Standard protocol approvals, registrations, and subject consents

The trial was conducted at the University of Massachusetts Chan Medical School (UMASS Chan) Neurology dementia clinic in accordance with the principles of Good Clinical Practice guidelines and approved by the UMASS Chan institutional review board according to their ethical standards for human research. Written informed consent was provided by the subjects or their legal representatives. Data were collected and analyzed by the investigators. All the authors approved the manuscript, had full access to the trial data, and vouch for the accuracy and completeness of the data, for the fidelity of the trial to the protocol, and for the reporting of adverse events (except Dr. Drachman who passed away prior to the completion of the data analysis).

Subjects

Subjects were enrolled in the study between 1/1/2011 and 2/6/2016 and were eligible for enrollment in the study if they were between 55 and 85 years of age and if they had mild AD or mild cognitive impairment (MCI) according to the specifications by the National Institute of Neurological and Communicative Disorders and Stroke (NINCDS) and the Alzheimer’s Disease and Related Disorders Association (ADRDA) Workgroup [14]. Subjects could have been receiving an acetylcholinesterase inhibitor, memantine, or both, provided that they had received a stable dose for at least 3 months prior to study entry. Baseline MRI was performed prior to initiation of any study drugs. Subjects taking statin medication before study entry were allowed study entry after a washout period of at least 8 weeks before the baseline MRI.

Exclusion criteria were known allergy to any of the study drugs; significant psychiatric disorder; history of stroke; current use of any of the test medications (statin, THB, l-arginine); active malignancy; renal insufficiency (elevated creatinine above 1.3 mg/dL); other serious diseases including coronary insufficiency or congestive heart failure (ICD-9 criteria); known carotid stenosis, active peptic ulcer; urinary tract infection; and inability to come to the study site for follow-up.

Trial design

This was a single-center, single-arm, prospective proof-of-concept study that assessed adverse events and effects on CBF (primary aim) and cognition (secondary aim) of sequential treatment with simvastatin, l-arginine, and THB in subjects with AD or MCI. We chose these drugs based on their known interaction with the eNOS pathway in a potentially synergistic manner:

(1).

The HMG-CoA reductase synthesis inhibitor simvastatin has been shown to upregulate eNOS expression [15] as well as inhibit the Rho-kinase (ROCK) pathway, which leads to rapid phosphorylation and activation of eNOS via the phosphatidylinositol-3 kinase (PI3K)/protein kinase B (PKB/Akt) [16, 17]. This results in enhanced eNOS activity, which promotes nitric oxide (NO) production and subsequently improves cerebral perfusion [18,19,20,21].

(2).

l-Arginine, a semi-essential amino acid, is the substrate used by eNOS to produce NO in the vascular endothelium [22, 23]. Following simvastatin-induced eNOS upregulation, l-arginine amplifies and sustains cerebral hyperemia [21].

(3).

THB is an essential cofactor of the eNOS. Low bioavailability of THB leads to uncoupling of eNOS favoring the production of the superoxide oxide over NO. Conversely, supplementation of THB improves endothelium-dependent vasodilation and treatment with simvastatin elevates endothelial THB through inhibition of the ROCK pathway in vitro.

After informed consent, eligible subjects underwent formal history taking, physical and neurologic examination, psychometric assessment, blood work, and brain MRI (Table 1). Seven subjects also underwent lumbar puncture, and 1 subject underwent nuclear imaging to rule out other possible causes of dementia as part of their routine care. Although we included MCI as well as AD in our inclusion criteria, only 1 subject was diagnosed with MCI at the start of the study. This subject transitioned to AD early in the study according to the NINCDS and ADRDA Workgroup definition. Therefore, no attempt was undertaken to stratify the analyses in this study according to the diagnosis of MCI versus AD due to our small study size and lack of power for this type of analysis.

Table 1 Baseline demographics

After baseline assessment, subjects were sequentially treated once daily with the three study drugs. From weeks 0 to 16, subjects received oral simvastatin at a dose of 40 mg at bedtime; from weeks 4 to 16, subjects received oral l-arginine at a dose of 2 g three times a day and at bedtime; from weeks 8 to 16, subjects received oral THB at a dose of 20 mg/kg once a day. The study timeline is shown in Fig. 1. After completion of the 16-week study, subjects had the choice of continuing to take the three study medications or to taper and discontinue them over 8-day periods for each drug (THB first, l-arginine second, simvastatin third).

Fig. 1

Study design and timeline. A Study flowchart. B Study timeline depicting the timing of MRI and specific psychometric analysis relative to treatment initiation. 1Included Alzheimer’s Disease Assessment Scale-Cognitive 13 (ADAS-cog 13), Clinical Dementia Rating Scale (CDR), Cognitive Assessment Screening Test (CAST), and Mini-Mental State Exam (MMSE). 2Included Clinician Interview-Based Impression of Change plus caregiver input (CIBIC-plus) and MMSE. 3Included ADAS-cog 13, CDR, CIBIC-plus, and MMSE

Imaging protocol and image analysis

Dynamic susceptibility contrast (DSC)-MRI was performed on a Philips Achieva 3.0T/60-cm bore magnet (Philips Healthcare, Andover, MA, USA) scanner with gadolinium (0.1 mmol/kg) for all subjects to assess brain perfusion at 4 time points after initiation of the treatment regimen: baseline, 4 weeks, 8 weeks, and 16 weeks. The imaging protocol included DSC-MRI (TR/TE = 1700/40ms, FA = 75°, 100 dynamics, matrix = 128×128) and T1-MPRAGE (TR/TE = 7/3ms, FA = 8°, matrix = 256×256). In an exploratory post hoc analysis, the MRI perfusion parameters were analyzed and stratified by the degree of cognitive change in subjects as assessed on the ADAS-cog 13.

Volumetric brain analysis was performed at baseline using BrainSuite software [24]. Image analysis was performed using ImageJ (National Institutes of Health, Bethesda, MD, USA) and DSC-MRI toolbox in MATLAB (Mathworks, Natick, MA, USA) with semi-automated arterial input function selection and deconvolution algorithms [25, 26]. CBF, cerebral blood volume (CBV), and mean transit time (MTT) maps were generated for each subject at each time point. To ensure an objective comparison of CBF maps between time points and across all subjects, a relative CBF (rCBF) map was calculated by normalizing the CBF maps relative to the whole brain CBF. Similarly, the CBV and MTT maps were normalized to generate relative CBV (rCBV) and relative MTT (rMTT) maps. CBF analyses were focused on five regions of interest (ROIs, eFigure 1 in Supplement) that comprised the limbic system (hippocampus plus amygdala) and three cortical regions (middle temporal, middle frontal, and inferior parietal lobes). These regions were chosen because prior studies demonstrated reduced perfusion in these areas among subjects with MCI and AD [27, 28]. ROIs were manually placed on perfusion MRI images by a radiologist (Z.V.). Finally, whole brain white matter (WM), gray matter (GM), cerebrospinal fluid (CSF) volumes as well as regional volumes of the limbic system, and select cortical regions were calculated for all the subjects.

Psychometrics

Cognitive function was evaluated using a battery of psychometric measures including the Mini-Mental State Exam (MMSE), a 30-point scale of cognitive function where higher scores indicate better cognition [29]; the Cognitive Assessment Screening Test (CAST), a 40-point self-administered cognitive screen assessing memory, general intellect, visuospatial functioning, and mathematics where higher scores indicate better cognition [30]; the Clinical Dementia Rating Scale (CDR/sum of boxes), a semi-structured interview of the subject and their primary caregiver to rate impairment in six categories on a 0–3-point scale including memory, orientation, judgment, problem solving, community affairs, home and hobbies, and personal care, with higher scores showing more impairment [31]; and the Alzheimer’s Disease Assessment Scale-Cognitive (ADAS-cog) 13 [32, 33], to assess the severity of cognitive impairment in multiple cognitive domains, with higher scores showing more impairment. We used the Clinician Interview-Based Impression of Change plus caregiver input (CIBIC-plus) [34] to gauge a global impression of change from the caregiver, the subject, and the physician. To reduce variability, each subject was tested by the same trained research assistant.

Outcomes

The primary predetermined outcome of interest was the change in CBF from baseline to 16 weeks as measured by MRI.

There were five secondary predetermined outcomes: the change from baseline to week 16 in the MMSE, CDR, and ADAS-cog 13 and the change in CIBIC-plus scores between 4 and 16 weeks.

To gain deeper insight into the possible link between CBF augmentation and cognitive outcome measures in the entire cohort, we conducted exploratory post hoc analyses stratified according to the cognitive trajectories as assessed by the ADAS-cog 13. For this analysis, we chose the ADAS-cog 13 (scores range from 0 to 85, with higher scores indicating worse cognition) because it was specifically designed to assess the efficacy of treatments based on cognitive performance for AD patients [33, 35], used as a primary outcome of the landmark donepezil studies, and shown to be sensitive to short-term cognitive changes [36]. We stratified included subjects into three groups (eTable 1 in Supplement): improvement (decline in ADAS-cog 13 by at least one point between baseline and 16 weeks: group 1), stable (16-week ADAS-cog 13 remained within a half point from baseline, group 2), and deterioration (increase in ADAS-cog 13 by at least one point from baseline to 16 weeks: group 3).

Statistical analysis

Unless otherwise stated, continuous variables are reported as mean ± standard deviation or as median (interquartile range [IQR]). Normality of data was examined using the Shapiro–Wilk test. One-way repeated measures analysis of variance (ANOVA) tests were used to analyze the rCBF and psychometric data for statistical significance over time. rCBF differences between the three subject groups were examined by performing two-sample t-tests using the SPM12 software (Statistical Parametric Mapping version 12, Wellcome Department of Imaging Neuroscience, University College London). For the rCBF, rCBV, and rMTT values across the 4 time points for the 3 subgroups, ANOVA for mixed models was used to determine whether there was a significant change in the perfusion parameters across time and between the groups for the limbic system and cerebral cortex. Two-sided significance tests were used throughout and unless stated otherwise, a two-sided P<0.05 was considered statistically significant. All statistical analyses were performed using IBM® SPSS® Statistics Version 26 (IBM®-Armonk, NY) and GraphPad Prism (V9.0.2 for Windows, GraphPad Software, La Jolla, CA).

Protocol and statistical analysis plan

The study protocol and statistical analysis plan were published [37].