From: Factors associated with brain ageing - a systematic review
Reference | Study (Design, country) | n, Mean age ± SD (Range), Sex, Other information | Modality (Protocol) | Features | Model (Cross-validation) | Exposure | Main findings outcome | Adjustments |
---|---|---|---|---|---|---|---|---|
[65] | Cognitively normal, younger/ amyloid negative adults from 6 studies (US) | ♀: n = 108; ♂: n = 76; Age unknown | PET (18F-FDG; 15O-O2, −CO, −H2O) | CMRGIc, CMRO2, CBF, AG | RF [10-fold] | Gender | When trained on ♂ only, GAP ↓ in ♀ (−3.8 yrs., p < 0.01). When trained on ♀ only, GAP ↑ in ♂ (+ 2.4 yrs., p < 0.04) | aAge |
[53] | UK Biobank | n = 19,000, 10,112♀; Age unknown | MRI (T1, rfMRI, tfMRI, T2 FLAIR, dMRI, swMRI [3 T]) | IDP | Non-LR [10-fold] | Gender; cognitive function; imaging features | GAP 0.7 yrs. ↑ in ♀ than ♂. GAP correlates ↓ psychomotor &/or executive function (DSST: r = − 0.06; reaction time: r = 0.05; TMT-B: r = 0.08), fluid intelligence (r = − 0.05), & visual memory (pairs matching: r = 0.07; all -log10P > 8). Correlates with GM & WB volume (r ~ − 0.59 to − 0.49), & microstructure (r ~ 0.25 to 0.41), with differences between sexes (no p-value) | aAge, age2, gender |
[19] | Lothian Birth Cohort 1936 (LBC1936) (Scotland) | n = 669, 72.7 ± 0.7 yrs., 317♀ | MRI (T1[3T]) | Voxel-wise WB volume | GPR [10-fold (×1000)] | Gender; MRI; epigenetic clock; TL | ♀ PAD (− 1.29 yrs) ↓ than ♂ (4.29 yrs., p < 0.001). NS association with epigenetic clock (ρ = − 0.007), or TL (r = 0.04). Correlates ↑ CSF, WMH & FA (r ~ 0.27 to 0.49); ↓ GM, WM, CT & MD (r ~ − 0.16 to − 0.47; no p-value) | None |
[31] | Dunedin Longitudinal Study (New Zealand) | n = 869, 45.2 ± 0.7 (43.5–47.0)yrs.; Cognitive data collected at 3, 7, 9, & 11 yrs. of age; Sex unknown | MRI (T1[3T]) | CT, SA & subcortical volume | Stacked RF (SVR) | Adult cognitive function & decline; ageing; brain health at age 3 | BA score associated with ↓ total & sub-domains of cognitive function (β ~ − 0.09 to − 0.20) & decline (β ~ − 0.07 to − 0.12; all p < 0.05). ↓ Brain health (β = − 0.12, p < 0.05). ↑ Ageing (pace: β = 0.22; facial: β = 0.15; both p < 0.05) | bGender |
[84] | Community dwelling adults recruited at university medical center (Germany) | n = 34, 68.8 ± 5.3 (61–80) yrs., 20♀ | MRI (T1[3T]) | Voxel-wise WB volume | RVR | Cognitive function | BranAGE NS correlated with working memory (r = 0.01, p = 0.98) | None |
[90] | Community dwelling adults from DEU [1], CR/RANN [2], & TILDA [3] studies (Turkey, unknown, Ireland) | 1) n = 175, 69.0 ± 8.6 (47.6–93.5) yrs., 104♀; 2) n = 380, 52.4 ± 17.1 (19–80) yrs., 210♀; 3) n = 470, 68.6 ± 7.2 (50–88) yrs., 260♀ | MRI (T1[1.5/3 T]) | Voxel-wise GM density | E-Net [Nested 10-fold] | Cognitive function | 1&2) GAP correlates ↓ general cognition (1: ρ = − 0.32; 2: ρ = − 0.14), semantic verbal fluency (1: ρ = − 0.25; 2: ρ = − 0.20) & executive function (TMT-B minus A: 1: ρ ~ 0.12 to 0.27; replicated p < 0.05). 1–3) ↓ psychomotor & executive function (TMT-B: ρ ~ 0.09 to 0.27; replicated p < 0.05) | aAge; bAge, gender |
[54] | UK Biobank | n = 14,701, 62.6 ± 7.5 yrs., 7914♀ | MRI (T1, rfMRI, tfMRI, T2 FLAIR, dMRI, swMRI [3 T]) | IDP | LASSO [10-fold] | Cognitive function | ↑ GAP associated with ↓ fluid intelligence (B = -0.15), psychomotor &/or executive functions (TMT-B: B = 0.002; tower rearranging: B = -0.12) & non-verbal fluid reasoning (matrix pattern completion: B = -0.22; all p < 0.001) | abAge; bage2, gender, height, volumetric scaling, & tfMRI head motion |
[32] | UK Biobank | Discovery: n = 12,378, 46-79 yrs.; Replication: n = 4456, 47-80 yrs.; Sex unknown | MRI (T1[3T]) | Voxel-based MNI, Jacobian map, GM and WM volume | CNN [Data splitting] | Cognitive function | GAP associated with ↓ psychomotor and/or executive functions (DSST: r = − 0.08; reaction time: r = 0.03; TMT-A, B, & minus A: r ~ 0.05 to 0.08; all p < 0.0056). Fluid intelligence, numeric/prospective/visual memory NS | aAge, age2, gender, TICV, 40 PCs, head motion, genotyping, study site |
[60] | Community dwelling adults from 1 of 6 studies (US) | n = 185, 64.9 ± 8.3 yrs., 91♀ | MRI (T1[3T]) | Voxel-wise WB volume | RVR | Cognitive function | ↑ BrainAGE correlates ↓ semantic verbal fluency (r = − 0.25, p = 0.006) | bAge, gender, diabetes duration |
[91] | Harvard Ageing Brain Study (US) | AA: n = 43, 62-88 yrs., 32♀; Matched NHW: n = 43, 65-90 yrs., 30♀; Unmatched NHW: n = 43, 64-86 yrs., 29♀ | MRI (T1[3T]) | CT (Racially different regions in high amyloid group) | SVR [LOO] | Race | GAP −1.05 yrs. for AA & matched NHW; or 0.92 yrs. when unmatched (no p-value). GAP remained when controlling for WMH (AA&Matched: − 0.88 yrs.; Unmatched: − 0.64 yrs.; no p-value) | aWith/without adjust WMH |