic nerves. Autonomic nerve function was assessed with cardiovascular re ex-based autonomic nerve function tests (heart rate response to Valsalva maneuver, heart rate response to deep breathing, heart rate response to standing, and blood pressure response to sustained handgrip) and sympathetic skin response (SSR). Exercise modi ed slow repetitive supra-maximal stimulation was used to assess the function of the neuromuscular junction.
The Neuropack MEB-9400A/K EMG/EP (Nihon Koden) system was used for electrophysiological assessments, which were carried out by the trained principal investigator (SJ EMG studies were performed on the deltoid and the rst dorsal interosseous muscle on the dominant side. A concentric needle electrode was inserted into the above muscles at rest and during contraction.
Attention was paid to spontaneous activity at rest, amplitude of the motor units, polyphasia, and the interference pattern during contraction of the muscle 22 .

Muscle power
Muscle power was assessed in exors and extensors of the wrist, biceps, triceps, dorsi exors, and plantar exors of the feet and toes, exors and extensors of the knee, exors and extensors of the hip, abductors and adductors of the thigh. The muscle power was graded according to the Medical Research Council scale for grading muscle function from 0 -5 24 .

Autonomic nerve function
Autonomic function tests were based on the cardiovascular re exes to a variety of stimuli as described by D J Ewing and B F Clarke (1982) 25 . These tests are the heart rate response to Valsalva manoeuvre, heart rate response to deep breathing, heart rate response to standing, blood pressure response to standing, and blood pressure response to sustained handgrip.
According to Neuropack S1, QP948BK user guide heart rate variability (R-R interval) based autonomic function tests were performed with reference, recording, and ground surface electrodes attached to the second intercostal space at the right border of the sternum, left anterior axillary line at the lowest rib, and right anterior axillary line at the lowest rib, respectively. The R-R intervals during the test were analysed by autonomic nervous system testing software. (Neuropack S1, QP948BK) For the blood pressure response to sustained handgrip, resting diastolic blood pressure was measured.
The subject was then instructed to squeeze the hand dynamometer as hard as possible and then to maintain 30% of maximum voluntary pressure for up to 5 minutes; diastolic pressure was recorded at the end of each minute. If the rise reached 16 mmHg within 5 minutes, it was considered normal and the procedure was discontinued. If not, the rise of diastolic blood pressure just before handgrip release at 5 minutes was recorded 25-27 .
In sympathetic skin response, the median nerve was stimulated with an intensity of 25 mA as described by Neuropack S1, QP948BK user guide. Six simulations were given with at least a 30 seconds interval between consecutive stimuli. Acquired wave forms were superimposed and recti ed to get the maximum amplitude. Maximum SSR amplitude and average SSR latency were recorded. Percentage change in amplitude between the rst and n th action potential of the third train of stimuli (C 1 -

NMJ function
Post exercise facilitation was assessed by calculating the ratio of the amplitude of the rst action potential in the second train to the amplitude of the rst action potential of the rst train (B 1 /A 1 ) 28, 30 and post exercise exhaustion by calculating the ratio of the amplitude of the rst action potential of the third train to the amplitude of the rst action potential of the rst train (C 1 /A 1 ) 28 .

Estimation of plasma paraquat
Blood was collected within 12 hours of exposure. Liquid chromatography-tandem mass spectrometry (LC-MS/MS) was used to estimate the plasma paraquat levels 31 .
The severity of paraquat poisoning was quanti ed with the Severity Index of Paraquat poisoning (SIPP) (time since ingestion in hours multiply by plasma paraquat level in µg/ml) 32 .

Statistical analysis
Data from the controls and the patients were compared with unpaired T-tests. Comparisons of the rst (at one week following exposure) and the second (at six weeks following exposure) assessment of patients used paired T-tests. F-wave in the rst assessment of the patients vs controls and the rst assessment vs the second assessment were analysed with Mann-Whitney U test and Wilcoxon Signed Ranks test respectively. Bonferroni correction was used in multiple comparisons. Correlation between SIPP score with the neurotoxicity outcomes was analysed with Spearman's correlation coe cient.

Results
A total of 99 patients with paraquat poisoning were admitted to the hospital during the study period. Forty nine patients died during the hospital stay and nineteen left against medical advice. Three patients did not consent for electrophysiological assessment. Electrophysiological assessments were carried out in 28 patients at one week (mean 9 ± 4 days -rst assessment) and 18 patients turned up for the second assessment at six weeks after the exposure. There were 56 controls.
None of the patients had co-ingested any other agrochemical.
Among 28 patients, six patients received activated charcoal, 14 patients received Fuller's earth, 12 patients received both and seven patients did not receive either of them.
Some of the patients were treated with immunosuppressants; two patients received cyclophosphamide (intravenous (IV) 1g stat, oral 250 mg twice daily for three days), dexamethasone (IV 8mg three times daily for three days followed by oral dexamethasone for a maximum of six days) and methylprednisolone (IV 1g daily for three days). Six patients received cyclophosphamide (IV 1g stat, oral 250 mg twice daily) and methylprednisolone (IV 1g daily for three days).

Somatic nerve function
Comparison of the results of somatic nerve function in the patients and the controls is shown in Table 1.
EMG was performed in 20/28 patients in the rst assessment (eight patients had painful cannula sites on the dominant hand) and 15/18 in the second assessment. None of the patients showed spontaneous activity, brillation potentials, high amplitude of the motor units, polyphasia or reduced interference pattern.

Autonomic nerve function
The results of the rst and the second assessment of autonomic function in the patients and the controls are shown in Table 2. Statistically signi cant autonomic dysfunction was found in the change of SBP three minutes after standing and SSR amplitude. All the parameters reverted to normal in the second assessment.
The test of heart rate response to Valsalva manoeuvre was performed only on 12 patients since the rest of the patients had mouth ulcers as a result of mucosal burns due to paraquat.
The blood pressure response to sustained hand grip was not able to perform in eight patients due to painful cannula sites on the hand. Only 4/20 patients were able to complete the test due to fatigue. This was signi cantly less than that of the second assessment of the controls (p<0.01), however no useful information on autonomic response was able to be derived from this test.

NMJ function
Signi cant decrement response was observed 30 seconds after an isometric exercise at the rst assessment of the patients compared to the controls ( Table 3).
None of the decrement responses was signi cant in the second assessment compared to the controls.
Maximum decrements observed in the patients at rest, just after isometric exercise, and two minutes after isometric exercise, were 28%, 20%, and 13% respectively.

Muscle power and tendon re exes
All patients and the controls showed muscle power of 5 and tendon re exes of 1 -2.
Statistically signi cant correlations were observed with SIPP and electrophysiological parameters at the rst assessment in heart rate response to standing, heart rate response to Valsalva manoeuvre and Fwave occurrence of tibial nerve with Spearman's correlation coe cient (Table 4).

Discussion
This was the rst study in humans to have measured the effects of acute paraquat poisoning on peripheral nerves (somatic and autonomic) and the NMJ. Signi cant impairment of function was found at the rst assessment in the SNCV of ulnar nerve, amplitude of ulnar nerve CMAP on distal stimulation, F-wave occurrence in median, ulnar and tibial nerves and change of systolic blood pressure three minutes after standing. The decrement response in NMJ after exercise augmentation was signi cantly greater in the patients at one week after the exposure. All these electrophysiological changes had become insigni cant at the time of the second assessment in six weeks. However, it needs to be noted that for any damage to be evident with electrophysiology, 50% of the nerve bres should be affected as unaffected bres compensate for the damaged ones 22 . Therefore, less than 50% of residual nerve injury cannot be excluded.
A study conducted in Japan showed that patients with paraquat self-poisoning had signi cantly lower nerve conduction velocity. The parameters returned to normal after 6 weeks which is compatible with the current study 12 .
SNCV and amplitude of CMAP of ulnar nerve of the patients showed signi cant reduction compared to the controls. The same ndings were not seen in the median and common peroneal nerves.
Accordingly, ulnar nerve seemed to be affected more than median and common peroneal nerves in this study. Vulnerability of nerves to toxic damage has been associated with smaller nerve diameter and increased bre length 33 . In a study that assessed median and sural nerve conduction velocities in workers exposed to phenoxy herbicides, sural nerve was signi cantly affected and it was attributed to the increased length, smaller diameter, presence of few bres and lesser degree of myelination of sural nerve compared to median nerve 33 . The ulnar nerve diameter is smaller than of median throughout its course, and signi cantly smaller than median nerve at several sites. Further, the number of fascicles was lower in ulnar nerve than median nerve 34 . Therefore, the ulnar nerve may be more vulnerable than the median or common peroneal nerves to paraquat induced damage.
In a study conducted by Hichor et al. (2017), it was revealed that mice exposed to paraquat displayed disorganization of myelin sheaths and deregulation of myelin genes. It was found that paraquat-induced oxidative stress in the nerve resulting in lipid peroxidation and myelin protein aggregation that leads to demyelination. Paraquat intoxication was found to disturb the steady-state level of myelin gene expression through alteration of positive and negative signals for myelination. They suggested that these ndings may explain the peripheral nerve defects observed in humans exposed to paraquat. Therefore, it was suggested that demyelination could be a cause of defects in conduction velocity of peripheral nerves observed in patients with paraquat intoxication 12 .
The electrophysiological ndings of demyelinating type peripheral neuropathy are slowing of conduction velocity, marked prolongation of distal latency, or both 35 . In our study we demonstrated signi cant reduction in SNCV which suggests demyelination of nerves following paraquat poisoning.
Another major mechanism of peripheral neuropathy is axonal loss 35 . Exposure to toxins is an accepted cause for axonal loss 22 . Axonal type peripheral neuropathy primarily affects amplitude of CMAP and amplitude sensory nerve action potential (SNAP) 35 . In this study we assessed only the amplitude of CMAP which showed a signi cant reduction for the ulnar nerve. Thus, this supports that paraquat poisoning may also cause axonal damage to peripheral nerves. However, amplitude changes can also occur with demyelination due to secondary axonal loss.
Our results for the F-wave occurrence showed a statistically signi cant reduction in all three nerves tested (median, ulnar and tibial) in the patients one week after exposure to paraquat compared to the controls.
F-waves are low amplitude responses produced by antidromic activation of motor neurons. F-waves are the most sensitive and reliable nerve conduction study for evaluating polyneuropathies 36 . Occurrence of F-waves re ects the excitability of motor nerves that determines the probability of a recruitment response in individual axons [36][37][38][39] . Thus the reduced occurrence of F-waves implies impairment of motor nerve excitability following acute exposure to paraquat.
The autonomic nerve assessment suggested that paraquat poisoning affects sympathetic integrity, with a signi cant reduction in SBP 3 min after standing and amplitude of SSR noted in patients one week after paraquat ingestion.. However, parasympathetic integrity was not affected.
Neuronal uptake of paraquat has not been studied but muscles have a very high uptake of paraquat 40 . However, we were not able to elicit evidence of myotoxicity. Reactive oxygen species generated in muscle would be expected to affect NMJ function 41 . The gold standard method to assess the function of NMJ is a single-bre electromyogram (SFEMG) 42 . Repetitive nerve stimulation is commonly used when facilities and expertise to do SFEMG are not available. The current study used slow (rather than fast) repetitive stimulation to minimize the discomfort to the participants, but exercise modi cation was introduced to augment the function of NMJ 43 . We observed signi cant decrement response just after the exercise but neither at rest nor few minutes after the exercise.
Animal studies suggest co-formulants of paraquat may also contribute to neurotoxicity. Commercial paraquat inhibited the amplitude of miniature end-plate potentials (m.e.p.ps) and end-plate potentials (e.p.ps) of mouse diaphragm. Interestingly pure paraquat did not have these effects. By the analysis of components in commercial paraquat, they found that paraquat by-products and the added emulsifying agent were responsible for these ndings. The study emphasized the clinical signi cance their ndings stating that neuromuscular blocking may contribute to the respiratory failure in paraquat intoxicated patient 44 .
Paraquat may also disrupt calcium homeostasis which is critical for both muscle and nerve function 45 .
Paraquat increased the basal activity of plasma membrane Ca 2+ -ATPase (PMCA), which plays a critical role in Ca 2+ homeostasis, and inhibited its sensitivity to calmodulin at low doses, while inhibiting both these components at high doses. It was suggested that PMCA is a sensitive target of oxidation in primary neurons and the inactivation of PMCA under prolonged oxidative stress could lead to altered Ca 2+ signalling 45 .
In ammatory cytokines such as interferon-γ may also contribute to neurodegenerative and neurochemical effects of paraquat 46 .
Paraquat-induced myopathy characterized by muscle necrosis and increased plasma creatine kinase has been reported in animals and humans 47,48 . However, EMG in this study did not show any evidence to support chemically induced myositis (spontaneous brillation potentials activity) or denervation ( brillation potentials, high amplitude motor units, polyphasia, and reduced interference). The reduction in amplitude of CMAP is consistent with either muscle or axonal damage 22 . The lack of any other evidence of myotoxicity may imply axonal damage.

Limitations
The major limitation was the limited number of paraquat survivors we studied. However, this is still the largest neurotoxicity study of paraquat to date as paraquat poisoning is highly lethal. Many patients refused further medical care and study involvement when they were informed about the dire prognosis and lack of any effective antidote. Electrophysiological investigations are time-consuming, cause discomfort, and are not required for the medical management of these potentially fatally poisoned patients. These di culties may explain the lack of any previous published longitudinal studies of neurotoxicity after paraquat poisoning.
The patients with clinical features of paraquat poisoning were recruited to the study irrespective of the severity of poisoning. However, most severely poisoned patients either died or discharged themselves against medical advice when it was clear there was no antidote. Therefore, neurotoxicity in severely poisoned patients was not able to be studied.
The rst assessment was not able to be performed exactly one week after exposure as patients needed to be transported to the Clinical Neuroscience Centre from the hospital and it was only possible to do this when they were clinically stable. As the effect of paraquat poisoning on the nerve and neuromuscular function is likely to be transient, it is unclear whether such a variation in the timing of conducting nerve and neuromuscular function assessment may have had impacts on the observed results.
Approximately 35% of the enrolled patients did not attend for follow-up testing despite reminders. Thus, our study can exclude clinically apparent long-term neurotoxicity but lacks the power to eliminate the possibility that there is a minor degree of persistent sub-clinical neurological damage.

Conclusions
The function of peripheral nerves and NMJ was impaired after single acute exposure to paraquat. There was evidence for both axonal and myelin sheath lesions. While long-term impairment was not demonstrated in this study with electrophysiological means, it may be worthwhile assessing the structural lesion with direct methods such as visualization of nerve bres with electron microscopy.