Sex-Specific Genetic Determinants of Asthma-COPD Phenotype and COPD in Middle-Aged and Older Canadian Adults: An Analysis of CLSA Data

Abstract

The etiology of sex differences in the risk of asthma-COPD phenotype and COPD is still not completely understood. Genetic and environmental risk factors are commonly believed to play an important role. This study aims to identify sex-specific genetic markers associated with asthma-COPD phenotype and COPD using the Canadian Longitudinal Study on Aging (CLSA) Baseline Comprehensive and Genomic data. There were a total of 1,415 COPD cases. Out of them, 504 asthma-COPD phenotype cases were identified. 20,524 participants without a diagnosis of asthma and COPD served as controls. We performed genome-wide SNP-by-sex interaction analysis. SNPs with an interaction p-value < 10⁻⁵ were included in a sex-stratified multivariable logistic regression for asthma-COPD phenotype and COPD outcomes. 18 and 28 SNPs had a significant interaction term p-value < 10⁻⁵ with sex in the regression analyses of asthma-COPD phenotype and COPD outcomes, respectively. Sex-stratified multivariable analysis of asthma-COPD phenotype showed that 7 SNPs in/near SMYD3, FHIT, ZNF608, RIMBP2, ZNF133, BPIFB1, and S100B loci were significant in males. Sex-stratified multivariable analysis of COPD showed that 8 SNPs in/near MAGI1, COX18, OSTC, ELOVL5, C7orf72 FGF14, and NKAIN4 were significant in males, and 4 SNPs in/near genes CAMTA1, SATB2, PDE10A, and LINC00908 were significant in females. An SNP in the ZPBP gene was associated with COPD in both males and females. Identification of sex-specific loci associated with asthma-COPD phenotype and COPD may offer valuable evidence toward a better understanding of the sex-specific differences in the pathophysiology of the diseases.

Keywords:

• Asthma-COPD phenotype
• chronic obstructive pulmonary disease
• sex
• polymorphisms
• CLSA

Introduction

Chronic obstructive pulmonary disease (COPD) is a chronic respiratory condition characterized by irreversible airflow limitation. It is a prevalent disease among middle-aged and elderly. In 2015, GINA and GOLD collaboration described a new chronic obstructive airway phenotype known as asthma-COPD overlap (1). The Asthma-COPD phenotype was characterized as a chronic obstructive airway phenotype with persistent airflow limitation and a clinical presentation of features associated with asthma and COPD. A number of studies have shown that patients with the overlap phenotype present with more frequent airflow obstruction symptoms, more exacerbation, lower quality of life, recurrent hospitalization, and more medical utilization than patients with classic COPD or asthma alone (2–7). In addition, the lack of clinical data from randomized controlled trials explicitly addressing treatment regimens for patients with the overlap phenotype makes definitive treatment difficult.

Many epidemiological studies have found significant sex differences in the risk of asthma-COPD phenotype and COPD, with females having a higher prevalence than males (8–12). COPD was initially considered a disease of middle-aged and older men; however, its prevalence and hospitalization rates in women have increased in the last 20 years (13–15). Although the etiology of sex differences is not fully understood, it is commonly believed that both environmental and genetic factors play an important role.

Cigarette smoking is the most consistent risk factor for COPD, especially in developed nations; exposure to biomass and noxious pollutants impacts COPD risk in developing countries (16,17). Apart from the strong influence of environmental factors on COPD risk, various genetic risk factors have been associated with COPD susceptibility (18–22). As genetic and environmental risk factors play crucial roles in COPD development, studies have revealed that COPD’s clinical presentation and susceptibility vary by gender (23–31). Women who were ex-smokers and current smokers had a higher risk of airflow obstruction than men who were ex-smokers and current smokers exposed to the same dose of tobacco (31). A study observed that females with early-onset COPD and < 20 pack-years of smoking had a greater reduction in lung function (FEV₁% predicted) and more severe disease than males (24). Females with COPD have been reported to have more dyspnea, less expectoration, worse exercise capacity, and more frequent diagnosis of concomitant asthma than males (29). Whereas in another study, males with COPD were reported to have more emphysema and lesser airway remodeling than females (28). Epidemiological studies have linked several socio-demographic, socio-economic, and lifestyle factors to asthma-COPD phenotype risk (8,9). In a study of the risk of asthma-COPD phenotype in Aboriginal people, daily smoking, working more than 40 h per week, and being separated, widowed, or divorced were associated with the risk of asthma-COPD phenotype in females but not in males (9). In addition, Aboriginal women who smoked at home compared to those who did not were almost three times more likely to be associated with asthma-COPD phenotype risk. However, this trend was not observed in men (9). Similarly, in another population-based study, women who smoked cigarettes for ten years or more had a higher prevalence of asthma-COPD phenotype than men (12). Nevertheless, the underlying mechanisms for the sex-specific differences in asthma-COPD phenotype and COPD remain unclear.

Previous research has identified various genetic loci associated with COPD and asthma-COPD phenotype (18–22,32–34). However, no genetic association studies have examined sex-specific genetic risk factors for asthma-COPD phenotype while including socio-demographic, socio-economic, and lifestyle factors in their analyses. Using the Canadian Longitudinal Study on Aging (CLSA) baseline questionnaire and genomic dataset, this study aimed to explore sex-specific genetic differences in asthma-COPD phenotype and COPD while accounting for various socio-demographic, lifestyle, socio-economic, and environmental factors.

Materials and method

Study design and population

The Canadian Longitudinal Study on Aging (CLSA) is a population-based cohort of 51,338 Canadians aged 45 to 85 years old. The CLSA is made up of two complementary cohorts: the Tracking cohort, which consisted of 21,241 people who were questioned over the phone, and the Comprehensive cohort, which consisted of 30,097 people who provided baseline data via an in-person home interview as well as other questionnaires, tests, physical measurements, and biospecimen (blood and urine) collected at the data collection sites (35). Our study focused on a subset of 26,622 subjects from the Comprehensive Cohort who were genotyped using the Affymetrix UK Biobank Axiom array (36).

Outcomes

We defined COPD and asthma from the positive responses to the questions, “Has a doctor told you that you have/had any of the following: emphysema, chronic bronchitis, chronic obstructive pulmonary disease (COPD), or chronic changes in the lungs due to smoking?” and “Has a doctor ever told you had asthma?”, respectively. Participants with positive responses to both self-reported physician-diagnosed asthma and COPD were categorized as asthma-COPD phenotype. Control subjects were those who responded “no” to self-reported physician diagnoses of asthma and COPD. Participants who self-reported a physician diagnosis of only asthma and those with missing responses were excluded from each study group. Asthma-COPD phenotype cases were also excluded from the COPD case-control group. On the other hand, COPD cases were excluded from the asthma-COPD phenotype case-control group (see Figure 1(A)).

Figure 1. Study flow charts. (A) Workflow for the genome-wide SNP-by-sex interaction and sex-stratified analysis. (B) Flow chart for sample and marker quality control.

Predictor variables

The CLSA questionnaire contains a wide range of socio-demographic and socio-economic data, information on lifestyle and health behaviors, as well as clinical and physical data. Our study considered the following potential confounders such as age group (45 to 54 years, 55 to 64 years, 65 to 74 years, and 75 years and above), biological sex, marital status (single or never married, married or in a common-law relationship, widowed/divorced/separated), an education level (less than post-secondary education, post-secondary but not university education, and university education/others), total personal income and total household income (Less than $20,000, $20,000 to less than $50,000, $50,000 to less than $100,000, $100,000 or more), province of recruitment (Prairies, British Columbia, Eastern provinces, Ontario and Quebec), retirement status (Retired completely and not retired/partly retired), smoking status (current, never, and former smokers), homeownership (owned and rented/others), and urban/rural dwelling.

Genotyping and quality control

Genotyping for 794,409 genetic markers was done using Affymetrix UK Biobank Axiom array. Quality control of genetic markers and samples was adequately described in another paper (36). SNPs and Samples that failed the QC metrics were removed. All genomic positions were in reference to the human genome build GRCh37/hg19. Using PLINK v1.90b6.25 (37), we excluded samples with discordant sex information and more than 5% genotype missingness. Autosomal SNPs were extracted, SNPs with less than 99% call rate, minor allele frequency less than 1% (MAF < 0.01), and SNPs deviating from the Hardy Weinberg equilibrium threshold of 1e-10 were excluded. In order to obtain SNPs that are in approximate linkage equilibrium with each other, we generated a subset of LD pruned SNPs using PLINK’s Indep-pairwise command (Indep-pairwise 50, 5, 0.5). Following quality control, we had 504 asthma-COPD phenotype cases, 911 COPD cases, 20,524 controls, and 416,562 SNPs for genome-wide SNP-by-sex interaction analysis (see Figure 1(AB)).

Genetic and statistical analysis

In order to investigate sex-specific variants for asthma-COPD phenotype and COPD outcomes, genome-wide SNP-by-Sex interaction was examined in logistic regression for the asthma-COPD phenotype and COPD, respectively, after adjusting for age, sex, smoking status, and the first four principal components of genetic ancestry using PLINK v1.90b6.25 (37). Those SNPs with an interaction p value less than 10⁻⁵ were further examined in a sex-stratified analysis for asthma-COPD phenotype and COPD outcomes using survey-specific logistic regression (Proc Surveylogistic) in SAS version 9.4. The survey-specific SAS procedure allows us to apply sampling weights, incorporate complex survey designs, and control for potential confounders. For SNPs in high linkage disequilibrium (D’>0.80), the SNP with the highest polymorphism information content (PIC) value was chosen.

For descriptive statistics of continuous and categorical variables, mean and standard error and frequency and percentage were used to describe the study population. CLSA’s trimmed inflation and analytic weights (CLSA Sample Weights Version 1.2) were used for descriptive and regression analysis. In order to identify potential confounders to be included in the final sex-stratified analysis, a univariate logistic regression model was used to identify significant risk factors. The predictor variables with a p value ≤ 0.20 from the univariate analysis were entered into an interim multivariate model. The least significant variable was then removed one at a time until only variables with significant p value (p ≤ 0.05) and clinically important factors remained in the model.

In order to discover genetic risk factors that are specific to either sex, a survey logistic regression was conducted with sex as a domain factor. For the SNPs with an interaction term p value less than 10⁻⁵, we included one SNP at a time and the first four principal components into the final model while controlling for potential environmental, demographic, and socio-economic risk factors. In addition to the four principal components, the final model for the asthma-COPD phenotype included adjustments for age, smoking status, marital status, homeownership, total household income, education level, and retirement status. For the COPD outcome, the variables adjusted for were age, smoking status, province of recruitment, marital status, homeownership, total personal income, and retirement status.

Three inheritance models (dominant, recessive, and additive) for each variant were considered. The model with the lowest Akaike information criterion values (AIC) was regarded as the best-fitting model. The results from the best-fitted model were presented.

In order to account for multiple comparisons, the Bonferroni correction was applied in the sex-stratified analyses of asthma-COPD phenotype and COPD based on the number of SNPs with a significant interaction term with the sex variable. We created regional association plots using Locuszoom (38) for the significant SNPs from the sex-stratified analyses. The strength of association was reported as odds ratios with 95% confidence intervals.

Results

Characteristics of the study population

Table 1 depicts the characteristics of the study population stratified by sex. A total of 26622 subjects (13,343 males and 13,279 females) represented 3,295,958 participants in the weighted sample (1,670,060 males and 1,625,898 females).

Table 1. Descriptive characteristics of the study population.

Variables	Groups	Males (n = 13,343)	Females (n = 13,279)	p-value
Asthma + COPD, N(%)	Yes	218 (1.4)	327 (2.3)	<0.0001
COPD, N(%)	Yes	451 (2.6)	508 (3.0)	<0.0001
Age, mean ± SEM		59.1 ± 0.1	59.9 ± 0.1	<0.0001
Age group (%)	45 to 54	44.3	39.7	<0.0001
	55 to 64	29.8	30.2
	65 to 74	15.8	18.3
	75 and above	10.2	11.7
Smoking Status (%)	Current Smoker	10.0	8.5	<0.0001
	Never Smoker	47.6	51.4
	Former Smoker	42.4	40.1
Marital status (%)	Single/never married	8.1	8.2	<0.0001
	Married/Living in common law	82.3	70.1
	Widowed/Separated/Divorced	9.6	21.6
Urban or Rural Dwelling (%)	Urban core	86.6	85.0	0.0028
	Rural/not Urban core	13.4	15.0
Homeownership (%)	Owned	86.6	84.9	0.0029
	Rented/Others	13.2	15.0
Total household income (%)	Less than $20,000	3.4	5.6	<0.0001
	$20,000 to less than $50,000	15.1	22.2
	$50,000 to less than $100,000	32.5	34.2
	$100,000 or more	49.1	38.0
Total personal income (%)	Less than $20,000	7.8	22.9	<0.0001
	$20,000 to less than $50,000	27.9	39.9
	$50,000 to less than $100,000	40.7	28.9
	$100,000 or more	23.6	8.2
Retirement Status (%)	Retired completely	29.0	38.0	<0.0001
	Not retired/partly retired	71.0	62.0
Province of recruitment (%)	Prairies	22.2	19.1	<0.0001
	British Columbia	28.9	30.0
	Eastern province^a	5.2	6.2
	Ontario	13.5	13.6
	Quebec	30.2	31.1
Highest Education Status (%)	No post-secondary Education	7.4	8.4	<0.0001
	Non-university post-secondary	31.4	34.8
	University post-secondary	61.4	56.8

Number of participants in the unweighted sample: 26622 (13,343 males and 13,279 females).

represent 3,295,958 participants (1,670,060 males and 1,625,898 females) in the weighted sample.

451 males and 508 Females with COPD in the unweighted sample represent 43722 males and 48348 females in the weighted sample.

218 males and 327 Females with asthma + COPD in the unweighted sample represent 24068 males and 38085 females in the weighted sample.

^a: Newfoundland and Labrador, Nova Scotia; COPD: chronic obstructive pulmonary disease; Asthma + COPD: Asthma-COPD phenotype; SEM: standard error of the mean.

In males and females, the prevalence of asthma-COPD phenotype was 1.4% and 2.3%, respectively. The prevalence of COPD, however, was 2.6% for males and 3.0% for females.

Males and females differed significantly in the distribution of age, age groups, smoking status, marital status, urban or rural dwelling, homeownership, total household income, total personal income, retirement status, province of recruitment, and highest education status.

Approximately 52.4% of males were smokers (current smokers: 10% and former smokers:42.4%), whereas, in females, 48.6% had a positive smoking history (current smokers: 8.5% and former smokers: 40.1%). In comparison to males, there were more females in the age groups of 55 to 64 (30.2% vs. 29.8%), 65 to 74 (18.3% vs. 15.8%), and 75 and over (11.7% vs. 10.2%). In contrast, more males (44.3%) than females (39.7%) were in the age group 45-54. A significantly higher proportion of females (21.6%) were widowed/divorced/separated than males (9.6%).

Males were more likely than females to have a total household income (49.1% vs. 38.0%) and a total personal income (23.6% vs. 8.2%) of $100,000 or more. In contrast, more females than males earned less than $20,000 (5.6% vs. 3.4%) and less than $50,000 (22.2% vs. 15.1%) for total household income. Similarly, for total personal income, a higher proportion of females than males earned less than $20,000 (22.9% vs. 7.8%) and less than $50,000 (39.9% vs. 27.9%). Females (38.0%) tended to be retired more than males (29.0%). In contrast, for the highest education status, males (61.4%) were more likely to have post-secondary education at the university level than females (56.8%).

Genome-wide SNP-by-sex interaction

In the SNP-by-sex interaction analysis, we examined after quality control, 416,562 SNPs, 21435 subjects for COPD (cases: n = 911, controls: n = 20524) and 21028 subjects for the asthma-COPD phenotype (cases: n = 504, controls: n = 20524) (see Figure 1(A,B)).

There were 18 and 28 distinct SNPs from the SNP-by-sex interaction terms with p value less than 10¯⁵ for asthma-COPD phenotype and COPD, respectively (see Tables 2 and 3). No indication of population stratification based on the genomic inflation factor was observed in the genome-wide interaction analysis for both phenotypes (see Figure 2(A,B)).

Figure 2. Quantile-Quantile plots (Q-Q plot) showing the distribution of observed versus Expected pvalue. (A) Q-Q plot for the genome-wide SNP-by-sex interaction for asthma-COPD phenotype. Genomic inflation factor, lambda (λ) = 1.00. (B) Q-Q plot for the genome-wide SNP-by-sex interaction for COPD. Genomic inflation factor, lambda (λ) = 1.00.

Table 2. 18 Signals of the SNP-by-Sex Interactions with pvalue less than 10^¯5 for asthma-COPD phenotype.

Chr	SNP	Effect Allele	MAF	Major Allele	OR	Interaction P value	Gene/Nearest Gene	Consequence
1	rs619928	A	0.387	G	0.57	5.74E-05	SPSB1	Intron variant
1	rs6666629	T	0.061	C	2.77	8.47E-05	HSD52	Intron variant
1	rs12062324	G	0.059	A	2.99	8.93E-05	RPS6KC1	Intergenic variant
1	rs61140467*	A	0.285	G	1.76	9.78E-05	SMYD3	Intron variant
1	rs11799559*	A	0.458	G	1.68	8.55E-05	SMYD3	Intron variant
2	rs77602394	G	0.064	A	2.96	3.99E-05	RAD51AP2	Intergenic variant
3	rs3821479	C	0.430	T	0.57	4.51E-05	FHIT	Intron variant
4	rs72901492	A	0.107	C	0.40	3.83E-05	NDST4	Intergenic variant
5	rs77800494	T	0.043	G	3.84	6.73E-05	ZNF608	Intergenic variant
7	rs7782108	C	0.190	T	0.48	8.00E-05	CARD11	Intron variant
7	rs34331963	C	0.159	T	2.05	4.52E-05	SDK1	Intergenic variant
8	rs6986253	A	0.254	G	1.81	9.60E-05	LRP12	Intron variant
12	rs11061082	A	0.399	G	1.71	6.16E-05	RIMBP2	Intergenic variant
14	rs2582577	T	0.199	C	1.88	7.47E-05	C14orf79	Intergenic variant
20	rs926718	A	0.374	G	0.55	2.18E-05	ZNF133	Intron variant
20	rs1884882	C	0.282	T	1.82	2.66E-05	BPIFB1	Intergenic variant
20	rs6013630	T	0.122	C	2.34	4.26E-05	TSHZ2	Intron variant
21	rs1051169	C	0.329	G	1.72	9.53E-05	S100B	Synonymous variant

SNP-by-Sex Interaction GWAS on asthma-COPD phenotype, controlling for age, sex, smoking status, and top 4 principal components.

*: SNP rs11799559 was in high linkage disequilibrium (D’=0.99) with rs61140467.

Chr: chromosome; SNP: single nucleotide polymorphism; MAF: minor allele frequency; OR: odds ratio.

Genes contain variant or are located within 500 kb from variants.

Table 3. 28 Signals of the SNP-by-Sex Interactions with pvalue less than 10⁻⁵ for COPD.

Chr	SNP	Effect Allele	MAF	Major Allele	OR	Interaction P value	Gene/Nearest Gene	Consequence
1	rs12025895	C	0.44	T	1.48	6.74E-05	CAMTA1	Intron variant
1	rs6429144	T	0.24	C	1.63	3.17E-05	CHRM3	Intron variant
2	rs10931835	C	0.43	T	1.52	3.01E-05	SATB2	Intergenic variant
2	rs7557339	T	0.39	C	0.65	1.85E-05	SLC19A3	Intergenic variant
3	rs13326145	T	0.05	C	0.32	3.86E-05	MAGI1	Intron variant
3	rs12633737	A	0.31	G	1.57	2.00E-05	ALCAM	Intergenic variant
4	rs56334611	G	0.27	A	0.63	3.69E-05	COX18	Intergenic variant
4	rs6816344	T	0.28	A	0.62	1.65E-05	COX18	Intergenic variant
4	rs73838466^a	T	0.15	C	1.86	9.35E-06	OSTC	Intron variant
4	rs17039240^a	T	0.26	C	1.57	5.84E-05	OSTC	Intergenic variant
4	rs11947441	C	0.03	T	0.19	9.65E-05	FRG1	Intergenic variant
5	rs76189059	A	0.06	G	0.38	8.89E-05	NA	Intergenic variant
6	rs6935314	C	0.33	T	0.64	3.93E-05	ELOVL5	Intergenic variant
6	rs117896837	T	0.02	C	4.80	9.10E-05	FAM46A	Intergenic variant
6	rs220806	T	0.31	C	1.58	2.78E-05	PDE10A	Intron variant
7	rs59794319	G	0.08	A	0.48	9.09E-05	NXPH1	Intergenic variant
7	rs10488084	C	0.08	A	0.48	9.95E-05	PLEKHA8	Intron variant
7	rs1911770	T	0.43	C	1.51	4.81E-05	ZPBP	Intron variant
7	rs13225543	T	0.41	C	0.67	5.48E-05	C7orf72	Intergenic variant
8	rs112593560	T	0.03	C	3.69	8.14E-05	KCNQ3	Intron variant
9	rs4076140	G	0.08	A	2.18	3.39E-05	TRPM3	Intron variant
9	rs74964803	G	0.05	T	0.39	7.95E-05	MED27	Intergenic variant
13	rs66468120	A	0.05	G	2.64	5.02E-05	RNF219-AS1	Intron variant
13	rs12869252	G	0.24	A	0.64	9.73E-05	FGF14	Intron variant
14	rs12879329	G	0.07	A	2.24	3.85E-05	NPAS3	Intron variant
18	rs77625370	T	0.05	C	2.75	7.03E-05	LINC00908	Intergenic variant
20	rs1891778	G	0.48	A	1.52	2.60E-05	CDH4	Intron variant
20	rs6090327	C	0.17	T	0.59	8.39E-05	NKAIN4	3 prime UTR variant

SNP-by-Sex Interaction GWAS on COPD, controlling for age, sex, smoking status, and top 4 principal components.

^aSNP rs17039240 was in high linkage disequilibrium (D’=0.99) with rs73838466,.

Chr: Chromosome; SNP: Single Nucleotide Polymorphism; MAF: Minor allele frequency; OR: Odds ratio NA: No annotation.

Genes contain variants or are located within 500 kb from variants.

SNP rs926718, an intronic variant in the ZNF133 gene, had the lowest interaction p value (p = 2.18 × 10¯⁵) for the asthma-COPD phenotype (Table 2). Regarding COPD outcome, the SNP with the lowest p value was an intronic variant [rs73838466 (p = 9.35 × 10¯⁶)] in the OSTC gene (Table 3).

Sex stratified analysis for asthma-COPD phenotype

Out of the 18 SNPs with a significant SNP-by-sex interaction term, pvalue less than 10¯⁵ for asthma-COPD phenotype, the polymorphisms rs61140467, which was in high LD (D’=0.99) with rs11799559, was excluded from the analysis due to its lower polymorphic information content. Table 4 demonstrates the results of significant SNPs from the sex-stratified multivariate analysis of asthma-COPD phenotype after Bonferroni correction (p ≤ 0.003 = 0.05/17). 7 SNPs were significant in males, and no SNP was significant in females (see Table 4 and Supplementary Tables S3 and S4). The regional plots of these 7 male-specific SNPs associated with asthma-COPD phenotype are shown in Figure 5.

Table 4. Result of sex-stratified analysis for asthma-COPD phenotype.

Chr	Gene/Nearest Gene	SNP (A1/A2)	Male		Female
			OR (95% CI)	p-value	OR (95% CI)	p-value
1	SMYD3	rs11799559 (A/G)^a	1.56 (1.18 − 2.07)zz^A	0.0021*	PM: 0.69 (0.50 − 0.94)^DWT: 1	0.0185
3	FHIT	rs3821479 (C/T)	0.55 (0.41 − 0.73)^A	<0.0001*	PM: 0.79 (0.57 − 1.08)^DWT: 1	0.141
5	ZNF608	rs77800494 (T/G)	2.08 (1.30 − 3.34)^A	0.0025*	0.42 (0.20 − 0.86)^A	0.0178
12	RIMBP2	rs11061082 (A/G)	PM: 2.15 (1.37 − 3.37)^RWT: 1	0.0009*	PM: 0.75 (0.48 − 1.16)^RWT: 1	0.199
20	ZNF133	rs926718 (A/G)	0.58 (0.43 − 0.79)^A	0.0005*	1.32 (1.07 − 1.64)^A	0.0101
20	BPIFB1	rs1884882 (C/T)	PM: 2.24 (1.32 − 3.81)^RWT: 1	0.0029*	0.93 (0.73 − 1.20)^A	0.5823
21	S100B	rs1051169 (C/G)	1.56 (1.18 − 2.06)^A	0.0019*	PM: 0.74 (0.55 − 1.01)^DWT: 1	0.0562

Sex-stratified multivariate logistic regression of top interaction single nucleotide polymorphisms for asthma-COPD phenotype, adjusting for age, smoking status, marital status, homeownership, total household income, education level, retirement status, and first four principal components.

^aSNP rs11799559 was in high linkage disequilibrium (D’=0.99) with rs61140467; hence rs61140467 was removed from sex-stratified analysis due to lower polymorphic information content (PIC).

*: statistically significant after adjusting for multiple comparisons (p ≤ 0.003).

^A: Additive; ^D: Dominant; ^R: Recessive; PM: Polymorphism; WT: wild type; Chr: chromosome.

SNP: single nucleotide polymorphism; A1/A2: minor allele/major allele; OR: odds ratio; CI: confidence interval.

Genes contain variants or are located within 500 kb from variants.

The top significant SNP amongst the males was rs3821479 in the FHIT gene [OR = 0.55, p < 0.0001, 95% CI (0.41 − 0.73)]. The polymorphisms rs926718 in the ZNF133 gene, which had the lowest SNP-by-sex interaction pvalue, was significantly associated with asthma-COPD phenotype amongst the males [OR = 0.58, p = 0.0005, 95% CI (0.43 − 0.79)]. Howbeit, the effects of these variants were protective (Table 4 and Figure 3).

Figure 3. Forest plot showing the direction of effects of the male-specific SNPs associated with asthma-COPD phenotype from the sex-stratified analysis.

As illustrated in Table 4, SNP rs11061082 in/near the RIMBP2 gene [OR = 2.15, p = 0.0009, 95% CI (1.37 − 3.37)] and SNP rs1884882 near BPIFB1 gene [OR = 2.24, p = 0.0029, 95% CI (1.32 − 3.81)] under the recessive inheritance model were significantly associated with asthma-COPD phenotype risk in males. While in the additive model, SNPs rs11799559 [OR = 1.56, p = 0.0021, 95% CI (1.18 − 2.07)] in the SMYD3 gene, rs1051169 [OR = 1.56, p = 0.0019, 95% CI (1.18 − 2.06)] in S100B gene and rs77800494 [OR = 2.08, p = 0.0025, 95% CI (1.30 − 3.34)] near ZNF608 gene were significantly associated with asthma-COPD phenotype risk in males (see Figure 3).

Sex-stratified analysis for COPD

Out of the 28 SNPs with a significant SNP-by-sex interaction term, pvalue less than 10¯⁵ for COPD, the polymorphism rs73838466, which was in high LD (D’=0.99) with rs17039240, was excluded from the analysis due to its lower polymorphic information content. Table 5 displays the results of significant SNPs from the sex-stratified multivariate analysis of COPD after Bonferroni correction (p ≤ 0.002 = 0.05/27).

Table 5. Result of sex-stratified analysis for COPD.

Chr	Gene/Nearest Gene	SNP (A1/A2)	Male		Female
			OR (95% CI)	p-value	OR (95% CI)	p-value
1	CAMTA1	rs12025895 (C/T	PM: 1.29 (0.99 − 1.68)^D WT: 1	0.0622	0.72 (0.61 − 0.84)^A	<0.0001*
2	SATB2	rs10931835 (C/T)	PM: 1.20 (0.92 − 1.57)^D WT: 1	0.1761	PM: 0.64 (0.50 − 0.81)^D WT: 1	0.0002*
3	MAGI1	rs13326145 (T/C)	0.38 (0.22 − 0.65)^A	0.0005*	PM: 0.61 (0.08 − 4.58)^R WT: 1	0.6267
4	COX18	rs56334611 (G/A)	0.69 (0.57 − 0.84)^A	0.0002*	PM: 1.37 (1.09 − 1.73)^D WT: 1	0.0073
4	COX18	rs6816344 (T/A)	0.71 (0.58 − 0.85)^A	0.0003*	1.26 (1.04 − 1.52)^A	0.0176
4	OSTC	rs17039240 (T/C)^a	PM: 2.48 (1.62 − 3.79)^R WT: 1	<0.0001*	0.83 (0.69 − 1.01)^A	0.0571
6	ELOVL5	rs6935314 (C/T)	0.74 (0.61 − 0.89)^A	0.0017*	PM: 1.30 (1.03 − 1.65)^D WT: 1	0.0276
6	PDE10A	rs220806 (T/C)	PM: 1.19 (0.93 − 1.51)^D WT: 1	0.1624	0.70 (0.58 − 0.84)^A	0.0001*
7	ZPBP	rs1911770 (T/C)^b	PM: 1.64 (1.24 − 2.16)^R WT: 1	0.0005*	0.76 (0.64 − 0.91)^A	0.002*
7	C7orf72	rs13225543 (T/C)	PM: 0.65 (0.51 − 0.83)^D WT: 1	0.0005*	PM: 1.36 (1.05 − 1.77)^D WT: 1	0.0196
13	FGF14	rs12869252 (G/A)	PM: 0.68 (0.53 − 0.87)^D WT: 1	0.002*	1.24 (0.99 − 1.56)^A	0.0194
18	LINC00908	rs77625370 (T/C)	1.61 (1.12 − 2.30)^A	0.0094	0.45 (0.28 − 0.74)^A	0.0014*
20	NKAIN4	rs6090327 (C/T)	0.65 (0.51 − 0.84)^A	0.001*	1.38 (1.11 − 1.71)^A	0.0038

Sex-stratified multivariate survey logistic regression of top interaction single nucleotide polymorphisms for COPD, adjusting for age, smoking status, province of recruitment, marital status, homeownership, total personal income, retirement status, and first four principal components.

^a: SNP rs17039240 was in high linkage disequilibrium (D’=0.99) with rs73838466; hence rs73838466 was removed from sex-stratified analysis due to lower polymorphic information content (PIC).

*:statistically significant after adjusting for multiple comparisons (p ≤ 0.002).

^bsignificant in both males and females.

^A: additive; ^D: dominant; ^R: recessive; PM: polymorphism; WT: wild type; Chr: chromosome; SNP: single nucleotide polymorphism; A1/A2: minor allele/major allele; OR: odds ratio; CI: confidence interval.

Genes contain variants or are located within 500 kb from variants.

As illustrated in Table 5, and Supplementary Tables (S1 and S2), eight variants (rs13326145, rs56334611, rs6816344, rs17039240, rs6935314, rs13225543, rs12869252, and rs6090327) mapped in/near MAGI1, COX18, OSTC, ELOVL5, C7orf72 FGF14, and NKAIN4 genes respectively were significantly associated with COPD in males. Whereas four SNPs (rs12025895, rs10931835, rs220806, and rs77625370) in/near CAMTA1, SATB2, PDE10A, and LINC00908 genes were significantly associated with COPD amongst females. Figure 6 presents a regional plot of significant sex-specific SNPs associated with COPD and their genomic positions relative to genes or nearby genes. The SNP rs1911770 in gene ZPBP was found to have a statistically significant association with COPD in males [OR = 1.64, p = 0.0005, 95% CI (1.24 − 2.16)] and females [OR = 0.76, p = 0.002, 95% CI (0.64 − 0.91)]; indicating that it increased the risk of COPD in males while conferring a protective effect in females (see Figure 4(A)).

Figure 4. Forest plots showing the direction of effects for the sex-specific SNPs associated with COPD from the sex-stratified analysis. (A) Males. (B) Females. Only significant SNPs after the Bonferroni adjustment were plotted.

Figure 5. Regional plots for the 7 male-specific SNPs associated with asthma-COPD phenotype showing their respective genomic locations. (A) rs11799559 in SMYD3 gene. (B) rs3821479 in FHIT gene. (C) rs77800494 near ZNF608 gene. (D) rs11061082 near RIMBP2 gene. (E) rs926718 in ZNF133 gene. (F) rs1884882 near BPIFB1 gene. (G) rs1051169 in S100B gene.

Figure 6. Regional plots for the sex-specific SNPs associated with COPD showing their respective genomic locations (A) rs12025895 in CAMTA1 gene. (B) rs10931835 near SATB2 gene. (C) rs13326145 in MAGI1 gene. (D) rs56334611 near COX18 gene. (E) rs6816344 near COX18 gene. (F) rs17039240 near OSTC gene. (G) rs6935314 near ELOVL5 gene. (H) rs220806 in PDE10A gene. (I) rs1911770 in ZPBP gene. (J) rs13225543 near C7orf72 gene. (K) rs12869252 in FGF14 gene. (L) rs77625370 near LINC00908 gene. (M) rs6090327 in NKAIN4 gene.

Out of the eight male-specific SNPs, an intergenic SNP rs17039240 near the OSTC gene was significantly associated with COPD risk in males [OR= 2.48, p < 0.0001, 95% CI (1.62 − 3.79)] but not in females (see Figure 4(A)). The top female-specific SNP was rs12025895 in the CAMTA1 gene [OR = 0.72, p < 0.0001, 95% CI (0.61 − 0.84)]. All the variants significantly associated with COPD amongst females had protective effects, with ORs ranging from 0.45 to 0.76 (see Figure 4(B)).

Discussion

This study found 18 and 28 distinct signals for a genome-wide SNP-by-Sex interaction on asthma-COPD phenotype and COPD outcomes, respectively. The SNPs with the lowest SNP-by-Sex interaction pvalue for asthma-COPD phenotype and COPD outcomes were located in the intronic regions of the ZNF133 and OTSC genes at 20p11.23 and 4q25 cytogenetic positions, respectively.

We discovered seven male-specific variants in or adjacent SMYD3, FHIT, ZNF608, RIMBP2, ZNF133, BPIFB1, and S100B genes that were significantly associated with asthma-COPD phenotype. Five of the seven SNPs (rs11799559, rs77800494, rs11061082, rs1884882, and rs1051169) were associated with an increased risk, with ORs ranging from 1.56 to 2.24, while the remaining two SNPs (rs3821479 and rs926718) showed protective effects, with OR of 0.55 and 0.58.

A total of eight male-specific variants (rs13326145, rs56334611, rs6816344, rs17039240, rs6935314, rs13225543, rs12869252, and rs6090327) in/near the MAGI1, COX18, OSTC, ELOVL5, C7orf72 FGF14, and NKAIN4 genes, and 4 female-specific variants (rs12025895, rs10931835, rs220806 and rs77625370) in/near the CAMTA1, SATB2, PDE10A, and LINC00908 genes were significantly associated with COPD. In addition, rs1911770 in the ZPBP gene, was associated with an increased risk of COPD in males and a reduced risk in females.

To the best of our knowledge, this is the first study to look at sex-specific genetic risk factors for asthma-COPD phenotype and COPD while also taking into account environmental, lifestyle, socio-economic, and socio-demographic factors. Previous genetic and gene-based association studies had demonstrated sex-specific genetic effects on COPD and asthma (39–41); however, these studies did not consider the influence of socio-demographic and socio-economic factors.

In this study, the strongest male-specific associations for asthma-COPD phenotype risk were from SNPs (rs1884882, rs11061082 rs77800494, rs11799559, rs1051169) in/near BPIFB1, RIMBP2, ZNF608, SMYD3, and S100B genes. The BPIFB1 gene on chromosome 20q11.21 encodes a protein secreted by goblet cells in the airway epithelium, trachea, submucosal glands of airways, and nasal cavities (42). This protein is believed to play a role in innate immunity against inhaled toxins and pathogens. BPIFBI is upregulated in several respiratory diseases. For example, after a segmental allergen challenge, higher levels of BPIFB1 were found in bronchoalveolar lavage fluid in asthmatic patients (43). In addition, BPIFB1 levels in the sputum of smokers with COPD were significantly higher than in smokers and nonsmokers without COPD (44). De Smet et al. (45) observed that the mRNA expression levels of BFIFB1 amongst COPD subjects were positively correlated with disease severity and that smokers with COPD had higher BPIFB1 mRNA and protein expression levels in lung tissue and airway epithelium than nonsmokers and smokers without COPD. BPIFB1 levels have also been found to be significantly inversely correlated to FEV₁% predicted, FEV_1/FVC ratio, and diffusing capacity of the lung for carbon monoxide (DLCO), all of which are proxies for COPD disease severity and emphysema (44,45). Additionally, human and animal studies have demonstrated that males are more prone to emphysematous changes than females (28,46).

Polymorphisms of RIMBP2, FHIT, and ZNF608 genes on 12q24.33, 3p14.2, and 5q23.2 cytogenetic positions have been associated with testosterone levels (47). Several studies have suggested the influence of sex hormones on lung diseases and inflammatory responses of the lungs to pathogens and inhaled toxins, including cigarette smoke and pollutants (48,49). Androgen (testosterone) is thought to have anti-inflammatory effects, which are mediated by interaction with androgen receptors (AR) and control the expression of transcriptions (50). For example, one study found that testosterone reduced pulmonary epithelial inflammation in rats with COPD (51). As androgens decline with advancing age, the binding of androgen and AR complexes to transcriptions might be attenuated. These might lead to altered expressions of genes, increased pro-inflammatory cytokines, and chronic inflammatory diseases.

The S100B gene on chromosome 21q22.3 belongs to the S100 family of proteins, which regulates calcium balance, cell apoptosis, migration, proliferation, differentiation, energy metabolism, and inflammation. Furthermore, research has indicated that S100B is a major ligand of the receptor for advanced glycation end products (RAGE), a pattern recognition receptor that is expressed in alveolar type I and type II epithelial cells, bronchiolar epithelium, and alveolar macrophages (52,53). S100 protein interaction with RAGE activates NF-κB, causing the production of pro-inflammatory cytokines and the migration of neutrophils, monocytes, and macrophages (54). For instance, In an in-vitro study, S100B was shown to stimulate the secretion of TNF-alpha and IL-6 in Alveolar Type-I (AT-I) derived cells from the pulmonary tissue of male fetuses of Han-Wistar rats (55). The SMYD3 gene, a protein-coding gene known for its histone methyltransferase activity, is more expressed in males’ dorsolateral frontal cortex and anterior cingulate cortex of the brain than in females (56). Other histone-encoding genes have been shown to be more expressed in males than females in the heart, kidney, and colon (56).

In this study, the strongest male-specific SNP for COPD was rs17039240. This SNP is an intergenic variant near the OSTC gene. OSTC plays a critical role in the generation and processing of amyloid-beta peptides (Aβ) from the amyloid precursor protein (APP) (57). APP and Aβ have not been adequately explored in lung diseases; however, studies have shown that APP and Aβ from human monocyte-derived macrophages regulate pro-inflammatory and anti-inflammatory mediators (58). Increased levels of amyloid-beta peptides have been observed in the serum and lungs of COPD patients compared to controls. In addition, higher serum Aβ negatively correlated with worse lung function in COPD patients (59,60). Studies have shown that androgens regulate amyloid beta-peptides (61–63). Gillett et al. (61) demonstrated that lower androgen levels were associated with increased plasma amyloid beta-peptide in older men with dementia. Furthermore, low total testosterone has been associated with worse pulmonary function in men with COPD (64). Given that a significant proportion of COPD patients are middle-aged and older men, the age-related decline in androgens (low testosterone) in males, as well as the proteolytic action of OSTC protein on amyloid precursor protein, may result in elevated levels of amyloid beta-peptide, suggesting a potential role of OSTC gene polymorphisms in increasing COPD risk in males.

We also observed intronic and intergenic SNPs (rs1911770 and rs13225543) in/near ZPBP and C7orf72 genes associated with COPD in males. ZPBP, a Zona Pellucida Binding Protein implicated in adult fertility, is expressed in the testis and ovary (65,66). This intronic SNP (rs1911770) in the ZPBP gene had the opposite effect on the risk of COPD in males and females (increasing COPD risk for males but protective for females). In a previous GWAS, variants in/near the ZPBP gene approached genome-wide significance for an association with pulmonary function amongst smokers (FEV₁ and FEV₁/FVC ratio) (67). In addition, ZPBP is a paralog to ZPBP2, which is located on chromosome 17q21.1. Research has shown that ZPBP2 is associated with asthma and childhood asthma (68,69). In a study conducted by Naumova et al. (70), they discovered sex-specific differences in the DNA methylation of the ZPBP2 gene in relation to asthma susceptibility. Specifically, males were revealed to have a lower average methylation than females in the ZPBP2 gene promoter region, implying that hypo-methylation of the ZPBP2 gene increases asthma risk in males. Furthermore, ZPBP/ZPBP2 deletion in a mouse model produces sperm abnormalities and infertility in men but not females (65).

The intergenic SNP (rs13225543) near the C7orf72 gene was found to be male-specific. C7orf72 gene, a spermatogenesis-associated protein, has been linked to spermatogenesis (71). Studies indicate that in male hamsters, pulmonary emphysema affects spermatogenesis, resulting in morphophysiological changes to the reproductive organs due to increased oxidative stress and testosterone imbalance (72). This suggests that the polymorphisms of ZPBP and C7orf72 may have a stronger impact on the development of diseases in males than in females.

Other male-specific associations, albeit protective in the direction of effect, were discovered within or near the MAGI1, COX18, ELOVL5, FGF14, and NKAIN4 genes in the 3p14.1, 4q13.3, 6p12.1, 13q33.1, and 20q13.33 cytogenetic bands, respectively. Variants of the FGF14 gene, which belongs to the fibroblast growth factor family, have been associated with post-BD FEV1 in children with asthma (73). Members of the FGF family have been associated with lung development and respiratory disease (74,75). For instance, polymorphisms in the FGF7 gene have been reported to be associated with COPD (74). FGFs 1,2,8,9 and 10 have been implicated in various levels of lung development (75). Interestingly, FGF10 expression was higher in males than females in a study examining the expression profile of androgen-regulated genes in murine fetal developing lungs (76). In our study, an SNP (rs12869252) in FGF14 was significantly associated with COPD in males and may, in combination with sex hormones, potentially play a sexually dimorphic role in COPD susceptibility.

MAGI1 is widely expressed in lung epithelial cells, where it functions as a scaffolding protein at intercellular junctions and maintains epithelial barrier function (77). The airway epithelial lining serves as the first line of defense against environmental insults such as cigarette smoke. MAGI1 gene has been implicated as a surfactant regulator with increased expression in the fetal lung of males compared to females (76). Cigarette smoke, a significant risk factor for COPD, adversely affects surfactants and airway epithelial architecture (78–80). In one study, the expression of the MAGI1 gene in the airway epithelium was significantly downregulated in smokers with COPD and healthy smokers compared to nonsmokers (80). This suggests that cigarette smoke compromises the integrity of airway epithelial cell-cell junction. With males and females having differential susceptibility to cigarette smoke, the distribution and population of MAGI1 proteins in airway epithelial cells’ tight junctions may play an important role in COPD pathogenesis in males and females.

Two SNPs, rs56334611 and rs6816344, located near the COX18 gene, were associated with COPD in males. COX18 gene encodes a cytochrome c oxidase assembly protein responsible for mitochondrial biogenesis, MT-CO2/COX2 maturation, and regulation of mitochondrial respiratory chain complex IV. Oxidative stress due to excessive reactive oxygen species in COPD patients has been linked to mitochondrial damage, reduced mitochondrial biogenesis, and mitochondrial homeostasis (81,82). Sex disparities in oxidative stress and reactive oxygen species generation have been reported, with males having more oxidative stress, more reactive oxygen species, and lower antioxidant capacity than females (83). ELOVL5 gene, widely expressed in the brain, lung, testis, adrenal gland, and prostate, also regulates mitochondrial functions and reactive oxygen species production (84,85). ELOVL5 has also been found to be overexpressed in prostate cancer. For instance, one experimental study discovered that ELOVL5 was significantly more expressed in prostate cancer cells than in normal/benign prostatic hyperplasia cells and that this upregulation was mediated via androgen receptors (84).

Variants (rs12025895, rs10931835, rs220806 and rs77625370) within/near the CAMTA1 SATB2 PDE10A and LINC00908 genes on chromosomes 1p36.31, 2q33.1, 6q27, and 18q23, respectively, showed female-specific associations with COPD. In a large GWAS of lung function using the UK biobank data, SNPs in SATB2 have been associated with an increase in FEV₁ and FVC (32). In an experimental study using ovariectomized rats, Wu et al. (86) demonstrated that the bone marrow stromal cells (BMSCs) of ovariectomized rats experienced weaker SATB2 expression, reduced bone formation capacity, and increased senescence. On the other hand, estrogen increased SATB2 expression, slowed down cellular aging, and increased the osteogenicity of bone marrow stromal cells. Estrogen deficiency has been associated with osteoporosis during post-menopause (87). Osteoporosis is a major comorbid condition in females with COPD (88). It is likely that the expression of the SATB2 gene may decrease as estrogen levels decrease in menopausal and post-menopausal females with COPD, potentially resulting in a decline in lung function.

CAMTA1 gene, another female-specific association with COPD in our study, has been associated with lung function and COPD (89,90). Kang et al. (91) suggested that the CAMTA1 gene plays a regulatory role in the nuclear factor of activated T cells pathway. The nuclear factor of activated T cells (NFAT), identified in activated T-cells, regulates the expression of IL-2, IL-4, and IL-5 (92,93). T-cells play a central role in an adaptive immune response. Also, Innate and adaptive immune responses differ between sexes. Females have been reported to have more activated T-cells and T-cell proliferation than males (94). Studies have shown that the increased number of T cells in the lungs and airways of patients with COPD correlates with disease severity (95). Furthermore, it has been found that female smokers appear to experience a higher level of inflammatory responses in the airways than their male counterparts (27). These suggest that the CAMTA1 gene may modulate inflammatory mechanisms differently for males and females with COPD.

An SNP (rs220806) in the PDE10A gene was one of the female-specific loci for COPD. This gene’s protein belongs to the cyclic nucleotide phosphodiesterase family (PDEs), which plays an important role in controlling intracellular cyclic nucleotide by hydrolyzing cAMP and cGMP second messengers involved in regulating airway smooth muscle function (96). PDE10A plays an essential role in lung inflammation by promoting macrophage activation and neutrophil infiltration (97). For example, PDE10A knockout mice exhibited reduced IL-1b, MCP-1, IL-6, and TNF-alpha protein levels in lung tissues than in PDE010-WT mice after exposure to lipopolysaccharide (97). Sexual dimorphism in the PDE10A gene has been demonstrated in an experimental animal study. PDE10A knockout mice were confirmed to have decreased body weight compared to their wild-type counterparts, with females being more affected than males (98).

In our study, the identified sex-specific loci associated with asthma-COPD phenotype and COPD may have direct or indirect sexually dimorphic roles. This suggests that the complex interplay between these sex-specific gene signatures and sex hormones or lifestyle factors, such as cigarette smoking, may influence the varying expression and pathobiological functions of these genes in males and females, thus leading to differences in susceptibility to asthma-COPD phenotype and COPD.

Our study had some limitations. The identification of COPD and asthma-COPD phenotype cases as self-reported physician-diagnosed COPD and concomitant diagnosis of COPD and asthma without objective spirometry measurements are subject to misclassification. In this case, a non-differential misclassification (i.e., equal chance of misclassification of our outcome between those with polymorphisms and those with wild-type), might result in reduced statistical power and an underestimation of the observed association. However, large GWAS and population-based studies have widely used self-reported obstructive airway disease diagnosed by physicians to identify genetic and clinical characteristics (11,12,99–102). Self-reported asthma has been shown to have high reliability and validity (103). In addition, self-reported physician-diagnosed COPD/emphysema/chronic bronchitis has been demonstrated to have very high specificity and low sensitivity (104). A recent population study of individuals aged 50–64 years concluded that self-reported physician-diagnosed COPD is a valid tool in studies of risk factors for COPD in the general population, not in studies of prevalence of COPD, due to its high specificity (>95%) and low sensitivity (<13%) (104). This implies that false positives will be reduced. However, the low sensitivity will cause the prevalence of COPD to be underestimated. Nonetheless, our primary goal was to identify sex-specific genetic risk factors.

The sample size for our study was moderately large. However, it appeared relatively underpowered to identify variants (rare or common) with interaction p value significant at the genome-wide significance threshold. Genome-wide interaction studies, in general, necessitate a much larger sample size and more statistical power than standard GWAS. Due to the high proportion of missing values in the pre-bronchodilator spirometry variables and the absence of post-bronchodilator spirometry parameters in the dataset, we could not use spirometry criteria to define our outcomes. Our study lacked an independent replication cohort to validate our findings. Future replication of our findings in subsequent studies could improve generalizability.

Conclusion

Our study identified novel sex-specific loci associated with asthma-COPD phenotype and COPD. These findings are potential precursors to deepening our understanding of sex-related genetic differences in asthma-COPD phenotype and COPD pathology. Future research exploring the expression quantitative trait loci (eQTLs) and discovering the functional sex-specific roles of these genetic signatures may improve disease endotyping in individuals with asthma-COPD phenotype and COPD.

Ethical approval

This CLSA project received ethics approval at two levels. Consent to participate was obtained for all participants under the CLSA harmonized multi-university ethics process approved by the Hamilton Integrated Research Ethics Board (HiREB), Hamilton Health Sciences/McMaster University. Simon Fraser University (SFU) was a participating institution in the CLSA data collection, and the SFU Office of Research Services Ethics Committee reviewed all consent material prior to data collection (SFU ORS #2018s0139). Ethics approval for the current study was obtained from the University of Alberta Health Research Ethics Board (Pro00091377_REN3) and Memorial University Health Research Ethics Board (HREB # 2019.072).

Disclaimer

The opinions expressed in this manuscript are the author’s own and do not reflect the views of the Canadian Longitudinal Study on Aging.

Acknowledgment

This research was made possible using the data/biospecimens collected by the Canadian Longitudinal Study on Aging (CLSA). This research has been conducted using CLSA’s Baseline Comprehensive Dataset version 4.0 and Genomics Dataset version 3.0, under Application Number 19CA006. The CLSA is led by Drs. Parminder Raina, Christina Wolfson, and Susan Kirkland. The opinions expressed in this manuscript are solely those of the author and do not represent those of the Canadian Longitudinal Study on Aging.

Disclosure statement

The authors declare no conflicts of interest.

Data availability statement

The datasets used in this study are not readily available. Data are only available from the Canadian Longitudinal Study on Aging (www.clsa-elcv.ca) for researchers who meet the criteria for access to de-identified CLSA data.

Additional information

Funding

The Canadian Institutes of Health Research (CIHR) Catalyst Grant provided funding for this project (ACD 162989). This was also supported by The Government of Canada provides funding for the Canadian Longitudinal Study on Aging (CLSA) through the Canadian Institutes of Health Research (CIHR) under grant reference: LSA 94473 and the Canada Foundation for Innovation, as well as the following provinces, Newfoundland, Nova Scotia, Quebec, Ontario, Manitoba, Alberta, and British Columbia.