Overall, there are 29 patients with acute CO poisoning included and categorized into the CO group. These patients were treated for acute CO poisoning in our hospital at different time points from December 2020 to March 2021. There were 12 men and 17 women, aged between 6 and 48 years (average, 19.4 ± 12.5 years). The inclusion criteria (38) included (1) a history of exposure to high concentrations of CO; (2) symptoms and signs of acute central nervous system injuries; (3) timely blood COHb content conforming to the national diagnostic criteria. For MRI, the time interval to be exposed to high concentrations of CO was required to be smaller than 3 days. Hyperbaric oxygen and correction for electrolyte disturbance treatment were performed upon admission. Adult patients with CO poisoning were treated with hormone therapy for 3 days after admission to prevent immune inflammatory responses. The patients received 1 h hyperbaric oxygen therapy (0.2 MPa) 1–2 h after admission and then received hyperbaric oxygen therapy (0.2 MPa) once a day, 1 H each time. The duration was generally 7 days. After treatment, the patient still showed clinical symptoms and cognitive decline. Although the clinical symptoms recovered, the patient did not meet the criteria for recovery. In the meantime, each patient was subjected to a Mini-Mental State Examination (MMSE), and the period in coma and COHb concentration were recorded (Supplementary Table 1). At 1 month follow-up, onsite and telephone interviews were conducted and the Hamilton Anxiety Scale (HAMA) score was obtained (Supplementary Table 1). Healthy volunteers matched for gender, age, and educational level were recruited from our hospital as the control group (control group, n = 21), including 9 men and 12 women aged between 8 and 30 years (average, 20.7 ± 7.4 years). In this population, there was no history of cerebral injury, psychiatric disorder, alcohol abuse/substance dependence, or diseases of the nervous system (including stroke, seizure, and somatic disease). Table 1 shows the detailed demographic and clinical information. All patients gave written informed consent. The project was approved by the Ethics Committee of The Second Affiliated Hospital of Shantou University Medical College.

The structural MRI and CEST MRI data were collected using a 3.0 Tesla MRI scanner (Sigma; GE Healthcare, Milwaukee, WI, USA), using an 8-channel phased-array head coil. The sponge padding was used to limit head motion. The T2-weighted images (T2WI) [repetition time (TR) = 4,600 ms, echo time (TE) = 120 ms, 20 slices, and acquisition time: 1 min 35 s], the T2WI fluid-attenuated inversion recovery images (TR = 8,600 ms, TE = 155 ms, inv. time = 2,100 ms, 20 slices, and acquisition time: 1 min 53 s), and diffusion-weighted images (TR = 5,200 ms; TE = minimum; b = 1,000, 20 slices, and acquisition time: 42 s) were obtained to acquire the information of the brains of all subjects.

The CEST scan was based on an MT-prepared gradient echo MRI sequence with the following settings: TR = 50 ms, TE = 3.1 ms, field of view = 240 × 240 mm², matrix = 128 × 128, 1 slice, slice thickness = 5 mm, and bandwidth = 15.63 kHz. The MT saturation pulse was a Fermi pulse with a 20 ms width and a B1 of 1.95 μT (39). The CEST imaging was performed on the brain slice shown in Figure 1. The regions of interest (ROI) consisted of seven bilateral standard regions in all patients with acute CO poisoning and healthy controls which are basal ganglia, thalamus, the genu of the corpus callosum, the frontal lobe, the occipital lobe, the cuneus, and the cingulate gyrus (Figure 1). Except for the basal ganglia region and the thalamus, the ROI was placed in the gray matter region, and for other brain regions, the ROI was placed in the white matter region of the genu of the corpus callosum, the frontal lobe, the occipital lobe, the cuneus, and the cingulate gyrus. The images were interpreted independently by two experienced radiologists who were blinded to the neurological manifestations and the results of the analyses.

All CEST image processing was performed using software routines written in Matlab 7 (Mathworks, Natick, MA, USA). The acquired images were corrected for B0 inhomogeneity using a water saturation shift referencing map. The corresponding correction algorithm referred to a previous detailed discussion (40). Then, the GluCEST contrast map was generated using the following equation (41).

where S (−3 ppm) and S (+3 ppm) are the images obtained at −3 ppm and +3 ppm, respectively. The Z-spectra were obtained from the normalized CEST images. The MTRasym maps were computed using the equation (41).

Data processing and analyses were completed with SPSS 20.0 (SPSS 20.0, IBM, Armonk, NY). Comparative results, MMES score, time in coma, COHb concentration, and postoperative HAMA score were represented by mean (interquartile range). Age (continuous variable) was displayed in mean ± standard deviation. The categorical variables are shown in integers. The normal distribution of continuous parameters was determined by the Shapiro–Wilk test, followed by the Kruskal–Wallis H-test for data comparison.

The average GluCEST% in the bilateral ROI in the two groups was measured by GluCEST, and the data that did not conform to normal distribution were tested by the Mann–Whitney U-test. A comparison of the age and educational levels were completed by the two-sample t-test and that for gender using the chi-square test. A Spearman correlational analysis was performed to calculate the association of GluCEST% with the time in coma, MMSE score, COHb concentration, and postoperative HAMA score. A p < 0.05 demonstrates a statistically significant difference.

The reason for CO poisoning in coma patients was the long bathing time in an enclosed bathroom using the gas stove. There were no statistically significant differences between the patients and the healthy volunteers in terms of gender (χ² = 0.011 and P = 0.917) and age (P = 0.65, two-sample t-test). The MMSE score was lower in patients with CO poisoning (P < 0.001, two-sample t-test). In the acute phase, patients presented with symptoms, including coma, headache, dizziness, fatigue, and memory decline after a lucid interval. Upon follow-up 1 month after the discharge, the HAMA scores between the two groups were not remarkably different. There was no evident sequela, except for one who still had dizziness and memory decline. The demographic and clinical information of the subjects of the study are shown in Table 1. There were no statistical differences between the subgroups of age regarding COHb concentration, the time in coma, and the HAMA score upon follow-up.

In routine MRI, abnormal images were shown only in two adult patients, majorly bilateral globus pallidus on T2WI (Figure 2) and signal enhancement on T2flair and DWI. One of them also showed abnormal images in the white matter of the right occipital lobe. No abnormality was observed in the other patients. Higher GluCEST imaging parameters were generally observed in patients with CO poisoning vs. healthy volunteers (Figure 2). In terms of the GluCEST% in each ROI, it was higher in patients in the basal ganglia (Z = 3.253), the occipital lobe (Z = 4.142), the genu of the corpus callosum (Z = 2.323), the cingulate gyrus (Z = 1.96), and the cuneus (Z = 2.849) as compared to the healthy controls (P < 0.05, Tables 2, 3). In the subgroup analysis for age, higher GluCEST% was observed in the basal ganglia (Z = 3.956), the thalamus (Z = 2.11), the occipital lobe (Z = 3.429), the cingulate gyrus (Z = 2.506), and the cuneus (Z = 3.165) in adult patients, when compared to the healthy adult volunteers (P < 0.05, Tables 2, 3). In children, higher GluCEST% was also observed in multiple ROIs in patients, while the differences among those in healthy children were not statistically significant. Further correlational analysis revealed that the increasing GluCEST% in the thalamus and the genu of the corpus callosum in patients with CO poisoning was remarkably associated with the MMSE score (Figure 3). There was no statistical association between the GluCEST% in each ROI and the time in coma, COHb concentration, and postoperative HAMA score.

CEST is a relatively new contrast mechanism of MRI. It uses the specific MR frequency (chemical shift) at the molecular level and the standard MRI technique to generate images of good spatial resolution. Some CEST techniques, such as GluCEST, can indirectly examine molecules by combining the specificity and the spatial resolution of MRI (42, 43). GluCEST is effective in measuring the glutamate content in the brain, which is generally higher than the concentration of other metabolites in the brain (42). In the current study, changes in cranial glutamate levels were examined in the acute phase after CO poisoning to study the potential effect on cranial neurotransmitters, which was not previously reported.

In a previous animal experiment, hypoxia was induced in rats by CO inhalation, and instantly, the release of glutamate and the production of hydroxyl free radicals increased in the cerebral cortex and the hippocampus, which were considered to be the cause of ischemic brain injury after CO poisoning (30). Other studies revealed that glutamate could aggravate cellular dysfunction and cell apoptosis by activating NMDAR (6). Here, we found that patients with CO poisoning had remarkably increasing GluCEST% in the basal ganglia, the occipital lobe, the corpus callosum, the cingulate gyrus, and the cuneus, indicative of more cranial excitatory neurotransmitters and a tendency of accumulation in the above brain regions. Autopsy findings suggested that the major pathological changes after CO poisoning were the necrosis of globus pallidus and demyelination in the cerebral white matter area (44–46), which were mainly attributed to cerebral ischemia and edema. Given the hyperactivity and sufficient blood supply of neurons in the gray matter regions, such as the deep brain nucleus, the gray matter is more susceptible to poisoning-induced brain ischemia than the white matter. In the subacute phase, DWI manifestations are a hypointense signal and a high ADC value in the globus pallidus and a hyperintense signal and a low ADC value in the cerebral white matter (47). This suggests that the injury in cerebral white matter is a result of both ischemic globus pallidus and progressive demyelination changes (48). In the present study, only two patients exhibited abnormal images in MRI and no abnormality was noted in the other patients. It is tempting to assume that the GluCEST technique can show early changes in cranial neurotransmitters after hypoxia induced by CO exposure, and the changes can involve the deep brain nucleus and white matter simultaneously. It is reported that the bilateral basal ganglia, especially the globus pallidus, is the most susceptible region to CO poisoning (45). Controversially, some researchers report that the changes in the striatum after CO poisoning were independent of glutamate receptor activation induced by the increasing extracellular glutamate content (49). Tambasco et al. (10) revealed that CO poisoning leads to impaired glutamate homeostasis and significantly affects neurons by producing neurotoxins or excitotoxins and activating the pathways of neuronal death. In this context, the specific mechanism of the action of glutamate elevation in the basal ganglia requires further research. Herein, we also found that the GluCEST% was abnormally increased in the bilateral occipital lobe areas, except for the common bilateral basal ganglia. The occipital lobe is the visual center and when an injury occurs, visual impairment will be developed. A previous study found that COHb >30% could result in visually evoked abnormal potentials and that CO poisoning came with a series of clinical symptoms, including ocular discomfort, blurred vision, and visual field defects, which are not entirely consistent with the findings in ophthalmic examinations (50). This is in line with our findings. In the injured areas of the cerebral white matter, the semioval center and the white matter next to the lateral ventricle are the most susceptible, and the corpus callosum can also be affected. Research revealed that corpus callosum could present with generalized atrophy after CO poisoning (18), which was once reported to be associated with neuropsychological presentations in other diseases (51). Our results showed that the GluCEST% in the genu of the corpus callosum was statistically different between the patients and healthy volunteers. In addition, the GluCEST% in the thalamus and the genu of the corpus callosum were negatively associated with the MMSE score, which was weak. This infers that the changes in glutamate levels in the genu of corpus callosum may indicate the disease condition in patients to some extent. The thalamus is key to cognitive tasks, consciousness, and awakening. Impairment of the thalamus and the connections may cause damage to a wide range of neurological functions, which might be clinically translated into significant cognitive, physical, and psychic disorders (52). We also noted that there was no correlation with the MMSE score in the other ROIs. The possible explanation might be that the MMSE score mainly targets the directional and verbal functions. There is a certain rate of false negative score when applied for moderate to critical cognitive disorders, and it is readily affected by the speech and educational levels of the patients.

The age range of patients included in this study was large and ranged from childhood to middle age. However, there are no unified reports on the changes in craniocerebral metabolites at different stages of craniocerebral development, especially glutamate. Although the patients included in this study have a large age span, in addition to the measurement analysis of all patients with CO poisoning, adults and children are also divided into different subgroups for analysis, and different results are obtained. In the subgroup analysis for age, GluCEST% in both the adult and child patients increased when compared to the corresponding healthy control. Notably, there was no statistical significance in the difference in children. We reasoned that the varying collateral circulation compensation mechanism between adults and children (53) may be used to explain the less damage to the basal ganglia region in the early acute phase after CO exposure in children. Some previous literature suggested that patients with CO poisoning could be divided into mild, moderate, and severe according to the blood COHb value, symptoms, and coma state (54, 55). However, this classification was not well-suited to the patients we included in the study. First of all, according to clinical symptoms, the conditions of our patients were relatively consistent, whose state of consciousness was manifested as mild to moderate coma, and they recovered after rescue without obvious complications. Previous literature state that the degree of poisoning in patients can be determined according to the standard of blood COHb concentration. However, clinical symptoms of acute CO poisoning and their severity do not always correlate with the concentrations of CO-Hb on admission (8, 56–58). The COHb level in the clinical diagnosis of CO poisoning is not significant (58). Herein, we also found that there was no statistical correlation between the GluCEST% and COHb concentration after CO poisoning.

At follow-up, patients made a good recovery of mental status according to postoperative HAMA scores. There was no evidence of significant sequela, and the time in coma was independent of the GluCEST values. This indicates that timely treatment was achieved in patients after coma was induced by CO poisoning and the early increase in GluCEST values might not cause irreversible damage to the brain. There are two possible explanations. First, the time in coma lacks subject assessment. Here the value was obtained mainly by the bath time, which is changeable and objective possibly affecting the data correlations. Second, patients included in this study had different degrees of CO exposure and are not only from the most critically affected populations. Only in the cases of severe CO exposure, the patients might be more likely to develop higher GluCEST values and severe clinical symptoms. More significant associations might be present, while further validation is in demand.