Discussion
Our study found that people with lung function impairments, such as PRISm and AO, have higher risks of lung cancer incidence and mortality. The risks of lung cancer increased as the severity of lung function impairment increased or as the FEV1 values lowered. These findings of dose responsiveness and reverse correlation remained consistent across sensitivity analyses, stratified analyses and cubic spline curves, demonstrating the robustness of the association.
In our study, individuals with AO had a higher risk of lung cancer incidence and mortality than individuals with PRISm. This could be attributed to the larger number of cases in low severity levels (FEV1>70%) of PRISm-type lung function impairment. At high severity levels (FEV1 value <50%), individuals with PRISm faced higher risks compared with those with AO. Our findings are consistent with the report published by Young et al in 2022, which found that individuals with PRISm have more advanced lung cancers and the greatest lung cancer lethality.21
Prevalence of PRISm and AO
The prevalence of PRISm and AO in previous population-based and hospital-based studies gave a wide range of results. In our study, the proportions of PRISm and AO were 14.3% and 7.9%, compared with 10.2% and 13.1% in a population-based cohort in Korea.22 These differences could be attributed to variations in the study population, such as a higher number of low-risk which means never-smokers in our cohort. Higher proportions of PRISm could be found in hospital-based or cohorts of high-risk groups. It was as high as 21.3% in the Korean lung cancer registry,23 16.7% in a Japanese study of individuals over 40 years of age24 and 17–24% in a hospital study from the University of Iowa.11 Restrictive patterns of lung function impairment can be observed in interstitial lung diseases such as idiopathic pulmonary fibrosis (IPF). However, the prevalence of IPF is relatively low in Taiwan, that is, 3.1–6.4 cases per 100 000 people.25 Therefore, the restrictive patterns observed in our cohort were more likely due to PRISm or AO rather than underlying interstitial lung diseases. Additionally, our study’s cohort was based on an apparently healthy general population enrolled in a private health screening programme and we excluded extreme values of spirometry test readings (FEV1 or FVC percentage >150 or <25) for analyses.
In our study, the prevalence of AO was 7.9%, which is close to 8.8% reported for the Southeast Asia Region/Western Pacific Region in a meta-analysis.26 However, a higher prevalence of AO 37.4% was observed in a Manchester study targeted at ever-smokers aged 55–74.27 Approximately one-third of smokers eligible for lung cancer screening in Germany had AO or PRISm,28 and 34.4% of the National Lung Screening Trial-American College of Radiology Imaging Network cohort had COPD.29 Importantly, the prevalence of AO can vary depending on the cut-off value used for diagnosis. Using a pre-bronchodilator FEV1/FVC<0.66 instead of 0.7 resulted in a 15% increase in the accuracy of AO diagnosis.30 Despite these variations, our study’s findings agree with previous literature on the increasing lung function impairments in the general population, either no matter PRISm or AO type.
Lung cancer incidence and mortality in PRISm and AO
The risks of lung cancer incidence and mortality for individuals with PRISm or AO in previous literature were higher than our study results. However, it is important to note that most of those studies were not based on general populations. In the overall cohort of our study, individuals with PRISm had a 14% higher chance of developing lung cancer which is relatively lower compared with the 27% higher risk observed in a nationwide population-based cohort study conducted in Korea.22 Additionally, a Swedish study on construction workers reported a relative risk of 2.0 (95% CI 1.6 to 2.5).31 Individuals with AO in our study had a 29% higher risk for lung cancer, however, the aforementioned Korean study reported two and half times higher risk in their analysis.22 In a Norway study conducted on former or current smokers between the ages of 40 and 76 years, the risk of developing lung cancer for AO patients was reported to be as high as five times.32
Individuals with PRISm in our overall cohort had a 23% higher risk for lung cancer mortality. In the population-based Tucson Epidemiological Study of Airway Obstructive Disease study of the 2048 cohort, participants with recurrent PRISm had 30% lower risk and inconsistent PRISm had 50% higher risk but the HRs were not significant due to a low number of cases identified.33 Patients with non-small cell and SCLC with PRISm had poorer overall survival with a 62% higher risk of mortality, compared with those with AO or normal lung function.23 In our cohort study, individuals with AO had a 30% higher risk of lung cancer mortality which was relatively low compared with other studies. For instance, in studies conducted on patients with lung cancer, AO was associated with poorer overall and progression-free survivals with more than two times higher mortality risks.34 Our study generally agreed with findings from previous literature and relatively weaker strength of associations could be explained by differences in cohort sizes, study design and analytical settings.
Sensitivity and stratified analysis
In sensitivity analyses that excluded smokers and lung cancer cases occurring within the initial 3-year period, we observed associations between the risk of lung cancer incidence and impairments in lung function. However, the risks were not significant among never-smokers with PRISm/AO, except for lung cancer mortality within the PRISm subgroup. The relatively small numbers of lung cancer cases and deaths among never-smokers in the cohort may explain the lack of significance observed in the analysis. We believe this finding is still relevant because even though no proven direct association, the UK Biobank study18 acknowledged the advantage of identifying lung cancer cases in never-smokers by including lung function status. The cubic spline curves in our study also showed the risk of lung cancer increased with the severity of lung function impairment in never-smokers with PRISm. These dose-response shapes and inverse correlation were more obvious in people with severely reduced FEV1 values, that is, <50% (figure 3).
Our stratified analyses revealed that PRISm at age <60 had lower risks for both lung cancer incidence and mortality compared with those ages >60. However, the result was insignificant with 95% CIs of over-riding 1. This may be attributed to the low number of lung cancer cases and fatalities in the age group under 60. The Okinawa COPD casE finding AssessmeNt study in Japan showed that a significant proportion of individuals aged <60 who attended their annual health examination had impaired lung function.24 Similarly in our study, 70% of those with PRISm or AO were under 60 and were potentially engaging with respiratory illnesses including lung cancer in the future. In terms of stratification by gender and lung cancer histological types, it was observed that women exhibited a greater mortality risk from lung cancer in comparison to men. Furthermore, non-smoking individuals with PRISm demonstrated higher risks of squamous cell carcinoma than adenocarcinoma. However, the German Lung Cancer Screening Intervention Study shed light on the reasons behind the greater reduction in lung cancer mortality through CT screening in women, explaining the higher prevalence of slow-progressing and peripherally located adenocarcinomas in women, whereas men exhibited a greater association with rapidly growing and centrally located squamous cell carcinomas.28 Additionally, the inverse relationship between FEV1 values (<80%) and lung cancer risks observed under cubic spline curves for non-smoking individuals with PRISm is also consistent with previous literature concluding the association of rapid FEV1 decline with lung cancer development.12 13 FEV1<80% is associated with inferior clinical outcomes and is also an independent risk factor for shorter overall survival in patients with NSCLC.35 36 Furthermore, the outcomes of patients with lung cancer with ongoing treatment can be estimated through lung function status.34 37
Mechanism
The lungs’ primary function is to exchange gases, and the quality of the air or substance being breathed in is very important for health. Inhaled hazardous substances can damage cilia which then fail to propel trapped pathogens,38 leading to prolonged exposure and eventually reducing respiratory function. In the absence of external factors, impaired lung function alone can initiate that vicious circle and consequently lung cancer-pro0moting events of inflammation, oxidative stress, DNA damage, fibrosis and connective tissue deposition in the airways.39 Genetic studies have also shown that pulmonary impairment can lead to lung cancer. Integrative analyses uncovered that pulmonary function can influence gene expression in lung tissue, which can then activate immune-related pathways that promote cancer growth.40 The relationship between impaired lung function and lung cancer is complex, as both conditions are often caused by cigarette smoking. However, a recent study using Mendelian randomisation analysis found that there is an independent link between lung function impairment and lung cancer risk, even after accounting for smoking. This suggests that there may be other factors, such as immune-related pathways, that contribute to the increased risk of lung cancer in people with impaired lung function.41 According to our study results, the feasibility of including impaired lung function as a criterion for determining eligibility for lung cancer screening remains uncertain. Lung cancer screening did not have a mortality reduction benefit in people with severe airflow limitation.21 42 By incorporating FEV1 into screening criteria, we may preferentially select people who are not healthy enough to undergo follow-up curative treatments for lung cancer, or whose life expectancy is so limited that they do not benefit from early detection of their cancer.
Strengths and limitations
Strengths: Our study has several notable strengths. First, it is a large cohort consisting of nearly half a million participants. Second, the long follow-up time of 23 years increased the statistical power and generalisability of findings. Third, the study used a rigorous method to exclude those with pre-existing lung cancer before data analysis, reducing reverse causality bias. Fourth, the robustness of the association was confirmed by consistent findings of association in sensitivity and stratified analyses. Finally, it is one of the rare studies that have examined the relationship between PRISm and AO for lung cancer incidence and mortality in the same cohort, with a particular focus on never-smokers.
Limitations: There also exist some limitations in our study. The study did not consider the possibility of transition to another type of lung function impairment, which may affect the findings. Our study did not include bronchodilator testing, which is necessary for COPD diagnosis, we used pre-bronchodilator spirometry and AO to describe individuals with compromised lung functions instead of COPD. The study cohort was based on a self-paying private health surveillance programme, mainly involving higher socioeconomic classes. The study cohort mainly represented residents in Taiwan, thus, the results may not be generalisable to other ethnicities or regions.