Noradrenergic Add-on Therapy with Extended-Release Guanfacine in Alzheimer’s Disease (NorAD): study protocol for a randomised clinical trial and COVID-19 amendments

The following is the primary outcome:

The following are the secondary outcomes:

The following is the Global Clinical Impression Index:

Between January 2019 and March 2020, all participants followed the assessments outlined in Table 2. At the week 2 (W2) visit, a telephone call is to check for any adverse events. At the week 12 (W12) visit, final assessments are carried out, and unused study medication is collected. One week after cessation of treatment, there is a final follow-up telephone call (W13). The neuropsychological assessments are completed by a trained rater, an individual trained specifically by a trained dementia research coordinator to conduct the outcome assessments at the research site.

The NorAD trial commenced in January 2019, with 48 patients randomised by March 2020. At that point, due to UK-wide lockdown and COVID-19-related research suspension and following Imperial College Healthcare NHS Trust (ICHT) guidance, NorAD suspended recruitment and all visits at the trial site. Those patients who commenced treatment between 01 January 2020 and 13 March 2020 continued treatment until the date of their next planned visit at which point a telephone assessment was conducted by a trained rater in place of attending at the trial site. The telephone assessment included a shorter version of the assessment battery but did not include any cognitive instruments or vital signs measurements (it should be noted that this was not considered a safety risk as treatment was discontinued at the time of the telephone visit). Telephone assessment consisted of completing safety questions (Adverse Event/Concomitant Medication, Side-Effect Questionnaire, Epworth Sleepiness Scale) and carer questionnaires (Neuropsychiatric Inventory, Alzheimer’s Disease co-operative Study – Activities of Daily Living Inventory, Zarit Burden Interview) where appropriate. Twelve patients prematurely ended trial participation. Of these, 7 had reached the week 4 (W4) visit and therefore had attended their second assessment with outcome data. Five participants ended participation before this stage, with no outcome data available.

NorAD was restarted in August 2020 according to the NIHR restart framework for clinical research (https://​www.​nihr.​ac.​uk/​documents/​restart-framework/​24886). Those participants that had previously commenced treatment but not reached the week 4 visit were eligible to restart participation (using the same randomisation code) to complete a full 12 weeks of treatment (from the date of recommencement). Three of those patients have successfully restarted, and the other 2 are considered withdrawn.

Recruitment resumed using local safety measures and a revised set of assessments to minimise COVID risk from multiple in-person attendances. These were approved as a protocol amendment by the Research Ethics Committee prior to the trial restart. Revisions included the week 8 (W8) visit being converted to a telephone call to check for any adverse events rather than an in-person visit. This was not considered a safety risk as at W8 patients have been on the same dose of treatment for 7 weeks, and to date, there have been no reports of new side effects or safety issues at this time point. In addition, following the restart, the week 12 (W12) visit includes the Alzheimer’s Disease Co-operative Study – Clinical Global Impression of Change (ADCS-CGIC) with the final follow-up telephone call (W13) including part 2 of the ADCS-CGIC.

Participants who drop out of the NorAD study are recorded as withdrawn. Participants may withdraw consent to participate in the trial and will no longer receive the study treatment or undergo assessments. The chief investigator may withdraw a participant because of clinical factors, discontinuation of the study treatment or a change in circumstances. A “lost to follow-up” would be no news of a participant after every effort to contact the participant or informant. In all cases, the investigator suggests the participant attends a study end visit, and substantial effort will be made to obtain follow-up data for all participants. All data until the point of withdrawal from the trial are recorded.

Trial participants, care providers, the study team (including the study coordinator, trained raters) and the chief investigator are blinded to each participant’s treatment allocation. The data analyst and clinical pharmacist are unblinded to the treatment allocation. Unblinding of all participants may take place prior to the end of the study on safety grounds, in the case of a serious adverse event or events. This would be initiated by the Trial Steering Committee, in discussion with the chief pharmacist and the Data Monitoring Committee. In a medical emergency, the chief investigator or delegated clinician may unblind individual participants for their own safety and for further medical care.

The primary outcome measure is the 11-item version of the ADAS-Cog used to measure cognition in patients with mild to moderate AD. It is considered the gold standard for assessing the efficacy of treatment in dementia clinical trials. The testing kit contains the study instruments with three sets of word lists for word recall and word recognition to be completed at week 0, week 4 and week 12 assessments.

The secondary outcome measures include the following:

Tests of Attention: Trails A and B is a validated timed test of complex attention. Trails Making Test Part A consists of 25 circles on a piece of paper with the numbers 1–25 written in random positions. It requires the subject to connect the circles in numeric order as quickly as possible within 90 s. Trails Making Test Part B consists of 25 circles on a piece of paper with the numbers 1–13 and the letters A–L written in random positions. It requires the subject to connect the circles in numeric and alphabetical order as quickly as possible, alternating between numbers and letters within 180 s. The trails are completed on worksheets with a pencil.

Digit-symbol substitution is a timed neuropsychological test sensitive to dementia. It consists of a key of nine digit-symbol pairs shown at the top of the page, followed by rows of numbers below missing symbols. It requires the subject to match symbols to numbers according to the given key as fast as possible. This is scored as the number of correct symbols within 60 s. The test is completed on a worksheet with a pencil.

Test of Everyday Attention (Map Search and Elevator Counting with Distraction) is a validated test of attention performed using everyday materials; the complete kit can be obtained through Pearsons including all materials and answer sheets [42]. In addition to the kit, the testing requires two coloured erasable markers (red and blue) and a CD player. The map search is a measure of visual selective attention and involves looking at a large map of Philadelphia in which the subject searches for a symbol within 2 min. Elevator counting with distraction is a measure of auditory selective attention, and it requires the subject to imagine they are in an elevator and listen to a series of tones to establish which floor the elevator has arrived at with the presence of distractor tones. There are three sets of symbols (A, B and C) and three CDs (A, B and C) to be completed at week 0, week 4 and week 12 assessments.

CANTAB-RVP (Rapid Visual Information Processing) [43] is a validated computerised measure of sustained attention administered in 7 min using an iPad tablet. The subject is required to look for a sequence of numbers; when they see the target sequence, they must respond by selecting the button in the centre of the screen as quickly as possible. To ensure that the subject can complete the RVP, they initially complete the CANTAB-MOT (Motor Screening Task), a 2-min task that screens for any sensorimotor deficits or lack of comprehension by asking the subject to select a coloured cross that appears in locations on the screen.

Carer Input and Activities of Daily Living: Neuropsychiatric Inventory (NPI) is a brief interview with the study partner to evaluate 12 behavioural areas commonly affected in patients with dementia; it is routinely used to evaluate the effects of treatment on these symptoms. Zarit Burden Interview (22 items) is a well-known measure of caregiver burden in caregivers of patients with dementia. It is a 22-item interview with the study partner. Each item on the interview is a statement which the subject is asked to answer on a 5-point scale. Alzheimer’s Disease Co-operative Study – Activities of Daily Living Inventory (ADCS-ADL) is commonly used to assess the performance of AD patients in 23 basic and instrumental activities of daily living (ADLs) giving a score from 0 to 78. It is completed with the study partner as a structured interview asking the subject to consider the past 4 weeks at the time of rating.

Toxicity and tolerability measurements including blood pressure and heart rate measurements are taken when the participants are sitting and after 1 min standing. Epworth Sleepiness Scale is widely used as a subjective measure of a subject’s sleepiness. It is a list of eight situations in which an individual rates their tendency to become sleepy on a 4-point scale giving a score from 0 to 24. The scale estimates whether the subject is experiencing excessive sleepiness and is completed with the patient and study partner. Common side effects of guanfacine (dry mouth, sedation, fatigue) are recorded using a patient and study partner reported Side Effects Questionnaire (SEQ). The participants are asked to report any side effects beginning with an open question addressing any new problems to then ask of five potential known side effects and a final question to ask of any new symptoms given a further five examples.

Observation of change following cessation of treatment as measured by Alzheimer’s Disease Co-operative Study – Clinical Global Impression of Change (ADCS-CGIC) [44]. Clinical Global Impression of Change (CGIC) is a measure of clinically meaningful change; it is a semi-structured interview based on the impression of the change completed with the patient and the study partner, using a modified version of the ADCS-CGIC with permission from the NIA Alzheimer’s Disease Cooperative Study (NIA Grant AG10483) to record the impression of change having ceased treatment.

The outcome measures are administered by a trained rater. Testing kits, study instruments and data collection forms are stored at the trial site. Where possible, to maximise standardisation and maintain consistency across administrations, the same trained rater will conduct the assessment in a similar testing environment at around the same time of day.

We will use the manuals on each questionnaire to determine the appropriate treatment of missing data items. When there is no information on this in the manuals, then we will replace the missing values by scaling up the outcome scale score according to the total possible scores from the missing and non-missing items on the questionnaire (providing the missing data does not account for more than 40% of the total for the specific patient).

All statistical models with numeric/questionnaire outcomes at follow-up will be analysed using mixed linear regression. There will be a separate model for each outcome variable, including one for ADAS-Cog. Outcome measurements taken at 4 weeks and at 12 weeks will form the mixed linear regression outcome variable, which will include a random intercept for each participant [45]. The regression model will be adjusted for a baseline measurement of the same outcome variable and for the concomitant medication (indicator variables for rivastigmine and galantamine use, rather than donepezil use, will be included), with indicator variables for time (4 weeks follow-up versus 12 weeks) and for the treatment group, and with an interaction between treatment and time (to allow for reporting of separate treatment effects at 4 weeks and at 12 weeks). Analyses will be further adjusted for severity of Alzheimer’s disease (mild or moderate, according to baseline MMSE score, as used in the stratified randomisation) and with indicator variables for the centre (if there is more than one full centre in the trial). All variables will be fitted as fixed effects initially. Time will be fitted as a random effects variable if the fit of the fixed effects model is poor (if p < 0.01 for improvement in fit from likelihood ratio model for primary outcome analysis, with the intention of reporting consistent models for all outcomes). For assessing the 4-week outcome, the indicator variable for time will be recoded as 1 for 12 weeks and 0 for 4 weeks. Log transformations will be applied, where appropriate, in order to render the outcome distributions closer to the normal. Boot-strap techniques will be undertaken to calculate the confidence intervals if neither the variable nor its log results in residuals from the regression analysis close to the normal distribution.

We will use multiple imputation to take account of any missing baseline data and follow-up data (when 4-week and 12-week data are both missing). Imputed values will be based on ADAS-Cog at baseline, on treatment group, and measurements within the same family (at available time points). MMSE and ADAS-COG are both in the “measuring severity of dementia” family, measures of attention form the “attention” family and measures of carer burden form another family, and measures of patient competency and behaviour form the final family. If fewer than 20% of patients have missing outcome data at either baseline or at both 4 weeks and 12 weeks follow-up (or fewer than 10% for ADAS-Cog and for MMSE), then reported results will be based on complete case analysis, on the basis that the multiple imputation results would be very similar in this situation.

A full, detailed statistical analysis plan has been written and signed. Deviations from this plan will be reported in the final analysis and report.

Patient demographics and other baseline information will be summarised by treatment group. Continuous variables following a normal distribution will be summarised using means (and standard deviation). Highly skewed variables will be summarised using medians (interquartile or full range). Categorical variables (binary and ordinal and multinomial) will be presented as frequencies and percentages. Descriptive analyses, including histograms, normal probability plots and box plots, will be used to assess the distributional assumptions and check for possible outliers.

Serious adverse events and any adverse events (occurring from 0 to 13 weeks) considered to be relevant will be presented. The number of participants who suffer at least one of each event type will be compared between the GXR group (those who received at least one dose) and those allocated to placebo (who did not receive any GXR) using Fisher’s exact test. The results will be interpreted taking account of multiple comparisons made.