A polygenic risk score and age of diagnosis of COPD

Jingzhou Zhang, Hanfei Xu, Dandi Qiao, Dawn L. DeMeo, Edwin K. Silverman, George T. O'Connor, Brian D. Hobbs, Josée Dupuis, Michael H. Cho, Matthew Moll
European Respiratory Journal 2022 60: 2101954; DOI: 10.1183/13993003.01954-2021

Abstract

Background Genetic susceptibility may be associated with earlier onset of chronic obstructive pulmonary disease (COPD). We hypothesised that a polygenic risk score (PRS) for COPD would be associated with earlier age of diagnosis of COPD.

Methods In 6647 non-Hispanic White (NHW) and 2464 African American (AA) participants from COPDGene, and 6812 participants from the Framingham Heart Study (FHS), we tested the relationship of the PRS and age of COPD diagnosis. Age at diagnosis was determined by: 1) self-reported age at COPD diagnosis or 2) age at visits when moderate-to-severe airflow limitation (Global Initiative for Chronic Obstructive Lung Disease (GOLD) grade 2–4) was observed on spirometry. We used Cox regression to examine the overall and time-dependent effects of the PRS on incident COPD. In the COPDGene study, we also examined the PRS's predictive value for COPD at age <50 years (COPD50) using logistic regression and area under the curve (AUC) analyses, with and without the addition of other risk factors present at early life (e.g. childhood asthma).

Results In Cox models, the PRS demonstrated age-dependent associations with incident COPD, with larger effects at younger ages in both cohorts. The PRS was associated with COPD50 (OR 1.55 (95% CI 1.41–1.71) for NHW, OR 1.23 (95% CI 1.05–1.43) for AA and OR 2.47 (95% CI 2.12–2.88) for FHS participants). In COPDGene, adding the PRS to known early-life risk factors improved prediction of COPD50 in NHW (AUC 0.69 versus 0.74; p<0.0001) and AA (AUC 0.61 versus 0.64; p=0.04) participants.

Conclusions A COPD PRS is associated with earlier age of diagnosis of COPD and retains predictive value when added to known early-life risk factors.

Abstract

A polygenic risk score is associated with earlier age of COPD diagnosis and is predictive for COPD occurring early in life, with clinical implications of individualised risk stratification and preventive measures https://bit.ly/3KIB8Z6

Introduction

Characterised by persistent airflow limitation and respiratory symptoms, chronic obstructive pulmonary disease (COPD) is a leading cause of mortality and morbidity worldwide [1, 2]. Emerging evidence indicates that the pathogenesis of COPD can begin in early life and even the prenatal period [3, 4]. Many patients with COPD are undiagnosed or initially present with advanced disease, missing a potential opportunity for early intervention [5].

COPD in younger individuals is associated with poor clinical outcomes. In a recent large Danish population study, the prevalence of COPD early in life (defined as age <50 years with a forced expiratory volume in 1 s (FEV1)/forced vital capacity (FVC) ratio below the lower limit of normal (LLN) and a cumulative cigarette smoking history of ≥10 pack-years) was estimated to be 15%, and these individuals had more frequent respiratory hospitalisations and increased risk for early death compared with age-adjusted controls [6]. In addition, COPD in young adults does not equate to mild disease, as severe airflow limitation and symptoms can be present [7]. The US Preventive Services Task Force currently recommends against screening for COPD with spirometry in asymptomatic adults. Thus, a risk stratification tool that 1) can identify individuals who are at high risk for developing COPD at young ages and 2) does not depend on early-life screening spirometry or imaging data is of great importance for both individualised preventive measures and effective case-finding efforts.

COPD susceptibility is influenced by both environmental and genetic factors [8, 9]. Genetics explains a sizeable proportion of phenotypic variance of adulthood lung function and risk of COPD (∼30–60% for FEV1, FEV1/FVC and moderate-to-severe COPD) in population-based, cohort and family-based studies, as well as in families specifically identified through probands with severe, early-onset COPD [10–12]. In genome-wide association studies (GWASs) of lung function, each single nucleotide polymorphism (SNP) accounts for a small amount of phenotypic variance; however, variants can be combined into a polygenic risk score (PRS), which accounts for more phenotypic variability and is predictive of prevalent COPD and lung imaging phenotypes [13–16]. We recently derived a PRS based on lung function that was associated with prevalent COPD, emphysema, and lung function growth and decline trajectories [17]. However, it is unknown if the PRS can inform which patients are at high risk for COPD acquired early in life. Furthermore, the magnitude of polygenic effects across age ranges is nonuniform in other diseases [18] and whether polygenic effects are uniform across age ranges in COPD remains unclear.

We hypothesised that a higher PRS would be associated with an earlier age of diagnosis of COPD and COPD before age 50 years (COPD50) in an ascertained sample of smokers and a population-based cohort. We examined the effects of the PRS on the risk for incident COPD and examined age-dependent effects of the PRS. We additionally evaluated the predictive performance of the PRS for COPD50 compared with other early-life risk factors.

Materials and methods

Study design and populations

All studies obtained approval from local institutional review boards and participants provided written informed consent.

COPDGene

We included non-Hispanic White (NHW) and African American (AA) participants from the COPDGene (Genetic Epidemiology of COPD) study, which has been previously described [19]. Briefly, COPDGene is an ongoing multicentre prospective cohort study which enrolled 10 198 NHW and AA participants aged 45–80 years, with a smoking history ≥10 pack-years at baseline and without severe α1-antitrypsin deficiency. Participants were followed up at 5-year (phase 2 visit) and 10-year (phase 3 visit) intervals from baseline visits. Questionnaires for demographics and respiratory health conditions together with spirometry data were obtained at all visits. Further details can be found in the supplementary material.

Framingham Heart Study

To additionally assess the effect of the PRS in a population-based cohort with longitudinal spirometry, we used participants of European ancestry from the Framingham Heart Study (FHS) Offspring cohort and Third Generation cohort. Briefly, FHS was a large longitudinal population-based cohort which has been previously described [20, 21]. Further details can be found in the supplementary material.

Age at diagnosis of COPD

At the COPDGene baseline visit, all participants were asked about physician diagnosis of COPD, emphysema and chronic bronchitis, and for those who answered “yes” they were further asked about age at diagnosis of those conditions. We considered correctly diagnosed COPD as participants with 1) a self-reported physician diagnosis of COPD and/or emphysema and/or chronic bronchitis and 2) moderate-to-severe airflow limitation (FEV1/FVC <0.7 and FEV1 <80% predicted) on baseline post-bronchodilator spirometry. For participants with correctly physician-diagnosed COPD and a self-reported age at diagnosis >30 years, we used self-reported age of diagnosis for COPD, emphysema or chronic bronchitis, in that order of priority; otherwise, we used age at earliest visit when airflow limitation was identified as the age at diagnosis (supplementary figures S1 and S2 for NHW and AA participants, respectively). In FHS, questionnaire data about the age of diagnosis were not available. Thus, age of diagnosis was defined using age at earliest exam when moderate-to-severe COPD (modified Global Initiative for Chronic Obstructive Lung Disease (GOLD) grade 2–4 using pre-bronchodilator spirometry) was observed. We defined COPD occurring early in life as age at diagnosis <50 years [5, 22].

Predictive variables

As described previously, we developed PRSs based on results from GWASs of FEV1 and FEV1/FVC in the UK Biobank and SpiroMeta Consortium participants of European ancestry [17]. Briefly, we calculated individual PRSs separately for FEV1 and FEV1/FVC based on weighted effects of 1.7 and 1.2 million SNPs, respectively. We then calculated a composite PRS as a weighted sum of the two individual PRSs and standardised the composite PRS to facilitate statistical analyses. Details regarding construction of the PRS can be found in the supplementary material. The composite COPD PRS for this study was calculated in COPDGene and FHS, which are external datasets that were not used in the derivation or tuning of the PRS.

A number of early-life risk factors have been identified for COPD [3, 4]. In COPDGene, we included previously described risk factors available by early adulthood that could potentially be used to guide risk stratification for acquiring COPD early in life. These available risk factors included maternal smoking during pregnancy, childhood asthma, active smoking during adolescence, childhood pneumonia, education and family history of COPD. Details of definitions of the these risk factors can be found in the supplementary material.

Statistical analysis

Continuous and categorical variables were shown as median (interquartile range (IQR)) and counts (percentages), respectively. Mann–Whitney U-tests and Chi-squared tests (Fisher's exact tests when appropriate) were used to examine differences for continuous and categorical variables, respectively.

We first performed a time-to-event analysis to examine the effect of the PRS on incident COPD diagnosis in COPDGene. Age in years was used as the underlying timescale and time was censored at age of COPD diagnosis or age at last follow-up. We plotted cumulative incidence curves of COPD diagnosis among tertiles (low versus middle versus high) of the PRS using the Kaplan–Meier estimator and tested differences between curves using a log-rank test. We fitted a Cox model and evaluated the proportional hazards assumption using diagnostics based on Schoenfeld residuals [23]. We additionally fitted Cox models separately for different age intervals (age <50, 50–59, 60–69 and ≥70 years) to allow time-dependent coefficients (β) of the PRS, which was a fixed covariate (see the supplementary material for more details). We tested for an association between the PRS and sex, age, pack-years of smoking at the baseline visit and age at smoking initiation to determine whether these variables could be potential confounding factors.

We also performed logistic regression analysis with COPD50 as a binary outcome (see the supplementary material for more details). We constructed three prediction models for COPD50 with predictive variables of 1) the PRS, 2) other early-life risk factors (see the predictive variables in the previous section) that were significantly associated with COPD50 in univariate analysis, and 3) a combination of 1) and 2). We evaluated the discriminatory accuracy of predictive models by comparing the area under the curve (AUC) of receiver operating characteristic (ROC) curves using DeLong tests [24].

To examine the effect of the PRS on incident COPD in FHS, we constructed similar models as in COPDGene. First, we built a Cox model in participants without COPD at baseline (Offspring cohort exam 5 and Third Generation cohort exam 1), adjusting for baseline age, and then performed regressions across age intervals (age <50, 50–59, 60–69 and ≥70 years). We also performed linear regression between PRS and age of diagnosis of COPD in participants with COPD, and fitted a logistic regression model to examine the association between the PRS and COPD50.

All models including the PRS were adjusted for principal components of genetic ancestry. Models in FHS additionally accounted for study cohort and familial relatedness using mixed models and generalised estimating equations, as appropriate. We performed analyses using R version 4.0.3 (R Foundation for Statistical Computing, Vienna, Austria) and the packages “survival”, “survminer”, “coxme”, “gee” and “pROC”. We considered a two-tailed p-value <0.05 as statistically significant for all tests.

Sensitivity analysis

To assess for potential bias from censoring and using two definitions of COPD diagnosis, we performed a range of sensitivity analyses including stratified analyses and analyses using varying COPD and control definitions in COPDGene NHW participants. See the supplementary material for more details.

Results

Sample characteristics

In COPDGene, 6647 NHW (52.3% males) and 2464 AA (57.6% males) participants were included (table 1); 2115 and 444 had physician-diagnosed COPD prior to their baseline visit, and 696 and 309 were newly diagnosed at the time of study enrolment, respectively. Compared with those with newly diagnosed COPD, participants with physician-diagnosed COPD were more likely to have GOLD spirometry grades 3 and 4 at baseline (supplementary figure S3).

View this table:
TABLE 1

Sample characteristics by status of chronic obstructive pulmonary disease (COPD)# diagnosis at baseline visit

We included 6812 FHS participants of European ancestry who had at least two longitudinal spirometry measures from the Offspring and Third Generation cohorts (table 1). COPD diagnoses were ascertained in 811 participants, including 491 at baseline, and 154, 67, 65 and 34 at each follow-up visit, respectively.

Further details regarding COPDGene and FHS baseline characteristics can be found in the supplementary material.

PRS and incident COPD diagnosis

In COPDGene, cumulative incidence curves (Kaplan–Meier estimator) for COPD diagnosis stratified by tertiles of the PRS (figure 1) demonstrate an increased risk for incident COPD diagnosis among individuals with higher PRSs across all ages (log-rank p<0.0001 for difference between curves for both NHW and AA participants). However, we found evidence against the proportional hazards assumption for a Cox model of the PRS on incident COPD across all age ranges (p=0.048 for the scaled Schoenfeld residuals test) in COPDGene NHW participants. Therefore, we fitted a Cox model allowing different coefficients of the PRS within time intervals of ages <50, 50–59, 60–69 and ≥70 years (p=0.88 for the proportional hazards assumption of the PRS on incident COPD); the hazard ratio (HR) estimates per 1 sd increase in the PRS were 1.51 (95% CI 1.41–1.63), 1.39 (95% CI 1.30–1.47), 1.36 (95% CI 1.28–1.45) and 1.33 (95% CI 1.21–1.47), respectively (figure 2). In COPDGene AA participants, a 1 sd increase of the PRS was associated with a 28% (HR 1.28, 95% CI 1.19–1.37) increased hazard for incident COPD across all ages (p=0.37 for testing proportional hazards assumption of the PRS on incident COPD).

FIGURE 1
FIGURE 1

Cumulative incidence curves of chronic obstructive pulmonary disease (COPD) diagnosis by tertiles of polygenic risk score (PRS: low, middle and high) in COPDGene a) non-Hispanic White and b) African American participants.

FIGURE 2
FIGURE 2

Risk estimates associated with the polygenic risk score (per 1 sd) for incident chronic obstructive pulmonary disease (COPD) among age intervals of <50, 50–69, 60–79 and ≥70 years in a) COPDGene non-Hispanic White and b) Framingham Heart Study participants.

Similarly, in FHS, we found evidence against the proportional hazards assumption for a Cox model examining the effects of the PRS on incident COPD across all age ranges (p=0.005 for the scaled Schoenfeld residuals test). For Cox models allowing different coefficients of the PRS within time intervals of ages <50, 50–59, 60–69 and ≥70 years (p=0.45 for the proportional hazards assumption), the HR estimates per 1 sd increase in the PRS were 2.32 (95% CI 1.81–2.97), 1.70 (95% CI 1.41–2.04), 1.37 (95% CI 1.11–1.67) and 1.28 (95% CI 1.01–1.61), respectively (figure 2).

In COPDGene, we did not detect a differential effect of the PRS by sex (p=0.07 and p=0.91 for the interaction term for COPDGene NHW and AA participants, respectively). In a sex-stratified analysis in NHW participants, a 1 sd increase of the PRS was associated with a 36% (HR 1.36 (95% CI 1.30–1.43)) and a 44% (HR 1.44 (95% CI 1.37–1.52)) increased hazard for incident COPD in male and female participants, respectively. We found no association between the PRS and sex, age, pack-years of smoking at baseline visit or age at initiation of smoking. Results did not change significantly with further adjustment of these factors in the model.

Prediction for COPD diagnosed early in life

In COPDGene, COPD50 diagnoses were observed for 491 NHW and 194 AA participants. These individuals with COPD50 had higher PRSs and were more likely to have early-life risk factors compared with those without COPD50 (table 2). The PRS was associated with an increased risk for COPD50 in univariable (OR per 1 sd: 1.60 (95% CI 1.46–1.76) for NHW and 1.23 (95% CI 1.06–1.43) for AA participants) and multivariable models including other early-life risk factors (OR per 1 sd: 1.55 (95% CI 1.41–1.71) for NHW and 1.23 (95% CI 1.05–1.43) for AA participants) (tables 3 and 4). See the supplementary results for more details. In FHS, COPD50 diagnoses were observed in 186 participants and a 1 sd increase of the PRS was associated with an increased odds of COPD50 (OR 2.47 (95% CI 2.12–2.88); p<0.0001). In FHS participants with COPD, a 1 sd increase of the PRS was associated with a 1.52-year (95% CI 0.84–2.19 years; p<0.0001) earlier COPD diagnosis.

View this table:
TABLE 2

Distribution of risk factors for chronic obstructive pulmonary disease (COPD) occurring early in life (age <50 years; COPD50) in COPDGene non-Hispanic White (NHW) and African American (AA) participants

View this table:
TABLE 3

Associations between the polygenic risk score (PRS) and other early-life risk factors and chronic obstructive pulmonary disease (COPD) occurring early in life (age <50 years; COPD50) in COPDGene non-Hispanic White participants

View this table:
TABLE 4

Associations between the polygenic risk score (PRS) and other early-life risk factors and chronic obstructive pulmonary disease (COPD) occurring early in life (age <50 years; COPD50) in COPDGene African American participants

In COPDGene, ROC curves of the three prediction models for COPD50 (model 1: PRS; model 2: other early-life risk factors; model 3: PRS and other early-life risk factors) are shown in figure 3, and the AUCs were 0.659 (95% CI 0.636–0.683), 0.692 (95% CI 0.668–0.716) and 0.739 (95% CI 0.718–0.761) for NHW participants and 0.571 (95% CI 0.530–0.612), 0.609 (95% CI 0.572–0.646) and 0.635 (95% CI 0.595–0.675) for AA participants, respectively. No significant difference was found between AUCs of model 1 and model 2 for NHW (p=0.055) or AA (p=0.17) participants. There were significant differences between AUCs of model 2 and model 3 for both NHW (p<0.0001) and AA (p=0.043) participants.

FIGURE 3
FIGURE 3

Receiver operating characteristic curves of prediction models for chronic obstructive pulmonary disease (COPD) occurring early in life (age <50 years; COPD50) with predictive variables of polygenic risk score (PRS) (model 1), other early-life risk factors (model 2), and PRS and other early-life risk factors (model 3) in COPDGene a) non-Hispanic White and b) African American participants. Models with PRS were adjusted for principal components of genetic ancestry. AUC: area under curve.

Sensitivity analysis

All sensitivity analyses were performed in COPDGene NHW participants. In participants who had COPD at the baseline visit (n=2811), a 1 sd increase in the PRS was associated with a 0.95-year (95% CI 0.59–1.30 years; p<0.0001) earlier diagnosis of COPD. Compared with participants with the lowest PRS tertile, those who were of the highest PRS tertile had a 1.89-year (95% CI 1.01–2.78 years; p<0.0001) earlier diagnosis. A 1 sd increase of the PRS was associated with a 0.93-year (95% CI 0.52–1.33 years; p<0.0001) and a 1.07-year (95% CI 0.41–1.72 years; p=0.0016) earlier COPD diagnosis in participants with physician-diagnosed COPD and newly diagnosed COPD at the baseline visit, respectively. In participants who were ≥50 years old at the baseline visit (n=5929), COPD50 was reported in 353 participants. As shown in supplementary table S1, the association between the PRS and COPD50 in univariable and multivariable models which included other early-life risk factors was similar compared with the results in table 3. No significant difference was found between AUCs of model 1 and model 2 (AUC 0.654 (95% CI 0.626–0.682) versus 0.692 (95% CI 0.664–0.720); p=0.060). There was a significant difference between AUCs of model 2 and model 3 (AUC 0.692 (95% CI 0.664–0.720) versus 0.736 (95% CI 0.711–0.762); p<0.0001). Additional sensitivity analysis results can be found in the supplementary results.

Discussion

In two cohorts, one of NHW and AA smokers with and without COPD, predominately using self-reported age of diagnosis, and a second cohort, a population-based study relying on longitudinal lung function, we found that a PRS for COPD is associated with an increased risk for earlier age of COPD diagnosis. In addition, we found an age-dependent effect of the PRS on risk for COPD, observing larger effect estimates at younger ages (negative age dependency of the PRS).

Different lung function trajectories distinguished by combinations of failure to attain maximal lung function and/or accelerated decline of lung function may lead to airflow obstruction [25, 26]. Derived from more than 1 million SNPs from a large GWAS of lung function in population-based studies, our PRS could theoretically represent the cumulative effect of common genetic variants associated with growth, plateau and/or early decline of lung function. The genetic variants included in the PRS may affect gene regulation in fetal lung [13]. We previously demonstrated that the PRS was associated with reduced lung growth in children with asthma; these children with reduced lung growth developed spirometric obstruction early in adulthood [17]. Thus, it is possible that the observed age-dependent effects of the PRS are driven by variants important for lung growth and development. In addition, the PRS may be associated with genetic susceptibility to environmental exposure-induced injury and consequently the rate of early lung function decline. For example, genetic factors may modify the effect of cigarette smoking on the development of COPD [27–30].

Our findings may also be consistent with the increased relative effects of cigarette smoking in later age. While this question was not directly addressed in our study, it has been shown that heritability of lung function decreases with age in the UK Biobank [31]. The underlying mechanisms of the time-varying effect of the PRS warrant further research.

COPD occurring at earlier ages is often underdiagnosed and underdiagnosis is associated with unfavourable clinical outcomes [5, 32]. To address this issue, an international panel of experts defined early COPD as occurring in individuals aged <50 years with ≥10 pack-years cigarette smoking, with either abnormal lung function (FEV1/FVC <LLN or accelerated FEV1 decline) or abnormal chest imaging findings (visual emphysema, air trapping or bronchial thickening) [22]. In the present study, we defined COPD occurring early in life using age, smoking exposure and moderate-to-severe spirometric criteria for obstruction. Additional studies examining the association of the PRS with the development of destructive emphysema and airway pathology in younger cohorts may help elucidate the specific phenotype identified by our PRS.

While a previous study suggested that the PRS could be used for early detection and prevention of COPD [17], the present study offers direct evidence that the PRS can predict COPD occurring in younger age groups. Under the current case-finding strategy for COPD, respiratory symptoms are essential for physicians to suspect a diagnosis of COPD. However, a substantial proportion of patients are underdiagnosed with already present or under-reported respiratory symptoms, and have increased risk for respiratory hospitalisations and mortality compared with people without obstruction [32–34]. Thus, the current findings suggest that genetics offer a critical tool to identify young people at high risk for COPD occurring early in life. Knowledge of an individual's PRS may enable physicians to make targeted inquiries about patients’ respiratory symptoms and prioritise pulmonary function testing for patients with symptoms. Also, a known unfavourable PRS might further motivate an individual to be compliant with medical interventions [35], although the behavioural and psychosocial reactions to the awareness of an individual's own genetic risk of COPD should be carefully studied prior to clinical application of the PRS.

Many early-life risk factors for COPD have been identified, including maternal smoking during pregnancy [36–38], childhood asthma [39–42], active smoking during adolescence [43], childhood pneumonia [44], socioeconomic status [45–47] and family history [48, 49]. The odds ratio of the PRS was comparable with other early-life risk factors (table 3), but the PRS offers several distinct advantages compared with these clinical variables. First, the PRS can be determined at birth and earlier than most of the other risk factors, and interventions targeting modifiable early-life risk factors for COPD can be implemented. Second, the PRS is a continuous predictor; as we have previously demonstrated, identifying those at the highest and lowest predicted risk as would likely be done in clinical implementation yields larger effect sizes. Third, the AUC estimates demonstrate that the PRS alone had a comparable performance for predicting COPD occurring early in life compared with a combination of early-life risk factors and the addition of the PRS to early-life risk factors significantly improved the predictive performance. Notably, the PRS was derived from external cohorts, whereas early-life risk factors were modelled and tested in the same cohort, which might have resulted in an underestimated additive value of the PRS. However, early-life risk factors were self-reported and may suffer from recall bias, and the predictive performance of early-life risk factors may have been under- or overestimated. The association of the PRS with COPD early in life was amplified when excluding participants with an age of COPD diagnosis ≥50 years from controls, suggesting that the observed association of the PRS with COPD early in life could be a conservative estimate.

While other early-life risk factors for COPD were not available for this study (e.g. low birthweight) [50], detailed knowledge of a person's early-life risk factors is often unknown or difficult to measure accurately (e.g. cumulative exposure to air pollution). By contrast, a PRS can be calculated using genome-wide genotyping, which can be done once in an individual's lifetime, is of low cost and is potentially relevant for a large number of diseases. In addition, the performance of the PRS will likely improve with future genetic association studies.

Strengths of this study include a large sample of well-phenotyped smokers with post-bronchodilator spirometry data, which is essential to confirm irreversible airflow limitation and to differentiate COPD from asthma, especially in young participants. In addition, we were able to include a relatively extensive panel of known early-life risk factors to compare with the PRS for the predictive performance of COPD occurring early in life. We were also able to replicate the findings in a population-based cohort with both smokers and nonsmokers.

This study has several limitations. We acknowledge a potential measurement bias introduced from using two different COPD outcome definitions (physician-diagnosed versus spirometry-defined); however, sensitivity analyses suggest that our findings are robust because 1) the linear association between the PRS and age of COPD diagnosis is consistent among participants with physician-diagnosed and spirometry-defined COPD, 2) removing participants with baseline COPD (i.e. those most likely to be physician-diagnosed) did not attenuate the association of the PRS with incident COPD, and 3) stratified analyses within individuals with physician-diagnosed COPD demonstrated similar results. In addition, our findings were further supported by FHS results that used only spirometry-defined COPD. We used age at diagnosis as a proxy of age at onset of COPD. We would expect a large variability in the interval between the two time-points since the severity of COPD at the time of diagnosis is highly variable. This issue could result in misclassification bias of the outcome variable. However, our sensitivity analyses did not reveal an association between the PRS and either timing of physician diagnosis in the course of clinical COPD or participants’ ages at baseline visit. Therefore, misclassifications would generally bias toward the null and are unlikely to account the observed results.

Our case definition was for moderate-to-severe COPD (GOLD grades 2–4), and thus we included GOLD grade 1 and preserved ratio impaired spirometry (PRISm) subjects as controls in our calculations of incidence of moderate-to-severe disease. Having demonstrated age-dependent PRS effects, derivation and testing of PRSs in younger populations is needed. Findings of AA smokers were of the same direction of association compared with those of NHW participants in COPDGene, although with appreciably smaller effect size estimates. This ∼50% reduction in the magnitude of the regression coefficients (β) parallels the reduced predictive performance observed when applying the European-derived PRS for COPD to AA participants (OR per 1 sd: 1.50 (95% CI 1.37–1.65) for AA and 2.20 (95% CI 2.03–2.37) for NHW participants). In addition to under-representation of non-European ancestry individuals in previous GWASs, the difference in the case–control ratio (milder disease in AA participants in COPDGene) and access to healthcare might also contribute the different performance of the PRS on age of diagnosis of COPD. These results highlight the need to improve multi-ancestry polygenic prediction.

In conclusion, in a large sample of smokers and a general population cohort, a higher COPD PRS is associated with an increased risk for incident COPD, and the effect of the PRS is age-dependent and larger at younger ages. A higher PRS is associated with an earlier age of diagnosis of COPD. The PRS adds substantial value to other early-life risk factors in prediction for COPD occurring early in life.

Supplementary material

Supplementary Material

Please note: supplementary material is not edited by the Editorial Office, and is uploaded as it has been supplied by the author.

Supplementary material ERJ-01954-2021.Supplement

Shareable PDF

Supplementary Material

This one-page PDF can be shared freely online.

Shareable PDF ERJ-01954-2021.Shareable

Footnotes

  • Author contributions: Study design: J. Zhang, M. Moll, B.D. Hobbs and M.H. Cho. Acquisition, analysis or interpretation of the data: all authors. Drafting of the manuscript: J. Zhang, M. Moll and M.H. Cho. Critical revision of the manuscript for important intellectual content: all authors. Obtained funding: E.K. Silverman and M.H. Cho.

  • Conflict of interest: E.K. Silverman received grant support from GlaxoSmithKline and Bayer. M.H. Cho has received grant support from GlaxoSmithKline and Bayer, consulting fees from Genentech and AstraZeneca, and speaking fees from Illumina. D.L. DeMeo has received support from Bayer and Honoraria from Novartis. J. Dupuis received NIH funding for salary coverage paid to Boston University. J. Zhang, H. Xu, D. Qiao, G.T. O'Connor, B.D. Hobbs and M. Moll have no conflict of interest to declare.

  • Support statement: M. Moll is supported by T32HL007427. B.D. Hobbs is supported by K08HL136928, U01HL089856, R01HL147148 and R01HL135142. M.H. Cho is supported by R01HL137927, R01HL135142, R01HL147148 and U01HL089856. E.K. Silverman is supported by R01HL137927, R01HL147148, U01HL089856, R01HL133135, P01HL132825 and P01HL114501. D.L. DeMeo is supported by P01HL132825 and P01HL114501, and a grant from the Alpha-1 Foundation. The COPDGene project described was supported by U01HL089897 and U01HL089856 from the National Heart, Lung, and Blood Institute (NHLBI). The content is solely the responsibility of the authors and does not necessarily represent the official views of the NHLBI or the National Institutes of Health. COPDGene is also supported by the COPD Foundation through contributions made to an industry advisory board that has included AstraZeneca, Bayer Pharmaceuticals, Boehringer Ingelheim, Genentech, GlaxoSmithKline, Novartis, Pfizer and Sunovion. The Framingham Heart Study is conducted and supported by the NHLBI in collaboration with Boston University (N01-HC-25195, HHSN268201500001I and 75N92019D00031). Funding information for this article has been deposited with the Crossref Funder Registry.

  • Received July 12, 2021.
  • Accepted January 14, 2022.

References

  1. ↵
    1. Rabe KF,
    2. Watz H
    . Chronic obstructive pulmonary disease. Lancet 2017; 389: 1931–1940. doi:10.1016/S0140-6736(17)31222-9
    OpenUrlCrossRefPubMed
  2. ↵
    1. Lõpez-Campos JL,
    2. Tan W,
    3. Soriano JB
    . Global burden of COPD. Respirology 2016; 21: 14–23. doi:10.1111/resp.12660
    OpenUrlCrossRefPubMed
  3. ↵
    1. Postma DS,
    2. Bush A,
    3. Van Den Berge M
    . Risk factors and early origins of chronic obstructive pulmonary disease. Lancet 2015; 385: 899–909. doi:10.1016/S0140-6736(14)60446-3
    OpenUrlCrossRefPubMed
  4. ↵
    1. Martinez FD
    . Early-life origins of chronic obstructive pulmonary disease. N Engl J Med 2016; 375: 871–878. doi:10.1056/NEJMra1603287
    OpenUrlCrossRefPubMed
  5. ↵
    1. Soriano JB,
    2. Polverino F,
    3. Cosio BG
    . What is early COPD and why is it important? Eur Respir J 2018; 52: 1801448. doi:10.1183/13993003.01448-2018
    OpenUrlAbstract/FREE Full Text
  6. ↵
    1. Çolak Y,
    2. Afzal S,
    3. Nordestgaard BG, et al.
    Prevalence, characteristics, and prognosis of early chronic obstructive pulmonary disease the Copenhagen General Population Study. Am J Respir Crit Care Med 2020; 201: 671–680. doi:10.1164/rccm.201908-1644OC
    OpenUrlCrossRefPubMed
  7. ↵
    1. Hersh CP,
    2. DeMeo DL,
    3. Al-Ansari E, et al.
    Predictors of survival in severe, early onset COPD. Chest 2004; 126: 1443–1451. doi:10.1378/chest.126.5.1443
    OpenUrlCrossRefPubMedWeb of Science
  8. ↵
    1. Silverman EK
    . Genetics of COPD. Annu Rev Physiol 2020; 82: 413–431. doi:10.1146/annurev-physiol-021317-121224
    OpenUrlPubMed
  9. ↵
    1. Lange P,
    2. Ahmed E,
    3. Lahmar ZM, et al.
    Natural history and mechanisms of COPD. Respirology 2021; 26: 298–321. doi:10.1111/resp.14007
    OpenUrl
  10. ↵
    1. Zhou JJ,
    2. Cho MH,
    3. Castaldi PJ, et al.
    Heritability of chronic obstructive pulmonary disease and related phenotypes in smokers. Am J Respir Crit Care Med 2013; 188: 941–947. doi:10.1164/rccm.201302-0263OC
    OpenUrlCrossRefPubMed
    1. Klimentidis YC,
    2. Vazquez AI,
    3. de los Campos G, et al.
    Heritability of pulmonary function estimated from pedigree and whole-genome markers. Front Genet 2013; 4: 174. doi:10.3389/fgene.2013.00174
    OpenUrl
  12. ↵
    1. Silverman EK,
    2. Chapman HA,
    3. Drazen JM, et al.
    Genetic epidemiology of severe, early-onset chronic obstructive pulmonary disease: risk to relatives for airflow obstruction and chronic bronchitis. Am J Respir Crit Care Med 1998; 157: 1770–1778. doi:10.1164/ajrccm.157.6.9706014
    OpenUrlCrossRefPubMedWeb of Science
  13. ↵
    1. Sakornsakolpat P,
    2. Prokopenko D,
    3. Lamontagne M, et al.
    Genetic landscape of chronic obstructive pulmonary disease identifies heterogeneous cell-type and phenotype associations. Nat Genet 2019; 51: 494–505. doi:10.1038/s41588-018-0342-2
    OpenUrlCrossRefPubMed
    1. Shrine N,
    2. Guyatt AL,
    3. Erzurumluoglu AM, et al.
    New genetic signals for lung function highlight pathways and chronic obstructive pulmonary disease associations across multiple ancestries. Nat Genet 2019; 51: 481–493. doi:10.1038/s41588-018-0321-7
    OpenUrlCrossRefPubMed
    1. Wain LV,
    2. Shrine N,
    3. Artigas MS, et al.
    Genome-wide association analyses for lung function and chronic obstructive pulmonary disease identify new loci and potential druggable targets. Nat Genet 2017; 49: 416–425. doi:10.1038/ng.3787
    OpenUrlCrossRefPubMed
  16. ↵
    1. Oelsner EC,
    2. Ortega VE,
    3. Smith BM, et al.
    A genetic risk score associated with chronic obstructive pulmonary disease susceptibility and lung structure on computed tomography. Am J Respir Crit Care Med 2019; 200: 721–731. doi:10.1164/rccm.201812-2355OC
    OpenUrl
  17. ↵
    1. Moll M,
    2. Sakornsakolpat P,
    3. Shrine N, et al.
    Chronic obstructive pulmonary disease and related phenotypes: polygenic risk scores in population-based and case-control cohorts. Lancet Respir Med 2020; 8: 696–708. doi:10.1016/S2213-2600(20)30101-6
    OpenUrl
  18. ↵
    1. Archambault AN,
    2. Su YR,
    3. Jeon J, et al.
    Cumulative burden of colorectal cancer-associated genetic variants is more strongly associated with early-onset vs late-onset cancer. Gastroenterology 2020; 158: 1274–1286. doi:10.1053/j.gastro.2019.12.012
    OpenUrlCrossRefPubMed
  19. ↵
    1. Regan EA,
    2. Hokanson JE,
    3. Murphy JR, et al.
    Genetic Epidemiology of COPD (COPDGene) study design. COPD 2010; 7: 32–43. doi:10.3109/15412550903499522
    OpenUrlCrossRefPubMedWeb of Science
  20. ↵
    1. Kannel WB,
    2. Feinleib M,
    3. McNamara PM, et al.
    An investigation of coronary heart disease in families: the Framingham Offspring Study. Am J Epidemiol 1979; 110: 281–290. doi:10.1093/oxfordjournals.aje.a112813
    OpenUrlCrossRefPubMedWeb of Science
  21. ↵
    1. Splansky GL,
    2. Corey D,
    3. Yang Q, et al.
    The Third Generation Cohort of the National Heart, Lung, and Blood Institute's Framingham Heart Study: design, recruitment, and initial examination. Am J Epidemiol 2007; 165: 1328–1335. doi:10.1093/aje/kwm021
    OpenUrlCrossRefPubMedWeb of Science
  22. ↵
    1. Martinez FJ,
    2. Han MK,
    3. Allinson JP, et al.
    At the root: defining and halting progression of early chronic obstructive pulmonary disease. Am J Respir Crit Care Med 2018; 197: 1540–1551. doi:10.1164/rccm.201710-2028PP
    OpenUrlPubMed
  23. ↵
    1. Grambsch PM,
    2. Therneau TM
    . Proportional hazards tests and diagnostics based on weighted residuals. Biometrika 1994; 81: 515–526. doi:10.1093/biomet/81.3.515
    OpenUrlCrossRefWeb of Science
  24. ↵
    1. DeLong ER,
    2. DeLong DM,
    3. Clarke-Pearson DL
    . Comparing the areas under two or more correlated receiver operating characteristic curves: a nonparametric approach. Biometrics 1988; 44: 837. doi:10.2307/2531595
    OpenUrlCrossRefPubMedWeb of Science
  25. ↵
    1. Lange P,
    2. Celli B,
    3. Agustí A, et al.
    Lung-function trajectories leading to chronic obstructive pulmonary disease. N Engl J Med 2015; 373: 111–122. doi:10.1056/NEJMoa1411532
    OpenUrlCrossRefPubMed
  26. ↵
    1. Agusti A,
    2. Faner R
    . Lung function trajectories in health and disease. Lancet Respir Med 2019; 7: 358–364. doi:10.1016/S2213-2600(18)30529-0
    OpenUrl
  27. ↵
    1. Rangasamy T,
    2. Cho CY,
    3. Thimmulappa RK, et al.
    Genetic ablation of Nrf2 enhances susceptibility to cigarette smoke-induced emphysema in mice. J Clin Invest 2004; 114: 1248–1259. doi:10.1172/JCI200421146
    OpenUrlCrossRefPubMedWeb of Science
    1. DeMeo DL,
    2. Campbell EJ,
    3. Barker AF, et al.
    IL10 polymorphisms are associated with airflow obstruction in severe α1-antitrypsin deficiency. Am J Respir Cell Mol Biol 2008; 38: 114–120. doi:10.1165/rcmb.2007-0107OC
    OpenUrlCrossRefPubMedWeb of Science
    1. Molloy K,
    2. Hersh CP,
    3. Morris VB, et al.
    Clarification of the risk of chronic obstructive pulmonary disease in α1-antitrypsin deficiency PiMZ heterozygotes. Am J Respir Crit Care Med 2014; 189: 419–427. doi:10.1164/rccm.201311-1984OC
    OpenUrlCrossRefPubMed
  30. ↵
    1. Kim W,
    2. Moll M,
    3. Qiao D, et al.
    Smoking interaction with a polygenic risk score for reduced lung function. medRxiv 2021; preprint [https://doi.org/10.1101/2021.03.26.21254415]. doi:10.1101/2021.03.26.21254415
  31. ↵
    1. Ge T,
    2. Chen CY,
    3. Neale BM, et al.
    Phenome-wide heritability analysis of the UK Biobank. PLoS Genet 2017; 13: e070177. doi:10.1371/journal.pgen.1006711
    OpenUrl
  32. ↵
    1. Diab N,
    2. Gershon AS,
    3. Sin DD, et al.
    Underdiagnosis and overdiagnosis of chronic obstructive pulmonary disease. Am J Respir Crit Care Med 2018; 198 : 1130–1139. doi:10.1164/rccm.201804-0621CI
    OpenUrlCrossRefPubMed
    1. Martinez CH,
    2. Mannino DM,
    3. Jaimes FA, et al.
    Undiagnosed obstructive lung disease in the United States associated factors and long-term mortality. Ann Am Thorac Soc 2015; 12: 1788–1795. doi:10.1513/AnnalsATS.201506-388OC
    OpenUrlCrossRefPubMed
  34. ↵
    1. Gershon AS,
    2. Hwee J,
    3. Chapman KR, et al.
    Factors associated with undiagnosed and overdiagnosed COPD. Eur Respir J 2016; 48: 561–564. doi:10.1183/13993003.00458-2016
    OpenUrlAbstract/FREE Full Text
  35. ↵
    1. Carpenter MJ,
    2. Strange C,
    3. Jones Y, et al.
    Does genetic testing result in behavioral health change? Changes in smoking behavior following testing for alpha-1 antitrypsin deficiency. Ann Behav Med 2007; 33: 22–28. doi:10.1207/s15324796abm3301_3
    OpenUrlCrossRefPubMedWeb of Science
  36. ↵
    1. Hanrahan JP,
    2. Tager IB,
    3. Segal MR, et al.
    The effect of maternal smoking during pregnancy on early infant lung function. Am Rev Respir Dis 1992; 145: 1129–1135. doi:10.1164/ajrccm/145.5.1129
    OpenUrlCrossRefPubMedWeb of Science
    1. Gilliland FD,
    2. Berhane K,
    3. McConnell R, et al.
    Maternal smoking during pregnancy, environmental tobacco smoke exposure and childhood lung function. Thorax 2000; 55: 271–276. doi:10.1136/thorax.55.4.271
    OpenUrlAbstract/FREE Full Text
  38. ↵
    1. Hayatbakhsh MR,
    2. Sadasivam S,
    3. Mamun AA, et al.
    Maternal smoking during and after pregnancy and lung function in early adulthood: a prospective study. Thorax 2009; 64: 810–814. doi:10.1136/thx.2009.116301
    OpenUrlAbstract/FREE Full Text
  39. ↵
    1. McGeachie MJ,
    2. Yates KP,
    3. Zhou X, et al.
    Patterns of growth and decline in lung function in persistent childhood asthma. N Engl J Med 2016; 374: 1842–1852. doi:10.1056/NEJMoa1513737
    OpenUrlCrossRefPubMed
    1. Sears MR,
    2. Greene JM,
    3. Willan AR, et al.
    A longitudinal, population-based, cohort study of childhood asthma followed to adulthood. N Engl J Med 2003; 349: 1414–1422. doi:10.1056/NEJMoa022363
    OpenUrlCrossRefPubMedWeb of Science
    1. Morgan WJ,
    2. Stern DA,
    3. Sherrill DL, et al.
    Outcome of asthma and wheezing in the first 6 years of life follow-up through adolescence. Am J Respir Crit Care Med 2005; 172: 1253–1258. doi:10.1164/rccm.200504-525OC
    OpenUrlCrossRefPubMedWeb of Science
  42. ↵
    1. James AL,
    2. Palmer LJ,
    3. Kick E, et al.
    Decline in lung function in the Busselton Health Study: the effects of asthma and cigarette smoking. Am J Respir Crit Care Med 2005; 171: 109–114. doi:10.1164/rccm.200402-230OC
    OpenUrlCrossRefPubMedWeb of Science
  43. ↵
    1. Gold DR,
    2. Wang X,
    3. Wypij D, et al.
    Effects of cigarette smoking on lung function in adolescent boys and girls. N Engl J Med 1996; 335: 931–937. doi:10.1056/NEJM199609263351304
    OpenUrlCrossRefPubMedWeb of Science
  44. ↵
    1. Hayden LP,
    2. Hobbs BD,
    3. Cohen RT, et al.
    Childhood pneumonia increases risk for chronic obstructive pulmonary disease: the COPDGene study. Respir Res 2015; 16: 115. doi:10.1186/s12931-015-0273-8
    OpenUrl
  45. ↵
    1. Prescott E,
    2. Lange P,
    3. Vestbo J
    . Socioeconomic status, lung function and admission to hospital for COPD: results from the Copenhagen City Heart Study. Eur Respir J 1999; 13: 1109–1114. doi:10.1034/j.1399-3003.1999.13e28.x
    OpenUrlAbstract/FREE Full Text
    1. Yin P,
    2. Zhang M,
    3. Li Y, et al.
    Prevalence of COPD and its association with socioeconomic status in China: findings from China Chronic Disease Risk Factor Surveillance 2007. BMC Public Health 2011; 11: 586. doi:10.1186/1471-2458-11-586
    OpenUrlCrossRefPubMed
  47. ↵
    1. Kanervisto M,
    2. Vasankari T,
    3. Laitinen T, et al.
    Low socioeconomic status is associated with chronic obstructive airway diseases. Respir Med 2011; 105: 1140–1146. doi:10.1016/j.rmed.2011.03.008
    OpenUrlCrossRefPubMed
  48. ↵
    1. Hersh CP,
    2. Hokanson JE,
    3. Lynch DA, et al.
    Family history is a risk factor for COPD. Chest 2011; 140: 343–350. doi:10.1378/chest.10-2761
    OpenUrlCrossRefPubMed
  49. ↵
    1. Moll M,
    2. Lutz SM,
    3. Ghosh AJ, et al.
    Relative contributions of family history and a polygenic risk score on COPD and related outcomes: COPDGene and ECLIPSE studies. BMJ Open Respir Res 2020; 7: e000755. doi:10.1136/bmjresp-2020-000755
    OpenUrlAbstract/FREE Full Text
  50. ↵
    1. Baumann S,
    2. Godtfredsen NS,
    3. Lange P, et al.
    The impact of birth weight on the level of lung function and lung function decline in the general adult population. The Inter99 study. Respir Med 2015; 109: 1293–1299. doi:10.1016/j.rmed.2015.08.011
    OpenUrlCrossRefPubMed
Back to top
View this article with LENS
Email
Print
Citation Tools

A polygenic risk score and age of diagnosis of COPD
Jingzhou Zhang, Hanfei Xu, Dandi Qiao, Dawn L. DeMeo, Edwin K. Silverman, George T. O'Connor, Brian D. Hobbs, Josée Dupuis, Michael H. Cho, Matthew Moll
European Respiratory Journal Sep 2022, 60 (3) 2101954; DOI: 10.1183/13993003.01954-2021
del.icio.us logo Digg logo Reddit logo Technorati logo Twitter logo CiteULike logo Connotea logo Facebook logo Google logo Mendeley logo
Full Text (PDF)

Jump To

Subjects