Pulmonary Function and Risk of Alzheimer Dementia: Two-Sample Mendelian Randomization Study

Abstract

Dementia is a major growing global public health problem.1Prince M.J. Wimo A. Guerchet M. Ali G.-C. Wu Y.-T. Prina A.M. World Alzheimer Report 2015. The Global Impact of Dementia: an analysis of prevalence, incidence, cost and trends. Alzheimer's Disease International, London2015Google Scholar Alzheimer disease (AD) risk is thought to be raised in the presence of relatively few environmental and genetic factors that include lower educational attainment, hypertension, obesity, diabetes mellitus, cigarette smoking, and the APOE e4 allele.2Livingston G. Huntley J. Sommerlad A. et al.Dementia prevention, intervention, and care: 2020 report of the Lancet Commission.Lancet. 2020; 396: 413-446Abstract Full Text Full Text PDF PubMed Scopus (516) Google Scholar Recent findings also suggest that impaired pulmonary function is associated consistently with approximately 40% elevation in later dementia risk.3Russ T.C. Kivimaki M. Batty G.D. Respiratory disease and lower pulmonary function as risk factors for dementia: a systematic review with meta-analysis.Chest. 2020; 157: 1538-1558Abstract Full Text Full Text PDF PubMed Scopus (7) Google Scholar Although there is mechanistic evidence to support this—including hypoxia from extended suboptimal ventilator function4Horsburgh K. Wardlaw J.M. van Agtmael T. et al.Small vessels, dementia and chronic diseases – molecular mechanisms and pathophysiology.Clin Sci. 2018; 132: 851-868Crossref Scopus (28) Google Scholar—crucially, given the observational nature of these studies, it is unclear whether this relationship is causal. An obstacle to drawing causal inference from such studies is the perennial problem of confounding, that characteristics of people with poorer pulmonary function differ from the unexposed in various ways that may explain the association. Investigators attempt to include as many relevant covariates as possible, but the possibility of confounding by unmeasured or imprecisely quantified factors is universal. Mendelian randomization (MR) has been seen as a possible remedy to this problem5Davey Smith G. Ebrahim S. ‘Mendelian randomization’: can genetic epidemiology contribute to understanding environmental determinants of disease?.Int J Epidemiol. 2003; 32: 1-22Crossref PubMed Scopus (1968) Google Scholar and has been extended to two-sample MR where genetic associations for the exposure and outcome are obtained from independent samples.6Hartwig F.P. Davies N.M. Hemani G. Davey Smith G. Two-sample Mendelian randomization: avoiding the downsides of a powerful, widely applicable but potentially fallible technique.Intl J Epidemiol. 2017; 45: 1717-1726Crossref Scopus (120) Google Scholar Accordingly, for the first time to our knowledge, we present a two-sample MR study to clarify whether the observed association between poorer pulmonary function and subsequent AD is causal. We ran a two-sample MR using summary data from the UKBiobank/SpiroMeta Consortium Genome-Wide Association Study (GWAS) comprising 400,102 individuals.7Shrine N. Guyatt A.L. Erzurumluoglu A.M. et al.New genetic signals for lung function highlight pathways and chronic obstructive pulmonary disease associations across multiple ancestries.Nat Genet. 2019; 51: 481-493Crossref PubMed Scopus (87) Google Scholar We derived two genetic instruments for lung function: FEV1 (liters) and FVC (liters). Of the 279 single nucleotide polymorphisms (SNPs) associated with lung function, but not smoking, only those related to the relevant trait with P < 5 × 10-8 and the same direction of effect in UKBiobank and SpiroMeta were used as genetic instruments. In addition, we included a more exploratory measure: the FEV1/FVC ratio, which has been used in the diagnosis of COPD whereby lower values are more suggestive of this condition.8Rabe K.F. Hurd S. Anzueto A. et al.Global strategy for the diagnosis, management, and prevention of chronic obstructive pulmonary disease: GOLD executive summary.Am J Respir Crit Care Med. 2007; 176: 532-555Crossref PubMed Scopus (4225) Google Scholar For the outcome, we used summary data from the most recent GWAS that included 21,982 people with AD and 41,944 control subjects.9Kunkle B.W. Grenier-Boley B. Sims R. et al.Genetic meta-analysis of diagnosed Alzheimer’s disease identifies new risk loci and implicates Aβ, tau, immunity and lipid processing.Nat Genet. 2019; 51: 414-430Crossref PubMed Scopus (567) Google Scholar The models used the TwoSampleMR R package.10Hemani G. Tilling K. Davey Smith G. Orienting the causal relationship between imprecisely measured traits using GWAS summary data.PLoS Genet. 2017; 13e100708Crossref Scopus (150) Google Scholar Because this study used publicly available data, no ethical approval was required. Table 1 shows the relationship between lung function and subsequent AD risk. There was no evidence of a causal effect of poorer lung function, using FEV1 or FVC, on AD risk (both P > .35). However, each SD increase in FEV1/FVC ratio (indicating superior lung function) was associated with an increased AD risk (OR, 1.12; 95% CI, 1.02 - 1.23; P = .016). The MR Egger intercept for the latter indicates little horizontal pleiotropy (β = 0.0002; P = .96) and the inverse-variance weighted Q-value (177.7; P = .08) suggests no substantial heterogeneity. Using the weighted median method gave a similar result (OR, 1.15; 95% CI, 1.00-1.31; P = .048).Table 1Estimates of the Association Between Pulmonary Function (FEV1, FVC, and FEV1/FVC Ratio) and Alzheimer Dementia From a Two-Sample Mendelian RandomizationVariableSNPs, nF-StatisticMethodOR (95% CI)PFEV117970.0Inverse variance weighted1.06 (0.94-1.19).355FVC13365.1Inverse variance weighted0.98 (0.85-1.14).815FEV1/FVC ratio154128.3Inverse variance weighted1.12 (1.02-1.23).016……MR Egger estimate1.11 (0.90-1.38).3188……Weighted median1.15 (1.00-1.31).0480SNP = single nucleotide polymorphisms. Open table in a new tab SNP = single nucleotide polymorphisms. We found that the observed association between lower pulmonary function and AD risk was not supported as being causal. Thus, it is possible that the original relationship resulted from confounding by one or more unmeasured or poorly measured confounders. Multiple candidates exist that include an adverse intrauterine environment that led to reduced maximal lung function, exposure to environmental factors (eg, tobacco smoke, atmospheric pollution) that affected lung function and development and socioeconomic factors (poverty, educational failure, and less-advantaged social class). In our systematic review and meta-analysis, most of the included studies took account of smoking and cardiovascular disease risk factors; slightly fewer included height.3Russ T.C. Kivimaki M. Batty G.D. Respiratory disease and lower pulmonary function as risk factors for dementia: a systematic review with meta-analysis.Chest. 2020; 157: 1538-1558Abstract Full Text Full Text PDF PubMed Scopus (7) Google Scholar Socioeconomic position was accounted for variably, and there was little coverage of the whole life course in terms of all included covariables. The FEV1/FVC ratio has not been examined routinely in relation to dementia risk.3Russ T.C. Kivimaki M. Batty G.D. Respiratory disease and lower pulmonary function as risk factors for dementia: a systematic review with meta-analysis.Chest. 2020; 157: 1538-1558Abstract Full Text Full Text PDF PubMed Scopus (7) Google Scholar However, we found a link, albeit a weak one, between higher pulmonary function captured by this measure and increased AD risk. This may possibly be explained by survivor bias, with participants with poorer pulmonary function dying before they reach late life; however, a false-positive result must also be considered. MR uses genetic variants that are allocated randomly at conception and therefore are generally independent of confounders that may otherwise bias an association when observational methods are used as proxies for environmental exposures. This assumes that genetic variants are (1) associated with the exposure, (2) only associated with the outcome of interest via their effect on the exposure, and (3) independent of confounders. It also relies on the exposure being measured accurately in the GWAS from which the instrument is derived. Pulmonary function was measured accurately with rigorous quality control in both UKBiobank (87.2% participants) and the individual studies of the SpiroMeta consortium.7Shrine N. Guyatt A.L. Erzurumluoglu A.M. et al.New genetic signals for lung function highlight pathways and chronic obstructive pulmonary disease associations across multiple ancestries.Nat Genet. 2019; 51: 481-493Crossref PubMed Scopus (87) Google Scholar Pathway analysis suggested biologic plausibility for the SNPs used as instruments, with enrichment of genes relating to extracellular matrix organization and ciliogenesis.7Shrine N. Guyatt A.L. Erzurumluoglu A.M. et al.New genetic signals for lung function highlight pathways and chronic obstructive pulmonary disease associations across multiple ancestries.Nat Genet. 2019; 51: 481-493Crossref PubMed Scopus (87) Google Scholar Furthermore, a genetic risk score comprising all 279 lung function SNPs predicted COPD.7Shrine N. Guyatt A.L. Erzurumluoglu A.M. et al.New genetic signals for lung function highlight pathways and chronic obstructive pulmonary disease associations across multiple ancestries.Nat Genet. 2019; 51: 481-493Crossref PubMed Scopus (87) Google Scholar It is unlikely that collider bias due to smoking and height adjustment in the lung function GWAS explains the observed association, because SNPs associated with smoking behavior were excluded, and a sensitivity analysis that excluded the 12 SNPs included in our instrument which were associated with height in UKBiobank did not affect our conclusions. The AD GWAS included 46 case-control studies from four consortia; rates of APOE e4 carriage are not reported.9Kunkle B.W. Grenier-Boley B. Sims R. et al.Genetic meta-analysis of diagnosed Alzheimer’s disease identifies new risk loci and implicates Aβ, tau, immunity and lipid processing.Nat Genet. 2019; 51: 414-430Crossref PubMed Scopus (567) Google Scholar These studies used various methods of ascertaining dementia, with multiple diagnostic criteria being applied. Some studies used clinical diagnoses, and some identified Alzheimer-type pathology after death. This variation is likely to affect the applicability of the GWAS findings in our analysis. In contrast to the instrumental AD variable used here, most observational studies use a more general category of “dementia.”3Russ T.C. Kivimaki M. Batty G.D. Respiratory disease and lower pulmonary function as risk factors for dementia: a systematic review with meta-analysis.Chest. 2020; 157: 1538-1558Abstract Full Text Full Text PDF PubMed Scopus (7) Google Scholar,9Kunkle B.W. Grenier-Boley B. Sims R. et al.Genetic meta-analysis of diagnosed Alzheimer’s disease identifies new risk loci and implicates Aβ, tau, immunity and lipid processing.Nat Genet. 2019; 51: 414-430Crossref PubMed Scopus (567) Google Scholar This lack of clarity is common, and the multiple diseases that cause the dementia syndrome (eg, AD, cerebrovascular disease, Lewy body disease, and Fronto-Temporal Lobar Degenerative syndromes) frequently are conflated. Depending on the methods used, clarifying an individual’s precise diagnosis can be challenging. For example, death certificates frequently record only the broad dementia syndrome. Thus, although we can conclude that there is no causal link between impaired pulmonary function and AD, our study sheds less light on potential links with other types of dementia. It is plausible that there may be a different relationship between pulmonary function and vascular dementia, for instance.