Lung cancer•

Impaired lung function and lung cancer risk in 461 183 healthy individuals: a cohort study

Abstract

Background It has been known that smoking and various lung diseases including lung cancer can cause lung function impairment. However, the impact of different types of lung function impairments, such as preserved ratio impaired spirometry (PRISm) and airflow obstruction (AO), on the incidence and mortality of lung cancer in both general and never-smoker populations remains unclear. We wished to examine the effect of lung function impairments on lung cancer risks.

Methods This was a retrospective cohort study (1 January 1994 to 31 December 2017) of individuals from a health surveillance programme in Taiwan who underwent baseline spirometry tests at the entry point. PRISm was defined as an FEV1/FVC (forced expiratory volume in 1 s/forced vital capacity) ratio >0.7 and FEV1 <0.8, while AO was defined as an FEV1/FVC ratio <0.7. Cox proportional hazards models and cubic spline curves were used to examine the associations between lung function impairments and lung cancer risks.

Results The study included 461,183 individuals, of whom 14.3% had PRISm and 7.9% had AO. A total of 4038 cases of lung cancer and 3314 lung cancer-related deaths were identified during the 23 years of follow-up. Individuals with PRISm and AO exhibited a higher risk of lung cancer incidence and mortality compared with those with normal lung function. The adjusted HRs and 95% CIs were 1.14 (1.03 to 1.26) and 1.23 (1.10 to 1.37) in the overall cohort, and 1.08 (0.93 to 1.24), and 1.23 (1.05 to 1.45) in the never-smoker cohort. The risks of both developing and dying of lung cancer increased with the severity levels of lung function impairments and lower FEV1 values.

Conclusion Impaired lung function is associated with increased risks of developing lung cancer and subsequent mortality. The study highlights the importance of considering lung function in lung cancer screening for better candidate selection.

What is already known on this topic

  • Impaired lung function is a common condition that can be observed in people with smoking habits and other respiratory illnesses including lung cancer. However, the effect of lung function impairment alone on the risks of lung cancer incidence and mortality is not clear. Early detection of lung cancer is essential for effective disease management, and lung cancer screening is a key preventive measure that can help achieve this. However, current lung cancer screening guidelines only consider age and smoking history, not lung function status.

Why this study adds

  • Impaired lung function is associated with an increased risk of lung cancer in a large Asian cohort. Nonetheless, among never-smoking individuals exhibiting spirometrically defined preserved ratio impaired spirometry (PRISm)/airflow obstruction, the observed risks were found to lack statistical significance, with the exception of lung cancer mortality within the non-smoking PRISm subgroup.

How this study might affect research, practice or policy

  • Lung function status should be taken into consideration in lung cancer screening criteria. The importance of monitoring and addressing lung function impairment in lung cancer risk management should also be widely shared with the medical community and the public.

Introduction

Lung cancer is the deadliest cancer, with the highest mortality rate and second-highest incidence rate. In 2020 alone, it accounted for 2 206 771 new cases (11.4%) and caused 1 796 144 deaths (18%) worldwide.1 Cigarette smoking stands out as the most significant contributor, being responsible for approximately 80–90% of all lung cancer cases.2 Other well-known risk factors include exposure to carcinogens like tobacco smoke, radon and air pollution, as well as pre-existing pulmonary conditions such as chronic obstructive pulmonary disease (COPD) and age-related intrinsic factors.3 Lung function status is not taken into consideration in current screening criteria which are based only on age and pack-year smoking history.4

It is well-recognised that individuals with smoking habits, lung cancer or respiratory illnesses have impaired lung function. For example, tuberculosis and asthma are risk factors for lung function impairment among Korean non-smokers,5 and air pollution, even at very low levels, has adverse effects on lung function in adults.6 There have been studies on lung function and lung cancer however most studies targeted smokers or patients with COPD or preserved ratio impaired spirometry (PRISm) or lung cancer. A retrospective study of COPD infradiagnosis in lung cancer (RECOIL) study reported that COPD is often underdiagnosed in patients with lung cancer and this can lead to more advanced cancer stages and possibly worse outcomes.7 The severity of COPD is independently associated with the risk of developing lung cancer8 and smoking status modifies this association.9 Another form of lung function impairment, the restrictive type PRISm, was also linked to an elevated mortality risk. This association appears unrelated to smoking, obesity or pre-existing lung disease, despite PRISm being transient in many patients.10 The prevalence of PRISm also was as high as 17–24% in a retrospective analysis of over 20 000 spirometries.11

The forced expiratory volume in 1 s (FEV1) decline rate may be a biomarker for developing lung cancer.12 It may still be a predictor of lung cancer, especially for women,13 however, genetically decreased FEV1 is not causally correlated with lung cancer incidence.13 In patients with small cell lung cancer (SCLC), a low FEV1, not COPD, is a predicting factor for poor treatment outcomes.14 Some studies, however, have proposed that the variance in lung function measures among smokers is not fully explained by the smoking intensity and the number of pack-years smoked did not affect the lung cancer rate in patients with COPD.15

A study in 201716 said the annual frequency of lung cancer in never-smokers had more than doubled in the past 7 years, from 13% to 28%. This increase was due to an absolute increase in the number of never-smokers developing lung cancer, not simply a change in the ratio of never-smokers to current and former smokers. Patients with lung cancer who had never smoked typically presented with non-specific symptoms and most were detected on incidental imaging.16 A previous study found that the proportion of non-smoking patients with non-small cell lung cancer (NSCLC) increased between 1990 and 2013. This increase was seen in a large, diverse patient population and independent of sex, stage and race/ethnicity. This suggests that the actual incidence of lung cancer in never-smokers is increasing.17 A UK Biobank study on 222 274 never-smokers aged 40–69 years with 4 years of follow-up found that lung function can be a modest predictor of lung cancer risk in never-smokers, however, there acknowledged the potential selection bias of containing more male and white participants in the study cohort.18 Overall, the relationship between lung function and lung cancer is complex and multifactorial.

We hypothesised that the presence of impaired lung function statuses alone may be an independent risk factor for lung cancer. The objective of our study was to investigate the potential causal relationship between lung function impairments, particularly restricted type (PRISm) and lung cancer risks. The implication of our study was to suggest the inclusion of lung function status in future lung cancer screening criteria.

Method

It was a retrospective cohort using baseline data of participants enrolled in a health screening programme in Taiwan between 1994 and 2017. Participants were followed-up for lung cancer development and death until December 2017.

This study included 646 987 healthy individuals (aged ≥20). Getting signed consent forms, baseline data on demographics, lifestyle and medical history were collected using a self-administered questionnaire. A standardised physical examination and laboratory tests were performed including spirometry for assessing lung function using HI-501, HI-701 or HI-801 spirometers from Chest M.I. Tokyo, Japan. Results were measured as percentages of FEV1 and forced vital capacity (FVC). Participants were excluded based on criteria of missing values of FEV1 or FVC (n=109 122), FEV1/FVC>1 (n=71 818), FEV1 or FVC percentage >150 or <25 (n=4864) and already diagnosed with lung cancer (n=152). The final cohort consisted of 461 031 participants (figure 1). Ethical approval for this study was obtained from the Institutional Review Board at the China Medical University (Approval Number: CMUH111-REC2-118). All participants gave consent to analyse their unidentifiable data before being included in this study. All data analyses were conducted at the Data Science Center of the Ministry of Health and Welfare in Taiwan, where all data were de-identified to protect participant confidentiality.

Figure 1
Request permissionrequest permission icon
Figure 1

Study Flowchart. FEV1, forced expiratory volume in 1 s; FVC, forced vital capacity.

Patient and public involvement

Patients and/or the public were not involved in the design, conduct, reporting or dissemination of this research.

Variables and definitions

Based on spirometry results, participants were classified into normal, PRISm and airflow obstruction (AO). PRISm was defined as FEV1/FVC ratio >0.7 and FEV1<80% predicted, AO as FEV1/FVC ratio <0.7 and normal as FEV1/FVC ratio >0.7 and FEV1>80% predicted. Severity levels of PRISm19 were determined based on FEV1 values: >70, 60–69, 50–59 and <50. For AO, Global Initiative for Chronic Obstructive Lung Disease stages I–IV were applied, stage I (mild) for FEV1 values >80, II (moderate) for 50–79, III (severe) for 30–49 and IV (very severe) for <30.

The primary outcomes of the study were newly diagnosed lung cancer cases after the entry time point and lung cancer-related deaths during the follow-up period by matching patients’ identification numbers with the Nationwide Registry of Patients with Catastrophic Illness and National Death File until December 2017. Causes of death were coded according to the International Classification of Diseases, ninth version.

Several covariates were adjusted during analyses. Diabetes and hypertension cases were identified based on medical history or positive screening results such as fasting blood glucose 126 mg/dL or higher and systolic blood pressure 140 mm Hg or higher. Other variables included age, sex, educational levels, smoking status, drinking status, body mass index (BMI) and physical activity.

Statistical processing and analysis

Descriptive statistics were used to tabulate the baseline characteristics of participants by lung function impairment categories. The Multivariable Cox proportional hazards model was applied to investigate the relationship between lung function impairment and the risks of lung cancer incidence and mortality. HRs and 95% CIs were calculated, adjusting for confounding variables of age, sex, education level, smoking status, alcohol consumption, BMI, physical activity level, hypertension and diabetes status.

To ensure the credibility of our findings and to eliminate any potential bias arising from reverse causation, we conducted sensitivity analyses. We analysed specific subgroups by excluding individuals who smoked and those who developed lung cancer within 3 years of the health screening programme entry. For instance, the exclusion of participants with baseline hypertension in a study from Hong Kong20 and the exclusion of participants with asthma in the UK Biobank study.10 Stratified analyses by age, gender, smoking status and histological cell types were also performed. Furthermore, HRs were plotted against continuous values of FEV1 on cubic spline curves to observe the correlation trends. SAS analytical software V.9.4 was applied, and all tests were two-sided with a significance level of α=0.05.

Results

Out of 461 031 individuals, 14.3% had PRISm, 7.9% had AO and 77.8% had normal lung function. Among the never-smoker subgroup, 13.8% had PRISm, 3.5% had AO and 82.7% had normal lung function. During the 23-year follow-up period of the study, 4038 cases of lung cancer and 3314 lung cancer-related deaths were identified. The incidence rate of lung cancer was 28.55 cases per 100 000 person-years, and the mortality rate was 21.90 cases per 100 000 person-years.

Characteristics of participants by lung function statuses

The main characteristics of the study participants by lung function statuses are shown in table 1. In comparison to the AO group, the individuals in the PRISm group were relatively older age, that is, over 40 and more likely to be women, never-smokers, non-drinkers, higher BMI, that is, above 25, physically less active, with comorbidities of diabetes and hypertension. However, there were no significant differences in respiratory symptoms between the two groups. Characteristics of the study subjects by severity levels of restrictive lung condition (PRISm) were presented in online supplemental table 1.

Table 1
|
Characteristics of participants by lung function status

Risks of lung cancer and lung function impairments

Findings for the overall cohort in table 2 showed individuals diagnosed with PRISm exhibited a 14% higher risk of developing lung cancer and a 23% higher risk of dying from lung cancer compared with individuals with normal lung function. Similarly, individuals with AO had a 29% increased risk of lung cancer incidence and a 30% higher risk of lung cancer mortality. Notably, the risks of developing lung cancer were found to escalate in proportion to the severity levels of lung function impairment.

Table 2
|
Risk of lung cancer incidence and mortality by lung function impairment severities in overall and never-smoker cohorts

Lung cancer risks by lung function impairments in sensitivity analyses

After excluding smoking individuals, the risks of developing lung cancer were still higher among those with impaired lung function. Analyses for the subgroup of never-smokers in table 2 revealed individuals with PRISm had an 8% higher chance of developing lung cancer and a 23% higher chance of dying from lung cancer compared with those with normal lung function. Similarly, for individuals with AO, the increased risks were 16% for lung cancer incidence and 23% for lung cancer mortality. These findings remained consistent even when lung cancer cases occurring within the first 3 years were excluded to avoid reverse causation bias, as demonstrated in online supplemental figure 1A and B.

Lung cancer risks by lung function impairments in subgroups analyses

Stratification on the individuals with PRISm by age, gender and smoking status, individuals with impaired lung function were consistently found to have elevated risks of both lung cancer incidence and mortality (figure 2). Those aged over 60 had higher risks compared with those under 60. Women exhibited a higher risk of lung cancer mortality compared with men. Current and ex-smokers had a higher risk of lung cancer incidence compared with never-smokers. In stratified analyses by histological types of lung cancer for never-smokers, the risks of developing or dying from squamous cell carcinoma were found to be higher than those of adenocarcinoma (online supplemental figure 2).

Figure 2
Request permissionrequest permission icon
Figure 2

Adjusted HRs of lung cancer (A) incidence and (B) mortality for impaired lung function individuals compared with normal ones by age, gender and smoking status in PRISm. PRISm, preserved ratio impaired spirometry.

Relationship between continuous values of FEV1 and HRs

Cubic spline models were employed to analyse the correlation between FEV1 values and the likelihood of lung cancer incidence and mortality within the PRISm group of never-smokers. The findings from figure 3 revealed a noteworthy inverse relationship, suggesting that under the predicted FEV1 80%, individuals with lower FEV1 values had elevated risks of developing and dying from lung cancer. Under the FEV1 value of 50%, never-smokers with PRISm exhibited significantly higher risks for both lung cancer incidence and mortality. Moreover, the slopes of cubic spline curves in squamous cell carcinoma were steeper, indicating greater risks, compared with adenocarcinoma (figures not shown).

Figure 3
Request permissionrequest permission icon
Figure 3

Cubic spline curves for risk of lung cancer (A) incidence and (B) mortality in overall PRISm and (C) incidence and (D) mortality in never-smokers PRISm by FEV1 values. FEV1, forced expiratory volume in 1 s; PRISm, preserved ratio impaired spirometry.

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.

Conclusion

There is a significant association between impaired lung function and increased risks of lung cancer incidence and mortality, in the overall cohort. That relationship is independent of other potential confounding factors. However, the association was insignificant in the subcohort of never-smokers.

Article metrics
Altmetricopen-url
Dimensionsopen-url