Nitrous oxide‐related neurological disorders: Clinical, laboratory, neuroimaging, and electrophysiological findings

Abstract Background Recreational N2O abuse is an important etiology of neurological impairment in young patients, which may easily be ignored clinically. Few current studies have investigated the characteristics or the effects experienced by its users. We aimed to explore any correlation between the clinical severity and biomarkers and spinal magnetic resonance imaging (MRI) abnormalities, identify independent factors associated with spinal MRI abnormalities, and ascertain factors affecting depression/anxiety in patients with N2O‐related neurological disorders. Methods Patients with N2O‐related neurological disorders were enrolled retrospectively between February 2017 and July 2020. Their demographic, clinical, laboratory, neuroimaging, electrophysiological, and neuropsychological findings were analyzed. Correlation analyses were conducted using Spearman's or Pearson's correlation and linear regression analysis. Independent factors associated with spinal MRI abnormalities were identified using univariate and multivariate analyses. Results The principal clinical manifestations of N2O‐related neurological disorders (n = 63; 38 men, 25 women; mean age ± SD: 22.60 ± 4.46 years) were sensory disturbance, followed by gait disturbance and pyramidal tract damage. A significant negative correlation existed between serum vitamin B12 levels and clinical severity (r = −0.309, p = .014), which disappeared after linear regression. An interval of less than 6 months between initial N2O abuse and hospitalization was independently associated with spinal MRI abnormalities (39.47% vs. 72.00%, respectively; χ 2 = 6.40, p = .01). Thirty‐eight (60.32%) and 40 (63.49%) patients experienced anxiety and depression, respectively. Moreover, the higher the clinical scores/serum homocysteine levels, the greater the severity of anxiety/depression (r = 0.442, p < .01; r = 0.346, p < .01; r = 0.477, p < .01; r = 0.324, p < .01). Conclusions The significant inverse correlation between initial vitamin B12 levels and clinical severity could aid prognosis prediction in patients with N2O‐related neurological disorders. Spinal MRI abnormalities were not related to clinical severity but depended on the time interval between initial N2O abuse and hospitalization. Anxiety and depression were common comorbidity in these patients, and their severity increased with the intensity of clinical impairment and/or serum homocysteine levels.

depended on the time interval between initial N 2 O abuse and hospitalization. Anxiety and depression were common comorbidity in these patients, and their severity increased with the intensity of clinical impairment and/or serum homocysteine levels.

K E Y W O R D S
homocysteine, magnetic resonance imaging, nitrous oxide, subacute combined degeneration, vitamin B 12 deficiency INTRODUCTION Nitrous oxide (N 2 O), also known as laughing gas, is a stable, colorless, and sweet-tasting gas. Historically, N 2 O has played a very important role in providing general anesthesia in clinical settings, including dentistry, ambulance transport, and childbirth (Lew et al., 2018).
Recent studies have provided evidence of a sharp surge in N 2 O abuse for recreational purposes, especially among adolescents, owing to its euphoric effect, easy availability, and low cost (Bethmont et al., 2019;Oussalah et al., 2019). Epidemiological studies showed that the frequency of N 2 O abuse rose from 4% of respondents in 2013 to 16% in 2017 (Entwistle & Burns, 2013;O'Donnell & Uporova, 2017). A New South Wales survey on participants' exposure to stimulants found that 75% of respondents recently used N 2 O in 2018, which increased from 20% in 2013 (Winstock et al., 2017). Accordingly, there have been sporadic reports of the risk of subacute combined degeneration (SCD), peripheral neuropathy, and neuropsychiatric disturbances resulting from exposure to N 2 O (Garakani et al., 2016;Hew et al., 2018;Lan et al., 2019;Yuan et al., 2017).
Despite the increase in its abuse among the youth and reports of N 2 O-induced neurological complications, current studies remain limited by extremely small sample sizes and incomplete exploration of the clinical severity, biomarkers, neuroimaging, and neuropsychology in these patients (Bao et al., 2020;Zheng et al., 2020). These shortcomings account for the insufficient consideration of N 2 O abuse as an important etiology of polyneuropathy and SCD in young patients, leading to inadequate diagnosis and treatment. Therefore, we conducted a retrospective study to analyze any correlation between the relevant biomarkers, extent of spinal magnetic resonance imaging (MRI) abnormalities, Hamilton Depression/Anxiety (HAMD/HAMA) Rating Scale, and clinical severity in patients with N 2 O-related neurological disorders, and to identify factors affecting their spinal MRI presentations.

Demographic and clinical evaluation
The patients' demographic data, including sex, age of onset, history of N 2 O abuse, routes of self-administration, approximate amount used per session, intake frequency (number of time(s)/week), and time interval between first N 2 O use and hospitalization were collated. Two experienced neurologists independently determined the severity of neurological impairment based on the clinical scoring system described by Hemmer et al., which has been widely used by other researchers (Hemmer et al., 1998;Jain et al., 2014). The scoring system comprises five items: gait disturbance (0 = normal, 1 = positive Romberg's sign, 2 = impaired ambulation but able to walk unsupported, 3 = substantial support required for ambulation, and 4 = wheelchair-bound or bedridden), sensory disturbances (hypesthesia, dysesthesia, vibration/jointposition impairment; 0 = normal, 1 = impairment in toes and fingers, 2 = impairment in the ankle and wrists, and 3 = impairment in the upper arms and legs), mental impairment (0 = normal, 1 = intellectual or behavioral impairment without the need for social support, 2 = partial dependence for activities of daily living, and 3 = complete dependence for all activities of daily living), neuropathy (0 = no loss or reduction of reflex, 1 = areflexic or hyporeflexic ankle jerk, 2 = areflexic or hyporeflexic patellar jerk, and 3 = areflexic or hyporeflexic biceps jerk), and pyramidal tract signs (0 = no involvement of the pyramidal tract, 1 = positive Babinski sign, 2 = spastic paraparesis, and 3 = spastic tetraparesis). The overall score ranges from 0 to 16.
The higher the clinical score, the greater the severity of neurological impairment.

Laboratory and conventional MRI assessments
All patients underwent laboratory tests that assessed their complete blood cell count, vitamin B 12 , folate, and homocysteine levels. Hepatic, renal, and thyroid function; ion levels; blood sugar; antibodies related to rheumatism, paraneoplastic syndrome and autoimmune encephalitis; syphilis; HIV infection, and cerebrospinal fluid were analyzed as needed for the differential diagnosis.
All patients underwent conventional spinal MRI using 1.5-T/3.0-T magnetic resonance devices (Magnetom H-15 and Vision; Siemens, Erlangen, Germany) on admission. The cervical, thoracic, and/or lumbar segments of the spinal cord were scanned, depending on the patient's clinical signs and symptoms, with sagittal and axial reconstruction. Spinal T1-weighted spin-echo (repetition time/echo time: 500−550/10−15) and T2-weighted imaging (3000−4000/100−120) were performed with echo train lengths of 5. Other MRI scan parameters included a 3-mm section thickness and 1-mm scanning interval.
We recorded the extent and location of the affected spinal cord segments with MRI abnormalities.

Electrophysiological and neuropsychological examinations
Each participant underwent electrophysiological examinations of the median, ulnar, peroneal, tibial, and sural nerves, depending on his/her clinical manifestations. The amplitude, distal latency, and conduction velocity of the compound muscle action potential (CMAP) and amplitude and conduction velocity of the sensory nerve action potential (SNAP) were measured. Neuropsychological assessment was conducted in all patients. The HAMA Scale was used to evaluate the severity of anxiety symptoms, which includes 14 items that are rated from 0 to 4. The severity of depression was evaluated using the HAMD Scale, which consists of 17 items. Items 1,2,3,7,8,9,10,11,and 15 are rated from 0 to 4, while the other items are rated from 0 to 2. We defined anxiety as a HAMA score of above 6, and depression as a HAMD score of above 7.

Correlation and regression analysis
We explored any correlation between clinical severity and sensitive biomarkers, including serum red blood count (

Patient profiles
Sixty-three patients with N 2 O-related neurological disorders (38 men, 25 women) aged between 15 and 33 (mean ± SD: 22.60 ± 4.46) years were enrolled in this study. The patients' exposure to N 2 O entailed inhalation from bulbs/balloons (7.5 g/10 ml/bulb) or larger canisters (1000 ml/canister); two patients also took N 2 O in the solid form. Accurate quantification of N 2 O inhalation was challenging because users typically inhaled in groups or were unable to describe the precise volumes of the balloons. The amount of N 2 O consumed per session was 4000 (2400−7000) ml (median [IQR]), and the intake frequency was 3.33 ± 1.69 times per week. The time interval between initial N 2 O abuse to hospitalization was 6 (3−12) (median [IQR]) months.

Clinical impairments and severity
Neurological manifestations of the participants are presented in

Biomarkers and correlation with clinical severity
The changes in serum levels of RBC, hemoglobin, MCV, vitamin B 12 , folate, and homocysteine of the study population are presented in

Conventional MRI findings
All

Electromyographical and neuropsychological examinations
Fifty-one (80.95%) patients in this study presented with typical mixed axonal and demyelinating neuropathy. Reduced amplitude, prolonged distal latency, and slow conduction velocity of CMAP were observed most frequently in the peroneal and tibial nerves (n = 47, 74.60%), while the proportion of patients with reduced amplitude, prolonged distal latency, and slow conduction velocity of SNAP in the

DISCUSSION
Recent case studies have reported that N 2 O abuse can result in several neurological and psychiatric disorders, such as SCD, myelopathy, demyelinating polyneuropathy, and even death (Backstrom et al., 2015;Chen et al., 2018;Choi et al., 2019;Edigin et al., 2019;Neveu et al., 2019). Fortunately, these conditions are treatable, and full clinical recovery is also possible if early detection and adequate treatment are implemented (Green et al., 2017;Jain et al., 2014). However, the adverse effects of N 2 O exposure are often ignored clinically, despite its widespread abuse amongst the youth and the surge in the number of reported N 2 O-induced neurological complications (Kaar et al., 2016). is, an interval of less than 6 months between the first N 2 O abuse and hospitalization, were susceptible to spinal cord abnormalities on MRI. Finally, anxiety/depression is more common in these patients than generally believed, and the higher the clinical severity scores or serum homocysteine levels, the greater the degree of anxiety or depression.
Each of the 63 patients included in the present study complained of sensory disturbances, including hypesthesia, dysesthesia, vibration, and impaired proprioception, followed by gait disturbance, which was consistent with previously published studies (Patel et al., 2018;Zheng et al., 2020). Moreover, most instances of sensory disturbance occurred bilaterally in the upper arms and legs. This corresponded to the electromyographical results showing that sensorimotor neuropathy was common in patients with N 2 O-related neurological disorders and more prominent in the lower extremities. Our findings concurred with the results of Zheng et al. (2020) and Keddie et al. (2018), that is, altered sensation in the extremities was the predominant presenting feature of N 2 O-related neurological disorders.
Vitamin B 12 deficiency is a well-established cause of neurological disorders in elderly individuals with malabsorption syndromes and inadequate intake or bioavailability of vitamin B 12 (Green et al., 2017), and its incidence generally increases after the age of 60 years (Allen, 2009;Hunt et al., 2014). Clinically, the classical hematological or neurological manifestations of B 12 deficiency are relatively uncommon because the substantial hepatic stores of vitamin B 12 can delay clinical manifestations for up to 10 years after the onset of deficiency (Carmel, 2000). This is why patients with vitamin B 12 deficiency are likely to be elderly individuals. The present study demonstrated that N 2 O-related neurological disorders were more prevalent among young adults and even adolescents, and the median interval between the first N 2 O abuse and hospitalization was only 6 months, which differs from the established knowledge of patients with conventional B12 defi-  (Hathout & El-Saden, 2011), resulting in two significant dysfunctions. First, the oxidized cobalt cation inhibits the action of cobalamin, a coenzyme for methionine synthase, and further impairs the generation of tetrahydrofolate (THF) and methionine from homocysteine and 5-methyl-tetrahydrofolate (5-methyl-THF) (Green et al., 2017;Hunt et al., 2014). THF is the precursor to thymidine monophosphate (TMP) required for DNA synthesis. Therefore, the failure of this pathway can result in megaloblastic anemia due to the interference in DNA synthesis and blood cell turnover (Nunn, 1987). Second, vitamin B 12 acts as a cofactor in the conversion of methylmalonyl-CoA to succinyl-CoA (Green et al., 2017;Hunt et al., 2014). Methylmalonyl-CoA aggregation occurs due to vitamin B 12 deficiency, which in turn results in methylmalonic acid (MMA) accumulation, thus disrupting lipid synthesis and causing neuronal demyelination (Richardson, 2010). Hence, elevated serum levels of homocysteine and/or MMA serve as better biomarkers than decreased serum vitamin B 12 for the precise and timely detection of cellular vitamin B 12 deficiency, consistent with the findings of previous studies (Bao et al., 2020;Buizert et al., 2017;Lan et al., 2019). However, we did not find any linear correlation between the serum homocysteine levels and severity of impairment. On the contrary, a significant negative correlation was observed between the initial serum vitamin B 12 level and clinical severity. Although this relationship was not reflected on linear regression, this correlation has not been described before in patients with N 2 Orelated neurological disorders, which suggests that serum vitamin B 12 levels could aid in the prediction of the severity of clinical impairment and even prognosis.
In this study, MRI depicted spinal cord abnormalities in 38 (60.31%) patients, with relatively low sensitivity. Symmetrical T2-signal abnormalities involving the posterior column of the spinal cord with an inverted V-sign were predominant and relatively specific manifestations in these patients, consistent with previous studies (Bao et al., 2020;Johnson et al., 2018;Keddie et al., 2018;Patel et al., 2018), which tended to involve the cervical or cervico-thoracic spinal cord (Lim, 2011;Zheng et al., 2020). This is also in accordance with the characteristics of SCD of the spinal cord, that is, swelling of the myelin sheath and patchy myelopathic spongy vacuolation (Powers, 1996). As noted above, vitamin B 12 is essential for the maintenance of the myelin sheath; thus, N 2 O-induced vitamin B 12 deficiency can account for the progressive demyelination and even axonal loss. Moreover, the highest density of myelinated fibers in the fasciculus gracilis in the cervical area can explain the susceptibility of the cervical spinal cord to N 2 O neurotoxicity (Ohnishi et al., 1976). Interestingly, our study also found that patients with an interval of less than 6 months between the first Increased homocysteine can reduce the synthesis of catecholaminergic and non-catecholaminergic neurotransmitters related to Sadenosylmethionine that contribute to depression (Zaric et al., 2019).
Excess homocysteine also leads to the production of homocysteic acid and cysteine sulfinic acid, which exert excitotoxic effects on the Nmethyl-D-aspartate receptor and neurotoxic effects on dopaminergic neurons (Bhatia & Singh, 2015). Current research on the relationship between anxiety and homocysteine is sparse compared to that between homocysteine and depression (Coplan et al., 2015). One possible mechanism underlying the link between homocysteine and anxiety may be the oxidative status of the brain. A recent animal study indicated that hyperhomocysteinemia induced by dietary overload of methionine increases anxiety-related behaviors in rats, and the proanxiogenic effects could have resulted from oxidative stress in the rat brain (Hrncic et al., 2016).
Our study has some limitations. First, this single-center retrospective study used consecutively collected data. Therefore, our findings may not be applicable to other settings due to the inherent selection bias. Second, it was difficult to identify the exact amount of N 2 O abused because patients usually inhale it in a group, or they cannot describe the precise volumes contained within the balloons. Thus, we could not analyze the relationship between the inhalation dose and clinical severity and MRI/laboratory findings. Finally, MMA evaluation was not conducted in this study, and concurrent drug/toxin/vitamin-related diseases were also not incorporated in the analysis.

CONCLUSION
Recreational N 2 O abuse that resembles the symptoms of vitamin B 12 deficiency is more common in young and otherwise healthy patients than previously believed. The negative correlation between serum vitamin B 12 levels with clinical severity suggests that initial serum vitamin B 12 level may help in determining the prognosis of these patients. Conventional

CONFLICT OF INTEREST
The authors declare no conflict of interest.

DATA AVAILABILITY STATEMENT
The anonymized data that support the findings of this study are available from the corresponding authors upon reasonable request.