A Phase 1B Trial in GBA1‐Associated Parkinson's Disease of BIA‐28‐6156, a Glucocerebrosidase Activator

Loss‐of‐function mutations in the GBA1 gene are one of the most common genetic risk factors for onset of Parkinson's disease and subsequent progression (GBA‐PD). GBA1 encodes the lysosomal enzyme glucocerebrosidase (GCase), a promising target for a possible first disease‐modifying therapy. LTI‐291 is an allosteric activator of GCase, which increases the activity of normal and mutant forms of GCase.


Introduction
Parkinson's disease (PD, MIM: 168600) is the second most common neurodegenerative disorder and has a likely multifactorial disease etiology, consisting of both environmental and genetic risk factors. 1 A diseasemodifying treatment is lacking. Mutations in the GBA1 gene are one of the most common genetic risk factors for Parkinson's disease (GBA-PD). 2-6 GBA-PD presents at a slightly younger age than idiopathic PD, with a greater prevalence of nonmotor symptoms. [7][8][9] Clinically, however, on an individual level GBA-PD is indistinguishable from iPD. GBA1 encodes the lysosomal enzyme glucocerebrosidase (Fig. 1A, GCase; EC 3.2.1.45), and risk-associated GBA1 mutations cause a loss of enzymatic activity. This gene is primarily known for causing the rare lysosomal storage disorder Gaucher's disease (GD) in the case of biallelic mutations. In GD, more than 400 pathogenic variants have been described (Hruska et al., 2008;Stenson et al., 2017), 10,11 likely with a "dose-effect" based on severity of the variant, however, with a large variability in the genotype-phenotype relationship. For most relatively common variants, the increased risk for PD has been determined, which is impossible for the long tail of rare variants, but these are expected to behave similarly. In addition to their role in PD risk, GBA1 mutations have been related to more rapid progression of motor 12,13 and cognitive 14,15 symptoms of PD. Activation of GCase may be a promising strategy for a possible first disease-modifying therapy in PD.
GCase functions at the luminal face of the lysosomal membrane and catalyzes one step in the multistep hydrolytic degradation of glycosphingolipids (GSLs), leading ultimately to sphingosine, the building block for the synthesis of new GSLs (Fig. 1A). 16,17 GSLs are essential for the maintenance of membrane properties that play a role in many diverse cellular functions. 18 GCase hydrolyzes glucosylceramide (GluCer), producing glucose and ceramide. Ceramide is subsequently deacylated to produce sphingosine, which is transported to the cytosol and may be recycled into ceramide The SL pathway is a closed system, with de novo synthesis as input and sphingosine phosphate lyase (the final enzyme in the sphingolipid degradative pathway) as output. The former pathway is endogenously inhibited but may play a role in maintaining GSL (glycosphingolipid) flux when GCase activity is low. Two pools of GluCer (lysosomal and non-lysosomal) exist, which are not distinguishable in our measures. (B) GBA-PD patients recycle ceramide slowly, so de novo synthesis of ceramide is increased and GluCer levels are maintained (left panel). Treatment with LTI-291 increases the availability of cytosolic ceramide, resulting in a transient increase in GluCer synthesis (middle panel). However, increased ceramide levels are known to result in decreased de novo synthesis, bringing steadystate levels of GluCer and Cer back to the pre-dose levels at 28 days. However, the pre-dose and day 28 pathways differ, in that the day 28 pathway has greater GSL flux (comparable amount but faster rate) and reduced de novo synthesis. GCase, glucocerebrosidase; Cer, ceramide; GluCer, glucosylceramide; LacCer, lactosylceramide; GluSph, glucosylsphingosine; S1P lyase, sphingosine phosphate lyase. [Color figure can be viewed at wileyonlinelibrary.com] again (Fig. 1A). Ceramide and all of its conjugates, including GluCer, comprise a group of acyl chain isomers that are produced from sphingosine, by acylation with activated fatty acids of diverse chain lengths by ceramide synthases. 19 The acyl chain isomers of ceramide do not interconvert (eg, by addition or removal of carbons from the acyl chain). All ceramide isomers can be glucosylated by glucosylceramide synthase to produce the GluCer isomers, the starting points for ganglioside synthesis in the Golgi.
LTI-291 (now designated BIA-28-6156) is a smallmolecule GCase allosteric activator. GCase activation has been measured in-vitro in lysed cells using the fluorogenic probe (4-MUG; 4-Methylumbelliferyl β-Dgalactopyranoside), however given the reversible nature of the compound, a method in intact cells would be more appropriate (using the existing probe: PFB-FDGlu; 5-(Pentafluorobenzoylamino) Fluorescein Dibeta-D-Glucopyranoside). This is however not feasible as this method does not allow the measurement of BIA 28-6156 effect, possibly because the fluorophore is occupying BIA-28-6156 binding site. As an alternative, we hypothesized that time-dependent changes in the levels of substrate GluCer and related molecules may be used to infer enzyme activity/activation. In this exploratory study, we measured GluCer and lactosylceramide (LacCer) in peripheral blood mononuclear cells (PBMCs), plasma, and cerebrospinal fluid (CSF), and we measured glucosylsphingosine (GluSph) in PBMCs and plasma. We did not measure ceramide, the product (in addition to glucose) of hydrolyzation by GCase, because this is also regulated by several other pathways and was therefore considered too variable. It is important to note that, as predicted by Michaelis-Menten kinetic theory, 20 GluCer levels are not sensitive to GCase activity when GCase activity exceeds about 30% of the normal, or average, level, 21 as is the case in GBA-PD (in contrast, GD, which is characterized by very low GCase activity, is characterized by the accumulation of GluCer isomers in peripheral cells and tissue).
LTI-291/BIA-28-6156 increases V max and decreases K m of GCase wild-type and at least some mutant enzymes, such that in vitro activity is increased by up to threefold (LTI, unpublished). When administered to healthy volunteers for 14 days, with a maximal single dose of 90 mg and multiple daily doses of 60 mg, LTI-291 was generally well tolerated, without any treatment-emergent serious adverse events (SAEs) or any adverse events (AEs) that led to discontinuation. 22 No AEs were attributed as being likely related to the administration of LTI-291/BIA-28-6156. CSF unbound drug concentrations were estimated to be in an approximate ratio of 1:1 with the unbound plasma drug concentration, across all doses, indicating excellent central penetrance. Based on in vitro studies, the central exposures reached by multiple LTI-291 doses (10-60 mg) were sufficient to at least double in vitro GCase activity (LTI, unpublished). Doubling of GCase activity is expected to restore 100% of average non-GBA-PD activity in most, if not all, GBA-PD patients. In an earlier study of LTI-291/BIA-28-6156 in healthy elders, 22 intracellular GluCer isomers (in PBMCs) did not change significantly over 14 days of dosing. This may be attributable to the possibility that these healthy volunteers had "normal" GSL flux, which cannot be increased by further GCase activation. The same GluCer isomers were measured again as exploratory biomarkers for the current 28-day study in GBA-PD patients, with significantly different results. This paper describes these studies, assessing safety, tolerability, pharmacokinetics (PKs), and pharmacodynamics, in GBA-PD in a 28-day treatment trial of LTI-291.

Patients and Methods
This was a randomized, double-blind, and placebocontrolled trial. The study was approved by the Independent Ethics Committee of the Foundation "Evaluation of Ethics in Biomedical Research" (Stichting Beoordeling Ethiek Biomedisch Onderzoek), Assen, the Netherlands. The trial is registered in the Dutch Trial Registry (Nederlands Trial Register) under study number NTR6960. The trial was conducted between January and June 2018 at the Centre for Human Drug Research, Leiden, the Netherlands. All participants signed an informed consent form before any study-related activity, in accordance with the Declaration of Helsinki.

Participants
GBA-PD patients (minimum age of 18 years), with Hoehn and Yahr (H&Y) stages 1 to 4 and a Mini-Mental State Exam (MMSE) score ≥18, male and female of nonchildbearing potential, were enrolled for 28 consecutive daily oral doses of LTI-291 or placebo. Stable treatment with antiparkinsonian treatments from 1 month before the screening (2 months for monoamine oxidase B inhibitors) was allowed. Other prior concomitant medication was allowed only at the discretion of the investigator. The following dose levels were investigated: 10, 30, and 60 mg of LTI-291. Treatment was administered as powder in a capsule. Each treatment arm consisted of 10 patients. Patients were randomly assigned in 10 blocks of 4 to receive one of the three dose levels of LTI-291 or placebo in the ratio of 1:1:1:1. The randomization code was generated using SAS version 9.4 by a study-independent statistician. Patients visited the clinical research unit at the start of dosing and after 1, 2, and 4 weeks. A safety call was performed after 3 weeks. Between visits, patients self-administered LTI-291 daily. A safety follow-up visit was performed 7 to 14 days after the last dose.

Safety
A medical screening (medical history, record of prior concomitant medication, participant demographics, height and weight, 12-lead electrocardiography [ECG], vital signs, routine hematology, biochemistry/ electrolytes and urinalysis, urine pregnancy test [for females], virology, urine drug screen, ethanol breath test, physical examination, MMSE, and H&Y staging) was performed to assess a participant's eligibility. During study periods, safety was assessed using monitoring of AEs, concomitant medication, vital signs, ECG, physical examination, and safety chemistry and hematology blood sampling.

Pharmacokinetics
LTI-291/BIA-28-6156 levels were measured in K 2 EDTA plasma and in CSF. Plasma PK samples were taken pre-dose and 2, 4, and 6 hours after the first and last doses and a single sample after the 7th (AE2) and 14th (AE2) doses. CSF was taken pre-dose and 4 hours after the last (28th) dose. Noncompartmental analysis was performed on the plasma data from each participant as data permitted.

MDS-UPDRS Part III, MMSE, and Neurocognitive Biomarkers
No clinical effect was expected after 28 days of LTI-291/BIA-28-6156 dosing, but the Movement Disorder Society-Unified Parkinson's Disease Rating Scale (MDS-UPDRS), 24 Part III (motor assessment), in on and off state and the MMSE were performed at baseline and at the end of dosing as pharmacodynamic parameters for safety.
The NeuroCart, 25 a central nervous system (CNS) test battery, was used to exclude any adverse effects of LTI-291/BIA-28-6156 on CNS function. This was performed at baseline and after 2 weeks of dosing, to spread the burden of different measurements over different visits. Steady-state exposures of LTI-291/BIA-28-6156 were expected to be achieved after 7 days of dosing. The test battery consists of neurophysiological, psychomotor, and cognitive tests and has been extensively used previously in clinical drug development. [25][26][27][28][29] In short, measurements consist of saccadic and smooth pursuit eye movements, the adaptive tracking test (a visuomotor task sensitive to disturbances in vigilance and attention), the body sway (a test of postural stability), the Bond and Lader test (visual analog scale of alertness, calmness, and mood), the Visual Verbal Learning Test (a test of immediate and delayed memory), and pharmacoelectroencephalogram (EEG, measured separately after the last dose instead of after 2 weeks). Tests were performed in a quiet room with ambient illumination with only one participant in the same room (and a research assistant) per session.

GBA1 Genotyping
GBA1 genotype was determined in a previous largescale GBA1 screening in the Netherlands. 4 In short, full gene sequencing was performed on saliva-derived DNA, using next-generation sequencing and a primer set unique for the functional gene, thereby preventing amplification of the nearby pseudogene. For this trial, GBA1 genotypes were confirmed by repeating sequencing in a whole blood sample (Table 1).
GBA1 genotypes were categorized into two: (1) carriers of one allele that has been reported in at least a single GD or (2) carriers of a non-GD GBA1 allele linked to PD risk, for alleles associated with PD or reported in PD patient(s), but never GD. It is important to emphasize that, although GBA1 genotype is related to average residual GCase activity, there is considerable interindividual variation and overlap between genotypes.

Statistical Analysis
Neurocognitive pharmacodynamic data were analyzed with an analysis of covariance with fixed factor treatment and average pre-dose value as covariate.  Biochemical pharmacodynamic data were analyzed using a linear mixed model, with fixed factors treatment, time and treatment by time; random factor participant; and covariates average baseline value, sex, age, and GBA1 type (GD or non-GD carriers). An overall treatment effect was assessed, and an effect over time, both for all active dose levels combined and per-dose level compared to placebo. Statistical programming was conducted using R version 3.6.2 (December 12, 2019).
This was an exploratory study; therefore, the sample size was not based on statistical considerations. Ten patients per dose level and 10 placebo patients were considered adequate to define initial safety and tolerability and to explore pharmacodynamics in the target patient population over 28 days of dosing.
Biomarkers were measured in an exploratory hypothesis-generating setting and were therefore not corrected for multiple testing.

Results
Forty-nine participants signed the informed consent form and underwent a medical screening. Seven participants were not enrolled because they were excluded based on the inclusion and exclusion criteria or withdrew consent to participate. A total number of 42 participants were enrolled. Two participants were excluded before the first dose based on physician decision (significant ECG abnormalities, not visible at screening). In total 40 participants were treated in the study, and all completed the study, including the follow-up visit (Supplementary Figure 1).

Demographics and Baseline Characteristics
In total 20 men and 20 women were included in the study. The mean weight ranged from 69.0 kg (10 mg of LTI-291) to 81.0 kg (30 mg of LTI-291). Participants in the different dose levels were comparable regarding mean age, mean height, mean MMSE, and mean MDS-UPDRS Part III score. Demography data are summarized in Table 1.  Table 2 for a full list of all AEs after dosing. Back pain was reported only in LTI-291/BIA-28-6156 dose groups (N = 4) and not in placebo; however, there is no clear rationale for this, and there was no dose-dependent increase in frequency; therefore, this was considered unlikely related to the administration of LTI-291/BIA-28-6156. Other frequently reported AEs like fatigue and headache occurred in a similar or higher frequency in the placebo group, and no dose-dependent increase was observed; therefore, these are also considered unlikely related to the administration of LTI-291/ BIA-28-6156. Five participants (LTI-291 n = 4, placebo n = 1) reported a mild subjective worsening of Parkinson's disease symptoms. Three participants related this to a stressful period. In 4 of 5 participants this subjective worsening of symptoms resolved before the last dose of LTI-291/BIA-28-6156 or placebo. Most AEs were mild in severity. Only three moderate AEs were reported, namely urinary tract infection (10 mg of LTI-291), tendonitis (placebo),   and paronychia (30 mg of LTI-291). Both infections were successfully treated with antibiotics, and the participant with tendonitis was referred for physiotherapy. These three AEs were considered to be unlikely related to LTI-291/BIA-28-6156 treatment.

Safety and Tolerability
Pharmacokinetics PK analysis of LTI-291/BIA-28-6156 showed a maximum plasma concentration (T max ) ranging from 2 to 6 hours. C max and AUC 0-6 increased dose proportionally. Half-life could not be determined due to limited sampling, but the PK profile otherwise was similar to results from previous studies in healthy volunteers. 22 The CSF to plasma concentration ratios range from 0.00634 to 0.0187 and were similar at all the dose levels. See Supplementary Tables 3 and 4 for details.

Group Average Levels of Intracellular GluCer Isomers in PBMCs Significantly Increase 14 Days after Dosing with LTI-291/BIA-28-6156 and Then Partially Return to the Pre-Dose Level
In PBMCs, GluCer concentrations (C16:0, C22:0, C24:0, and C24:1) showed a statistically significant overall treatment-associated increase in all active treatment groups (doses were combined because all doses were expected to at least double activity based on estimated brain exposure) at all times combined compared to placebo (Table 2A). The effect was significant in the 10-mg-LTI-291/BIA-28-6156 and the 60-mg-LTI-291/BIA-28-6156 dose groups but not in the 30-mg-treated group (Table 2A). The effect was largest on day 14 (Table 2B; Fig. 2; Supplementary  Table 5). Age, sex, and GBA1 genotype were not significant covariates. LacCer and GluSph in PBMCs were omitted from analysis, because of influence of leukocyte subtype ratios (including granulocyte contamination), which vary between blood draws. No significant changes were detected in extracellular GluCer levels at any time. Plots of raw measurements over time are provided in the Supplementary Material.

MDS-UPDRS Part III, MMSE, and Neurocognitive Biomarkers Were Unchanged by 28 Days of Dosing
No clinically significant changes were observed in MDS-UPDRS Part III (ON state) or MMSE total score in any dosing group compared to placebo (Table 3). See Supplementary Table 6 for details.
There were no dose-dependent effects of LTI-291 on any of the neurocognitive biomarkers, indicating that 28 consecutive oral doses in participants with GBA-PD were not observed to cause any effects on CNS functioning. Some isolated differences from placebo were observed in single, mostly submaximal, dose levels, but due to the lack of dose dependency these were considered chance findings due to multiple testing (Supplementary Table 7).

Discussion
Here we report the first administration of LTI-291 (now designated BIA-28-6156), a centrally penetrant small molecule, aimed at increasing GCase activity in patients with GBA-PD. Safety, tolerability, PKs, and pharmacodynamics of LTI-291/BIA-28-6156 were evaluated. LTI-291/BIA-28-6156 was administered in 28 consecutive daily doses at 10, 30, or 60 mg. This was generally well tolerated, no treatment-related SAEs or deaths occurred, and no participants withdrew due to AEs.
We were not able to demonstrate a decrease in GluCer as target engagement of the compound. In contrast, an unexpected significant and transient increase in four of five intracellular GluCer isomers was detected in PBMCs in dosed participants as compared to placebo (the fifth, which is also the lowest in abundance, was also increased but not statistically significantly so). No change in extracellular GluCer was observed in plasma at any time or in the CSF at 28 days (not shown). Intracellular GluCer levels and plasma GluCer levels do not correlate. 23 The observed increase in intracellular GluCer was slow, with no change at 6 hours, a mild increase after 7 days of dosing and a significant increase after 14 days of dosing. A second phase of the response was suggested by the fact that GluCer levels seemed to return toward pre-dose levels by day 28 (Table 2B; Fig. 1B). The observed changes in intracellular GluCer levels were unexpected, since it had been hypothesized by us and others that activation of GCase would reduce GluCer levels. 30,31 While physiological and methodological variability should always be considered as an explanation for the observed signal in this exploratory setting, our straightforward analysis shows a clear difference in treated patients compared to placebo. Furthermore, our preclinical studies confirmed that BIA-28-6156 is a highly selective GCase activator (LTI unpublished), supporting that the observed effects derive from GCase activation. We have refined our initial hypothesis to include the fact that GCase is part of a cyclic pathway (Figure 1), 32 where GluCer lies both downstream as well as upstream of GCase; we thus propose a datadriven hypothesis: It should be noted that intracellular GluCer measures include lysosomal GluCer (the GBA1 substrate), as well as non-lysosomal or cytoplasmic GluCer. 27 Because the majority of GluCer is non-lysosomal, we hypothesize that the activation of GCase activity by dosing with LTI-291/BIA-28-6156 may cause a transient increase in salvaged ceramide available for GluCer synthase in the cytosol (Fig. 1B). 16,17 We hypothesize that as the systemic ceramide (and GluCer) levels increase, de novo synthesis, which is known to be sensitive to ceramide, 33 is downregulated. This effect subsides as the system returns to a new homeostasis (the return of GluCer toward pre-dose levels by day 28). This hypothesis is data driven based on these unexpected results. Ceramide levels, or other molecules in the GSL metabolism more distal to GCase, were not measured. No additional target engagement biomarkers were available. Additional experiments assessing flux in the GSL pathway are being performed to further evaluate the validity of our hypothesis.
Comparison to other (experimental) drugs targeting the same mechanism could be informative, but unfortunately no adequate data are available. Enzyme replacement therapy is given only to GD patients, which has a much lower residual GCase activity with the expected substrate accumulation (based on Michaelis-Menten kinetics), 19 as opposed to GBA-PD. Ambroxol, a GCase chaperone, was shown to decrease GluCer ex vivo in cultured GBA-PD macrophages, 31 but in vivo data are not yet available. Venglustat, a GluCer synthase inhibitor, reduced GluCer levels in a phase 2 GBA-PD trial (through a different mechanism, namely blockade of synthesis), 34 but the phase 3 trial was terminated due to lack of efficacy (ClinTrials NCT02906020 35 ).
GluCer transient elevation was observed for the 10and 60-mg treated groups, but small increases in the 30-mg group did not reach statistical significance. Based on preclinical experiments (LTI unpublished), an effect was expected with a C max of $360 ng/mL, with similar responses for higher dose levels, indicating a flattening of the dose response. In the 10-mg group, the mean C max was 554 ng/mL, showing all dose levels reached expected active concentrations. Lack of a clear signal in the 30-mg group can have different explanations: the inherent physiological or methodological variability in the biomarker, combined with a small sample size of subgroups; a predominance of nonresponders in this group; and the found treatment effect in other groups may be chance findings (however, the treatment effect in the total group seems the most robust assessment).
Variability in GluCer in PBMCs as a biomarker can also be observed in placebo data. No change over 28 days is expected in placebo-treated participants, so fluctuations in the placebo group likely reflect natural and methodological variability. The strongest signal at day 14 seems driven by both a GluCer increase in LTI-291-treated participants and a random trough in the placebo-treated group (Fig. 2). Nevertheless, the overall treatment effect is still statistically significant different in LTI-291-treated participants compared to placebo, accounting for this variability over time (Table 2). Considering the exploratory setting of these pharmacodynamic measurements, without correction for multiple testing, these effects require validation in a larger cohort.
Measurements in PBMCs of GluSph and various LacCer isoforms were heavily influenced by the cell subtype composition of the PBMC isolate. 23 This composition also varied within individuals between samples. Because this variation could not be distinguished from a potential treatment effect, these were omitted from analysis. GluSph may still be of interest for future trials, considering its hypothesized toxic effect in GD, so cell subtypes should be considered when measuring this.
GBA1 genotype category (GD risk [n = 25] vs. PD risk [n = 15]) was investigated as covariate. GD risk showed a trend for a stronger effect in all GluCer isoforms in PBMCs but did not reach statistical significance (data not shown). It can be speculated that patients with a larger GCase deficiency may have more benefit of treatment. Subgroups were small however, and GCase activity is known to vary between individuals with the same mutation. Whether this translates to a clinical effect will be determined in an upcoming trial.
PK sampling was limited to three plasma samples up to 6 hours post-dose on days 1 and 28, showing a dose-proportional increase in C max and AUC 0-6 . The mean CSF to plasma concentration ratios ranged from 0.0113 to 0.0122 at 4 hours after the 28th dose and were similar at all dose levels (Supplementary Table 4). This ratio corresponds with a free distribution of unbound LTI-291 between plasma and CSF, which again is in distribution equilibrium with brain tissue, as was shown in preclinical rat neuro PK experiments. This PK profile is similar to what was determined in healthy volunteers, 22 which also showed a median halflife of 28.0 hours, favoring daily single dosing.
A neurocognitive test battery showed no adverse effect on CNS functioning, performed after the 14th dose, during which steady-state LTI-291 plasma concentration was already achieved. No clinical improvement was expected after 28 days of dosing, and no deterioration was observed, as confirmed by MDS-UPDRS, Part III (motor assessment), and MMSE testing. The MDS-UPDRS was performed in ON state, because the burden of testing in OFF state was not considered justified, as no clinical change was expected. In a long-term study to assess clinical improvement, OFF state measures will be appropriate.
In 5 participants, a mild subjective worsening of PD symptoms was reported, which resolved before the end of dosing in 4 participants (3 active treatment and 1 placebo). Considering the natural variation in Parkinson's disease symptom severity and the progressive disease course, these complaints were considered unlikely to be caused by the administration of LTI-291. No relationship was observed between worsening of PD symptoms and changes in GluCer levels.
Lack of a control group with PD without a GBA1 mutation limits interpretation. Patients with GBA-PD are most likely to have an affected GCase and GSL metabolism but not necessarily exclusively so. A future study should assess the effect in PD without a GBA1 mutation.
It remains an open debate what the most effective therapeutic target may be based on GBA-PD pathogenic mechanisms. Other than increasing wild-type GCase activity, this may be refolding of mutant enzyme, supporting adequate transport of enzyme, preventing endoplasmatic reticulum stress, substituting with recombinant enzyme, and other options reviewed elaborately elsewhere. 36,37 In conclusion, LTI-291 was observed to be well tolerated when given orally once daily for 28 consecutive days at all dose levels tested in this GBA-PD population. Plasma concentrations that are expected to be active in at least doubling GCase activity were reached. A transient increase in GluCer was observed, possibly due to target engagement, but this requires further validation. A long-term (1 year) dosing study is being planned to assess clinical benefit.
A P H A S E 1 B T R I A L O F L T I -2 9 1 / B I A -2 8 I N G B A -P D