Clinical and Radiologic Correlations of Central Pontine Myelinolysis Syndrome

Abstract

OBJECTIVE: To characterize clinical and radiologic features of patients with central pontine myelinolysis (CPM) and identify variables that predict outcome.

PATIENTS AND METHODS: We retrospectively studied patients diagnosed as having CPM identified by a search of Mayo Clinic medical records from January 1, 1999, through December 31, 2010. Diagnosis was made by clinical and radiologic features. Favorable outcome was defined by a modified Rankin Scale score of 2 or lower. Volume of signal abnormality on brain magnetic resonance imaging (MRI) was quantified by a neuroradiologist blinded to outcomes. Wilcoxon rank sum tests were used to assess association between volume of signal abnormality and outcomes at discharge and last follow-up.

RESULTS: Of 24 patients, 14 (58%) had only CPM, and 10 (42%) had extrapontine involvement. Hyponatremia was documented in 18 patients (75%), with median sodium nadir of 114 mmol/L. Eighteen patients (75%) had alcoholism, and malnutrition was documented in 12 (50%). Presenting symptoms included encephalopathy (n=18 [75%]), ataxia (n=11 [46%]), dysarthria (n=7 [29%]), eye movement abnormalities (n=6 [25%]), and seizures (n=5 [21%]). Favorable outcome was seen in 15 patients (63%) at last follow-up. Findings on initial brain MRI were normal in 5 patients, but all MRI scans were abnormal with serial imaging. The volume of radiologic signal abnormality was not associated with outcome at discharge or last follow-up (P=.67 and P=.37, respectively).

CONCLUSION: Clinical outcome in patients with CPM is not predicted by the volume of radiologic T2 signal abnormality on MRI or the severity of hyponatremia. Serial brain imaging is of value because a substantial proportion of patients have normal findings on initial MRI.

CPM = central pontine myelinolysis; DICOM = Digital Imaging and Communications in Medicine; MRI = magnetic resonance imaging; mRS = modified Rankin Scale; ROI = regions of interest

In 1959, Adams et al¹ described 4 cases of quadriplegia and pseudobulbar palsy and coined the term central pontine myelinolysis (CPM). Subsequently, this disorder was found to involve extrapontine sites as well.² In 1983, Laureno³ demonstrated the importance of serum sodium fluctuations as a causative factor of CPM in dogs by rapidly correcting induced hyponatremia. In a report of 58 cases of CPM, 27 were confined to the pons, 18 involved the pons and extrapontine areas, and 13 were only extrapontine, with the cerebellum and lateral geniculate nucleus being the most frequent extrapontine sites.⁴

Clinically, CPM presents with a biphasic course of encephalopathy or seizures with improvement followed by deterioration, including quadriparesis, dysarthria, dysphagia, and occasionally oculomotor abnormalities, depending on the extent of the lesions.⁵ Few studies have assessed the correlation of clinical or radiologic factors and clinical outcome. One recent study found that higher Glasgow Coma Scale sum scores, less severe levels of hyponatremia, and absence of hypokalemia predicted favorable outcome.⁶ Another study reported no correlation between the size of the pontine lesion on brain magnetic resonance imaging (MRI) and clinical course; however, besides a single case description, no details or data were provided to support this conclusion.⁷ Therefore, we performed this study to evaluate whether any clinical or laboratory data could predict functional outcome in patients with CPM, with special interest in determining whether more extensive pontine T2 signal abnormality was an indicator of poor outcome in these patients.

PATIENTS AND METHODS

We performed a search for the terms central pontine myelinolysis, extrapontine myelinolysis, and/or osmotic demyelination using the Mayo Clinic medical records linkage system to identify patients with possible CPM who were evaluated at our institution between January 1, 1999, and December 31, 2010. This study was approved by the Mayo Clinic Institutional Review Board. Given our search, any case in which CPM was mentioned by the physician was identified (N=51). These medical records were reviewed to determine whether the patient’s clinical and radiologic features were consistent with CPM. Twenty-seven cases were excluded because CPM was considered in the differential diagnoses but was not the final diagnosis or patients had a remote history of probable CPM with no details of the clinical or radiologic features. Clinical data were abstracted in all 24 cases that met our inclusion criteria, including age, sex, presence of encephalopathy, seizures, extrapyramidal findings, dysarthria, ataxia, presence of hypertension, hyponatremia (serum sodium concentration <135 mmol/L; to convert to mEq/L, multiply by 1.0), correction of sodium (change by at least 10 mmol/L within 48 hours), diuretic use, hypokalemia (serum potassium concentration <3.5 mmol/L; to convert to mEq/L, multiply by 1.0), endocrinopathy, presence of diabetes mellitus, alcoholism, liver transplant, and sepsis. Laboratory data were recorded, including sodium concentration at presentation, sodium nadir, sodium peak within 48 hours after nadir, albumin level, and glucose level.

MRI Analysis

An American Board of Radiology–certified neuroradiologist (T.J.K.) who was blinded to patient outcome and clinical characteristics reviewed each brain MRI. The volume of signal abnormality (microliter) attributed to osmotic myelinolysis in the brainstem and in the deep gray structures if present was quantified on serial axial T2-weighted images (4- to 5-mm section thickness from spin echo or fast spin echo pulse sequences) using Analyze 9.0 software (Biomedical Imaging Resource, Mayo Clinic, Rochester, MN). DICOM (Digital Imaging and Communications in Medicine) standard images were imported into the regions of interest (ROI) module of this software program, and the “Auto Trace” function was used to draw ROI that conformed precisely to the signal abnormality seen on each involved axial slice through seed placement and growing of the ROI through thresholding (Figure 1). Manual editing of these semiautomated ROI was then performed as needed. Signal abnormality attributed to osmotic myelinolysis of extrapontine structures was also recorded qualitatively.

FIGURE 1.

Example of quantification of signal abnormality on sequential axial T2-weighted images in the pons (A and B) and midbrain (C) totaling 3194 μL. Volumes of signal abnormality were obtained by tracing regions of interest around areas of signal abnormality with a semiautomated technique involving seed placement and growing through thresholding.

Statistical Analyses

Categorical variables were expressed as number (percentage). Continuous variables were expressed as median (minimum, maximum). A modified Rankin Scale (mRS) score of 2 or lower was considered a favorable outcome, and an mRS score greater than 3 was considered a poor outcome. Because of small sample size and skewed data (ie, non-Gaussian distribution), comparisons of key variable of interest (volume of radiologic signal abnormality) between favorable vs poor outcomes at discharge and at last follow-up were made using Wilcoxon rank sum tests. All tests were 2-sided, and P<.05 was considered statistically significant. Analysis was performed using SAS software version 9.1 (SAS Institute, Cary, NC).

RESULTS

Of 24 patients, 13 (54%) were men, and the median age of the patients was 56 years (35, 92). Fourteen patients (58%) had only CPM, and 10 (42%) had extrapontine involvement. The suspected underlying causes of CPM were rapid correction of sodium (n=16 [67%]), hyperosmolar hyperglycemia (n=1 [4%]), and hyperammonemia (n=1 [4%]) and were unclear in 6 patients (25%). Eighteen patients (75%) had chronic alcoholism, and 12 (50%) had documented malnutrition with a median albumin level of 2.6 g/dL (1.6, 3.1) (to convert to g/L, multiply by 10). Eleven patients (46%) had chronic hypertension, 8 (33%) were taking diuretic medications, 7 (29%) had sepsis, 4 (17%) had diabetes mellitus, and 1 (4%) had a combined liver and kidney transplant. Presenting clinical symptoms included encephalopathy (n=18 [75%]), ataxia (n=11 [46%]), dysarthria (n=7 [29%]), eye movement abnormalities (n=6 [25%]), seizures (n=5 [21%]), extrapyramidal symptoms (n=2 [8%]), and chorea (n=1 [4%]). Among 18 patients with documented hyponatremia, 7 (39%) had associated hypokalemia. Median sodium level at presentation in those with hyponatremia was 118 mmol/L (95, 135), and median sodium nadir was 114 mmol/L (95, 129). The median increase in serum sodium over 48 hours was 21 mmol/L (10, 28).

Radiologic Findings

Of the 10 patients (42%) who underwent head computed tomography, initial findings were hypodensity in the pons in 3 (30%). All patients underwent brain MRI. The images of 22 patients were available electronically for our review. DICOM standard images were not available for quantitative analysis in 1 of these patients. In 5 patients (23%), the initial MRI showed no evidence of pontine signal abnormality; on repeated imaging, abnormalities were evident in all 21 patients. The pons was involved in all 22 patients (100%), followed by the thalamus (n=8 [36%]), midbrain (n=6 [27%]), cortical gray matter (n=3 [14%]), hippocampus (n=3 [14%]), caudate (n=2 [9%]), putamen (n=2 [9%]), and middle cerebral peduncle (n=2 [9%]). Of 19 cases in which diffusion-weighted sequences were obtained, 10 (53%) demonstrated evidence of restricted diffusion. Four (29%) of 14 patients who underwent gadolinium contrast had enhancing lesions. More remote MRI scans obtained before patients’ CPM episodes were available to review for 6 patients (27%), and no marked pontine signal abnormality was seen. In 19 patients (86%), there was no or mild associated hemispheric small vessel ischemic white matter changes.

Outcomes and Associations

Ten patients (42%) achieved a favorable outcome (mRS score ≤2) at the time of hospital discharge; 14 patients (58%) had a poor outcome (mRS score >2), 2 of whom died. At last clinical follow-up at a median of 22 months (0, 125), 15 patients (63%) had achieved a favorable outcome, 4 had an mRS score of 3, and 5 had died. The clinical and radiologic characteristics in patients with CPM in association with outcome are summarized in the Table. Radiologically, there was no association between the volume of signal abnormality and clinical outcome either at time of discharge from the hospital (P=.67) or at last clinical follow-up (P=.37) as assessed by Wilcoxon rank sum test (Figures 2 and 3).

TABLE.

Clinical and Radiologic Characteristics in Patients With Central Pontine Myelinolysis Syndrome^a

FIGURE 2.

Example of lack of association between pontine signal and clinical outcome. Axial T2-weighted fluid-attenuated inversion recovery images demonstrate (A) a mild area of pontine hyperintensity (1825 μL total volume of abnormality) in a 67-year-old man with significant ataxia and hyperreflexia who remained moderately disabled (mRS [modified Rankin Scale] score, 3), and (B) severe pontine involvement measuring a total of 6431 μL in a 45-year-old woman with dysarthria and ataxia who achieved a favorable recovery (mRS score, 2).

FIGURE 3.

Lack of association between quantified radiologic volume of signal abnormality (microliter) attributed to osmotic myelinolysis on T2-weighted brain magnetic resonance images and clinical outcome at last follow-up. A score of 2 or lower on the modified Rankin Scale was considered favorable.

DISCUSSION

In this retrospective series of patients with CPM and quantified MRI lesions, we found that the extent of T2 signal abnormality was not associated with clinical outcome. Although a prior study mentioned a similar clinicoradiologic disconnect (based on a single patient whose MRI results did not change despite clinical improvement),⁷ previous studies have not attempted to correlate severe radiologic abnormalities on MRI with clinical outcome. A persistently abnormal MRI signal after clinical improvement in CPM has been previously appreciated and has been attributed to residual gliosis.⁸ Our finding is of vital importance to clinicians who care for patients with CPM because this information can help them avoid a premature pessimistic prognostication based on severe radiologic abnormalities.

Central pontine myelinolysis is classically attributed to rapid correction of hyponatremia, particularly in the setting of malnutrition or alcoholism. The findings in our series are consistent with this because the cause of CPM was rapid correction of sodium in 67% of our patients, and fluctuating serum osmolarity (hyperglycemia and hyperammonemia) accounted for the other known causes. The disorder is rare: in one pathologic series of 3000 unselected patients, 0.5% had evidence of CPM.⁹ As a result of its rare occurrence, outcome prediction in patients with CPM has not been well studied, and only one study has reported potential prognostic factors. One recent study found that higher Glasgow Coma Scale sum scores, less severe levels of hyponatremia, and absence of hypokalemia predicted favorable outcome.⁶ Only the severity of hyponatremia was found to be an independent predictor of outcome in the multivariate analysis. In one prospective MRI study of 13 patients with hyponatremia that was corrected over 24 hours, the rate of correction was significantly faster in those who developed pontine lesions.¹⁰ These findings taken together support the premise that the rate of sodium correction is more important than the nadir sodium concentration.¹¹

Our study corroborated the finding that outcomes in patients with CPM are not as poor as traditionally believed. Although early reports emphasized very poor outcomes,^1,4,12 several studies have since demonstrated that this is not inevitable. In our series, a favorable outcome was seen in 60%, and overall mortality was 8% in the acute setting. Similarly, in a large series of 34 patients with CPM, 65% achieved a good or moderate outcome (no functional deficit or independence despite minor deficits), and a poor outcome was seen in 27%, including a 6% mortality rate.⁷ Another recent study of 25 patients found that 46% achieved a favorable outcome.⁶ The increasing availability of MRI scans aiding in early diagnosis and improvements in intensive care treatments are likely contributing factors to the noted improvements in outcomes.

We also found that serial brain imaging is useful for cases of high clinical suspicion, as results of initial imaging may be unrevealing. The first MRI in 5 of our patients (21%) showed no pertinent abnormalities; however, all 21 patients had characteristic pontine signal abnormality on T2-weighted images on repeated imaging. This finding is in agreement with those of a recent study that reported that early MRI findings were normal in 25% of cases.⁶

The current study has several limitations. Lack of availability of laboratory information in some patients prevented us from reaching a conclusion regarding the association between nadir sodium level and its rate of correction and clinical outcome. Although we did not find lesion size statistically associated with clinical outcome, we cannot exclude the possibility of a type II error because the study may be underpowered. Still, the absence of any hint of correlation between the extent of pontine lesions and the likelihood of clinical recovery when evaluating our cases individually or as a group supports our conclusion. The central pons is a common site of chronic small vessel ischemic disease in older individuals and manifests radiologically as signal abnormality on T2-weighted images. This could potentially confound our results, but we consider this to be highly unlikely because of the relatively young age of our population and the dense, confluent, geographic characteristics of signal abnormality seen (as opposed to the usually more patchy changes produced by chronic small vessel disease). Furthermore, 27% of patients had prior MRI scans with no evidence of marked pontine signal abnormality, and associated hemispheric small vessel ischemic white matter changes were uncommon (86% had none or mild). A major strength of our study is that a neuroradiologist blinded to patient outcomes identified and quantified the radiologic abnormalities.

CONCLUSION

Neurologists are often asked to provide a prognosis after CPM because the clinical presentation can be catastrophic. Our study demonstrates that, although MRI is helpful in diagnosing CPM, the extent of pontine signal abnormality should not be considered a prognostic indicator. Patients with extensive pontine lesions can have a complete recovery. Further research is needed to help determine what factors might be important determinants of outcome after CPM.