Genetics

OR 3.2 (95% CI 2.8-3.8) vs e3/e3 (odds ratio (genotype vs e3/e3))

• Allele: e4
• Frequency: 0.138
• Cases: 5,930
• Controls: 8,607

Who was studied: Meta-analysis of 40 studies: 5,930 probable/definite AD cases and 8,607 non-demented controls (clinic, community and brain-bank sources); ORs adjusted for age and study

One copy of e4 (heterozygote) vs e3/e3, Caucasian clinic/autopsy-based series. APOE e4 is defined by rs429358-C together with rs7412-C. GnomAD genomes (Ensembl release 116) frequency of rs429358-C: African/African American 0.2157, non-Finnish European 0.1379, Admixed American 0.1133, South Asian 0.1046, East Asian 0.0977, Middle Eastern 0.0685. Rs7412-T (e2-defining): African 0.1045, non-Finnish European 0.0794, East Asian 0.0801, Admixed American 0.0419.

Source: Farrer et al. 1997, JAMA (APOE and Alzheimer Disease Meta Analysis Consortium) · 1997 · PubMed 9343467

OR 14.9 (95% CI 10.8-20.6) vs e3/e3 (odds ratio (genotype vs e3/e3))

• Allele: e4
• Frequency: 0.138
• Cases: 5,930
• Controls: 8,607

Who was studied: Meta-analysis of 40 studies: 5,930 probable/definite AD cases and 8,607 non-demented controls (clinic, community and brain-bank sources); ORs adjusted for age and study

Two copies of e4 (homozygote) vs e3/e3, Caucasian clinic/autopsy-based series. Compare e3/e4 OR 3.2: the dose effect is strongly super-additive. APOE e4 is defined by rs429358-C together with rs7412-C. GnomAD genomes (Ensembl release 116) frequency of rs429358-C: African/African American 0.2157, non-Finnish European 0.1379, Admixed American 0.1133, South Asian 0.1046, East Asian 0.0977, Middle Eastern 0.0685. Rs7412-T (e2-defining): African 0.1045, non-Finnish European 0.0794, East Asian 0.0801, Admixed American 0.0419.

Source: Farrer et al. 1997, JAMA (APOE and Alzheimer Disease Meta Analysis Consortium) · 1997 · PubMed 9343467

OR 2.6 (95% CI 1.6-4.0) vs e3/e3 (odds ratio (genotype vs e3/e3))

• Allele: e4
• Frequency: 0.138
• Cases: 5,930
• Controls: 8,607

Who was studied: Meta-analysis of 40 studies: 5,930 probable/definite AD cases and 8,607 non-demented controls (clinic, community and brain-bank sources); ORs adjusted for age and study

E2/e4 vs e3/e3, Caucasian. APOE e4 is defined by rs429358-C together with rs7412-C. GnomAD genomes (Ensembl release 116) frequency of rs429358-C: African/African American 0.2157, non-Finnish European 0.1379, Admixed American 0.1133, South Asian 0.1046, East Asian 0.0977, Middle Eastern 0.0685. Rs7412-T (e2-defining): African 0.1045, non-Finnish European 0.0794, East Asian 0.0801, Admixed American 0.0419.

Source: Farrer et al. 1997, JAMA (APOE and Alzheimer Disease Meta Analysis Consortium) · 1997 · PubMed 9343467

OR 0.6 (95% CI 0.5-0.8) vs e3/e3 (odds ratio (genotype vs e3/e3))

• Allele: e2
• Frequency: 0.079
• Cases: 5,930
• Controls: 8,607

Who was studied: Meta-analysis of 40 studies: 5,930 probable/definite AD cases and 8,607 non-demented controls (clinic, community and brain-bank sources); ORs adjusted for age and study

E2/e3 vs e3/e3, Caucasian. Abstract: 'The e2/e3 genotype appears equally protective across ethnic groups.' APOE e4 is defined by rs429358-C together with rs7412-C. GnomAD genomes (Ensembl release 116) frequency of rs429358-C: African/African American 0.2157, non-Finnish European 0.1379, Admixed American 0.1133, South Asian 0.1046, East Asian 0.0977, Middle Eastern 0.0685. Rs7412-T (e2-defining): African 0.1045, non-Finnish European 0.0794, East Asian 0.0801, Admixed American 0.0419.

Source: Farrer et al. 1997, JAMA (APOE and Alzheimer Disease Meta Analysis Consortium) · 1997 · PubMed 9343467

OR 0.6 (95% CI 0.2-2.0) vs e3/e3 (odds ratio (genotype vs e3/e3))

• Allele: e2
• Frequency: 0.079
• Cases: 5,930
• Controls: 8,607

Who was studied: Meta-analysis of 40 studies: 5,930 probable/definite AD cases and 8,607 non-demented controls (clinic, community and brain-bank sources); ORs adjusted for age and study

E2/e2 vs e3/e3, Caucasian; CI crosses 1 because e2 homozygotes are <1% of the population (see Reiman 2020 for a powered estimate). APOE e4 is defined by rs429358-C together with rs7412-C. GnomAD genomes (Ensembl release 116) frequency of rs429358-C: African/African American 0.2157, non-Finnish European 0.1379, Admixed American 0.1133, South Asian 0.1046, East Asian 0.0977, Middle Eastern 0.0685. Rs7412-T (e2-defining): African 0.1045, non-Finnish European 0.0794, East Asian 0.0801, Admixed American 0.0419.

Source: Farrer et al. 1997, JAMA (APOE and Alzheimer Disease Meta Analysis Consortium) · 1997 · PubMed 9343467

OR 3.46 (95% CI 3.27-3.65) vs APOE*33 (odds ratio (genotype or allele dosage vs APOE*33))

• Allele: e4
• Frequency: 0.138

Who was studied: Non-Hispanic White individuals (n = 34,021; 59.3% female; mean age 77.0 y); case-control logistic regression, reference APOE*33

Read from the paper's main Table (Europe PMC full text). Case/control split per group is not given in the abstract or table, so the case and control counts are left empty; group n is in 'population'. Abstract: 'stepwise pattern' of APOE*34 ORs attenuating East Asian > White > Black > Hispanic; APOE*22+23 not associated with AD in East Asian or Hispanic individuals. Within Black individuals, more European global ancestry was associated with increasing APOE*4 risk. These ancestry differences are directly relevant to genetic-risk transferability and disparities work.

Source: Belloy et al. 2023, JAMA Neurol (ADSP/ADGC/A4/others; 68,756 individuals) · 2023 · PubMed 37930705

OR 13.04 (95% CI 11.31-15.04) vs APOE*33 (odds ratio (genotype or allele dosage vs APOE*33))

• Allele: e4
• Frequency: 0.138

Who was studied: Non-Hispanic White individuals (n = 34,021; 59.3% female; mean age 77.0 y); case-control logistic regression, reference APOE*33

Read from the paper's main Table (Europe PMC full text). Case/control split per group is not given in the abstract or table, so the case and control counts are left empty; group n is in 'population'. Abstract: 'stepwise pattern' of APOE*34 ORs attenuating East Asian > White > Black > Hispanic; APOE*22+23 not associated with AD in East Asian or Hispanic individuals. Within Black individuals, more European global ancestry was associated with increasing APOE*4 risk. These ancestry differences are directly relevant to genetic-risk transferability and disparities work.

Source: Belloy et al. 2023, JAMA Neurol (ADSP/ADGC/A4/others; 68,756 individuals) · 2023 · PubMed 37930705

OR 2.23 (95% CI 1.90-2.62) vs APOE*33 (odds ratio (genotype or allele dosage vs APOE*33))

• Allele: e4
• Frequency: 0.138

Who was studied: Non-Hispanic White individuals (n = 34,021; 59.3% female; mean age 77.0 y); case-control logistic regression, reference APOE*33

Read from the paper's main Table (Europe PMC full text). Case/control split per group is not given in the abstract or table, so the case and control counts are left empty; group n is in 'population'. Abstract: 'stepwise pattern' of APOE*34 ORs attenuating East Asian > White > Black > Hispanic; APOE*22+23 not associated with AD in East Asian or Hispanic individuals. Within Black individuals, more European global ancestry was associated with increasing APOE*4 risk. These ancestry differences are directly relevant to genetic-risk transferability and disparities work.

Source: Belloy et al. 2023, JAMA Neurol (ADSP/ADGC/A4/others; 68,756 individuals) · 2023 · PubMed 37930705

OR 0.53 (95% CI 0.48-0.59) vs APOE*33 (odds ratio (genotype or allele dosage vs APOE*33))

• Allele: e2
• Frequency: 0.079

Who was studied: Non-Hispanic White individuals (n = 34,021; 59.3% female; mean age 77.0 y); case-control logistic regression, reference APOE*33

Read from the paper's main Table (Europe PMC full text). Case/control split per group is not given in the abstract or table, so the case and control counts are left empty; group n is in 'population'. Abstract: 'stepwise pattern' of APOE*34 ORs attenuating East Asian > White > Black > Hispanic; APOE*22+23 not associated with AD in East Asian or Hispanic individuals. Within Black individuals, more European global ancestry was associated with increasing APOE*4 risk. These ancestry differences are directly relevant to genetic-risk transferability and disparities work.

Source: Belloy et al. 2023, JAMA Neurol (ADSP/ADGC/A4/others; 68,756 individuals) · 2023 · PubMed 37930705

OR 0.39 (95% CI 0.25-0.61) vs APOE*33 (odds ratio (genotype or allele dosage vs APOE*33))

• Allele: e2
• Frequency: 0.079

Who was studied: Non-Hispanic White individuals (n = 34,021; 59.3% female; mean age 77.0 y); case-control logistic regression, reference APOE*33

Read from the paper's main Table (Europe PMC full text). Case/control split per group is not given in the abstract or table, so the case and control counts are left empty; group n is in 'population'. Abstract: 'stepwise pattern' of APOE*34 ORs attenuating East Asian > White > Black > Hispanic; APOE*22+23 not associated with AD in East Asian or Hispanic individuals. Within Black individuals, more European global ancestry was associated with increasing APOE*4 risk. These ancestry differences are directly relevant to genetic-risk transferability and disparities work.

Source: Belloy et al. 2023, JAMA Neurol (ADSP/ADGC/A4/others; 68,756 individuals) · 2023 · PubMed 37930705

OR 3.48 (95% CI 3.32-3.64) vs APOE*33 (odds ratio (genotype or allele dosage vs APOE*33))

• Allele: e4
• Frequency: 0.138

Who was studied: Non-Hispanic White individuals (n = 34,021; 59.3% female; mean age 77.0 y); case-control logistic regression, reference APOE*33

Read from the paper's main Table (Europe PMC full text). Case/control split per group is not given in the abstract or table, so the case and control counts are left empty; group n is in 'population'. Abstract: 'stepwise pattern' of APOE*34 ORs attenuating East Asian > White > Black > Hispanic; APOE*22+23 not associated with AD in East Asian or Hispanic individuals. Within Black individuals, more European global ancestry was associated with increasing APOE*4 risk. These ancestry differences are directly relevant to genetic-risk transferability and disparities work.

Source: Belloy et al. 2023, JAMA Neurol (ADSP/ADGC/A4/others; 68,756 individuals) · 2023 · PubMed 37930705

OR 0.55 (95% CI 0.50-0.60) vs APOE*33 (odds ratio (genotype or allele dosage vs APOE*33))

• Allele: e2
• Frequency: 0.079

Who was studied: Non-Hispanic White individuals (n = 34,021; 59.3% female; mean age 77.0 y); case-control logistic regression, reference APOE*33

Read from the paper's main Table (Europe PMC full text). Case/control split per group is not given in the abstract or table, so the case and control counts are left empty; group n is in 'population'. Abstract: 'stepwise pattern' of APOE*34 ORs attenuating East Asian > White > Black > Hispanic; APOE*22+23 not associated with AD in East Asian or Hispanic individuals. Within Black individuals, more European global ancestry was associated with increasing APOE*4 risk. These ancestry differences are directly relevant to genetic-risk transferability and disparities work.

Source: Belloy et al. 2023, JAMA Neurol (ADSP/ADGC/A4/others; 68,756 individuals) · 2023 · PubMed 37930705

OR 0.13 (95% CI 0.05-0.36) (odds ratio (genotype comparison as stated))

• Allele: e2
• Frequency: 0.079

Who was studied: More than 5,000 clinically and neuropathologically characterized AD dementia cases and controls (ADGC); ORs from logistic regression; predominantly European-ancestry ADGC autopsy series (ancestry composition not stated in abstract)

Text: 'APOE2 homozygotes ... Had ... An extremely low 0.13 OR (95% CI 0.05-0.36) compared to the most common APOE3/3 genotype'. In the clinical (unconfirmed) group the e2/e2 vs e3/e3 OR was 0.52 (0.30-0.90).

Source: Reiman et al. 2020, Nat Commun (ADGC neuropathologically confirmed cases and controls) · 2020 · PubMed 32015339

OR 0.34 (95% CI 0.12-0.95) (odds ratio (genotype comparison as stated))

• Allele: e2
• Frequency: 0.079

Who was studied: More than 5,000 clinically and neuropathologically characterized AD dementia cases and controls (ADGC); ORs from logistic regression; predominantly European-ancestry ADGC autopsy series (ancestry composition not stated in abstract)

Table 2: e2 homozygotes have lower odds than e2/e3 heterozygotes, i.e. A dose effect for e2 as well.

Source: Reiman et al. 2020, Nat Commun (ADGC neuropathologically confirmed cases and controls) · 2020 · PubMed 32015339

OR 0.004 (95% CI 0.001-0.014) (odds ratio (genotype comparison as stated))

• Allele: e2
• Frequency: 0.079

Who was studied: More than 5,000 clinically and neuropathologically characterized AD dementia cases and controls (ADGC); ORs from logistic regression; predominantly European-ancestry ADGC autopsy series (ancestry composition not stated in abstract)

Table 2: the full span of APOE genotype risk in autopsy-confirmed material.

Source: Reiman et al. 2020, Nat Commun (ADGC neuropathologically confirmed cases and controls) · 2020 · PubMed 32015339

OR 15.92 (95% CI 11.85-21.38) (odds ratio (genotype comparison as stated))

• Allele: e4
• Frequency: 0.138

Who was studied: More than 5,000 clinically and neuropathologically characterized AD dementia cases and controls (ADGC); ORs from logistic regression; predominantly European-ancestry ADGC autopsy series (ancestry composition not stated in abstract)

Table 2 uses e2/e3 as reference; e3/e3 vs e2/e3 = 2.60 (2.00-3.38), e2/e4 = 6.96 (4.06-11.92).

Source: Reiman et al. 2020, Nat Commun (ADGC neuropathologically confirmed cases and controls) · 2020 · PubMed 32015339

OR 81.05 (95% CI 41.39-158.68) (odds ratio (genotype comparison as stated))

• Allele: e4
• Frequency: 0.138

Who was studied: More than 5,000 clinically and neuropathologically characterized AD dementia cases and controls (ADGC); ORs from logistic regression; predominantly European-ancestry ADGC autopsy series (ancestry composition not stated in abstract)

Table 2. Mean onset age in neuropath-confirmed cases: 69.9 +/- 6.1 y (e4/e4) to 79.3 +/- 9.0 y (e2/e3 + e2/e2).

Source: Reiman et al. 2020, Nat Commun (ADGC neuropathologically confirmed cases and controls) · 2020 · PubMed 32015339

OR 6.00 per e4 allele (neuropathologically confirmed group) (odds ratio (genotype comparison as stated))

• Allele: e4
• Frequency: 0.138

Who was studied: More than 5,000 clinically and neuropathologically characterized AD dementia cases and controls (ADGC); ORs from logistic regression; predominantly European-ancestry ADGC autopsy series (ancestry composition not stated in abstract)

Additive-model per-allele OR in the confirmed group was 6.00 vs 3.43 in the unconfirmed group and 3.61 combined; misclassification in clinical series dilutes APOE effects.

Source: Reiman et al. 2020, Nat Commun (ADGC neuropathologically confirmed cases and controls) · 2020 · PubMed 32015339

OR 0.38 per e2 allele (neuropathologically confirmed group) (odds ratio (genotype comparison as stated))

• Allele: e2
• Frequency: 0.079

Who was studied: More than 5,000 clinically and neuropathologically characterized AD dementia cases and controls (ADGC); ORs from logistic regression; predominantly European-ancestry ADGC autopsy series (ancestry composition not stated in abstract)

Confirmed 0.38, unconfirmed 0.64, combined 0.59.

Source: Reiman et al. 2020, Nat Commun (ADGC neuropathologically confirmed cases and controls) · 2020 · PubMed 32015339

Near-full penetrance of AD biology: almost all e4 homozygotes had AD pathology; by age 65 nearly all had abnormal CSF amyloid and 75% positive amyloid PET; symptom onset age 65.1 y (penetrance / biomarker positivity (descriptive))

• Allele: e4
• Frequency: 0.138

Who was studied: NACC neuropathology (n = 3,297) and five biomarker cohorts (n = 10,039); predominantly non-Hispanic White cohorts

Authors argue e4/e4 should be considered a distinct genetic form of AD, with onset predictability resembling autosomal-dominant AD and Down syndrome. Whether this holds outside European-ancestry cohorts is untested (see Belloy 2023 e4/e4 OR 6.49 in Black and 3.62 in Hispanic individuals).

Source: Fortea et al. 2024, Nat Med · 2024 · PubMed 38710950

OR 5.6 (95% CI 3.9-8.0) vs e3/e3 (odds ratio (genotype vs e3/e3))

• Allele: e4
• Frequency: 0.098
• Cases: 5,930
• Controls: 8,607

Who was studied: Meta-analysis of 40 studies: 5,930 probable/definite AD cases and 8,607 non-demented controls (clinic, community and brain-bank sources); ORs adjusted for age and study

Japanese subjects; abstract: 'The APOE e4-AD association in Japanese subjects was stronger than in Caucasian subjects.' Abstract also states the association was weaker among African Americans and Hispanics with significant heterogeneity among African American studies (P<.03); the numeric AA/Hispanic ORs are in the paywalled full text and were NOT loaded, so they are not recorded here.

Source: Farrer et al. 1997, JAMA (APOE and Alzheimer Disease Meta Analysis Consortium) · 1997 · PubMed 9343467

OR 33.1 (95% CI 13.6-80.5) vs e3/e3 (odds ratio (genotype vs e3/e3))

• Allele: e4
• Frequency: 0.098
• Cases: 5,930
• Controls: 8,607

Who was studied: Meta-analysis of 40 studies: 5,930 probable/definite AD cases and 8,607 non-demented controls (clinic, community and brain-bank sources); ORs adjusted for age and study

Japanese subjects; abstract: 'The APOE e4-AD association in Japanese subjects was stronger than in Caucasian subjects.' Abstract also states the association was weaker among African Americans and Hispanics with significant heterogeneity among African American studies (P<.03); the numeric AA/Hispanic ORs are in the paywalled full text and were NOT loaded, so they are not recorded here.

Source: Farrer et al. 1997, JAMA (APOE and Alzheimer Disease Meta Analysis Consortium) · 1997 · PubMed 9343467

OR 4.54 (95% CI 3.99-5.17) vs APOE*33 (odds ratio (genotype or allele dosage vs APOE*33))

• Allele: e4
• Frequency: 0.098

Who was studied: East Asian individuals (n = 21,852; summary statistics from published studies; demographic data not available); reference APOE*33

Read from the paper's main Table (Europe PMC full text). Case/control split per group is not given in the abstract or table, so the case and control counts are left empty; group n is in 'population'. Abstract: 'stepwise pattern' of APOE*34 ORs attenuating East Asian > White > Black > Hispanic; APOE*22+23 not associated with AD in East Asian or Hispanic individuals. Within Black individuals, more European global ancestry was associated with increasing APOE*4 risk. These ancestry differences are directly relevant to genetic-risk transferability and disparities work.

Source: Belloy et al. 2023, JAMA Neurol (ADSP/ADGC/A4/others; 68,756 individuals) · 2023 · PubMed 37930705

OR 0.97 (95% CI 0.77-1.23) vs APOE*33 (odds ratio (genotype or allele dosage vs APOE*33))

• Allele: e2
• Frequency: 0.08

Who was studied: East Asian individuals (n = 21,852; summary statistics from published studies; demographic data not available); reference APOE*33

Read from the paper's main Table (Europe PMC full text). Case/control split per group is not given in the abstract or table, so the case and control counts are left empty; group n is in 'population'. Abstract: 'stepwise pattern' of APOE*34 ORs attenuating East Asian > White > Black > Hispanic; APOE*22+23 not associated with AD in East Asian or Hispanic individuals. Within Black individuals, more European global ancestry was associated with increasing APOE*4 risk. These ancestry differences are directly relevant to genetic-risk transferability and disparities work.

Source: Belloy et al. 2023, JAMA Neurol (ADSP/ADGC/A4/others; 68,756 individuals) · 2023 · PubMed 37930705

OR 2.18 (95% CI 1.90-2.49) vs APOE*33 (odds ratio (genotype or allele dosage vs APOE*33))

• Allele: e4
• Frequency: 0.216

Who was studied: Non-Hispanic Black individuals (n = 7,145; 70.8% female; mean age 78.4 y); reference APOE*33

Read from the paper's main Table (Europe PMC full text). Case/control split per group is not given in the abstract or table, so the case and control counts are left empty; group n is in 'population'. Abstract: 'stepwise pattern' of APOE*34 ORs attenuating East Asian > White > Black > Hispanic; APOE*22+23 not associated with AD in East Asian or Hispanic individuals. Within Black individuals, more European global ancestry was associated with increasing APOE*4 risk. These ancestry differences are directly relevant to genetic-risk transferability and disparities work.

Source: Belloy et al. 2023, JAMA Neurol (ADSP/ADGC/A4/others; 68,756 individuals) · 2023 · PubMed 37930705

OR 6.49 (95% CI 5.07-8.31) vs APOE*33 (odds ratio (genotype or allele dosage vs APOE*33))

• Allele: e4
• Frequency: 0.216

Who was studied: Non-Hispanic Black individuals (n = 7,145; 70.8% female; mean age 78.4 y); reference APOE*33

Read from the paper's main Table (Europe PMC full text). Case/control split per group is not given in the abstract or table, so the case and control counts are left empty; group n is in 'population'. Abstract: 'stepwise pattern' of APOE*34 ORs attenuating East Asian > White > Black > Hispanic; APOE*22+23 not associated with AD in East Asian or Hispanic individuals. Within Black individuals, more European global ancestry was associated with increasing APOE*4 risk. These ancestry differences are directly relevant to genetic-risk transferability and disparities work.

Source: Belloy et al. 2023, JAMA Neurol (ADSP/ADGC/A4/others; 68,756 individuals) · 2023 · PubMed 37930705

OR 1.63 (95% CI 1.24-2.13) vs APOE*33 (odds ratio (genotype or allele dosage vs APOE*33))

• Allele: e4
• Frequency: 0.216

Who was studied: Non-Hispanic Black individuals (n = 7,145; 70.8% female; mean age 78.4 y); reference APOE*33

Read from the paper's main Table (Europe PMC full text). Case/control split per group is not given in the abstract or table, so the case and control counts are left empty; group n is in 'population'. Abstract: 'stepwise pattern' of APOE*34 ORs attenuating East Asian > White > Black > Hispanic; APOE*22+23 not associated with AD in East Asian or Hispanic individuals. Within Black individuals, more European global ancestry was associated with increasing APOE*4 risk. These ancestry differences are directly relevant to genetic-risk transferability and disparities work.

Source: Belloy et al. 2023, JAMA Neurol (ADSP/ADGC/A4/others; 68,756 individuals) · 2023 · PubMed 37930705

OR 0.69 (95% CI 0.56-0.84) vs APOE*33 (odds ratio (genotype or allele dosage vs APOE*33))

• Allele: e2
• Frequency: 0.104

Who was studied: Non-Hispanic Black individuals (n = 7,145; 70.8% female; mean age 78.4 y); reference APOE*33

Read from the paper's main Table (Europe PMC full text). Case/control split per group is not given in the abstract or table, so the case and control counts are left empty; group n is in 'population'. Abstract: 'stepwise pattern' of APOE*34 ORs attenuating East Asian > White > Black > Hispanic; APOE*22+23 not associated with AD in East Asian or Hispanic individuals. Within Black individuals, more European global ancestry was associated with increasing APOE*4 risk. These ancestry differences are directly relevant to genetic-risk transferability and disparities work.

Source: Belloy et al. 2023, JAMA Neurol (ADSP/ADGC/A4/others; 68,756 individuals) · 2023 · PubMed 37930705

OR 0.78 (95% CI 0.38-1.58) vs APOE*33 (odds ratio (genotype or allele dosage vs APOE*33))

• Allele: e2
• Frequency: 0.104

Who was studied: Non-Hispanic Black individuals (n = 7,145; 70.8% female; mean age 78.4 y); reference APOE*33

Read from the paper's main Table (Europe PMC full text). Case/control split per group is not given in the abstract or table, so the case and control counts are left empty; group n is in 'population'. Abstract: 'stepwise pattern' of APOE*34 ORs attenuating East Asian > White > Black > Hispanic; APOE*22+23 not associated with AD in East Asian or Hispanic individuals. Within Black individuals, more European global ancestry was associated with increasing APOE*4 risk. These ancestry differences are directly relevant to genetic-risk transferability and disparities work.

Source: Belloy et al. 2023, JAMA Neurol (ADSP/ADGC/A4/others; 68,756 individuals) · 2023 · PubMed 37930705

OR 2.39 (95% CI 2.16-2.65) vs APOE*33 (odds ratio (genotype or allele dosage vs APOE*33))

• Allele: e4
• Frequency: 0.216

Who was studied: Non-Hispanic Black individuals (n = 7,145; 70.8% female; mean age 78.4 y); reference APOE*33

Read from the paper's main Table (Europe PMC full text). Case/control split per group is not given in the abstract or table, so the case and control counts are left empty; group n is in 'population'. Abstract: 'stepwise pattern' of APOE*34 ORs attenuating East Asian > White > Black > Hispanic; APOE*22+23 not associated with AD in East Asian or Hispanic individuals. Within Black individuals, more European global ancestry was associated with increasing APOE*4 risk. These ancestry differences are directly relevant to genetic-risk transferability and disparities work.

Source: Belloy et al. 2023, JAMA Neurol (ADSP/ADGC/A4/others; 68,756 individuals) · 2023 · PubMed 37930705

OR 0.72 (95% CI 0.60-0.86) vs APOE*33 (odds ratio (genotype or allele dosage vs APOE*33))

• Allele: e2
• Frequency: 0.104

Who was studied: Non-Hispanic Black individuals (n = 7,145; 70.8% female; mean age 78.4 y); reference APOE*33

Read from the paper's main Table (Europe PMC full text). Case/control split per group is not given in the abstract or table, so the case and control counts are left empty; group n is in 'population'. Abstract: 'stepwise pattern' of APOE*34 ORs attenuating East Asian > White > Black > Hispanic; APOE*22+23 not associated with AD in East Asian or Hispanic individuals. Within Black individuals, more European global ancestry was associated with increasing APOE*4 risk. These ancestry differences are directly relevant to genetic-risk transferability and disparities work.

Source: Belloy et al. 2023, JAMA Neurol (ADSP/ADGC/A4/others; 68,756 individuals) · 2023 · PubMed 37930705

E4 x rs10423769-A interaction beta -0.54 (SE 0.12, p 7.5e-6); e4/e4 OR falls from 7.2 (no A allele) to 2.1 (>=1 A allele) (interaction beta; genotype-stratified OR)

• Allele: A (protective, interacts with e4)
• Frequency: 0.114
• Cases: 1,850
• Controls: 4,331

Who was studied: Discovery 1,850 AD / 4,331 cognitively intact African Americans; replication Ibadan (63/648) and Puerto Rican (273/275)

African-ancestry-specific locus ~2 Mb upstream of APOE that reduces the e4 risk; a candidate explanation for the attenuated e4 effect in African-ancestry populations. The A allele is essentially absent in Europeans (gnomAD nfe A = 0.0017), so it cannot be studied in European-only GWAS.

Source: Rajabli et al. 2022, PLoS Genet · 2022 · PubMed 35788729

OR 1.90 (95% CI 1.65-2.18) vs APOE*33 (odds ratio (genotype or allele dosage vs APOE*33))

• Allele: e4
• Frequency: 0.113

Who was studied: Hispanic individuals, all races (n = 5,738; 68.2% female; mean age 75.4 y); reference APOE*33

Read from the paper's main Table (Europe PMC full text). Case/control split per group is not given in the abstract or table, so the case and control counts are left empty; group n is in 'population'. Abstract: 'stepwise pattern' of APOE*34 ORs attenuating East Asian > White > Black > Hispanic; APOE*22+23 not associated with AD in East Asian or Hispanic individuals. Within Black individuals, more European global ancestry was associated with increasing APOE*4 risk. These ancestry differences are directly relevant to genetic-risk transferability and disparities work.

Source: Belloy et al. 2023, JAMA Neurol (ADSP/ADGC/A4/others; 68,756 individuals) · 2023 · PubMed 37930705

OR 3.62 (95% CI 2.56-5.11) vs APOE*33 (odds ratio (genotype or allele dosage vs APOE*33))

• Allele: e4
• Frequency: 0.113

Who was studied: Hispanic individuals, all races (n = 5,738; 68.2% female; mean age 75.4 y); reference APOE*33

Read from the paper's main Table (Europe PMC full text). Case/control split per group is not given in the abstract or table, so the case and control counts are left empty; group n is in 'population'. Abstract: 'stepwise pattern' of APOE*34 ORs attenuating East Asian > White > Black > Hispanic; APOE*22+23 not associated with AD in East Asian or Hispanic individuals. Within Black individuals, more European global ancestry was associated with increasing APOE*4 risk. These ancestry differences are directly relevant to genetic-risk transferability and disparities work.

Source: Belloy et al. 2023, JAMA Neurol (ADSP/ADGC/A4/others; 68,756 individuals) · 2023 · PubMed 37930705

OR 1.11 (95% CI 0.72-1.72) vs APOE*33 (odds ratio (genotype or allele dosage vs APOE*33))

• Allele: e4
• Frequency: 0.113

Who was studied: Hispanic individuals, all races (n = 5,738; 68.2% female; mean age 75.4 y); reference APOE*33

Read from the paper's main Table (Europe PMC full text). Case/control split per group is not given in the abstract or table, so the case and control counts are left empty; group n is in 'population'. Abstract: 'stepwise pattern' of APOE*34 ORs attenuating East Asian > White > Black > Hispanic; APOE*22+23 not associated with AD in East Asian or Hispanic individuals. Within Black individuals, more European global ancestry was associated with increasing APOE*4 risk. These ancestry differences are directly relevant to genetic-risk transferability and disparities work.

Source: Belloy et al. 2023, JAMA Neurol (ADSP/ADGC/A4/others; 68,756 individuals) · 2023 · PubMed 37930705

OR 0.89 (95% CI 0.71-1.10) vs APOE*33 (odds ratio (genotype or allele dosage vs APOE*33))

• Allele: e2
• Frequency: 0.042

Who was studied: Hispanic individuals, all races (n = 5,738; 68.2% female; mean age 75.4 y); reference APOE*33

Read from the paper's main Table (Europe PMC full text). Case/control split per group is not given in the abstract or table, so the case and control counts are left empty; group n is in 'population'. Abstract: 'stepwise pattern' of APOE*34 ORs attenuating East Asian > White > Black > Hispanic; APOE*22+23 not associated with AD in East Asian or Hispanic individuals. Within Black individuals, more European global ancestry was associated with increasing APOE*4 risk. These ancestry differences are directly relevant to genetic-risk transferability and disparities work.

Source: Belloy et al. 2023, JAMA Neurol (ADSP/ADGC/A4/others; 68,756 individuals) · 2023 · PubMed 37930705

OR 1.02 (95% CI 0.41-2.51) vs APOE*33 (odds ratio (genotype or allele dosage vs APOE*33))

• Allele: e2
• Frequency: 0.042

Who was studied: Hispanic individuals, all races (n = 5,738; 68.2% female; mean age 75.4 y); reference APOE*33

Read from the paper's main Table (Europe PMC full text). Case/control split per group is not given in the abstract or table, so the case and control counts are left empty; group n is in 'population'. Abstract: 'stepwise pattern' of APOE*34 ORs attenuating East Asian > White > Black > Hispanic; APOE*22+23 not associated with AD in East Asian or Hispanic individuals. Within Black individuals, more European global ancestry was associated with increasing APOE*4 risk. These ancestry differences are directly relevant to genetic-risk transferability and disparities work.

Source: Belloy et al. 2023, JAMA Neurol (ADSP/ADGC/A4/others; 68,756 individuals) · 2023 · PubMed 37930705

OR 1.90 (95% CI 1.70-2.13) vs APOE*33 (odds ratio (genotype or allele dosage vs APOE*33))

• Allele: e4
• Frequency: 0.113

Who was studied: Hispanic individuals, all races (n = 5,738; 68.2% female; mean age 75.4 y); reference APOE*33

Read from the paper's main Table (Europe PMC full text). Case/control split per group is not given in the abstract or table, so the case and control counts are left empty; group n is in 'population'. Abstract: 'stepwise pattern' of APOE*34 ORs attenuating East Asian > White > Black > Hispanic; APOE*22+23 not associated with AD in East Asian or Hispanic individuals. Within Black individuals, more European global ancestry was associated with increasing APOE*4 risk. These ancestry differences are directly relevant to genetic-risk transferability and disparities work.

Source: Belloy et al. 2023, JAMA Neurol (ADSP/ADGC/A4/others; 68,756 individuals) · 2023 · PubMed 37930705

OR 0.90 (95% CI 0.74-1.10) vs APOE*33 (odds ratio (genotype or allele dosage vs APOE*33))

• Allele: e2
• Frequency: 0.042

Who was studied: Hispanic individuals, all races (n = 5,738; 68.2% female; mean age 75.4 y); reference APOE*33

Read from the paper's main Table (Europe PMC full text). Case/control split per group is not given in the abstract or table, so the case and control counts are left empty; group n is in 'population'. Abstract: 'stepwise pattern' of APOE*34 ORs attenuating East Asian > White > Black > Hispanic; APOE*22+23 not associated with AD in East Asian or Hispanic individuals. Within Black individuals, more European global ancestry was associated with increasing APOE*4 risk. These ancestry differences are directly relevant to genetic-risk transferability and disparities work.

Source: Belloy et al. 2023, JAMA Neurol (ADSP/ADGC/A4/others; 68,756 individuals) · 2023 · PubMed 37930705

39% lower odds of AD with African-derived ancestry at APOE vs European-derived, after adjusting for APOE genotype, age and genome-wide ancestry (odds reduction (as stated in abstract))

Who was studied: 3,067 Caribbean Hispanics and 3,028 individuals of European descent

APOE e2 and e4 effects were significant but 'substantially attenuated' in Caribbean Hispanics vs European-ancestry individuals. The odds ratio left empty because the abstract reports a percentage reduction, not an OR.

Source: Blue et al. 2019, Alzheimers Dement · 2019 · PubMed 31606368

Single PSEN1 E280A carrier homozygous for APOE3ch did not develop MCI until her seventies, ~3 decades after expected onset, despite very high amyloid burden (case report (n = 1))

• Allele: R136S (protective)
• Cases: 1

Who was studied: One member of the Colombian (Antioquia) PSEN1 E280A kindred

Anecdotal-level evidence (n = 1) but mechanistically influential; do not use quantitatively.

Source: Arboleda-Velasquez et al. 2019, Nat Med · 2019 · PubMed 31686034

OR 1.13 (95% CI 1.11-1.15) per T allele (odds ratio per effect allele)

• Chr 1
• Risk allele: T
• OR 1.13 (1.11–1.15)
• p = 7e-46
• Cases: 85,934

First reported by Lambert 2009; Lambert 2013 lead rs6656401-A OR 1.18 (1.14-1.22). Complement receptor 1.

Source: Bellenguez et al. 2022, Nat Genet (EADB/IGAP/UKB stage 1) · 2022 · PubMed 35379992

OR 1.18 (95% CI 1.14-1.22) per A allele (odds ratio per effect allele)

• Chr 1
• Risk allele: A
• OR 1.18 (1.14–1.22)
• p = 5.999999999999999e-24
• Cases: 17,008

Lambert 2013 stage-1 estimate for the original CR1 lead SNP.

Source: Lambert et al. 2013, Nat Genet (IGAP stage 1) · 2013 · PubMed 24162737

OR 1.17 (95% CI 1.16-1.19) per T allele (odds ratio per effect allele)

• Chr 2
• Risk allele: T
• OR 1.17 (1.16–1.19)
• p = 5.9999999999999995e-118
• Cases: 85,934

Strongest non-APOE common locus (P 6e-118). Lambert 2013: OR 1.22 (1.18-1.25). GnomAD T allele: nfe 0.385, afr 0.396, eas 0.399.

Source: Bellenguez et al. 2022, Nat Genet (EADB/IGAP/UKB stage 1) · 2022 · PubMed 35379992

OR 0.91 (95% CI 0.90-0.92) per T allele (odds ratio per effect allele)

• Chr 8
• Risk allele: T
• OR 0.91 (0.9–0.92)
• p = 2e-44
• Cases: 85,934

Effect allele T is protective (OR 0.91). Lambert 2013 lead rs9331896 OR 1.16 for the C allele (catalog stores inverted 1.1628 for T). GnomAD rs9331896-C: nfe 0.403, afr 0.562, eas 0.225 -- the CLU risk allele is far commoner in African and rarer in East Asian populations.

Source: Bellenguez et al. 2022, Nat Genet (EADB/IGAP/UKB stage 1) · 2022 · PubMed 35379992

OR 0.90 (95% CI 0.89-0.92) per T allele (odds ratio per effect allele)

• Chr 11
• Risk allele: T
• OR 0.9 (0.89–0.92)
• p = 3e-48
• Cases: 85,934

Effect allele T protective (OR 0.90). Lambert 2013 lead rs10792832. GnomAD rs3851179-T: nfe 0.365, afr 0.155, eas 0.398.

Source: Bellenguez et al. 2022, Nat Genet (EADB/IGAP/UKB stage 1) · 2022 · PubMed 35379992

OR 0.84 (95% CI 0.81-0.87) per C allele (odds ratio per effect allele)

• Chr 11
• Risk allele: C
• OR 0.84 (0.81–0.87)
• p = 1e-21
• Cases: 85,934

Effect allele C protective (OR 0.84, freq 0.039). Miyashita 2013 first reached genome-wide significance for SORL1 by combining Japanese, Korean and Caucasian samples (P 1.77e-9); gnomAD rs11218343-C: eas 0.299, nfe 0.039, afr 0.080 -- much commoner in East Asians.

Source: Bellenguez et al. 2022, Nat Genet (EADB/IGAP/UKB stage 1) · 2022 · PubMed 35379992

OR 1.19 (95% CI 1.13-1.25) per G allele (odds ratio per effect allele)

• Chr 11
• Risk allele: G
• OR 1.19 (1.13–1.25)
• p = 3e-11
• Cases: 85,934

Second, rare SORL1 signal (OR 1.19, freq 0.0186). Rare SORL1 loss-of-function variants (not we hold) carry far larger effects.

Source: Bellenguez et al. 2022, Nat Genet (EADB/IGAP/UKB stage 1) · 2022 · PubMed 35379992

OR 1.10 (95% CI 1.09-1.12) per A allele (odds ratio per effect allele)

• Chr 19
• Risk allele: A
• OR 1.1 (1.09–1.12)
• p = 2e-37
• Cases: 85,934

European ABCA7 lead. In African Americans the lead is a different, higher-frequency variant rs115550680 with OR 1.79 (Reitz 2013), see separate record.

Source: Bellenguez et al. 2022, Nat Genet (EADB/IGAP/UKB stage 1) · 2022 · PubMed 35379992

OR 1.15 (95% CI 1.11-1.19) per A allele (odds ratio per effect allele)

• Chr 19
• Risk allele: A
• OR 1.15 (1.11–1.19)
• p = 1e-15
• Cases: 17,008

Lambert 2013 stage-1 ABCA7 lead. GnomAD rs4147929-A: nfe 0.172, afr 0.043, eas 0.335.

Source: Lambert et al. 2013, Nat Genet (IGAP stage 1) · 2013 · PubMed 24162737

OR 0.91 (95% CI 0.90-0.92) per A allele (odds ratio per effect allele)

• Chr 11
• Risk allele: A
• OR 0.91 (0.9–0.92)
• p = 4e-42
• Cases: 85,934

MS4A gene cluster (chr11q12); effect allele A protective (OR 0.91). Lambert 2013 lead rs983392 (MS4A6A) OR 1.11 for the A allele. GnomAD rs983392-G: nfe 0.411, afr 0.078, eas 0.057.

Source: Bellenguez et al. 2022, Nat Genet (EADB/IGAP/UKB stage 1) · 2022 · PubMed 35379992

OR 1.11 (95% CI 1.09-1.15) per A allele (odds ratio per effect allele)

• Chr 11
• Risk allele: A
• OR 1.111 (1.09–1.15)
• p = 6e-16
• Cases: 17,008

Lambert 2013 stage-1 MS4A6A lead (catalog stores OR 1.1111 for the A allele).

Source: Lambert et al. 2013, Nat Genet (IGAP stage 1) · 2013 · PubMed 24162737

OR 1.06 (95% CI 1.04-1.10) per C allele (odds ratio per effect allele)

• Chr 19
• Risk allele: C
• OR 1.06 (1.04–1.1)
• p = 3e-06
• Cases: 17,008

CD33: Lambert 2013 stage-1 P 3e-6 (not genome-wide significant in stage 1 alone; reached significance in combined stages in earlier IGAP/Naj-Hollingworth 2011 reports). Wightman 2021 rs1354106 P 2e-10 and Jansen 2019 rs3865444 P 6e-9 (z-score) support the locus; Bellenguez 2022 did not list a CD33 lead among catalog-curated associations. GnomAD rs3865444-A: nfe 0.316, afr 0.087, eas 0.188.

Source: Lambert et al. 2013, Nat Genet (IGAP stage 1) · 2013 · PubMed 24162737

OR 1.08 (95% CI 1.06-1.09) per T allele (odds ratio per effect allele)

• Chr 6
• Risk allele: T
• OR 1.08 (1.06–1.09)
• p = 8e-22
• Cases: 85,934

Known locus; also genome-wide significant in African-ancestry MVP (different lead rs7738720, Sherva 2023).

Source: Bellenguez et al. 2022, Nat Genet (EADB/IGAP/UKB stage 1) · 2022 · PubMed 35379992

OR 1.07 (95% CI 1.06-1.08) per T allele (odds ratio per effect allele)

• Chr 8
• Risk allele: T
• OR 1.07 (1.06–1.08)
• p = 4e-22
• Cases: 85,934

Source: Bellenguez et al. 2022, Nat Genet (EADB/IGAP/UKB stage 1) · 2022 · PubMed 35379992

OR 0.95 (95% CI 0.93-0.96) per A allele (odds ratio per effect allele)

• Chr 7
• Risk allele: A
• OR 0.95 (0.93–0.96)
• p = 3e-14
• Cases: 85,934

Effect allele A protective.

Source: Bellenguez et al. 2022, Nat Genet (EADB/IGAP/UKB stage 1) · 2022 · PubMed 35379992

OR 0.93 (95% CI 0.92-0.95) per G allele (odds ratio per effect allele)

• Chr 2
• Risk allele: G
• OR 0.93 (0.92–0.95)
• p = 4e-18
• Cases: 85,934

Effect allele G protective.

Source: Bellenguez et al. 2022, Nat Genet (EADB/IGAP/UKB stage 1) · 2022 · PubMed 35379992

OR 1.10 (95% CI 1.07-1.12) per G allele (odds ratio per effect allele)

• Chr 14
• Risk allele: G
• OR 1.1 (1.07–1.12)
• p = 8e-16
• Cases: 85,934

Source: Bellenguez et al. 2022, Nat Genet (EADB/IGAP/UKB stage 1) · 2022 · PubMed 35379992

OR 0.93 (95% CI 0.92-0.95) per A allele (odds ratio per effect allele)

• Chr 14
• Risk allele: A
• OR 0.93 (0.92–0.95)
• p = 4e-21
• Cases: 85,934

Effect allele A protective.

Source: Bellenguez et al. 2022, Nat Genet (EADB/IGAP/UKB stage 1) · 2022 · PubMed 35379992

OR 0.89 (95% CI 0.87-0.91) per G allele (odds ratio per effect allele)

• Chr 20
• Risk allele: G
• OR 0.89 (0.87–0.91)
• p = 4e-21
• Cases: 85,934

Effect allele G protective.

Source: Bellenguez et al. 2022, Nat Genet (EADB/IGAP/UKB stage 1) · 2022 · PubMed 35379992

OR 0.91 (95% CI 0.90-0.93) per G allele (odds ratio per effect allele)

• Chr 6
• Risk allele: G
• OR 0.91 (0.9–0.93)
• p = 7e-20
• Cases: 85,934

HLA class II; Kunkle 2019 fine-mapping implicated the HLA-DR15 haplotype.

Source: Bellenguez et al. 2022, Nat Genet (EADB/IGAP/UKB stage 1) · 2022 · PubMed 35379992

OR 1.11 (95% CI 1.09-1.13) per T allele (odds ratio per effect allele)

• Chr 15
• Risk allele: T
• OR 1.11 (1.09–1.13)
• p = 2e-25
• Cases: 85,934

Gamma-secretase component; missense variant.

Source: Bellenguez et al. 2022, Nat Genet (EADB/IGAP/UKB stage 1) · 2022 · PubMed 35379992

OR 1.06 (95% CI 1.04-1.07) per A allele (odds ratio per effect allele)

• Chr 11
• Risk allele: A
• OR 1.06 (1.04–1.07)
• p = 5e-14
• Cases: 85,934

PU.1 transcription factor (microglia).

Source: Bellenguez et al. 2022, Nat Genet (EADB/IGAP/UKB stage 1) · 2022 · PubMed 35379992

OR 1.08 (95% CI 1.05-1.10) per A allele (odds ratio per effect allele)

• Chr 15
• Risk allele: A
• OR 1.075 (1.05–1.1)
• p = 7.000000000000001e-09
• Cases: 21,982

New locus in Kunkle 2019 (OR curated as 1.0753 for the A allele, P 7e-9); alpha-secretase.

Source: Kunkle et al. 2019, Nat Genet (IGAP stage 1) · 2019 · PubMed 30820047

OR 0.95 (95% CI 0.94-0.97) per C allele (odds ratio per effect allele)

• Chr 21
• Risk allele: C
• OR 0.95 (0.94–0.97)
• p = 1e-12
• Cases: 85,934

Common protective variant at APP itself (OR 0.95); the same gene carries causal autosomal-dominant mutations (see causal records).

Source: Bellenguez et al. 2022, Nat Genet (EADB/IGAP/UKB stage 1) · 2022 · PubMed 35379992

OR 1.38 (95% CI 1.24-1.54) per C allele (odds ratio per effect allele)

• Chr 1
• Risk allele: C
• OR 1.38 (1.24–1.54)
• p = 8e-09
• Cases: 85,934

New in Bellenguez 2022; rare (freq 0.0038), OR 1.38 (1.24-1.54).

Source: Bellenguez et al. 2022, Nat Genet (EADB/IGAP/UKB stage 1) · 2022 · PubMed 35379992

OR 1.47 (95% CI 1.33-1.63) per C allele (odds ratio per effect allele)

• Chr 2
• Risk allele: C
• OR 1.47 (1.33–1.63)
• p = 2e-13
• Cases: 85,934

New in Bellenguez 2022; rare (freq 0.0051), OR 1.47 (1.33-1.63).

Source: Bellenguez et al. 2022, Nat Genet (EADB/IGAP/UKB stage 1) · 2022 · PubMed 35379992

OR 1.13 (95% CI 1.09-1.16) per A allele (odds ratio per effect allele)

• Chr 8
• Risk allele: A
• OR 1.13 (1.09–1.16)
• p = 2e-16
• Cases: 85,934

Missense; linear ubiquitin chain assembly complex / TNF-alpha pathway highlighted by Bellenguez 2022. Also genome-wide significant cross-ancestry in Rajabli 2025 with 13.7% of the sample size.

Source: Bellenguez et al. 2022, Nat Genet (EADB/IGAP/UKB stage 1) · 2022 · PubMed 35379992

OR 1.07 (95% CI 1.06-1.09) per T allele (odds ratio per effect allele)

• Chr 17
• Risk allele: T
• OR 1.07 (1.06–1.09)
• p = 2e-20
• Cases: 85,934

Progranulin 3'UTR variant shared with FTLD-TDP risk; note the Bellenguez phenotype includes proxy 'AD and related dementias'.

Source: Bellenguez et al. 2022, Nat Genet (EADB/IGAP/UKB stage 1) · 2022 · PubMed 35379992

OR 0.96 (95% CI 0.94-0.97) per A allele (odds ratio per effect allele)

• Chr 7
• Risk allele: A
• OR 0.96 (0.94–0.97)
• p = 4.9999999999999995e-11
• Cases: 85,934

FTLD-TDP risk gene; effect allele A protective. Shared with proxy-dementia phenotype.

Source: Bellenguez et al. 2022, Nat Genet (EADB/IGAP/UKB stage 1) · 2022 · PubMed 35379992

OR 1.06 (95% CI 1.04-1.08) per G allele (odds ratio per effect allele)

• Chr 9
• Risk allele: G
• OR 1.06 (1.04–1.08)
• p = 2e-09
• Cases: 85,934

New in Bellenguez 2022; cholesterol efflux.

Source: Bellenguez et al. 2022, Nat Genet (EADB/IGAP/UKB stage 1) · 2022 · PubMed 35379992

OR 0.95 (95% CI 0.94-0.96) per G allele (odds ratio per effect allele)

• Chr 4
• Risk allele: G
• OR 0.95 (0.94–0.96)
• p = 8e-12
• Cases: 85,934

New in Bellenguez 2022; effect allele G protective.

Source: Bellenguez et al. 2022, Nat Genet (EADB/IGAP/UKB stage 1) · 2022 · PubMed 35379992

OR 1.28 (95% CI 1.18-1.39) per T allele (odds ratio per effect allele)

• Chr 15
• Risk allele: T
• OR 1.28 (1.18–1.39)
• p = 1e-09
• Cases: 85,934

New in Bellenguez 2022; rare (freq 0.0091), OR 1.28.

Source: Bellenguez et al. 2022, Nat Genet (EADB/IGAP/UKB stage 1) · 2022 · PubMed 35379992

OR 1.16 (95% CI 1.11-1.21) per G allele (odds ratio per effect allele)

• Chr 3
• Risk allele: G
• OR 1.16 (1.11–1.21)
• p = 2e-11
• Cases: 85,934

New in Bellenguez 2022 (neprilysin, amyloid-degrading enzyme); OR 1.16 (1.11-1.21), freq 0.026.

Source: Bellenguez et al. 2022, Nat Genet (EADB/IGAP/UKB stage 1) · 2022 · PubMed 35379992

OR 2.02 (95% CI 1.57-2.60) per T allele (odds ratio per effect allele)

• Chr 4
• Risk allele: T
• OR 2.02 (1.57–2.6)
• p = 4e-08
• Cases: 85,934

Rarest and largest-effect common-GWAS hit in Bellenguez 2022 (freq 0.0013, OR 2.02, 95% CI 1.57-2.60, P 4e-8) -- borderline significance, treat with caution.

Source: Bellenguez et al. 2022, Nat Genet (EADB/IGAP/UKB stage 1) · 2022 · PubMed 35379992

OR 0.96 (95% CI 0.95-0.97) per T allele (odds ratio per effect allele)

• Chr 5
• Risk allele: T
• OR 0.96 (0.95–0.97)
• p = 9.000000000000001e-09
• Cases: 85,934

New in Bellenguez 2022; effect allele T protective.

Source: Bellenguez et al. 2022, Nat Genet (EADB/IGAP/UKB stage 1) · 2022 · PubMed 35379992

OR 1.06 (95% CI 1.05-1.08) per G allele (odds ratio per effect allele)

• Chr 10
• Risk allele: G
• OR 1.06 (1.05–1.08)
• p = 1e-18
• Cases: 85,934

Source: Bellenguez et al. 2022, Nat Genet (EADB/IGAP/UKB stage 1) · 2022 · PubMed 35379992

OR 1.07 (95% CI 1.05-1.09) per A allele (odds ratio per effect allele)

• Chr 16
• Risk allele: A
• OR 1.07 (1.05–1.09)
• p = 6e-10
• Cases: 85,934

Microglial survival cytokine; missense.

Source: Bellenguez et al. 2022, Nat Genet (EADB/IGAP/UKB stage 1) · 2022 · PubMed 35379992

OR 1.05 (95% CI 1.04-1.07) per C allele (odds ratio per effect allele)

• Chr 19
• Risk allele: C
• OR 1.05 (1.04–1.07)
• p = 4e-11
• Cases: 85,934

New in Bellenguez 2022.

Source: Bellenguez et al. 2022, Nat Genet (EADB/IGAP/UKB stage 1) · 2022 · PubMed 35379992

OR 1.05 (95% CI 1.03-1.07) per T allele (odds ratio per effect allele)

• Chr 19
• Risk allele: T
• OR 1.05 (1.03–1.07)
• p = 5e-09
• Cases: 85,934

New in Bellenguez 2022.

Source: Bellenguez et al. 2022, Nat Genet (EADB/IGAP/UKB stage 1) · 2022 · PubMed 35379992

OR 2.90 (95% CI 2.16-3.91) per T allele (odds ratio per effect allele)

• Chr 6
• Risk allele: T
• OR 2.9 (2.16–3.91)
• p = 2e-12
• Cases: 3,550

R47H. Abstract: Iceland discovery OR 2.92 (95% CI 2.09-4.09, P 3.42e-10); combined discovery + replication OR 2.90 (2.16-3.91, P 2.1e-12); control frequency 0.46% in Icelanders >=85 y. Catalog effect-allele frequency 0.0063.

Source: Jonsson et al. 2013, N Engl J Med (deCODE) · 2013 · PubMed 23150908

OR 5.05 (95% CI 2.77-9.16), P 9.0e-9 (entire series, additive model); OR 4.59 (2.49-8.46) in newly genotyped participants (odds ratio per allele)

• Chr 6
• Risk allele: T (R47H)
• OR 5.05 (2.77–9.16)
• p = 9e-09
• Cases: 1,887

Read from PMC3631573 (author manuscript). Early small-sample estimate; later large meta-analyses give OR ~2.4 (Sims 2017, Bellenguez 2022) -- classic winner's-curse shrinkage. The case and control counts give the direct-genotyping series.

Source: Guerreiro et al. 2013, N Engl J Med · 2013 · PubMed 23150934

OR 2.46 (95% CI 2.06-2.92) per T allele (odds ratio per effect allele)

• Chr 6
• Risk allele: T
• OR 2.456 (2.06–2.92)
• p = 5e-24
• Cases: 16,097

R47H in the IGAP exome-array study (stage 1). Catalog OR 2.4557 (95% CI 2.06-2.92), P 5e-24.

Source: Sims et al. 2017, Nat Genet (exome-array, IGAP) · 2017 · PubMed 28714976

OR 2.39 (95% CI 2.09-2.73) per T allele (odds ratio per effect allele)

• Chr 6
• Risk allele: T
• OR 2.39 (2.09–2.73)
• p = 3e-37
• Cases: 85,934

R47H in the largest European GWAS: OR 2.39 (2.09-2.73), P 3e-37, effect-allele frequency 0.0031. R47H is essentially absent in East Asian and African populations (gnomAD eas 0, afr 0.0004), so TREM2 risk in those populations is carried by other variants.

Source: Bellenguez et al. 2022, Nat Genet (EADB/IGAP/UKB stage 1) · 2022 · PubMed 35379992

OR 1.67 (95% CI 1.46-1.90) per T allele (odds ratio per effect allele)

• Chr 6
• Risk allele: T
• OR 1.667 (1.46–1.9)
• p = 2e-14
• Cases: 16,097

R62H (p.Arg62His). Abstract: OR 1.67, P 1.55e-14, MAF cases 0.0143 vs controls 0.0089.

Source: Sims et al. 2017, Nat Genet (exome-array, IGAP) · 2017 · PubMed 28714976

OR 1.41 (95% CI 1.32-1.50) per T allele (odds ratio per effect allele)

• Chr 6
• Risk allele: T
• OR 1.41 (1.32–1.5)
• p = 3.0000000000000002e-25
• Cases: 85,934

R62H in Bellenguez 2022: OR 1.41 (1.32-1.50), effect-allele frequency 0.0126.

Source: Bellenguez et al. 2022, Nat Genet (EADB/IGAP/UKB stage 1) · 2022 · PubMed 35379992

OR 1.55 (95% CI 1.33-1.80) per G allele (odds ratio per effect allele)

• Chr 6
• Risk allele: G
• OR 1.55 (1.33–1.8)
• p = 2e-08
• Cases: 85,934

Third independent signal at the TREM2 locus in Bellenguez 2022 (OR 1.55, 1.33-1.80; frequency 0.0042).

Source: Bellenguez et al. 2022, Nat Genet (EADB/IGAP/UKB stage 1) · 2022 · PubMed 35379992

P = 2e-7 in MVP African-ancestry ADRD/proxy meta-analysis (GWAS Catalog GCST90267627; no OR curated); not genome-wide significant (p-value only)

• Chr 6
• Risk allele: T (T96K)
• p = 2e-07
• Cases: 6,796

T96K is common in African ancestry (~12% of alleles) and near-absent in Europeans, so it is invisible to European GWAS. Included as a suggestive (p 2e-7) signal only; effect direction/OR were not curated in the catalog record and are not asserted here.

Source: Sherva et al. 2023, Mol Psychiatry (Million Veteran Program African ancestry ADRD + proxy, meta-analysed with ADGC) · 2023 · PubMed 36543923

OR 1.48 (95% CI 1.25-1.67) (effect allele not specified in catalog record; OR may be inverted) (odds ratio per effect allele)

• Chr 16
• OR 1.476 (1.25–1.67)
• p = 5e-10
• Cases: 16,097

P522R. Abstract: protective, OR 0.68, P 5.38e-10, MAF cases 0.0059 vs controls 0.0093. The catalog stores the inverted OR (1.4756) because the effect allele was not specified; use the abstract value 0.68 for the G (Arg) allele.

Source: Sims et al. 2017, Nat Genet (exome-array, IGAP) · 2017 · PubMed 28714976

OR 0.74 (95% CI 0.68-0.81) per G allele (odds ratio per effect allele)

• Chr 16
• Risk allele: G
• OR 0.74 (0.68–0.81)
• p = 8e-12
• Cases: 85,934

P522R in Bellenguez 2022: OR 0.74 (0.68-0.81) per G allele; frequency 0.0078. Microglial gain-of-function variant; also protective for FTD/DLB in other studies (not loaded).

Source: Bellenguez et al. 2022, Nat Genet (EADB/IGAP/UKB stage 1) · 2022 · PubMed 35379992

OR 1.43 (95% CI 1.30-1.60) per T allele (odds ratio per effect allele)

• Chr 17
• Risk allele: T
• OR 1.43 (1.3–1.6)
• p = 5e-10
• Cases: 16,097

S209F. Abstract: OR 1.43, P 4.56e-10, MAF cases 0.011 vs controls 0.008.

Source: Sims et al. 2017, Nat Genet (exome-array, IGAP) · 2017 · PubMed 28714976

OR 1.32 (95% CI 1.23-1.42) per T allele (odds ratio per effect allele)

• Chr 17
• Risk allele: T
• OR 1.32 (1.23–1.42)
• p = 3e-14
• Cases: 85,934

S209F in Bellenguez 2022: OR 1.32 (1.23-1.42), frequency 0.0122.

Source: Bellenguez et al. 2022, Nat Genet (EADB/IGAP/UKB stage 1) · 2022 · PubMed 35379992

P = 3e-09 (effect size not curated) (p-value only)

• Chr 6
• Risk allele: A
• p = 3.0000000000000004e-09
• Cases: 6,796

Abstract: 'MVP/ADGC meta-analysis yielded additional significant SNPs near known AD risk genes TREM2 (rs73427293, p = 2.95e-9)'; the peak variant differs from the European R47H. GnomAD afr frequency of the T allele 0.124, nfe 0.001.

Source: Sherva et al. 2023, Mol Psychiatry (Million Veteran Program African ancestry ADRD + proxy, meta-analysed with ADGC) · 2023 · PubMed 36543923

OR 1.79 (95% CI 1.47-2.12) per G allele (odds ratio per effect allele)

• Chr 19
• Risk allele: G
• OR 1.79 (1.47–2.12)
• p = 2e-09
• Cases: 1,968

Abstract: 'rs115550680, allele G; frequency 0.09 cases and 0.06 controls; OR 1.79 (95% CI 1.47-2.12); P 2.2e-9', an effect 'comparable with that of the APOE e4-determining SNP rs429358' in the same sample (OR 2.31, 2.19-2.42). GnomAD G allele: afr 0.057, nfe 0.0002 -- an African-ancestry-specific ABCA7 risk allele.

Source: Reitz et al. 2013, JAMA (ADGC African American) · 2013 · PubMed 23571587

OR 1.58 (95% CI 1.32-1.88) (effect allele not specified in catalog record; OR may be inverted) (odds ratio per effect allele)

• Chr 19
• OR 1.58 (1.32–1.88)
• p = 4e-07
• Cases: 2,784

Same African-American ABCA7 variant in the expanded 2021 ADGC analysis (OR 1.58, 1.32-1.88; P 4e-7).

Source: Kunkle et al. 2021, JAMA Neurol (ADGC African American, African Genome Resources panel) · 2021 · PubMed 33074286

OR 1.93 (95% CI 1.72-2.17) (effect allele not specified in catalog record; OR may be inverted) (odds ratio per effect allele)

• Chr 19
• OR 1.93 (1.72–2.17)
• p = 2e-25
• Cases: 2,784

APOE-region lead in African Americans (OR 1.93, 1.72-2.17).

Source: Kunkle et al. 2021, JAMA Neurol (ADGC African American, African Genome Resources panel) · 2021 · PubMed 33074286

OR 0.10 (95% CI 0.05-0.22) (effect allele not specified in catalog record; OR may be inverted) (odds ratio per effect allele)

• Chr 15
• OR 0.1 (0.05–0.22)
• p = 2e-09
• Cases: 2,784

Abstract: 'genome-wide significant intergenic locus near IGF1R at 15q26 (P = 1.7e-9)'; rare variant, catalog OR 0.10 (0.05-0.22). Rare-variant ORs of this size in ~8,000 samples are unstable; not replicated in Europeans (variant absent).

Source: Kunkle et al. 2021, JAMA Neurol (ADGC African American, African Genome Resources panel) · 2021 · PubMed 33074286

OR 0.78 (95% CI 0.71-0.86) (effect allele not specified in catalog record; OR may be inverted) (odds ratio per effect allele)

• Chr 3
• OR 0.78 (0.71–0.86)
• p = 9e-07
• Cases: 2,784

Suggestive (P 9.3e-7) novel common locus in African Americans; ALCAM expression associated with brain amyloid load. Not genome-wide significant; listed because these are the candidate African-American-specific loci highlighted by the authors.

Source: Kunkle et al. 2021, JAMA Neurol (ADGC African American, African Genome Resources panel) · 2021 · PubMed 33074286

OR 1.73 (95% CI 1.39-2.15) (effect allele not specified in catalog record; OR may be inverted) (odds ratio per effect allele)

• Chr 13
• OR 1.73 (1.39–2.15)
• p = 4e-07
• Cases: 2,784

Suggestive (P 4.1e-7). Not genome-wide significant; listed because these are the candidate African-American-specific loci highlighted by the authors.

Source: Kunkle et al. 2021, JAMA Neurol (ADGC African American, African Genome Resources panel) · 2021 · PubMed 33074286

OR 1.39 (95% CI 1.23-1.56) (effect allele not specified in catalog record; OR may be inverted) (odds ratio per effect allele)

• Chr 19
• OR 1.39 (1.23–1.56)
• p = 4e-07
• Cases: 2,784

Suggestive (P 3.5e-7). Not genome-wide significant; listed because these are the candidate African-American-specific loci highlighted by the authors.

Source: Kunkle et al. 2021, JAMA Neurol (ADGC African American, African Genome Resources panel) · 2021 · PubMed 33074286

P = 3e-10 (effect size not curated) (p-value only)

• Chr 19
• Risk allele: A
• p = 3e-10
• Cases: 6,796

P 3.26e-10; peak differs from European and prior AA leads. Phenotype is ADRD (incl. Vascular/non-specific dementia) plus parental-proxy dementia in a predominantly male Veteran cohort; ORs were not curated by the catalog.

Source: Sherva et al. 2023, Mol Psychiatry (Million Veteran Program African ancestry ADRD + proxy, meta-analysed with ADGC) · 2023 · PubMed 36543923

P = 1e-09 (effect size not curated) (p-value only)

• Chr 6
• Risk allele: T
• p = 1e-09
• Cases: 6,796

P 1.14e-9. Phenotype is ADRD (incl. Vascular/non-specific dementia) plus parental-proxy dementia in a predominantly male Veteran cohort; ORs were not curated by the catalog.

Source: Sherva et al. 2023, Mol Psychiatry (Million Veteran Program African ancestry ADRD + proxy, meta-analysed with ADGC) · 2023 · PubMed 36543923

P = 2e-08 (effect size not curated) (p-value only)

• Chr 3
• Risk allele: C
• p = 2e-08
• Cases: 6,796

Novel in African ancestry (P 1.63e-8), common-variant meta-analysis of MVP ADRD + proxy. Phenotype is ADRD (incl. Vascular/non-specific dementia) plus parental-proxy dementia in a predominantly male Veteran cohort; ORs were not curated by the catalog.

Source: Sherva et al. 2023, Mol Psychiatry (Million Veteran Program African ancestry ADRD + proxy, meta-analysed with ADGC) · 2023 · PubMed 36543923

P 3.68e-9, MAF 0.002 (novel African-ancestry locus); catalog beta 3.5 (95% CI 2.17-4.83), P 2e-7 for the within-dataset meta-analysis record (p-value; beta as curated (scale not stated))

• Chr 9
• p = 3.68e-09
• Cases: 2,903

Very rare variant; gnomAD C allele afr 0.0024, nfe 0. Authors: association patterns at chr12q23.2 (ASCL1) differed by degree of African ancestry, i.e. Local ancestry modulates some associations. Classified 'disputed' pending replication.

Source: Ray et al. 2024, Alzheimers Dement (ADGC African ancestry, African Genome Resources panel) · 2024 · PubMed 38958117

OR 0.65 (95% CI 0.57-0.75) per C allele (odds ratio per effect allele)

• Chr 4
• Risk allele: C
• OR 0.65 (0.57–0.75)
• p = 5e-09
• Cases: 3,962

Novel Japanese locus (OR 0.65, 0.57-0.75, P 5e-9 in meta-analysis with replication). GnomAD C allele: eas 0.043, nfe 0.009, afr 0.089.

Source: Shigemizu et al. 2021, Transl Psychiatry (Japanese GWAS) · 2021 · PubMed 33654092

P = 2e-08 (effect size not curated) (p-value only)

• Chr 6
• p = 2e-08
• Cases: 25,944

Novel locus from Japanese + IGAP trans-ethnic meta-analysis (P 2e-8); effect size not curated.

Source: Shigemizu et al. 2021, Transl Psychiatry (Japanese + IGAP trans-ethnic meta-analysis) · 2021 · PubMed 33654092

P 1.77e-9 (combined Japanese + Korean + Caucasian); OR not in abstract (p-value only)

• Chr 11
• Risk allele: C (protective)
• p = 1.77e-09
• Cases: 14,072

First genome-wide-significant SORL1 association, made possible by the much higher East Asian allele frequency. PICALM rs3851179 (P 1.7e-5) and BIN1 rs744373 (P 1.4e-4) replicated in Japanese with the same risk alleles as Caucasians. The case and control counts sum all stages.

Source: Miyashita et al. 2013, PLoS One (JGSCAD/ADGC) · 2013 · PubMed 23565137

OR 0.69, P 3.07e-19 (odds ratio (allele not stated in text loaded))

• Chr 5
• OR 0.69
• p = 3.07e-19
• Cases: 3,913

ORs read from the PMC full text (PMC8041344); CIs and effect alleles were in supplementary tables not loaded. Not in the GWAS Catalog. Position from Ensembl release 116.

Source: Jia et al. 2021, Brain (Chinese GWAS) · 2021 · PubMed 33188687

OR 0.74, P 2.49e-23; also genome-wide significant in IGAP stage 1 (odds ratio (allele not stated in text loaded))

• Chr 5
• OR 0.74
• p = 2.49e-23
• Cases: 3,913

ORs read from the PMC full text (PMC8041344); CIs and effect alleles were in supplementary tables not loaded. Not in the GWAS Catalog. Position from Ensembl release 116.

Source: Jia et al. 2021, Brain (Chinese GWAS) · 2021 · PubMed 33188687

OR 1.71, P 1.35e-67 (odds ratio (allele not stated in text loaded))

• Chr 14
• OR 1.71
• p = 1.35e-67
• Cases: 3,913

ORs read from the PMC full text (PMC8041344); CIs and effect alleles were in supplementary tables not loaded. Not in the GWAS Catalog. Position from Ensembl release 116.

Source: Jia et al. 2021, Brain (Chinese GWAS) · 2021 · PubMed 33188687

OR 1.23, P 4.81e-9 (odds ratio (allele not stated in text loaded))

• Chr 21
• OR 1.23
• p = 4.81e-09
• Cases: 3,913

ORs read from the PMC full text (PMC8041344); CIs and effect alleles were in supplementary tables not loaded. Not in the GWAS Catalog. Position from Ensembl release 116.

Source: Jia et al. 2021, Brain (Chinese GWAS) · 2021 · PubMed 33188687

OR 3.10 (95% CI 2.75-3.49) per T allele (odds ratio per effect allele)

• Chr 19
• Risk allele: T
• OR 3.1 (2.75–3.49)
• p = 1e-320
• Cases: 24,495

Cross-ancestry random-effects estimate for the e4-defining SNP: OR 3.10 (2.75-3.49) per allele. Ancestry-specific: African American stratum chr19:44908684-T OR 0.376 (0.34-0.39) for the non-e4 T allele (frequency 0.759); Hispanic stratum OR 0.443 (0.41-0.46) for T (freq 0.797).

Source: Rajabli et al. 2025, Genome Biol (ADGC cross-ancestry random-effects meta-analysis) · 2025 · PubMed 40676597

OR 0.54 (95% CI 0.50-0.60) per T allele (odds ratio per effect allele)

• Chr 19
• Risk allele: T
• OR 0.54 (0.5–0.6)
• p = 4e-106
• Cases: 24,495

Cross-ancestry estimate for the e2-defining SNP: OR 0.54 (0.50-0.60) per T allele.

Source: Rajabli et al. 2025, Genome Biol (ADGC cross-ancestry random-effects meta-analysis) · 2025 · PubMed 40676597

OR 1.12 (95% CI 1.08-1.16) per A allele (odds ratio per effect allele)

• Chr 11
• Risk allele: A
• OR 1.12 (1.08–1.16)
• p = 5e-09
• Cases: 24,495

Novel cross-population locus 11p12 (abstract: LRRC4C). OR 1.12 (1.08-1.16), P 5e-9. GnomAD A allele frequency: nfe 0.164, eas 0.419, afr 0 (not observed) -- detection depended on non-African populations.

Source: Rajabli et al. 2025, Genome Biol (ADGC cross-ancestry random-effects meta-analysis) · 2025 · PubMed 40676597

OR 0.63 (95% CI 0.32-0.81) per T allele (odds ratio per effect allele)

• Chr 12
• Risk allele: T
• OR 0.63 (0.32–0.81)
• p = 1e-08
• Cases: 24,495

Novel cross-population locus 12q24.13 (abstract: LHX5-AS1); OR 0.63 (0.32-0.81), P 1e-8. Driven by the Hispanic stratum where chr12:113643122-T has OR 0.44 (0.34-0.50), P 5e-10, frequency 0.039; gnomAD T allele nfe 0.0002, amr 0.0022, eas 0.

Source: Rajabli et al. 2025, Genome Biol (ADGC cross-ancestry random-effects meta-analysis) · 2025 · PubMed 40676597

OR 0.47 (95% CI 0.36-0.54) per A allele (odds ratio per effect allele)

• Chr 2
• Risk allele: A
• OR 0.472 (0.36–0.54)
• p = 2e-08
• Cases: 3,112

Hispanic-specific genome-wide-significant locus (OR 0.47, 0.36-0.54, P 2e-8, frequency 0.038). The abstract names GRB14 and PTPRK as Hispanic-specific loci; the catalog record gives coordinates only, so no gene is asserted here.

Source: Rajabli et al. 2025, Genome Biol (ADGC Hispanic stratum) · 2025 · PubMed 40676597

OR 0.63 (95% CI 0.54-0.69) per A allele (odds ratio per effect allele)

• Chr 9
• Risk allele: A
• OR 0.635 (0.54–0.69)
• p = 2e-08
• Cases: 1,430

East Asian stratum genome-wide-significant locus (OR 0.63, 0.54-0.69, P 2e-8) in only 1,430 cases / 1,802 controls; gene not curated; treat as provisional.

Source: Rajabli et al. 2025, Genome Biol (ADGC East Asian stratum) · 2025 · PubMed 40676597

OR 1.20 (95% CI 1.04-1.39) per A allele (odds ratio per effect allele)

• Chr 8
• Risk allele: A
• OR 1.2 (1.04–1.39)
• p = 2e-09
• Cases: 24,495

Abstract: 'we detect the SHARPIN locus with only 13.7% of the sample size of the NHW GWAS study (n = 409,589) in which this locus was first observed' -- illustrates the power of diverse-ancestry haplotypes.

Source: Rajabli et al. 2025, Genome Biol (ADGC cross-ancestry random-effects meta-analysis) · 2025 · PubMed 40676597

P = 7e-08 (effect size not curated) (p-value only)

• Chr 3
• Risk allele: A
• p = 7e-08
• Cases: 101,061

Chromosome 3 locus in the Lake 2023 multi-ancestry meta-analysis; abstract reports 'two independent novel disease-associated loci on chromosome 3'. In the catalog record this variant is P 7e-8 / 2e-7 (just above 5e-8), and effect sizes were not curated -- provisional.

Source: Lake et al. 2023, Mol Psychiatry (multi-ancestry meta-analysis) · 2023 · PubMed 37198259

P = 2e-07 (effect size not curated) (p-value only)

• Chr 3
• Risk allele: G
• p = 2e-07
• Cases: 101,061

Chromosome 3 locus in the Lake 2023 multi-ancestry meta-analysis; abstract reports 'two independent novel disease-associated loci on chromosome 3'. In the catalog record this variant is P 7e-8 / 2e-7 (just above 5e-8), and effect sizes were not curated -- provisional.

Source: Lake et al. 2023, Mol Psychiatry (multi-ancestry meta-analysis) · 2023 · PubMed 37198259

AUC 0.782 (95% CI 0.77-0.80) with APOE, PRS, sex and age as predictors

Who was studied: Training: IGAP 17,008 cases / 37,154 controls; prediction subset 3,049 cases / 1,554 controls with APOE genotypes; European ancestry

Polygenic component significant even after excluding APOE and known loci (P 3.4e-19). This 0.78 figure is the most-cited PRS benchmark; note it includes age and sex.

Source: Escott-Price et al. 2015, Brain · 2015 · PubMed 26490334

AUC 0.84 in a pathological case-control series (vs <0.80 in clinical series)

Who was studied: Pathologically confirmed AD cases and controls (n not in abstract); European ancestry

Authors 'suggest ... There is limited room for further improvement using genetic data'. Higher AUC reflects cleaner phenotypes, not a better score.

Source: Escott-Price et al. 2017, Ann Neurol · 2017 · PubMed 28727176

AUC 74.1% (best model); APOE alone AUC 70.0%; PRS without APOE 61.3%; individuals at +/-2 SD extremes OR 124 (95% CI 6-2707)

Who was studied: IGAP-derived weights applied to a European-ancestry case-control sample (n in Table 2 of paper; not in abstract)

Read from PMC8302739. Key limits: prediction accuracy is similar across PRS methods but individual rankings differ; scores must be standardised against a population (not sample) mean to be comparable. The extreme-tail OR has an enormous CI.

Source: Leonenko et al. 2021, Nat Commun · 2021 · PubMed 34301930

1.6- to 1.9-fold increase in AD risk from lowest to highest decile, in addition to age and APOE e4 effects

Who was studied: Independent European cohorts (Rotterdam, Three-City etc.; abstract) -- validation of a score built from 111,326 cases / 677,663 controls

Also predicted progression from MCI to AD/dementia. Modest gradient: the non-APOE polygenic component moves risk <2-fold across the whole distribution.

Source: Bellenguez et al. 2022, Nat Genet · 2022 · PubMed 35379992

AUC 0.72-0.76; microglia-region-only PRS AUC 0.71-0.74; individual risk scores differed by up to 70% depending on which GWAS summary statistics were used

Who was studied: ADNI (N = 568) and BioFINDER (N = 766); European-ancestry research cohorts

Most recent benchmark; shows PRS instability at the individual level even within European ancestry.

Source: Bellou et al. 2025, Alzheimers Res Ther · 2025 · PubMed 39762974

OR 1.95 (95% CI 1.40-2.72) for AD dementia per PRS unit, independent of APOE e4 (dataset 1); OR 1.85 (1.05-3.32) in dataset 2; amyloid PET positivity OR 1.81 (1.32-2.48)

Who was studied: 1,634 Koreans: 716 AD dementia, 222 aMCI, 699 cognitively unimpaired

Shows partial transferability of a European PRS to East Asians; a trans-ancestry (European + Japanese) meta-analysis PRS was 'slightly' more accurate. AUC not in abstract.

Source: Jung et al. 2022, JAMA Netw Open · 2022 · PubMed 36520433

Maximal AUC 0.79 (multi-ancestry PRS incl. APOE) vs 0.75 (European PRS incl. APOE); 0.68 vs 0.63 excluding APOE

Who was studied: Colombian three-way admixed cohort: 281 AD cases / 87 neurologically normal controls (TANGL study)

Read from PMC10615750. Small test cohort (87 controls); shows that adding even under-powered non-European GWAS improves transfer to admixed populations.

Source: Lake et al. 2023, Mol Psychiatry · 2023 · PubMed 37198259

Peak AUC 0.73 (95% CI 0.71-0.74); range 0.63-0.73 across 11 models; validated in a longitudinal cohort of 5,474

Who was studied: Chinese two-stage GWAS 3,913 cases / 7,593 controls; validation cohort 5,474

Few-SNP models rather than genome-wide PRS; AUC is dominated by APOE.

Source: Jia et al. 2021, Brain · 2021 · PubMed 33188687

PRS 'available today are several times more accurate in individuals of European ancestry than other ancestries' -- 'an inescapable consequence of Eurocentric biases in GWAS'

Who was studied: Cross-trait analysis (not AD-specific)

The core transferability caveat: clinical use of current PRS 'would systematically afford greater improvement for European-descent populations'.

Source: Martin et al. 2019, Nat Genet · 2019 · PubMed 30926966