Posterior Reversible Encephalopathy Syndrome in Clinical Toxicology: A Systematic Review of Published Case Reports

Background: Posterior reversible encephalopathy syndrome (PRES) is a rare clinical and radiological entity characterized by a typical brain edema. Although several case reports have described PRES in a context of poisoning, to our knowledge, a comprehensive assessment has not been performed. The aim of this systematic review was to raise awareness on poisoning-specific PRES features and to encourage consistent and detailed reporting of substance abuse–and drug overdose–associated PRES. Methods: Medline/PubMed, Web of Science, and PsycINFO were screened through May 31, 2019, to systematically identify case reports and case series describing PRES associated with poisoning (i.e., alcohol, drugs, illicit drugs, natural toxins, chemical substances) in accidental context, intentional overdose, and substance abuse. The methodological quality of eligible case reports/series was assessed. Patients and exposure characteristics were recorded; relevant toxicological, radiological, and clinical data were extracted. Results: Forty-one case reports and one case series reporting 42 unique cases were included. The median time to PRES onset from the start of exposure was 3 days (IQR 2–10). Acute high blood pressure, visual disturbance, and seizure were reported in 70, 55, and 50% of patients, respectively. The initial clinical presentation was alertness disorders in 64% of patients. Nine patients (21%) required mechanical ventilation. One-third of patients had at least one risk factor for PRES such as chronic hypertension (17%) or acute/chronic kidney failure (24%). The main imaging pattern (67%) was the combination of classical parieto-occipital edema with another anatomical region (e.g., frontal, basal ganglia, posterior fossa involvement). Vasogenic edema was found in 86% of patients. Intracranial hemorrhage occurred in 14% of patients. Both brain infarction and reversible cerebral vasoconstriction syndrome were diagnosed in 5% of patients. Three patients (12%, 3/25) had non-reversible lesions on follow-up magnetic resonance imaging. The median time required to hospital discharge was 14 days (IQR 7–18). Mortality and neurological recurrence rate were null. Conclusions: Comorbidities such as chronic hypertension and kidney failure were less frequent than in patients with other PRES etiologies. Imaging analysis did not highlight a specific pattern for poisoning-induced PRES. Although less described, PRES in the context of poisoning, which shares most of the clinical and radiological characteristics of other etiologies, is not to be ignored.

The pathophysiology of PRES is still debated through hypoperfusion and hyperperfusion theories. Impaired cerebral autoregulation responsible for disruption of the bloodbrain barrier (BBB) is one of the two main hypotheses, the other one being endothelial dysfunction (5). A recent review suggests that arginine vasopressin (AVP) axis stimulation could precipitate PRES development through an increase in AVP secretion or AVP receptor density. Activation of vasopressin V1a receptors leads to cerebral vasoconstriction, endothelial dysfunction, and subsequent brain edema (5).
Thus, PRES is a syndrome whose circumstances of occurrence and radiological features are associated with significant clinical and radiological polymorphism, making it difficult to characterize. Drug toxicity is one of the potential etiologies. Data are available in the literature regarding PRES occurring at therapeutic drug doses, but to the best of our knowledge, no review focused specifically on cases of accidental or intentional poisoning-associated PRES.
In order to investigate the occurrence of PRES in poisoned patients, the authors systematically reviewed the scientific literature of case reports and case series concerning PRES associated with poisoning (i.e., alcohol, drugs, illicit drugs, natural toxins, chemical substances) in accidental context, intentional overdose, and substance abuse. The authors did not include cases of calcineurin inhibitor overexposure, which has already been covered (6). The purpose of this study is to raise awareness of the features of PRES in poisoned patients and encourage consistent and detailed reporting of PRES in a context of overdose and substance abuse.

METHODS
A systematic review was conducted in accordance with the Preferred Reporting Items for Systematic reviews and Meta-Analyses (PRISMA) guidelines (7).

Eligibility Criteria
All language case reports and cases series were eligible if they concerned human subjects and included full-text. Records were screened if they were explicitly labeled as PRES based on head MRI (magnetic resonance imaging) or CT (computed tomography) scans. Case reports were included if PRES occurred in temporal connection with poisoning (i.e., alcohol, drugs, illicit drugs, chemical substances, natural toxins) in accidental context, intentional overdose, or substance abuse and/or a causal relationship was assumed by the authors of the report. [cannabis] OR [alcohol])). A manual search and screening of the bibliographies of selected articles was performed, in addition to the computerized search. The search strategy is summarized in Supplementary Figure 1.

Data Sources, Search Strategy, and Data Extraction
Extracted data included the following clinical and toxicological considerations: age, gender, risk factors of PRES occurrence, and exposure characteristics to the causative agent. Clinical and radiological data on PRES, its time course, and outcomes were also extracted. Unreported data were considered absent. Neuroimaging results were classified according to three different patterns. Briefly, the classical PRES pattern, involving only parieto-occipital lobes; the variant, including isolated various anatomical regions (e.g., frontal, cerebellum, brainstem, basal ganglia); and finally, the combined pattern, which includes combination of various regions.

Methodological Quality Assessment
The methodological quality of case reports and cases series was assessed using the tool proposed by Murad et al. (9) modified to adapt it to the analysis of toxicological exposures associated with PRES. The selected items were based on scores used for causality assessment in drug overdose (10,11). This new tool converges into 10 items that can be categorized into six domains: selection, ascertainment, diagnosis, causality, follow-up, and reporting (Supplementary Table 1). Briefly, items included time to onset of PRES, exposure characteristics (e.g., dose, drug detection, identification of species), radiological features, clinical data (e.g., symptoms, risk factors), dechallenge phenomenon, differential diagnosis, pharmacological properties of the causative agent, and clinical/radiological follow-up. The results of this modified tool have been summarized by summing the scores of the 10 binary responses into an aggregate score (Supplementary Tables 1, 2).

Case Selection
The literature search revealed a total of 149 non-duplicate records of which 95 were excluded because they did not report poisoningassociated PRES or were not case reports or case series. Out of 54 full-texts assessed for eligibility, 13 were excluded because they did not report poisoning-associated PRES or outcomes of interest (i.e., symptoms, brain edema features, follow-up). Finally, 40 case reports and 1 case series reporting a total of 42 unique cases were included (Supplementary Figure 1).

Characteristics According to the Causative Agent
Exposure characteristics, clinico-radiological features, and follow-up data of the included cases are shown in Table 1.

Chemical Substance
Phatake et al. (47) reported a patient complaining of headache and blurred vision 4 days after coma induced by consumption of organophosphorus compounds with suicidal intention. Brain MRI showed multifocal hyperintensities mainly in subcortical areas of parietal and occipital areas with increase ADC, suggesting PRES.

DISCUSSION
Are Toxic PRES Different From Other Etiologies?
As in other etiologies, poisoning-associated PRES are very polymorphic in terms of both exposure characteristics and patient comorbidities. Regardless of the substances involved, the median of 3 days for time to onset of PRES can dichotomize the presentation of this syndrome into two distinct situations. Indeed, PRES occurring after 3 days seems to be more related to a rebound effect of intoxication or to complications related to the management of the poisoning, whereas a shorter period would support a direct link between the causative agent and PRES.

Clinical and Biological Findings
Clinically, the prevalence of symptoms usually reported in PRES patients was consistent with the published literature (1). Indeed, acute high blood pressure, visual disturbance, and seizure were reported in 70, 55, and 50% of patients, respectively. Seizure is the most common symptom found in children (53), and all children had seizures (12)(13)(14)(15). However, comorbidities such as chronic hypertension (17%, 7/42) and acute or chronic kidney failure (24%, 10/42) were less frequently reported than in patients with other PRES etiologies, where their estimated prevalence is about 50% (1). In the literature, the frequency of isolated CSF protein elevation without pleocytosis, as a biomarker of permeability disruption of the BBB, varies from 60% (54) to 75% (55). In this review, although few cases reported CSF analysis, only one case (1/10, 10%) (34) described elevated CSF protein.
Imaging analysis did not highlight a specific pattern for poisoning-associated PRES but reinforces the fact that PRES is neither only posterior nor always reversible.

Vasoconstriction and endothelial dysfunction
Most of the pharmacological and toxicological agents involved in this review are known to induce either or both vasoconstriction and endothelial dysfunction. High alcohol concentrations have been associated with dose-related vasoconstriction and impaired dilatation of cerebral vessels (22). In addition, chronic alcoholism and long-term lithium exposure increase reactive oxygen species and nitric oxide in brain endothelial cells. This oxidative stress can induce endothelial dysregulation, leading to the BBB breakdown (27,34). Cocaine, amphetamines, lysergic acid amide, mephedrone, kratom, and phenylpropanolamine have sympathetic and/or serotoninergic effects that cause vasoconstriction or vasculitis, leading to severe hypertension (36,59).
In alcohol withdrawal, the hypothalamic-pituitary-adrenal system is stimulated, leading to increased levels of catecholamine and AVP, which can induce acute hypertension (27,28). Similarly, the biologically active component of licorice, glycyrrhizic acid, inhibits 11 β-hydroxysteroid dehydrogenase, leading to hypervolemic hypertension (16). After snakebite envenomation and organophosphate severe poisoning, neurotoxins (14) and nicotinic stimulation (60), respectively, lead to autonomic dysregulation with massive release of catecholamines and angiotensin II. These effects, in combination with the presence of sympathetic denervated vasculature in the brain posterior area, may induce failure of the cerebral autoregulatory system (14). The increase in proinflammatory mediators, either exogenous, originating from the substance (e.g., venom itself), or endogenous (e.g., in acute alcoholic pancreatitis, induced by venom), can damage the BBB and is likely to cause the leakage of blood contents into the surrounding brain tissue (14,30).

AVP axis hyperstimulation
A recent review highlighted that AVP overstimulation seems to be involved in PRES development and subsequent symptoms, in particular because of both its pathophysiological mechanism in brain edema formation and its involvement in most PRES etiologies (5). AVP hypersecretion could be the trigger of PRES through a dysregulation of ionic/water transglial flux via astrocytic ion channel dysfunction and subsequent brain edema. In the periphery, AVP receptor stimulation could be responsible for symptoms usually reported in PRES such as hyponatremia, acute hypertension, and impaired renal function (5,61). The use of cocaine, amphetamine (62), and lysergic acid diethylamide (63) and co-administration of alcohol with disulfiram (5) are known situations to stimulate AVP neurons. Thus, these agents could be directly responsible for the pharmacological cascade described above. In several cases, patients received multiple psychotropic drugs such as quetiapine (20,33,34), duloxetine (33), sertraline (33), amitriptyline (33), escitalopram (51), and valproate (51); it could be argued that these drugs, known to induce AVP release (64) and cerebrovascular effects, may serve as a contributing factor in the genesis of PRES.
In hepatic encephalopathy, in addition to dysregulation of cerebral blood flow and consequent cerebral vasodilation induced by hyperammonemia (32), ammonia reaching the astrocytes is detoxified in glutamine, and its overproduction promotes, through AVP stimulation, astrocytic swelling, resulting in brain edema (65). In contrast, acute alcohol intoxication (62) and licorice (66) inhibit AVP release, and lithium inhibits renal effects of AVP (5). In PRES associated with alcohol withdrawal (26)(27)(28), licorice (44), and lithium intoxication (33,34), the onset of neurological symptoms ranged from 2 to 9 days after intoxication; this chronology of events may suggest a rebound phenomenon on the AVP axis (5).

Multiple risk factors in substance abuse patients
In substance abuse-related PRES, patients had several risk factors for PRES such as chronic hypertension (18,19), chronic kidney disease (18,19), and HIV infection (18). In these cases, the different risk factors should act synergistically on the different pathophysiological pathways leading to PRES. For instance, cocaine has the ability to synergistically increase the pathologic processes induced by HIV infection (18) including endothelial dysfunction and disruption of the BBB integrity. Interestingly, single nucleotide polymorphisms in AVP gene and AVP V1a receptor have been associated with drug abuse disorders (67), suggesting a different probability of PRES occurrence in patients with substance abuse.
In patients with multiple risk factors of PRES or in the context of multiple drug intake (20), causality assessment is difficult. In addition, combined drug intoxication also exposes to a risk of drug cocktail effect. As with drug-drug interactions, illicit drug-drug interactions can occur both at the pharmacodynamic (i.e., interactions in which drugs influence each other's effects directly) and pharmacokinetic level (i.e., modification of drug absorption, distribution, metabolism, or excretion). In pharmacodynamic illicit drug-drug interactions, the neurotoxicity and/or endothelial toxicity of both the drug and illicit drug act synergistically, which can promote the occurrence of PRES. Legriel et al. (21) reported a case of lysergic acid amide-induced PRES in a depressed patient treated with clomipramine. The analysis of urinary catecholamines and serotonin metabolites showed a massive sympathetic storm with high urinary serotonin levels, supporting the pharmacodynamic convergence of these two adrenergic and serotoninergic agents. Castillo et al. (20) reported a case of kratom-induced PRES in a patient on fluoxetine for depression with dextroamphetamine misuse. Kratom exerts α2-adrenoreceptor agonistic effects and is known to induce high blood pressure, particularly when combined with other drugs (68). Mitragynine, the major indole-based alkaloid of kratom, is extensively metabolized by hepatic cytochrome P450 (CYP) isoform 3A4 and 2D6 (69). Amphetamine metabolism primarily involves CYP2D6 (70). Fluoxetine and its main metabolite, norfluoxetine, are wellknown inhibitors of CYP2D6 and CYP3A4, respectively (71). Taken together, a probable pharmacokinetic illicit drug-drug interaction occurred between mitragynine/dextroamphetamine and fluoxetine, causing overexposure in mitragynine and dextroamphetamine. This case of kratom/amphetamine interaction shares the pharmacological characteristics of PRES induced by clonidine, another α2-adrenoreceptor agonist, and dextroamphetamine overdose described by Minhaj et al. (35).

Coagulopathy
Characteristics of PRES associated with snake bite included normal blood pressure (38,39,42), coagulopathy (12,38,39), and respiratory failure (12,38). The mechanism of PRES associated with snake venom does not appear to be related to direct toxic effects of the venom within the central nervous system, because venom proteins do not cross the BBB. Instead, neurological manifestations are most often related to blockage of the neuromuscular transmission, causing paralysis, and abnormalities in the coagulation cascade, producing cerebrovascular events (38).

A Lack of Standardization in Reported Data on Poisoning-Induced PRES
The assessment of the methodological quality of the cases showed that the data reported are very heterogeneous, with a median overall score of 5/10 (IQR 3-6). In addition, none of the case reports met the domain selection criteria (Supplementary Table 1). An unclear selection approach leaves the reader with uncertainty as to whether this is the whole experience of the researchers or only the most serious case, and suggests possible selection bias. As the scientific literature on PRES is almost exclusively represented by case series and case reports, we propose a checklist with various items that should at least be reported in case reports of substance-induced PRES (Supplementary Table 4). Indeed, it seems essential to standardize the reporting of outcome so that studies collecting a large amount of data can be conducted.

CONCLUSIONS
PRES in a context of poisoning does not exhibit a specific imaging pattern. The predominance of various anatomical implications outside the parieto-occipital lobes in PRES, including toxic etiologies, is a key message for clinicians and radiologists. The prevalence of the neurological symptoms was also consistent with the published literature on other etiologies of PRES. In addition to toxic exposure, one-third of patients had at least one other risk factor of PRES. Chronic hypertension (17%) was less frequent than reported in other causes of PRES.
As in iatrogenic PRES, toxicology cases do not always have a temporally close relationship. PRES occurring after 3 days seems to be more related to a rebound effect of poisoning, to secondary brain damage, or to complications related to the management of the poisoning, whereas a shorter period would support a direct link between the causative agent and PRES. Poisoned patients may have a lower threshold for developing PRES. It may also be caused by the convergence of various pathophysiological processes (e.g., high blood pressure, endothelial dysfunction, AVP axis overstimulation) induced directly by the poison and/or indirectly by acting in concert with other factors such as drugdrug interaction or hypomagnesemia that ultimately causes the cerebrovascular cascade leading to PRES.
Although less described, PRES in a context of poisoning, which shares most of the clinical and radiological characteristics of other etiologies, is not to be ignored. The characterization of the pathophysiological mechanisms is essential to individualize management according to the presence of risk factors and/or specific features of PRES. Standardization of data reported in future case series of substance-induced PRES is required in order to better characterize and thus manage this syndrome.

DATA AVAILABILITY STATEMENT
Publicly available datasets were analyzed.

AUTHOR CONTRIBUTIONS
BL, CS, and SE conceived the idea. BL wrote the manuscript and performed the selection and summary of published literature on PRES in clinical toxicology. DB, CV-V, CS, and SE helped to design, write, and revise the paper. CC contributed in improving the article, notably enriching the analysis of neuroimaging data.