(M0930-11-63) Evaluation of Phosphorylated Neurofilament Heavy Chain (pNF-H) as a Biomarker to Bridge Riluzole Treatment and Motor Outcomes in Patients with Acute Spinal Cord Injury (SCI)

Purpose: Riluzole in Acute Spinal Cord Injury Study (RISCIS) Phase II/III trial (ClinicalTrials.gov- NCT01597518) was conducted to determine the relationship between pharmacokinetics of riluzole and neurological outcomes in patients with Spinal Cord Injury (SCI). However, since these clinical outcomes were measured at the admission and 6-month after injury, it is quite challenging to correlate 14-day riluzole exposure with its clinical response months after the therapy. To overcome the temporal gap between the treatment and motor outcomes, we performed further analysis by using a novel biomarker, phosphorylated neurofilament heavy chain (pNF-H). The pNF-H is one of the structural proteins of neurons, and its elevated serum levels indicate the breakdown of axons. Previous studies in literature show the possibility of pNF-H as a diagnostic biomarker for SCIs by demonstrating the positive correlation between the severity of SCI and plasma pNF-H levels: pNF-H levels appear to be elevated in accordance with the severity of SCI due to a greater magnitude of axonal damage [1][2]. Based on this evidence in previous research, the aim of this study is to scrutinize the possibility of pNF-H as a prognostic and predictive biomarker in patients with SCI by assessing the correlation between the improvement of motor scores in 6 months and the 2-week change in pNF-H concentration after 14-day riluzole treatment.

Methods: Patient data were obtained from RISCIS, a multi-center, randomized, placebo controlled, double-blinded, phase 2/3 clinical trial. Patients in the treatment arm received a 100 mg of riluzole twice on day 1, then 50 mg of riluzole twice daily for the following of 13 days. Plasma sampling was conducted at 3 hours post dose and before the next dose, on days 3, 7, 10 and 14. With these samples, the concentrations of riluzole were quantified using a validated LC-MS/MS assay with LLOQ of 0.5 ng/ml, and pNF-H levels were measured by sandwich enzyme-linked immunosorbent assays (ELISA). PK analysis was performed using Phoenix NLME version 8.2. Lastly, to examine clinical response of riluzole, Total Motor Score (TMS) were assessed by International Standards for Neurological Classification of Spinal Cord Injury Examination (ISNCSCI) at baseline and 6 months.

Results: Plasma samples from thirty-two SCI patients were analyzed in the PK sub-study of the trial (N=18 in placebo and N=14 in treatment groups). The time course of pNF-H progression in the placebo and treatment group demonstrated that pNF-H in the placebo group was more rapidly increased, while the treatment arm showed diminished increase in pNF-H concentrations (Table 1). In other words, the placebo group reflects a natural SCI progression in patients without any treatment induced by neurodegeneration over time, while the treatment group with neuroprotective riluzole experienced an offset against axonal breakdown causing less production of pNF-H. To characterize the time-variant parameters of the pNF-H production rate from the natural progression, and the pharmacodynamics of riluzole and its inhibition of pNF-H production, one of the indirect pharmacodynamic response (IDR) models was utilized (Fig. 1). Since riluzole indirectly prevented from axonal degradation and consequently slowed down the production of pNF-H, ‘Inhibition of K_in’ model was adapted for this study. Via the PK/PD model, the Area Between the Baseline and Effect Curves (ABEC) was derived as a secondary parameter. ABEC can be described as: ABEC=|pNFH_nat×t_r-AUEC_0-tr|
where pNFH_nat×t_r means the baseline values of pNF-H levels over the time interval of 0 to t_r in the placebo group, AUEC_0-tr defines the area under the response vs. time curve over the time interval of 0 to t_r. These ABEC values reflect the area between the natural disease progression without riluzole and the delayed onset of neurodegeneration due to the riluzole treatment. The multiple linear regression model showed that ABEC was a statistically significant predictor of change in TMS while baseline score was not (Fig.2 Bottom). The correlation between change in score at 6 month and ABEC of pNF-H inhibition was further established through a linear progression with a marginally statistical significance (Fig. Top). This demonstrated that it was feasible to predict the motor outcomes using the drug exposure and time courses of pNF-H levels during the 14-day riluzole treatment.

Conclusion: The pharmacodynamic models of pNF-H demonstrated the efficacy of riluzole in patients with SCI, as the treatment group exhibited a reduced increase in pNF-H due to a reduced axonal damage and a consequent offset against natural disease progression. The indirect response model successfully defined the time course of pNF-H in relation to the riluzole exposure. The natural disease progression and inhibition of axonal degradation over time were matched to the trends in patients’ pNF-H levels. Based on the PK/PD model, ABEC served as a statistically significant predictor of total motor score at 6-month. By bridging the temporal gap between the treatment and motor scores through a biomarker pNF-H, the correlation between longitudinal motor outcomes and riluzole treatment was apparent in SCI patients.

References: 1. Hayakawa K, Okazaki R, Ishii K, Ueno T, Izawa N, Tanaka Y, Toyooka S, Matsuoka N, Morioka K, Ohori Y, Nakamura K, Akai M, Tobimatsu Y, Hamabe Y, Ogata T. Phosphorylated neurofilament subunit NF-H as a biomarker for evaluating the severity of spinal cord injury patients, a pilot study. Spinal Cord. 2012 Jul;50(7):493-6. doi: 10.1038/sc.2011.184. Epub 2012 Jan 24. PMID: 22270191.
2. Shaw G, Yang C, Ellis R, Anderson K, Parker Mickle J, Scheff S et al. Hyperphosphorylated neurofilament NF-H is a serum biomarker of axonal injury. Biochem Biophys Res Commun 2005; 336: 1268–1277.

Acknowledgements: US Army Medical Research Acquisition Activity, Contract No. W81XWH-16-C-0031


Table 1. Descriptive statistics of pNF-H levels over time in placebo and treated patients.


Fig.1. Proof-of-concept model to characterize the time course of motor score improvement (clinical outcome) in relation to riluzole exposure via PK/PD correlation with pNF-H


Fig. 2. Linear regression of 6-month changes in total motor score vs. ABEC with confidence and prediction interval (Top). Summary of the multiple linear regression model for change in total motor score with ABEC and baseline score as predictors (Bottom).