Determinants of Recovery from Severe Posterior Reversible Encephalopathy Syndrome

Objective Few outcome data are available about posterior reversible encephalopathy syndrome (PRES). We studied 90-day functional outcomes and their determinants in patients with severe PRES. Design 70 patients with severe PRES admitted to 24 ICUs in 2001–2010 were included in a retrospective cohort study. The main outcome measure was a Glasgow Outcome Scale (GOS) of 5 (good recovery) on day 90. Main Results Consciousness impairment was the most common clinical sign, occurring in 66 (94%) patients. Clinical seizures occurred in 57 (81%) patients. Median mean arterial pressure was 122 (105–143) mmHg on scene. Cerebral imaging abnormalities were bilateral (93%) and predominated in the parietal (93%) and occipital (86%) white matter. Median number of brain areas involved was 4 (3–5). Imaging abnormalities resolved in 43 (88%) patients. Ischaemic and/or haemorrhagic complications occurred in 7 (14%) patients. The most common causes were drug toxicity (44%) and hypertensive encephalopathy (41%). On day 90, 11 (16%) patients had died, 26 (37%) had marked functional impairments (GOS, 2 to 4), and 33 (56%) had a good recovery (GOS, 5). Factors independently associated with GOS<5 were highest glycaemia on day 1 (OR, 1.22; 95%CI, 1.02–1.45, p = 0.03) and time to causative-factor control (OR, 3.3; 95%CI, 1.04–10.46, p = 0.04), whereas GOS = 5 was associated with toxaemia of pregnancy (preeclampsia/eclampsia) (OR, 0.06; 95%CI, 0.01–0.38, p = 0.003). Conclusions By day 90 after admission for severe PRES, 44% of survivors had severe functional impairments. Highest glycaemia on day 1 and time to causative-factor control were strong early predictors of outcomes, suggesting areas for improvement.


Introduction
Posterior reversible encephalopathy syndrome (PRES) is a clinicoradiologic entity characterized by a variable combination of consciousness impairment, seizure activity, headaches, visual abnormalities, nausea/vomiting, and focal neurological signs. [1,2,3,4,5] Cerebral imaging abnormalities are often symmetric and predominate in the posterior white matter. Oedema is an occasional finding in the frontal and temporal lobes, basal ganglia, cerebellum, brainstem, and cortical grey matter. [4,5,6] Recognition of PRES is improving with the increasing availability of magnetic resonance imaging (MRI) and recent reappraisal of the imaging abnormality spectrum. [6].
The pathophysiology of PRES remains controversial, and the two main hypotheses contradict each other. One involves impaired cerebral autoregulation responsible for an increase in cerebral blood flow, whereas the other incriminates endothelial dysfunction with cerebral hypoperfusion. This hypoperfusion hypothesis may be most relevant to cases of PRES associated with cytotoxic therapy. Under both hypotheses, the cerebral blood perfusion abnormalities result in blood-brain barrier dysfunction with cerebral vasogenic oedema. [4,5,7] PRES can develop in association with a vast array of clinical conditions and is typically reversible once the cause is removed. [3,8,9] However, its reversible nature has been challenged based on reports of permanent neurological impairments and of mortality rates reaching 15% [8,9]. No studies focusing specifically on patients with severe PRES requiring life-sustaining treatments [10,11,12,13] have been published to date. Although the pathophysiological mechanisms may be the same as in less severe forms, knowledge of factors influencing the outcome of severe PRES might result in improved early management. [7].
Here, our objective was to identify predictors of functional outcome on day 90 in adults with severe PRES, with special attention to factors amenable to improvement. Thus, we thought to provide intensivists and other clinicians a realistic picture of severe PRES with useful management pathways in daily clinical practice. To this end, we conducted a multicentre retrospective cohort study.

Methods
The ethics committee of the French Society for Critical Care approved the constitution of this retrospective cohort of patients with severe PRES.

Patients
Patients admitted to one of the 24 participating ICUs (Appendix) between May 2001 and May 2010 and exhibiting clinical and neuroimaging features consistent with severe PRES, as defined below, were eligible for the study.
All neuroimaging documents were reviewed by two independent and certified neurologists with specialist stroke-experience and trained in MRI diagnosis of PRES and its pitfalls. Neuroimaging review was performed blinded to the clinical findings. Patients were included by consensus between two neurologists (FP and EM). Patients with no brain imaging studies available for review by the study neurologists were not included.

Definitions
PRES was defined as a variable combination of acute neurologic clinical changes including consciousness impairment, seizure activity, headaches, visual abnormalities, nausea/vomiting, and focal neurological signs [4,5] associated with neuroimaging findings consistent with PRES [4], including vasogenic oedema by MRI diffusion sequences (measurement of apparent diffusion coefficient (ADC)) or at least partial reversibility on follow-up imaging when diffusion sequences were not available. [6].
PRES was severe when associated with neurological failure defined by any neurological disorder of central origin among impairment of consciousness, seizure with or without status epilepticus, focal sign, encephalopathy, and meningeal symptoms, which required intensive care management for monitoring or life support management.
Coma was defined as the absence of arousal and consciousness with a Glasgow Coma Scale (GCS) score , 9. [14,15].
Visual abnormalities consisted of blurred vision, visual neglect, homonymous hemianopsia, visual hallucinations, and cortical blindness.
Focal neurological signs consisted of symptoms or signs related with damage to, or dysfunction of, a specific anatomic site in the central nervous system. [16] These signs were categorized as unifocal or multifocal, and as transient or persistent.
Convulsive status epilepticus was defined as continuous motor seizure activity for at least 5 minutes (continuous) or as more than two motor seizures without full recovery of consciousness in the interval (intermittent). [17,18] Refractory status epilepticus was defined as continuous or intermittent seizures despite treatment with an intravenous benzodiazepine (clonazepam or diazepam) and intravenous phenytoin, fosphenytoin, or phenobarbital. [19] Electrical status epilepticus was diagnosed in comatose patients with or without subtle convulsive movements (rhythmic twitching of the arms, legs, trunk, or facial muscles; tonic eye deviation; or nystagmoid eye jerking) [20] but with generalized ictal discharges on the electroencephalogram (EEG). [21].
Seizure activity on the EEG was defined as continuous or recurrent rhythmic focal or generalized spikes; sharp waves; spike waves; or rhythmic waves changing in amplitude, frequency, and/ or spatial distribution. [22].

Diagnosis and Treatment
The management combined symptomatic life-supporting treatments and control of the factor causing PRES. [10,11] Efforts were made to control systemic secondary brain insults and to limit effects of potential cranial hypertension Hypoglycemia was routinely checked and corrected. If glucose was given, 100 mg of thiamine was administered concomitantly, most notably when there was evidence of vitamin B1 deficiency. Patients were also routinely evaluated for hyperthermia, hyperglycemia, hypo-or hyper-carbia, anemia, metabolic disturbances, epileptic activity and aspiration pneumonia that may complicate the initial consciousness disorders and which required prompt correction. Patients with status epilepticus were managed as previously described. [24] Control of severe hypertension, if present, was an important part of the symptomatic management. Intravenous antihypertensive drugs including labetolol, nicardipine, or urapidil were given. [25].
A neurologist consultant was available for advice. Extensive diagnostic investigations were performed. CT was easier to obtain first. MRI was performed in most patients, as either the first or the second imaging study. Both local radiologists and neurologist consultants read the neuroimaging studies. In addition, a neuroradiologist was consulted if deemed necessary. Electroencephalography (EEG) was performed routinely to look for nonconvulsive status epilepticus. Cerebrospinal fluid was examined in patients with a fever or clinical suspicion of meningitis and when deemed appropriate by the attending physicians. Laboratory tests were also obtained routinely. Plasma anticonvulsant drug assays, including magnesium assays, and qualitative tests for toxic agents or medications associated with seizures and other symptoms of PRES were performed at the discretion of the attending physicians.
Interventions to control identified causative factors were initiated promptly. These interventions included blood pressure control, withdrawal of cancer chemotherapy or immunosuppressive agents, caesarean section, and/or dialysis, as appropriate.

Data Collection
A standardized form was used to collect the variables listed in Tables 1-7 and Figure 1. Clinical features of PRES were collected retrospectively based on data in the pre-hospital notes, emergencyroom chart, and ICU chart.
A standardized form was used to collect the neuroimaging features of PRES: uni/bilateral involvement, grey/white matter involvement, lesion distribution (i.e., parietal and/or occipital and/or temporal and/or frontal lobes, ganglia, brainstem, cerebellum), presence of ischaemic and/or haemorrhagic complications, reversibility of lesions, and recurrences ( Figure 1).
The primary cause of PRES was classified as hypertensive encephalopathy, exposure to a toxic agent, toxaemia of pregnancy (preeclampsia/eclampsia), autoimmune disease, or miscellaneous causes. Toxicity was considered when PRES occurred during or after a pharmacological treatment known to induce neurological toxicity (Table 3). Hypertensive encephalopathy was diagnosed in patients presenting the association of severe hypertension and signs of neurological failure [23], and absence of evidence of other etiologic category as described above.
The time from symptom onset to control of the causative factor was clocked (e.g., time to achieve blood pressure values within the normal range for hypertension, time to treatment discontinuation for toxicity, or time of C-section for toxaemia).
Severity and organ dysfunction at ICU admission were assessed using the Simplified Acute Physiology Score II (SAPS-II) [26] and the Logistic Organ Dysfunction (LOD) score. [27].

Assessment of Outcome
The primary evaluation criterion was the structured Glasgow Outcome Scale (GOS) score on day 90 (67 days) after onset of PRES ( Table 1). The GOS score reflects both mortality and morbidity. [28] Each surviving patient was interviewed for GOS score determination on day 90 and day 180 by a trained physician, when allowed by the time since inclusion. Otherwise, the GOS score was extracted from the hospital charts or determined by interviewing the patient's general practitioner or neurologist. For this study, we defined a favourable outcome as a GOS score of 5, that is, alive with good function enabling the return to former occupational or academic activities, with or without minor physical or mental deficits.

Statistical Analysis
Quantitative parameters are reported as median and interquartile range (IQR, 25th-75th percentile) and qualitative parameters as numbers and percentage. Categorical variables were compared using the x2 test or Fisher's exact test, as appropriate. Continuous variables were compared using the Mann-Whitney U test or the Wilcoxon test, as appropriate. The day-90 GOS score was known for all study patients. Associations between patient characteristics and 90-day GOS score were assessed using a logistic regression model (Tables 5 and 6). Multivariable analysis was performed using stepwise forward selection to introduce variables whose P values were smaller than 0.20 by univariate analysis. The following variables were entered into the model: age (year), pre-existing co-morbidity (Y/N), epileptic seizure (Y/N), status epilepticus (Y/N), GCS score at ICU admission, headache (Y/N), acute hypertension (Y/N), mean arterial pressure on scene (mmHg), highest glycaemia value (mmol/L), LOD score, grey matter involvement (Y/N), brainstem involvement (Y/N), ganglia involvement (Y/N), total number of brain areas involved, haemorrhagic complication at first imaging (Y/N), time from PRES onset to causative-factor control (hours), SAPS II score, duration of mechanical ventilation (days), mechanical ventilation (Y/N), refractory status epilepticus (Y/N), length of ICU stay (days), length of hospital stay (days), toxaemia of pregnancy (Y/N), and exposure to toxic agent (Y/N). Then, the absence of a significant increase in the likelihood value after omission of each of the remaining variables was checked. Variables were tested for co-linearity and interactions before inclusion in the multivariable model. Goodness of fit was evaluated by the Hosmer-Lemeshow statistic. The area under the ROC curve was estimated by the c statistic (Association of Predicted Probabilities and Observed Responses). Odds ratios (ORs) and their 95% confidence intervals (95%CIs) were computed. Values of p less than 0.05 were considered statistically significant. Analyses were done using the SAS 9.1 software package (SAS Institute, Cary, NC, USA).

Patients (n = 70)
The patient flow chart is shown in Figure 1. Of the 86 patients admitted to the 24 study ICUs for possible PRES during the 9year study period, 7 did not meet our inclusion criteria and 9 had no neuroimaging documents available for review, leaving 70 patients for the study. Table 2 reports the main patient characteristics.

On-scene Clinical Presentation (Figure 1)
Consciousness impairment was the most common clinical feature, being present in all but 4 (94%) patients. Clinical seizures occurred in 57 (81%) patients. Median value of mean arterial pressure was 122 (105-143) mmHg on scene. Acute hypertension was noted in 62 (89%) patients.

Discussion
In this retrospective multicentre study of 70 patients with severe PRES, 11 (16%) patients died before day 90 and only 33 (56%) survivors had a good recovery defined as a GOS score of 5. Both of the factors independently associated with a poor outcome (GOS,5) were directly linked to the early management, and both may offer hope for improving outcomes (Panel).
Patient characteristics were consistent with previous studies. PRES preferentially affects women and relatively young individuals with serious co-morbidities. [9] The main clinical manifestations were consciousness impairment (94%) and seizure activity (81%). Consciousness was often severely impaired, with a median Glasgow Coma Scale score of 9 (3)(4)(5)(6)(7)(8)(9)(10)(11)(12)(13)(14) at ICU admission. The 44% prevalence of status epilepticus was considerably higher than in previous reports of unselected cases of PRES. [1,9] The other clinical signs such as headaches, vision impairment, nausea/ vomiting, and focal neurological signs at scene were in accordance with recent reports. [1,6,8,9,29] Acute hypertension was found onscene in 83% of patients, with a median value of mean arterial pressure level of 122 (105-143) mmHg, in keeping with earlier data. [1,3,8,9] The neuroimaging findings in our study were also consistent with current knowledge. Despite a marked predominance of posterior parietal-occipital lobe involvement, atypical patterns were seen in a substantial proportion of patients and included strictly unilateral lesion distribution and involvement of the grey matter, anterior lobes, basal ganglia, cerebellum, and brainstem. Thorough familiarity with these atypical patterns is crucial, especially when the clinical features are not suggestive [6].
All patients received early symptomatic management concomitantly with prompt and extensive investigations to identify the cause. The on-scene presentation was severe as evidenced by the

3
Severe disability Patient is unable to live independently but can follow commands. This indicates that a patient is conscious but needs the assistance of another person for some activities of daily living every day. This may range from continuous total dependency to the need for assistance with only one activity.

4
Moderate disability Patient is capable of living independently but unable to return to work or school. Such a patient is able to look after himself at home, to get out and about to the shops and to travel by public transport. However, some previous activities, either at work or in social life, are now no longer possible by reason of either physical or mental deficit. high number of organ failures, and mechanical ventilation was required in 71% of patients. Antihypertensive drugs and anticonvulsants were given early, as necessary. At least one cerebral imaging study was performed, and 83% of patients had MRI, the reference standard for diagnosing PRES. [30] At least one cause was identified in all patients, allowing early etiological treatment. Median time from PRES onset to control of the cause was only 30  hours. Exposure to toxic agents was the most common cause, with 44% of patients. The already long list of toxic agents associated with PRES is growing steadily. In our study, immunosuppressants such as anticalcineurin agents (cyclosporine A and tacrolimus FK506) were the most common toxic causes. Hypertensive encephalopathy was the second most common causative factor, with 41% of patients. Toxaemia of pregnancy and autoimmune diseases were identified in 23% and 11% of patients, respectively, potentially explaining the female predominance of the syndrome. Finally, other factors reported in  [14,15].  Refractory status epilepticus was defined as continuous or intermittent seizures despite treatment with an intravenous benzodiazepine (clonazepam or diazepam) and intravenous phenytoin, fosphenytoin, or phenobarbital [19].

{
Four deaths directly ascribable to PRES: one patient each had brain death related to cardiac arrest complicating the treatment of status epilepticus, refractory status epilepticus with multi-organ failure, cerebral haemorrhage with ventricular flooding, and brain death related to cerebral herniation complicating cerebral ischaemia. doi:10.1371/journal.pone.0044534.t002 association with PRES, sometimes only anecdotally, include hypomagnesaemia, sickle cell disease, and infection/sepsis/shock. We identified three factors independently associated with the day-90 functional outcome. Among them, time to control of the causative factor provides the greatest room for improvement. Although a role for this variable was suggested previously [11,31], our study is the first to demonstrate an independent association with the outcome. Patients may require blood pressure control, withdrawal of cancer chemotherapy or immunosuppressants, caesarean section, dialysis, or other interventions. Prompt control of the cause is crucial to decrease the risk of ischaemia or bleeding, thereby avoiding permanent disabilities or death. [1] Although  control of the causative factor was achieved within a median of only 30 hours in our study, ischaemic and haemorrhagic complications occurred in 8% and 12% of patients, respectively. Moreover, reversibility of the imaging abnormalities was partial in 43% and complete in 45% of patients. Several studies have suggest that toxaemia of pregnancy may be associated with better outcomes. [32] In our study, toxaemia of pregnancy was independently associated with a higher likelihood of a favourable day-90 outcome (GOS = 5). Maternal mortality after eclampsia is only about 1%. However, extra-neurological complications (placental abruption and HELLP syndrome) and neurological complications (ischaemia and haemorrhage) can occur. Another risk after eclampsia is the development of post-traumatic stress disorder. Hyperglycaemia was the last factor independently associated with the 90-day outcome in our patients. Hyperglycaemia, a common occurrence in neurocritical care patients [33], was significantly associated in several studies with increased mortality and impaired functional outcomes after events such as stroke [34,35], intracerebral haemorrhage [36], traumatic brain injury, [37] and spinal cord injury. [38] However, whether hyperglycae-mia is an independent prognostic factor or a marker for brain injury severity remains unclear. [33]. None of the clinical or neuroimaging features evaluated in our study was independently associated with the day-90 outcome. The two independent risk factors for a poor day-90 outcome were related to early management. Our results support a goal-directed strategy combining prompt symptomatic treatment and control of the causative factor as soon as it is identified, regardless of on-scene severity of the clinicoradiologic presentation.
Previous studies provided limited data on the management and outcome of PRES. The mortality rate reached 15% [8,9]. An assessment of functional impairment was reported in a single study, which found a median modified Rankin scale of 2.5 at discharge, indicating mild-to-moderate disability [8]. None of these studies focused on severe PRES. Moreover, the relative roles in fatal outcomes of the syndrome itself and of associated factors remained unclear. Our multicentre study accurately delineates the presentation, management, and determinants of functional outcomes in patients requiring ICU admission for severe forms of PRES.  with no follow-up imaging studies, 3 with persistent haemorrhagic abnormalities by follow-up imaging, and 3 with resolution of the haemorrhagic abnormalities Y According to the Logistic Organ Dysfunction (LOD) score [27] {The primary outcome measure was the score on the Glasgow Outcome Scale [39] (GOS) 90 days after onset of severe posterior reversible encephalopathy syndrome. A score of 1 indicates death; 2, a vegetative state (the patient is unable to interact with the environment); 3, severe disability (the patient is unable to live independently but can follow commands); 4, moderate disability (the patient is capable of living independently but unable to return to work or school); and 5, mild or no disability (the patient is able to return to work or school). A favourable outcome was defined as a score of 5 and an unfavourable outcome as a score lower than 5. The day-90 GOS score was known in all 70 patients. doi:10.1371/journal.pone.0044534.g001 Our study has several limitations. First, the extent to which our findings apply to the full spectrum of patients with PRES is unclear. Patients were included in 24 ICUs over a 10-year period, yielding a great variability in the radiologic technology used for diagnosis and follow-up. However the year of admission, as well as the center, were not determinants of functional outcome. However, despite the retrospective study design, the participation of 24 ICUs provided a broad picture of the management of severe PRES. Second, PRES was diagnosed based on a combination of clinical and neuroimaging features previously described as consistent with PRES. MRI was obtained in most of the patients, all neuroimaging studies were reviewed by two certified senior neurologists blinded to clinical data, and patients were included by consensus between the two neurologists. Consensus agreement between the two neurologists that performed independent blinded review of all available neuroimaging was strong with a value of 1 (considering a proportion of concordant classifications expected by chance of 0.05). Given the 7 cases discarded after neuroimaging expertise, kappa coefficient agreement between initial selection of cases and final enrolment was excellent with a value of 0.9 (considering a proportion of concordant classifications expected by chance of 0.05). Follow-up neuroimaging studies were lacking in 30% of patients, but this proportion was even higher in previous reports [6,8]. Third, the GOS in its structured form has been found valid, practical, and reliable [39], but indirect GOS evaluation via charts or physician interviews has not been studied. Qualitative data from structured interviews may better assess the cognitive and emotional burden after PRES. Fourth, A GOS score of 4 may be viewed as a favorable outcome. A GOS score of 4 is ''moderate disability'' in a ''patient capable of living independently but unable to return to work or school''. This category is associated with functional impairments, as ''some previous activities, either at work or in social life, are now no longer possible by reason of either physical or mental deficit''. Thus, in keeping with other studies of outcomes of neurological conditions in critically ill patients, we defined a poor outcome as a GOS score smaller than 5. [40,41] Fifth, even if global mortality rate was of 15.7%, assessment of direct imputability of PRES indicates a lower value of 5.7%. This interesting finding reinforces results of our multivariate analysis of factors associated with functional outcome in patients with severe PRES. Indeed, toxemia of pregnancy, as opposed to others causes, was associated with good outcome, underlying burden of co-morbidities in this syndrome. Moreover, reasons for death directly ascribable to PRES demonstrate the potential of aggravation in course of disease of patients presenting with severe PRES. Sixth, impact on 90-days functional outcome of other organ dysfunction than respiratory failure was not assessed during the whole ICU stay. Seventh, due to the retrospective design of the study, the GOS score was extracted from the hospital charts or determined by interviewing the patient's general practitioner or neurologist in 90% of patients. In the remaining 10% of patients (corresponding to latest inclusions), the GOS score was evaluated by the local investigator, who administered the scale directly to the patient during a phone interview. The day-90 GOS score was known for all study patients whereas the day-180 GOS score was known for 62 (88.6%) patients. Last, 90 days after ICU discharge may be too early to determine the final functional outcome. However, 90 days is the most widely used interval for assessing overall outcomes after brain injuries.

Conclusions
In conclusion, 90 days after severe PRES only 56% of survivors had achieved a good recovery and the outcome was predicted chiefly by factors available within a few hours after ICU admission. Among these factors, hyperglycaemia during the first 24 hours and delayed control of the cause may be amenable to improvement. These results should be confirmed in a large multicentre prospective study. Department, CH de Roubaix, Roubaix, France); Emmanuel Michelin (Neuro Radiology Department, CH de Roubaix, Roubaix, France).