Development and validation of a peripheral cell ratio and lactate score for differentiating status epilepticus from prolonged psychogenic nonepileptic seizures

Abstract

Objective: Differentiating status epilepticus (SE) from prolonged psychogenic nonepileptic seizures (pPNES) can be difficult clinically. We aimed to define the utility of peripheral cell counts, cell ratios, and lactate levels in distinguishing SE from pPNES.

Methods: Retrospective two-center study investigating the sensitivity and specificity of acute (≤12 h of event offset) peripheral cell counts, cell ratios (neutrophil-lymphocyte ratio, neutrophil-monocyte ratio, monocyte-lymphocyte ratio, platelet-lymphocyte ratio, systemic immune-inflammatory index [SII], systemic inflammatory response index [SIRI]), and lactate levels in differentiating SE from pPNES. Patients were identified from two tertiary hospitals, with one forming the development cohort and the other the validation cohort. Using generalized additive models to generate biomarker vs time curves, optimal blood collection times were defined for set parameters. Three diagnostic scores combining neutrophil count, SII, or SIRI with lactate levels were developed and validated in separate cohorts.

Results: For the development cohort, 1262 seizure-like events were reviewed and 79 SE and 44 pPNES events were included. For the validation cohort, 241 events were reviewed and 20 SE and 11 pPNES events were included. Individually, the biomarkers generally had low sensitivity and reasonable specificity for differentiating SE from pPNES, with the neutrophil count, SIRI, and SII performing best with sensitivities of 0.65-0.84, specificities of 0.64-0.89, and ROC AUCs of 0.78-0.79. Lactate levels peaked at 60 min, while cell counts and ratios peaked after 240 min. Combining early peaking lactate levels and later peaking neutrophil count, SIRI or SII resulted in three scores that improved predictive potential with sensitivities of between 0.75 and 0.79, specificities between 0.93 and 1.00, and ROC AUCs of 0.89-0.91.

Significance: Lactate levels peak early post-SE, whereas cell counts and ratios do so later. The differing post-event time profiles of lactate levels vs neutrophil count, SIRI, and SII allow incorporation into three separate scores which can assist in differentiating SE from pPNES.

Keywords: PNES; lactate; lymphocyte; monocyte; neutrophil; seizures.

FIGURE 1

Study flowchart. Refractory status epilepticus is defined as ongoing seizure activity despite the use of two intravenous antiseizure medications, one of which was from a nonbenzodiazepine class. Super‐refractory SE is defined as ongoing seizure activity lasting over 24 h despite the use of an anesthetic agent or recurrence of seizures following weaning of anesthetic agent after over 24 h of use. pPNES, prolonged psychogenic nonepileptic seizures; SE, status epilepticus.

FIGURE 2

Level versus time curves. Lactate, neutrophil count, neutrophil‐lymphocyte ratio (NLR), and systemic inflammatory index (SII) versus time curves. The solid line represents the calculated estimated marginal mean; the shaded area represents the 95% confidence interval. Blue is the status epilepticus cohort and yellow is the prolonged psychogenic nonepileptic seizure cohort. Vertical black broken lines mark the calculated optimal collection times in minutes postevent onset using timepoints where there were significant differences between the calculated estimated marginal means of the two arms (P < 0.05). The red vertical line marks the time of peak variable measurement in the SE cohort. Lactate optimal collection time onset and time of peak variable measurement in the SE cohort coincided at 60 min. NLR, neutrophil–lymphocyte ratio; SII, systemic immune inflammatory index.

FIGURE 3

Diagnostic nomograms. Three separate nomograms combining lactate and neutrophil count (Neut‐Lac Score) (A) or lactate and SIRI (SIRI‐Lac Score) (B) or lactate and SII (SII‐Lac Score) (C) to predict the probability an SE diagnosis (as opposed to a PNES diagnosis). The probability of SE is calculated by adding the corresponding points from each variable to give a total point score from which the probability of SE can be read.