Prognostic factors of subacute comprehensive encephalitis: a retrospective study
Article information
Abstract
Purpose
The etiology of encephalitis is unknown in 40%–50% of cases, so a comprehensive examination of encephalitis would be significant and meaningful. The short-term outcomes in appropriately managed patients are also unknown. Short-term clinical outcomes following onset can provide clinicians with clues regarding the clinical course in the immediate future. We investigated cases of encephalitis, including viral and autoimmune encephalitis, to determine the predictable risk factors that can be assessed to determine a short-term prognosis.
Methods
We studied 90 patients with encephalitis. Poor and good outcomes were defined as scores of ≥3 and ≤2 on the modified Rankin scale, respectively. Multivariate logistic regression analysis using 19 independent variables was performed.
Results
Multivariate logistic regression analysis identified cranial magnetic resonance imaging (MRI) lesions (odds ratio [OR], 3.119; 95% confidence interval [CI], 1.166–8.344; p = 0.023) and the need for mechanical ventilation (OR, 4.461; 95% CI, 1.685–11.813; p = 0.003)) as being significantly associated with poor outcomes. In 57 patients with subacute encephalitis presenting with cranial MRI lesions, bilateral lesions on cranial MRI (OR, 5.078; 95% CI, 1.516–17.007; p = 0.008) and the need for mechanical ventilation (OR, 4.461; 95% CI, 1.135–13.584; p = 0.031) were significantly associated with poor outcomes.
Conclusion
The location of brain lesions, lateral or bilateral, on the initial MRI during the acute phase of encephalitis may be a useful predictor of the outcome during the first 2 months after encephalitis onset, even in cases of encephalitis of unknown etiology.
Introduction
Abnormal immunological and inflammatory reactions in the brain can lead to encephalitis. The mortality rate for encephalitis ranges from 7% to 18%, and 50% of survivors develop severe disabilities [1]. Although the outcomes of autoimmune encephalitis have improved with the development of immunotherapies, the prognosis remains unfavorable. Certain individuals with autoimmune encephalitis, such as those with anti–N-methyl-ᴅ-aspartate receptor (NMDAR) encephalitis, have poor results, defined as a modified Rankin scale (mRS) score of 4 or over [2]. Additionally, approximately 70% of patients with untreated herpes simplex encephalitis die; and 97% of survivors never regain their prior level of functioning [3]. Physicians commonly encounter patients with encephalitis, particularly limbic or autoimmune encephalitis, that can present with a variety of symptoms including seizures, psychosis, memory loss, and altered consciousness. Developing a differential diagnosis and making initial treatment decisions can be challenging owing to the many etiologies, only half of which are known [1,4,5]. The International Classification of Diseases, 10th Revision (ICD-10) (G04) provides several classifications for encephalitis. Types of encephalitis (G04-8) include autoimmune limbic encephalitis and anti-NMDAR encephalitis. The large number of encephalitis cases that are unclassifiable or have an unknown cause are grouped together (G04-9). The cases with an unknown cause may have diverse etiologies. Many facilities frequently do not provide initial test results regarding the causative virus or the presence of autoantibodies required for ICD-10 classification. The causative etiology for encephalitis remains unknown in 40% to 50% of cases [6,7].
The FilmArray meningoencephalitis panel (BioFire Diagnostics) has high diagnostic accuracy. However, according to recent studies, herpes simplex virus (HSV) 1 and 2, enterovirus, and Cryptococcus neoformans/gattii have the highest percentages of false-negative results. Therefore, the treatment decision needs to be individualized [8]; and a comprehensive examination of encephalitis would be significant and meaningful.
Previous studies have identified the need for admission to the intensive care unit (ICU) [1,5], unilateral hyper-perfusion on single-photon emission computed tomography analysis [9], and abnormal electroencephalographic findings [1,4] as risk factors for poor outcomes in encephalitis cases. The outcome endpoints also vary; for example, some studies use data from the admissions to the ICU [1] or from more than 1 year of follow-up after discharge [4]. However, the short-term outcomes in appropriately managed patients are unknown. Short-term clinical outcomes following onset can provide clinicians with clues regarding the clinical course in the immediate future. In this study, we investigated cases of encephalitis, including cases of viral and autoimmune encephalitis, to determine the predictable risk factors for short-term prognosis.
Methods
The study protocol was approved by the Medical Ethics Committee of Nara Medical University (No. 3831), including the use of the opt-out method as appropriate for a retrospective study where written consent was not obtained.
Participants
A total of 90 enrolled patients had a mean ± standard deviation age of 47.6 ± 18.9 years with a range of 16 to 87 years. Thirty-seven were females. All patients were treated for subacute encephalitis. Subacute was defined as neurological or neuropsychiatric syndromes that rapidly developed over 4 weeks. The inclusion criteria were white blood cell count >5 in the cerebrospinal fluid (CSF) [1] and the presence of neurological manifestations, such as altered mental status (altered level of consciousness or personality change); abnormal behavior; seizure or involuntary movement; and memory impairment at disease onset. All patients were examined for viral antibodies using screening assays, like those for HSV, and CSF culture testing for bacteria, fungi, and mycobacteria (tuberculosis). Polymerase chain reaction (PCR) confirmed HSV in 13 patients, varicella zoster virus in four, and human herpes virus 6 in three. Twenty-eight patients were examined for the presence of cell surface antibodies. As indicated in Table 1, 15 cases of HSV encephalitis, seven cases of varicella zoster virus encephalitis, and nine cases of anti-NMDA receptor antibody-associated encephalitis were identified. When the PCR results for varicella zoster virus or HSV were negative, acyclovir, generally administered as the initial treatment for suspected HSV encephalitis, was discontinued. Steroid treatment and intravenous immunoglobulin were administered to 49 and 22 patients, respectively. The dosage and duration of these treatments were at the discretion of the treating physician.
Basic characteristics of patients with subacute encephalitis
The exclusion criteria were detection of bacteria, tuberculosis, or fungi on CSF culture; detection of syphilis on serological testing; presence of a suspected brain tumor on computed tomography (CT) or initial/follow-up magnetic resonance imaging (MRI) as assessed by an experimental radiologist; subsequent diagnosis of brain tumor; history of systemic immunological disease; presence of acute clinical episodes following vaccinations; or multiple clinical multiple episodes with abnormal white matter lesions in the central nervous system suggestive of acute disseminated encephalomyelitis or multiple sclerosis. Furthermore, these diseases and conditions were ruled out based on the results of serological and radiological investigations: cerebral thromboembolism and hemorrhage, cerebral venous sinus thrombosis, cerebral aneurysm, cerebral arteriovenous malformation, mitochondrial encephalopathy, diabetic encephalopathy, hepatic encephalopathy, systemic lupus erythematosus, anti-phospholipid antibody syndrome, sarcoidosis, and Hashimoto’s encephalopathy.
Statistical analysis
Poor and good outcomes were defined as scores of ≥3 and ≤2 on the mRS [1] 2 months after onset, respectively. For patients who were discharged with excellent outcomes within 2 months, the outcome at hospital discharge was used in the analysis. The following 19 independent variables were evaluated and scored: age; sex (female = 1); detection of virus or autoantibodies (absent = 0 and present = 1); initial neurological symptoms of seizure, altered mental status, and abnormal behavior or memory impairment (absent = 0 and present = 1); Glasgow Coma Scale (GCS) score at admission; leukocyte count in the CSF (/mm3); protein level in the CSF (mg/dL); detection of focal lesions on initial CT (absent = 0 and present = 1); focal lesions on cranial MRI (absent = 0 and present = 1); lateralized periodic discharges (LPDs) on EEG (absent = 0 and present = 1); use of antiviral, steroid or intravenous immunoglobulin therapy during the acute phase of encephalitis (not administered = 0 and administered = 1); duration from neurological onset to initiation of acyclovir or immune treatments; generalized seizure during the disease (absent = 0 and present = 1); brain infarction or hemorrhage during the disease course (absent = 0 and present = 1); and the need for use of mechanical ventilation during the acute stage (not administered = 0 and administered = 1). In the second cohort presenting with cranial MRI lesions, in addition to assessments of these variables, three of the five cortical brain lobes, frontal, parietal, temporal, occipital, and insular, were analyzed.
Variables showing a statistically significant correlation (p < 0.05) with poor outcomes on univariate logistic regression analysis were entered into a multivariate logistic regression analysis using forced entry. The corresponding odds ratios (ORs) and 95% confidence intervals (CIs) were calculated. Correlations between individual variables were evaluated using Spearman rank correlation test. Differences in clinical characteristics between patients with poor outcomes and those with good outcomes were evaluated using the Mann-Whitney or chi-square tests. All statistical analyses were performed using SPSS version 24 (IBM Corp.).
Results
The clinical characteristics of the 90 enrolled patients with subacute encephalitis are summarized in Table 1. The proportion with cranial MRI lesions (p = 0.004), bilateral lesions on MRI (p = 0.006), and need for mechanical ventilation (p < 0.001) in the poor outcome group were significantly higher than those in the good outcome group (Table 2). Significant variables in univariate logistic regression analysis included cranial MRI lesions (OR, 3.942; 95% CI, 1.554–10.002; p = 0.004), and need for mechanical ventilation during the acute stage (OR, 5.333; 95% CI, 2.076–13.701; p = 0.001) (Table 3). Multivariate logistic regression analysis adjusted for presence of cranial MRI lesions, and the use of mechanical ventilation identified cranial MRI lesions (OR, 3.119; 95% CI, 1.166–8.344; p = 0.023), and the need for mechanical ventilation (OR, 4.461; 95% CI, 1.685–11.813; p = 0.003) as being significantly associated with poor outcomes. These predictors did not significantly interact with each other. Furthermore, age [10] and GCS score at admission [11], which have previously been reported as risk factors for poor prognosis in patients with subacute encephalitis, were entered into the multivariate logistic regression analysis; and the presence of cranial MRI lesions (OR, 2.740; 95% CI, 0.988–7.604; p = 0.053) and the need for of mechanical ventilation (OR, 5.591; 95% CI, 1.931–16.183; p = 0.002) were associated with poor outcomes.
Comparison of subacute encephalitis patients with poor or good outcomes
Independent variables predictive of outcomes in 90 patients with subacute encephalitis
Subacute encephalitis with cranial MRI lesions showed a significantly higher proportion of patients that were female (p = 0.015); had the presence of memory impairment (p = 0.04), focal lesions on initial CT (p = 0.002), or LPDs on electroencephalography (p = 0.006); needed mechanical ventilation (p = 0.024); and had an mRS score of >2 (p = 0.004) than the proportion of patients without cranial MRI lesions (Table 4). In 57 patients with subacute encephalitis presenting with cranial MRI lesions, bilateral lesions on cranial MRI (OR, 5.078; 95% CI, 1.516–17.007; p = 0.008) and need for mechanical ventilation (OR, 3.927; 95% CI, 1.135–13.584; p = 0.031) were significantly associated with poor outcomes in adjusted multivariate logistic regression analysis (Table 5). When age, GCS score at admission, and sex were entered into the multivariate logistic regression analysis, bilateral lesions on cranial MRI (OR, 5.003; 95% CI, 1.406–17.801; p = 0.013) and the need for mechanical ventilation (OR, 7.747; 95% CI, 1.580–37.997; p = 0.012) showed significant differences.
Comparison of subacute encephalitis patients with/without cranial MRI lesions (n = 90)
Independent variables predictive of outcomes in 57 patients with subacute encephalitis presenting with cranial MRI lesions
These results were observed in patients with a discernible etiology, particularly in patients with viral encephalitis, but not in patients with unknown etiologies (Tables 6 and 7).
Difference between patients with poor or good outcomes by subgroups of subacute encephalitis
Crude ORs on univariate logistic regression analysis for predictive outcomes in patients with subgroups of subacute encephalitis
Discussion
The results of this study demonstrated that the need for mechanical ventilation and the presence of cranial MRI lesions were significantly associated with poor prognosis in a heterogeneous group of patients with subacute viral or autoimmune encephalitis or with encephalitis of unknown cause. Analysis of data from the cohort of patients with MRI lesions also showed that the presence of bilateral lesions and the need for mechanical ventilation were significantly associated with poor outcomes. This study is the first to investigate the short-term outcomes following clinical onset of viral and autoimmune encephalitis or encephalitis with unknown etiologies. As such, this study can provide clinicians with insights into clinical management during the active phase of encephalitis.
Like our study, a recent retrospective study of 209 patients with comprehensive encephalitis based on ICD-10 diagnostic codes evaluated the clinical predictors of mortality and poor outcomes [1]. That study found that the presence of cranial MRI abnormalities, defined as lobe lesions; presence of increased T2 and fluid-attenuated inversion recovery signals; increased contrast enhancement; and abnormal electroencephalograms were risk factors for mortality on multivariate logistic regression models. Abnormal imaging findings were observed in 78 patients; but detailed information regarding the MRI lesions, such as lesion laterality, was not available. That study also documented that poor outcomes, defined as mRS scores of 4–5, were not associated with MRI abnormalities but were associated with advanced age, number of rescues, and tubercular infection. The inconsistent poor outcome result in our study may be explained by the fact that their study population included patients with bacterial, fungal, and tubercular encephalitis and excluded those with autoimmune encephalitis. In addition, the definition of poor outcomes and the setting of the hospital discharge endpoint differed from those in our study. Furthermore, assessment of the need for mechanical ventilation was not conducted in their study. Another retrospective study of 161 patients with encephalitis of various etiologies showed that the need for mechanical ventilation for >10 days was associated with in-hospital mortality in multivariate analyses [5].
The etiology of encephalitis was not significantly associated with poor outcomes in our study. In HSV encephalitis cases, which have the highest mortality rate among adult viral encephalitides, more than three brain lobes affected [12,13], restriction of diffusion identified on MRI [10], and presence of diffusion-weighted MRI signal abnormalities in the left thalamus [12] have previously been reported as predictive risk factors for poor outcomes. In 45 patients with HSV encephalitis, bilateral involvement on MRI was not significantly associated with unfavorable or favorable outcomes at discharge (6–12 months) [10]. Another recent study of 138 patients with HSV encephalitis showed that bilateral abnormalities on MRI were associated with poor outcomes; 34 patients with bilateral abnormalities had poor outcomes based on mRS score at 90 days after ICU admission [12]. This was consistent with our findings. Another study showed that the need for mechanical ventilation was independently associated with poor outcome [13]. For patients with comprehensive autoimmune encephalitis, previous studies have reported that the need for mechanical ventilation was associated with a longer length of hospital stay and a slower recovery process and could serve as an independent risk factor of poor short-term outcomes [14]. In patients with anti-NMDA encephalitis, the most observed form of autoimmune encephalitis, one study showed that the proportion of abnormal brain MRI signals was significantly higher in the poor prognosis group [15]. Central hypoventilation, which is associated with a high risk of admission to the ICU due to respiratory failure, was associated with poor prognosis [16,17]. The need for long-term respiratory support increased the risk of infection and pneumonia and the development of ICU-acquired weakness; these contributed to worse acute outcomes and long-term mortality [18]. These disease-specific results may support the association between causative etiology and poor outcomes in subacute encephalitis. Our study failed to detect the etiology of some encephalitides. However, in the subgroup of patients with a discernible etiology, the need for mechanical ventilation, and the presence of bilateral cranial MRI lesions were associated with poor outcome; and there was a similar trend in patients with HSV encephalitis.
Treatment with antiviral therapy, steroids, and intravenous immunoglobulins showed no significant effect on univariate logistic regression analysis. We evaluated heterogeneous patients with autoimmune and viral encephalitis and encephalitis of unknown etiology. The rate of detecting causative viruses is low; only 50% of all encephalitis cases have an identified cause [19]. The incidence of autoimmune encephalitis has been increasing in recent years, but its etiology and pathogenesis are unclear [20]. First-line treatment with steroids or immunoglobulins commonly fails to improve the severity of autoimmune encephalitis, resulting in a need for second-line treatments such as immunosuppressants. However, the regimens and durations of these treatments were inconsistent in our study; but specific treatments such as acyclovir for HSV encephalitis or immune therapies for autoimmune encephalitis are generally effective for acute encephalitis.
This study had several limitations. Owing to the retrospective design, none of the patients without a confirmed etiology underwent neuronal antibody testing or PCR for viruses, including HSV; and causes were unknown in half of the patients. However, bacterial, fungal, and tuberculosis infections were ruled out as causes in these patients on repeated CSF cultures or repeated cranial MRI suggesting a viral infection cause. Second, the final follow-up outcomes were not examined, particularly for anti-NMDAR encephalitis with protracted disease. Extending follow-up would provide a more comprehensive understanding of prognostic variables. In this study, the mRS score was used to determine the prognosis of encephalitis. The prognosis of encephalitis can vary due to the wide variations in symptoms, including memory disorders, higher dysfunction, and motor paralysis. Currently, there is no specific prognostic index for encephalitis that considers these symptoms. Nonetheless, the prognosis of encephalitis has been extensively assessed using the mRS [21]. Moreover, comparing patient improvement over the first two-month period using patient initial clinical condition as judged by baseline Glasgow Coma Scale or mRS score would help identify factors associated with recovery.
In conclusion, lesion laterality on the initial, acute phase, brain MRI could serve as a useful predictor of the outcome within 2 months of encephalitis onset, even in cases of encephalitis of unknown etiology.
Notes
Conflicts of Interest
No potential conflict of interest relevant to this article was reported.
Acknowledgments
We are extremely grateful to all clinicians and medical staff members of Nara Medical University.