The Changing Face of Osmotic Demyelination Syndrome

Abstract

Objective

To describe the long-term outcomes of osmotic demyelination syndrome (ODS) in an updated cohort.

Methods

We performed a retrospective medical records review of cases of ODS at the Massachusetts General and Brigham and Women's Hospitals using International Classification of Diseases–9th edition codes and a text-based search for central pontine myelinolysis, extrapontine myelinolysis, and osmotic demyelination syndrome (1999–2018). Cases were individually selected based on patients having neuroimaging and symptoms consistent with ODS and no other potentially explanatory etiology. Modified Rankin scale (mRS) scores were extracted at prehospitalization, hospital discharge, 6 months post discharge, and the most recently available clinical visit.

Results

We identified 45 cases of ODS (mean age 48.4 years, range 0.07–75 years; 58% female patients). Common comorbidities included liver disease (27%, n = 12), alcoholism (44%, n = 20), and kidney failure (20%, n = 9). Twenty-nine percent of patients had a rapid correction of hyponatremia. Twenty-nine percent had other electrolyte abnormalities. Only 59% (24/41) of patients with complete electrolyte data had abnormalities that could explain their ODS. At the 6-month follow-up, 16% of the patients were dead and 60% of patients had minimal-to-no disability (mRS 0–2).

Conclusions

ODS has a diverse range of clinical presentations. Not all patients have electrolyte abnormalities. The prognosis is generally favorable, although 1 in 6 patients had died at 6 months, likely because of underlying disease states.

In 1959, a new disorder was described in 3 alcoholic and chronically malnourished adults with lesions of the pons.¹ These lesions were pathologically characterized by destruction of myelin and oligodendrocytes with relative preservation of axons. Each case was attributed to the small pontine lesion. Given the distinct pathologic findings, the term “central pontine myelinolysis” (CPM) was coined.

Today, the classical description of CPM is a syndrome in which a flaccid paralysis occurs after a rapid correction of hyponatremia.² However, this is a simplification of the disorder. The term CPM has been replaced with osmotic demyelination syndrome (ODS) to reflect the fact that both CPM and extrapontine myelinolysis (EPM) are a continuum of the same underlying pathology.

Unfortunately, most of the current literature does not reflect the current epidemiology of ODS. Of the 15 published studies that comprise more than 15 individuals, only 7 studies include any type of neuroimaging criteria for diagnosis and only 7 are from after the year 2000.^3–17 The largest study to date is a nationwide Swedish study that includes 83 individuals.⁴ In this study, 60% of patients were functionally independent 3 months after diagnosis. Although this work was a large step toward updating the literature on ODS, the authors relied on a historic diagnosis of ODS, which may have led to a heterogeneous sample.

Our study sought to describe baseline factors and long-term outcomes of individuals with ODS. We performed a retrospective review of our referral institution's medical records to identify cases of ODS. We used strict clinical criteria to depict a reproducible clinical phenotype and allow generalizable understanding of the prognosis of ODS.

Methods

Standard Protocol Approvals, Registrations, and Patient Consents

The study protocol was approved by the Partners Healthcare, Inc. Institutional Review Board (IRB) before study commencement. Individual written consent by each patient was waived by the IRB for this retrospective study without direct participant interactions.

Case Identification

We searched the Partners Research Patient Data Registry (RPDR) for unique case records diagnosed from June 1, 1999, to a date of administrative censoring of August 1, 2018. We used 2 methods to identify incident cases of ODS at the Massachusetts General Hospital and Brigham and Women's Hospital in Boston, MA. First, we used the International Classification of Disorders–9th edition (ICD-9) codes for demyelinating disorders of the CNS. Because the ICD-9 code for ODS was only adopted in 2015, many patients with other demyelinating disorders were found. Because of the high number of patients with this code, we then limited this search to patients given the diagnosis while hospitalized. We also did a text-based search for “central pontine myelinolysis,” “extrapontine myelinolysis,” and “osmotic demyelination.” Charts were reviewed manually by one of the authors (WF) for cases of ODS.

Case Inclusion and Exclusion Criteria

The following criteria were used to identify ODS: (A) either a tissue pathologic diagnosis of ODS or (B) all of the following criteria: (1) neuroimaging characteristics typical of CPM or EPM, (2) neurologic symptoms and signs, (3) ODS on the treating clinician's differential diagnosis, and (4) no other diagnosis was more likely.

Cases in which the diagnosis of ODS occurred at our institution but the precipitating event occurred at an outside institution were included. Cases with radiologic evidence of an acute pontine lesion (i.e., imaging study before hospitalization with no evidence of ODS) were included. We excluded individuals with “chronic CPM,” meaning pontine lesions that occur gradually over time and are asymptomatic.⁹ To do this, cases of pontine lesion of unclear chronicity and altered mental status were excluded unless the neurologic symptoms and signs persisted after the metabolic derangements were restored. If there was evidence that the lesion was acute (i.e., an imaging study within the past few months with no previous evidence of ODS), individuals were included. Similarly, cases of neurologic complaints that were clearly distinct from their imaging findings (such as migraines, seizures in the setting of alcohol withdrawal, or slowly progressive memory problems in an elderly individual) were excluded.

Date of Diagnosis

The date of ODS diagnosis was defined as the date of the first appearance of abnormality on brain MRI or head CT. Although this is likely later than the first symptomatic manifestations of CNS demyelination, this date was chosen to standardize the date of diagnosis across cases. Because many individuals were critically ill with multiple metabolic derangements, it was frequently unclear when their neurologic impairment began. For cases in which the MRI was performed months after the suspected incident, the date of diagnosis was estimated based on neurologic abnormalities during the initial hospitalization. When the diagnosis was made by pathology alone, the date of diagnosis was estimated based on the onset of neurologic symptoms.

Medical Records Review

Charts were independently examined by 2 coauthors (WF and ACV). A “low sodium” value was defined based on the lowest sodium value documented in the chart any time within the 3 weeks before diagnosis (i.e., nadir). “High sodium” was defined as the highest sodium value documented within the 3 weeks before diagnosis (i.e., peak). Sodium was corrected for hyperglycemia with serum glucose >200 mg/dL, when glucose measurements were available. If serum glucose measurements were not available, the uncorrected sodium was used. Other electrolytes, including serum potassium, were also extracted. The 24-hour and 48-hour sodium correction rates were determined in reference to the “low sodium” value. “Rapid correction” was defined as an increase greater than 8 mmol/L in 24 hours or greater than 18 mmol/L in 48 hours, based on the published expert consensus guidelines.¹⁸ Hypernatremia was defined as a value >155 mmol/L. Hyperglycemia was defined as >400 mg/dL. Modified Rankin scale (mRS) scores¹⁹ were ascribed for prehospitalization, time of discharge, 6 months postdischarge, and the most recent visit.²⁰ Children were scored according to the pediatric modifications for the mRS.²¹ Individuals were said to be “critically ill” if they were treated in an intensive care unit during their hospitalization when they developed ODS.

Data Analysis

Summary statistics were used to describe the prevalence of baseline demographic and clinical characteristics. Categorical variables were described by their proportions. Continuous variables were described using mean and standard deviation for normally distributed variables and median and interquartile range for variables with skewed distributions. Tests of 2 proportions were used to compare 2 proportions of interest. Exploratory regressions were performed using univariate models with either (1) death or (2) combined moderate-severe disability and death (mRS 4–6) as outcomes. Given the small sample size, there was no formal hypothesis testing of putative prognostic factors; however, directional associations were used to confirm presumed negative prognostic factors being associated with worse outcomes. A 2-tailed p-value of <0.05 was considered statistically significant.

Data Availability

The authors will provide deidentified data on reasonable request from a qualified investigator.

Results

There were 368 possible cases identified in the RPDR by ICD-9 code and 487 cases by the free text search. Of a total of 816 unique charts that were searched (because of the overlap between search results), 45 charts met the prespecified criteria for ODS.

Baseline Characteristics

Of the 45 identified cases, the average age was 48.4 years (standard deviation 18.6 years, range 0.07–75 years) (table 1). There were 4 cases less than 18 years old. Common baseline diseases included liver disease (27%, n = 12), alcoholism (44%, n = 20), and kidney failure (20%, n = 9). Five patients had both alcoholism and liver disease. At the time of diagnosis of ODS, 42% (n = 19) had a systemic infection, 69% (n = 31) were critically ill, 9% (n = 4) had a malignancy, and 13% (n = 6) had malnutrition. The median baseline mRS score before hospitalization was zero (25th and 75th percentiles: 0 and 1).

Table 1.

Baseline Demographic and Clinical Characteristics of Patients With Osmotic Demyelination Syndrome (n = 45)

ODS Characteristics

The most common symptom noted was altered mental status (62%, n = 28), followed by upper motor neuron pattern weakness in 58% (n = 26) (table 1). All except for one patient had neuroimaging. Ninety-one percent (n = 41) had an MRI, and 7% (n = 3) had a CT scan. Four patients had ODS diagnosed on pathology, 3 of whom additionally had premortem neuroimaging (figure 1). Forty-seven percent (n = 21) of patients had exclusively central pontine involvement. Thirteen percent (n = 6) had exclusively extrapontine involvement. Forty percent (n = 18) had both central pontine and extrapontine involvement.

Figure 1. Flowsheet of Diagnostic Modalities to Characterize Osmotic Demyelination Syndrome, 1999–2018.

Electrolyte Abnormalities

Forty-one of 45 patients had electrolyte data available. A total of 12 individuals had a “rapid” sodium increase (4 meeting criteria of >8 in the first 24 hours, 2 meeting criteria of >18 in the first 48 hours, and 6 meeting criteria for rapid correction at both 24 and 48 hours). Twelve additional individuals had other potentially explanatory electrolyte abnormalities (3 with sodium >155 mmol/L, 6 with hyperglycemia >400 mg/dL, and 3 with both hyperglycemia and hypernatremia), leading to a total of 59% (24/41, 95% confidence interval [CI]: 38%–69%) of participants with electrolyte data, who had electrolyte abnormalities that could potentially explain their ODS. Electrolyte data are summarized in table 2.

Table 2.

Electrolyte Data in Patients With Osmotic Demyelination Syndrome

Long-term Outcomes

At the time of discharge, 13% of individuals (n = 6, 95% CI: 3%–24%) were dead. The median mRS score of survivors increased from zero at baseline (25th and 75th percentiles: 0 and 1) to 3 at discharge (25th and 75th percentiles: 2 and 4).

At 6 months postdischarge, an additional participant had died, leading to a case fatality of 16% (n = 7, 95% CI: 4%–27%), although 10 individuals were lost to follow up by this time. The median mRS score improved to 1 (25th and 75th percentiles: 0 and 2.5). Follow-up mRS scores were available on 35 of 45 cases. Of those with follow-up data points at 6 months, 60% (n = 21/35) of patients had no limitations in their daily activities (mRS ≤ 2) (figure 2). Of the 7 participants who died by 6 months posthospital discharge, 3 (43%) had a history of liver disease, 1 (14%) had a history of alcoholism, 1 had a history of cancer, 2 (29%) were immunosuppressed, and 2 (29%) had a history of kidney failure. All 7 patients had overlapping comorbidities (e.g., a history of both liver disease and kidney failure).

Figure 2. Distribution of mRS Scores Before Hospitalization, at Discharge, and at 6 Months Postdischarge.

mRS = modified Rankin scale.

Of the 15 patients who had an mRS of 4 or 5 (significant disability) at 6 months postdischarge, 2 (13%) had a history of liver disease, 1 (7%) had a history of alcoholism, 2 (13%) were immunosuppressed, and 3 (20%) had a history of kidney failure. None had cancer.

We did not observe a difference in the 6-month mortality or combined severe disability and mortality between patients who had a rapid sodium change and those who did not. Of the 35 participants who had an evaluable mRS at 6 months, 10 had a rapid sodium increase and 25 did not. Of these 10 participants, 1 had an mRS of 0, 3 had an mRS of 1, 4 had an mRS of 2, 1 had an mRS of 3, and 1 died (mRS = 6). On univariate linear regression analysis, a rapid sodium increase was not a predictor of death at 6 months postdischarge (p = 0.29).

Discussion

We sought to examine the baseline factors, presentations, and outcomes of patients with ODS in a new cohort in the MRI era. Despite the fact that ODS is typically characterized as a disorder of electrolyte derangements, 41% of patients did not have an explanatory electrolyte abnormality. ODS is typically considered an iatrogenic condition of electrolyte derangements and corrections. Our results show that this common conceptualization of ODS is an oversimplification and that electrolyte abnormalities are not a sufficient explanation for some cases of ODS. Because ODS is considered a disorder of electrolyte levels, physicians may not consider it as part of a differential diagnosis without a noted rapid change in serum sodium. Such a simplification could be detrimental to patient outcomes.

Similarly, the typical description²² of ODS as a flaccid paralysis leading to locked in syndrome is rare. Only one patient in our cohort had this classic presentation. More commonly, a wide range of neurologic signs and symptoms occurred including altered mental status and upper motor neuron pattern weakness.

Consistent with previous studies,^3,4,10,13,17 60% of ODS cases had a history of alcoholism and liver disease. The prevalence of liver disease patients with ODS is lower in our cohort than in cohorts reported elsewhere. As an example, a 2012 study found that 86% of patients with CPM had a history of alcoholism.¹³ This may be due to the fact that we excluded cases of “chronic CPM.” “Chronic CPM” lesions are common in patients with a history of alcohol dependence, with one study reporting that 80% of asymptomatic patients with CPM lesions incidentally found on autopsy had a history of alcohol dependence.

In contrast to previous studies, we found high rates of patients who were immunosuppressed because of post-transplantation antirejection therapies or chemotherapy for malignancy. This may represent the increasing number of immunosuppressed individuals within our medical system or the changing demographic characteristics of hospitalized patients in the United States in general. However, because our study was observational, future studies are required to determine whether there is an association between immunosuppression and ODS and the possible mechanisms behind this finding.

Six months postdischarge, 16% of our patients were dead. It is classically cited that ODS will have a case fatality of approximately 1 in 3.¹³ However, mortality rates varied widely in previous studies. Of studies with more than 15 patients each, mortality rates ranged from 5.8%¹⁵ to 48%.¹² This wide range most likely relates to the heterogeneous definitions of ODS used and the heterogeneous populations studied.

Among cases (n = 35) with available follow-up data at 6 months postdiagnosis, 57% of patients had a favorable functional outcome and were not impaired in their activities of daily living. Unfortunately, due to the fact that 10 of our participants were lost to follow up at the 6-month point, these results are subject to follow-up bias. Our finding is similar to previous studies that estimated favorable functional outcomes in roughly 50% of survivors.^12,13,15

We were not able assess the neuroimaging findings of ODS in a standardized way. We are aware of no grading system for the severity of neuroimaging findings in patients with ODS. It would be useful for future—likely prospective—research to include MRI brain images that could be adjudicated at baseline and repeated at specific time points and given a severity ranking by neuroradiologists, based on standard, prespecified criteria.

Our study had several strengths. We used stringent criteria with a strict a priori definition of ODS with a physician review of each case. This sample was more likely to represent acute ODS and exclude chronic pontine lesions. This cohort is among one of the largest observed to date. Most of the other studies include autopsy-based cases. Given that 84% of our sample was alive at 6 months, the generalizability of autopsy-based series to living patients is debatable.

We also recognize several weaknesses in this retrospective and observational approach that is subject to several biases. Our sample is not population based. Electrolytes and full medical histories were not available for certain patients, and the electrolyte values reported were not systematically drawn at specified intervals. This limits the accuracy and reproducibility of our cohort. Performing a systematic study for a disease as rare as ODS is difficult, so this limitation occurs across comparable studies. We were unable to report prognostic factors for survival in our cohort because of its small size and the incomplete 6-month discharge vitality status, which was missing in 10 patients. Given the low sample size, we were not able to properly test for any hypotheses on outcomes and cannot confirm baseline variables that may hold prognostic value.

Although there are clear pathologic characteristics of ODS,^1,23 the brain MRI characteristics are less specific,²⁴ making the diagnosis vulnerable to clinicians' inferences.

The presence of imaging findings consistent with ODS may cause providers to defer further diagnostic evaluation. As imaging findings of ODS can lag behind clinical symptoms,²⁴ its diagnosis may be delayed if imaging alone is required for diagnosis. Finally, inappropriate diagnosis may lead to inappropriate counseling about long-term outcomes and withdrawal of care. Given the high rates of mortality and disability associated with ODS, this is very relevant. One study found that 31% of patients had life-sustaining measures withdrawn with the likelihood of severe cerebral disability cited as the main reason in 10/11 cases.¹³

Our study showed a gap between the classic view of ODS and the clinical spectrum of the disease. Despite the fact that it is typically characterized as a disorder of electrolyte abnormalities, a large proportion of our sample did not have significant electrolyte derangements. Many individuals had relatively mild neurologic symptoms. This discrepancy may lead to missed diagnoses and other inappropriate interventions.

Future research should involve larger sample sizes. A recent nationwide study in Sweden included 83 individuals with ODS. Unfortunately, they only used a historic chart-based diagnosis of ODS, so their sample may include individuals with pontine lesions of other etiologies. To obtain large samples with a priori diagnoses using consistent clinical criteria, collaboration across multiple centers will be necessary.

Finally, although we established guidelines for our own definition of ODS in this study, no consensus clinical standard exists. Strict clinical criteria for ODS are necessary for consistency across reports. Only with such consistency will we be able to provide a reproducible clinical prognosis for patients with ODS.

Appendix. Authors

Study Funding

No targeted funding reported.

Disclosure

The authors report no disclosures relevant to the manuscript. Full disclosure form information provided by the authors is available with the full text of this article at Neurology.org/cp.