We previously reported that the ApoE e4e4 genotype was associated with COVID-19 test positivity (odds ratio [OR] = 2.31, 95% CI: 1.65–3.24, p = 1.19 × 10−6) (1) in the UK Biobank (UKB) cohort, during the epidemic peak in England, from March 16 to April 26, 2020 (2). With more COVID-19 test results (March 16 to May 31, 2020) and mortality data (to March 31, 2020, with incomplete data for April 2020) linked to UKB, we reevaluated the ApoE e4 allele association with COVID-19 test positivity, and with all-cause mortality following test-confirmed COVID-19.
We restricted analyses to European-ancestry participants (3) (n = 451 367, 90% of sample) attending baseline assessment centers in England (n = 398 073) and excluded participants who died before the pandemic (set at February 1, 2020, n = 22 384). Single-nucleotide polymorphism data for rs429358 and rs7412 were used to determine ApoE genotypes. Our outcomes of interest were: (a) COVID-19 test positive versus the rest of the sample meeting inclusion criteria (ie, including untested samples and tested negative), and (b) tested positive and died versus the rest of the sample as above, but with additional exclusion of test positive participants who survived. Some of the excluded samples could have died but cannot be identified due to incomplete mortality data for April 2020. Logistic regression models compared ApoE e4e4 participants (or e3e4s) to e3e3s with adjustment for sex; age on April 26 or age at death; baseline UKB assessment center in England (accounting for geographical differences in viral exposures); genotyping array type; and the top five genetic principal components (accounting for possible population admixture).
The mean attained age was 68.2 years (SD = 8.0) with 174 667 females (55%). Of 219 747 e3e3 participants, 663 participants tested positive (302 per 100 000), of whom, 79 later died. Similarly, of 8767 e4e4 participants, 59 tested positive (673 per 100 000), of whom 13 later died (Table 1). In logistic models, ApoE e4e4 genotype was associated with increased risks of test positivity (OR = 2.24, 95% CI: 1.72–2.93, p = 3.24 × 10−9) and of mortality with test-confirmed COVID-19 (OR = 4.29, 95% CI: 2.38–7.72, p = 1.22 × 10−6), compared to e3e3s. For e3e4s versus e3e3s, these two associations were nominally statistically significant (at p < .05), but with much smaller effect sizes. The e4e4 associations were similar after excluding 50 566 participants related to the third degree or closer for test positivity (e4e4 OR = 2.30, 95% CI: 1.73–3.07, p = 1.39 × 10−8) and for mortality with test-confirmed COVID-19 (e4e4 OR = 4.53, 95% CI: 2.39–8.61, p = 3.87 × 10−6). Additionally, the e4e4 association with either COVID-19 outcome was little changed after removing participants with diseases associated with ApoE e4 alleles (4) and COVID-19 severity (5), including dementia, hypertension, coronary artery disease (myocardial infarction or angina), or type 2 diabetes (Table 1), based on diagnoses recorded from baseline self-reports and hospital discharge records during follow-up to March 2017. ApoE e3e4s were modestly associated with test positivity overall, and the association tended to be less marked in disease-free samples (Table 1). In additional analyses, we tested associations with ApoE e2 alleles, which have been linked to beneficial health outcomes (4). No associations were found between e2e3 and either of our COVID-19 outcomes (p > .05, vs e3e3). Analyses for e2e2s associations were underpowered (n = 2427, 4 positives, and 1 positive death).
Table 1.
Risk of COVID-19 Test Positivity and Mortality, Comparing Participants With ApoE e3e4 or e4e4 to e3e3 Genotypes, in UK Biobank
| COVID-19 Positive vs Rest of Study Sample | COVID-19 Positive and Died vs Rest† of Study Sample | |||||||||
|---|---|---|---|---|---|---|---|---|---|---|
| n | Negative or Untested | Positive | Positive and Dead | Positivity Rate per 105 | Positivity and Death Rate per 105 | OR (95% CI)* | p Value | OR (95% CI)* | p Value | |
| All | ||||||||||
| e3e3 | 219 747 | 219 084 | 663 | 79 | 302 | 36 | ||||
| e3e4 | 88 882 | 88 561 | 321 | 42 | 361 | 47 | 1.20 (1.05, 1.37) | .009 | 1.35 (0.92, 1.96) | .121 |
| e4e4 | 8767 | 8708 | 59 | 13 | 673 | 148 | 2.24 (1.72, 2.93) | 3.24E-9 | 4.29 (2.38, 7.72) | 1.22E-6 |
| Excluding dementia | ||||||||||
| e3e3 | 219 392 | 218 744 | 648 | 76 | 295 | 35 | ||||
| e3e4 | 88 558 | 88 263 | 295 | 34 | 333 | 39 | 1.13 (0.98, 1.29) | .093 | 1.14 (0.76, 1.70) | .536 |
| e4e4 | 8676 | 8618 | 58 | 13 | 669 | 151 | 2.27 (1.74, 2.98) | 2.42E-9 | 4.53 (2.51, 8.16) | 5.21E-7 |
| Excluding hypertension | ||||||||||
| e3e3 | 147 332 | 146 958 | 374 | 31 | 254 | 21 | ||||
| e3e4 | 59 655 | 59 483 | 172 | 17 | 288 | 29 | 1.13 (0.94, 1.35) | .186 | 1.39 (0.77, 2.51) | .278 |
| e4e4 | 5918 | 5881 | 37 | 5 | 625 | 85 | 2.45 (1.75, 3.44) | 2.10E-7 | 4.25 (1.65, 10.95) | .003 |
| Excluding coronary artery disease | ||||||||||
| e3e3 | 201 003 | 200 435 | 568 | 62 | 283 | 31 | ||||
| e3e4 | 80 850 | 80 590 | 260 | 33 | 322 | 41 | 1.14 (0.98, 1.32) | .090 | 1.36 (0.89, 2.08) | .153 |
| e4e4 | 7973 | 7923 | 50 | 10 | 627 | 126 | 2.23 (1.67, 2.98) | 6.21E-8 | 4.23 (2.16, 8.26) | 2.43E-5 |
| Excluding type 2 diabetes | ||||||||||
| e3e3 | 208 374 | 207 795 | 579 | 61 | 278 | 29 | ||||
| e3e4 | 84 620 | 84 342 | 278 | 30 | 329 | 36 | 1.18 (1.02, 1.36) | .024 | 1.24 (0.80, 1.92) | .338 |
| e4e4 | 8 391 | 8336 | 55 | 12 | 655 | 144 | 2.36 (1.79, 3.12) | 1.23E-9 | 5.05 (2.72, 9.39) | 3.08E-7 |
Notes: CI = confidence interval; OR = odds ratio.
*Adjusted for sex, age at death or age on April 26, 2020 (the last date of death), assessment center in England, genotyping array type, and the top five genetic principal components.
†Comparison group excluded participants testing positive and surviving.
The results presented imply a recessive effect of the ApoE e4 allele. Only modest associations were present between the much more common e3e4 genotype and COVID-19 outcomes, similar to results for rs429358 (which separates 0, 1, and 2 copies of e4 alleles, OR = 1.3, p = .0026) reported for severe COVID-19 with respiratory failure in a recent additive effect genome-wide analysis (6). ApoE e4e4 associations with test positivity and mortality were little affected by excluding dementia and other ApoE e4-associated diagnoses reported before March 2017: future work should include recent preexisting diagnoses. More data are needed on ApoE and COVID-19 associations in other ancestry groups, as numbers of UKB participants of such groups are unfortunately too small for this analysis.
In conclusion, ApoE e4e4 genotype is associated with COVID-19 test positivity at genome-wide significance (ie, p < 5 × 10−8) in UKB, using data covering a longer period than previously reported. Similarly, the e4e4 genotype was associated with a 4-fold increase in mortality after testing positive for COVID-19, in UKB. Independent replications are needed to confirm our findings and mechanistic work is needed to understand how ApoE e4e4 results in the marked increase in vulnerability, especially for COVID-19 mortality. These findings also demonstrate that risks for COVID-19 mortality are not simply related to advanced chronological age or the comorbidities commonly seen in aging.
Funding
C.L.K., L.C.P., G.A.K., and D.M. are supported in part by an R21 grant (R21AG060018) funded by National Institute on Aging, National Institutes of Health, USA. UK Medical Research Council award MR/S009892/1 (PI Melzer) supports J.L.A. J.A.H.M. is supported by National Institute for Health Research, UK Doctoral Research Fellowship DRF-2014-07-177. The views expressed in this publication are those of the author(s) and not necessarily those of the NHS, the National Institute for Health Research or the Department of Health.
References
- 1. Kuo C-L, Pilling LC, Atkins JL, et al. . APOE e4 genotype predicts severe COVID-19 in the UK Biobank community cohort. J Gerontol A Biol Sci Med Sci [Internet]. 2020. doi: 10.1093/gerona/glaa131 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2. Armstrong J, Rudkin JK, Allen N, et al. . Dynamic linkage of COVID-19 test results between Public Health England’s Second Generation Surveillance System and UK Biobank. Microb Genom. 2020. doi: 10.1099/mgen.0.000397 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3. Pilling LC, Tamosauskaite J, Jones G, et al. . Common conditions associated with hereditary haemochromatosis genetic variants: cohort study in UK Biobank. BMJ. 2019;364:k5222. doi: 10.1136/bmj.k5222 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4. Kuo C-L, Pilling LC, Atkins JL, Kuchel GA, Melzer D. ApoE e2 and aging-related outcomes in 379,000 UK Biobank participants. Aging (Albany NY). 2020;12(12):12222–12233. doi: 10.18632/aging.103405 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5. Atkins JL, Masoli JAH, Delgado J, et al. . Preexisting comorbidities predicting severe covid-19 in older adults in the UK Biobank community cohort. medRxiv. 2020. http://medrxiv.org/content/early/2020/05/08/2020.05.06.20092700.abstract. Accessed June 15, 2020.
- 6. Ellinghaus D, Degenhardt F, Bujanda L, et al. . Genomewide association study of severe Covid-19 with respiratory failure. N Engl J Med [Internet]. 2020. doi: 10.1056/NEJMoa2020283 [DOI] [PMC free article] [PubMed] [Google Scholar]