Out with prolactin, in with neurofilament light and glial fibrillary acidic protein

Abstract

Elevated Plasma Neurofilament Light and Glial Fbrillary Acidic Protein in Epilepsy Versus Nonepileptic Seizures and Nonepileptic Disorders.

Dobson H, Al Maawali S, Malpas C, Santillo AF, Kang M, Todaro M, Watson R, Yassi N, Blennow K, Zetterberg H, Foster E, Neal A, Velakoulis D, O'Brien TJ, Eratne D, Kwan P. Epilepsia. 2024 Sep;65(9):2751-2763. doi: 10.1111/epi.18065. Epub 2024 Jul 20. PMID: 39032019.

Objective: Research suggests that recurrent seizures may lead to neuronal injury. Neurofilament light chain protein (NfL) and glial fibrillary acidic protein (GFAP) levels increase in cerebrospinal fluid and blood in response to neuroaxonal damage, and they have been hypothesized as potential biomarkers for epilepsy. We examined plasma NfL and GFAP levels and their diagnostic utility in differentiating patients with epilepsy from those with psychogenic nonepileptic seizures (PNES) and other nonepileptic disorders. Methods: We recruited consecutive adults admitted for video-electroencephalographic monitoring and formal neuropsychiatric assessment. NfL and GFAP levels were quantified and compared between different patient groups and an age-matched reference cohort (n = 1926) and correlated with clinical variables in patients with epilepsy. Results: A total of 138 patients were included, of whom 104 were diagnosed with epilepsy, 22 with PNES, and 12 with other conditions. Plasma NfL and GFAP levels were elevated in patients with epilepsy compared to PNES, adjusted for age and sex (NfL p = .04, GFAP p = .04). A high proportion of patients with epilepsy (20%) had NfL levels above the 95th age-matched percentile compared to the reference cohort (5%). NfL levels above the 95th percentile of the reference cohort had a 95% positive predictive value for epilepsy. Patients with epilepsy who had NfL levels above the 95th percentile were younger than those with lower levels (37.5 vs 43.8 years, p = .03). Significance: An elevated NfL or GFAP level in an individual patient may support an underlying epilepsy diagnosis, particularly in younger adults, and cautions against a diagnosis of PNES alone. Further examination of the association between NfL and GFAP levels and specific epilepsy subtypes or seizure characteristics may provide valuable insights into disease heterogeneity and contribute to the refinement of diagnosis, understanding pathophysiological mechanisms, and formulating treatment approaches.

Commentary

Psychogenic non-epileptic seizure (PNES) is no longer the proper name to describe events that may resemble epileptic seizures (ES) but are not in keeping with the pathophysiological mechanisms of ES. To align with the broader field of functional neurological disorders, PNES has been rebranded to “functional seizures” (FS). The name is not the only thing that changed, as prolactin levels, previously considered very useful in the diagnosis of FS, has fallen in popularity. Elevated serum prolactin had been used as a marker of ES, while FS was not associated with abnormal levels of prolactin. ¹ However, more research has shown that serum prolactin can be elevated in almost 30% of FS (false-positive) and levels may be normal in 15% of ES (false-negative), including non-convulsive seizures. ² Furthermore, temporal lobe seizures can cause a rise of prolactin levels, but frontal lobe seizures may not. ³ Finally, prolactin levels rapidly return to baseline in a few hours after a seizure, making it less useful outside of emergency rooms. Based on clinical, imaging and neuropathological evidence that seizures can lead to neuronal injury, it is reasonable to consider that markers of neuronal injury may be better tools to differentiate ES from FS. ⁴ So far, neurofilament light chain protein (NfL) and glial fibrillary acidic protein (GFAP) have been in the spotlight in matters of neuronal injury markers and are easily measured in plasma. These cytosolic proteins are expressed in myelinated axons (NfL) and astrocytes (GFAP). Protein levels of both NfL and GFAP are increased in cerebrospinal fluid and blood, in response to neuroaxonal damage. Indeed, some research has shown that increased levels of NfL are seen in status epilepticus, and even in chronic epilepsy. ⁵ So, the question is: could NfL and GFAP substitute prolactin as a better tool for diagnosing FS versus ES?

Dobson and colleagues ⁶ set out to determine the usefulness of plasma levels of NfL and GFAP in aiding the diagnosis of ES or FS. They evaluated patients admitted to their epilepsy monitoring unit (EMU) in whom a final diagnosis had been obtained, based on their video-EEG, plus imaging and neuropsychological findings. The cohort was then classified into “epilepsy”, “functional seizures” and “other” groups. In short, 152 patients admitted to their EMU were enrolled between June 2018 and July 2022. Fourteen patients were excluded from the final analyses because medical records were unavailable in 3, and because blood had been obtained after neurosurgical intervention in 11 cases. Their final cohort of 138 cases was composed of 104 patients with epilepsy (3 of whom had both ES and FS), 22 patients with FS and 12 patients with other conditions such as cardiac (n = 5), migraine (n = 3), dementia (n = 1), idiopathic primary hypersomnolence (n = 1), provoked seizure (n = 1) and nondiagnostic non-epilepsy (n = 1). Patients with epilepsy were having 1-3 seizures per month in the 12 months prior to EMU admission. The mean age of patients in the epilepsy group was 42.5 years (SD = 15.1), FS group was 38.5 years (SD = 11.8), and other group was 48.8 years (SD = 21.4). Females represented 56.7% of the epilepsy, 90.9% of the FS and 75% of the other groups.

NfL and GFAP evaluation: blood was collected during the EMU admission in 119 cases. In the remainder of patients, collection was done in outpatient clinic or during a new admission, prior to surgical procedure.

Both NfL and GFAP levels were higher in the epilepsy group compared to the FS group, even after adjusting for age and sex (p = .04 in both cases). However, the absolute NfL plasma level had a weak performance in distinguishing between epilepsy and FS (AUC = .63, 95% CI = .50-.75). GFAP also had a weak performance in distinguishing between epilepsy and FS (AUC = .63, 95% CI = .51-.75). Using different cutoff levels, specificity and sensitivity changed. For instance, with NfL cutoff of 4.2 mg/ml, a 36% specificity and 88% sensitivity were obtained, but with a cutoff of 10.3pg/ml, the specificity increased to 91% while sensitivity decreased to 29%. With a GFAP cutoff of 93.8pg/ml the specificity was 91% and sensitivity 31%. When cutoff was set at 39.1pg/ml, the specificity decreased to 32% while sensitivity increased to 88%.

The NfL data was compared to a large reference control cohort of 1926 people (aged 5-90 years old). That comparison showed that a greater number of patients with epilepsy (20%) had NfL levels greater than the 95th percentile for their age, compared to 5% with levels above the 95th percentile in the FS group. The Other group however had 2/12 (17%) of cases with NfL levels higher than the 95th percentile. In the Other group, one patient had alcohol-related brain injury and another had dementia. Interestingly, the authors decided to evaluate the results in a subgroup of patients stratified based on their age. This comparison appeared to be more fruitful. Using the cutoff of 95th percentile for age matched and looking only at patients younger than 60 years of age, high NfL level was associated with 95% specificity and 24% sensitivity, 4.72 positive likelihood ratio, 0.8 negative likelihood ratio, 95% positive predictive value and 22% negative predictive value for a diagnosis of epilepsy. Further evaluation of those patients whose NfL was above the 95th percentile, showed that they were younger and prescribed more medication than those whose levels were at or less than the 95th percentile (37.5 vs 43.8 years, p = .03 and 3.19 vs 2.29 drugs, p = .01, respectively). A similar control data set was not available for GFAP, such that a similar comparison could not be done. The authors also looked for differences in clinical variables between quartiles of GFAP levels. There was none.

This is an important study that builds on previous knowledge of neuronal damage caused by seizures.^5,7 There is still debate regarding the detectable amount of neuronal injury caused by focal versus generalized seizures, or if a structural abnormality (such as stroke) may further increase the levels of markers of neuronal injury. However, this study suggests that markers such as NfL and GFAP, may have a role in determining if a patient has epilepsy or functional seizures. In fact, there is a high (95%) specificity seen with a cutoff above the age-matched 95^th percentile in patients 60 years or younger. This finding can be extremely helpful, suggesting that only a very small proportion of patients aged 60 years or younger with FS could be incorrectly diagnosed as having epilepsy, based on NfL alone. The evidence for GFAP is not that strong, however if larger control cohorts with broad age ranges are available, that marker may also be useful in the future.

Despite the exciting results, several things still need to improve before NfL (and GFAP) levels can be widely used to help in the diagnosis of FS. For instance, how much do these levels vary in relation to seizure occurrence? There seems to be a weak correlation between the number of days between the last seizure and NfL and GFAP levels. ⁵ A previous study had suggested that NfL and GFAP levels are higher in patients with seizures within 2 months of testing, compared to those that were seizure free for 12 months. However, if blood is collected within 2 weeks rather than 2 h after the “event”, how helpful is NfL in determining if the event was an epileptic or FS? Can the levels be different if a seizure was generalized compared to a focal onset? Could seizure duration or clustering also influence the levels of markers of neuronal injury? Another potential complication is that levels of NfL increase with age in control groups. Perhaps this is because those “controls” in fact already have some degree of neuronal injury due to atherosclerosis or initial neurodegenerative processes not related to epilepsy, or just “ageing”. As this study showed, plasma markers of neuronal injury are not very helpful in patients older than 60 years of age. It is also unclear how those markers would behave in the pediatric population. Finally, a head-to-head comparison with prolactin is still warranted to determine the usefulness of NfL and GFAP in aiding the diagnosis of FS.

Despite these few limitations, the work presented by Dobson and colleagues is very innovative and paves the way to a new era of investigating ES and FS. We look forward to the next steps, hoping it will not be too long before we can incorporate markers of neuronal injury in FS investigations as new practice standards.