Article Figures & Data
Figures
- Figure 1. Schematic of our approach.
Age-sensitive transcripts are determined from a large training cohort of disease-free brains. These are used to create “molecular ages” and applied to four additional cohorts, one of which includes subjects with neurodegenerative diseases. The deviation of molecular age from chronological age (Δ age) is used to test associations of diseases and phenotypes with rates of aging of the brain.
- Figure 2. Molecular ages were significantly predictive of chronological ages in all cohorts.
(A–H) P-values corresponding to the R-values depicted are (A) P = 8.3 × 10−49. (B) P = 2.1 × 10−6. (C) P = 3.2 × 10−20. (D) P = 1 × 10−10. (E) P = 1.2 × 10−9. (F) P = 1.3 × 10−18. (G) P = 9.8 × 10−10. (H) P = 1.7 × 10−19. R-values were determined by Pearson correlation. The age-sensitive transcripts used to predict molecular ages are listed in Table S2. Note that the PE cohorts were analyzed as two separate cohorts because of confounding variables from different collections (see the Materials and Methods section).
- Figure S1. Delta Age predicts age at death in subjects over 60 yr of age
(A, B) Delta age versus chronological age in old and young subjects in CM.
- Figure S2. Sex differences in delta age in CM.
- Figure 3. Methylation ages and Δages in BC and ROS-MAP
(A, B) Methylation ages are predictive of chronological ages in BrainCloud (A) and ROS-MAP (B). (C, D) Methylation and molecular delta ages correlate in BrainCloud (C) and ROS-MAP (D). R values were determined by Pearson correlation.
- Figure 4. Characterization of age-sensitive transcripts and molecular ages.
(A, B) The 537 increasing and 834 decreasing age-sensitive transcripts are visualized in the heat map (A). Top Ingenuity functional categories are shown for increasing or decreasing transcripts (A). (B, C) Venn diagrams show the intersection of age down-regulated transcripts and neuronal-specific transcripts (B) and age up-regulated transcripts and glial-specific transcripts (C).
- Figure 5. Relationship of Δ age to clinical variables.
(A–I) Representative plots are shown using raw data (A–I). P values were determined by linear regression with relevant covariates. (A) Postmortem final clinical diagnosis of AD, (B) mini mental examination score, (C) tangles density, (D) Global PD score, a composite score for 4 signs: tremor, rigidity, bradykinesia, and gait, (E) rigidity score, (F) PD pathology is present if Lewy bodies are present and there was moderate-to-severe neuronal loss in the substantia nigra, (G) Global cognition slope, a composite slope of the longitudinal changes over time in five domains of cognition: working memory, visual-spatial ability, perceptual speed, episodic memory, and semantic memory, (H) change in episodic memory over time, and (I) association of Δ age & APOE ε4.
- Figure 6. APOE ε4 and Δage are synergistic risk factors for AD
(A) Odds ratio of Δ age and APOE ε4 for having AD diagnosis. (B) Model showing synergistic effects of Δ age and APOE ε4 on AD.
Tables
Cohort No. subjects No. male No. Caucasians No. African American No. Hispanic No. Asian Mean age Mean PMI Mean RIN Mean pH Mean Education CM 239 147 179 39 17 3 63 yr 15 h 7.7 6.7 N/A BrainCloud 127 85 58 69 4 3 48 yr 35 h 8.1 N/A 10 yr PE 216 145 211 0 2 2 70 yr 11 h N/A 6.5 N/A GTEx 87 62 76 11 0 0 58 yr 14 h 7.4 N/A N/A ROS-MAP 438 163 438 0 0 0 89 yr 7 h 7.2 N/A 17 yr - Table 2.
False discovery corrected P-values obtained from regression of either Δ age (main method) or APOE ε4 with indicated disease or aging variables of interest.
Variable Δ Age APOE ε4 P-value P-value ΔGlobal cognition/yr 5.1 × 10−5a 1.9 × 10−5a ΔEpisodic memory/yr 2.9 × 10−4a 2.5 × 10−6a ΔVisual-spatial ability/yr 0.003a 0.02a ΔPerceptual speed/yr 8.7 × 10−5a 0.02a ΔSemantic memory/yr 5.1 × 10−5a 1.6 × 10−5a ΔWorking memory/yr 0.03a 0.006a Global cognition level 3.2 × 10−5a 2.4 × 10−6a Episodic memory level 9 × 10−5a 2.2 × 10−6a Dementia grade 0.006a 7.6 × 10−5a AD clinical diagnosis 3.2 × 10−5a 7.9 × 10−4a Mini mental examination score 0.0017a 1.7 × 10−4a Depression score 0.69 0.85 General pathology 0.035a 5.2 × 10−10a Plaque level 0.06 8.1 × 10−8a Tangles level 0.04a 5.4 × 10−7a Amyloid level 0.15 5.4 × 10−7a Amyloid angiopathy 0.07a 9.4 × 10−7a PD diagnosis 0.88 0.28 PD sign score 0.038a 0.02a Gait 0.031a 0.02a Bradykinesia 0.41 0.02a Rigidity 0.0059a 0.035a Tremor 0.86 0.17 PD Pathology 0.006a 0.88 Lewy body pathology 0.02a 0.69 Stroke diagnosis 0.56 0.40 “Heart problem” history 0.17 0.90 Hypertension at baseline 0.003b 0.67 Arteriolar sclerosis 0.75 0.84 Cerebral infarction gross 0.47 0.40 Cerebral infarction micro 0.33 0.58 Cancer history 0.91 0.95 Thyroid disease history 0.08 0.60 Smoking (lifetime pack-years) 0.75 0.46 APO ε4 alleles 0.035a 0 Δ Age 0 0.035a
