Safety and efficacy of an oral, fixed-dose sodium phenylbutyrate and taurursodiol combination (PB and TURSO) in amyotrophic lateral sclerosis (ALS) were evaluated in the multicentre phase 2 CENTAUR trial encompassing randomised placebo-controlled and open-label extension phases.1–3 The randomised phase primary end point was met, with PB and TURSO significantly slowing functional decline as measured by the ALS Functional Rating Scale-Revised (ALSFRS-R) over 24 weeks compared with placebo.2 Changes in plasma phosphorylated neurofilament heavy chain and neurofilament light chain (NfL) concentrations were evaluated as a secondary outcome and showed no between-group differences.2 4 Plasma samples were prospectively collected from CENTAUR participants for potential future biomarker analyses.4

Neuroinflammatory biomarkers may add value beyond other biomarkers for assessing disease progression and therapeutic response in ALS. Chitinases, a class of hydrolases expressed by activated microglia and astrocytes, have emerged as potential prognostic biomarkers in ALS given prominence of glial activation in the neuroinflammatory response.5 ALS is also characterised by a peripheral immune response that includes chitinase expression by activated myeloid cells, largely in response to cytokines.5 6 PB and TURSO was previously shown to significantly reduce concentration of the chitinase YKL-40 (also known as chitinase-3-like protein 1) in cerebrospinal fluid (CSF) compared with placebo in adults with mild cognitive impairment and mild to moderate Alzheimer’s disease (AD) dementia in a 24-week phase 2, multicentre, randomised trial (PEGASUS).7 While most YKL-40 data in ALS are also in CSF, limited data in blood (mainly serum)5 8 prompted our interest in evaluating plasma YKL-40 in an interventional trial setting.

Here, we describe results of exploratory post hoc analyses of neuroinflammatory biomarkers, including YKL-40, chitinase 1 (CHIT1) and the systemic inflammatory biomarker C reactive protein (CRP), in plasma samples from participants with ALS from CENTAUR. Results of these analyses were presented at the 2022 Annual Northeast Amyotrophic Lateral Sclerosis Consortium Meeting.4

Detailed methods for the CENTAUR randomised phase (NCT03127514) are reported elsewhere.1 2 CENTAUR enrolled adults with definite ALS (revised El Escorial criteria9) ≤18 months from symptom onset and slow vital capacity >60% predicted. Participants were randomised 2:1 to receive PB and TURSO (3 g PB/1 g TURSO) or matching placebo by mouth or feeding tube for 24 weeks. Continuation of stable-dose riluzole and/or edaravone was permitted.

Blood samples were drawn at baseline and every 6 weeks thereafter through week 24 (or early discontinuation) to obtain plasma and stored in the NEALS biorepository.10 YKL-40, CHIT1 and CRP immunoassays were conducted in a blinded manner using 0.5-mL plasma samples. YKL-40 and CHIT1 assays were developed by the Bowser Laboratory at the Barrow Neurological Institute, while the CRP assay was from Meso Scale Discovery (MSD). All assays were performed within the Clinical Laboratory Improvement Amendments–certified laboratory at nVector (previously Iron Horse Diagnostics) on the MSD platform and demonstrated a <5% coefficient of variation (CV) for intra-assay and interassay quality controls and standards; the average CV across all samples was <4%.

As for all prespecified efficacy outcomes in CENTAUR, the participant population for these analyses was the modified intention-to-treat (mITT) population (ie, all participants who received ≥1 dose of study medication and had ≥1 postbaseline ALSFRS-R assessment). Specifically, participants with ≥1 postbaseline plasma sample collected over the 24-week randomised phase were included. First, log10-transformed plasma biomarker measurements8 were analysed using a random-slope, shared-baseline, linear mixed-effects model adjusted for age and ALSFRS-R slope (ie, rate of ALSFRS-R total score change from symptom onset) interacting with time. Geometric least squares (LS) mean biomarker concentrations were calculated for each treatment group; given paucity of data at other time points (weeks 6 and 18), results for only weeks 12 and 24 are presented. Second, change-from-baseline analyses that did not assume a shared baseline were performed for all biomarkers. Geometric mean ratios from a mixed model for repeated measures without linear trend assumption were also calculated for weeks 12 and 24. Finally, Pearson correlation coefficients were calculated to assess the correlation between plasma concentrations of YKL-40, CHIT1 and CRP paired with ALSFRS-R total score and between concentrations of each biomarker paired with ALSFRS-R slope.

Of 135 participants in the mITT population (PB and TURSO, n=87; placebo, n=48), 126 (PB and TURSO, n=81; placebo, n=45) had plasma samples available for these analyses. Summary baseline biomarker concentrations are shown in table 1. Additional baseline characteristics for this population are summarised in online supplemental table 1 and were generally similar to those in the overall mITT population.2

Plasma biomarker concentrations*

Geometric LS mean YKL-40 plasma concentration was 10% lower (ratio, 0.90; 95% CI 0.83 to 0.97) at week 12 and approximately 20% lower (ratio 0.81; 95% CI 0.69 to 0.94) at week 24 in the PB and TURSO versus placebo group (p=0.008; table 1). Furthermore, YKL-40 concentration correlated with ALSFRS-R total score (r of –0.21; p<0.0001) and ALSFRS-R slope (r of 0.11; p=0.034). Change-from-baseline analyses also showed significant reduction in plasma YKL-40 concentration in the PB and TURSO versus placebo group (p=0.002; figure 1).

Change in YKL-40 plasma concentration relative to baseline over 24 weeks. Log10-transformed YKL-40 concentration measurements8 were analysed using a random-slope, shared-baseline, linear mixed-effects model adjusted for age and ALSFRS-R slope interacting with time (coloured lines, with shading reflecting SEs). Geometric mean ratios (solid dots) and 95% CIs (bars) from an MMRM model without linear trend assumption are shown at weeks 12 and 24. ALSFRS-R, Amyotrophic Lateral Sclerosis Functional Rating Scale-Revised; MMRM, mixed model for repeated measures; PB and TURSO, sodium phenylbutyrate and taurursodiol; YKL-40, chitinase-3-like protein 1.

Geometric LS mean CHIT1 plasma levels were not significantly different between treatment groups (table 1). CHIT1 concentration did not correlate with ALSFRS-R total score (r of –0.05; p=0.326) or slope (r of 0.10; p=0.061). Change-from-baseline analyses showed no significant differences between groups (p=0.247; online supplemental figure 1). Geometric LS mean CRP concentration was 17% lower (ratio 0.83; 95% CI 0.69 to 1.00) at week 12 and approximately 30% lower (ratio 0.69; 95% CI 0.48 to 1.00) at week 24 in the PB and TURSO versus placebo group (p=0.048; table 1). CRP concentration correlated with ALSFRS-R total score (r of –0.19; p=0.0002) and slope (r of 0.21; p<0.0001). Change-from-baseline analyses showed significant reduction in plasma CRP concentration in the PB and TURSO versus placebo group (p=0.018; online supplemental figure 2).

In our analyses of neuroinflammatory biomarkers in participants with ALS from CENTAUR, PB and TURSO significantly reduced plasma concentrations of YKL-40 and CRP but not CHIT1 relative to placebo over the 24-week randomised phase, with reductions observed as early as week 12. YKL-40 and CRP concentration further correlated significantly with disease progression as measured by the ALSFRS-R, the most widely used functional end point in ALS trials11 and the primary efficacy outcome in CENTAUR.

CSF YKL-40 concentration has been shown to be elevated in ALS and correlates with clinical measures of disease severity, including the ALSFRS-R, progression rate and survival.5 12 However, blood-based biomarkers offer the advantages of being more cost- and time-efficient and less invasive than CSF-based biomarkers.13 While few studies have explored longitudinal blood levels of YKL-40 in ALS,8 blood YKL-40 levels have been shown to be a treatment-responsive biomarker for multiple sclerosis.14 Notably, in line with the current analysis, PB and TURSO significantly reduced CSF YKL-40 levels compared with placebo in the PEGASUS trial in AD, a condition in which plasma and CSF YKL-40 levels have been shown to modestly correlate.15

CRP is a sensitive marker of systemic inflammation that is widely used given its availability and reliability. A systematic review found that CRP was significantly elevated and correlated with disease progression and survival in the majority of studies evaluating this biomarker in ALS, primarily in blood.6 CRP level was also noted to significantly decrease after riluzole administration in an unblinded surveillance study of people with ALS presenting to a single clinic setting, suggesting CRP may be a potential biomarker for assessing treatment responsiveness in this population.16 However, CRP may have limitations as a neuroinflammatory biomarker in ALS, including potential for confounding by comorbid inflammatory conditions6 and the high variability observed for this biomarker.17 Like YKL-40, CHIT1 has been shown to be elevated in ALS5 12 but is primarily expressed by circulating cells of myeloid lineage.5 18 In this analysis, CHIT1 levels did not differ significantly between treatment groups.

In contrast to YKL-40 and CRP in the current analyses, plasma neurofilament levels were not significantly changed with PB and TURSO treatment in CENTAUR.2 PB and TURSO similarly had no significant effect on CSF NfL levels over a similar duration in the PEGASUS trial in AD.7 While neurofilaments are a promising biomarker for ALS, the mechanistic linkage to ALS remains under study and other biomarkers may capture different aspects of ALS pathophysiology; though some studies have shown concomitant effects on clinical outcomes and neurofilament levels, others showed clinical effects without changes in neurofilament levels.2 19 20 Ultimately, translation of the differential effects of PB and TURSO on plasma biomarkers in CENTAUR into the exact mechanism of PB and TURSO in ALS requires further study.

We demonstrated significantly reduced YKL-40 and CRP plasma levels in participants receiving PB and TURSO in CENTAUR, with lower YKL-40 and CRP concentrations correlating with higher ALSFRS-R total scores and lower rate of ALSFRS-R total score decline through 24 weeks. Analyses of neuroinflammatory biomarkers in the ongoing 48-week phase 3 clinical trial of PB and TURSO in ALS (PHOENIX) are planned to confirm these results. Ultimately, further delineation of the utility of these biomarkers in ALS may provide a tool for evaluating therapies targeting neuroinflammation in ALS trials.

Deidentified participant data will be made available on reasonable request. Requests for data sharing can be sent to info@amylyx.com.

Not applicable.

The CENTAUR trial involved human participants and was approved by a central institutional review board, the Partners Human Research Committee, for all trial sites (no reference number or ID associated). Participants gave informed consent to participate in the study before taking part.

The authors would like to thank the people living with ALS who participated in the CENTAUR trial, as well as their families and caregivers, and the Northeast Amyotrophic Lateral Sclerosis Consortium (NEALS). Lara Primak, MD, and Nikhilesh Sanyal, PhD, of PRECISIONscientia provided medical writing assistance with the development and revision of the manuscript under the direction of the authors, with financial support from Amylyx Pharmaceuticals, Inc., and in compliance with international Good Publication Practice guidelines.