Rationale and study-design of a randomized, placebo-controlled, double-blind phase 2b trial to Evaluate E cacy, Safety and Tolerability of an oral glutaminyl cyclase inhibitor Varoglutamstat (PQ912) in Study participants With MCI and Mild AD-VIVIAD

E.G.B. Vijverberg Brain Research Center T.M. Axelsen Sanos Clinic A.R. Bihlet NBCD K. Henriksen University of Copenhagen: Kobenhavns Universitet F. Weber Vivoryon Therapeutics K. Fuchs Vivoryon Therapeutics J.E. Harrison Amsterdam UMC Locatie VUMC: Amsterdam UMC Locatie VUmc K. Kühn-Wache Vivoryon Therapeutics P. Alexandersen SANOS clinic N.D. Prins Brain Research Center Philip Scheltens (  p.scheltens@amsterdamumc.nl ) Amsterdam Universitair Medische Centra https://orcid.org/0000-0002-1046-6408


Introduction
Alzheimer's disease (AD) affects approximately 47 million people worldwide and the prevalence will triple by 2050 (1). No disease modifying therapy for AD has, so far, been approved. The failure of previous trials to show the desired e cacy has led to questions regarding the validity of amyloid hypothesis and whether the mechanism of action of the therapies and/or the study designs have been appropriate. Due to the rising demand for effective treatment, the scienti c eld is highly invested in the development of new therapeutic targets associated with amyloid beta aggregation in the brain as well as other possible disease modifying targets (2).
pGluAB42 is generated from full-length AB by various peptidases (eg. neprilysin (NEP) (12) and insulin degrading enzyme (IDE) (13) as part of the normal AB-degradation pathway and are then further modi ed by the enzyme glutaminyl cyclase (QC) (14). In human autopsy studies on AD-patients, pGluAB42 was found to constitute 10-50% of the total amyloid burden, suggesting a signi cant role in the development of AD (15)(16). Consequently, making the processes involved with pGluAB42 an interesting therapeutic target, highlighted by the recent encouraging news regarding donanemab, a monoclonal antibody also targeting pGluAB42 (17).
The small molecule QC-inhibitor varoglutamstat inhibits the activity of QC and reduces the level of pGluAB42. This was established in a large phase 1 program evaluating up to 14 days of treatment in healthy young and aging individuals being treated with up to 1800mg of varoglutamstat twice daily (BID). The study found varoglutamstat to be well tolerated with few and mild adverse events (AE), and demonstrated clear pharmacodynamic effects in terms of QC-inhibition in plasma and CSF (18).
In a phase 2a, randomized, double-blind, placebo-controlled, proof of concept trial in biomarker con rmed AD-patients (n = 120), study participants were treated with varoglutamstat in doses of 800 mg for 12 weeks to further evaluate safety and effects on biomarkers (19). Varoglutamstat showed an acceptable safety and tolerability pro le in lower doses and with slower titration. The varoglutamstat-treatment group were also found to have a signi cant improvement of working memory, reduction of synaptotoxicity measured by less theta-wave activity on EEG and reduced neurogranin-levels as well as improvements on various other experimental biomarkers (19). Post-hoc results further supported the concept of enhance synaptoplasticity with the treatment of varoglutamstat in a network analysis.
The trials with varoglutamstat indicated a bene cial safety and tolerability-pro e resulting in favourable bene t-risk ratio, biological effect on QC inhibition, reduced synaptic toxicity, data suggestive of a clinical effect and provided the rationale to design a state-of-the-art phase 2b trial with endpoints in cognitive function, biomarkers, and long-term safety and tolerability. In the following we describe the study design of the VIVIAD trial, a study to evaluate safety and tolerability of different doses and e cacy of varoglutamstat in study participants with MCI and Mild AD.

Study setting and design
This multinational, double-blind, randomized, placebo-controlled study is evaluating the disease modifying effects of PQ912 and serves as a dose nding study on the drug in patients suffering from MCI or mild dementia due to AD. The study aims to enroll 250 patients and is currently enrolling study participants in all participating centers.
The trial consists of an initial dose-nding phase. All participants enrolled will be randomized in a 1:1:1 fashion between 600mg BID varoglutamstat, 300mg BID varoglutamstat or placebo BID. When the 90th subject has completed the evaluation at week 24, an interim analysis will be conducted in order to establish the lowest effective dose of varoglutamstat. Hereafter, all participants receiving varoglutamstat will be changed to the highest well-tolerated dose and all further randomization will be done 1:1 between the chosen dose of varoglutamstat and placebo. Study participants recruited early into the study will be kept on treatment for up to 96 weeks or until the last subject recruited will have completed visit (V) 8 (48 weeks of treatment), whichever comes rst, (see Fig. 1 for graphic overview of the study design). The trial is registered at clinicaltrials.gov: NCT04498650

Study population
The study's inclusion criteria are study participants, male or female, from ≥ 50 to ≤ 80 years with a diagnosis of MCI or mild dementia due to AD as per the 2018 NIAA research framework (20) (30), with a Mini Mental State Examination (MMSE) score of ≥ 20, a Wechsler Adult Intelligence Scale (WAIS) IV Coding Test score (21) (31) ≤ 0.5 standard deviations below the reference score adjusted for age. Our purpose in selecting the Coding Test requirement was to include only study participants likely to to have a rescuable cognitive de cit. Also, the participants must be able to have a study partner accompany them at study visits, as well as be in a stable state regarding both AD and potential AD-medication.
Exclusion criteria include exclusion of conditions which may affect cognition, including vascular dementia or other cerebral lesions as evaluated by magnetic resonance imaging. No major psychiatric conditions are allowed, there should be no known presence of insu ciently treated vitamin-B, folatede ciency or hypothyroidism. There should be no signs of severe hepatic or renal failure, neither should the participants have suffered stroke or have evidence of other forms of dementia including atypical presentations of AD (e.g. posterior cortical atrophy, frontal or language types of AD) or previous seizures.
Patients having had cancer (besides from non-metastatic basal cell cancer or non-metastatic squamous cell cancer) within 2 years, myocardial infarction within the past 6 months or signs of addiction disorder, are also excluded (see table 1).

Study procedures
As a means of evaluating activities of daily living in this cognitively impaired patient population, the Amsterdam Instrumental of Activities of Daily Living questionnaire (A-IADL-Q) (22) (32) will be employed. This instrument is completed by the study partner, hence study participants are required to have a study partner able to accompany them on visits to clinic. The screening for the trial consists of several elements in order to ensure adherence to the inclusion and exclusion criteria (see table 1). A medical history, physical and neurological examination, vital signs, ECG, and cognitive measures including the WAIS-IV Coding test, to test executive function and working memory, the MMSE and a Neuropsychological Test Battery (NTB) (23-25)(33-35) composed of CogState tests of episodic verbal memory (the International Shopping List Test), episodic visual memory (the One Card Learning test), working memory (the One Back test) and two measures of attention ('Detection' and 'Identi cation'). A composite score of perfomance on these latter three tests is the primary e cacy measure for this study.
To assess AD-speci c speech changes the Winterlight speech assessment (WLA) will be employed regularly to assess any change from baseline in the study participants (36). Also, blood will be drawn for hematology, blood biochemistry and assessment of blood biomarkers. Also, a lumbar puncture will be performed to assess inclusion criteria in the form of AB42 and p-Tau-levels, and have an extra matrix to assess longitudinal pharmacodynamic effects of varoglutamstat by means of investigational biomarkers. A cerebral MRI will be performed to rule out relevant cerebral vascular disease, or other pathological changes.
Upon passing screening, the rst 90 participants will be randomized 1:1:1 by a computerized allocation system to receive either PQ912 300mg BID, 600mg BID or placebo BID following a 12-week escalation period. Escalation occurs in up to ve steps; weeks 1-2: 50 mg/placebo once daily, weeks 3-4: 50 mg/placebo BID, weeks 5-8: 150 mg/placebo BID, weeks 9-12: 300 mg/placebo BID, weeks 13-24: 300 mg/placebo or 600 mg/placebo BID, in order to minimize possible side-effects. At the baseline visit, study participants will be assessed again on the NTB, WAIS-IV coding test and the WLA. A full physical examination will also be performed, as well as en EEG and blood draws for hematology, biochemistry and apolipoprotein E (ApoE), Human Leukocyte Antigen (HLA) and CYP2c19 genotyping.
As illustrated in Fig. 1, the participants will be seen in the clinics 4 weeks after baseline and assessed for AE's, drug accountability and vital signs. In addition, blood will be sampled for hematology and biochemistry and a physical exam will be conducted. At week 12, the above procedures will be repeated as well as the NTB, WLA and WAIS-IV Coding test. These procedures will be repeated every 12 weeks.
The participants attend clinic visits once every 12 weeks until study termination, where several key measurements are made at timepoints, including physical and neurological examination, WAIS-IV Coding test, WLA, NTB, Amsterdam IADL-Q. Every 24 weeks blood will be analyzed for biomarkers. Blood levels of PQ912 will be measured at week 12, 24 and 48 as well as at EOT, for study participants enrolled early in the trial. In addition, an EEG will be performed at week 24 and 48 and an additional cerebral MRI will be performed at EOT. CSF will be collected for biomarker measurement at week 48.
Should the patients experience signi cant intolerable side effects (starting from a maximal dose of 300 mg BID/placebo or above), the investigators may decrease drug dosage by 50% for a period or temporarily suspend treatment with IMP. Should treatment-naïve study participants progress signi cantly in their symptoms during the study, initiation of standard AD medication is allowed.

Endpoints
The safety analysis will be based on signi cant changes on i) physical and neurological examinations, ii) signi cant changes on blood samples, iii) amyloid-related imaging abnormalities (ARIA) and iv) reported adverse events.
The primary e cacy analysis consists of the pooled Z-score of the CogState 'Detection', 'One Back' and 'Identi cation' tests (see 'Statistical Evaluation' for method of evaluation).
In addition to the primary analysis of e cacy and safety evaluation, several secondary and exploratory endpoints have been de ned for this study. Secondary endpoints have been de ned as linear change in overall cognition based on the complete NTB and in the CogState battery by itself, using the same statistical model as the primary e cacy analysis. At week 48, changes in global relative theta power (4)(5)(6)(7)(8) in the EEG-data as well as changes from baseline score of A-IADL-Q will also be analyzed.
As an exploratory endpoint neuronal activity will be evaluated in the same manner as described by Briels et al. 2020 (26) (36) (19) in regards to global relative power in the delta (0.5-4 Hz), alpha (8)(9)(10)(11)(12)(13) and beta (13-30Hz) frequency bands, looking into global posterior dominant peak frequency, Amplitude Envelope Correlation (AEC) in the 4-13 Hz band and functional network topology measures such as centrality, modularity, minimum spanning tree in an attempt to acquire new EEG AD-speci c biomarkers in this modality.
Besides collection at screening and baseline, blood for biomarker measurement will be collected at week 24, 48 and the EOT-visit, whereas CSF will be collected at screening and at EOT. There will be a broad assessment of biomarkers with potential use for quantifying neuroin ammation, synaptic toxicity and neurodegeneration. The panel of biomarkers assessed will include levels of: YKL-40 (in ammatory marker), Neurogranin (synaptic marker), Beta secretase 1 protein [BACE-1], Tau and pTau, Protein fragments of Tau, GFAP and extracellular matrix (ECM) molecules (neurocan, brevican and Tenascin-R). In addition to the earlier mentioned well-established biomarkers, exploratory biomarkers in both plasma and CSF will be investigated e.g. Neuro lament light chain (NFL); protein fragments of Tau, GFAP and extracellular matrix (ECM) molecules (neurocan, brevican and Tenascin-R); pGlu-peptide substrates of QC (e.g. pGlu-and total C-C motif chemokine ligand 2 [CCL2], Orexin A) and Aβ peptides (including full length and truncated A peptides) will be assessed.

Statistical evaluation
The main statistical evaluation will be performed in the Full Analysis Set population, such that all participants who are randomized and received at least one dose of study medication will be included in the main statistical analysis.
An interim analysis is planned to be performed shortly following the 90th participant reach 24 weeks post-baseline. At this point a data safety monitoring board (DSMB) consisting of experts within relevant medical and scienti c elds will conduct an unblinded safety analysis, in order to decide on study continuation and secondly which dose of investigational medicinal product (IMP) should be used for the remainder of the study.
As the study is planned to have differences in participation duration and dosage, a linear model of the primary e cacy parameter will be employed to assess effectiveness. The primary e cacy analysis will consist of the pooled Z-score based on the cognitive scores from the CogState 'One Back', 'Identi cation' and 'Detection' tests (see Fig. 2). The two different dose groups will be pooled and the slope coe cient of the Z-score over time will be compared to that of the placebo group. This approach enables incorporation of all data on participating individuals, irrespective of their follow-up time, thus ensuring a maximum amount of information obtained from study participants over the course of the trial.
Based on extrapolation of the results in the SAPHIR study, evaluation of other empirical data, and expert opinion, it was determined that at 48 weeks, an effect size of approximately 0.35 is a reasonable expectation. The effect size, Cohen's D, of 0.35 corresponds to an abolishment of the decline in the combined Z-score for cognition in the PQ912 treated study participants. Based on the assumptions above, and an allocation ratio of 11:14 between active and placebo, by comparison via a two-sample Ttest, a total sample size of around 250 should provide a power close to or exceeding 80%.
The sample size is not adjusted for the rate of drop-out as it is anticipated that all participating individuals will have at least two measurements of the combined Z-score, irrespective of their follow-up time, and therefore contribute to the analysis of the primary e cacy endpoint.

Discussion
The VIVIAD trial is a phase 2b dose-nding study further evaluating the safety and e cacy of varoglutamstat, as a potential disease-modifying treatment in AD. Varoglutamstat inhibits the QC activity and reduces the level of pGluAB42, which is expected to alleviate the acute and chronic neurotoxic effects of pGluAB42. In addition to its main purpose, the trial will explore potential associations between novel and established biomarkers and their individual and composite relation to disease characteristics.
Dose nding for the optimal bene t-risk ratio is an important part of the VIVIAD trial. The previous 2a study established 800 mg varoglutamstat BID to be the maximum tolerable dose. It is expected that a slower dose titration may improve safety and tolerability. The rationale behind a gradual titration is that most AEs from the SAPHIR trial was deemed of immunological origin (19), a titration period may allow a gradual habituation of the immune system to the drug. So, the current trial has a gradual dose escalation period, followed with an interim analysis in order to establish the most suitable dose of varoglutamstat while limiting the risk of potential negative consequences of QC-inhibition in the periphery for rest of the trial.
One of the primary e cacy endpoints is the evaluation of varoglutamstat on working memory and attention, and secondarily on cognition and activities of daily living, using the Cogstate Neuropsychological Test Battery and the A-IADL-Q, both of which have been validated for use as serial measurements (22)(23)(24). Exploratory measure include the WAIS-IV Coding test (21,28), part of the WAIS-IV intelligence test and which has also been validated for exploring executive function and working memory as a stand-alone test and which is recognised by the EMA as a measure of 'timed executive function'. The WAIS-IV Coding test serves in this study as an inclusion criteria with a speci c cut-off in the screening process, selecting only participants with a certain degree of cognitive impairment. Using the WAIS-IV Coding test score as inclusion criterion and an exploratory endpoint (longitudinal evaluation) is a novel approach in AD research.
Results from SAPHIR trial indicated bene cial effects of varoglutamstat on synaptic toxicity measured by different biomarkers. Therefore, the e cacy of varoglutamstat on brain activity will be assessed as a secondary objective. Based on the positive ndings in the SAPHIR study (19,26), resting-state brain activity will be evaluated by using EEG which has been found to be associated with synaptic activity enabling a macroscale assessment of neuronal circuit integrity (26,29). Besides that, EEG is both costeffective and widely available, making it very suitable for trials in AD due to the method's ability to detect early changes in oscillatory activity of the cortex (30). Another argument for including EEG-changes as an endpoint is that the functional neurophysiological changes of a treatment can happen within days to weeks, as shown in the classical cholinesterase studies. In the VIVIAD trial EEG data will be assessed for changes in theta-wave activity (theta power), and other sensitive oscillatory activity markers (31,32). Incorporation of this modality in the current study may further validate EEG-changes as endpoints in clinical trials Another outcome in the VIVIAD trial is an automated speech recognition method-based measure of disease affecting language (including AD), the WLA. Most individuals living with AD exhibit some degree of speech impairment (33). This worsens over time (34,35), making speech and language assessments relevant for appraisal of disease severity and activity. However, changes over time of each individual's language and in early stages of AD can be subtle and di cult to detect. Previous speech evaluation has relied on cumbersome tests unsuitable for widespread clinical use (37). By employing an algorithm evaluating factors such as length of sentences, richness of vocabulary, repetitiveness and spectrum of acoustics it is possible to extract accurate data in speech passages as brief as 150 words. Digital speech biomarkers have, so far, only been used in an experimental diagnostic setting (25,38,39). The longitudinal setting of this trial allows for analysis of the WLA as an exploratory endpoint, as well as a tertiary measurement of e cacy of varoglutamstat.
Changes in biochemical biomarkers are important outcome measurements to detect biological/pharmacodynamic effect in phase 2b AD trials. In the SAPHIR trial (19), YKL-40, a neuroin ammation marker in AD (40), was shown to decrease approximately 5% from baseline levels within 12 weeks when using varoglutamstat. Additionally, the synaptic biomarker neurogranin was reduced by approximately 4%. Both of these biological signals were meaningful when considered in relation of increase in general AD patients. In the current trial, biochemcial biomarkers serve several purposes: They act as an inclusion criterion, minimizing the risk of including non-AD-patients into the trial and also function as outcome measures on week 48/and or end of treatment. Serum and CSF will be assessed for QC activity, in ammatory-, synaptic-, neurodegeneration and AD biomarkers.
The statistical approach to the primary and some of the secondary endpoints allows for interpretation of the effect of varoglutamstat in a larger sample of individuals, as it does not require all study participants to be evaluated at a speci c time point. This is highly bene cial, as it allows inclusion of the data from the dose-nding phase into the primary endpoint analysis. Based on the phase 1 trial the difference between 600mg BID and 300mg BID in CSF QC-inhibition will be approximately 10% (18), whether this difference is clinically signi cant is unknown, but as AB is a poor substrate of glutaminyl cyclase, even minor reductions in QC-activity are expected to inhibit formation of neurotoxic pyro-GluAB. In addition, the inclusion of the dose nding phase in the data-analysis allows a longer follow-up of the treated individuals and provide information on the long-term treatment effect regarding both safety and e cacy data without affecting the time to result.
In conclusion, the present VIVIAD trial is a state-of-the-art design in the eld of AD and is a combined dose-nding, proof-of-concept e cacyand safety-analysis of varoglutamstat with a strong exploration platform for development of new important biomarker-based diagnostics in one and the same trial.