Introduction
Chronic kidney disease (CKD) and chronic respiratory diseases impose a significant global burden. The prevalence of CKD has risen alongside the increasing prevalence of risk factors, such as obesity and diabetes mellitus (DM), affecting more than 10% of the global population [1]. CKD ranks as the 12th leading cause of death globally and is a significant risk factor for cardiovascular disease [2]. Chronic respiratory disease affects over 7% of the global population, accounts for 7% of deaths, and is the third leading cause of mortality [3]. CKD and chronic respiratory disease share several common risk factors and frequently coexist. Patients with CKD exhibit a higher prevalence of pulmonary hypertension and chronic obstructive pulmonary disease (COPD), suggesting a close interaction between these two systems [4–6]. Nevertheless, lung and kidney function assessments are not routinely performed in clinical practice when managing patients with unrelated conditions.
Previous cross-sectional and prospective cohort studies have suggested an association between lung function decline and abnormal kidney function [7–9]. While a recent bidirectional Mendelian randomization study confirmed a causal relationship between abnormal kidney function and obstructive lung function [10], numerous observational studies reported that impaired lung function increases the risk of CKD [4,11–16]. Thus, there is a need to explore the precise nature and direction of the longitudinal relationship between kidney and lung function, which have yet to be elucidated. This study aimed to investigate the bidirectional association between kidney and lung function, emphasizing their interactive effects and the impact of longitudinal changes in both systems, within a community-based prospective cohort study in Korea.
Methods
Study participants
We recruited participants from the Ansan and Anseong cohorts of the Korean Genome and Epidemiology Study, a series of ongoing community-based prospective cohort studies conducted by the National Research Institute of Korea. The study is designed to provide a large-scale health database for investigating the etiologies of chronic diseases in South Korea. The cohort comprised 10,030 residents of Ansan (rural area) and Anseong (urban area) in Gyeonggi Province, South Korea, aged 40–69 years. Recruitment was conducted via invitations, and informed consent was obtained from all participants.
Data from health examinations and biochemical tests were collected biennially between 2001 and 2014 [17]. Owing to the lack of quality-controlled spirometry data at baseline (2001–2002) and the first follow-up (2003–2004), 7,515 participants from the second follow-up (2005–2006) were initially screened. Of these, 5,487 participants with ≥2 valid spirometry and ≥2 estimated glomerular filtration rate (eGFR) measurements between the second and sixth follow-ups were selected for further analysis. Participants with moderately reduced baseline eGFR (<60 mL/min/1.73 m2), a history of chronic lung disease, or missing data on chronic lung disease were excluded. Chronic lung disease was defined as either a ratio of forced expiratory volume in 1 second (FEV1) to forced vital capacity (FVC) <70% or current use of inhalers. A total of 4,388 participants were enrolled in the final analysis (Fig. 1).
Anthropometric and biochemical data
Demographic and socioeconomic data, including age, sex, height, body mass index, smoking history (in pack-years), alcohol use, residential area, education, income, physical activity, and systolic and diastolic blood pressure, were collected through a comprehensive health and lifestyle questionnaire [17].
We categorized the participants into three educational levels based on their most recent educational degrees. Participants whose educational level was lower than middle school, lower than high school, and higher than high school were classified as “low,” middle,” and “high,” respectively. Participants were also categorized into three economic levels based on monthly income: those earning less than $850, between $850 and $1,700, and above $1,700 were classified as “low,” “middle,” and “high,” respectively.
Biochemical data, including serum hemoglobin, total cholesterol, high-density lipoprotein cholesterol, triglycerides, and creatinine levels, were measured from blood samples collected after an 8-hour fast. Blood specimens were collected in serum separator tubes and two ethylenediaminetetraacetic acid tubes. Proteinuria was defined as a protein level ≥+1 in the dipstick urine test (URISCAN Pro II; YD Diagnostics Corp.) using a 10-mL midstream urine sample [18].
Comorbidities, including hypertension, DM, and dyslipidemia, were diagnosed based on clinical data and medical history. Participants with hypertension consisted of those taking antihypertensive medications or with blood pressure readings ≥140/90 mmHg. A trained technician measured blood pressure using a mercury sphygmomanometer after the participants rested for 5 minutes in a seated position. Participants were instructed to abstain from alcohol consumption for 7 days and smoking for 24 hours before the assessment. Blood pressure was initially measured in both arms. Subsequently, two additional measurements were taken from the arm with the higher blood pressure. The second blood pressure measurement was used to establish the diagnosis [19]. Participants with DM referred to those with a confirmed and ongoing diagnosis of DM. Individuals meeting any of the following criteria were categorized as having DM: a fasting blood glucose level ≥126 mg/dL after an 8-hour fast, a 2-hour post-load glucose level ≥200 mg/dL from a 75-g oral glucose tolerance test, a hemoglobin A1c level ≥6.5%, or the use of antidiabetic medications, including insulin. Finally, participants with a history of dyslipidemia or those using lipid-lowering medications were classified as having dyslipidemia.
Assessment of lung function
Lung function was assessed at each visit using spirometry (Vmax-2130; Sensor-Medics). Participants visited the Korea University Ansan Hospital or Ajou University Center for Clinical Epidemiology to undergo testing under skilled supervision. To minimize measurement errors, the lung function parameter was taken from at least three acceptable measurements at each follow-up. The largest FVC and FEV1 values were used in the analysis, according to American Thoracic Society guidelines [20]. Morris and Polgar’s equation referenced normal lung function [21].
Estimated glomerular filtration rate
Kidney function was evaluated using eGFR, which was measured at each visit. The eGFR values were calculated according to the CKD-EPI (Chronic Kidney Disease Epidemiology Collaboration) equation as follows: 142 × min (serum creatinine/κ)α × max (serum creatinine /κ)–1.200 × 0.9938Age × (1.012 if female), where min ( ) is the minimum between input and 1.0, max ( ) is the maximum between input and 1.0, κ = 0.9 (male) or 0.7 (female), α = –0.302 (male) or –0.241 (female) [22].
Statistical analysis
Baseline characteristics are summarized as mean ± standard deviation or median (first quartile [Q1]–third quartile [Q3]) for continuous variables and as frequency and percentage (%) for categorical variables. Sex differences for each variable were assessed using the Student t test, Mann-Whitney U test, chi-square test, or Fisher exact test, as appropriate. Pearson’s correlation coefficient was used to evaluate the association between baseline eGFR and baseline lung function.
The primary aim of this study was to analyze longitudinal changes in patient responses to covariates, including eGFR, FVC, FEV1, and FEV1/FVC. Linear mixed-effects models were employed to estimate the effects of eGFR on lung function and vice versa. In our linear mixed-effects model, random intercepts for each subject were included to account for the variability among individuals, and an autoregressive correlation structure of order 1, denoted as AR(1), was applied to account for the temporal correlation among repeated measurements within subjects. Lung function and eGFR were treated as time-varying variables. When each variable (FVC, FEV1, FEV1/FVC, and eGFR) was treated as an independent variable, it was decomposed into “between-” and “within-effects” [23]. The decomposed “between” term represents the average of covariate values repeatedly measured within a patient, allowing for the examination of the overall effect of the covariates on the response variable. In contrast, the decomposed “within” term represents deviations of the covariate from the patient’s mean, allowing for an examination of the effect of these deviations on the response.
The long-term association between eGFR and lung function, with eGFR as the independent variable, was adjusted for age, residential area, height, pack-years smoked (in males only), baseline lung function, physical activity, income, and educational level. Adjustments for pack-years smoked were made exclusively for males because of the disproportionately low proportion of female smokers. Consequently, the inclusion of ever-smoking females in the analysis was deemed impractical and omitted. Baseline lung function was adjusted to account for interpersonal variations in the magnitude of longitudinal decline [24]. To explore the longitudinal association between lung function and eGFR, with lung function (FVC, FEV1, and FEV1/FVC) as independent variables, a multiple linear mixed-effects model was employed, adjusted for age, residential area, height, pack-years smoked (in males only), baseline eGFR, physical activity, income, educational level, alcohol consumption, systolic blood pressure, hypertension, and DM. The detailed information for each analytical model is presented in Table 1.
Statistical analyses were performed using SAS version 9.4 (SAS Institute). Statistical significance was set at p < 0.05.
Ethics approval and consent to participate
The Korea Disease Control and Prevention Agency obtained written informed consent from all participants for the collection of their data, and the Institutional Review Board of Severance Hospital approved the study protocol (4-2022-0558). All studies were conducted in accordance with approved protocols and relevant guidelines and regulations.
Results
Baseline characteristics
The baseline characteristics of the 4,388 participants are presented in Table 2. The participants’ mean age was 53.4 ± 7.7 years, and 2,092 were males (47.7%). The demographic and laboratory data exhibited statistically significant variations based on sex. A higher proportion of males were ever-smokers (73.7% vs. 2.3%, p < 0.001) and alcohol drinkers (81.2% vs. 30.5%, p < 0.001) compared with females. Males had greater FVC values (4.27 L vs. 3.02 L, p < 0.001) and FEV1 values (3.37 L vs. 2.46 L, p < 0.001) than females. Conversely, females exhibited a higher FEV1/FVC ratio than males (81.6% vs. 78.9%, p < 0.001). Additionally, males had a higher eGFR (80.2 mL/min/1.73 m2 vs. 78.2 mL/min/1.73 m2, p < 0.001) and greater incidence of proteinuria (2.0% vs. 1.0%, p = 0.008) than females.
The median follow-up period for the study participants was 8 years (interquartile range, 6–8 years), with spirometry and eGFR measurements taken at a median of four times (interquartile range, 3–5 times) during this period.
Cross-sectional association between estimated glomerular filtration rate and lung function
Cross-sectional associations between eGFR and lung function at baseline for males and females are presented in Fig. 2. Both sexes exhibited a modest positive correlation between eGFR and FVC as well as FEV1 at baseline. In contrast, the FEV1/FVC ratio was positively correlated with eGFR only in females. In males, the Pearson correlation coefficients for eGFR with FVC, FEV1, and FEV1/FVC were r = 0.049 (p = 0.02), r = 0.051 (p = 0.02), and r = 0.011 (p = 0.61), respectively. The corresponding values for females were r = 0.045 (p = 0.03), r = 0.062 (p = 0.003), and r = 0.057 (p = 0.006).
Longitudinal association between estimated glomerular filtration rate and lung function
We conducted primary analyses to examine the effect of eGFR on lung function (Table 3) as well as the effect of lung function on eGFR (Table 4). Additionally, sensitivity analyses were performed by incorporating adjustment variables to assess the consistency of the results. The details of the analyses are presented in Table 1, and the corresponding results are presented in Supplementary Tables 1–8 (available online). All analyses were conducted separately for males and females.
In Analysis 1, decreases in eGFR variation within patients were longitudinally associated with reductions in all three lung function parameters (Table 3). In males, for each 1 mL/min/1.73 m2 decrease in eGFR, the FVC decreased by 8.1 mL, FEV1 decreased by 10.6 mL, and the FEV1/FVC ratio decreased by 0.11% (all p < 0.001). In females, for each 1 mL/min/1.73 m2 decrease in eGFR, the FVC decreased by 5.1 mL, FEV1 decreased by 6.0 mL, and the FEV1/FVC ratio decreased by 0.07% (all p < 0.001). The estimated values remained consistent despite stringent adjustments. Detailed analyses of each model are presented in Supplementary Table 1 (available online).
The longitudinal association between reductions in all three lung function parameters and a decrease in eGFR was consistent when analyzed separately in three age groups (≤50, ≤60 and >50, and >60 years). Meanwhile, in males, there was a greater decrease in FVC (7.6 mL for age ≤50 years, 7.8 mL for age ≤60 and >50 years, and 10.0 mL for age >60 years) and FEV1 (9.7 mL for age ≤50 years, 11.1 mL for age ≤60 and >50 years, and 11.9 mL for age >60 years) for each 1 mL/min/1.73 m2 decrease in eGFR with increased age. For females, decrease in FVC (5.4 mL for age ≤50 years, 4.2 mL for age ≤60 and >50 years, and 6.0 mL for age >60 years) and FEV1 (6.2 mL for age ≤50 years, 5.5 mL for age ≤60 and >50 years, and 6.5 mL for age >60 years) for each 1 mL/min/1.73 m2 decrease in eGFR was greatest in age >60 years, followed by age ≤50 years, and age ≤60 and >50 years, respectively (Supplementary Tables 2–4, available online).
In the reverse analysis (Analysis 2), the within-patient decline in all three lung function parameters was longitudinally associated with a reduction in eGFR (Table 4). In males, for each 1-unit decrease in FVC, FEV1, and FEV1/FVC, the eGFR decreased by 11.30 mL/min/1.73 m2, 13.87 mL/min/1.73 m2, and 0.80 mL/min/1.73 m2, respectively (all p < 0.001). In females, for each 1-unit decrease in the same parameters, the eGFR decreased by 10.96 mL/min/1.73 m2, 13.62 mL/min/1.73 m2, and 0.69 mL/min/1.73 m2, respectively (all p < 0.001). Similar results were observed after adjusting for total cholesterol, hemoglobin, and proteinuria levels. A complete analysis of each model is presented in Supplementary Table 5 (available online).
The longitudinal association between reductions in eGFR and all three lung function indexes was consistent when analyzed separately in three age groups (≤50 years, ≤60 and >50 years, and >60 years). However, no pattern was observed across the age groups for both sexes (Supplementary Tables 6–8, available online).
Discussion
In this study, we demonstrated a bidirectional and longitudinal correlation between kidney and lung function in the general population. Baseline cross-sectional analysis revealed a positive correlation between eGFR and FVC and FEV1 in both sexes, as well as a positive correlation between eGFR and FEV1/FVC ratio in females. Multiple linear mixed-effects model analyses produced similar results, indicating that the longitudinal decline in eGFR was associated with a concurrent decline in FVC, FEV1, and FEV1/FVC, and vice versa, even after adjusting for potential confounding factors. While the study did not establish a causal relationship, the large sample size and comprehensive data collected at four follow-up visits over 8 years strongly support a close association between the two systems.
The findings of this study are consistent with previous research indicating that a decline in lung function may be a potential risk factor for kidney function [4,11–16]. In a cross-sectional study involving 356 Italian patients with COPD aged 65 years or older, the prevalence of CKD was found to be higher compared to that in healthy participants [4]. Another prospective cohort study involving 14,949 White and Black Americans, followed over 26 years at 5-year intervals, suggested that baseline FVC may serve as a long-term predictor for end-stage kidney disease [13]. Similar results were reported in a study conducted in a Swedish population, where baseline FEV1 and FVC predicted the incidence of CKD in males over a 30-year interval [16]. Additionally, a retrospective cohort study of 10,128 South Koreans conducted over 7 years indicated that a decrease in baseline FEV1/FVC was associated with a greater long-term risk of developing CKD [15].
However, previous studies designated dependent variables as endpoints and considered lung function measurements only once at baseline, leading to inter-individual comparisons rather than examining the interplay between the two systems within a single individual over time. As a result, there is still a limited understanding of the continuous relationship between lung and kidney function over an extended period. In our study, we defined both independent and dependent variables as continuous values of eGFR and lung function, aiming for a more comprehensive understanding of how a decline in lung function serves as a risk factor for kidney function.
Conversely, research on the potential risk of decreased lung function due to a decline in kidney function is relatively limited. In a prospective cohort study involving Swedish patients with abnormal kidney function or CKD, lower baseline eGFR was a stronger indicator for comorbidities, including obstructive and particularly restrictive lung dysfunction [9]. Proteinuria, which is evidence of kidney injury, is associated with decreased hemoglobin-adjusted %DLCO [7]. A recent bidirectional Mendelian randomization analysis revealed that a genetic predisposition for greater FEV1/FVC was not associated with higher eGFR, while the reverse analysis showed statistical significance [10]. However, the results are based upon Mendelian assumptions, which consider that an individual’s genotype is directly translated into their phenotype [25]. Furthermore, this study exclusively focused on European ethnicities, which limits its generalizability. In contrast, we verified the relationship between kidney and lung function based on actual clinical data from the Korean general population. This finding highlights a relationship that has largely been overlooked in previous studies, thereby ensuring the reliability of our results.
Multiple hypotheses exist regarding the biological and physiological factors that contribute to the interaction between the lung and kidney systems. One hypothesis suggests that abnormal kidney function may exacerbate pulmonary edema, leading to lung congestion and impaired alveolar compliance, which can result in the over-distension of fluid-filled alveoli [26,27]. This hypothesis is supported by the observed relief of obstructive lung patterns in patients with CKD following hemodialysis treatment [28]. Another hypothesis highlights the role of hypoxia [29,30] and right ventricular dysfunction [31] resulting from impaired lung function in stimulating the renin-angiotensin-aldosterone system, which is closely linked to kidney function. However, as the eGFR values of our participants were predominantly in the normal range, this proposed mechanism should be interpreted with caution when applied to our findings.
Alternatively, aging is associated with chronic low-grade inflammation, with elevated levels of proinflammatory cytokines such as interleukin (IL)-1β, IL-6, and tumor necrosis factor alpha [32]. With the kidneys playing an essential role in their excretion, a physiological decrease in eGFR may contribute to persistent inflammation, which further modulates glomerular and tubule-interstitial scarring [33]. In addition, such mediators of inflammation have been linked to endothelial leaking and organ inflammation resulting in progressive organ dysfunction, especially in the lungs given their extensive vascularity [34].
Meanwhile, there have been numerous other studies that explain the lung-kidney and kidney-lung interactions through various mechanisms, including but not limited to oxidative stress and dysregulation of acid-base balance [35]. Further research into the molecular mechanism is needed to strengthen the holistic understanding of the bidirectional relationship between the two organs.
The strengths of our study lie in its longitudinal design, which tracks individual participants at multiple time points over 8 years. In contrast to previous studies, we used a large-scale, community-based cohort with ≥4 spirometry and eGFR measurements during the study period. Through multiple linear mixed-effects model analysis, the study effectively demonstrated that a decline in kidney function is longitudinally correlated with a decline in lung function and vice versa. This provides valuable insights into how physiological changes in one system are associated with alterations in the other system over the long term. Furthermore, the use of a large-scale community-based cohort is relatively uncommon in this field of research.
Nevertheless, this study had several limitations. First, our participants were relatively healthy individuals with lung function and eGFR values within the normal range and without overt respiratory symptoms. Hence, the generalization of our findings to patients with CKD or severe lung disease should be done with caution. Additionally, this study did not address the longitudinal association between eGFR and subjective respiratory symptoms, which are critical indicators of obstructive lung disease. Finally, we assessed lung function exclusively using pre-bronchodilator spirometry without incorporating post-bronchodilator spirometry or imaging studies. However, by conducting spirometry under strict quality control, we ensured the reliability of our lung function assessment.
In conclusion, a decline in lung function is longitudinally associated with a decline in kidney function and vice versa, underscoring a potential correlation between the two systems. Clinical guidelines recommend close monitoring of both systems in patients with impaired function. Further research should investigate the benefits of preventive measures targeting one system to enhance the health of the other.
PDF Links
PubReader
ePub Link
XML Download
Full text via DOI
Download Citation
Supplement table 1
Print
