St George's COPD and CAD brain cohort

This cohort comes from St George's, University of London, where patients with chronic obstructive pulmonary disease (COPD) and coronary artery disease (CAD) had multi-modal brain MRI, retinal photography, lung function tests, vascular measurements and blood tests at one study visit. The aim was to work out whether brain changes in these smoking-related diseases track respiratory, vascular or inflammatory factors. Spilling et al. (PLoS ONE, 2021) report the main analysis. Two figshare records share spreadsheets of derived values for the same 100 patients under CC BY 4.0; the images themselves are not released.

Composition

Patients were recruited from respiratory and cardiology outpatient clinics of St George's Hospital NHS Foundation Trust between December 2015 and October 2017 through two complementary studies (NoVasC2 and NoVasC3). Inclusion required age over 40 and at least 10 pack-years of smoking. Of 103 enrolled, 100 had MRI: 53 with COPD only, 24 with CAD only and 23 with both. The cohort is 69% male with a mean age of 68 years (SD 8).

Acquisition

All brain MRI was done on a 3 T Philips Achieva dual TX: sagittal 3D T1-weighted fast field echo (1 x 1 x 1.25 mm), axial FLAIR (1 x 1 x 3 mm), diffusion tensor imaging with 32 directions at b = 1000 s/mm2 acquired twice (2 mm isotropic), and pseudo-continuous arterial spin labelling with a proton density reference image. Retinal fundus photographs were taken with a Topcon 50IX camera after pupil dilation; images of 16 patients could not be graded.

Annotations

No image annotations are shared. The spreadsheets hold per-patient grey matter, white matter, CSF and ventricle volumes, white matter hyperintensity volume, fractional anisotropy and mean diffusivity histogram measures, grey matter blood flow, retinal vessel calibre, tortuosity, branching angle and fractal dimension, plus clinical, blood and (in the 2022 record) Montreal Cognitive Assessment scores.

Known limitations

• MR images and retinal photographs are not shared, and no request route is published.
• Age and sex were removed from the open spreadsheets.
• No control group without COPD or CAD.
• Per-sequence counts here come from non-missing derived values, so a missing value may reflect failed processing rather than a missing scan.