NULL HYPOTHESIS OPEN ACCESS CLASS OF EVIDENCE

INTREPADA randomized trial of naproxen to slow progress of presymptomaticAlzheimer disease

Pierre-Franccedilois Meyer MSc Jennifer Tremblay-Mercier MSc Jeannie Leoutsakos PhD Cecile Madjar MSc

Marie-Elyse Lafaille-Maignan PhD Melissa Savard MSc Pedro Rosa-Neto MD PhD Judes Poirier PhDdagger

Pierre Etienne MDdagger and John Breitner MD MPHdagger for the PREVENT-AD research group

Neurologyreg 201992e1-e11 doi101212WNL0000000000007232

Correspondence

Dr Breitner

johnbreitnermcgillca

AbstractObjectiveTo evaluate the safety and efficacy of low-dose naproxen for prevention of progression inpresymptomatic Alzheimer disease (AD) among cognitively intact persons at risk

MethodsInvestigation of Naproxen Treatment Effects in Pre-symptomatic Alzheimerrsquos Disease(INTREPAD) a 2-year double-masked pharmaco-prevention trial enrolled 195 AD familyhistoryndashpositive elderly (mean age 63 years) participants screened carefully to exclude cog-nitive disorder (NCT-02702817) These were randomized 11 to naproxen sodium 220 mgtwice daily or placebo Multimodal imaging neurosensory cognitive and (in 50) CSFbiomarker evaluations were performed at baseline 3 12 and 24 months A modified intent-to-treat analysis considered 160 participants who remained on-treatment through their firstfollow-up examination The primary outcome was rate of change in a multimodal compositepresymptomatic Alzheimer Progression Score (APS)

ResultsNaproxen-treated individuals showed a clear excess of adverse events Among treatment groupscombined the APS increased by 0102 pointsyear (SE 0014 p lt 10minus12) but rate of change showedlittle difference by treatment assignment (0019 pointsyear) The treatment-related rate ratio of 116(95 confidence interval 064ndash196) suggested that naproxen does not reduce the rate of APSprogression by more than 36 Secondary analyses revealed no notable treatment effects on in-dividual CSF cognitive or neurosensory biomarker indicators of progressive presymptomatic AD

ConclusionsIn cognitively intact individuals at risk sustained treatment with naproxen sodium 220mg twicedaily increases frequency of adverse health effects but does not reduce apparent progression ofpresymptomatic AD

Classification of evidenceThis study provides Class I evidence that for people who are cognitively intact low-dose nap-roxen does not significantly reduce progression of a composite indicator of presymptomatic AD

RELATED ARTICLE

EditorialNaproxen forpresymptomatic Alzheimerdisease Is this the end orshall we try again

Page 829

MORE ONLINE

Class of EvidenceCriteria for ratingtherapeutic and diagnosticstudies

NPuborgcoe

These authors contributed equally to this work

daggerThese authors contributed equally to the direction and supervision of this work

From the McGill Centre for Integrative Neuroscience Montreal Neurological Institute (CM) and McGill University Research Centre for Studies in Aging (MS PR-N) McGillUniversity (P-FM M-EL-M PR-N JP PE JB) StoP-AD Centre (P-FM JT-M M-EL-M PR-N JP PE JB) Douglas Mental Health University Institute Research CentreMontreal Canada and John Hopkins University (JL) Baltimore MD

Go to NeurologyorgN for full disclosures Funding information and disclosures deemed relevant by the authors if any are provided at the end of the article

Data used in preparation of this article were obtained from the program of 49 PRe-symptomatic EValuation of Novel or Experimental Treatments for 50 Alzheimerrsquos Disease (PREVENT-AD) at theCentre for Studies onPreventionof 51AlzheimerrsquosDisease (StoP-AD) DouglasMentalHealthUniversity Institute 52ResearchCenter Archive 5340 (June27 2017) The authorsand coinvestigators thank all the members of the PREVENT-AD Research Group who made contributions to the study

The Article Processing Charge was funded by the Fonds de Recherche du Quebec-Sante

This is an open access article distributed under the terms of the Creative Commons Attribution-NonCommercial-NoDerivatives License 40 (CC BY-NC-ND) which permits downloadingand sharing the work provided it is properly cited The work cannot be changed in any way or used commercially without permission from the journal

Copyright copy 2019 The Author(s) Published by Wolters Kluwer Health Inc on behalf of the American Academy of Neurology e1

Published Ahead of Print on April 5 2019 as 101212WNL0000000000007232

Alzheimer disease (AD) includes a decades-long period ofpresymptomatic biochemical imaging neurosensory andsubtle cognitive changes12 Because cognitive changes emergegradually AD prevention trials using cognitive endpointstypically follow thousands of individuals for several yearsImproved efficiency may result from use of multimodalcomposite indicators of early AD pathogenesis as well assubtle cognitive decline One such indicator the AlzheimerProgression Score (APS)3 assesses presymptomatic ADprogression using Item Response Theory modeling Pre-dictive validity of the APS method has been demonstratedrecently3

Among potential preventive interventions in presymptomaticAD nonsteroidal anti-inflammatory drugs (NSAIDs) haveretained interest Numerous observational studies have shownreduced incidence of AD in users of these drugs at least inrelatively young elderly4 However NSAID prevention trialshave failed to show benefit and instead have caused harm toolder individuals5 or those with imminent symptoms67

These findings suggest that NSAID prevention trials mustscreen (preferably younger-old) participants carefully forcognitive or other prodromal AD symptoms8 Such principleswere applied when designing Investigation of NaproxenTreatment Effects in Pre-symptomatic Alzheimerrsquos Disease(INTREPAD NCT-02702817) a 2-year double-maskedrandomized trial of oral naproxen sodium 220 mg twicedaily vs placebo for safety and efficacy against progression ofAD-related change

MethodsPrimary research questionINTREPAD was designed to test the safety and efficacy oflow-dose naproxen in reducing the rate of change of a com-posite indicator of presymptomatic AD To that end Class Ievidence is provided here An important additional objectiveof the work was to assess the practicality and utility of a moreefficient approach to AD prevention trials using novel meth-ods of recruitment data collection and analysis A notablefeature was the use of a composite primary efficacy outcomederived from a parallel observational program of cognitivestructural and functional MRI and neurosensory measuresalong with CSF biomarkers

Standard protocol approvals registrationsand patient consentsINTREPAD (NCT-02702817) was approved by the in-stitutional review board of McGill University Recruitmentoccurred between November 2011 and March 2015 Datagathering ended March 28 2017 All participants providedwritten informed consent for each trial procedure Data werecollected at the Douglas Mental Health University Institutean affiliate of McGill University (Montreal) All researchprocedures complied with the ethical principles of the Dec-laration of Helsinki A Data Monitoring Committee reviewedall safety and efficacy data prepared by a contract (unmasked)statistician on October 20 2016 and upon completion (June26 2017)

Overview of participants and trial regimenWe recruited 462 healthy older individuals with a parentalor multiple-sibling history of AD for participation in Pre-symptomatic Evaluation of Novel or Experimental Treatmentsfor Alzheimerrsquos Disease (PREVENT-AD) an observationalcohort study of healthy persons aged 55+ without evidenceof cognitive deficit9 Among these 195 eligible volunteerswere randomized in INTREPAD to receive either low-dosenaproxen sodium or placebo (figure 1) The primary effi-cacy analysis considered a modified intent-to-treat (mITT)group of 160 persons who remained on assigned treatmentsuntil their first follow-up evaluation 90 days after baselineIn all 166 (85) completed their participation per proto-col (154 mITT and 12 others in the intent-to-treat (ITT)group with 2 years of follow-up and biomarker assessment)Participants who completed the trial on study treatmentsnumbered 124

Initial integrity of participantsrsquo cognitive abilities was evalu-ated by telephone interview followed by in-person assess-ment using the Montreal Cognitive Assessment (MoCA) andthe Clinical Dementia Rating scale1011 In an early protocolchange we further verified intact cognitive status after base-line testing using the trialrsquos principal cognitive assessmentmeasure the Repeatable Battery for Assessment of Neuro-psychological Status (RBANS)12 As a consequence 14 enroll-ees (3 originally assigned to naproxen and 11 to placebo) wereconsidered unsuitable for further participation because ofnotable cognitive deficits that had escaped detection at base-line suggesting early mild cognitive impairment (MCI) Final

GlossaryAD = Alzheimer disease AE = adverse event APS = Alzheimer Progression Score CI = confidence interval INTREPAD =Investigation of Naproxen Treatment Effects in Pre-symptomatic Alzheimerrsquos Disease ITT = intent-to-treat LP = lumbarpuncture MCI = mild cognitive impairment mITT = modified intent-to-treat MoCA = Montreal Cognitive AssessmentNSAID = nonsteroidal anti-inflammatory drug p-tau = phosphorylated tau PREVENT-AD = Pre-symptomatic Evaluationof Novel or Experimental Treatments for Alzheimerrsquos Disease RBANS = Repeatable Battery for Assessment ofNeuropsychological Status SAE = serious adverse event t-tau = total tau UPSIT = University of Pennsylvania SmellIdentification Test

e2 Neurology | Volume 92 Number 18 | April 30 2019 NeurologyorgN

determination of cognitive eligibility relied for some on fullneuropsychological evaluation

Eligibility criteriaOther eligibility criteria included (1) at least 1 parent or 2siblings with AD (2) age ge 60 years or ge55 years if within 15years of youngest-affected relativersquos onset (3) health andsocial stability sufficient to enable participation for 5 years oflongitudinal study and (4) no contraindications to sustainedtreatment with naproxen sodium Family history was ascer-tained from an expertrsquos diagnosis of AD or when necessaryvia a brief structured questionnaire (e-Methods doiorg105061dryadr58d342) Exclusion criteria included regular use(more than 4 doses per week) of corticosteroids NSAIDsother anti-inflammatoryimmunosuppressant agents or as-pirin A complete list of inclusionexclusion criteria is avail-able in the e-Methods

Specification of primary efficacy outcome andinitial power analysisThe original INTREPAD analysis plan called for a primaryefficacy determination based on a composite of data that wasnot then specified This composite was under development indata from the parallel (nontrial) parent PREVENT-AD Co-hort with near-identical characteristics9 Initial power analy-ses considered a subset of data from the public use ADNIdatabase We simulated an attempt to demonstrate a true 25reduction in the slope of the CSF concentration of total tau

(t-tau) over 1 year specifying 85 statistical power Thisanalysis suggested that the needed power could be providedby a sample of 228 aged cognitively normal participantsassigned 11 to active drug vs placeboWe assumed that greaterpower would be provided by 2 (rather than 1) years of ob-servation and thus specified a target enrollment of 200 withan assumption that 80 would remain in an mITT primaryanalysis pool of persons who remained on study treatmentsthrough their first follow-up examination (but note below thatresults belied these assumptions)

When the primary efficacy outcome (APS)3 was fully speci-fied we performed a similar simulation in the parallel nontrialPREVENT-AD Cohort This simulation now relied on theCohortrsquos observed slope random intercept variance and errorvariance using a longitudinal random effects model It sug-gested that 160 participants would afford 85 power to detectonly a 50 difference in slope between 2 study arms or 68power to detect a 40 difference Because the PREVENT-ADdata did not include CSF biomarker measures however weexpected that the trialrsquos CSF assay results would providea substantial increase in power as had been the case in theobservational BIOCARD study13 (see Discussion)

Randomization masking and provision ofstudy drugUsing randomizationcom participants were randomized 11in 34 permuted blocks of 6 to identical-appearing tablets of

Figure 1 Consort flow

Fragile cognitive status indicates indi-viduals with cognitive deficits sugges-tive of earlymild cognitive impairmentGI = gastrointestinal mITT = modifiedintent-to-treat SAE = serious adverseevent

NeurologyorgN Neurology | Volume 92 Number 18 | April 30 2019 e3

naproxen sodium 220 mg or placebo both generouslydonated by Pharmascience (Montreal Canada) for admin-istration twice daily with meals The Douglas Hospital phar-macy stored and prepared study drug in sealed dosage packetsfor each participant visit Participants the principal in-vestigator study staff and all clinicians responsible forassessments or marker measures remained masked to treat-ment assignment until safety and efficacy analyses werecomplete A protocol for interim unmasking was available butnot needed Only an external statistician (Daniel Morissette)had advance access to treatment assignment

Assessment methodsFigure 2 shows the timeline for data gathering Completeevaluations were performed at baseline and after 3 12 and 24months of treatment

SafetyFollow-up interviews (on-site or by telephone) for adverseevents (AEs) were administered ad hoc or at aminimum every3 months using structured medical history and review-of-system questionnaires On-site safety evaluation includedroutine laboratory studies ECG and a brief physical andneurologic examination Potentially important incidental MRIand other newly discovered health risks were referred for expertreview Research nurses rated AEs using the Cancer TherapyEvaluation Program Common Terminology Criteria version30 AEs were graded as mild moderate or severe after phy-sician review and each AE was also assigned a Preferred Termand a System Organ Class using the Medical Dictionary forRegulatory Activities classification system version 191 SeriousAEs (SAEs) that were life-threatening or required hospitaliza-tion were reported in real time to the McGill Research EthicsBoard Relationship of AEs and SAEs to treatment (assumingassignment to naproxen) was assessed by a study physicianElective surgeries were not considered SAEs

ComplianceParticipants were asked to bring unused supplies of drug totrial visits Research nurses evaluated compliance at each visitinquiring when indicated into reasons for missed doses

Cognitive and neurosensory performanceAt baseline 3 12 and 24 months neuropsychological per-formance was measured using the RBANS which evaluates 5cognitive domains (immediate memory delayed memoryattention language and visuospatial abilities) and a totalsummary score The RBANS is available in 4 equivalent ver-sions (for longitudinal assessment) in both French and En-glish Version A was administered at baseline and alternateforms were used in random order at follow-up visits Wedeveloped correction factors to improve version equivalenceand scored results without correction for age (often usedclinically see e-Methods doiorg105061dryadr58d342)

At each visitrsquos fMRI session participants were also given al-ternate versions of an episodic memory task (encoding andretrieval of objects)14 They were asked to identify whetheritems had been observed during the encoding period or werenew at the retrieval session (details in e-Methods doiorg105061dryadr58d342) A correct response required a ldquohitrdquo orcorrect rejection Odor identification a neurosensory facultywas also tested using the 40-item ldquoscratch and sniffrdquo Univer-sity of Pennsylvania Smell Identification Test (UPSIT)1516

The latter comprising 40 items administered in 4 randomlyordered booklets was available in both French and English

Neuroimaging markersBrain structural and functional MRI were performed atbaseline and 3- 12- and 24-month visits on a Siemens TIMTrio 3T MRI system (Siemens Medical Solutions ErlangenGermany) using the Siemens standard 12-channel head coil(figure 2 and e-Methods doiorg105061dryadr58d342)

Figure 2 Trial timeline

Full assessment (on site) baseline (BL) 3 months 12 months 24 months Safety follow-up (on site) 1 month 6 months 18 months Safety follow-up(telephone) 9months 15months 21months T1-weighted (EN BL 3months 12months 24months) fluid-attenuated inversion recovery (EN 24months)diffusion-weighted imaging (EN 24 months) arterial spin labeling (BL 3 months 12 months 24 months) resting-state fMRI (BL 3 months 12 months24months) gradient echo quantitative T2 task (BL 3months 12months 24months) task fMRI (BL 3months 12months 24months) EN = enrollment LP =lumbar puncture RBANS = repeatable battery for assessment of neuropsychological status UPSIT = University of Pennsylvania Smell Identification Test

e4 Neurology | Volume 92 Number 18 | April 30 2019 NeurologyorgN

Using conventional pipelines averages of gray matter densitywere calculated for 78 regions based on T1-weighted imagesusing the SPM12 toolbox to define probabilistic gray mattermaps Cortical thickness was estimated from T1-weightedimages using version 112 of the CIVET pipeline17 Brainvolumes were computed from the same images using a volu-metric pipeline18 Regional cerebral blood flow was evaluatedusing quantitative pipelines from single-echo pseudo-continuous arterial spin labeling acquisitions19

CSF biomarkersA subset of 93 participants in the mITT pool volunteered forup to 4 serial lumbar punctures (LPs) over the 2-year trialinterval following their clinical and imaging evaluations Afteran overnight fast LPs were performed by PR-N using anintroducer and the Sprotte 24-gauge atraumatic needleSamples of 20ndash30 mL were withdrawn by syringe and ali-quoted (500 μL) into propylene cryotubes for storage atminus80degC We followed procedures specified by the BIOMARK-APD consortium of the EU Joint Program in Neurodegen-erative Disease to measure CSF concentrations of the ADbiomarkers Aβ1-42 t-tau and phosphorylated tau (p-tau) withthe Innotest ELISA assay kit (Fujirebio Ghent Belgium)

APOE genotypeAPOE genotype was determined using RT-PCR amplifiedDNA and the PyroMark Q96 pyrosequencer (Qiagen Tor-onto Canada) as described previously16

Primary efficacy outcome The composite APSThe primary efficacy outcomewas annual rate of change in theAPS using marker weights estimated beforehand in the non-trial PREVENT-AD Cohort The APS is a composite thatincorporates multimodal imaging neurosensory cognitiveand CSFmarkers based on an assumption that change in eachof these arises from a single underlying latent process (ADpathogenesis) Its scores are scaled as a standard normaldistribution with higher scores denoting increasing severityConstituent measures are summarized in table 1 At eachmeasurement a uniform scheme of weightings for individualmarkers yielded a composite summary score All available datawere used to estimate individual scores and missing data wereaccommodated in a process that essentially averaged overmissing values To verify robustness to missing data Grosset al20 had used iterative leave-one-out analyses in a similaroutcome measure comprising 6 markers Leoutsakos et al3

also examined effects of missing CSF data on the APS notingsimilarly satisfactory findings The APS approach had beenvalidated using data from the BIOCARD study13 before be-ing incorporated into INTREPAD efficacy analyses In BIO-CARD the APS approach showed substantial abilities topredict subsequent conversion to MCI or AD dementia3

Analyses there also showed temporal measurement in-variance Before applying the PREVENT-AD cohort-derivedmarker weights to the analysis of INTREPAD we demon-strated that they provided excellent performance in the trialsample3 However the trial data (table 1) included 2 variables

(CSF t-tau and Aβ1-42 levels) that were not available fromnontrial participants We incorporated these measures onlyafter verifying that doing so did not materially alter theweights for the remaining variables Full specification of var-iables and APS parameters preceded unmasking of treatmentassignment

Statistical analysisWe followed a statistical analysis plan finalized on June 82017 by JL and the PREVENT-AD Research Group Toassure consideration to potentially important AEs occurringduring the first 3 months of the trial safety analyses werebased on the full intention-to-treat population (n = 195)These analyses were based on summary listings of AEs usinga χ2 test for pairwise comparisons The baseline characteristicsof the treatment groups were compared pro forma usingFisher exact (for sex) χ2 (for number ofAPOE e4 alleles) and2-sample t tests (for age education parental age at AD onsetMoCA score and APS)

The primary efficacy analysis was based on the mITT sampleof participants who had remained on treatment through atleast one follow-up (3-month) assessment (n = 160) Sec-ondary outcomes were rate of change in cognition (RBANStotal score) olfaction (UPSIT score) and CSF biomarkers ofAD (Aβ1-42 t-tau p-tau t-tauAβ1-42 and p-tauAβ1-42 ratio)

Table 1 Variables included in the Alzheimer ProgressionScore

Measures Variables

Cognitive measures RBANS attention index score

RBANS immediate memory indexscore

Item recognition task

Neurosensory measure UPSIT total score

Gray matter density measures Bilateral entorhinal cortex

Bilateral lingual cortex

Bilateral putamen

Gray matter cortical thicknessmeasures

Right superior parietal gyrus

Right superior dorsal frontal gyrus

Cerebral blood flow measures Right rostral anterior cingulatecortex

Brain volume variables Bilateral hippocampus volumes

Lateral ventricular volume

CSF measures Total tau concentrations (pgmL)

β-amyloid1-42 concentrations (pgmL)

Abbreviations RBANS = Repeatable Battery for Assessment of Neuro-psychological Status UPSIT = University of Pennsylvania Smell IdentificationTest

NeurologyorgN Neurology | Volume 92 Number 18 | April 30 2019 e5

extending over the 24 months of treatment For both primaryand secondary efficacy analyses we used longitudinal linearrandom effects models (random intercepts) to assessbetween-group differences in rates of change We performedan additional post hoc analysis that included baseline APSscore as a covariate We also constructed additional explor-atory models to test treatment effects on APS rate of changeafter first assigning participants into tertiles based on theirbaseline CSF Aβ1-42

Version 94 of the SAS (SAS Institute Cary NC) statisticalpackage and R (Core Team 2014) were used for analysesexcept for post hoc exploratory analyses which were per-formed using MATLAB APS scores were calculated usingSTAT R and MPLUS21 All statistical tests were 2-sided Ap value le005 was considered to indicate statistical significance

Data sharingAll de-identified data and related documentation from thistrial are available upon request to qualified researchers with-out limit of time subject to a standard data sharing agreementThe PREVENT-AD program is currently developing a lessrestrictive approach to data sharing through the CanadianOpen Neuroscience Platform

ResultsEnrollment and study completion dataThe naproxen- and placebo-assigned groups included 102 and93 participants ThemITT analysis groups included 88 and 72of these The main reasons for exclusion of the remaining 35were apparent ineligibility (discovered typically followingreview of baseline cognitive testing n = 14) early appearanceof intolerable adverse effects (n = 6) or voluntary withdrawal(many reasons given n = 12) Compliance-to-completionrates (24 months on study drug) were 62 for participants

assigned to naproxen and 66 for those given placebo (p =0579 figure 1) After the 3-month run-in the most commoncause for drug discontinuation was occurrence of new AEs (12naproxen-assigned and 3 placebo-assigned individualsfigure 1)

Baseline characteristicsThere were no important differences between naproxen andplacebo-treated groups with respect to age sex or educationAll participants were Caucasian and there were no substantialimbalances across groups in APOE e4 status parental age atAD onset or MoCA score As noted below however weobserved an unexpected difference by treatment assignmentin baseline APS values in the mITT pool (table 2)

Concomitant medicationsOf the 102 naproxen-assigned persons in the ITT pool 85(83) initiated use of concomitant medicines regularly or forshort intervals over the interval of the trial The comparablefigure for those assigned to placebo was 6693 (71 p = 004table e-3 doiorg105061dryadr58d342) The apparentimbalance was attributable principally to initiation of lipid-lowering drugs during the trial but notably a similar imbal-ance was not evident in the mITT analysis pool Baseline oranytime use of such drugs was also similar across the treat-ment arms in the mITT group and accordingly post hocstatistical adjustment for their use brought no appreciablechange in the primary outcome results

Safety outcomesTable 3 shows the frequency of AEs occurring in ge10 ofeither treatment arm Gastrointestinal AEs prompted studydrug discontinuation in 9 naproxen-assigned and 3 placebo-assigned participants Constipation dyspnea hypertensionand petechiae were also substantially more common in per-sons receiving naproxen However as expected this groupreported pain less frequently Overall reports of any AE were

Table 2 Baseline characteristics of Investigation of Naproxen Treatment Effects in Pre-symptomatic Alzheimerrsquos Disease(INTREPAD) participants

Characteristics Total Naproxen Placebo p For difference

No of participants 195 102 93 NS

Age y mean plusmn SD 633 plusmn 56 640 plusmn 59 626 plusmn 50 NS

Male sex n () 51 (26) 25 (25) 26 (28) NS

Education y mean plusmn SD 152 plusmn 34 150 plusmn 32 155 plusmn 36 NS

Parental age at AD onset y mean plusmn SD 733 plusmn 77 728 plusmn 76 738 plusmn 78 NS

MoCA score (out of 30) mean plusmn SD 280 plusmn 16 280 plusmn 15 280 plusmn 16 NS

APOE laquo4 carriers n () 73 (37) 37 (36) 36 (39) NS

Baseline data (mITT) n = 160 n = 88 n = 72

APS minus009 plusmn 086 0021 plusmn 08 minus0214 plusmn 09 006

Abbreviations AD = Alzheimer disease APS = Alzheimer Progression Score mITT = modified intent-to-treat MoCA = Montreal Cognitive AssessmentNo observable difference between the 2 groups except for APS at baseline

e6 Neurology | Volume 92 Number 18 | April 30 2019 NeurologyorgN

more common in naproxen-assigned persons (98 vs 89p = 0015) The principal contributors to this excess werevascular disorders which were more common in the naproxen-assigned group (p = 0023)

Of 10 participants reporting SAEs over 2 years (table e-4 doiorg105061dryadr58d342) 8 had been assigned to nap-roxen Three SAEs required study drug discontinuation while

2 prompted treatment interruption A further 3 participantswere unable or unwilling (some upon physicianrsquos advice) tocontinue participation No suspected unexpected serious ad-verse reaction occurred during the trial

Among blood safety measures both hemoglobin and he-matocrit decreased in persons assigned to naproxen but notin placebo-treated individuals Between-group comparison ofthese measures showed strong differences at all time pointsexcept at 24 months (figures e-1 and e-2 doiorg105061dryadr58d342) No differences were found in the 2 treatmentgroups with respect to other safety measurements (weightpulse rate systolic and diastolic blood pressure)

Primary efficacy outcomeAnalyses of the primary outcome are summarized in figures 3and 4 Among the combined treatment groups the APSshowed a clear increase over the 2-year trial period (β = 0101standard unitsyear SE = 0014 95 confidence interval [CI]0074ndash0130 p lt 0001) However this change did not differmeaningfully between naproxen- and placebo-assigned par-ticipants after 3 12 or 24 months of treatment A longitudinallinear random effects model for APS showed a slight increaseif anything in rate of change among naproxen-assigned per-sons but this was well within chance expectation (time-by-treatment interaction β = +0019 APS unitsyear for naproxenvs placebo SE = 003 95 CI ndash0037 to +0074 p = 051)The APS ratio for rate of change comparing naproxen- vsplacebo-assigned mITT participants was 116 (bootstrapped95 CI 064ndash196)

Secondary efficacy analysesWe used similar linear random effects models in the mITTsample to evaluate treatment-related differences in rate ofchange of RBANS total index score UPSIT score and CSFAD biomarkers (Aβ1-42 t-tau p-tau t-tauAβ1-42 and p-tauAβ1-42) None of these comparisons suggested any benefit of

Table 3 Adverse events (AEs)

Naproxen(n = 102)

Placebo(n = 93) p Value

AEs

Constipation 19 (19) 6 (6) 0011

Dyspnea 23 (23) 10 (11) 0028

Heartburn 31 (30) 17 (18) 005

Peripheral edema 25 (25) 14 (15) 01

Hypertension 19 (19) 8 (9) 004

Petechiae 12 (12) 2 (2) 0009

Pain 20 (20) 28 (30) 009

Serious AEs 8 (8) 2 (2) 007

Vascularcardiacevents

4a 1

Cancers 1 1

Musculoskeletalinjuries

3 mdash

AE table represents n () of participants If an individual had the sameadverse experiences multiple times during the trial he or she was countedonly oncea Two SAEs in this category were judged to have a clear or possible re-lationship with study drug For detailed description of SAEs see e-Results(doiorg105061dryadr58d342)

Figure 3 Treatment effects on Alzheimer Progression Score (APS)

Results for primary and exploratory secondary outcomes are represented (A) There was nomeaningful difference in APS rate of change between treatmentgroups (B) Mean change from baseline (plusmn standard error of themean) in APS did not differ between the 2 treatment groups at any time during the treatmentinterval However APS for both groups increased after 12 and 24months (p lt 005) Data are represented as point estimates (groupmeans) with error bars(standard error of the mean) BL = Baseline M = months

NeurologyorgN Neurology | Volume 92 Number 18 | April 30 2019 e7

naproxen treatment Indeed UPSIT results suggested a trendtoward harm (β = minus0320 SE = 0196 95 CI ndash0704 to+0064 p = 0102 figure 2B) Over the treatment interval 4INTREPAD enrollees (3 naproxen-assigned and 1 placebo-assigned) developed MCI22 or other suggestive cognitivedeficiency sufficient to prompt discontinuation of study drug

Exploratory efficacy analysesAddition of covariates for APOE e4 carrier status age atbaseline years of education and sex as well as their interactionwith time did not materially affect the above findings

As noted above we observed that the naproxen-assignedgroup had higher baseline APS scores than placebo-assignedparticipants This finding prompted us to carry out a post hocaddition of a term for baseline APS in the model used for theprimary endpoint analysis Addition of this term brought noconsequential change from the primary endpoint results Wealso analyzed whether higher baseline APS was associatedwith an increased rate of change in the outcome regardless oftreatment assignment No such association was found Afterpartitioning participants into tertiles of Aβ1-42 concentrationand testing for a triple interaction with time and treatment

Figure 4 Treatment effects on neurosensory and CSF biomarker measures

Linear mixed effect models did not indicate any difference between naproxen- and placebo-assigned groups in rate of change of (A) cognitive or neuro-sensory or (B) biological markers of Alzheimer disease University of Pennsylvania Smell Identification Test (UPSIT) scores decreased over the 2-year trialperiod for the whole group (p lt 005) Data are represented as point estimates (group means) with error bars (standard error of the mean) BL = baselinep-tau = phosphorylated tau RBANS = Repeatable Battery for the Assessment of Neuropsychological Status t-tau = total tau

e8 Neurology | Volume 92 Number 18 | April 30 2019 NeurologyorgN

assignment we also observed no effect of naproxen treatmenton rate of change in APS among the different CSF Aβ1-42tertiles

DiscussionTo assess the potential of the common NSAID naproxensodium for prevention of AD we tested this agent inINTREPAD a 2-year randomized placebo-controlled trialamong cognitively healthy persons at increased risk of de-mentia We observed a clear increase over time in the trialrsquosprimary efficacy outcome (the APS composite representingseveral imaging metrics a neurosensory modality 3 cognitivemeasures and when available CSF biomarkers) Assignmentto naproxen produced an unambiguous increase in adverseevents but no meaningful alteration in the rate of change forthis primary efficacy outcome Overall our results suggest 2points for discussion (1) evaluation of a design intended toincrease efficiency of AD prevention trials and (2) the trialrsquossafety and efficacy results and their ethical implications

We employed several design features intended to improveefficiency for INTREPAD and by implication other ADprevention trials First we attempted to enrich participantsrsquoproportion of persons vulnerable to AD by requiring thatenrollees have an affected parent or multiple siblings23 En-richment by requirement of a first-degree relative with AD isnot novel but the INTREPAD family history criteria werestronger Especially in todayrsquos environment of greater lon-gevity and increased awareness of AD-affected status (thusready identification) this more restrictive method is in-creasingly practical It would seem easier to implement thana requirement of amyloid pathology demonstrable by PET24

or CSF analysis or even homozygosity at the APOE e4 riskallele25 How this enhanced family history method of en-richment compares with the latter more costly or invasivemethods is unknown However we have recently shown thatproximity to an index relativersquos age at symptom onset is re-lated to increased AD biomarker load26 This same metric wasassociated in a separate cohort with brain changes predictiveof time to symptom onset27 Given the methodrsquos reducedcosts and subject burden a costbenefit comparison withother methods probably deserves consideration

Second we selected the composite APS as the trialrsquos primaryoutcome Relative to any single cognitive or biomarker in-dicator composite outcomes of this sort should logically offergreater inference especially in relatively early-stage pre-symptomatic AD We chose the APS in preference to severalsimilar composite indicators that had been less extensivelyvalidated2829 While additional efforts to validate the APS arewarranted it has shown demonstrable utility for evaluation ofpresymptomatic AD progress Specifically in BIOCARD a 1standard unit increase in APS predicted a 5-fold greater hazardof diagnostic progression over time3 In the parallelPREVENT-AD cohort we performed several simulations to

compare the statistical power of the APS to that of its con-stituent markers The APS provided more information thandid any single endpoint including all cognitive measures andit also offered improved performance over a simple summingof z scores of its individual components (data not shown) Wetherefore suggest that the APS or similar multimodal com-posites show promise for prevention trials like INTREPAD

Nonetheless post hoc analyses suggested that our methodsresulted in a trial with substantially less statistical power thanhad been originally projected Our original power estimateswere based in part on expectation that considerable in-formation would be contributed by CSF biomarker dataavailable from more than half of the participants That turnedout not to be the case and accordingly the trial outcomersquos CIwas sufficiently broad to suggest a notable possibility of type IIerror Now having data on rate of change in the trialrsquos out-comes we can estimate that 2250 person-years of observation(eg a sample of 1125 followed over 2 years or 563 followedfor 4 years) would have been required for 80 power todetect a 30 reduction in the rate of APS change Althoughmuch higher than originally estimated this number still rep-resents a considerable improvement over requirements ofconventional prevention trial designs For example the on-going A4 trial will follow 1150 persons over 46 years(5290 person-years) for its primary cognitive outcome2430

Similarly the TOMMORROW trial had originally estimateda requirement of 5800 persons over 4 years (29200 per-son-years)31 By comparison even ignoring costs of PETscans in the former or serial detailed psychometric assess-ments in the latter and assuming costs proportional only toperson-years of observation the INTREPAD design mayachieve cost savings of 57ndash90 over traditional methods

The INTREPAD safety results affirm prior data suggestingthat even in relatively low dosage among younger elderlypersons naproxen and other NSAIDs provoke harm in severalhealth outcomes32ndash34 Our findings that these risks can beheld within bounds by careful monitoring does not obviatethe ethical concern that NSAID treatments are potentiallyharmful and should be given for AD prevention only if theyproduce substantial reduction of AD risk The INTREPADresults provide no evidence for such a reduction This result isespecially salient inasmuch as the trial sample was relativelyyoung for this sort of work and chosen for a favorable riskbenefit balance in relation to NSAID treatment Specificallyparticipants were on average 10 years younger than theiraffected parent or first-affected sibling and were meticulouslyscreened for incipient cognitive disorder35 These attributesappear to weaken arguments that earlier NSAID trials failed toshow benefit because their aging samples were too old or toonear the cusp of symptom onset36ndash38

The null efficacy outcomes of INTREPAD are reinforced byseveral observations The CI around the trialrsquos efficacy rateratio suggests 95 certainty that the true treatment-relatedreduction of AD risk in this trial (or presumably in similar

NeurologyorgN Neurology | Volume 92 Number 18 | April 30 2019 e9

samples) does not exceed 36 This conclusion is buttressedby a consistent absence of apparent benefit on any of the trialrsquossecondary or exploratory outcomes These results recallobservations in ADAPT8 and prior NSAID treatment andprevention trials with null or negative results While we can-not exclude possible benefits of NSAIDs in middle life we cannow suggest that such benefits would be nearly impossible todemonstrate in randomized prevention trials

As regards our further objective of testing or demonstratingmethods for improving efficiency in AD prevention trials weoffer several observations Our enrichment strategy requiringparental or multiple-sibling family history of AD might havebeen improved by a further requirement specifying partic-ipantsrsquo age in relation to their index relativesrsquo onsets Whileevidently more informative than any single biomarker ora composite of only a few such markers our outcome couldtoday probably be improved by incorporation of newer salientmarkers of preclinical disease These might include thresholdvalues for the traditional CSF biomarkers Aβ1-42 t-tau orp-tau or possibly neurofilament light chain as a preconditionto enrollment Power estimation and sample size requirementfor such work should be clearer now than when we began thiswork In this last sense INTREPAD may be viewed as pro-viding an approximate benchmark for work with sampleshaving similar baseline characteristics

In all events this work has left us with extreme pessimismregarding any possible role of NSAIDs in AD preventionInstead our results may suggest reconsideration of inflam-matory diseases (or a proinflammatory diathesis) as a possibleexplanation for the reduced AD incidence among NSAIDusers in observational studies8

AcknowledgmentThe authors thank the research participants and their studypartners for their commitment to this work members of thePREVENT-ADResearch Group (preventadloriscaacknowl-edgementsacknowledgementsphpdate=2018-01-01) DanielMorissette (StatExpert Inc) for executing the trialrsquos dataanalysis plan the Data Monitoring Committee comprisingDenis Evans (Chair) Bruce Lamb James Neaton LonSchneider and Doug Galasko and Sharon Keays and RuthPoole Pharmascience Inc for assistance with drug supplyand management

Study fundingThis trial was supported by McGill University by an un-restricted gift from Pfizer Canada and by infrastructure sup-port from the Canada Fund for Innovation Dr Breitnerrsquosefforts were supported by a Canada Research Chair awardfrom the government of Canada Genetic and laboratory work(JP) were supported by the Fonds de Recherche duQuebecndashSante (FRQ-S) and the JL Levesque FoundationP-FM is supported by funding from the Canada First Re-search Excellence Fund awarded to McGill University for theHealthy Brains for Healthy Lives initiative M-EL-M and

P-FM received research support from the Lazlo amp EtelkaKollar fellowship fund for this project Additional support wasprovided by the Douglas Mental Health University InstituteFoundation The funding agencies had no role in study de-sign data analysis or interpretation of results

DisclosureThe authors report no disclosures relevant to the manuscriptGo to NeurologyorgN for full disclosures

Publication historyReceived by Neurology June 1 2018 Accepted in final form January 72019

Appendix 1 Authors

Name Location Role Contribution

Pierre-FranccediloisMeyer MSc

McGill UniversityMontreal Canada

Author Data analysis andinterpretation draftingand revising themanuscript

JenniferTremblay-MercierMSc

Douglas MentalHealth UniversityInstitute affiliatedwith McGillUniversityMontreal Canada

Author Study designsupervision of dataacquisition draftingand revising themanuscript

Jeannie-MarieLeoutsakosPhD

John HopkinsUniversityBaltimore MD

Author Development ofprimary outcomeAlzheimer ProgressionScore data analysis

CecileMadjar MSc

McGill Centre forIntegrativeNeuroscienceMcGill UniversityMontreal Canada

Author Data managementconceptualization ofdatabase support fordata analysis

Marie-ElyseLafaille-MagnanPhD

McGill UniversityMontreal Canada

Author Major role in dataacquisition (olfaction)study design(randomization)

MelissaSavard MSc

McGill UniversityMontreal Canada

Author Data analysis

PedroRosa-NetoMD PhD

McGill UniversityMontreal Canada

Author Major role in dataacquisition (lumbarpunctures)

JudesPoirier PhD

McGill UniversityMontreal Canada

Author Major role in dataacquisition (laboratorymethods) studydesign revising themanuscript

PierreEtienne MD

McGill UniversityMontreal Canada

Author Major role in dataacquisition(supervision of safetymonitoring) dataanalysis andinterpretation studydesign drafting themanuscript

JohnBreitnerMD MPH

McGill UniversityMontreal Canada

Author Study conceptualizationand designinterpretation of resultsdrafting and revising themanuscript studyfunding

e10 Neurology | Volume 92 Number 18 | April 30 2019 NeurologyorgN

References1 Jack CR Jr Knopman DS Jagust WJ et al Hypothetical model of dynamic

biomarkers of the Alzheimerrsquos pathological cascade Lancet Neurol 20109119ndash128

2 Bateman RJ Xiong C Benzinger TL et al Clinical and biomarker changes in dom-inantly inherited Alzheimerrsquos disease N Engl J Med 2012367795ndash804

3 Leoutsakos JM Gross AL Jones RN Albert MS Breitner JCS ldquoAlzheimerrsquos Pro-gression Scorerdquo development of a biomarker summary outcome for AD preventiontrials J Prev Alz Dis 20163229ndash235

4 in trsquo Veld BA Ruitenberg A Hofman A et al Nonsteroidal antiinflammatory drugsand the risk of Alzheimerrsquos disease N Engl J Med 20013451515ndash1521

5 Breitner JC Haneuse SJ Walker R et al Risk of dementia and AD with priorexposure to NSAIDs in an elderly community-based cohort Neurology 2009721899ndash1905

6 Aisen PS Davis KL Berg JD et al A randomized controlled trial of prednisone inAlzheimerrsquos disease Alzheimerrsquos Disease Cooperative Study Neurology 200054588ndash593

7 Thal LJ Ferris SH Kirby L et al A randomized double-blind study of rofecoxib inpatients with mild cognitive impairment Neuropsychopharmacology 2005301204ndash1215

8 Adapt Research Group Lyketsos CG Breitner JC Green RC et al Naproxen andcelecoxib do not prevent AD in early results from a randomized controlled trialNeurology 2007681800ndash1808

9 Breitner JCS Poirier J Etienne PE Leoutsakos JM PREVENT-AD Research GRationale and structure for a new center for studies on prevention of Alzheimerrsquosdisease (StoP-AD) J Prev Alz Dis 20163236ndash242

10 Hughes CP Berg L Danziger WL Coben LA Martin RL A new clinical scale for thestaging of dementia Br J Psychiatry 1982140566ndash572

11 Nasreddine ZS Phillips NA Bedirian V et al The Montreal Cognitive AssessmentMoCA a brief screening tool for mild cognitive impairment J Am Geriatr Soc 200553695ndash699

12 Randolph C Tierney MC Mohr E Chase TN The repeatable battery for the as-sessment of neuropsychological status (RBANS) preliminary clinical validity J ClinExp Neuropsychol 199820310ndash319

13 Albert M Soldan A Gottesman R et al Cognitive changes preceding clinicalsymptom onset of mild cognitive impairment and relationship to ApoE genotypeCurr Alzheimer Res 201411773ndash784

14 Rajah MN Languay R Grady CL Age-related changes in right middle frontal gyrusvolume correlate with altered episodic retrieval activity J Neurosci 20113117941ndash17954

15 Doty RL Shaman P Dann M Development of the University of Pennsylvania SmellIdentification Test a standardized microencapsulated test of olfactory functionPhysiol Behav 198432489ndash502

16 Lafaille-Magnan ME Poirier J Etienne P et al Odor identification as a biomarker ofpreclinical AD in older adults at risk Neurology 201789327ndash335

17 Zijdenbos AP Forghani R Evans AC Automatic ldquopipelinerdquo analysis of 3-D MRI datafor clinical trials application to multiple sclerosis IEEE Trans Med Imaging 2002211280ndash1291

18 Aubert-Broche B Fonov VS Garcia-Lorenzo D et al A new method for structuralvolume analysis of longitudinal brain MRI data and its application in studying thegrowth trajectories of anatomical brain structures in childhood Neuroimage 201382393ndash402

19 Wang J Alsop DC Song HK et al Arterial transit time imaging with flow encodingarterial spin tagging (FEAST) Magn Reson Med 200350599ndash607

20 Gross AL Mungas DM Leoutsakos JS Albert MS Jones RN Alzheimerrsquos diseaseseverity objectively determined and measured Alzheimers Dement 20164159ndash168

21 Muthen LK Muthen BO Mplus userrsquos guide Statistical analysis with latent variablesUserrsquos guide Los Angeles Muthen amp Muthen 2004

22 Winblad B Palmer K Kivipelto M et al Mild cognitive impairment beyond con-troversies towards a consensus report of the International Working Group on MildCognitive Impairment J Intern Med 2004256240ndash246

23 Fratiglioni L Ahlbom A Viitanen M Winblad B Risk factors for late-onset Alz-heimerrsquos disease a population-based case-control study Ann Neurol 199333258ndash266

24 Sperling RA Rentz DM Johnson KA et al The A4 study stopping AD beforesymptoms begin Sci Transl Med 20146228fs213

25 A study of CAD106 and CNP520 versus placebo in participants at risk for the onset ofclinical symptoms of Alzheimerrsquos Disease [online] Available at clinicaltrialsgovshowNCT02565511 Accessed May 2018

26 Villeneuve S Vogel JW Gonneaud J et al Proximity to parental symptom onset andamyloid-beta burden in sporadic Alzheimer disease JAMA Neurol 201875608ndash619

27 Vogel JW Vachon-Presseau E Pichet Binette A et al Brain properties predict proximity tosymptom onset in sporadic Alzheimerrsquos disease Brain 20181411871ndash1883

28 Jedynak BM Lang A Liu B et al A computational neurodegenerative disease pro-gression score method and results with the Alzheimerrsquos Disease Neuroimaging Ini-tiative cohort Neuroimage 2012631478ndash1486

29 Xiong C van Belle G Chen K et al Combining multiple markers to improve thelongitudinal rate of progression-application to clinical trials on the early stage ofAlzheimerrsquos disease Stat Biopharm Res 20135

30 Clinical trial of solanezumab for older individuals who may be at risk for memory loss[online] Available at clinicaltrialsgovshowNCT02008357 Accessed May 2018

31 Budur K Welsh-Bohmer K Burns D et al A pharmacogenetics-supported clinicaltrial to delay onset of mild cognitive impairment due to Alzheimerrsquos disease using low-dose pioglitazone an update on the TOMORROW study Alzheimers Dement 201410P809ndashP810

32 LangmanMJWeil JWainwright P et al Risks of bleeding peptic ulcer associated withindividual non-steroidal anti-inflammatory drugs Lancet 19943431075ndash1078

33 PirmohamedM James S Meakin S et al Adverse drug reactions as cause of admissionto hospital prospective analysis of 18 820 patients BMJ 200432915ndash19

34 Kearney PM Baigent C Godwin J Halls H Emberson JR Patrono C Do selectivecyclo-oxygenase-2 inhibitors and traditional non-steroidal anti-inflammatory drugsincrease the risk of atherothrombosis Meta-analysis of randomised trials BMJ 20063321302ndash1308

35 Hayden KM Zandi PP Khachaturian AS et al Does NSAID use modify cognitivetrajectories in the elderly The Cache County Study Neurology 200769275ndash282

36 Aisen PS Schafer KA GrundmanM et al Effects of rofecoxib or naproxen vs placeboon Alzheimer disease progression a randomized controlled trial JAMA 20032892819ndash2826

37 Breitner JC Baker LD Montine TJ et al Extended results of the Alzheimerrsquos diseaseanti-inflammatory prevention trial Alzheimers Dement 20117402ndash411

38 Leoutsakos JM Muthen BO Breitner JC Lyketsos CG ADAPT Research TeamEffects of non-steroidal anti-inflammatory drug treatments on cognitive decline varyby phase of pre-clinical Alzheimer disease findings from the randomized controlledAlzheimerrsquos Disease Anti-inflammatory Prevention Trial Int J Geriatr Psychiatry201227364ndash374

Appendix 2 Coinvestigators

Name Location Role Contribution

SergeGauthierMD PhD

McGill UniversityMontreal Canada

Executive member Study designconsulting

Denis AEvansMD

Rush UniversityChicago IL

Chair Data Safetyand MonitoringBoard

Reviewed allsafety data

NatashaRajahPhD

McGill UniversityMontreal Canada

fMRI expert Leader ofepisodicmemory taskdata

RickHogePhD

McGill UniversityMontreal Canada

fMRI expert Leader ofblood flow data

PierreBellecPhD

MontrealUniversityMontreal Canada

fMRI expert Leader ofresting statedata

LouisCollinsPhD

McGill UniversityMontreal Canada

MRI expert Leader of brainvolumes data

DavidFontainePhD

Douglas MentalHealth UniversityInstituteMontreal Canada

Neuropsychologist Cognitiveevaluations

VasavanNair MD

Douglas MentalHealth UniversityInstituteMontreal Canada

Psychiatrist Medicalevaluation

AlanEvansPhD

McGill UniversityMontreal Canada

Database and MRIexpert

DatabaseDevelopmentand Leader ofcorticalthickness data

NeurologyorgN Neurology | Volume 92 Number 18 | April 30 2019 e11

DOI 101212WNL0000000000007232 published online April 5 2019Neurology

Pierre-Franccedilois Meyer Jennifer Tremblay-Mercier Jeannie Leoutsakos et al Alzheimer disease

INTREPAD A randomized trial of naproxen to slow progress of presymptomatic

This information is current as of April 5 2019

ServicesUpdated Information amp

232fullhttpnneurologyorgcontentearly20190405WNL0000000000007including high resolution figures can be found at

Citations

232fullotherarticleshttpnneurologyorgcontentearly20190405WNL0000000000007This article has been cited by 5 HighWire-hosted articles

Subspecialty Collections

lled_consort_agreementhttpnneurologyorgcgicollectionclinical_trials_randomized_controClinical trials Randomized controlled (CONSORT agreement)

httpnneurologyorgcgicollectionalzheimers_diseaseAlzheimers disease

httpnneurologyorgcgicollectionall_cbmrt_null_hypothesisAll CBMRTNull Hypothesisfollowing collection(s) This article along with others on similar topics appears in the

Permissions amp Licensing

httpwwwneurologyorgaboutabout_the_journalpermissionsits entirety can be found online atInformation about reproducing this article in parts (figurestables) or in

Reprints

httpnneurologyorgsubscribersadvertiseInformation about ordering reprints can be found online

ISSN 0028-3878 Online ISSN 1526-632XWolters Kluwer Health Inc on behalf of the American Academy of Neurology All rights reserved Print1951 it is now a weekly with 48 issues per year Copyright Copyright copy 2019 The Author(s) Published by

reg is the official journal of the American Academy of Neurology Published continuously sinceNeurology