Effect of the Histone Deacetylase Inhibitor FRM-0334 on Progranulin Levels in Patients With Progranulin Gene Haploinsufficiency

Key Points Question What is the safety, tolerability, pharmacodynamic, and pharmacokinetic profile of the histone deacetylase inhibitor FRM-0334 in participants with progranulin gene (GRN) haploinsufficiency? Findings In this randomized placebo-controlled clinical trial including 27 participants with GRN haploinsufficiency, FRM-0334 was safe and well tolerated; however, it failed to increase plasma progranulin and cerebrospinal fluid progranulin and failed to demonstrate dose-dependent oral bioavailability. Meaning The studied formulation of FRM-0334 should not be investigated in future clinical trials enrolling participants with GRN haploinsufficiency; this study did not fully address the potential of histone deacetylase inhibition to alter GRN expression.


Introduction
Autosomal dominant variations in the progranulin (PGRN) gene GRN (OMIM 138945) [1][2][3][4][5][6] are among the most common causes of genetic frontotemporal degeneration (FTD). 7Pathogenic GRN variations result in haploinsufficiency, [7][8][9][10] with a variable, roughly 50% reduction in secreted PGRN protein in the plasma 11,12 and cerebrospinal fluid (CSF). 11The potential to restore or replace PGRN has spurred the development of a number of new GRN therapies that are entering human clinical trials, with 1 agent having recently entered a phase 3 trial. 13Clinical development programs have been aided by the ability to use blood and CSF PGRN concentrations as pharmacodynamics biomarkers in GRN haploinsufficiency clinical trials.However, the rarity of trial-ready participants and lack of established end points are barriers to successful development of new therapies.
The histone deacetylase (HDAC) inhibitor FRM-0334, formerly EVP-0334, was designed as a potent, orally available drug with brain-to-plasma concentration ratios of 3 to 6 in rodents. 14In preclinical studies, a similar HDAC inhibitor, suberoylanilide hydroxamic acid, enhanced PGRN transcription to near-normal levels in human-derived GRN -/+ (R493X) cells (−/+ denotes that the cells have heterozygous loss of function of the GRN gene) and prevented cytosolic TAR DNA-binding protein 43 accumulation in PGRN-deficient lymphoblasts. 15,16Although suberoylanilide hydroxamic acid is approved for treatment of cutaneous lymphoma, 17 its poor central nervous system availability 18 prioritized the exploration of FRM-0334 in GRN haploinsufficiency.In subsequent unpublished preclinical studies, FRM-0334 boosted GRN mRNA and PGRN in a dose-dependent manner in human GRN -/+ lymphoblasts and in the cortex of mice (dosed with oral FRM-0334).
Additionally, single and multiple doses for FRM-0334, ranging from 10 mg to 400 mg daily, were safe, well tolerated, and associated with dose-proportional plasma exposure in an unpublished, phase 1, placebo-controlled clinical trial of FRM-0334 in 87 healthy volunteers aged 18 to 65 years.We investigated the safety, tolerability, and plasma pharmacodynamic (PD) effects (change in PGRN) of FRM-0334 in a multinational clinical trial completed in PRGN GRN haploinsufficiency.We also trial implementing an international, multisite protocol as a mechanism for patient recruitment in the rare familial FTD space.frontotemporal lobar degeneration [FTLD] behavior and language domain sum of boxes <16) 19,20 FTD stages.Race and ethnicity information was recorded; participants self-identified their race and ethnicity.All participants were recruited and randomized from January

General Procedures
Participants and site staff were blinded to treatment assignment.A central Interactive Voice/Web Response System 21 generated randomization assignments in 2 sequential cohorts.Cohort 1 was randomized to placebo or low-dose (300 mg daily) FRM-0334.Cohort 2 began after review of cohort 1 safety data and included randomization to placebo or high-dose (500 mg daily) FRM-0334.Overall, 5 patients were randomized to placebo, 11 to low-dose FRM-0334, and 11 to high-dose FRM-0334.
The study drug (FRM-0334 drug substance blended with sodium lauryl sulfate, microcrystalline cellulose, magnesium stearate, and colloidal silicon dioxide) and placebo (containing microcrystalline cellulose) were presented as identical, opaque, white gelatin capsules and taken daily for 28 days.A central pharmacy was used to distribute packaged drugs to blinded site staff.Plasma collection occurred at screening visit 1 (8-30 days before the first dose), at screening visit 2 (1-7 days before the first dose), on study day 1 (first day of dose), in weekly intervals during dosing (days 7, 14, 21, and 28), and at 10 days after the final dose (day 38).Lumbar punctures for CSF collection were performed using standard clinical practice, aligned with Alzheimer's Disease Neuroimaging Initiative procedures, 22 on screening day 2 and study day 28 (after final study dose).Cerebrospinal fluid was collected in the afternoon using a gravity drip into a polypropylene collection tube.Magnetic resonance imaging was performed on screening day 1 for safety and diagnostic purposes using scanners, field strengths, and protocols employed for clinical use at each site (eTable 1 in Supplement 2).Brain FDG-PET was performed on screening day 2 and study day 28.

Outcomes
The primary outcome of safety and tolerability was assessed via treatment-emergent adverse events (AEs), serum chemistry, hematology, urinalysis, vital signs (including orthostatic measurements), 12-lead electrocardiogram, physical examination, and suicidality assessments. 23The coprimary  Non-normally distributed dependent variables were log converted before inclusion in our models.

FDG-PET Processing and Analysis
Baseline FDG-PET data were only available for 26 of 27 individuals randomized to the study drug.
Baseline FDG-PET data for the remaining participant could not be recovered after conclusion of the trial, and this participant was excluded from our imaging analyses.A summary of scanner types and

Magnetic Resonance Imaging Processing
Magnetic resonance imaging processing methods and scanner characteristics are further detailed in the eMethods and eTable 1 in Supplement 2. Though the heterogeneity of scanners and field strengths across sites limited the interpretability of volumetric data, each participant's relative degree of total brain atrophy was estimated by dividing total parenchymal volume by total intracranial volume.This parenchymal volume variable was exclusively used as a covariate in follow-up sensitivity analyses, controlling for the potential effect of atrophy in multivariate regression models analyzing bifrontal ROI FDG-SUVR.

Results
A No discontinuations or dose-limiting adverse effects occurred throughout treatment.Incidence of treatment-emergent AEs (Table 2) was similar between placebo (4 of 5; 80%), low-dose (7 of 11; 63.6%), and high-dose (7 of 11; 63.6%) cohorts (P > .99 for all comparisons).One patient randomized to 500 mg FRM-0334 experienced a serious AE (deep vein thrombosis with nonfatal pulmonary embolism), which was not felt to be related to the study drug in the opinion of the site investigator.
Patients randomized to FRM-0334 experienced higher incidences of cardiac, constitutional, dermatologic, infectious, psychiatric, and respiratory symptoms and a lower incidence of gastrointestinal symptoms relative to placebo-treated participants ( GRN variation carrier status, particularly symptomatic disease (but not prodromal carrier status alone), was associated with frontal FDG hypometabolism relative to controls (including left > right dorsal prefrontal, anterior cingulate, orbitofrontal, inferior frontal gyrus, and insular hypometabolism), with additional left-predominant hypometabolism noted in lateral parietal, lateral temporal, posterior cingulate, caudate, and thalamic regions (Figure 2; eTable 4 in Supplement 2).

Discussion
This report describes, to our knowledge, the first international multicenter randomized, placebocontrolled clinical trial in individuals with GRN haploinsufficiency and the first clinical exploration of HDAC inhibition in this cohort.Treatment with FRM-0334 was safe and well tolerated but was not associated with improvement in plasma PGRN concentration, CSF PGRN concentration, or exploratory pharmacodynamic measures.Moreover, the plasma PK profile of FRM-0334 did not change in proportion to the oral dose, which suggests inconsistent absorption or oral bioavailability   29 particularly in the liver, it is possible that GRN haploinsufficiency could give rise to a wide array of subtle phenotypic differences in a variety of cell types, including peripheral cells mediating drug absorption or metabolism.
Our FDG-PET analyses were consistent with previous clinical correlations 30 and anatomic descriptions of frontal hypometabolism 30,31 in GRN haploinsufficiency but reveal novel fluidbiomarker relationships to FDG-PET data.Cerebrospinal fluid NfL is interpreted to be a biomarker of axonal injury and neuronal degeneration 32 and correlates with disease severity 33,34 and aggressiveness in sporadic FTLD as well as in GRN haploinsufficiency. 35The observed relationship with NfL level may attest to the synaptic loss and neuronal metabolism documented on FDG-PET (via astrocyte-neuron metabolic coupling 36 ).Additionally, although GRN haploinsufficiency is not typified by tau deposition on autopsy, our findings support previous observations of elevated CSF t-tau in symptomatic disease. 37The relationship of FDG-PET to t-tau may reflect the lysosomal role of PGRN in supporting autophagy pathways, 38 which may impact tau clearance. 39uorodeoxyglucose hypometabolism was not associated with CSF p-tau 181 , therefore, comorbid Alzheimer pathology is unlikely to have mediated its relationship to t-tau.It is unclear how much of our FDG-PET fluid-biomarker correlations were mediated by brain atrophy alone, given our lack of complementary harmonized magnetic resonance images.Our sensitivity analyses, accounting for whole-brain parenchymal volume, did provide some evidence that FDG hypometabolism tracks clinical GRN deficiency severity even after the effect of volume is considered.

Limitations
Our study had several limitations affecting its interpretability.The inconsistent bioavailability of FRM-0334 made it difficult to reliably extrapolate the broader effect of consistent HDAC inhibition in PGRN deficiency or whether higher exposure levels of FRM-0334 may have affected plasma PGRN.The small sample size of patients randomized to placebo and the relatively short duration of this trial also limited our ability to detect modest deflections in biomarker trajectory.However, recent publicly presented open-label trial data do suggest that a similar duration and smaller sample size may be sufficient to demonstrate biomarker evidence of target engagement from alternative PGRN boosting strategies. 40,41With regard to our FDG-PET analyses, the heterogeneity of our data (acquired on differing scanners, with differing resolutions and protocols, in the absence of sitespecific, healthy controls) may have also limited the quality and resolution of our observations.Moreover, the short duration of our trial provided a relatively short interval to validly discern a true trajectory of change within our heterogeneous data set.Despite these limitations, our novel

Conclusions
In conclusion, this randomized clinical trial demonstrated that FRM-0334 did not have a PD effect on CSF or plasma PGRN concentration.However, this study provided useful data to inform future clinical trial methodology in similar cohorts.2][3][4][5][6] Such multicenter trials provide a unique opportunity to collect valuable observational data in rare familial forms of FTD.
characterized the peripheral pharmacokinetic (PK) profile and central PD effects (change in CSF PGRN) of FRM-0334 in GRN haploinsufficiency.Additionally, this trial's inclusion of a variety of exploratory PD end points allowed for novel analyses comparing cortical fluorodeoxyglucose positron emission tomography (FDG-PET), clinical measures of functional FTD severity, and candidate CSF biomarkers of neuronal injury, including neurofilament light chain (NfL), amyloid β 1-42 (Aβ 1-42 ), phosphorylated tau 181 (p-tau 181 ), and total tau (t-tau).We present data from the first JAMA Network Open | Neurology FRM-0334 and Progranulin Levels in Patients With Progranulin Gene Haploinsufficiency JAMA Network Open.2021;4(9):e2125584.doi:10.1001/jamanetworkopen.2021.25584(Reprinted) September 24, 2021 2/15 Downloaded from jamanetwork.comby guest on 03/23/2024 ) and time in determining patient outcome measures.The plasma FRM-0334 standard AUC from 0 to the maximum observed time was calculated using cubic splines.26Linear regression models were used to assess for baseline relationships between baseline PK values and age, sex, FTLD clinical severity (CDR plus NACC FTLD, sum of boxes), and follow-up PK values (day 1 vs day 7).Linear regression models (controlling for age and sex) were used to investigate baseline cross-sectional linear relationships between bifrontal FDG-SUVR, plasma PGRN, CSF biomarker concentrations, and clinical severity outcome measures.Linear regression models (controlling for age and sex) were also used to investigate linear relationships between magnitude of change in PGRN (plasma or CSF) and magnitude of change in FDG-SUVR, CSF biomarker concentrations, and clinical severity measures after treatment with FRM-0334.Univariate regression models were also used to compare baseline PK values by age and sex.The Shapiro-Wilks W test was used to confirm assumptions of normality (of dependent variables and residuals), and the Cameron and Trivedi decomposition of IM-test was used to confirm assumptions of residual homoscedasticity.

Figure 2 .A
Figure 2. Hypometabolism in GRN Variation Carriers Relative to Age-Matched Controls

JAMA
Network Open | Neurology FRM-0334 and Progranulin Levels in Patients With Progranulin Gene Haploinsufficiency observations of FDG hypometabolism support further exploration of FDG-PET as an imaging biomarker of FTLD biological severity in future GRN haploinsufficiency studies (ideally employing greater homogeneity and harmonization among scanners, acquisition protocols, and contemporaneous magnetic resonance imaging acquisition).
13, 2015, toApril 13, 2016, at 11clinical sites in the US, UK, France, Italy, the Netherlands, and Belgium (trial protocol in Supplement 1; eTable 1 in Supplement 2).Significant comorbid thyroid, hematologic, kidney, cardiovascular, inflammatory, or hepatic disease was exclusionary.The trial protocol specified that comorbid neurologic or psychiatric disease (if determined to be unrelated to GRN variations) was also exclusionary, though these criteria were not cited as the basis for exclusion of any participants recruited for screening (eFigure in Supplement 2).Ethics approval was obtained at each site from the local institutional review board and independent ethics committee.Written informed consent (from participants and legal proxies) was obtained at the start of recruitment, in compliance with the current revision of the Declaration of Helsinki, current International Conference on Harmonisation and Good Clinical Practice guidelines, the Consolidated Standards of Reporting Trials (CONSORT) reporting guideline, and local regulations.
-0334 and Progranulin Levels in Patients With Progranulin Gene Haploinsufficiency controls) of FDG-SUVR were assessed via analysis of covariance in SPM12, controlling for age and sex, in separate models including prodromal, symptomatic, and all gene variation carriers.Voxelwise associations between FDG-SUVR and 2 primary exploratory measures of FTD severity (CSF NfL and CDR plus NACC FTLD sum of boxes) were assessed in separate multiple regression models (with age and sex entered as covariates) in SPM12, including all variation carriers and excluding controls.
acquisition parameters is detailed in eTable 1 in Supplement 2. Digital Imaging and Communications in Medicine files from a total of 26 trial patients and 52 age-matched individuals without cognitive pathology from the Berkeley Aging Cohort Study were converted to Neuroimaging Informatics Technology Initiative format, warped to standard space with Statistical Parametric Mapping 12 (SPM12) via a PET-only pipeline (eMethods in Supplement 2) and normalized to pons to obtain parametric FDG-SUVR images.Given the heterogeneity of scanner and acquisition parameters (eTable 1 in Supplement 2), final FDG-PET SUVR images were downsampled to match the scan with the least spatial resolution using Analysis of Functional NeuroImages software, version 16.2.16(NIMH Scientific and Statistical Computing Core).Voxelwise group comparisons (gene variation carriers vs JAMA Network Open | Neurology FRMT-maps from voxelwise analyses were thresholded (primary threshold P < .001uncorrected for multiple comparisons, cluster-level familywise error [FWE]-corrected P < .05),converted to Pearson correlation coefficient r maps using the Computational Anatomy Toolbox, version 12 (Jena University and fluid biomarkers using linear regression models controlling for age and sex.The included ROIs were anterior, lateral, medial, and posterior orbital gyri, middle and superior frontal gyri, orbital and triangular part of the inferior frontal gyri, and the frontal pole.Change in bifrontal FDG-SUVR over time was also assessed via linear mixed-effects models as a follow-up sensitivity analysis, though the heterogeneity of PET sites and short duration of follow-up limited the potential interpretability of these analyses.The methods of FDG-PET collection and processing (including estimation of W-maps) are further detailed in the eMethods in Supplement 2.
Abbreviations: AE, adverse event; DVT, deep vein thrombosis; FTD, frontotemporal degeneration; PE, pulmonary embolism; SAE, severe adverse event.JAMA Network Open | NeurologyFRM-0334 and Progranulin Levels in Patients With Progranulin Gene Haploinsufficiency JAMA Network Open.2021;4(9):e2125584.doi:10.1001/jamanetworkopen.2021.25584(Reprinted) September 24, 2021 7/15 Downloaded from jamanetwork.comby guest on 03/23/2024 Downloaded from jamanetwork.combyguest on 03/23/2024 of the formulation used in this study.Although these results halted clinical development of the current formulation of FRM-0334 in GRN haploinsufficiency, this trial's international multicenter cohort enabled novel opportunities for additional biomarker analyses.Specifically, we found that bifrontal cortical FDG hypometabolism was associated with greater clinical disease severity and CSF measures of neurodegeneration including NfL and t-tau levels in patients with PGRN deficiency.The PD and PK profiles of oral FRM-0334 firmly suggest that further development of the tested formulation should not be pursued for GRN haploinsufficiency.Histone acetylation is primarily driven by FRM-0334 C max , and within this trial, only a minority of participants in both the low-and highdose cohorts experienced a C max above 1000 ng/mL (a magnitude equivalent to the minimum C max necessary to drive detectable histone acetylation in preclinical canine studies).Moreover, the observed variance in FRM-0334 C max did not predict change in PGRN (plasma and CSF) or other exploratory PD measures, further suggesting inadequate target engagement.Additionally, given the unreliable relationship between FRM-0334 dose level, plasma C max , and plasma AUC, it is unknown whether alternative regimens (including higher or more frequent doses) of the current formulation could reliably improve the bioavailability of FRM-0334 in future trials.Given the relationship between FRM-0334 exposure and age, future early phase trials of other orally available HDAC inhibitors may require design considerations that account for possible unique, age-related differences in drug oral bioavailability in GRN haploinsufficiency.Given the wide distribution of GRN expression,