Malignant cerebellar edema in three-year-old girl following accidental opioid ingestion and fentanyl administration

Abstract

A three-year-old girl was found altered with an unknown timeline. Gas chromatography mass spectrometry was positive for hydromorphone, dihydrocodeine, and hydrocodone. Initial computed tomography and magnetic resonance imaging suggested a malignant cerebellar edema not confined to a vascular distribution. She received fentanyl boluses on hospital days 0 and 1 before receiving a continuous infusion on day 1. On day 3, she had an episode of acute hypertension and bradycardia. Emergent computed tomography showed an evolving hydrocephalus and similar diffuse edema throughout both cerebellar hemispheres. External ventricular drain was placed to relieve the increased intracranial pressure. Following drain placement and fentanyl discontinuation, the patient recovered, though not without fine- and gross-motor deficits at the four-month follow-up. Our case adds to a handful of case reports of opioid toxicity in pediatric patients that present as toxic leukoencephalopathy. Though the mechanism is poorly understood, it has been suggested to be a consequence of the neurotoxic effects of the drug, which has particular affinity for µ opioid receptors—the primary opioid receptor found in the cerebellum. Clinicians would do well to recognize that this syndrome is primarily caused by direct toxicity rather than ischemia. This case adds insight by suggesting that lipophilic opioid analgesics may worsen this neurotoxicity. When intervening with mechanical ventilation, clinicians should consider avoiding lipophilic opioid drugs for analgesia until the pathogenesis of cerebellar edema is better understood.

Introduction

We are currently facing an opioid epidemic of historical proportions. Children are among its most tragic victims. Opioid exposure in the pediatric population is highest in the 0–5 years age range, and cases are mostly attributed to exploratory behaviors in this age group.¹ Although clinical presentation is variable, respiratory depression—potentially requiring intubation and mechanical ventilation—is common. Among children intubated for ingestion, opioid ingestion was the most common cause in those younger than two years of age.² Admission to pediatric intensive care units increased by 35% between 2004 and 2015, tracking the global increase in opioid-related hospitalizations in the general pediatric population.³ When an altered pediatric patient arrives in the emergency department, exploratory ingestion is always on the differential list.

Patients classically present in a stuporous state with depressed brain-stem reflexes and unresponsive pupils.⁴ The primary goal is often prevention of irreversible brain changes secondary to respiratory arrest. Initial treatment may include opioid reversal with antagonism by naloxone. Other management includes vasopressors and mechanical ventilation in the case of respiratory compromise. Patients may undergo imaging as part of routine workup for altered presentation, which may show ischemic changes on imaging due to apnea or hypopnea. However, a malignant cerebellar edema has been observed on rare occasions in pediatric patients. Eleven cases have been reported in the literature.^5–14 This edema likely results from a direct opioid-induced neurotoxicity, with hypoxia-ischemia playing a secondary role.¹⁵ As the underlying pathophysiology is not confined to a vascular distribution, toxicity-driven edema plays a bigger role than does ischemia-driven edema in these patients, as discussed in detail in the discussion section.

Case report

A three-year-old girl weighing 13.5 kg presented to the emergency department for altered mental status. She was found altered with her two siblings and one other child at 9:00 am on the day of presentation. Multiple opiate pills were discovered on the ground nearby. The quantity of opiates consumed was unknown. One of her siblings was pronounced dead of respiratory arrest at the local hospital. Our patient received two doses of naloxone, with response to the second dose, before being airlifted to our facility at 11:00 am. En route between 11:00 am and 12:30 pm, the patient was intubated and sedated. A total of 5 mg midazolam and 100 µg fentanyl was administered during the airlift by the paramedical staff. On arrival (hospital day 0), urine drug screen was positive for benzodiazepines and opiates (gas chromatography mass spectrometry later resulted in nonquantitative positive tests for hydromorphone, dihydrocodeine, and hydrocodone). Her Glasgow Coma Scale Score was 5T. She was minimally responsive but hemodynamically stable with a heart rate of 162.

A non-contrast computed tomography (CT) scan showed diffuse areas of hypoattenuation involving bilateral superior cerebellar hemispheres that were not confined to any vascular distribution. The imaging demonstrated ventral displacement of the vermis, resulting in mass effect on the fourth ventricle and subsequent mild proximal third and bilateral lateral ventricular dilatation (Figure 1(a) and (b)). Cerebellar tonsillar herniation associated with crowding at the level of the foramen magnum was seen. Subsequent magnetic resonance brain imaging on day 1 (Magnetom Aera, 1.5T; Siemens Healthineers) was performed with axial diffusion-weighted images with apparent diffusion coefficient maps, sagittal T1 magnetization prepared rapid acquisition gradient echo (MPRAGE) with axial and coronal reconstructions, axial T2, axial fluid attenuated inversion recovery (FLAIR), axial susceptibility-weighted imaging, and post-contrast sagittal T1 MPRAGE and axial T1 sequences. It showed restricted diffusion involving both cerebellar hemispheres, which was considered to represent malignant cerebellar edema subsequent to accidental opioid ingestion, given prior reports in the literature.¹³ Areas of diffusion restriction and altered T2/FLAIR signal in supratentorial white matter representing supratentorial leukoencephalopathy were seen—also a described entity in this condition. Additionally, effacement of cerebrospinal fluid space around the cervicomedullary junction representing transforaminal herniation of the cerebellar tonsils was seen. Mass effect on the fourth ventricle resulting in ventricular dilatation, increased from the prior CT head scan, was also appreciated (Figure 2). Both CT and magnetic resonance imaging (MRI) were read by a board-certified subspecialty neuroradiology faculty (T.R.) with 14 years of experience as a neuroradiologist.

Figure 1.

CT axial and coronal reconstructed (A and B) images show diffuse hypodensities in both cerebellar hemispheres (large arrows) with dilatation of third and both lateral ventricles representing early hydrocephalus (small arrow). Axial and coronal reconstructed (C and D) images obtained three days later show similar diffuse bilateral cerebellar edema but increasing hydrocephalus with increased third and both lateral ventricular dilatation (arrowheads).

Figure 2.

Magnetic resonance brain sagittal T1-weighted (A) image shows hypointense areas in cerebellar hemispheres, effacement of cerebrospinal fluid spaces at the cervicomedullary junction, and cerebellar tonsillar herniation (large arrow). Axial diffusion-weighted (B) and apparent diffusion coefficient maps (C) through the posterior fossa show diffusion restriction involving both cerebellar hemispheres (small arrow) with mass effect on the fourth ventricle representing malignant cerebellar edema. T2-weighted images (D) demonstrate increased T2 signal in both cerebellar hemispheres. Axial diffusion-weighted (E) and apparent diffusion coefficient maps (F) through the supratentorial white matter at the level of centrum semiovale show diffusion restriction representing supratentorial leukoencephalopathy (arrowheads).

After approximately a four-hour triage in the emergency room from 12:30 pm to 4:25 pm, our patient was transferred to the pediatric intensive care unit and connected to a ventilator at 3:16 pm. As she was waking up and ventilation required continued sedation, she received fentanyl boluses on day 0 (0.5 µg/kg, 7 µg, 1:07 pm) and day 1 (1 µg/kg, 15 µg, 11:10 am). She spontaneously awoke on both days, prompting providers to begin an as-needed fentanyl infusion (2 µg/kg/h, started day 1 at 12:30am) for sedation and analgesia. She was also started on an as-needed midazolam infusion (1 mg/h, started day 1 at 3:45 pm) to assist with sedation and agitation, the use of opioids and benzodiazepines being a routine choice in the pediatric intensive care setting.¹⁶ Per protocol, she received coverage for bacterial meningitis and aspiration pneumonia.

On day 3, the patient became acutely hypertensive and bradycardic, suggesting a Cushing reflex in response to increasing intracranial pressure and tonsillar herniation. Emergent CT showed an evolving hydrocephalus and similar diffuse edema throughout both cerebellar hemispheres, with resulting mass effect and herniation (Figure 1(c) and (d)). An emergent bedside twist drill craniotomy was performed with placement of an external ventricular drain. The patient had cumulatively received 229 µg of fentanyl prior to that point. The fentanyl infusion was discontinued per neurosurgery recommendations, as the neurosurgeons recognized that fentanyl may have been potentiating the edema on the follow-up imaging. Ventilatory sedation was later maintained primarily with midazolam (1.5 mg/h, started day 3 at 6:00 pm). The patient was extubated on day 6. Her drain was removed on day 12. She had no further neurological complications and was discharged on day 30.

At the patient’s four-month neurology follow-up, she was found to have fine- and gross-motor delays, including gait and speech difficulties. It is unclear which delayed milestones are attributable to her opioid exposure versus previous documented neglect by her foster parents, which included a referral to child protective services for frostbitten feet.

Discussion

Malignant cerebellar edema following opioid toxicity in pediatric patients is a rare phenomenon. To our knowledge, it has thus far been documented in only 11 case reports.^5–14 As we are in the midst of an opioid epidemic, exploratory ingestion has become a critical part of the differential diagnosis when a pediatric patient presents with altered mental status. In patients who present with malignant cerebellar edema and signs of increased intracranial pressure—which can mimic the pinpoint pupils and suppression of brain-stem reflexes seen in traditional opioid overdose—clinicians should recognize that a direct neurotoxic edema, not merely an isolated hypoxic-ischemic edema, may be at play. This edema may progress, resulting in tonsillar herniation. Malignant cerebellar edema requires timely diagnosis because the mass effect causes obstructive hydrocephalus and direct damage to the brain stem. If the brain stem is spared from ischemic insult/infarction, patients with malignant cerebellar edema development can achieve good outcomes.¹⁷ In cases of imminent herniation, posterior fossa decompression should be considered. Early neuroimaging in cases of pediatric opioid overdose is therefore critical.

This edema can mimic the physical presentation of opioid overdose and thus go unnoticed, especially in a patient who is intubated and sedated.⁵ Our patient required mechanical ventilation. Given her sister’s respiratory arrest and death, it is likely clinicians were sensitive to her respiratory status, lowering the threshold for mechanical ventilation. The patient appeared to be improving, so the acute worsening of symptoms on day 3 in the setting of malignant edema is confusing. The timeline suggests that another cause—in particular, her fentanyl boluses and subsequent continuous infusion—may have exacerbated her pathogenesis.

The neuropathology is likely driven by several mechanisms, including direct neurotoxicity mediated by cerebellar opioid receptors, with ischemia secondary to respiratory arrest aggravating the injury.¹⁵ Opioids such as heroin and morphine seem to induce neuronal apoptosis, with mitochondria acting as important mediators,¹⁵ but the precise mechanisms of this direct neurotoxicity and resultant edema are not currently well understood.

Ischemia of the cerebellum as the primary cause of our patient’s presentation is unlikely. During ischemia, autoregulatory mechanisms shunt blood from less metabolically active structures, such as the cerebral cortex and white matter, to more active structures, such as the cerebellum.¹⁸ Indeed, in hypoxic-ischemic encephalopathy, the cerebellum is often the final structure affected,¹³ suggesting ischemic-only encephalopathy as an unlikely cause of our patient’s cerebellar damage.

Additionally, our patient’s acute decline on day 3 does not seem to be driven by her exploratory ingestion. The half-lives of hydromorphone,¹⁹ hydrocodone,¹⁹ and dihydrocodeine²⁰—the three opiates found to be positive on gas chromatography mass spectrometry—are each less than six hours. More than 90% of these drugs would have been eliminated from her system within 24 hours of ingestion. However, pharmacokinetics of opioid overdoses may be more closely described by zero-order kinetics, which might in fact produce a delayed response.⁴ Even so, a delayed response does not seem to describe the patient’s clinical picture fully. She did not become hypotensive, as would be expected if it was driven primarily by opioid receptor agonism.⁴ Instead, she experienced a hypertensive Cushing reflex in response to worsening cerebellar edema—confirmed with a CT head. Following relief of this heightened pressure via an external ventricular drain, she went on to improve progressively, an unlikely result if the initial drug were still binding cerebellar receptors.

Given the likely toxic nature of her cerebellar damage combined with her acute clinical decline three days later, it is possible her condition was exacerbated by her continued exposure to fentanyl, as was suggested by our neurosurgical team. Her acute decline may have resulted simply from progressively worsening edema subsequent to her initial ingestion. Delayed decline following opioid ingestion has been noted in one pediatric case, although the cerebellar lesions were focal and did not appear on imaging until weeks after opioid exposure.²¹ In our case, there are reasons to believe her fentanyl exposure played a contributory role.

First, her condition did not progressively worsen, as would be expected if the expanding edema stemmed solely from her original ingestion. She improved clinically, regaining consciousness on days 1 and 2 before acutely declining on day 3. However, mechanical ventilation may have masked any worsening symptoms. Second, the temporal relationship between her decline and onset of continuous fentanyl infusion is suspicious. She experienced an acute decline within 24 hours after beginning a continuous fentanyl infusion. Additionally, upon relieving the increased intracranial pressure and discontinuing fentanyl, the patient progressively improved without further events. Finally, fentanyl is the most lipophilic of the opioid derivatives, implying a heightened propensity for crossing the blood–brain barrier and thereby resulting in direct toxicity. It also binds predominantly to µ opioid receptors. The cerebellum has a greater density of µ receptors than does any other brain structure, although the role of these receptors in toxic situations is not known.²² Therefore, fentanyl may be uniquely positioned to worsen opioid-induced cerebellar toxicity. Indeed, a case of toxic cerebellar edema in a pediatric patient following transdermal fentanyl patch exposure has been reported. This patient, like ours, also received fentanyl for sedation at the hospital.⁶

This pattern of opioid-induced toxic leukoencephalopathy—malignant cerebellar edema with overlying areas of supratentorial leukoencephalopathy—appears to be distinct to children.¹³ Cerebellar opioid receptor profiles differ between children and adults, which may explain this differential presentation in children. For example, newborn cerebellar µ receptors have twice the binding capacity of those in adults.²³

The more well-known “chasing the dragon” syndrome, seen with heroin inhalation in adults,²⁴ parallels the opioid-induced cerebellar edema described here, especially in that diffusion is restricted in both the cerebellum and watershed areas of the deep white matter.^7,25 However, it differs in important ways. First, unlike the acute edema seen in children, “chasing the dragon” syndrome usually presents over weeks to months after prolonged exposure,²⁶ although some cases with more acute onsets have been reported.^27,28 Second, the syndrome follows opioid inhalation rather than ingestion.²⁶ Histopathology examinations have revealed the presence of spongiosis with massive astrogliosis. Typical hypoxic lesions are absent,²⁴ consistent with a direct neurotoxic opioid effect, as hypothesized in our case. To date, no specimens of affected tissue in a pediatric patient have been examined. Histopathological examination will help clarify the precise mechanisms of this condition.

In opioid overdose, with the risk of brain ischemia secondary to respiratory arrest, it is understandable that clinicians would act swiftly to begin mechanical ventilation, which involves analgesia. Given the lack of evidence-based guidelines for analgesia and sedation in critically ill children, however, it is no surprise that there is a significant degree of variability in clinical practice. A prospective multicenter study of seven pediatric intensive care units showed up to 100-fold differences in baseline opioid doses, average daily doses, total doses, or peak infusion rates.²⁹ In general, fentanyl is often favored for its hemodynamic stability and titratability in the pediatric population³⁰ and is one of the few Food and Drug Administration–approved opioids for analgesic use in the pediatric population. A meta-analysis of intranasal fentanyl use, which has been shown to result in blood levels comparable to intravenous administration, found no evidence of toxicity across 313 subjects.³¹

However, there remains a paucity of studies on fentanyl toxicity in the pediatric population. The range of metabolic function, immature hepatic and renal clearance, and sensitive respiratory drive among the pediatric population, especially in young children, make it difficult to create conclusive guidelines. One study, for example, showed large interpatient variability in fentanyl clearance following continuous infusion, with patients requiring a 10-fold variability in infusion rates to achieve similar levels of sedation.³² In addition, many studies specifically exclude children in our patient’s age range. Although further research is necessary, in cases of opioid overdose, using non-lipophilic opioid analgesics or even non-opioid analgesics—such as acetaminophen or ketamine, which can have powerful opioid-sparing effects³³ and are safe in children³⁴—may prove to be the optimal management approach. If analgesia consists of lipophilic opioids, our case suggests the effects of this mode of analgesia should be carefully assessed for neurotoxic effects.

Conclusion

This case of a three-year-old who ingested opioids and developed malignant cerebellar edema provides an additional report of this rare complication of opioid toxicity in certain pediatric patients. Research clarifying the pathogenesis of opioid-induced neurotoxicity in pediatric patients is still needed. This case also brings new understanding by suggesting that lipophilic opioid analgesics used during mechanical ventilation may exacerbate the malignant edema. Although speculative, our patient’s acute hypertensive decline on hospital day 3 seems to be a direct result of worsening cerebellar edema secondary to fentanyl infusion. Fentanyl has a particular affinity for µ opioid receptors—the primary opioid receptor found in the cerebellum.³⁵ Investigations into the direct toxic effects of fentanyl and other lipophilic opioids can help establish clearer recommendations for managing opioid-induced neurotoxicity in the pediatric population.

Funding

This research received no specific grant from any funding agency in the public, commercial, or not-for-profit sectors.

Conflict of Interest

The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.