1. Membership in high-risk genetic groups predicts Alzheimer’s disease and age-at-onset Elizabeth CorderDuke University Shirley PodusloMedical College of Georgia

4. Background Some degree of AD brain changes (plaques and tangles) is almost universal by age 80 Extended longevity implies a strong! need to identify root causes and interventions I believe that risk pertains to many genes that have biologic plausibility but have been difficult to verify from sample to sample due to wide variation in frequencies of high-risk combinations

5. Goal Use GoM to define multilocus genotypes at high and low risk for AD Demonstrate that persons with high resemblance to high-risk ‘pure types’ are affected while those with low membership are OK

6. Grade of Membership analysis (Woodbury et al., 1978) Lambda coefficient ( ): probability that a specific variable outcome is associated with a particular pure type Grade of membership coefficients ( g ik ): estimate the degree to which a subject belongs to a pure type P ij   g ik kj k Internal variables and external (validating) variables The number of pure types that provide the best partition of the data matrix is determined by log likelihood tests

7. Data Age/ AD status APOE genotype Genotypes for plausible candidate loci: –APOE promoter polymorphisms at –491 and –427 –Adjacent gene APOCI –LDL receptor for APOE –Cystatin C –Cathepsin D

8. Table 1. Probabilities representing GoM groups I to V.* Attribute I II III IV V H AD case 100 100 100 0 0 0.68 Age (years) < 65 31 17 12 0 0 0.90 < 65 31 17 12 0 0 0.90 65- 69 30 0 0 0 65- 69 30 0 0 0 70 0 45 41 41 0 70 0 45 41 41 0 75 0 0 36 59 0 75 0 0 36 59 0 80+ 0 0 0 0 100 80+ 0 0 0 0 100

9. Group V: Long life without AD Permissive promotion of the APOE gene Several genotypes:  23,  33 and even  34! Heterozygosity for the LDL receptor for APOE

10. Table 1.cont I II III IV V H I II III IV V HAPOE  23 40 0 4 0 48 1.17  23 40 0 4 0 48 1.17  33 0 0 0 100 24  33 0 0 0 100 24  24 47 7 0 0 0  24 47 7 0 0 0  34 0 0 96 0 28  34 0 0 96 0 28  44 19 93 0 0 0  44 19 93 0 0 0 APOE – 491 AA 0 100 100 100 0 0.90 AA 0 100 100 100 0 0.90 AT 0 0 0 0 100 AT 0 0 0 0 100 TT 100 0 0 0 0 TT 100 0 0 0 0 APOE-427 TT 0 100 99 100 74 0.30 TT 0 100 99 100 74 0.30 TC 100 0 0 0 25 TC 100 0 0 0 25 CC 0 0 1 0 1 CC 0 0 1 0 1 APOCI AA n/a 100 0 0 0 0.91 AA n/a 100 0 0 0 0.91 AB n/a 0 100 0 100 AB n/a 0 100 0 100 BB n/a 0 0 100 0 BB n/a 0 0 100 0 I II III IV V H I II III IV V HLDLr8 GG 100 100 100 99 0 0.40 GG 100 100 100 99 0 0.40 AG 0 0 0 0 100 AG 0 0 0 0 100 AA 0 0 0 1 0 AA 0 0 0 1 0 LDLr13 TT 0 100 0 0 0 0.82 TT 0 100 0 0 0 0.82 TC 0 0 100 53 100 TC 0 0 100 53 100 CC 100 0 0 47 0 CC 100 0 0 47 0 CST3 GG 0 90 84 100 69 0.52 GG 0 90 84 100 69 0.52 GA 100 0 0 0 0 GA 100 0 0 0 0 AA 0 10 16 0 31 AA 0 10 16 0 31 CTSD CC 0 100 100 100 100 0.41 CC 0 100 100 100 100 0.41 CT 100 0 0 0 0 CT 100 0 0 0 0

11. Group I: Affected before age 70 Ultra-high expression of APOE High-risk homozygous combinations of APOE & LDL receptor genotypes Rare cathepsin D + cystatin C genotypes => that slow rate of amyloid degradation

12. Group II: Affected before age 75 High-risk APOE44 in combination with an alternate homozygous LDL exon 13 receptor genotypes, I.e. several high-risk APOE-LDL combinations

13. Group III: Affected before age 80 Common garden variety APOE34 Unaffected group IV of similar age carried APOE33 not APOE34

14. Table 2. AD status according to membership in the high- risk groups (I+II+III) AD Membership 0-20% 20-40%40-60% 60-80% 80-100% YES (n=180) 0(0%) 11(31%) 43(61%) 24(57%) 102(100%) NO (n=120) 50(100%) 24(69%) 28(39%) 18(43%) 0(0%) 50 35 71 42102

15. Conclusions Identification of high-risk combinations of gene variants jointly with the resemblance of study subjects to the to combinations may prove to be useful: –To predict the level of risk and likely age at onset of AD for individuals –Robust verification of candidate risk genes (the frequency of high-risk persons may vary from sample to sample while the risk groups rooted in biology are stable)