Markers in Status Epilepticus Prognosis

Abstract

Status epilepticus (SE) is a neurologic emergency with high morbidity and mortality. The assessment of a patient’s prognosis is crucial in making treatment decisions. In this review, we discuss various markers that have been used to prognosticate SE in terms of recurrence, mortality, and functional outcome. These markers include demographic, clinical, electrophysiological, biochemical, and structural data. The heterogeneity of SE etiology and semiology renders development of prognostic markers challenging. Currently, prognostication in SE is limited to a few clinical scores. Future research should integrate clinical, genetic and epigenetic, metabolic, inflammatory, and structural biomarkers into prognostication models to approach “personalized medicine” in prognostication of outcomes after SE.

Introduction

Status epilepticus (SE) is a life threatening condition of ongoing seizures with underlying heterogeneous etiologies and semiologies with various prognoses.(1, 2) Traditionally defined as 30 minutes of ongoing epileptic activity or seizures without recovery in-between, the Task Force of the International League Against Epilepsy (ILAE) in 2015 proposed a definition of SE as longer than 5 minutes for bilateral tonic-clonic seizures and longer than 10 minutes for focal seizures.(3) With annual incidence that varies from 9.9 to 41 per 100,000 people, SE case fatality is high and ranges from 5% to 46%.(4–13) Up to 43% of SE patients are refractory to treatment and progress to refractory status epilepticus (RSE) and super-refractory status epilepticus (SRSE).(14) Of RSE only 29% were reported to return to their functional baseline.(15)

Prognostic markers are helpful as they enable individualization of therapy and aid in decisions of treatment intensity. (16–18) SE prognosis is mostly driven by the complex pathophysiology and the underlying etiologies.(1) In this review, we evaluate the various markers used to prognosticate SE in terms of recurrence, mortality, and functional outcomes.

Markers in Status Epilepticus Prognosis- What We Know

Demographics and Clinical Characteristics

Sex-

The incidence of SE is higher in men. This might be related to the higher incidence of cerebrovascular events and brain trauma in males or be a result of hormonal influences.(19) Women were reported to have a higher risk of recurrent SE in a population-based study in Minnesota, Rochester (20) but a lower mortality in the Rochester Program Project data.(13) Studies evaluating the effect of gender on mortality have inconsistent results.(19, 21–23) In a recent population study by Leitinger et al., case fatality was 21.8% in women (95% CI 15.4–29.9) and 9.3% (95% CI 4.8–16.9) in men.(24) Studies have not evaluated the effect of gender on functional outcomes.

Age-

Age is a strong indicator of outcome in SE.(1, 5, 15, 25–28) Older age is an independent factor of death in SE.(29) In the Leitinger et al. population-based study, case fatality was 22.5% (95% CI 16.4–29.9) in elderly and 4.1% (95% CI 0.92–11.7) in patients younger than 60 year old. Therefore, age has been incorporated into various scoring systems for outcome prediction in SE such as the Status Epilepticus Severity Score (STESS), and the Epidemiology base Mortality score in SE (EMSE).(25, 29–31) Older age is also independently associated with poor outcome evaluated by the Glasgow Outcome Scale, with a GOS score <5 increasing the odds ratio for a poor outcome by 1.04/year (95% confidence interval, 10.02–1.05).(32, 33) Younger age, however, is associated with more refractory cases of SE, potentially driven by the underlying etiology.(33, 34)

Ethnicity-

There is a higher incidence of SE and lower mortality among blacks.(5, 21, 35, 36) This could be related to the underlying illness, access to medical care, compliance, or to other intrinsic biologic factors. Further studies are needed in minority populations to more closely evaluate the impact of race on prognosis.

Etiology and Comorbidities-

The underlying cause of SE is mostly related to structural, toxic, infectious or metabolic etiologies.(37) In most studies, cerebrovascular disease and low anti-seizure drug (ASD) levels were among the most common etiologies.(2) Overall, 30–40% of patients with SE have a history of epilepsy, and up to 60% of cases are due to acute symptomatic causes.(38, 39) Etiology is the main determinant of outcomes as reflected by STESS and EMSE scores.(25, 29, 31, 40) Low ASD levels (in epilepsy patients) and alcohol abuse have a relatively good prognosis (with reported mortality of less than 10% (2)) when compared to that for stroke (41, 42), meningitis (43), and post anoxic SE. (44, 45) Acute stroke in particular has higher mortality and morbidity rates compared with other causes, especially in Refractory SE patients. (41, 42) In patients with acute hypoxic brain injury, the prognosis is extremely poor with high mortality and poor functional recovery. (5, 46, 47) However, the presence of SE on continuous or intermittent EEG for a median of 48 hours of monitoring after cardiac arrest (CA) was followed by good functional recovery (Cerebral Performance Category CPC of 1 or 2) in 2 patients among 106 comatose CA patients.(48) The false positive rate was 0% for prediction of poor outcome (CPC 3–5) with seizures during hypothermia.(49, 50) Infectious encephalitis is associated with a high proportion of refractory SE with a substantial risk of developing epilepsy.(51, 52) The risks of morbidity and mortality are significant, as shown by the California Encephalitis Project, with 28% mortality within 2 years and 56% neurologically impaired or undergoing rehabilitation.(2, 53, 54) In other studies, mortality was reported as high as 71.4% (OR 13.88; P = 0.01).(27) Traumatic brain injury is an uncommon cause of SE, and overall it is associated with favorable outcomes.(2, 28) Cryptogenic SE is associated with variable outcomes, with generally low mortality and high risk of epilepsy.(5, 8, 33, 38, 51)

Studies looking at the role of prior comorbidities found them to be important in survival and in a return to baseline functional outcome.(28, 55–57) Higher comorbidity index was associated with higher mortality (OR 6.79, 95% 4.27 – 10.8).(28) Comorbidities such as diabetes mellitus, extracranial malignancy, sodium imbalance, CA, anoxic brain injury, pneumonia, sepsis, valvular heart disease, renal failure, and liver disease were associated with worse outcomes.(56, 57)

Level of Consciousness, Seizure Semiology, and Duration of SE –

Impaired level of consciousness at SE onset is independently associated with mortality.(58, 59) In a recent retrospective population-based study of adult patients with SE in Salzburg, impaired consciousness was associated with high case fatality (CF), 33%, as compared to 8.2% in awake patients(24). In the same study, CF was 2.8% in awake patients with prominent motor semiology versus 26.9% in somnolent, stuporous, or comatose patients. In nonconvulsive status epilepticus (NCSE) patients, CF was 0% in fully awake, 13.9% in awake with reduced cognition, 36.8% in somnolent, 46.2% in stuporous, and 42.9% in comatose patients.(24)

With respect to semiology, the new ILAE classification distinguishes NCSE from SE with prominent motor phenomena. In the study by Leitinger et al., prominent motor phenomena predisposed to good outcome but nCSE to bad outcome. Case fatality rates for NCSE, non-convulsive states at the end of a semiology sequence, nonconvulsive conditions at the beginning of SE, and only motor phenomena were 27.6%, 25.6%, 10%, and 3.5%, respectively.(24) The results of this study mirrored prior studies on the prognostic value of seizure semiology.(25, 29, 60) Myoclonic SE after CA was associated with poor outcomes and high mortality (61), but the characteristics of myoclonus in the studies were inconsistent.(44, 62, 63) It is also important to note that cases with good neurologic outcomes have been reported despite post-arrest myoclonus. (64–67)

It is difficult to know the exact duration of SE, as the onset is frequently not observed.(68) Overall, longer duration of SE was associated with higher mortality.(22, 27, 60, 69–71), However, new-onset refractory status epilepticus (NORSE) can be associated with significant recovery even after prolonged duration of SE.(72, 73) SE duration effect on outcome was heavily dependent on the underlying etiology.(70) Prolonged duration of SE can result in a significant volume reduction in the hippocampus, amygdala, or the entorhinal and peri-rhinal cortices.(74) Duration of SE was also a predictor of poor functional outcome and neurologic deficits (motor and cognitive).(32, 75–77)

Electrophysiological Characteristics -

The utility of different EEG features for prognosis is unclear. EEG findings were overall less important than the duration of SE and etiology as predictors of outcomes.(2) In a systematic review, periodic epileptiform discharges (PEDs) were found to be associated with poor outcome.(2) In an EEG study of 50 patients, PEDs were associated with higher mortality/vegetative state (44%) when compared to patients without PEDs (19%).(78) In a prospective study of 180 patients using at least 24 hours of continuous EEG recording, the authors found that the presence of a burst suppression (BS) pattern and post ictal discharges followed by PEDs was associated with higher mortality. Normalization of the EEG after SE correlated with good outcome, defined as an independent functional status on discharge.(79) Based on the former two studies, the EMSE score was introduced in 2015 with 4 EEG patterns as part of the score: PEDs, GPDs, after SE ictal discharges, and spontaneous burst suppression. The score had a negative predictive value (NPV) of 100%, positive predictive value (PPV) of 68.8% and classified 89.1% of in-hospital mortality.(31, 68) A prospective observational study compared the EMSE score to the STESS score (with no EEG data) and found EMSE to be superior for prediction of 30-day mortality and morbidity.(80) Interictal epileptiform discharges (IEDs) and periodic discharges/subtle status epilepticus patterns correlated with refractoriness of SE.(81) A reactive EEG at onset of prolonged refractory SE was associated with good outcome defined by mRS ≤ 3 at 6 months follow up.(82) EEG at 1 hour was found useful in pre ting SE recurrence within 24-hour, but not in predicting mortality.(83) Generalized periodic discharges (GPDs) were strongly associated with NCSE.(84) The absence of triphasic morphology, focality on EEG, and interburst suppression, correlated with higher risk of seizures in patients with GPDs.(85) Foreman et al., found that EEG interpreted as including triphasic waves had a similar risk of seizures as did GPDs without triphasic morphology.(86) In CA patients, the presence of seizures on EEG during hypothermia or after rewarming is associated with worse outcomes (CPC 3–5).(49, 50) Electrographic SE evolving from a burst suppression pattern was associated with poor outcome in all cases of CA, except when evolving from a continuous background (FPR 4%).(87)

Imaging-

The role of neuroimaging in prognostication is limited to recognizing acute etiologies of SE. In children, functional outcome (defined by neurologic sequelae) was worse in patients with abnormal neuroimaging.(88) Adults with prolonged refractory SE had good outcomes (mRS ≤ 3 at 6 months f/u) when neuroimaging was normal.(82) Abnormal neuroimaging was associated with increased risk of seizures in critically ill patients with GPDs and was associated with worse functional outcomes in SE measured by mRS.(85, 89) In a retrospective study of 277 patients, peri-ictal magnetic resonance imaging (MRI) abnormalities were associated with the duration of SE and with lateralized epileptiform discharges on EEGs. The alterations were unilateral and most commonly involved mesiotemporal structures.(90) MRI with contrast and T2-weighted signal protocols, as a surrogate for blood brain barrier dysfunction, were found to be sensitive and specific prognostic factors for developing epilepsy in animals.(91)

Biochemical markers-

Neuronal Damage Biomarkers (Neuron-specific enolase (NSE) and Tau proteins):

Elevated NSE levels can be indicative of neuronal injury.(92) NSE is best known for its role in prognosis in patients with CA, although this has changed since the more widespread use of hypothermia.(63, 93–101) NSE in SE was first reported in 1995 in 2 patients with NCSE.(102) DeGiorgio et al. found a serum peak 24–48 hours after NCSE in 19 patients; there was a possible association of higher levels (when compared to normal and epileptic controls) with worse outcome and longer duration of SE but overall the findings were inconclusive in this small cohort.(103) Correale et al. reported cerebral spinal fluid (CSF) NSE levels in patients with symptomatic SE. Levels were elevated in 9 of 11 patients when compared to controls. The authors also reported increased CSF/serum albumin ratios.(104) A prospective study of 8 patients with complex partial SE had 4 fold increases in serum levels (mean peak 21.81 ng/ml), with significant correlations with both seizure duration and outcomes (assessed using GOS). They reported one patient with a level of 50 ng/ml during the seizure who had a good outcome (GOS of 5 at 1 week after SE).(105) In 1999, the same authors studied 31 patients; levels were drawn 1, 2, 3, and 7 days after SE. NSE levels were elevated in all 4 subtypes of SE (complex partial, generalized convulsive, absence, and subclinical) but were highest in complex partial and subclinical SE.(106)

Tau protein is a phosphorylated protein involved in regulating microtubule assembly and disassembly. It is considered an indicator of both axonal and neuronal damage and can be a diagnostic and prognostic marker in traumatic brain injury.(107) Tau CSF levels were studied in 28 patients: 14 patients had high CSF t-tau levels, 6 patients had high CSF p-tau levels, and 3 patients had low AB_1–42 levels. CSF t-Tau level was associated with higher SE duration, the use of general anesthesia (propofol), higher risk of developing epilepsy, and higher disability rates.(108) The role of tau in SE needs further evaluation.

Inflammatory biomarkers (Procalcitonin PCT, C-reactive protein CRP, Albumin, Uric Acid UA, and Cytokines):

Seizure activity may cause systemic inflammatory reactions represented by changes in cytokine levels (interleukin-1B, interleukin-2, interleukin-6, and tumor necrosis factor TNF-alpha), increased circulating immune cells (neurophils, lymphocytes, natural killer cells), and blood brain barrier dysfunction.(109–119) Conversely, systemic inflammation can trigger or sustain seizures and influence the course of SE.(120–122) PCT, CRP and albumin are all acute-phase proteins and may reflect the inflammatory process in SE. These markers are routinely assessed and are broadly available. In a 5-year observational cohort study of 135 patients, albumin was assessed at the onset of SE, and CRP during the first 3 days. Higher levels of albumin had lower odds for progressing to RSE and death, regardless of seizure severity. CRP levels were associated with higher rates of RSE and death, but inconsistently after adjusting for confounders.(40) Serum albumin <35 g/l at SE onset was also found to be an independent predictor for RSE.(123) Higher CRP and TNF levels were linked to in-hospital nonconvulsive seizures in a prospective observational study of patients with acute subarachnoid hemorrhage. The association of inflammatory biomarkers with poor outcome was mediated in part through seizures.(119)

PCT is a prepropeptide precursor of the thyroid hormone calcitonin, which also increases in various inflammatory conditions, mostly reported in bacterial infections and sepsis.(124–126) In an observational cohort study, Sutter et al. studied serum levels of PCT, CRP, and albumin at SE onset in 91 patients. PCT predicted death and GOS 1–3 independently from acute etiology, infections, the Charlson Comorbidity Index, and the STESS score.(127) In this study, PCT was the only biomarker independently associated with unfavorable outcomes.(127) Determinations of CRP and PCT levels excluded SE patients with hospital acquired infections.(128)

More recently, uric acid (UA) levels were evaluated in 141 SE patients because of their roles as endogenous anti-oxidant and reactive oxygen/nitrite scavengers. UA levels on admission were predictive of refractoriness with low levels indicating a higher chance to respond to initial AED management . Higher levels indicated more difficult to control RSE. These patients had longer hospitalizations and worse functional outcomes (mRs >2).(129)

Inflammatory cytokines are usually not stable in blood, but they have a potential role in prognosis in SE.(130) For example, an elevated CSF/serum IL-1B ratio predicted post traumatic epilepsy.(131) Significant increases in blood IL-1, IL-1 RA, IL-6 and IL-8 within 72 hours in children with febrile SE were found in the FEBSTAT study. Interestingly, a low IL-1RA/IL-6 ratio was strongly associated with acute hippocampal injury on MRI.(132) CSF levels of 8-hydroxdeoxyguanosine (a marker of oxidative DNA damage) were increased in children with SE.(133)

What the Studies Tell us

The several biomarkers mentioned above can be used in three main areas of prognosis: resistance to treatment, morbidity, and mortality. Most of these markers are not modifiable and are baseline characteristics. Age, nonconvulsive semiology, longer duration of SE and etiology are among the most important determinants of SE outcomes.(22, 24, 25, 27, 29, 70, 71, 134, 135) Laboratory quantifiable biomarkers are more recent and may show promise for clinical endpoints.

The clinical bedside applications of markers in SE prognosis are so far limited to several prediction scores that have been created to prognosticate in-hospital mortality. A modified version of the STESS with an added premorbid functional status was proposed as a tool with greater accuracy.(136) Another score is the EMSE score, putting an emphasis on the different SE etiologies. EMSE (particularly, EMSE 64) appeared to be superior to STESS 3 and 4 thresholds for predicting 30 day mortality and morbidity.(29, 30, 80, 137, 138) More recently, the END-IT score was developed based on imaging findings and treatment response to first-line therapy.(89) The score was developed in China from a cohort of young patients with a high percentage of encephalitis, which may limit its use in other part of the world, and with other illnesses.

What We Want to Know and Future Research Directions

The ideal prognostic markers of SE would be readily available to clinicians at the bedside and provide high sensitivity and specificity in predicting outcomes associated with SE. These would inform about the likelihood of SE outcomes including progression into refractoriness, functional outcome, and mortality. Such prognostic markers are necessary as they would allow individualization of therapy and aid in decisions about treatment intensity. An ideal prognostic marker in SE does not exist, and we are presently limited to the SE scores mentioned above. Despite having a good negative predictive value, these scores have a poor positive predictive value preventing their use for decisions of withdrawal of life-sustaining therapy.(139)

Preclinical and clinical studies have examined different biomarkers for the diagnosis and prognosis of epilepsy.(140) Animal studies have revealed that genetic and epigenetic changes occur in patients with SE leading to over- or under-expression of certain genes. Genome-wide association studies have shown that certain genes, such as the SCN1A gene, are associated with certain types of epilepsies.(141) MicroRNAs, a class of small non-coding RNA that regulate post-transcriptional gene expression are present in the brain and were found to be deregulated in drug-refractory epilepsy.(142) These biomarkers are interesting as they can be measured non-invasively in biofluids. A study using a PCR-based technology to screen for microRNAs in the cerebrospinal fluid found that patients with SE had higher levels of certain microRNAs.(143)

Certain structural changes predispose patients to developing epilepsy such as hippocampal sclerosis.(140) Other studies have found that metabolic biomarkers such as glucose metabolism are altered in the setting of seizures.(144) Neuroinflammatory changes, such as reduced concentrations of anti-inflammatory modulators including soluble intercellular adhesion molecule 5 (sICAM5) have been linked to drug-resistant epilepsy.(145) Microvascular injury has also been linked to epileptogenesis.(140) Future clinical studies are needed to validate the sensitivity and specificity of these biomarkers in SE and assess their role in prognostication.

The heterogeneity of SE etiology and semiology renders development of prognostic markers challenging. Findings in animal studies should be assessed in clinical trials and more studies are needed to identify the most potential prognostic markers. The use of “big data” and the revolution of machine-learning models will allow the integration of these findings into “personalized medicine” systems. This would allow incorporation of clinical, genetic and epigenetic, metabolic, inflammatory and structural biomarkers into scoring systems that would allow more accurate prognosis.