Hippocampi and midbrain lesions in carbon monoxide intoxication: a rare case report

Abstract

Background

Carbon monoxide (CO) intoxication can lead to various brain lesions, with the globus pallidus being the most common site of injury. Herein, we report a rare case of CO poisoning with acute bilateral hippocampal lesions and delayed midbrain involvement of the substantia nigra.

Case presentation

A woman in her twenties who attempted suicide by burning charcoal briquettes presented with a Glasgow Coma Scale score of 3/15. Initial head computed tomography revealed low-density areas in the bilateral globus pallidus and hippocampi. Magnetic resonance imaging (MRI) on day 3 showed high signal intensity in the same regions on diffusion-weighted imaging (DWI), with a corresponding low signal on the apparent diffusion coefficient map and high signal intensity on fluid-attenuated inversion recovery imaging. The patient underwent hyperbaric oxygen therapy (HBOT) and gradually regained consciousness. However, the patient experienced persistent short-term memory loss. Follow-up MRI on day 23 showed improvement in the hippocampal and globus pallidus lesions but revealed new bilateral high-signal lesions in the substantia nigra on DWI. Despite these findings, the patient did not exhibit any extrapyramidal signs. Subsequent HBOT sessions led to further improvements in her cognitive function, as evidenced by an increase in her Mini-Mental State Examination score from 13/30 to 27/30.

Conclusions

This case highlights the importance of serial neuroimaging in CO poisoning, as delayed midbrain lesions may occur more frequently than previously thought, even in the absence of overt neurological symptoms. The patient’s cognitive recovery and lack of parkinsonism suggests that early intervention with HBOT may help mitigate the long-term consequences of CO-induced brain injury.

Introduction

Carbon monoxide (CO) intoxication can lead to various degrees of brain injury. Neuroimaging plays a crucial role in evaluating the extent and progression of brain injury. Among the available modalities, magnetic resonance imaging (MRI), particularly diffusion-weighted imaging (DWI), is highly sensitive for detecting early brain lesions associated with CO poisoning [1].

Classic radiological findings most commonly include basal ganglia structures, such as the globus pallidus, putamen, and caudate nucleus. Thalamus, periventricular and subcortical white matter, corpus callosum, cerebral cortex, and hippocampus are also typically affected by hypoxic-ischemic injury [2, 3]. These findings often correlate with clinical severity and neurological outcomes [4].

In contrast, midbrain involvement, especially in the substantia nigra, has rarely been observed [3, 5, 6], and its radiological and clinical implications remain underexplored. When present, these lesions are typically associated with profound neurological dysfunction or poor prognosis.

Here, we report a rare case of CO intoxication with acute hippocampal lesions, followed by delayed bilateral lesions in the substantia nigra on follow-up MRI. Memory function associated with the hippocampal lesion improved, and despite the midbrain lesion, the patient did not exhibit extrapyramidal signs. This case highlights the importance of serial neuroimaging in CO poisoning and expands the spectrum of radiological findings, suggesting that hippocampal lesions resulting in cognitive impairment could be plastic and that delayed midbrain lesions may occur more frequently than previously thought, albeit in a subclinical form.

Case presentation

A previously healthy woman in her twenties attempted suicide by burning charcoal briquettes and was discovered by emergency medical services. The duration of carbon monoxide (CO) exposure was unknown. On emergency medical service arrival, she had a Glasgow Coma Scale (GCS) score of 3/15, spontaneous respiration, and systolic blood pressure below 60 mmHg. After fluid resuscitation, the patient’s blood pressure improved to 140/90 mmHg. She was intubated and administered 100% oxygen because of her comatose state and unstable vital signs. Her initial arterial carboxyhemoglobin (CO-Hb) level was 17.5%.

Computed tomography (CT) of the head revealed bilateral low-density areas in the globus pallidus and hippocampus (Fig. 1). Brain MRI on day 3 showed high signal intensity in the same regions on DWI (Fig. 2, top), corresponding low values on the apparent diffusion coefficient (ADC) map (Fig. 2, middle), and high signal intensity on fluid-attenuated inversion recovery (FLAIR) imaging (Fig. 2, bottom). Cerebrospinal fluid (CSF) analysis on day 3 revealed markedly elevated levels of myelin basic protein (MBP) (> 500 pg/mL), neuron-specific enolase (NSE) levels of 135 ng/mL, and a CSF opening pressure of 30 cmH₂O.

Fig. 1.

Head computed tomography scan on admission. Low-density areas were observed bilaterally in the globus pallidus (left panel) and hippocampus (right panel), suggesting early hypoxic injury

Fig. 2.

Brain MRI on hospital day 3. The DWI (top row) shows high signal intensity in the globus pallidus and hippocampi. The ADC map (middle row) shows the corresponding low values in the same regions FLAIR imaging (bottom row) also shows high signal intensity in a larger area around the globus pallidus and hippocampi. MRI, magnetic resonance imaging; DWI, diffusion-weighted imaging; ADC, apparent diffusion coefficient; FLAIR, fluid-attenuated inversion recovery

Given the severity of CO intoxication, hyperbaric oxygen therapy (HBOT) was initiated with 2.8 atmospheres absolute (ATA) for 60 min on three consecutive days, followed by 2.0 ATA for 60 min on day 4. The patient’s consciousness gradually improved and she was successfully extubated.

Prominent short-term memory loss persisted. A follow-up MRI was performed on day 23, which showed improvement in the previously noted bilateral hippocampal and globus pallidus lesions with no white matter changes suggestive of delayed neurological sequelae (DNS). However, new bilateral DWI high-signal lesions with corresponding mildly decreased ADC values were observed in the substantia nigra of the midbrain (Fig. 3). Therefore, additional HBOT was administered on day 30 owing to persistent memory loss and residual MRI abnormalities.

Fig. 3.

Brain MRI on hospital day 23. DWI (top row) shows a newly developed bilateral high signal intensity in the substantia nigra. The ADC map (middle row) reveals the corresponding low values. FLAIR imaging (bottom row) demonstrates high signal intensity in the substantia nigra that remained, while the high signal intensity in the hippocampi has diminished. MRI, magnetic resonance imaging; DWI, diffusion-weighted imaging; ADC, apparent diffusion coefficient; FLAIR, fluid-attenuated inversion recovery

On day 37, the DWI abnormalities in the substantia nigra improved, and the FLAIR signal intensity in both the hippocampus and midbrain decreased (Fig. 4). The ADC values in the substantia nigra also returned toward normal, indicating partial reversibility of the diffusion abnormality. Along a total of additional 30 HBOT sessions (2.0 ATA for 60 min), her Mini-Mental State Examination (MMSE) score improved from 13/30 to 27/30 during the course of her recovery (Fig. 5). Around day 60, both MMSE and Hasegawa Dementia Scale-Revised (HDS-R) scores showed a transient decline compared with the previous assessment; however, they improved again on subsequent follow-up. At that time, the patient complained of fatigue and exhibited an attention deficit, which may have negatively affected her test performance. The patient was transferred to an affiliated rehabilitation hospital on day 133. During hospitalization and at outpatient follow-up, serial neurological examinations were performed by a board-certified neurologist. These examinations revealed none of masked face, bradykinesia, muscle rigidity, small steppage gait, postural instability, nor resting tremor. Thus, no clinical signs compatible with parkinsonism were observed at any time.

Fig. 4.

MRI on hospital day 37. DWI (top row) shows decreased high signal intensity in the substantia nigra compared to that on day 23. FLAIR imaging (bottom row) diminishes high intensity in the hippocampi and its atrophy. MRI, magnetic resonance imaging; DWI, diffusion-weighted imaging; FLAIR, fluid-attenuated inversion recovery

Fig. 5.

The clinical course of cognitive function scores and HBOT sessions. Time course of Mini-Mental State Examination (MMSE) and Revised Hasegawa Dementia Scale (HDS-R) scores in relation to the HBOT sessions. Circle: HBOT; yellow square: MRI scan. Cognitive function scores gradually improved over the course of hospitalization, particularly after serial HBOT sessions, suggesting a plausible therapeutic contribution of HBOT to cognitive recovery. HBOT, hyperbaric oxygen therapy; MRI, magnetic resonance imaging

Discussion

CO intoxication is a common method of suicide in East and Southeast Asia [7]. MRI plays a crucial role in detecting the brain lesions associated with CO poisoning. Although globus pallidus involvement is typical [3], cases involving the hippocampus are rare [1, 4]. Furthermore, lesions in the midbrain have been reported only in a few severe cases [3, 5, 6]. The globus pallidus is particularly susceptible to hypoxic-ischemic injury, possibly owing to its limited collateral blood supply. Moreover, this region contains a high concentration of iron, which may facilitate the direct binding of CO to heme-containing enzymes, further enhancing its vulnerability [8]. Notably, the substantia nigra also contains abundant iron, and concurrent involvement of both the globus pallidus and substantia nigra has been described in the literature as the so-called “pallidoreticular pattern” of brain injury [5, 6, 8]. This pattern has been associated with the development of parkinsonism following CO intoxication.

In our case, the initial head CT on admission demonstrated bilateral low-density lesions in the hippocampi, suggesting early hypoxic injury. Multiple HBOT sessions were performed during the acute phase by reference to Weaver et al. [9]. Subsequently, MRI performed on day 3 showed a high signal intensity in the hippocampi on DWI, consistent with restricted diffusion, and elevated signal on FLAIR imaging, indicating cytotoxic edema in the acute phase. CSF analysis on day 3 revealed markedly elevated levels of MBP [10, 11] and NSE, reflecting extensive demyelination and neuronal injury. The opening pressure was mildly elevated, which could be attributed to transient intracranial hypertension. These findings, combined with early imaging abnormalities, indicate that significant central nervous system injury had already occurred in the acute phase, although later MRI changes suggested that a considerable component of this injury was at least partially reversible. Follow-up MRI on day 23 revealed bilateral high-signal intensity in the substantia nigra on DWI, which appeared well beyond the acute phase. In our case, the diffusion abnormality in the substantia nigra was accompanied by only mildly decreased ADC values that later normalized, suggesting transient and at least partially reversible injury (such as edema and/or demyelination) rather than extensive irreversible neuronal loss. Although we cannot completely exclude the possibility that part of these diffusion changes represented non-specific or non-pathological components, the new, symmetric DWI hyperintensities with corresponding mildly decreased ADC values and their evolution on FLAIR imaging are more consistent with transient subclinical injury than with pure artifacts. Notably, these lesions were not observed on the initial imaging, and their delayed appearance underscores the importance of serial imaging [12] in capturing the evolving neuropathology of CO poisoning.

Although bilateral lesions in the substantia nigra, rich in dopaminergic neurons and central to extrapyramidal motor control, were evident on follow-up MRI, the patient did not develop parkinsonism, such as tremors, rigidity, or bradykinesia. This is particularly notable, as similar midbrain involvement in previous CO poisoning cases has been associated with delayed-onset movement disorders, including parkinsonism [13]. The absence of such symptoms in our patient suggests that the observed midbrain lesions may have been subclinical or functionally compensated and that, at least in this case, the injury involved a reversible component. In addition, lesion distribution alone cannot reliably predict clinical outcomes, and the absence of white matter lesions does not guarantee that parkinsonism will not occur in general, and the relatively young age of the patient, potentially providing greater neuroplasticity and resistance to injury, may have contributed to the absence of parkinsonism. HBOT may not only reduce acute hypoxic damage but also enhance angiogenesis and neurogenesis [14, 15], which has been proposed as a possible mechanism underlying cognitive recovery and mitigation of subcortical damage in some patients. Moreover, previous studies have indicated that younger patients may exhibit enhanced neuroplasticity and repair mechanisms [16], which may have contributed to the favorable neurological outcomes in this case. Taken together, the patient’s young age, absence of extensive white matter injury, and early intensive treatment including HBOT may have contributed to the lack of delayed parkinsonism in this patient; however, a causal relationship cannot be established in a single case. Several reports have suggested that HBOT may be beneficial in delayed neurological sequelae after hypoxic brain injury, including CO poisoning and opioid-related delayed post-hypoxic leukoencephalopathy [17, 18]. In these case reports, prolonged or consecutive HBOT sessions were temporally associated with parallel improvements in clinical status and white matter abnormalities on MRI. Although such clinical observations are anecdotal and cannot establish causality, they indicate that HBOT may modify the course of delayed myelin injury in selected patients. More direct mechanistic evidence comes from experimental models. An adult rat model of hypoxic–ischemic brain damage demonstrated that HBOT improved learning and memory performance, protected against myelin injury in the hippocampus and prefrontal cortex as shown by Luxol fast blue and myelin basic protein staining, attenuated oligodendrocyte loss, modulated NG2-positive oligodendrocyte precursor cell dynamics, and reduced neuroinflammation and oxidative stress [19]. These findings indicate that HBOT can ameliorate myelin injury and promote myelin repair under hypoxic–ischemic conditions, providing a biological rationale that may be relevant to CO-related white matter damage. In our single case, we therefore interpret the radiological resolution of hippocampal and midbrain lesions during repeated HBOT as being consistent with, but not proof of, a potential myelin-protective effect suggested by these experimental and clinical data.

The hippocampal lesions observed in the early phase in this case were also noteworthy. Although hippocampal involvement is less commonly described in CO poisoning literature than globus pallidus lesions, it has been associated with cognitive dysfunction and memory impairment [4]. In our patient, prominent short-term memory loss was evident, which gradually improved in parallel with radiological improvement and intensive HBOT. Given the role of the hippocampus in memory consolidation, it is plausible that HBOT contributed to both radiological recovery and improvement in MMSE scores. Further studies are required to determine whether early intervention can attenuate or reverse CO-related hippocampal damage and its effects on cognition. Interestingly, higher cognitive function scores showed a transient decline around day 60 after admission and then improved on later follow-up (Fig. 5). This pattern does not indicate progressive or sustained deterioration, but rather suggests a temporary fluctuation in neuropsychological performance. In patients with carbon monoxide poisoning, cognitive test scores can be influenced not only by structural brain injury but also by fatigue, emotional state, and attention problems. In our case, the patient complained of fatigue and demonstrated an attention deficit at the time of the day-60 assessment, which likely contributed to the poorer scores at that point. We therefore consider the day-60 worsening to reflect a transient functional fluctuation rather than delayed neurological deterioration.

The recovery of the MRI signals by day 37 suggested that the hippocampal and midbrain lesions contained an at least partially reversible component. This radiological evolution may reflect a delayed yet non-necrotizing inflammatory or metabolic process, rather than irreversible structural injury. Previous reports have shown that in cases of CO poisoning, DWI hyperintensity and low ADC values in the cerebral white matter may persist during not only the acute phase but also in the subacute and chronic phases. In patients who develop DNS, ADC values may further decline after symptom onset and remain low for a few months. This prolonged diffusion restriction differs markedly from cerebral infarction, in which the ADC typically normalizes within a month, and has been interpreted as reflecting progressive demyelination with cytotoxic edema [1]. However, the patient had no apparent white matter lesions, which might have contributed to the absence of DNS. These observations highlight the importance of monitoring DWI and ADC dynamics in the subacute phase to identify the early markers of progressive white matter injury. Experimental data from focal cerebral ischemia models support the concept that HBOT can beneficially influence the fate of potentially salvageable tissue. In rat middle cerebral artery occlusion models, early HBOT has been shown to increase oxygen delivery to the ischemic periphery (penumbra-like regions), reduce infarct volume, and improve neurological outcomes compared with normobaric oxygen [20, 21]. These studies suggest that HBOT can stabilize metabolically compromised but structurally viable cells at the border between infarction and recovery. Taken together with the clinical literature in CO poisoning, these findings provide a biological rationale for a possible contribution of HBOT to the favorable radiological and clinical evolution observed in our case. Nevertheless, because this is a single case without a control group, we cannot determine the relative contributions of HBOT versus spontaneous recovery and patient-related factors, and we therefore interpret the potential role of HBOT as hypothesis-generating rather than conclusive.

From a clinical perspective, our findings highlight the need for ongoing neurological and radiological monitoring in patients with CO intoxication. The presence of delayed or subclinical midbrain lesions may have long-term implications, including the potential development of parkinsonian features. Prospective studies incorporating serial imaging and long-term follow-up are warranted to better elucidate the prognostic significance of such lesions and to inform guidelines for the monitoring and rehabilitation of CO poisoning survivors.

This case highlights several important considerations. First, hippocampal lesions, though rare sites of damage, can be treated with HBOT. Second, although midbrain lesions following CO intoxication are rare, they can emerge late during the subacute phase, even in the absence of severe neurological symptoms. Third, serial MRI is essential in capturing dynamic radiological changes that may otherwise remain undetected. Finally, although the presence of bilateral substantia nigra lesions might suggest a risk of extrapyramidal dysfunction, our patient did not develop overt parkinsonian symptoms, suggesting that such lesions can remain subclinical. Nevertheless, careful long-term follow-up is essential for monitoring the possible emergence of DNS. This case expands the spectrum of radiological findings in CO intoxication and emphasizes the diagnostic and prognostic value of MRI beyond the acute phase.

This case report was written in accordance with the CARE (Case Report) guidelines [22, 23].

Conclusions

This case expands the recognized spectrum of CO induced brain injury by demonstrating both acute hippocampal involvement with reversible cognitive dysfunction and delayed bilateral substantia nigra lesions without associated clinical symptoms. The observed recovery of cognitive performance, despite clear hippocampal damage on early imaging, suggests that such injuries may not always predict permanent deficits, particularly in young patients receiving prompt and repeated HBOT. Additionally, the presence of clinically silent midbrain lesions highlights the importance of serial neuroimaging for detecting delayed or subclinical pathologies. These findings support the potential usefulness of early HBOT and close neurological monitoring to optimize outcomes in patients with CO poisoning, even in the presence of deep brainstem lesions that are traditionally associated with poor prognosis.

Abbreviations

CO

Carbon Monoxide

HBOT

Hyperbaric oxygen therapy

DNS

Delayed neurological sequelae

Authors’ contributions

YI and NJ took care of the patient, wrote the draft, and critically revised the manuscript. KI and RN took care of the patient and critically revised the manuscript. SO critically revised the manuscript. All authors read and approved the final manuscript.

Data availability

All data generated or analysed during this study are included in this published article.

Declarations

Ethics approval and consent to participate

Ethics approval or consent to participate was not applicable.

Consent for publication

Our patient gave written informed consent for their personal or clinical details along with any identifying images to be published in this study.

Competing interests

The authors declare no competing interests.