Foundational Study protocols and methods papers
ONDRI is a pan-Ontario research consortium that leads in the integration of a wide range of experimental, clinical, imaging and epidemiological expertise specifically to address the occurrence of degenerative cognitive impairment in the aging population (Farhan et al., 2017).
A founding objective of ONDRI is to understand the contribution of small vessel disease changes to disease presentation/cognition in aging and neurodegenerative diseases.
The ONDRI Foundational Study data represent a 5-year observational cohort study consisting of an enrollment period followed by up to 3 years of followup. Five hundred and twenty (520) participants who had one of the following diseases at baseline: Cerebrovascular disease (161), Parkinson disease (140), Alzheimer’s disease or mild cognitive impairment (126; 85 MCI, 41 AD), Frontotemporal dementia (53) or Amyotrophic lateral sclerosis (40), were enrolled into this study from fourteen (14) centres throughout Ontario. Each participant was matched with a care partner who was studied following a separate protocol.
Collected data are stored, managed, and shared on the Ontario Brain Institute’s secure Brain-CODE neuroinformatics platform to facilitate collaborative research across brain disorders and across multiple assessment modalities (Vaccarino et al., 2018).
Study participants underwent rigorous evaluations at screening and baseline. Assessment modality platforms include clinical, gait & balance, eye tracking, ocular imaging, neuropsychology, genetic markers and neuroimaging. These were repeated on an annual basis with the exception of genomics.
• Clinical data were collected using standardized assessments specific to each disease and Brain-CODE common data elements which capture a range of patient-reported characteristics including demographics, medical history, quality of life, depression, anxiety, and sleep quality.
• Eye movement was recorded while participants performed interleaved pro- and anti-saccade tasks and was characterized by reaction time and latency-specific saccade counts (Coe et al., 2017).
• Gait was assessed using accelerometry and electronic pressure-sensitive walkways under different dual-task conditions with varying levels of cognitive complexity. Balance was assessed using Wii boards adapted to record center of pressure displacement under different postural and visual conditions (Montero-Odasso et al., 2017).
• Genetic markers were collected through a validated targeted next-generation sequencing panel and curated through a resequencing, variant calling, and variant annotation workflow (Dilliott et al., 2018). Future releases will include annotated variant data for ONDRISeq panel (Farhan, Dilliott, et al., 2016) and NeuroX array (Ghani et al., 2014).
• Ocular images were acquired using traditional fundus photography as well as spectral domain optical coherence tomography. Retinal thickness measures were derived using automated software. In identification of retinal boundary lines, in the absence of obvious large errors resulting from poor image acquisition or retinal pathologies, the autosegmentation software showed excellent agreement with manually segmented images compared to manually corrected segmentation images by trained observers (Wong et al., 2019).
• Study coordinators were trained in the administration, scoring, and data entry for the assessment of psychological, behavioural, functional, and social characteristics of the participants. These data were double-scored; half the data were additionally reviewed by a study monitor, and a semi-automated pipeline was developed to flag missing values, out-of-range values, scoring errors, or other anomalies which triggered quality control queries to verify electronic data against source documents (McLaughlin et al., 2020).
• Using 3T MRI systems and following the harmonized Canadian dementia imaging protocol (Duchesne et al., 2018), neuroanatomical, microstructural, and functional imaging data were collected. A semi-automated brain region segmentation pipeline generated regional tissue volumes and markers of small vessel disease (Ramirez et al., 2020). Automated pipelines were used to perform quality control and remove artefacts from diffusion tensor images prior to calculation of diffusivity measures in grey and white matter within segmented brain regions (Haddad et al., 2020). Functional data were preprocessed using the OPPNI pipeline (Churchill et al., 2012). Future external releases will include defaced T1 images (Theyers et al., 2021) and functional data.
• ONDRI data have been cleaned, curated, and packaged into a standardized format to allow maximum utility of within- and cross-platform analysis. All tabular ONDRI data were additionally examined using robust multivariate outlier detection procedures that identified multivariate anomalies which assessment platform experts examined for any indication of error (Sunderland et al., 2019).