Delayed posthypoxic leukoencephalopathy: a case series and review of the literature

Background Delayed posthypoxic leukoencephalopathy (DPHL) is a rare and underrecognized entity where patients manifest a neurological relapse after initial recovery from an acute hypoxic episode. We sought to describe the magnetic resonance imaging (MRI) findings in a group of patients with DPHL and review the available literature. Methods Retrospective case series including patients who presented with neurological and/or psychiatric symptoms after recovery from an acute hypoxic episode. The history and clinical presentation were reviewed from the electronic medical records. MRI scans were evaluated from the picture archiving and communication system. We performed a comprehensive review of the English medical literature for prior published cases of DPHL and describe the key imaging findings that have been reported related to this condition. Results A total of five patients were identified, including four patients with respiratory failure due to drug overdoses from benzodiazepines, opioids, and/or barbiturates, and one patient who presented after cardiopulmonary arrest due to pulmonary embolism. All patients showed diffuse, extensive, and confluent white matter signal abnormalities including prominent restricted diffusion, extending to the subcortical white matter and respecting the U-fibers. There was no gyral edema or contrast enhancement. In one case histopathology was available, which highlighted patchy subcortical myelin loss with sparing of U-fibers and demonstrated prominent macrophage/microglial inflammation with extensive axonal damage. Of the other four patients, two were at their neurological baselines and two had persistent neurological deficits at the time of discharge. Conclusions The described constellation of MRI findings is highly suggestive of DPHL in the appropriate clinical setting.


Introduction
Delayed posthypoxic leukoencephalopathy (DPHL) is a rare and underrecognized entity characterized by neurological relapse following a period of clinical stability or improvement after an episode of hypoxia. Patients usually present between one and 4 weeks after the initial event with relatively lucid intervening periods of variable lengths. While the exact mechanism behind its delayed manifestation has not been elucidated, DPHL is considered a distinct process from various direct causes of acute leukoencephalopathy such as toxic and metabolic injury. The majority of DPHL cases reported to date have been associated with carbon monoxide intoxication, where the incidence is approximately 2.8% (Choi 1983). However, delayed neurological sequelae have also been documented in other causes of hypoxia, particularly due to respiratory failure in drug overdoses (Rozen 2012;Salazar and Dubow 2012;Meyer 2013).
Histopathologically, DPHL is characterized by widespread demyelination with axonal preservation (Gottfried et al. 1997). Diffuse and confluent white matter changes are present on magnetic resonance imaging (MRI), most notably with extensive, symmetric, and often striking restricted diffusion (Molloy et al. 2006). Such a pattern is distinctly different from that seen in acute hypoxic-ischemic injury in the adult, which involves predominantly the gray matter structures (Huang and Castillo 2008).
Although DPHL as a clinical phenomenon has been known for many years, the presence of extensive restricted diffusion has only been recognized after the relatively recent advent of diffusion-weighted imaging (DWI) and probably constitutes the most remarkable MRI feature. To date, the literature regarding MRI findings is relatively scarce and consists of scattered case reports on patients with various etiologies of hypoxia and two small series of DPHL after carbon monoxide poisoning (Kim et al. 2003;Hsiao et al. 2004). Herein, we describe the MRI characteristics of DPHL in a series of five adult patients, the majority of whom presented following drug overdoses, with histopathologic correlation in one case. We also offer an extensive review of the literature including cases where MRI was performed.

Case series
This retrospective case series was approved by our institutional review board which waived requirement for informed consent. Adult patients (≥18 years of age) who fulfilled the following criteria were included in the study: (1) neurological deterioration caused by an initial hypoxic event; (2) subsequent clinical improvement with return to (prehypoxic event) baseline or near baseline; and (3) neurological relapse or new neurological or psychiatric symptoms following clinical improvement. Patients with continued deterioration after the initial hypoxic event without a lucid period or a clear relapse or with an alternative explanation for neurologic deterioration were excluded.

Data collection and analysis
The electronic medical records of each patient were reviewed to determine the clinical presentation, history, etiology of hypoxia, time to neurological relapse, and clinical manifestations during relapse. MRI studies were extracted from our picture archiving and communication system. Time from the initial hypoxic event to identification of diffuse white matter abnormalities on MRI (FLAIR and DWI) was recorded. The following imaging characteristics were visually analyzed: morphology (patchy or homogeneous), symmetry, relative extent of FLAIR and ADC abnormalities, spared structures, presence of mass effect or gyral edema, contrast enhancement, or hemorrhage.

Histopathology
For the autopsy case, the brain was fixed in formalin for 2 weeks. Following brain cutting representative sections were taken for microscopic assessment. Histochemical staining was done with hematoxylin and eosin, with Luxol fast blue added to label myelin. Immunohistochemistry was performed to detect expression of SM31 (Sternberger) for neurofilaments and CD68 (Ventana) for macrophages.

Literature review
We queried the PubMed database using the following terms for articles written in the English language: "delayed hypoxic encephalopathy", "delayed hypoxic leukoencephalopathy", "delayed leukoencephalopathy", and "delayed hypoxic-ischemic leukoencephalopathy", as well as various permutations substituting "reversible" for "delayed" and "post-hypoxic" for "hypoxic" and "post-anoxic." All articles were reviewed for redundancy of patients and only those who had a clear delayed presentation fulfilling the criteria above and who underwent MRI were included.

Patient characteristics
A total of five patients (three men and two women) fulfilled criteria for inclusion in this case series. Median age (interquartile range) at presentation was 63 years (59-64). All patients were 59 years or older except for one patient who was 32 years old.

Clinical presentation and course
Four patients were brought to our facility following an episode of respiratory failure due to drug overdose with opioids, benzodiazepines, and/or barbiturates. Three of them were found unresponsive and one had altered mental status in the setting of respiratory failure. One of five patients developed pulmonary embolism at home 8 days after being discharged for bowel surgery and was brought to the hospital in cardiorespiratory arrest. After this initial admission for respiratory failure/hypoxia, all patients were discharged from the hospital at their baseline or near baseline. Relapse was made manifest with neuropsychiatric symptoms such as erratic behavior, ataxia, urinary and/or fecal incontinence, delusions, akinetic mutism, and deficits in executive functioning including memory and attention. Two patients presented with pyramidal signs at relapse, consisting of triple flexion response on plantar stimulation (patient 2) and a pronator drift (patient 3). Other symptoms included increased tone, tremors, and cogwheel and leadpipe rigidity. Median time to relapse was 23 days (14-32). One patient continued to worsen after relapse and died 24 days after readmission. The other four patients progressively improved and were discharged at a median 24.5 days (21.5-33.75) after relapse. Two out of four patients were at their neurological baselines at discharge. The other two patients had shown significant improvement but had persistent deficits (Table 1).

MRI findings
Median time to identification of white matter abnormalities since the initial hypoxic event was 40 days (30-50). All cases demonstrated extensive and confluent T2 and FLAIR hyperintensity involving predominantly the periventricular white matter and centrum semiovale, bilaterally and in a symmetric fashion (Fig. 1). In two out of the five patients, restricted diffusion matched the extent of the T2-FLAIR hyperintensity, while in three patients the restrictive abnormalities were relatively less extensive. The white matter lesions were confluent and homogeneous in two patients. Three patients had evidence of more heterogeneous, patchy lesions which still followed an overall symmetric distribution. In all patients the T2 abnormalities involved the subcortical white matter but spared the U-fibers (Fig. 2), and were confined to the supratentorial brain without affecting the basal ganglia, thalami, brain stem, or cerebellum. There was no gyral edema or sulcal effacement, and there were no areas of contrast enhancement following the intravenous administration of gadolinium (Table 2). These findings were new in four out of the five patients in whom baseline MRI studies from their initial presentation were available for review. In one out of five patients, we did not have the initial imaging studies from an outside institution; however, these reportedly did not show an acute abnormality. Initial baseline MRI on patient number four showed three small subacute-appearing embolic infarcts in the supratentorial brain, without the confluent and symmetric white matter abnormalities that were seen on an MRI study performed 9 days later. Of the other patients who had a baseline study, two demonstrated chronic white matter ischemic changes and one did not show any significant findings.

Histopathologic findings
Autopsy examination of the brain in case 5 demonstrated no significant edema or sulcal effacement. Microscopically, there was extensive white matter injury with myelin loss and axonal swelling, as well as abundant reactive astroglia, in a mildly vacuolated background neuropil. These changes involved the subcortical white matter in a patchy distribution, although the U-fibers were preserved (Fig. 3).

Discussion
The gray matter of the adult brain demonstrates selective vulnerability to acute hypoxic-ischemic injury, which is triggered by a complex cascade of cellular events resulting in glutamate excitotoxicity, release of free radicals, and apoptosis (Won et al. 2002). Hypoxia preferentially affects the basal ganglia, thalami, neocortex, and hippocampus, with relative sparing of the brainstem, and results in a consistent pattern of abnormalities on MRI which are most evident on DWI within the first few hours following injury (Huang and Castillo 2008). On the contrary, white matter is relatively resilient to the effects of hypoxia, with the exception of carbon monoxide poisoning, which can cause acute demyelination preferentially damaging the globi pallidi and subcortical white matter (Adams et al. 2000).
In the setting of drug overdose, there are two principal mechanisms by which white matter can be injured. First, a form of spongiform leukoencephalopathy with intramyelinic vacuolization has been described from inhalation of the pyrolysate vapors produced after heating heroin in a practice known as "chasing the dragon", which primarily presents as a cerebellar syndrome causing ataxia, dysarthria, and bradykinesia (Won et al. 2002;Keogh et al. 2003). White matter injury in heroin inhalation encephalopathy favors the cerebellum, brainstem, posterior cerebral white matter, and posterior limb of the internal capsule (Keogh et al. 2003). This is in contradistinction to a second mechanism consisting of delayed demyelination, seen in DPHL, which occurs weeks after the initial hypoxic event and typically spares the cerebellum, brainstem, and basal ganglia. This particular distribution was evident in our study and is also concordant with the MRI findings of DPHL in the majority of published case reports and series. Findings reported in the literature are described in detail in Table 3 (Weinberger et al. 1994;Lee and Lyketsos 2001;Arciniegas et al. 2004;Molloy et al. 2006;Shprecher et al. 2008;Mittal et al. 2010;Wallace et al. 2010;Nzwalo et al. 2011;Huisa et al. 2013;Meyer 2013;Tormoehlen 2013;Geraldo et al. 2014).
On histopathologic examination, DPHL demonstrates widespread demyelination with axonal sparing as well as macrophages and reactive astrocytes (Plum et al. 1962;Gottfried et al. 1997). The U-fibers and cerebral cortex are noticeably spared (Plum et al. 1962). As opposed to the spongiform leukoencephalopathy seen in heroin pyrolysate (Wolters et al. 1982;Kriegstein et al. 1999), there is usually no intramyelinic vacuolization in DPHL (Plum et al. 1962;Gottfried et al. 1997).
To date, the pathophysiology of DPHL remains elusive. A proposed mechanism relates to the fact that the turnover rates for some myelin-related proteins range between 19 to 22 days, which is close to the average time for clinical relapse after initial injury (Meyer 2013). This would be consistent with our study, where the median time to neurological relapse was 23 days. However, while this is plausible, it would not explain why DPHL is such an uncommon phenomenon. Additionally, there have been reports of patients with decreased levels of arylsulfatase A, which is deficient in metachromatic leukodystrophy, suggesting that this could represent a predisposing factor (Weinberger et al. 1994;Gottfried et al. 1997). However, levels of this enzyme in other reported cases have been normal (Salazar and Dubow 2012). We did not measure arylsulfatase A in our patients. It is also worth mentioning that DPHL appears to occur after mild-to-moderate episodes of hypoxia, as severe hypoxia would likely result in acute hypoxic-ischemic injury with typical damage to gray matter structures.
In our series, the white matter abnormalities were extensive, bilateral, and symmetric, and invariably involved the subcortical white matter while preserving the U-fibers. In particular, a remarkable imaging finding was the presence of extensive restricted diffusion, which has been partially described in some cases Arciniegas et al. 2004;Molloy et al. 2006;Shprecher et al. 2008;Lou et al. 2009;Betts et al. 2012;Salazar and Dubow 2012;Huisa et al. 2013). The majority of reports available, however, do not include DWI sequences (Hori et al. 1991;Weinberger et al. 1994;Gottfried et al. 1997;Lee and Lyketsos 2001;Hsiao et al. 2004;Mittal et al. 2010;Wallace et al. 2010;Nzwalo et al. 2011;Rozen 2012;Choi et al. 2013;Meyer 2013;Tormoehlen 2013;Figure 2. Cropped and digitally magnified region from Fig. 1E demonstrates that the white matter signal abnormality (asterisk) involves the subcortical white matter but spares the U-fibers which appear as a curvilinear dark band (arrowhead). The adjacent gray matter is also visualized (arrow).  Geraldo et al. 2014), and those that do, lack the level of detail that we include in our series, specifically in terms of spared structures, symmetry, and morphology of the signal abnormalities, and presence of T2-FLAIR/ADC mismatch. We have shown that the FLAIR abnormalities were more extensive than the areas of restricted diffusion in three of our patients, although both showed a similar distribution and were symmetric. These findings were not present on baseline MRI, which was available in 80% of our cases, further substantiating the fact that such changes constitute a delayed manifestation rather than a direct effect of acute injury. While none of the studies in the literature have reported signal abnormalities in the cerebellum or brainstem, there are a few instances where lesions were present in the basal ganglia, presumably related to the hypoxic injury (Hori et al. 1991;Gottfried et al. 1997;Kim et al. 2002;Hsiao et al. 2004;Lou et al. 2009;Betts et al. 2012;Rozen 2012;Salazar and Dubow 2012;Choi et al. 2013). Specific injury to the globi pallidi has been described in several cases, not only in the setting of carbon monoxide poisoning (Kim et al. 2003;Hsiao et al. 2004), but also following drug overdose and arrest after massive hemorrhage (Gottfried et al. 1997;Lou et al. 2009;Betts et al. 2012;Rozen 2012;Salazar and Dubow 2012). The authors of one case report describe isolated injury to the basal ganglia in a patient with delayed encephalopathy after strangulation (Hori et al. 1991). However, this is the earliest case with MRI and it is possible that subtle abnormalities may have been missed. Additionally, no DWI was available at that time. A delayed-onset dystonia following anoxic injury, without white matter abnormalities and progressive over time, has also been reported in two cases (Kuoppamaki et al. 2002).    The lack of gyral edema in our series also supports a delayed presentation rather than acute toxic or metabolic injury. The absence of contrast enhancement argues against an active inflammatory or demyelinating process, which is possible if myelin injury has already occurred. None of the reviewed publications described contrast enhancement, and if such is present an alternative diagnosis should be sought. Additionally, the clinical presentation in DPHL is different from that of acute hypoxia, with most patients showing bizarre behavior, akinetic mutism, psychomotor retardation, and deficits of executive functioning. The presence of pyramidal signs as well as Parkinsonism and other movement disorders is also relatively common. We believe that, in the appropriate clinical setting, the constellation of MRI findings described above is highly suggestive of DPHL, which usually warrants supportive treatment and bears a relatively good prognosis in most patients. Our findings are similar to those presented by Kim et al. (2003), although that study only included patients with DPHL after carbon monoxide poisoning.
The main limitations of our study are related to the small number of subjects and its retrospective design, which are difficult to avoid given the rarity of this disorder. While a prospective cohort study (e.g., including all patients with neurological deterioration caused by an initial hypoxic event) would be ideal, this type of research is not well suited for rare diseases as an impractically high number of study subjects would be required (Mann 2003). It is possible that we could have identified more cases but the delayed presentation of this entity and the fact that many patients may present primarily with psychiatric symptoms (Tormoehlen 2013) may cause it to go underrecognized. Additionally, the time from symptom onset to MRI is variable among different patients, which is an inherent drawback related to the retrospective nature of this study. We could also not follow patients longitudinally to assess the reversibility of imaging findings over time. A prior case report in a patient with DPHL from carbon monoxide intoxication showed gradual resolution of restricted diffusion but persistence of the abnormal periventricular T2 signal abnormality which remained largely unchanged for over a year . Another case report documented incomplete resolution of the white matter abnormalities 1 year postoverdose (Shprecher et al. 2008), while separate reports showed improved but persistent abnormalities 6 months and 8 years after initial presentation, respectively (Molloy et al. 2006;Betts et al. 2012). Two-year clinical follow-up on another study, in patients with carbon monoxide-related DPHL, showed 75% recovery within 1 year (Choi 1983). Therefore, the use of the term "reversible" as in some of the prior descriptions of the disease might not be appropriate. Finally, DPHL has been shown to usually occur in older individuals. None of the patients in a large study of DPHL after carbon monoxide poisoning was less than 30 years of age (Choi 1983). Our patients were close to or above 60 years of age, with the exception of one patient who was 32 years old. Interestingly, this younger patient showed the most rapid clinical improvement of our cohort and was at his neurological baseline at discharge.

Conclusion
The characteristics and distribution of imaging findings in DPHL can be striking on MRI. We have performed an exhaustive review of the literature on this entity and present our findings on its imaging aspects in great detail. In the appropriate clinical setting, bilateral and symmetric white matter signal abnormalities confined to the supratentorial white matter without gyral edema or enhancement, are highly suggestive of DPHL, which carries a relatively favorable prognosis compared to other acute toxic or metabolic causes of white matter injury.