Nitrous‐oxide‐induced polyneuropathy and subacute combined degeneration of the spine: clinical and diagnostic characteristics in 70 patients, with focus on electrodiagnostic studies

Abstract Background and purpose Nitrous oxide (N2O) induced neurological symptoms are increasingly encountered. Our aim is to provide clinical and diagnostic characteristics with a focus on electrodiagnostic studies. Methods Patients with neurological sequelae due to N2O presenting in our hospital between November 2018 and December 2021 reporting clinical and diagnostic data were retrospectively reviewed. Results Seventy patients (median 22 years) were included. Median N2O usage was 4 kg/week during 12 months. Patients’ history revealed a higher rate of sensory symptoms compared to motor (97% vs. 57%) and 77% walking difficulties. Clinical diagnosis was polyneuropathy (PNP) in 44%, subacute combined degeneration (SCD) of the spine in 19%, both in 37%. Median vitamin B12 level was low (159 pmol/L), normal in 16%. The median methylmalonic acid was increased (2.66 μmol/L). Electrodiagnostic abnormalities were observed in 91%, with 72% fulfilling axonal PNP criteria, 20% showing mild to intermediate slowing. One patient fulfilled demyelinating PNP criteria not related to N2O abuse (Charcot−Marie−Tooth type 1a). More prominent motor nerve conduction abnormalities were found; lower limbs were more affected. In 64% with normal conduction, myography showed signs of axonal loss. Magnetic resonance imaging showed cervical myelopathy in 58% involving generally five to six segments. Conclusions Nitrous oxide (N2O) leads to neurological symptoms by causing PNP and/or SCD primarily involving the legs. Distinguishing PNP and SCD clinically was shown to be insufficient. Electrodiagnostic studies showed axonal PNP. Demyelinating PNP due to N2O abuse was not present in our cohort. Therefore, further diagnostic work‐up is warranted if demyelinating features are present.

Still, little is known about the clinical characteristics of N 2 Oinduced PNP and/or SCD and the correlation between clinical presentation and outcome of diagnostic examinations.In addition, only few data are published about electrophysiological features of N 2 Oinduced PNP.Therefore, the aim of this study was to evaluate these characteristics in a large cohort of patients with special focus on electrodiagnostic studies including myography.

ME THODS
A retrospective analysis was performed of patients (≥18 years) who

Electrodiagnosis (EDx) was performed according to the local
EDx protocol.This protocol consisted of motor nerve conduction (median/abductor pollicis brevis muscle, peroneal/extensor digitorum brevis muscle and tibial/abductor hallucis muscle, including F waves) and antidromic sensory nerve conduction (median/index or middle finger, ulnar/little finger, superficial peroneal and sural nerve), Hoffmann reflex (tibial/soleus muscle) and myography (tibialis anterior muscle and/or interosseous dorsalis pedis muscle).The limbs of patients were not actively warmed up and values were not corrected for temperature.All electrophysiological measurements were compared to locally used nerve conduction study (NCS) reference values [26].Axonal PNP was defined as reduced sensory and/ or motor amplitudes (<80% of lower limit of normal) in at least two nerves, without demyelinating features [19,26].Myography was not part of the criteria for axonal PNP.PNP was considered demyelinating if demyelinating features were present in at least two nerves, as specified by the European Academy of Neurology and Peripheral Nerve Society (EAN/PNS) chronic inflammatory demyelinating polyneuropathy electrodiagnostic criteria [27].These criteria can be found in Data S1.
Patients were clinically classified in one of the following categories, based on history taking and neurological examination: (i) PNP, (ii) SCD of the spinal cord and (iii) combination of PNP and SCD.
Clinical PNP was defined according to the American Association of Electrodiagnostic Medicine definition for clinical research [28], with neuropathic symptoms (sensory and/or motor) together with normal to absent reflexes and/or decreased distal sensation and/or distal muscle weakness.Patients were concluded to have clinical SCD if at least one of the following signs was present: abnormal gnostic sensibility with normal vital sensibility, abnormal Romberg's sign, sensory level, positive Lhermitte sign, hyperreflexia, Babinski sign and/or spasticity [29].In the case of clinical features of both, this was concluded to be a combination of PNP and SCD.
All data on demographic and clinical presentation, laboratory tests, MRI data and EDx findings are presented as median and range.
All analyses were performed using SPSS (version 24, Armonk, NY, USA; IBM Corp.).Intergroup comparisons were made with the oneway ANOVA test in the case of a numerical variable and the chisquared test in the case of a categorical variable.p values of less than 0.05 were considered to be statistically significant.The study protocol was approved by the regional ethics committee (METC ZuidHolland, number G20.112).

Demographic baseline
In total 70 patients were included, 45 male (64%), 55 non-Caucasian (79%).The median age was 22 years (range 18-50).The average use of N 2 O was 4.0 kg/per week (range 0.1-56), with a median usage duration of 12 months (range 1-96).The median onset of symptoms was 35 days (range 1-552) before presentation at the hospital (PNP group 52.5 days; SCD group 15 days).All of these variables were not statistically different between groups.Admission at the ward of the hospital was necessary in 20 patients (30%) due to the severity of symptoms (Table 1).Twenty-eight patients (40%) were referred to a rehabilitation doctor.

Follow-up
All patients were advised to stop inhaling N 2 O and were started on either intramuscular injection or oral hydroxocobalamin according to local protocol.The median follow-up was 12 weeks (range 0-56 weeks).Eighteen patients (26%) did not show up at their followup appointment at around 3 months.Of the remaining 52 patients (74% of the total cohort) with at least a follow-up at 3 months 16 (31%) reported full recovery, and 26 (50%) reported only mild symptoms (mild paresthesia or hypoesthesia not interfering with daily activity).Ten patients (19%) still experienced invalidating symptoms: nine of them experienced foot drop with difficulty walking and one patient was unable to work due to cognitive problems related to N 2 O abuse.In total 28 patients (40%) were lost to follow-up since another 10 patients (14%) were lost to follow-up after the appointment at 3 months (either at the neurologist or rehabilitation physician).

Clinically isolated PNP
Thirty-one patients (31 patients) were clinically diagnosed with isolated PNP.According to our clinical definition of PNP all patients reported sensory complaints of which the majority were in both legs and arms (77%).Four patients (13%) had a positive Romberg sign of which one had a myelopathy with normal EDx and two had a normal MRI but abnormal EDx.Weakness was reported in 14 patients (45%) with objective weakness in eight (26%) on neurological examination (median Medical Research Council sum score of 60, range 42-60).Seventeen patients (55%) reported isolated sensory symptoms.Hyporeflexia was seen in 22 patients (71%).Twenty patients (65%) reported difficulty walking, of whom one patient was found on the ground with impaired consciousness and inability to stand up.Four patients (13%) experienced difficulties with micturition (inability to completely empty the bladder) of whom all had a normal MRI of the cervical spine with an abnormal EDx.Three (10%) suffered obstipation.

Clinically isolated SCD of the spinal cord
Thirteen patients (19%) were clinically diagnosed as SCD.According to our clinical definition of SCD, six patients (46%) had pathological reflexes (either Babinski sign and/or hyperreflexia), six patients (46%) had a sensory level (all with a thoracic level between the third and twelfth thoracic), seven patients (54%) had severe deficit in proprioception and three patients (23%) reported Lhermitte sign.Eight patients (62%) showed a combination of these signs.Seven patients (54%) had a positive Romberg.

Combined PNP and SCD of the spinal cord
A combination of PNP and SCD was clinically diagnosed in 26 patients (37%).The majority of reported symptoms were sensory signs (25  showed a myelopathy.
See Table 2 for a detailed overview of all clinical characteristics.

Laboratory investigations
The

Magnetic resonance imaging features
Magnetic resonance imaging of the cervical spine was performed for 52 patients (74%) with abnormalities in 30 (58%).All showed T2weighted signal enhancement affecting the dorsal columns of the cervical spine including edema of the cervical spine in four (13%).
The majority (21 patients, 70%) had a cervical myelopathy involving five or six segments.One patient had a cervical and thoracic myelopathy going as far as the tenth thoracic.In 4/17 patients (24%) with a clinically isolated PNP the MRI showed a myelopathy.In clinical signs of SCD (with or without PNP) MRI of the spine was normal in 9/35 patients (26%).

Electrophysiological findings
Electrodiagnosis was performed in 57/70 patients (81%).The median time between first presentation and EDx was 19 days (range 1-120 days) with a median of 71 days between first symptoms and EDx (range 2-577).Figure 1 gives an overview of the EDx results.
Motor NCS

Sensory NCS
Table 4 shows the sensory NCS results of 57 patients.The amount of sensory nerves tested per patient varied (median 3, range 0-4).
In the legs sensory nerve action potential (SNAP) amplitudes were often reduced or absent (61% of the superficial peroneal nerves and 38% of the sural nerves).In 62% of sural nerves amplitudes were normal (reference value >4 μV), with amplitudes >10 μV in 39.6% (21/53).In 22.6% (12/53) sural amplitudes were between 4 and 10 μV.Normal SNAP amplitudes of the arm nerves were seen in the majority of patients (78% of the median and 83% of the ulnar nerves).
In 43% (24/57 patients) the sensory NCS was normal in arms and legs.In 39% (22/57 patients) abnormal sensory NCS was confined to the nerves of the lower extremities.In only one patient sensory NCS was abnormal with normal motor NCS and myography.This patient with a proven VB12 deficiency also reported drinking 6 alcohol units daily.

Myography
Myography was performed in 96% (55 patients), with tibial anterior muscle tested in all patients (100%) and first dorsal interosseous muscle of the foot in 93% (51/55).All patients with abnormal motor conduction showed signs of denervation.No fasciculation potentials were observed.Signs of re-innervation were seen in 16% (9/55 patients); all showed polyphasic motor unit potentials with or without prolonged duration.In 24% (13/55 patients) severely reduced recruitment of motor units was seen in the tibial anterior muscle, with increased firing rate on the remaining motor units.In 64% (9/14 patients) with normal sensorimotor conduction, myography showed signs of denervation (57%) and/or re-innervation (7%).
In 24% (10/42) dCMAP amplitude of the peroneal nerve and/or tibial nerve were (close to) absent (0%-10% of the lower limit of normal) with relatively spared sensory amplitudes of sural and/or superficial peroneal nerves (median amplitudes 81% of the lower limit of normal).

DISCUSS ION
A large retrospective cohort of 70 patients with N 2 O-induced PNP and/or myelopathy is presented.This study accurately describes the clinical and diagnostic features including MRI of the spine, extensive EDx data with myography performed in the majority of patients.
Our cohort, in concordance with the literature, involved a young aged population with a male, non-Caucasian predominance [5,6,14,16,17,19,20,23,30].There was a wide range of duration and total N 2 O intake before presentation, which has been published previously [17,20,23,31,32].Most patients presented with a delay of at least 1 month after onset of symptoms.Data on prognosis in the literature are often insufficient because either follow-up is not conducted [25] or there is a high rate of loss to follow-up [23,24].For example, Yu et al. reported 54% loss to follow-up in their cohort study of 110 patients [23].On average the prognosis is found to be good with percentages of recovery varying between 67% and 95% [12,23,24].In our study, a large proportion of patients (81%) improved completely or with mild residual symptoms.This highlights the importance of education and VB12 supplementation [6] and prompt abstinence of N 2 O abuse [8,11,14,17,23,24,32].
Clinically, the majority of patients presented with sensory more than motor complaints, with legs more involved than arms and with walking difficulties [5,6,8,14,16,20,24,31].Micturition problems were reported in 20%, in patients with both proven PNP and/or SCD.
It is hypothesized that micturition problems in patients with N 2 O abuse can be caused by PNP with autonomic involvement or due to spinal cord damage [16].Patients were clinically diagnosed with isolated PNP in 44% (31 patients), isolated SCD in 19% (13 patients) and combined PNP and SCD in 37% (26 patients).However, our results showed that clinical discrimination between PNP and SCD is insufficient, because patients with clinically isolated SCD showed EDx abnormalities in 80% and in cases with clinically isolated PNP 24% of MRI revealed a myelopathy.This underlines the importance of further investigation by EDx and MRI of the (cervical) spine when neurological sequelae due to N 2 O are suspected.
On MRI the dorsal column of the cervical spine is known to be most prone for N 2 O damage and in a smaller proportion also the thoracic spine [6,8,22].In our cohort MRI of the cervical spine showed a myelopathy in 58% in, on average, five to six segments which is also reported in the literature [24,34].Normal MRI was observed in 26% of cases with clinical signs of myelopathy.The latter might be explained by supplementation of VB12 prior to MRI [35].
Electrodiagnosis abnormalities were present in 91% of our patients (52/57) with or without clinical symptoms of PNP.All but one (CMT1a) classified as an axonal PNP [5,12,19].Although 20% showed mild to intermediate conduction slowing, the EDx demyelinating criteria were not met.This mild conduction slowing might well be explained by axonal loss and/or low limb temperature (<30°C) in 80%.Pathology studies showed primarily axonal damage with chronic axonal damage with secondary segmental demyelinating in late stages at sural nerve biopsy in both VB12 PNP and specifically in N 2 O-induced PNP [22,31].Pure demyelinating PNP in N 2 O abuse has been described in up to 30% of cases in the literature [9,12,19,20,23].Despite this, in our cohort demyelinating PNP due to N 2 O abuse was not present.The discrepancy is likely to be explained by the use of different, less strict, demyelination EDx criteria [12,19,23] or these criteria were not specified [5,6,18,20,24].
Motor EDx abnormalities were more prominent than sensory abnormalities in 62% of patients with EDx-proven PNP.Seventeen As reference values for sensory NCSs are known to be quite variable, this could also partly contribute to the deviation between motor and sensory NCSs.Therefore, normal sensory amplitudes according to strict reference values might be an underestimation of abnormal sensory NCSs.This was also suggested by Lee et al.Our study has limitations.This is a retrospective study with clinical and diagnostic data not performed in a standardized manner.Despite this, the majority of patients underwent both EDx and MRI of the spine.It is uncertain whether the reported amount and duration of N 2 O use is correct and therefore no correlation could be made with the severity of symptoms and EDx changes.Unfortunately, it is well known that in the N 2 O-abuse population there is a large proportion lost to follow-up, also seen in our cohort (40%) [14,23,24].
Therefore, data about ongoing N 2 O abuse were insufficient.N 2 O is increasingly recognized as an addictive substance and this might explain the loss to follow-up [39].

CON CLUS ION
visited our large urban teaching hospital between November 2018 and December 2021 diagnosed with N 2 O-induced PNP and/or SCD, based on clinical presentation in combination with low functional VB12 levels and/or the patients acknowledged using N 2 O and by exclusion of other causes.Patients were identified with the use of our electronic patient database from the Department of Neurology at the HagaZiekenhuis, The Netherlands.Clinical data, including demographics and detailed information on the amount (average number of 2 kilograms N 2 O tanks/week, with one balloon estimated to contain 8 gram N 2 O) and frequency of N 2 O use, were collected.Standardized laboratory testing included VB12 and methylmalonic acid (MMA).Levels of VB12 were divided into normal (>250 pmol/L), low end of normal (160-250 pmol/L) and low (<160 pmol/L).In the case of symptoms suggesting central nerve damage a magnetic resonance imaging (MRI) of the (cervical) spine was performed according to our local protocol (including sagittal T1/T2 and short tau inversion recovery together with an axial T2 weighted sequence).All MRI scans were reviewed by a neuroradiologist.

[ 36 ]
. Further research is warranted to clarify the more prominent motor NCS abnormalities found in N 2 O-abuse patients.Follow-up EDx studies might help to partially overcome this problem of sensory reference values, as ongoing sensory axonal degeneration will give rise to further decrement of sensory amplitudes and eventually become abnormal.

Table 3
shows the motor NCS results.Motor NCSs were performed in 55 patients.The median amount of motor nerves tested per patient was 3 (range 0-3).In 22% (12/55 patients) motor NCSs were normal.The peroneal nerve showed a severely reduced distal compound muscle action potential (dCMAP) amplitude or was absent in 65% (36/55) and in 67% (36/54) for the tibial nerve.The median nerve was less often involved; the dCMAP amplitude was severely reduced in 26% (14/53) for the median nerve.All these patients also had severely reduced or absent dCMAP amplitudes in the peroneal and tibial nerves.
E 1 Neurological sequelae due to N 2 O predominate in a young, male and non-Caucasian population.VB12 levels can be normal, for example due to self-initiated supplementation; therefore MMA and/or HC should be tested.The clinical distinction between PNP and SCD is often hampered by subclinical involvement of the peripheral nerves or dorsal columns, apparent from EDx and MRI studies.N 2 O abuse causes an axonal PNP, with more prominent motor than sensory abnormalities and in some patients abnormalities on myography with normal sensorimotor NCS studies.A demyelinating PNP features a red flag and should promptly point towards other causes.