Articles

Efficacy and safety of tarenflurbil in mild to moderate

Alzheimer’s disease: a randomised phase II trial
Gordon K Wilcock*, Sandra E Black*, Suzanne B Hendrix, Kenton H Zavitz, Edward A Swabb, Mark A Laughlin, on behalf of the Tarenflurbil
Phase II Study investigators†

Summary
Background The amyloid-β peptide Aβ42 has been implicated in the pathogenesis of Alzheimer’s disease (AD). We Lancet Neurol 2008; 7: 483–93
aimed to test the effects of tarenflurbil, a selective Aβ42-lowering agent (SALA), on cognition and function in Published Online
patients with mild to moderate AD. April 30, 2008
DOI:10.1016/S1474-
4422(08)70090-5
Methods 210 patients living in the community who had a mini-mental state examination (MMSE) score of 15–26
This online publication
were randomly assigned to receive tarenflurbil twice per day (400 mg [n=69] or 800 mg [n=70]) or placebo (n=71) has been corrected.
for 12 months in a phase II, multicentre, double-blind study. Primary efficacy outcomes were the AD assessment The corrected version
scale cognitive subscale (ADAS-cog), the Alzheimer’s Disease Cooperative Study activities of daily living scale first appeared at
thelancet.com/neurology
(ADCS-ADL), and the clinical dementia rating sum of boxes (CDR-sb). In a 12-month extended treatment phase,
on March 21, 2011
patients who had received tarenflurbil continued to receive the same dose, and patients who had received placebo
See Reflection and Reaction
were randomly assigned to tarenflurbil at 800 mg or 400 mg twice per day. Primary efficacy analyses were done by page 468
intention to treat. This trial is registered with Health Canada (084527) and the Medicines and Healthcare products *Contributed equally
Regulatory Agency in the UK (20365/0001/A 69316).
†Investigators listed in full at end
of report
Findings A prespecified interaction analysis revealed that patients with mild AD (baseline MMSE 20−26) and University of Oxford, John
moderate AD (baseline MMSE 15−19) responded differently to tarenflurbil in the ADAS-cog and the ADCS-ADL Radcliffe Hospital, Oxford, UK
(p≤0·10); therefore, these groups were analysed separately. Patients with mild AD in the 800 mg tarenflurbil group (G K Wilcock DM); Sunnybrook
Health Sciences Centre,
had lower rates of decline than did those in the placebo group in activities of daily living (ADCS-ADL difference in
University of Toronto, Toronto,
slope 3·98 [95% CI 0·33 to 7·62] points per year, effect size [reduction from placebo decline rate] 46·4%, ON, Canada (S E Black MD); and
Cohen’s d 0·45; p=0·033) and global function (CDR-sb difference –0·80 [–1·57 to –0·03] points per year, effect Myriad Pharmaceuticals, Salt
size 35·7%, Cohen’s d 0·42; p=0·042); slowing of cognitive decline did not differ significantly (ADAS-cog difference Lake City, UT, USA
(S B Hendrix PhD, K H Zavitz PhD,
–1·36 [–4·07 to 1·36] points per year, effect size 33·7%, Cohen’s d 0·20; p=0·327). In patients with moderate AD,
E A Swabb PhD,
800 mg tarenflurbil twice per day had no significant effects on ADCS-ADL and ADAS-cog and had a negative effect M A Laughlin MD)
on CDR-sb (–52%, Cohen’s d –1·08; p=0·003). The most common adverse events were diarrhoea (in seven, nine, Correspondence to:
and five patients in the 800 mg, 400 mg, and placebo groups, respectively), nausea (in seven, seven, and four Gordon K Wilcock,
patients), and dizziness (in five, nine, and four patients). Patients with mild AD who were in the 800 mg tarenflurbil Nuffield Department of
Medicine, University of Oxford,
group for 24 months had lower rates of decline for all three primary outcomes than did patients who were in the
Level 7, John Radcliffe Hospital,
placebo group for months 0–12 and a tarenflurbil group for months 12–24 (all p<0·001), and had better outcomes Headley Way, Headington,
than did patients who were in the placebo group for months 0–12 and the 800 mg tarenflurbil group for months Oxford OX3 9DU, UK
12–24 (all p<0·05). [email protected]

Interpretation 800 mg tarenflurbil twice per day was well tolerated for up to 24 months of treatment, with evidence
of a dose-related effect on measures of daily activities and global function in patients with mild AD.

Funding Myriad Pharmaceuticals.

Introduction model prevented defects in learning and memory.5,8 The

Deposition of amyloid-β (Aβ) in the brain is a concentrations of tarenflurbil required to produce Aβ42-
neuropathological hallmark of Alzheimer’s disease lowering effects in vitro and in vivo5 can be achieved in
(AD) and a potential cause of neuronal damage. The human beings at doses that have been well tolerated.9–11
molecule believed to initiate the process is the 42-amino- This phase II, multicentre, randomised, double-blind,
acid peptide Aβ42,1–3 which is found in soluble oligomeric placebo-controlled, parallel-group study was designed
forms and in neuritic and diffuse plaques. Tarenflurbil to evaluate the effects of tarenflurbil on cognition and
(MPC-7869; former non-proprietary name R- function in patients with mild to moderate AD over
flurbiprofen) is a selective Aβ42-lowering agent (SALA) 12 months. We also did prespecified analyses for
that modulates γ-secretase activity and reduces interactions of treatment groups with baseline scores of
production of Aβ42 without effect on production of cognition and function, and a prespecified analysis to
Aβ40.4–7 In a mouse model of AD, tarenflurbil reduced assess the effect of tarenflurbil on patients with mild
brain concentrations of Aβ42, and chronic dosing in this AD and moderate AD separately. At the end of the

http://neurology.thelancet.com Vol 7 June 2008 483

 Articles

12-month study, eligible patients were enrolled in an Patients taking acetylcholinesterase inhibitors (AChEI)
additional 12-month blinded phase to investigate the were eligible for participation if their treatment dose had
effects of longer-term treatment and delayed treatment. been stable for at least 3 months before screening.
Patients receiving chronic aspirin therapy for
Methods cardioprotection (≤325 mg per day) or stable doses for at
Patients least 3 months of antidepressants, antipsychotics,
English-speaking adults 55 years of age or older who anxiolytics, vitamin E, or ginkgo biloba extract were also
were living in the community were eligible to participate. eligible. Patients were excluded if they had any of the
Patients were required to meet the Diagnostic and following: evidence of epilepsy, focal brain lesion, or head
statistical manual of mental disorders, 4th edition injury with loss of consciousness or immediate confusion
(DSM-IV, text revision) criteria for dementia12 and the after the injury; any major psychiatric disorder (eg, DSM-
National Institute of Neurological and Communicative IV-defined psychosis, major depression, bipolar disorder,
Disorders and Stroke and Alzheimer’s Disease and or alcohol or substance abuse); history of hypersensitivity
Related Disorders Association (NINCDS-ADRDA) to any non-steroidal anti-inflammatory drug (NSAID) or
criteria for probable AD.13 The following were additional cyclo-oxygenase-2-specific inhibitor; recent history of
inclusion criteria at screening: no clinically significant chronic use of NSAIDs or aspirin (>325 mg per day);
focal intracranial lesion on CT or MRI scans within the history of upper gastrointestinal bleeding that required
previous 12 months; MMSE14 score of 15–26, inclusive; transfusion or surgery within the previous 3 years;
a modified Hachinski ischaemic score of less than 4; at documented evidence of an active gastric or duodenal
least 6 years of education or sufficient work history to ulcer within the previous 3 months; history of NSAID-
exclude mental retardation; adequate vision and hearing associated ulcers; history of active malignancy except for
to participate in study assessments; and an English- basal cell carcinoma or squamous cell carcinoma of the
speaking caregiver who saw the patient at least 4 days skin, or prostate cancer, within the preceding 24 months;
per week and accompanied him or her to clinic visits. a chronic or acute renal, hepatic, or metabolic disorder;

254 assessed

44 excluded
36 did not meet inclusion criteria
1 refused to participate
7 excluded for other reasons

210 randomised

71 assigned to placebo 69 assigned to 400 mg tarenflurbil 70 assigned to 800 mg tarenflurbil

(66 in safety group)* twice per day (71 in safety group)† twice per day (70 in safety group)

2 untreated 1 untreated

6 had adverse events 12 had adverse events 8 had adverse events

3 withdrew consent 3 withdrew consent 2 at site closed by sponsor
2 at site closed by sponsor 3 at site closed by sponsor 1 withdrew consent
2 discontinued for other reasons 3 discontinued for other reasons 1 lost to follow-up
1 discontinued for other reasons

56 completed intervention 47 completed intervention 57 completed intervention

4 at site with compliance issues 3 at site with compliance issues 2 at site with compliance issues
3 at site with procedural issues 2 at site with procedural issues 1 at site with procedural issues
1 no post-baseline data 1 no post-baseline data 1 no post-baseline data

61 in ITT analysis 62 in ITT analysis 66 in ITT analysis

Figure 1: Trial profile

No protocol deviations prevented inclusion in the ITT analysis. *Did not include two patients who were untreated or three patients who were given the wrong kit on day 1 and received
400 mg tarenflurbil twice per day for part of the study. †Did not include one patient who was untreated but did include three patients from the placebo group who received wrong kit.

484 http://neurology.thelancet.com Vol 7 June 2008

 Articles

All patients in ITT population at 12 months (n=189) Patients with mild AD* (n=130)
Placebo Tarenflurbil Tarenflurbil Placebo Tarenflurbil Tarenflurbil
(n=61) 400 mg† (n=62) 800 mg† (n=66) (n=46) 400 mg† (n=36) 800 mg† (n=48)
Age (years) 74·4 (7·5) 73·4 (7·9) 75·8 (7·8) 75·6 (6·8) 74·8 (6·4) 75·7 (7·6)
Male 32 (52%) 32 (52%) 32 (48%) 27 (59%) 21 (58%) 24 (50%)
Ethnic origin
White 60 (98%) 60 (97%) 62 (94%) 45 (98%) 35 (97%) 44 (92%)
Black 0 (0%) 0 (0%) 2 (3%) 0 (0%) 0 (0%) 2 (4%)
Asian 0 (0%) 1 (2%) 0 (0%) 0 (0%) 0 (0%) 0 (0%)
Other 1 (2%) 1 (2%) 2 (3%) 1 (2%) 1 (3%) 2 (4%)
Weight (kg) 72·1 (13·0) 70·4 (12·7) 69·3 (13·4) 73·4 (13·3) 73·8 (12·2) 69·4 (13·6)
Time since AD diagnosis (months) 25·5 (19·8) 30·0 (22·2) 28·8 (21·0) 22·6 (19·4) 30·2 (22·8) 29·5 (22·3)
MMSE 21·6 (3·1) 20·5 (3·6) 21·3 (2·9) 22·9 (2·1) 23·1 (1·9) 22·8 (1·7)
ADAS-cog 30·8 (1·2) 33·5 (1·2) 31·2 (1·2) 27·5 (1·1) 28·6 (1·2) 28·3 (1·0)
ADCS-ADL 57·6 (1·6) 56·2 (1·6) 57·5 (1·5) 58·9 (1·6) 61·4 (1·8) 59·8 (1·6)
CDR-sb 6·20 (0·34) 6·53 (0·33) 6·39 (0·32) 5·66 (0·33) 5·01 (0·37) 5·96 (0·32)
NPI 10·67(1·32) 9·87 (1·30) 10·50 (1·26) 9·76 (1·49) 7·64 (1·66) 10·15 (1·44)
NPI-CD 5·98 (0·85) 6·10 (0·83) 6·21 (0·81) 5·62 (0·96) 5·14 (1·07) 6·04 (0·93)
Current AChEI use 59 (97%) 58 (94%) 62 (94%) 45 (98%) 34 (94%) 44 (92%)
Duration of current AChEI use 18·2 (15·7) 20·7 (14·6) 18·7 (15·6) 16·0 (12·3) 19·7 (13·7) 16·9 (14·5)
(months)

Data are mean (SD) or number (%). ITT=intention to treat. MMSE=mini-mental state examination. ADAS-cog=Alzheimer’s disease assessment scale cognitive subscale.
ADCS-ADL=Alzheimer’s Disease Cooperative Study activities of daily living. CDR-sb=clinical dementia rating sum of boxes. NPI=neuropsychiatric inventory.
NPI-CD=NPI-caregiver distress. AChEI=acetylcholinesterase inhibitors. *MMSE≥20 at baseline. †Twice per day.

Table 1: Baseline demographic and clinical characteristics

recent history of investigational drug use or major drug was supplied in tablets containing 400 mg. All
surgery; an uncontrolled cardiac condition (New York participants, irrespective of group and dose, were to
Heart Association class III or IV); anticoagulant therapy take two tablets of identical appearance twice per day.
within 12 weeks or therapy with any CYP2C9 inhibitor Randomisation was stratified on the basis of current
within 2 weeks; or memantine therapy within 30 days use or non-use of AChEI. Kit numbers at each site were
before screening. assigned by a web-based randomisation system, and
The study was done in accordance with the principles Myriad Genetics generated the sequence for allocation
of the Declaration of Helsinki and the International of patients to treatment groups. The Myriad Genetics
Conference on Harmonisation of Technical employee who generated two copies of the randomisation
Requirements for Registration of Pharmaceuticals for scheme had no further role in the study. One copy of
Human Use guideline for good clinical practice. The the randomisation list was stored in a sealed, tamper-
protocol was approved by institutional review boards at proof envelope at Myriad Pharmaceuticals, and the
each study site. Patients who completed the first other was sent to Clinical Research Networks (Salt Lake
12 months of the study were allowed by the ethics City, UT, USA). Each study site enrolled participants,
boards in Canada to enter the 12-month extension and a web-based system managed by Clinical Research
phase, because no concerns were raised by the data Networks assigned participants to treatment groups and
safety monitoring (DSM) committee. The UK kits to patients on the basis of the randomisation.
multicentre research ethics committee did not approve The double-blind, extended treatment phase of the
the extended treatment phase because data on safety study included eligible patients from the 16 sites in
and efficacy from the first 12 months were not available Canada. Patients initially randomised to tarenflurbil
when the extension phase began. All patients and their continued their established doses of 800 mg twice per
caregivers provided written informed consent. day (800 mg0–24 group) or 400 mg twice per day (400 mg0–24
group); those initially randomised to placebo (placebo0–12
Procedures group) were re-randomised (1:1) to receive tarenflurbil
31 sites in Canada and the UK enrolled patients from at 800 mg twice per day (800 mg12–24 group) or 400 mg
November 3, 2003, to April 24, 2006. Patients were twice per day (400 mg12–24 group). This second
randomised (1:1:1) in blocks of six to receive 12 months’ randomisation was implemented in the same way as
treatment with 400 mg tarenflurbil twice per day, the first randomisation. Investigators and patients
800 mg tarenflurbil twice per day, or placebo. Active remained blinded to tarenflurbil dose throughout the

http://neurology.thelancet.com Vol 7 June 2008 485

 Articles

because the study populations were similar. A sample

Tarenflurbil Placebo Difference in slope Tarenflurbil 800 mg* vs
800 mg* (95% CI)† placebo size of 67 patients per group was calculated to be
required to achieve 80% power to detect an effect size of
Effect size Cohen’s d p
80% on both the ADAS-cog (assuming a mean decline
Mild AD‡ n=48 n=46 of 5·5 points with SD 6·4) and the CDR-sb (assuming a
ADCS-ADL –4·60 (1·30) –8·57 (1·32)§ 3·98 (0·33 to 7·62) 46·4% 0·45 0·033 mean decline of 1·57 with SD 2·2), with a two-sided α
CDR-sb 1·33 (0·28) 2·13 (0·28) –0·80 (–1·57 to –0·03) 35·7% 0·42 0·042 of 0·05, a 28% dropout rate, and a correlation of 0·50
ADAS-cog 2·67 (0·97) 4·02 (0·99) –1·36 (–4·07 to 1·36) 33·7% 0·20 0·327 between ADAS-cog decline and CDR-sb decline.
Moderate AD¶ n=18 n=15 As prespecified in the statistical analysis plan, a
ADCS-ADL –13·40 (2·90) –8·31 (3·03) –5·13 (–13·44 to 3·17) –38·2% –0·44 0·223 statistically significant change in one measure of
CDR-sb 4·63 (0·57) 2·20 (0·58) 2·42 (0·82 to 4·03) –52·0% –1·08 0·003 cognition (ADAS-cog) and one measure of function
ADAS-cog 7·63 (1·72) 6·86 (1·79) 0·77 (–4·13 to 5·67) –10·1% –0·11 0·756 (CDR-sb or ADCS-ADL) was required for an overall
Data are mean (SE) rates of change in points per year, unless otherwise indicated. ADCS-ADL=Alzheimer’s Disease
positive result. The rates of change (slopes) for ADAS-
Cooperative Study activities of daily living scale. CDR-sb=clinical dementia rating sum of boxes. ADAS-cog=Alzheimer’s cog, CDR-sb, and ADCS-ADL were compared for the
disease assessment scale cognitive subscale. *Twice per day. †Slope of change in tarenflurbil group minus slope for 800 mg tarenflurbil and placebo groups by use of a
placebo group. Owing to rounding, some differences are not exactly equal to the differences in annual rates. ‡Patients repeated-measures mixed model with terms for
with MMSE ≥20 at baseline. §n=45. ¶Patients with MMSE ≤19 at baseline.
treatment, time, baseline score, baseline score by
Table 2: Rates of change in primary efficacy outcome measures over months 0–12 and comparison of treatment interaction, and time by treatment interaction.
800 mg tarenflurbil with placebo Because this model fits a covariance structure and
assumes that missing observations are missing at
study. The primary efficacy outcomes measured random within the subgroups defined by the model, no
cognition and function. last-observation-carried-forward imputation was done.
Cognition was assessed with the Alzheimer’s disease For the ADAS-cog and CDR-sb, worsening is indicated
assessment scale cognitive subscale (ADAS-cog),15 by an increase in test score; for the ADCS-ADL,
including the delayed recall subscale,16 at baseline (day 1 worsening is indicated by a decrease in test score.
clinic visit) and at 3-month intervals thereafter until The primary efficacy analysis was done with data from
month 24. Functional ability was evaluated with the the intention-to-treat (ITT) population, defined as all
Alzheimer’s Disease Cooperative Study (ADCS) activities patients randomly assigned to treatment who received
of daily living scale (ADCS-ADL),17 and global function at least one dose of study drug and who had at least one
with the clinical dementia rating sum of boxes (CDR-sb).18 post-baseline efficacy measurement, or who
Global function was also assessed as an exploratory discontinued the study because of an adverse event
outcome measure with the clinician-interview-based before an efficacy assessment. Two-tailed p values were
impression of change plus caregiver input (CIBIC+).19 compared with α=0·05. Interaction terms were removed
The ADCS-ADL was done on day 1 and at months 6 and if non-significant at α=0·1. For all analyses, the higher
12, and then at 3-month intervals until month 24; the dose was designated as primary, and comparisons with
CDR-sb and CIBIC+ were done at 3-month intervals the 400 mg group were secondary. Both doses were
from day 1 to month 24. The neuropsychiatric inventory included in the follow-on phase to maintain blinding
(NPI),20 including the caregiver distress scale (NPI-CD)21 and because no data on the relative efficacy of the doses
and the 11-question MMSE,14 were exploratory outcomes. were available when patients entered this phase. Data
Information on adverse events and compliance was from the 800 mg and 400 mg groups also were compared
collected at all post-baseline study visits (month 1 and for descriptive purposes. Because the comparison
every 3 months until the end of the study) and via between the 800 mg tarenflurbil group and the placebo
additional telephone conversations with caregivers at group was prospectively designated as primary, no
week 2 and every month until the end of the study. overall test of significance was done to compare all
Blood and urine samples were collected regularly for three study groups and no adjustments were made for
clinical laboratory analysis, and for measurement of multiple comparisons.
plasma concentrations of tarenflurbil. Physical assess- Effect size is reported as a mean percentage difference
ments were made at baseline, month 1, month 3, and in slopes (the difference between rates of change in the
then at 3-month intervals. Standard 12-lead resting treated and placebo groups, divided by the rate for the
electrocardiograms were done at screening and at placebo group, reported as a percentage; a 100% effect
month 12. Unblinded safety data were reviewed size would suggest a halting of decline, and a 50% effect
quarterly by an independent DSM committee. size would suggest that the treated group declined at
half the rate of the placebo group). Cohen’s d (the
Statistical analysis difference in slopes between the groups divided by the
Data from the ADCS study of NSAIDs in patients with SD of the slope of the comparison group) is also
mild to moderate AD22 were used to estimate placebo provided. A small value for Cohen’s d can be caused by
decline rates for the power calculation in this trial, a small effect or a scale with large variability. A positive

486 http://neurology.thelancet.com Vol 7 June 2008

 Articles

effect size and positive Cohen’s d indicate a positive Results

drug effect. The exploratory efficacy variables NPI, NPI- 210 patients were randomly assigned to treatment and
CD, and MMSE were analysed with the same 207 received at least one dose of tarenflurbil or placebo
longitudinal model as were the primary efficacy data. (safety population; figure 1). Efficacy data were excluded
Analyses of ADAS-cog, CDR-sb, and ADCS-ADL were
repeated separately for prespecified subgroups of
patients with mild AD (MMSE score 20−26) or moderate A
n=48
0 800 mg tarenflurbil
AD (MMSE score 15−19) at baseline. The relation
400 mg tarenflurbil
between maximum plasma concentrations of tarenflurbil n=36 Placebo
(Cmax) and these primary efficacy outcomes was analysed –5
n=45 n=23 Placebo→tarenflurbil
with a partial Pearson correlation from a mixed model. n=16
For the analysis of efficacy from 12 to 24 months, the

Mean change in ADCS-ADL

effect of 800 mg twice per day versus placebo for –10 n=20
24 months could not be estimated directly because a
concurrent placebo group was present for only
12 months. Two analyses were prespecified for the –15
comparison of decline (slopes): months 0–24 for the
800 mg0–24 group versus months 0–12 for the placebo0–12
–20
group; and months 0–24 for the 800 mg0–24 group versus
months 0–24 for the placebo0–24 group (ie, the second
randomisation was ignored). All available data were UK and Canada Canada only
–25
used for both of these analyses, including those from
patients in the UK who did not enter the follow-on B
phase. The 800 mg and 400 mg groups were compared n=48
0
to provide additional insight into the drug effect against
n=46
a lower dose. In a randomised-start analysis, data from
n=36
the extended treatment phase were used to assess 1 n=22
whether effects in patients who started to receive n=19
Mean change in CDR-sb

800 mg tarenflurbil twice per day at 12 months

2
(800 mg12–24 group) were less than those in patients who n=18
started to receive this dose at 0 months (800 mg0–24
group), which would suggest disease modification.23–25 3
For safety analyses, we used data from all patients
who received at least one dose of study drug. The
incidence of adverse events was compared between the 4
placebo group and each active treatment group by use
of Fisher’s exact test for events of sufficient frequency UK and Canada Canada only
or substantial differences between groups. 5
This trial is registered with Health Canada (084527) C
and the Medicines and Healthcare products Regulatory –2
Agency in the UK (20365/0001/A 69316). n=48
0
Role of the funding source 2 n=46
n=23
The study sponsor and the primary investigators were n=36 n=18
Mean change in ADAS-cog

together responsible for study design, data analysis, 4

n=20
interpretation, and writing of the report. All authors had
6
full access to all the data in the study and had final
responsibility for the decision to submit for publication. 8

Figure 2: Mean change in ADCS-ADL, CDR-sb, and ADAS-cog scores over time 10
for patients with mild impairment at baseline (MMSE ≥20)
(A) ADCS-ADL. Rate of decline (slope) during months 0–24 in the 800 mg0–24 12
group was 39% of the rate during months 0–12 in the placebo0–12 group. (Cohen’s
14
d=0·41; p=0·015). (B) CDR-sb. Rate of decline during months 0–24 in the 800
mg0–24 group was 46% of the rate during months 0–12 in the placebo0–12 group. UK and Canada Canada only
16
(Cohen’s d=0·58; p<0·001). (C) ADAS-cog. Rate of decline during months 0–24
0 3 6 9 12 15 18 21 24
in the 800 mg0–24 group was 39% of that during months 0–12 in the placebo0–12
group (Cohen’s d=0·27; p=0·109). For rates of decline during months 0–12 see Month
table 2, and for rates of decline for months 0–24 see table 3.

http://neurology.thelancet.com Vol 7 June 2008 487

 Articles

completed the 12-month study. The other 47 patients

A (23%) withdrew from the study prematurely, 26 (13%)
0·0
because of an adverse event. Baseline demographic and
–2·0
clinical characteristics were similar across study groups
(table 1).
ADCS-ADL 12-month slope

–4·0 The repeated-measures mixed model showed a

significant qualitative (or directional) interaction
–6·0 between baseline score and treatment group for the
p=0·028 ADAS-cog (effect size 94·8%, p=0·044) and ADCS-ADL
–8·0 scale (effect size 112·2%, p=0·100), with a non-
significant effect for the CDR-sb (effect size 76·7%,
–10·0 Placebo p=0·309), indicating that the treatment effect depended
on baseline scores and that the key assumption of
–12·0
equality of slopes between the groups was violated
B enough to be an issue for two of the three primary
0·5 outcomes. The only prespecified separation by baseline
severity was based on MMSE score. To assess whether
this subgroup analysis adequately addressed the issue
1·0
CDR-sb 12-month slope

of the interactions between baseline score and treatment

group, we did two additional tests for interactions with
treatment group: one based on the three primary
1·5
analyses run with baseline MMSE score rather than
p=0·025 baseline ADAS-cog, ADCS-ADL, or CDR-sb score; and
2·0 one based on the three primary analyses run with
baseline mild or moderate classification rather than
Placebo baseline score (webtable 1). The interactions between
2·5 baseline MMSE score and treatment group were all
C significant (ADCS-ADL p=0·048, CDR-sb p<0·001, and
0 ADAS-cog p=0·009), as were the interactions between
mild versus moderate classification and treatment
1·0
group for the ADCS-ADL (p=0·026) and the CDR-sb
ADAS-cog 12-month slope

(p<0·001). The interaction for ADAS-cog results was

not significant (p=0·429). Further efficacy analyses were
2·0
done separately for the mild and moderate subgroups
p=0·162 because the interaction analyses suggested that pooling
3·0 of the groups could produce misleading inferences for
the primary endpoints.
4·0 130 patients had mild AD (MMSE ≥20) at baseline.
Placebo These patients had demographic characteristics similar
5·0
to those of the overall population, with the exception of
–20 0 20 40 60 80 100 120 less severe disease and a shorter time on AChEI therapy
Cmax (μg/mL) quintile average at baseline (table 1). In this subgroup, treatment with
800 mg tarenflurbil was associated with a significantly
Figure 3: Relation of 12-month effect sizes to maximum plasma concentrations of tarenflurbil (Cmax) lower rate of decline than was treatment with placebo
(A) ADCS-ADL. (B) CDR-sb. (C) ADAS-cog. P values measure the strength of the linear relation between Cmax and in activities of daily living (ADCS-ADL; table 2,
each efficacy outcome.
figure 2A) and global function (CDR-sb; table 2,
figure 2B); the difference between the two groups in
See Online for webtables 1 and 2 for two study sites: one site was closed early owing to cognitive decline was not significant (ADAS-cog; table 2,
non-compliance with good clinical practice; the other figure 2C) but the ADAS-cog effect size was consistent
site was excluded from the ITT population before with that for CDR-sb (table 2). Differences between the
unblinding of the study because discrepancies in placebo and 800 mg tarenflurbil groups in the rates of
signatures on some source documents raised concerns decline in ADCS-ADL and CDR-sb increased over time
about data authenticity. The ITT population consisted (tables 2 and 3). Rates of decline in the placebo group
of the 189 patients who received at least one dose of over 12 months were consistent with those in a similar
study drug, were not at one of the two excluded sites, group of historical controls (webtable 2; data from
and had at least one post-baseline efficacy measurement. Aisen and colleagues22 and unpublished data from the
Of the safety population, 160 (77%) participants ADCS).

488 http://neurology.thelancet.com Vol 7 June 2008

 Articles

Tarenflurbil Tarenflurbil Placebo0–24† Placebo0–24† vs tarenflurbil 800 mg0–24* Placebo0–24† vs tarenflurbil 400 mg0–24*
800 mg0–24* 400 mg0–24* (n=46)
(n=48) (n=36)
Effect size Cohen’s d p Effect size Cohen’s d p

ADCS-ADL –5·16 (0·61) –6·41 (0·70) –9·26 (0·70)‡ 44·3% 0·87 <0·001 30·6% 0·60 0·006
CDR-sb 1·12 (0·12) 1·45 (0·14) 1·81 (0·15) 37·9% 0·69 <0·001 20·0% 0·37 0·090
ADAS-cog 2·20 (0·40) 3·72 (0·45) 5·64 (0·44) 60·9% 1·14 <0·001 34·1% 0·64 0·003

Data are mean (SE) rates of change in points per year, unless otherwise indicated. Mild AD was defined as MMSE ≥20 at baseline. ADCS-ADL=Alzheimer’s Disease Cooperative
Study activities of daily living scale. CDR-sb=clinical dementia rating sum of boxes. ADAS-cog=Alzheimer’s disease assessment scale cognitive subscale. *Data for 24 months
from patients who received the stated tarenflurbil dose twice per day for months 0–24. †Data for 24 months from patients who received placebo for months 0–12 and
tarenflurbil (400 mg or 800 mg twice per day) for months 12–24. ‡n=45.

Table 3: Annual rates of change in primary efficacy outcome measures over months 0–24 in patients with mild AD at baseline

Maximum plasma concentrations of tarenflurbil (Cmax)

Tarenflurbil Placebo0–24† Difference‡ p
had significant linear relationships with ADCS-ADL 800 mg0–24* (n=11)
(estimate 0·041 points/year per μg/mL; p=0·028) and (n=22)
CDR-sb (estimate –0·007 points/year per μg/mL; ADCS-ADL –9·8 (2·5) –19·5 (3·6)§ 9·7 0·032
p=0·025), indicating that patients with higher plasma CDR-sb 1·9 (0·7) 4·3 (0·9) 2·4 0·038
concentrations of tarenflurbil had slower rates of ADAS-cog 2·9 (1·7)¶ 11·5 (2·4) 6·6 0·005
decline (figure 3). The rate of decline on ADAS-cog was
not significantly lower for patients with a higher Cmax Data are mean (SE) changes in score from baseline to 24 months.
ADCS-ADL=Alzheimer’s Disease Cooperative Study activities of daily living scale.
(estimate –0·015; p=0·162). CDR-sb=clinical dementia rating sum of boxes. ADAS-cog=Alzheimer’s disease
59 patients had moderate AD (MMSE≤19) at baseline. assessment scale cognitive subscale. *Data for 24 months from patients who
In this group, treatment with placebo was associated received 800 mg tarenflurbil twice per day for months 0–24. †Data for
with a significantly lower rate of decline in global 24 months from patients who received placebo for months 0–12 and tarenflurbil
(400 mg or 800 mg twice per day) for months 12–24. ‡Difference between
function than was 800 mg tarenflurbil (CDR-sb, table 2). changes from baseline in the tarenflurbil 800 mg0–24 and placebo0–24 groups.
Non-significant effects were recorded for activities of §n=12. ¶n=20.
daily living and cognition (ADCS-ADL and ADAS-cog,
Table 4: Change in primary efficacy outcomes over months 0–24 in
table 2). Effect sizes were not consistent across measures. patients with mild AD at baseline (randomised-start analysis)
In addition, decline rates in the placebo group were
smaller than expected on the basis of data from historical
controls (webtable 2; data from Aisen and colleagues22 or had no change in CIBIC+ at month 12, compared with
unpublished data from the ADCS). only 8 of 42 (19%) patients on placebo (effect size 15%,
Although there was no evidence of an interaction p=0·208). Of patients with moderate AD at baseline, 3 of
between treatment group and baseline NPI or MMSE 14 (21%) in the placebo group had an improvement or no
score for the exploratory outcome measures (NPI, change in CIBIC+ over the 12 months, compared with
NPI-CD, and MMSE) at 12 months, results for these 0 of 16 in the 800 mg tarenflurbil group (difference
analyses and for the analysis of the CIBIC+ are between groups p=0·070), consistent with the smaller
presented for patients with mild or moderate AD than expected decline in the moderate placebo group that
separately for consistency with the primary analysis was recorded for the other measures.
subgroups. There were no significant differences 86 of the 107 eligible patients in Canada were
between the 800 mg and placebo groups for patients randomised in the extended treatment phase and
with mild AD on exploratory efficacy measures over 85 (80%) were treated (webfigure 1). 15 patients (18%) See Online for webfigure 1
12 months: NPI effect size 0%, p=1·000; NPI-CD effect withdrew prematurely from the follow-on phase, five
size –24%, p=0·751; and MMSE effect size –15%, (6%) because of an adverse event, six (7%) because of
p=0·761. No significant effects were seen for patients withdrawn consent, three (4%) because of disease
with moderate AD on these same exploratory efficacy progression, and one (1%) because of admission to a
measures over 12 months: NPI effect size –124%, nursing home. The similar dropout rates between
p=0·239; NPI-CD effect size –137%, p=0·166; and groups in both phases of the study suggest that blinding
MMSE effect size 1%, p=0·980. The large negative of the study was not jeopardised by drug tolerability
effect sizes in this subgroup were related to the smaller issues. None of the patients who discontinued because
than expected decline in the primary measures for of disease progression in the follow-on phase had mild
patients on placebo who had moderate AD. AD at baseline, and four of those who discontinued
Of patients with mild AD at baseline, 13 of 42 (31%) owing to adverse events in the follow-on phase had mild
recipients of 800 mg tarenflurbil twice per day improved AD at baseline (one in the 800 mg0–24 group and three in

http://neurology.thelancet.com Vol 7 June 2008 489

 Articles

group for all three primary efficacy outcomes at

Placebo Tarenflurbil Tarenflurbil Total
(n=66) 400 mg* 800 mg* (N=207) 24 months (all p≤0·032) and between baseline MMSE
(n=71) (n=70) and treatment group, indicating that treatment effects
Any event 56 (85%) 59 (83%) 63 (90%) 178 (86%) differed by baseline severity as they did in the first
Diarrhoea 5 (8%) 9 (13%) 7 (10%) 21 (10%) 12-month phase. Therefore, further efficacy analyses
Nausea 4 (6%) 7 (10%) 7 (10%) 18 (9%) were done separately on the mild and moderate
Dizziness 4 (6%) 9 (13%) 5 (7%) 18 (9%)
subgroups according to baseline MMSE.
Urinary tract infection 5 (8%) 7 (10%) 5 (7%) 17 (8%)
63 (74%) of the patients in the follow-on phase had
Agitation 5 (8%) 7 (10%) 4 (6%) 16 (8%)
mild AD (MMSE ≥20) at baseline. Of these, during the
first 12 months, 24 (38%) were in the 800 mg0–24 group,
Confused state 2 (3%) 10 (14%)† 4 (6%) 16 (8%)
18 (29%) were in the 400 mg0–24 group, and 21 (33%)
Upper respiratory tract infection 7 (11%) 4 (6%) 2 (3%) 13 (6%)
were in the placebo0–12 group. Of those on placebo,
Vomiting 3 (5%) 3 (4%) 6 (9%) 12 (6%)
12 (19% of patients in the extension phase) were
Headache 6 (9%) 3 (4%) 3 (4%) 12 (6%)
randomised to the 800 mg12–24 group at month 12 and
Constipation 3 (5%) 4 (6%) 4 (6%) 11 (5%)
nine (14%) were randomised to the 400 mg12–24 group.
Depression 4 (6%) 1 (1%) 6 (9%) 11 (5%)
For the patients with mild AD, analysis of data from
Eosinophilia 0 (0%) 3 (4%) 6 (9%)‡ 9 (4%)
the extension phase according to original randomisation
Nasopharyngitis 4 (6%) 0 (0%) 5 (7%)§ 9 (4%)
showed a lower rate of decline for all three scales in the
Pneumonia 0 (0%) 3 (4%) 5 (7%) 8 (4%)
800 mg0–24 group than in the placebo0–24 group (table 3). In
Rash 0 (0%) 5 (7%) 3 (4%) 8 (4%)
patients with mild AD, the treatment difference between
Urinary incontinence 6 (9%)¶ 0 (0%) 2 (3%) 8 (4%)
the 800 mg0–24 group and the placebo0–24 group was 4·10
Blood glucose increased 4 (6%) 0 (0%) 1 (1%) 5 (2%) points per year (95% CI 2·21 to 5·99). Mean rate of
Data are number (%) of patients. *Twice per day. †p=0·032 vs placebo. ‡p=0·028 vs placebo. §p=0·028 vs 400 mg change in global function (CDR-sb) over 24 months was
tarenflurbil. ¶p=0·011 vs 400 mg tarenflurbil. P values not reported are >0·05. 0·69 points per year (95% CI –1·08 to –0·29) less in the
800 mg0–24 group than in the placebo0–24 group. Scores on
Table 5: Adverse events that occurred in ≥5% of patients in any treatment group during months 0–12
all three primary endpoints were stable from month 18 to
month 24 in the 800 mg0–24 group but declined in the
Placebo Tarenflurbil Tarenflurbil Total 400 mg0–24 group (figure 2). Mean rate of change in
(n=66) 400 mg* 800 mg* (N=207) cognition (ADAS-cog) over 24 months was –3·44 points
(n=71) (n=70)
per year (95% CI –4·64 to –2·23; table 3) different
Gastrointestinal disorders 3 (5%) 2 (3%) 2 (3%) 7 (3%) between the 800 mg0–24 group and the placebo0–12 group.
Metabolism and nutritional disorders 1 (2%) 2 (3%) 2 (3%) 5 (2%) In the randomised-start analysis, change from
Psychiatric disorders 3 (5%) 0 (0%) 2 (3%) 5 (2%) baseline to 24 months differed between the 800 mg0–24
Data are number (%) of patients. *Twice per day.
and the placebo0–24 groups for all three scores in patients
with mild AD (table 4). Higher maximum concentrations
Table 6: Most common types of adverse event that led to discontinuation during months 0–12 of tarenflurbil in the blood during the first 12 months
were associated with a greater response on all three
the group that received placebo for months 0–12 and primary outcomes over 24 months (all p<0·05).
tarenflurbil [800 mg or 400 mg twice per day] for No consistent effects were seen in patients with moderate
months 12–24). AD who enrolled in the follow-on phase. Dropout rates
The baseline demographic and clinical characteristics were higher for these patients (9/22, 41%) than for those
of patients who entered the extended treatment phase with mild disease at baseline (9/63, 14%).
See Online for webtables 3 and 4 were similar across treatment groups (webtable 3) and In patients with mild AD at baseline, rate of decline
similar to those of the population enrolled at baseline. in MMSE, but not that in NPI or NPI-CD, was
There were no statistically significant interactions significantly lower in the 800 mg0–24 group than in the
between country and treatment group for any of the placebo0–24 group (webtable 4). MMSE scores were
three primary endpoints (all p>0·10), indicating that stabilised significantly from 12–24 months in the
the population of patients in Canada who were eligible 800 mg0–24 group compared with the 400 mg0–24 and
See Online for webfigure 2 for the follow-on study was not substantially different placebo0–24 groups (webfigure 2).
from the overall population of patients in the UK and Data from the 207 participants who received at least
Canada. All patients enrolled in the extended treatment one dose of study drug in the first 12 months of the
phase were receiving stable AChEI therapy at baseline: study were included in the safety analysis. The mean
44 on donepezil (52%), 28 on galantamine (33%), and duration of exposure was 10·5 (SD 3·32) months and was
13 on rivastigmine (15%). similar across treatment groups. 160 (79%) patients
The primary model, in which the 800 mg0–24 group received study drug for 12 months and most were at least
was compared with the placebo0–12 group, showed an 90% compliant with treatment during months 0–12
interaction between baseline scores and treatment (178/207, 86%) and months 12–24 (77/85, 91%).

490 http://neurology.thelancet.com Vol 7 June 2008

 Articles

Compared with the placebo group, during months 0–12

30 3 months
significantly more patients in the 400 mg tarenflurbil 6 months
group experienced a confused state (table 5; p=0·032) 9 months
and significantly more patients in the 800 mg 12 months
25
tarenflurbil group had eosinophilia (table 5; p=0·028).
Occurrences of eosinophilia were transient and not
associated with characteristic symptoms or signs. Other
adverse events that occurred with a non-significantly 20

Patients (cumulative %)
greater frequency in the tarenflurbil groups than in the
placebo group included anaemia, lower respiratory tract
infection or pneumonia, rash, and episodic 15
hypertension.
Five deaths occurred during months 0–12: four in the
400 mg tarenflurbil group and one in the 800 mg group. 10
There was no pattern to events that led to death, with
two malignancies (both in patients on 400 mg
tarenflurbil), one myocardial infarction (in a patient on 5
400 mg), one cerebral haemorrhage (in a patient on
400 mg), and one sepsis of unknown origin (in a patient
on 800 mg). 25 patients had one or more serious adverse
0
events (total 36 events). Frequency of discontinuation Placebo 400 mg tarenflurbil 800 mg tarenflurbil
was similar for the three study arms (800 mg tarenflurbil,
n=13; 400 mg tarenflurbil, n=21; placebo, n=13; figure 4). Figure 4: Cumulative discontinuation over the first 12 months of the study
The most common types of adverse event leading to
discontinuation were also similar across study arms and between mild versus moderate classification and
(table 6). treatment group for the ADCS-ADL and CDR-sb but
There were small changes in some laboratory not the ADAS-cog reassured us that the prespecified
parameters during months 0–12 that were statistically separation of mild and moderate patients by MMSE
significant but not medically significant. Haemoglobin score addresses the interactions of baseline severity of
was about 0·5 g/dL lower at month 3 in the 800 mg AD with treatment and is the most appropriate way to
tarenflurbil group than in the placebo group (p<0·001) interpret the results of the study. However, the
and subsequently returned towards baseline. Other significant treatment interaction in the primary models
laboratory values showed small changes within the and reliance on a subgroup analysis mean that this
normal range, but were not medically meaningful. phase II study is unlikely to be elevated to pivotal status
Data on adverse events from all patients who received in support of an efficacy claim.
at least one dose of study treatment during the extended The greater rate of decline in the CDR-sb over
phase were included in the safety analysis for this phase 12 months for patients with moderate AD on active
(webtable 5). Two deaths from cancer occurred in the treatment compared with those on placebo was See Online for webtable 5
follow-on phase but neither was deemed drug related. unexpected. This outcome might be partly explained by
the decline in the CDR-sb in patients with moderate
Discussion AD on placebo, which was less than in historical
In this phase II study, treatment with 800 mg of controls and similar to the decline in patients with mild
tarenflurbil twice per day had a significant benefit, AD on placebo. Additionally, this moderate subgroup
which increased over time, on global function (35·7% was small (table 2), because this trial enrolled patients
effect size) and activities of daily living (46·4% effect with a relatively mild baseline MMSE score of 15–26, in
size) in patients with mild AD (95% of whom were whom a disease-modifying effect would probably be
already being treated with stable doses of concomitant more apparent. Some baseline differences in disease
AChEI at entry and throughout the study). There was severity among the patients with moderate AD might
no significant effect on cognition. Although there are have contributed to the smaller than expected decline
no universally agreed-upon criteria to define a clinically rates in the placebo group; however, the sample sizes
relevant effect size, a position paper from the European were too small for us to state this definitively. Rates of
Alzheimer’s Disease Consortium Task Force26 on decline in patients with mild AD on placebo were
disease-modifying trials in AD suggested that a 30% or consistent with historical controls, suggesting that the
greater reduction in the rate of cognitive and functional effects we recorded in patients with mild AD are not
decline should be judged clinically relevant. likely to be due to an atypical placebo group.
The significant interactions between baseline MMSE The association between higher maximum plasma
score and treatment group for all primary endpoints concentrations of tarenflurbil during months 0–12 and

http://neurology.thelancet.com Vol 7 June 2008 491

 Articles

a greater response on all three primary outcomes over Conflicts of interest

24 months was consistent with the concentration- GKW and SEB have received grants and honoraria from Myriad
Pharmaceuticals. SBH, KHZ, EAS, and MAL have an equity or
response effect seen in the first 12 months. The
ownership interest in Myriad Pharmaceuticals. SBH, KHZ, and EAS
randomised-start analysis suggests that, in patients are currently employees of Myriad Pharmaceuticals, and MAL is a
with mild AD, treatment for 24 months resulted in former employee of Myriad Pharmaceuticals.
greater benefit than did delayed treatment for 12 months Acknowledgments
for all three primary efficacy measures. This smaller The authors thank David Moskovic for overseeing of clinical
benefit in a delayed-start group is consistent with operations and Oxford PharmaGenesis for editorial and writing
assistance. We acknowledge the following members of the data
modification of the underlying disease process rather safetymonitoring committee: Howard Feldman, Clive Ballard,
than a purely symptomatic effect. This effect did not Weichung Joe Shih, and Robert Makuch. GKW was partly funded by
seem to be due to differential dropout in the follow-on the NIHR Biomedical Research Centre Programme, Oxford.
phase, because none of the patients who dropped out References
owing to disease progression in the follow-on phase 1 Hardy JA, Higgins GA. Alzheimer’s disease: the amyloid cascade
hypothesis. Science 1992; 256: 184–85.
had mild AD at baseline.
2 Gandy S. The role of cerebral amyloid β accumulation in common
Although the follow-on phase was done only in forms of Alzheimer disease. J Clin Invest 2005; 115: 1121–29.
Canada, the additional effect from 12–24 months cannot 3 Golde TE. The Aβ hypothesis: leading us to rationally-designed
be primarily due to loss of the UK patients, because the therapeutic strategies for the treatment or prevention of
Alzheimer disease. Brain Pathol 2005; 15: 84–87.
Canadian patients had a smaller treatment effect than 4 Beher D, Clarke EE, Wrigley JD, et al. Selected non-steroidal anti-
did those in the UK during the first 12 months (data not inflammatory drugs and their derivatives target γ-secretase at a
shown). Any selection bias introduced at enrolment novel site. Evidence for an allosteric mechanism.
J Biol Chem 2004; 279: 43419–26.
into the follow-on phase would probably make a 5 Eriksen JL, Sagi SA, Smith TE, et al. NSAIDs and enantiomers of
treatment effect more difficult to identify. Additionally, flurbiprofen target γ-secretase and lower Aβ42 in vivo.
there was no significant interaction between country J Clin Invest 2003; 112: 440–49.
and treatment group during the first 12 months. 6 Morihara T, Chu T, Ubeda O, Beech W, Cole GM. Selective
inhibition of Aβ42 production by NSAID R-enantiomers.
In summary, this phase II study provides preliminary J Neurochem 2002; 83: 1009–12.
evidence of a possible therapeutic effect of tarenflurbil, 7 Weggen S, Eriksen JL, Das P, et al. A subset of NSAIDs lower
the first SALA to complete a phase II evaluation. In amyloidogenic Aβ42 independently of cyclooxygenase activity.
Nature 2001; 414: 212–16.
patients with mild AD, 800 mg tarenflurbil twice per 8 Kukar T, Prescott S, Eriksen JL, et al. Chronic administration of
day was associated with less decline in functional ability R-flurbiprofen attenuates learning impairments in transgenic
than was placebo. Moreover, tarenflurbil was generally amyloid precursor protein mice. BMC Neurosci 2007; 8: 54.
9 Murray ED, Quiggle D, Gibson KM, Leipold DD, Loughman BE,
safe and well tolerated in this small study for up to Wechter WJ. Phase I single-dose pharmacokinetics and lack of
24 months of treatment. These findings justify phase enantiomer bioinversion of E-7869 (R-flurbiprofen).
III studies of tarenflurbil at the 800 mg twice daily dose Clin Pharmacol Ther 2000; 67: 103 (abstr).
10 Swabb EA, Quiggle DD, Gutierrez I, et al. Multi-dose phase 2a
in patients with mild AD. trial of MPC-7869 in prostate cancer patients with increasing
Contributors prostate specific antigens (PSA). Proc Am Assoc Cancer Res 2002;
All authors made substantial contributions to the design of the trial 43: 749.
protocol, the planned analysis of the data, and writing and review of the 11 Galasko DR, Graff-Radford N, May S, et al. Safety, tolerability,
manuscript. GKW and SEB were principal investigators for this research pharmacokinetics and Abeta levels after short-term administration
and finalised the manuscript. SBH did the biostatistical analysis. of R-flurbiprofen in healthy elderly individuals.
Alzheimer Dis Assoc Disord 2007; 21: 292–99.
Tarenflurbil Phase II Study investigators 12 American Psychiatric Association. Dementia of the Alzheimer’s
Canada—Y Bacher (McGill University, Montreal), M Borrie (Parkwood type. Diagnostic and statistical manual of mental disorders, fourth
Hospital, London), S Cohen (Neurology Research Inc, Toronto), edition, text revision (DSM-IV-TR). Washington DC: American
E DeCoteau (Saskatoon City Hospital, Saskatoon), K Farcnik (Toronto Psychiatric Association, 2000: 157.
Western Hospital, Toronto), L Fornazzari (Centre for Addiction and 13 McKhann G, Drachman D, Folstein M, Katzman R, Price D,
Mental Health, Toronto), A Garcia (Kingston General Hospital, Stadlan EM. Clinical diagnosis of Alzheimer’s disease: report of
Kingston), D A Guzman (Elisabeth Bruyere Health Centre, Ottawa), the NINCDS-ADRDA Work Group under the auspices of
J Ingram (1559744 Ont Inc, Peterborough), A Kirk (Royal University Department of Health and Human Services Task Force on
Hospital, Saskatoon), C MacKnight (Queen Elizabeth II Health Alzheimer’s disease. Neurology 1984; 34: 939–44.
Sciences Centre, Halifax), S Martin and T Mendis (Parkinson’s and 14 Folstein MF, Folstein SE, McHugh PR. “Mini-mental state”. A
Neurodegenerative Disorder Clinic, Ottawa), N P V Nair (Douglas practical method for grading the cognitive state of patients for the
Hospital, Verdun, Canada), Z Nasreddine (Neuro-Rive-Sud, Greenfield clinician. J Psychiatr Res 1975; 12: 189–98.
Park), P Shelton (Winnipeg Clinic, Winnipeg), F Veloso (Pasqua 15 Rosen WG, Mohs RC, Davis KL. A new rating scale for
Hospital, Regina), J Willmer (Can-Med Clinical Research Inc, Victoria), Alzheimer’s disease. Am J Psychiatry 1984; 141: 1356–64.
M Winger (Optimum Clinical Research, Oshawa). UK—A Bayer 16 Mohs RC, Knopman D, Petersen RC, et al. Development of
(University of Wales College of Medicine, Pernarth, Vale of cognitive instruments for use in clinical trials of antidementia
drugs: additions to the Alzheimer’s disease assessment scale that
Glamorgan), R Bullock (Victoria Hospital, Swindon), R Coles (Kildean
broaden its scope. The Alzheimer’s Disease Cooperative Study.
Hospital, Stirling), P Crome (The City General Hospital, Alzheimer Dis Assoc Disord 1997; 11 (suppl 2): S13–21.
Staffordshire), M Evans (Leighton Hospital, Crewe), A Hughes
17 Galasko D, Bennett D, Sano M, et al. An inventory to assess
(Inverclyde Royal Hospital, Greenock), R Jones (St. Martin’s Hospital, activities of daily living for clinical trials in Alzheimer’s disease.
Bath), A Larner (The Walton Centre for Neurology and Neurosurgery, The Alzheimer’s Disease Cooperative Study.
Liverpool), I McKeith (Newcastle General Hospital, Newcastle-upon- Alzheimer Dis Assoc Disord 1997; 11 (suppl 2): S33–39.
Tyne), D Wilkinson (Moorgreen Hospital, West End Southampton).

492 http://neurology.thelancet.com Vol 7 June 2008

 Articles

18 Hughes CP, Berg L, Danziger WL, Coben LA, Martin RL. A new 23 Leber P. Slowing the progression of Alzheimer disease:
clinical scale for the staging of dementia. Br J Psychiatry 1982; methodologic issues. Alzheimer Dis Assoc Disord 1997;
140: 566–72. 11 (suppl 5): S10–21.
19 Schneider LS, Olin JT, Doody RS, et al. Validity and reliability of 24 Hendrix S, Gutin A, Zavitz K, Carron G, Laughlin M, Horton S.
the Alzheimer’s Disease Cooperative Study-clinical global Modification of the staggered start and randomized withdrawal
impression of change. The Alzheimer’s Disease Cooperative clinical trial designs: toward a practical demonstration of disease
Study. Alzheimer Dis Assoc Disord 1997; 11 (suppl 2): S22–32. modification in Alzheimer’s disease (AD).
20 Cummings JL, Mega M, Gray K, Rosenberg-Thompson S, Alzheimers Dement 2007; 3: S160 (abstr).
Carusi DA, Gornbein J. The neuropsychiatric inventory: 25 Horton S, Gutin A, Hendrix S. A mathematical comparison of a
comprehensive assessment of psychopathology in dementia. “randomized withdrawal” clinical trial design and a parallel
Neurology 1994; 44: 2308–14. groups design to demonstrate disease modification in
21 Kaufer DI, Cummings JL, Christine D, et al. Assessing the impact Alzheimer’s disease (AD). Alzheimer Dement 2007; 3: S161 (abstr).
of neuropsychiatric symptoms in Alzheimer’s disease: the 26 Vellas B, Andrieu S, Sampaio C, Wilcock G. Disease-modifying
neuropsychiatric inventory caregiver distress scale. trials in Alzheimer’s disease: a European task force consensus.
J Am Geriatr Soc 1998; 46: 210–15. Lancet Neurol 2007; 6: 56–62.
22 Aisen PS, Schafer KA, Grundman M, et al. Effects of rofecoxib or
naproxen vs placebo on Alzheimer disease progression: a
randomized controlled trial. JAMA 2003; 289: 2819–26.

http://neurology.thelancet.com Vol 7 June 2008 493