Clinical Utility of Computed Tomography and Magnetic Resonance Imaging for Diagnosis of Posterior Reversible Encephalopathy Syndrome after Stem Cell Transplantation in Children and Adolescents

Abstract

Posterior reversible encephalopathy syndrome (PRES) is a clinical syndrome characterized by vision changes, altered mental status, and seizures, typically caused by an acute rise in blood pressure. PRES has been reported after hematopoietic stem cell transplantation (HSCT) in association with hypertension from calcineurin inhibitors and corticosteroids. The imaging evaluation of PRES after HSCT in children and young adults has not been well described. We performed a retrospective review of all HSCT recipients presenting to the intensive care unit with new neurologic symptoms. A neuroradiologist reviewed all radiologic images and compared computed tomography (CT) versus magnetic resonance imaging (MRI) findings indicative of diagnosis of PRES. Alternative imaging diagnoses explaining the patients’ symptoms were also recorded. Fifty-four transplant recipients were admitted to the intensive care unit with new neurologic symptoms. Thirty-nine percent (21 of 54) of subjects had imaging findings consistent with PRES, 24% (13 of 54) had imaging findings consistent with an alternative diagnosis, 9% (5 of 54) had a nonspecific finding, and 28% (15 of 54) had no acute imaging findings. PRES was diagnosed at a median of 49 days (interquartile range, 29 to 94) after HSCT. The presenting symptom for the majority of patients with PRES was seizures (86%), whereas 14% presented with acute encephalopathy. Ninety-five percent of subjects diagnosed with PRES (20 of 21) underwent a head CT as their initial imaging evaluation. CT scan was diagnostic of PRES in 40% (8 of 20). Subsequently, 16 patients underwent brain MRI with 12 additional patients being diagnosed with PRES on MRI. The median time elapsed between negative CT and a positive MRI examination was 20 hours (range, 3.6 hours to 9 days). CT serves as an excellent screening test for acute pathology, such as intracranial hemorrhage; however, it lacks sensitivity for the diagnosis of PRES. Patients with clinical symptoms suggestive of PRES who have a negative CT should be treated appropriately for PRES and should undergo MRI of the brain as soon as clinically stable to confirm the diagnosis.

INTRODUCTION

Hematopoietic stem cell transplantation (HSCT) remains the only curative option for a subset of patients with malignant and nonmalignant diseases [1]. Central nervous system complications, however, continue to be an important cause of morbidity and mortality in patients undergoing transplantation [2,3]. Posterior reversible encephalopathy syndrome (PRES) is a clinical and radiologic syndrome characterized by vision changes, acute encephalopathy, nausea and vomiting, focal neurologic deficits, and seizures [4]. PRES has been reported after HSCT in association with hypertension and renal dysfunction, particularly in patients receiving calcineurin inhibitors and corticosteroids [5–7]. Although PRES usually is self-limiting, sequelae such as epilepsy and coma may persist [8,9].

In addition to PRES, other pathologies may lead to neurologic symptoms after HSCT, including infection, primary disease recurrence, hemorrhage, ischemia, and post-transplantation lymphoproliferative disease (PTLD) [10]. The majority of physicians choose computed tomography (CT) as the initial imaging evaluation for patients who present with neurologic symptoms because of its wide availability, timeliness, and lack of need for sedation in younger children. The CT and magnetic resonance imaging (MRI) manifestations of PRES have been well described in a variety of disease backgrounds [8,11–20]. However, the diagnostic utility of CT compared with that of MRI for PRES has not been studied in adults or children.

MATERIALS AND METHODS

HSCT recipients who underwent transplantation from January 2004 through December 2013 requiring intensive care unit admission for acute changes in neurologic status, including seizures, encephalopathy, and visual disturbances, and who underwent neuroimaging evaluation (CT/MRI) were included in the study. After institutional review board approval, patients were identified using the stem cell transplant database followed by a query of the electronic imaging record and imaging report (Softek Illuminate; Softek Solutions, Prairie Village, KS) databases to determine whether neuroimaging was performed. Demographic and transplantation data collected included age, gender, diagnosis, stem cell source, HSCT conditioning regimen, diagnosis of graft-versus-host disease (GVHD) at the time of PRES, and clinical stage of GVHD at day +100. Additionally, symptoms at the time of PRES diagnosis were reviewed.

An attending pediatric neuroradiologist reviewed the radiology reports and neuroimaging. Based on the imaging findings, patients were classified into 1 of 4 categories: (1) imaging findings consistent with PRES, (2) alternative acute diagnosis, (3) nonspecific imaging findings possibly representing PRES with considerable possibility of alternative etiologies to explain the imaging findings, and (4) no acute imaging findings or normal study. The specific alternative acute diagnoses were tabulated.

Diagnostic Criteria for PRES

For CT/MRI study to be diagnostic for PRES, patients had to fulfill 1 of 2 imaging criteria: criterion 1, cortical/subcortical signal abnormalities predominantly located in locations reported in the literature as typical for PRES (between arterial territories) with relative symmetric involvement of both cerebral hemispheres [8] or criterion 2, signal abnormalities in atypical locations (brainstem/cerebellum), as previously reported in the literature with subsequent resolution of imaging findings following PRES-specific therapy[8].

Neuroimaging Review

Neuroimaging for all patients was reviewed and a determination of whether findings of PRES or alternative diagnoses were present was made. A CT/MRI was considered negative when both the radiology report and review of images by the pediatric neuroradiologist were considered negative. All CT imaging was performed without contrast. Images were reconstructed at 5 mm axial contiguous slices. No multiplanar reconstructions (coronal and sagittal planes) were performed. The CT images were reviewed independently of the subsequent MRI, if available. For those patients with a negative CT and positive MRI, the number of hours between the initial CT and the follow-up MRI examinations were recorded. Additional imaging information for all category 1 patients was tabulated, including regions of brain involved, associated hemorrhage and enhancement, or diffusion restriction (when applicable imaging sequences were available for review). All brain MRI included fluid-attenuated inversion recovery, which has been shown to be most sensitive for detecting PRES [11]. Contrast administration, diffusion-weighted imaging, and hemorrhage-sensitive imaging were performed variably and are reported in the results section.

RESULTS

Eight hundred thirty-eight patients underwent HSCT over this period. Sixty patients were admitted to the intensive care unit with neurologic symptoms and 54 (6.4%) patients underwent brain imaging at time of presentation (Figure 1). Thirty-nine percent (21 of 54) of patients had imaging findings consistent with PRES (category 1). Demographic data for these patients are presented in Table 1. PRES was diagnosed at a median of 49 days (interquartile range, 29 to 94) after transplantation; presenting symptoms were seizures (86%) and acute encephalopathy (14%). The median age at HSCT was 9.6 years (interquartile range, 5.5 to 12.6) with a majority (62%) being male. Eight patients had a marrow failure syndrome (38%), 7 of which underwent HSCT for Fanconi anemia. Seven patients had an immune deficiency (33%) and 5 patients had an underlying malignancy (24%). All 21 patients who developed PRES underwent allogeneic transplantation and the majority (86%) received grafts from an unrelated donor. Nine patients (43%) were diagnosed with GVHD before the diagnosis of PRES, and 9 patients (43%) had grade II to IV GVHD before day 100. The overall mortality in patients who developed PRES was 62% (8 of 21 survived). Ten patients died from GVHD, 2 from disseminated fungal infection, and 1 from disease relapse.

Figure 1.

Diagnostic imaging category of the 54 patients who underwent neuroimaging for new neurologic symptoms. All patients underwent neuroimaging at time of presentation to the pediatric intensive care unit. Category 1 PRES was diagnosed in 21 patients. One patient was diagnosed with MRI alone; 4 were diagnosed with CT alone; 4 were diagnosed with CT and MRI; and 12 patients had normal CT, but positive MRI. In category 2, 13 patients had an alternative diagnosis other than PRES; all 13 underwent CT and MRI. In category 3, 5 patients had nonspecific findings on neuroimaging; all 5 underwent CT and MRI. In category 4, 15 patients had no acute neuroimaging findings; CT was performed in all 15 patients, 5 received both CT and MRI.

Table 1.

Demographics of Patients Diagnosed with PRES (n = 21)

Total Patients	Value
Gender
Male	13 (62%)
Age at HSCT, median (interquartile Range)	9.6 yr (5.5–12.6)
Diagnosis
Malignancy	5 (24%)
Immune deficiency	7 (33%)
Bone marrow failure	8 (38%)
Benign hematology	1 (5%)
Stem cell source
Bone marrow	12 (57%)
Peripheral blood	4 (19%)
Cord blood	5 (24%)
Donor source
Related	3 (14%)
Unrelated	18 (86%)
Autologous	0
HLA match status
Fully matched	13 (62%)
Mismatched	8 (38%)
Conditioning regimen
Ablative	11 (52%)
Reduced intensity	10 (48%)
GVHD at time of PRES diagnosis (grade 1–4)	9 (43%)
Grade 2–4 GVHD at +100 days	9 (43%)
Symptoms at time of PRES diagnosis
Seizures	18 (86%)
Mental status change	3 (14%)

Data presented are n (%), unless otherwise indicated.

Specific findings, such as hypertension and thrombotic microangiopathy, are only available on patients who developed PRES after 2010.We were able to obtain blood pressure data on 11 of the 21 patients diagnosed with PRES (52%). All 11 patients had refractory hypertension, requiring at least 2 antihypertensive medications in the 24 hours before the onset of symptoms, 8 of the 11 (73%) were being treated with calcineurin inhibitors, and 9 of the 11 (82%) met diagnostic criteria for thrombotic microangiopathy.

All but 1 patient (20 of 21, 95%) diagnosed with PRES had a head CT scan done as a first imaging study, with 40% (8 of 20) of patients having CT findings diagnostic of PRES (Figure 2). MRI was performed in 16 of these 21 patients (76%) and all studies were consistent with PRES, including 12 patients with negative CT findings during initial evaluation (Figure 3). The median time between negative CT examination and a positive MRI examination was 20 hours (range, 3.6 hours to 9 days). Diffusion-weighted imaging was performed in all 16 patients with diagnosis of PRES who underwent MRI. Areas of diffusion restriction were identified in 4 of the 16 patients. Intravenous contrast was administered in 8 of the patients receiving MRI, 1 of which showed abnormal contrast enhancement. Hemorrhage-sensitive MRI sequences (susceptibility-weighted imaging or gradient echo) were performed in 11 of the 16 patients, 3 of which demonstrated intraparenchymal hemorrhage in the regions of signal abnormality. Hemorrhage was identified on CT in 2 of these 3 patients. Brain regions affected in the 22 subjects diagnosed with PRES were as follows: frontal (14 of 21, 66%), parietal-occipital (19 of 21, 90%), temporal (8 of 21, 38%), cerebellum (3 of 21, 14%), and brainstem (2 of 21, 10%). Ten of the 21 patients (48%) had follow-up neuroimaging: all showed complete or near-complete resolution.

Figure 2.

Four-year-old female with severe combined immune deficiency who developed seizures 5 days after HSCT. Axial CT images show large areas of decreased attenuation and loss of the gray-white junction in a nonvascular, watershed distribution, most consistent with PRES. No MRI was performed.

Figure 3.

Seven-year-old male who developed seizures 68 days after HSCT. (A) Head CT examination was normal. (B) Axial fluid-attenuated inversion recovery images from a brain MRI examination performed 4 hours later demonstrates multiple cortical and subcortical foci of signal abnormality consistent with PRES.

One quarter of patients (24%, 13 of 54) in this cohort had neuroimaging suggestive of an alternative diagnosis (category 2) including intracranial hemorrhage (3 patients), hydrocephalus (2 patients), metastatic disease (2 patients), acute infarction (2 patients), nonalcoholic Wernicke encephalopathy (1 subject), orbital abscess (1 subject), PTLD (1 subject), and nonspecific basal ganglia/thalamic lesions (1 subject). Eight of the 13 patients (62%) with acute alternative diagnoses had findings evident on CT, including intracranial hemorrhage, hydrocephalus, PTLD, infarction, and orbital abscess. Five patients (38%) with acute alternative diagnoses had imaging findings only seen on MRI, including metastatic disease, infarction, nonalcoholic Wernicke encephalopathy, and nonspecific basal ganglia/thalamic lesions. Five (9%, 5 of 54) had new nonspecific imaging findings (category 3). The remainder of patients (28%, 15 of 54) had normal or no acute imaging findings (category 4). Of these patients, 10 had CT only, and it is unknown if additional MRI imaging would have identified PRES.

DISCUSSION

CT is a good first test in HSCT patients with acute neurologic changes when intracranial hemorrhage, space occupying lesions, and hydrocephalus need to be ruled out quickly. However, our findings suggest that CT lacks sensitivity for diagnosis of PRES. Only 40% of patients diagnosed with PRES on MRI had radiologic evidence of PRES on CT.

Our study validates others’ findings and suggests that MRI may be more sensitive to detect microhemorrhage within the affected areas of the brain [21,22]. Despite a negative initial head CT, if clinical symptoms are suggestive of PRES, HSCT patients should receive prompt and aggressive hypertension management and anticonvulsant therapy, and a diagnostic MRI should be obtained when clinically feasible.

Limitations of this study include 10 of the 15 patients with no acute imaging findings (category 4) who underwent CT imaging only. It is possible that additional acute diagnoses would have been revealed if follow-up MRI had been performed in these patients. This retrospective study is a nonuniform evaluation of subjects with neurologic symptoms, leading to the possibility that some pathology, including PRES, was missed. Additionally, the time between CT and MRI examinations was variable, allowing the possibility that some findings of PRES evolved in between the CT and MR examinations. However, a number of false-negative CT examinations performed only a few hours before the positive MRI examinations support our observation that MRI is the preferred study for diagnosis of PRES (eg, Figure 3). MRI can identify subtle cortical/subcortical signal abnormalities characteristic of PRES better than CT. Finally, at the time of this study, CT examinations were routinely reconstructed only in the axial plane. At our institution, it is now standard of care to reconstruct head CT examinations in the sagittal and coronal planes, which may increase the sensitivity and accuracy of this examination for detection of PRES.

We propose that patients with acute neurologic symptoms after HSCT in the setting of hypertension or thrombotic microangiopathy receive aggressive hypertension management [23]. Patients with neurologic changes should receive an MRI if rapidly available; otherwise, a CT scan should be obtained to evaluate for acute pathology. If PRES is suspected despite a negative CT scan, or if PRES is found on CT, PRES-specific therapy should be initiated, including aggressive hypertension management and anticonvulsant therapy. MRI should be obtained when clinically stable (Figure 4).

Figure 4.

Diagnostic algorithm for patients with new neurologic symptoms after stem cell transplantation.

In summary, CT examinations are a good initial diagnostic test for HSCT patients presenting with acute mental status changes or seizures. CT often identifies an alternative diagnosis or can make a definitive diagnosis of PRES. However, our findings demonstrate that CT is insensitive compared with MRI for PRES. Patients with negative CT findings but clinical symptoms highly suggestive of PRES should undergo an MRI of the brain when clinically stable and receive aggressive hypertension control. MRI results diagnostic of PRES or alternative diagnoses may prompt physicians to make appropriate management changes regarding hypertension control, aid in decision-making regarding invasive diagnostic procedures (including lumbar puncture), and/or intensify ongoing empiric antimicrobial/antifungal therapy.