Combined effects of lung function, blood gases and kidney function on the exacerbation risk in stable COPD: Results from the COSYCONET cohort

Open ArchivePublished:June 10, 2019DOI:https://doi.org/10.1016/j.rmed.2019.06.007

      Abstract

      Rationale

      Alterations of acid-base metabolism are an important outcome predictor in acute exacerbations of COPD, whereas sufficient metabolic compensation and adequate renal function are associated with decreased mortality. In stable COPD there is, however, only limited information on the combined role of acid-base balance, blood gases, renal and respiratory function on exacerbation risk grading.

      Methods

      We used baseline data of the COPD cohort COSYCONET, applying linear and logistic regression analyses, the results of which were implemented into a comprehensive structural equation model. As most informative parameters it comprised the estimated glomerular filtration rate (eGFR), lung function defined via forced expiratory volume in 1 s (FEV1), intrathoracic gas volume (ITGV) and (diffusing capacity for carbon monoxide (DLCO), moreover arterial oxygen content (CaO2), partial pressure of oxygen (PaCO2), base exess (BE) and exacerbation risk according to GOLD criteria. All measures were adjusted for age, gender, body-mass index, the current smoking status and pack years.

      Results

      1506 patients with stable COPD (GOLD grade 1–4; mean age 64.5 ± 8.1 y; mean FEV1 54 ± 18 %predicted, mean eGFR 82.3 ± 16.9 mL/min/1.73 m2) were included. BE was linked to eGFR, lung function and PaCO2 and played a role as indirect predictor of exacerbation risk via these measures; moreover, eGFR was directly linked to exacerbation risk. These associations remained significant after taking into account medication (diuretics, oral and inhaled corticosteroids), whereby corticosteroids had effects on exacerbation risk and lung function, diuretics on eGFR, BE and lung function.

      Conclusion

      Even in stable COPD acid-base metabolism plays a key integrative role in COPD risk assessment despite rather small deviations from normality. It partially mediates the effects of impairments in kidney function, which are also directly linked to exacerbation risk.

      Keywords

      1. Introduction

      Chronic obstructive pulmonary disease (COPD) and chronic kidney disease (CKD) are major health issues [
      • Inker L.A.
      • Astor B.C.
      • Fox C.H.
      • et al.
      KDOQI US commentary on the 2012 KDIGO clinical practice guideline for the evaluation and management of CKD.
      ,
      • Saran R.
      • Robinson B.
      • Abbott K.C.
      • et al.
      US renal data system 2016 annual data report: Epidemiology of kidney disease in the United States.
      ], with cigarette smoking and age as common risk factors [
      • Vogelmeier C.F.
      • Criner G.J.
      • Martinez F.J.
      • et al.
      Global strategy for the diagnosis, management, and prevention of chronic obstructive lung disease 2017 report: GOLD executive summary.
      ,
      • Haroun M.K.
      • Jaar B.G.
      • Hoffman S.C.
      • et al.
      Risk factors for chronic kidney disease: a prospective study of 23,534 men and women in Washington County, Maryland.
      ,
      • Fox C.S.
      • Larson M.G.
      • Leip E.P.
      • et al.
      Predictors of new-onset kidney disease in a community-based population.
      ]. According to the latest WHO estimates, currently 64 million people have COPD [

      WHO, Chronic obstructive pulmonary disease. Secondary WHO, Chronic obstructive pulmonary disease.

      ], and approximately 4–8% of the COPD population additionally suffer from renal impairment [
      • Baty F.
      • Putora P.M.
      • Isenring B.
      • et al.
      Comorbidities and burden of COPD: a population based case-control study.
      ,
      • Gjerde B.
      • Bakke P.S.
      • Ueland T.
      • et al.
      The prevalence of undiagnosed renal failure in a cohort of COPD patients in western Norway.
      ,
      • Yoshizawa T.
      • Okada K.
      • Furuichi S.
      • et al.
      Prevalence of chronic kidney diseases in patients with chronic obstructive pulmonary disease: assessment based on glomerular filtration rate estimated from creatinine and cystatin C levels.
      ]. In COPD, exacerbations exert a major impact on health status and disease progression [
      • Wedzicha J.A.
      • Singh R.
      • Mackay A.J.
      Acute COPD exacerbations.
      ]. During these, hypercapnic respiratory failure (HRF) is common and associated with poor clinical outcome [
      • Seiler F.
      • Trudzinski F.C.
      • Kredel M.
      • et al.
      [Update: acute hypercapnic respiratory failure].
      ], whereas adequate metabolic compensation and renal function have been shown to decrease mortality in exacerbations [
      • Ucgun I.
      • Oztuna F.
      • Dagli C.E.
      • et al.
      Relationship of metabolic alkalosis, azotemia and morbidity in patients with chronic obstructive pulmonary disease and hypercapnia.
      ]. Both, the renal and the pulmonary system are major modulators of the acid-base balance, complementing each other to maintain pH homeostasis [
      • Hopkins E.
      • Sharma S.
      Physiology, Acid Base Balance.
      ]. Impairment in renal ammonia-genesis and bicarbonate reabsorption might lead to metabolic dysregulation [
      • Kraut J.A.
      • Kurtz I.
      Metabolic acidosis of CKD: diagnosis, clinical characteristics, and treatment.
      ], especially in patients with limited ventilatory capacity. Ventilatory demands and tissue oxygen delivery are closely linked to acid-base balance disturbances via the oxygen-haemoglobin dissociation curve [
      • Hodgkin J.E.
      • Soeprono F.F.
      • Chan D.M.
      Incidence of metabolic alkalemia in hospitalized patients.
      ]. Tissue oxygenation and oxygen content of arterial blood also depend on haemoglobin concentration and its regulation by renal secretion of erythropoietin [
      • Babitt J.L.
      • Lin H.Y.
      Mechanisms of anemia in CKD.
      ]. Diuretics and corticosteroids that are often administered in COPD patients can interfere with the acid-base system due to their influence on renal ion transport [
      • Luke R.G.
      • Galla J.H.
      It is chloride depletion alkalosis, not contraction alkalosis.
      ]. In stable COPD a number of comorbidities are known to be linked to exacerbation risk [
      • Westerik J.A.
      • Metting E.I.
      • van Boven J.F.
      • et al.
      Associations between chronic comorbidity and exacerbation risk in primary care patients with COPD.
      ,
      • Kahnert K.
      • Alter P.
      • Young D.
      • et al.
      The revised GOLD 2017 COPD categorization in relation to comorbidities.
      ], but the role of acid-base disturbances and kidney function, including their interactions with other risk factors, has not been analysed in detail. The present study aimed to delineate the relationship between lung function, blood gases, kidney function and exacerbation risk in a comprehensive manner. For this purpose, we used data from the COPD cohort COSYCONET (Systemic Consequences - Comorbidities Network), which is a multicentre prospective study focusing on the role of comorbidities and systemic inflammation in COPD [
      • Karch A.
      • Vogelmeier C.
      • Welte T.
      • et al.
      The German COPD cohort COSYCONET: aims, methods and descriptive analysis of the study population at baseline.
      ].

      2. Methods

      2.1 Study population

      2,741 patients with stable COPD (age ≥40 years) were enrolled in the COSYCONET cohort [
      • Karch A.
      • Vogelmeier C.
      • Welte T.
      • et al.
      The German COPD cohort COSYCONET: aims, methods and descriptive analysis of the study population at baseline.
      ]. The study was approved by the respective ethical committees, and all patients gave their written informed consent. In all visits, patients were required to be in a stable clinical condition without exacerbation within the preceding four weeks. The present analysis used data from the baseline visit (V1) of patients of GOLD grades 1–4, who had complete data of forced expiratory volume in 1 s (FEV1), functional residual capacity in terms of intrathoracic gas volume (ITGV), diffusing capacity for carbon monoxide (DLCO), packyears of smoking, body-mass index (BMI), partial pressure of oxygen (PaO2), partial pressure of carbon dioxide (PaCO2), oxygen saturation (SO2), haemoglobin content (Hb), and creatinine serum level.

      2.2 Pulmonary function tests, blood gas analysis and kidney function

      According to the COSYCONET protocol [
      • Karch A.
      • Vogelmeier C.
      • Welte T.
      • et al.
      The German COPD cohort COSYCONET: aims, methods and descriptive analysis of the study population at baseline.
      ] and following recommendations [
      • Miller M.R.
      • Hankinson J.
      • Brusasco V.
      • et al.
      Standardisation of spirometry.
      ,
      • Criee C.P.
      • Berdel D.
      • Heise D.
      • et al.
      [Recommendations on spirometry by deutsche atemwegsliga].
      ,
      • Criee C.P.
      • Sorichter S.
      • Smith H.J.
      • et al.
      Body plethysmography--its principles and clinical use.
      ,
      • Wanger J.
      • Clausen J.L.
      • Coates A.
      • et al.
      Standardisation of the measurement of lung volumes.
      ,
      • Macintyre N.
      • Crapo R.O.
      • Viegi G.
      • et al.
      Standardisation of the single-breath determination of carbon monoxide uptake in the lung.
      ], pulmonary function tests (spirometry, body plethysmography, CO diffusing capacity) were performed 45 min after inhalation of 400 μg salbutamol and 80 μg ipratropium bromide. GOLD groups ABCD were formed on the basis of mMRC (modified Medical Research Council) dyspnea scale [
      • Mahler D.A.
      • Wells C.K.
      Evaluation of clinical methods for rating dyspnea.
      ] and exacerbation history [
      • Vogelmeier C.F.
      • Criner G.J.
      • Martinez F.J.
      • et al.
      Global strategy for the diagnosis, management, and prevention of chronic obstructive lung disease 2017 report: GOLD executive summary.
      ]. Patients showing ≥2 exacerbations without hospital admission or ≥1 exacerbation leading to hospital admission within the last year were categorized as groups C or D, depending on symptoms; the combined groups C and D defined the binary exacerbation risk variable versus groups A and B of low exacerbation risk.
      The values of PaO2, PaCO2, pH and SaO2 were determined from arterialized capillary blood from the earlobe. Blood gas analyzers were those available in the study centers (supplement S1). Adequate calibration and quality control were ensured by standardised operating procedures. Base excess (BE) values were provided by the integrated algorithms and were retrospectively checked by the van Slyke equation. Bicarbonate (HCO3) was derived via the Henderson-Hasselbalch equation [
      • Siggaard-Andersen O.
      • Fogh-Andersen N.
      Base excess or buffer base (strong ion difference) as measure of a non-respiratory acid-base disturbance.
      ,
      • Siggaard-Andersen O.
      • Gothgen I.H.
      • Wimberley P.D.
      • et al.
      The oxygen status of the arterial blood revised: relevant oxygen parameters for monitoring the arterial oxygen availability.
      ], and the oxygen content (CaO2) via the conventional formula: CaO2 = (1.34 * Hb * SaO2) + (0.0031 * PaO2) [
      • Duke J.W.
      • Davis J.T.
      • Ryan B.J.
      • et al.
      Decreased arterial PO2, not O2 content, increases blood flow through intrapulmonary arteriovenous anastomoses at rest.
      ].
      Patients with an eGFR <60 mL/min/1.73 m2 at study inclusion and at the 6-month visit were considered as having CKD, as per the Kidney Disease Outcome Quality Initiative (KDOQI) guidelines [
      • Inker L.A.
      • Astor B.C.
      • Fox C.H.
      • et al.
      KDOQI US commentary on the 2012 KDIGO clinical practice guideline for the evaluation and management of CKD.
      ].

      2.3 Comorbidities and medication

      Kidney function was quantified using the estimated glomerular filtration rate (eGFR), based on the creatinine equation from the Chronic Kidney Disease Epidemiology Collaboration (CKD-EPI) [
      • Levey A.S.
      • Stevens L.A.
      • Schmid C.H.
      • et al.
      A new equation to estimate glomerular filtration rate.
      ]. All other comorbidities and medication were assessed by structured interviews, whereby additional information on the presence of comorbidities was obtained by the evaluation of medication [
      • Karch A.
      • Vogelmeier C.
      • Welte T.
      • et al.
      The German COPD cohort COSYCONET: aims, methods and descriptive analysis of the study population at baseline.
      ,
      • Lucke T.
      • Herrera R.
      • Wacker M.
      • et al.
      Systematic analysis of self-reported comorbidities in large cohort studies - a novel stepwise approach by evaluation of medication.
      ].

      2.4 Statistical analysis

      Data in the tables are presented as numbers and percentages, or mean values and standard deviations (SD). Comparisons between COPD groups were performed by analysis of variance (ANOVA), or by chi-square tests in case of categorical variables. In case of continuous outcome variables, associations were determined by multiple linear regression analysis, for the exacerbation variable by binary logistic regression analysis. In addition, exploratory factor analysis was used to analyse the relationship between lung function parameters. To account for the mutual relationships between variables and to differentiate between direct and indirect effects, we employed structural equation modelling (SEM) as repeatedly used for the analysis of COPD comorbidities [
      • Kahnert K.
      • Lucke T.
      • Huber R.M.
      • et al.
      Relationship of hyperlipidemia to comorbidities and lung function in COPD: results of the COSYCONET cohort.
      ,
      • Alter P.
      • Watz H.
      • Kahnert K.
      • et al.
      Airway obstruction and lung hyperinflation in COPD are linked to an impaired left ventricular diastolic filling.
      ,
      • Kahnert K.
      • Alter P.
      • Welte T.
      • et al.
      Uric acid, lung function, physical capacity and exacerbation frequency in patients with COPD: a multi-dimensional approach.
      ]. Using pathophysiological knowledge and the results of the regression analyses, we constructed an SEM based on data adjusted for age, gender, BMI, pack years of smoking and the current smoking status excluding medication. In a second step medication was added in terms of diuretics, oral (OCS) and inhaled (ICS) corticosteroids. Besides determining the effects of medication, we checked by this approach, whether the structure of the model was robust against the inclusion of medication. The SEM analyses were performed using the software package AMOS (Version 24, IBM Corp., Armonk, NY, USA) with generalized least squares estimation (GLS), quantifying the goodness of fit by chi-square statistics, comparative fit index (CFI) and root mean square error of approximation (RMSEA), taking CFI ≥0.95 and RMSEA ≤0.05 as indicators of a good fit. The approach primarily aims at excluding models not adequately fitting the data, thus the chi-square statistics describing the deviation from the model is therefore acceptable if p ≥ 0.05, i.e. there is no deviation. All other analyses were performed with SPSS version 24 (IBM Corp., Armonk, NY, USA). P-values < 0.05 were considered as statistically significant.

      3. Results

      3.1 Study subjects, lung function and medication

      Among the 2741 patients of COSYCONET, 1506 were eligible for this analysis; 450 patients did not fit into the GOLD grades 1-4, and among the remaining patients 785 did not have complete or valid data for the variables analysed in the present study. Patients’ characteristics regarding anthropometric data, lung function and medication are given in Table 1. The numbers across spirometric GOLD grades 1-4 were 145/677/562/122, those of ABCD groups 618/357/202/329. The majority of patients (63%) were male; there were no significant differences regarding ABCD categories between males and females. Mean (±SD) BMI was 26.7 ± 5.2 kg/m2, age 64.5 ± 8.1 y, FEV1 54.1 ± 18.3 %predicted, FEV1/FVC 51.7 ± 10.7%, and ITGV 149.2 ± 34.2 %predicted. All parameters except gender significantly differed between groups ABCD (Table 1) and all parameters except gender and packyears between grades 1-4 (Table S2) (ANOVA). Medication potentially affecting acid-base balance was frequent, as 20.5% of patients used diuretics, 10.3% oral, and 64.8% inhaled corticosteroids. Nephroprotective medication with blockade of the renin-angiotensin system with either angiotensin-converting enzyme inhibitors (ACEi) or angiotensin II receptor blockers (ARBs) was administered in 27.2% und 17.9% respectively.
      Table 1Anthropometric data, lung function and medication with GOLD A-D (mMRC) categorization.
      ParameterAllGOLD AGOLD BGOLD CGOLD Dp
      N1506618357202329
      Gender (m/f)955/551396/222236/121129/73194/1350.256
      Age (y)64.5 ± 8.164.5 ± 8.065.9 ± 8.062.9 ± 9.164.1 ± 8.0p ≤ 0.001
      BMI (kg/m2)26.7 ± 5.226.3 ± 4.527.7 ± 5.625.9 ± 4.527.0 ± .6.0p ≤ 0.001
      Packyears49.2 ± 36.648.0 ± 35.153.0 ± 37.944.0 ± 32.950.2 ± 39.60.033
      FEV1%pred54.1 ± 18.362.5 ± 17.849.0 ± 16.154.2 ± 16.743.7 ± 14.8p ≤ 0.001
      FEV1/FVC51.7 ± 10.754.9 ± 9.849.7 ± 10.850.6 ± 10.648.6 ± 10.7p ≤ 0.001
      ITGV %pred149.2 ± 34.2140.1 ± 29.8152.9 ± 36.3151.7 ± 32.2160.6 ± 36.5p ≤ 0.001
      DLCO %pred56.3 ± 21.263.4 ± 20.652.0 ± 19.859.0 ± 20.546.2 ± 18.9p ≤ 0.001
      Diuretics308/20.5%76/12.3%89/24.9%34/16.8%109/33.1%p ≤ 0.001
      OCS155/10.3%27/4.435/9.8%20/9.9%73/22.2%p ≤ 0.001
      ICS976/64.8%314/50.8%252/70.8%146/72.3%264/80.2%p ≤ 0.001
      ACEi (N/%)*409/27.2%137/22.3%109/30.5%61/30.2%102/31.10.004
      ARBs (N/%)*270/17.9%108/17.5%68/19.0%30/14.9%64/19.5%0.523
      Abbreviations: BMI body mass index; FEV1 forced expiratory volume in 1 s; ITGV intrathoracic gas volume; DLCO diffusing capacity for carbon monoxide; OCS oral corticosteroids; ICS inhaled corticosteroids, ACEi angiotensin-converting enzyme inhibitor; ARBs angiotensin II receptor blockers **N = 1505. Values are presented as mean ± SD or number (%).The table shows mean values, standard deviations or absolute numbers. The comparisons of age, BMI, packyears and lung function parameters between GOLD A-D categories were performed by analysis of variance (ANOVA); the comparison of gender and medication between GOLD A-D categories were performed by chi-squared statistics.

      3.2 Blood gas analyses and kidney function

      Patients showed mean (±SD) pH of 7.431 ± 0.027, PaCO2 of 37.8 ± 4.6 mmHg, PaO2 of 66.4 ± 8.2 mmHg, CaO2 of 18.9 ± 1.8 (mL/100 mL), and haemoglobin of 14.7 ± 1.3 mg/dL. Long-term oxygen therapy was prescribed in 134 (8.9%) of patients; this treatment was not considered in the analysis as blood gases had to be assessed without oxygen flow. The values of acid-base balance are given in Table 2. Regarding groups ABCD, all parameters were significantly different. The mean (±SD) eGFR was 82.3 ± 16.9 ml/min/1,73 m2. 6.9% of patients had chronic kidney disease per the Kidney Disease Outcome Quality Initiative (KDOQI) guidelines [
      • Inker L.A.
      • Astor B.C.
      • Fox C.H.
      • et al.
      KDOQI US commentary on the 2012 KDIGO clinical practice guideline for the evaluation and management of CKD.
      ].
      Table 2Blood gas analysis and kidney function measurements with GOLD A-D (mMRC) categorization.
      ParameterAllGOLD AGOLD BGOLD CGOLD Dp
      N1506618357202329
      pH7.43 ± 0.037.43 ± 0.037.43 ± 0.037.43 ± 0.037.43 ± 0.030.642
      PaCO2 (mmHg)37.8 ± 4.637.2 ± 4.138.0 ± 4.837.6 ± 4.638.8 ± 4.9p ≤ 0.001
      PaO2 (mmHg)66.4 ± 8.267.9 ± 7.964.6 ± 7.367.6 ± 8.764.8 ± 8.5p ≤ 0.001
      HCO3-(mmol/L)24.3 ± 2.723.9 ± 2.524.4 ± 2.824.2 ± 2.624.8 ± 2.8p ≤ 0.001
      SO2 (%)93.9 ± 2.594.4 ± 2.193.5 ± 2.694.1 ± 2.693.3 ± 2.7p ≤ 0.001
      Hb (mg/dL)14.7 ± 1.314.8 ± 1.314.7 ± 1.414.6 ± 1.314.5 ± 1.40.009
      BE (mmol/L)1.1 ± 2.30.8 ± 2.21.2 ± 2.41.1 ± 2.31.6 ± 2.5p ≤ 0.001
      CaO2 (ml/100 mL)18.9 ± 1.819.2 ± 1.718.9 ± 1.818.9 ± 1.618.6 ± 1.8p ≤ 0.001
      eGFR (ml/min/1,73 m2)82.3 ± 16.983.4 ± 15.280.7 ± 17.083.8 ± 16.381.2 ± 17.30.024
      CKD (N/%)*85/6.9%29/%5.625/8.8%7/4.0%24/9.6%0.046
      Abbreviations: PaCO2 partial pressure of carbon dioxide; PaO2 partial pressure of oxygen; HCO3- bicarbonate, SO2 oxygen saturation; Hb haemoglobin; BE base excess; CaO2 oxygen content; eGFR estimated glomerular filtration rate; CKD Chronic Kidney Disease. CKD categories were defined in accordance with the National Kidney Foundation–Kidney Disease Outcomes Quality Initiative (KDOQI) guideline Abbreviations, *N = 1144. The table shows mean values, standard deviations or absolute numbers and percentage. The comparisons were performed by analysis of variance (ANOVA) the comparison of CKD between GOLD A-D was performed by chi-squared statistics.

      3.3 Relationship between variables

      To identify the relationships between variables, we performed linear and logistic regression analyses, always adjusting for age, gender, BMI, smoking status and pack years. First, each of the single variables was tested on its association with exacerbation risk via logistic regression; the results are given in terms of odds ratios in Fig. 1. In particular, exacerbation risk was strongly associated with BE and lung function; an elevation of BE by 2 mmol/L or a reduction of FEV1 %predicted by 5% was linked to a marked increase in exacerbation risk, despite the comparatively small changes. Significant associations with exacerbation risk were also found for CaO2, PaO2, PaCO2, DLCO and ITGV and a borderline significance for eGFR, but no significance for pH.
      Fig. 1
      Fig. 1The figure shows the odds ratios for the outcome variable exacerbation risk, as derived from logistic regression analyses for each single depicted variable as predictor, while adjusting for age, gender, BMI, smoking status and packyears. In the bars, the odds ratios are given for changes by the amounts indicated for each of the variables (please note the sign that was chosen in order to indicate increases in risk). Additionally, the original odds ratios, their 95% confidence intervals and the corresponding p values are shown as numbers. The results of a multiple logistic regression analysis comprising all predictors were similar with regard to the magnitude and relationship of estimates but due to collinearity statistically less reliable. The complex relationship between variables is delineated in detail in Fig. 2, Fig. 3.
      To determine the association between the variables influencing exacerbation risk, multiple linear and logistic regression analyses were employed. BE turned out to be related to FEV1, pH, PaCO2 and GFR (p < 0.001 each). PaCO2 was associated with BE, pH, CaO2 (p < 0.001 each), eGFR (p = 0.006) and FEV1 (p = 0.027). CaO2 was linked to PaCO2, eGFR, exacerbation risk (p ≤ 0.002 each), as well as DLCO (p = 0.044). pH was related to PaCO2 and BE (p < 0.001 each), and to FEV1 (p = 0.031). Exacerbation risk was associated with FEV1 and CaO2 (p < 0.001 each), as well as DLCO and eGFR (p ≤ 0.029 each). Conversely, eGFR was linked to exacerbation risk (p = 0.022) and PaCO2 and CaO2 (p < 0.001 each) but not to BE; if PaCO2 was omitted as predictor, there was a significant association also with BE (p < 0.001). This pointed toward an intricate relationship between these measures.
      The lung function parameters FEV1, ITGV and DLCO were highly correlated with each other, thereby posing the problem of collinearity. To clarify the issue, we performed an exploratory factor analysis comprising all variables. The dominant factor (eigenvalue 2.51) was constituted by the three lung function measures, whereas the other variables were distributed over three other factors with eigenvalues between 1.00 and 1.29. Based on this, we decided to introduce the latent variable (construct) “lung function impairment” comprising the three lung function measures mentioned above into the SEM modelling.

      3.4 Results of structural equation modeling

      The results indicated a convoluted relationship between the variables analysed. To describe this as precisely and accurately as possible we constructed a structural equation model (SEM) by stepwise combination of the results of the regression analyses, keeping the model as parsimonious as possible. Lung function was included as a latent variable with FEV1, ITGV and DLCO as indicators, however we allowed for additional relationships of these indicators to other variables if suggested by the regression analyses. Kidney function in terms of the estimated glomerular filtration rate (eGFR) was considered as variable not affected by others, and exacerbation risk as final outcome variable affected by others but not affecting them itself. The resulting SEM is shown in Fig. 2. The model without medication (Fig. 2) fitted the data with a chi-square value of 23.,3 and 20 degrees of freedom (p = 0.273) and was also well-fitting according to a bootstrap procedure using 2000 samples and the Bollen-Stine method (p = 0.290). The CFI was 0.998 and the RMSEA was 0.010 (90%CI 0.000; 0.025). The corresponding regression coefficients and their p values are given in Table 3.
      Fig. 2
      Fig. 2Results of Structural Equation Modelling (SEM)
      Rectangles contain observed variables, the oval a latent variable, which is a hypothetical combination (construct) of the indicator variables connected with it. The SEM comprises FEV1, intrathoracic gas volume (ITGV), diffusing capacity for carbon monoxide (DLCO) as indicator variables, their combination “lung function impairment” as latent variable, glomerular filtration rate (eGFR), oxygen content (CaO2), partial pressure of carbon dioxide (PaCO2), pH value, and exacerbation risk according to GOLD criteria. Only relationships which turned out to be (simultaneously) statistically significant (p < 0.05) are shown, and each arrow symbolizes an association equivalent to a linear regression, whereby the variable at the arrowhead is treated as dependent. SEMs are a combination between multiple regression and factor analysis approaches, allowing for the simultaneous assessment of direct and indirect relationships. The error terms of dependent variables that are needed for statistical reasons are omitted for the sake of clarity.
      Table 3Numerical results of the final Structural Equation Model (Fig. 2).
      RegressionEstimateS.E.C.R.StandardizedP
      BEeGFR0,0200,0044,5730,117≤0.001
      Lung function impairmentBE−2,2940,191−12,009−0,305≤0.001
      DLCO %predLung function impairment0,6560,03717,7340,567≤0.001
      PaCO2eGFR0,0180,0062,8500,0550,004
      PaCO2BE1,1640,04029,2780,594≤0.001
      PaCO2Lung function impairment−0,0460,006−8,030−0,177≤0.001
      CaO2PaCO2−0,0690,009−7,503−0,195≤0.001
      CaO2eGFR0,0110,0033,8620,098≤0.001
      CaO2DLCO %pred0,0050,0022,5820,0670,010
      FEV1 %predLung function impairment1,0000,972
      ITGV %predLung function impairment−0,9920,057−17,323−0,547≤0.001
      pHPaCO2−0,0060,000−64,365−1,080≤0.001
      pHBE0,0120,00060,3431,013≤0.001
      Exacerbation riskLung function impairment−0,0070,001−9,215−0,256≤0.001
      Exacerbation riskeGFR−0,0020,001−2,130−0,0530,033
      Exacerbation riskCaO2−0,0240,007−3,235−0,0810,001
      Abbreviations: BE base excess; DLCO diffusing capacity for carbon monoxide; PaCO2 partial pressure of carbon dioxide; CaO2 oxygen content; ITGV intrathoracic gas volume; FEV1 forced expiratory volume in 1 s; eGFR estimated glomerular filtration rate. “Lung function impairment” denotes a latent variable (construct), with FEV1, ITGV and DLCO as indicator variables. Exacerbation risk was defined via the combined GOLD groups C and D as in GOLD criteria. The table refers to the directed arrows (regression terms) as shown in Fig. 2, whereby the left part lists the arrows shown in this Figure. The right part shows the results of the statistical estimation. The first column of the right part shows the non-standardized estimate of the respective regression coefficient, the second column the standard error (S.E.) of this coefficient, the third column the ratio of these two values (critical ratio, C.R.) which is used for significance testing. The forth column shows the standardized estimates of the regression coefficients shown in the first column, which is useful to compare the relative magnitude of relationships. The last column shows the significance level based on the generalized least squares (GLS) procedure of AMOS.
      In a second step, we analysed the potential effect of medication with diuretics, oral and inhaled corticosteroids, to reveal whether the SEM was robust against their influence. The correlations between the administration of OCS and ICS, and between OCS and diuretics, were incorporated via corresponding correlation coefficients. The resulting model is shown in Fig. 3. This model fitted the data with a chi-square value of 75.94 and 39 degrees of freedom (p < 0.001); the bootstrap procedure using 2000 samples and the Bollen-Stine method yielded a similar result. The CFI was 0.981, and the RMSEA 0.021 (90%CI 0.0.017; 0.033). All regression coefficients identified as significant in the SEM without medication (Table 3) remained significant and virtually unchanged, indicating the robustness of the SEM against the confounding influence of medication. The lower degree of fit was probably due to the introduction of binary variables into a model adapted for continuous variables.
      Fig. 3
      Fig. 3Results of Structural Equation Modelling including medication with potential impact on acid base metabolism
      Symbols are identical to those of , additionally medication is included as influencing factor in terms of binary variables (present/not present). For this, diuretics, oral and inhaled corticosteroids were chosen as potentially relevant. The arrows from the medication boxes illustrate directed relationship as explained for . Importantly, the structure identified between the measured variables as shown in remained robust when medication was included. OCS had no direct effect on eGFR, and ICS had direct effects only on “lung function impairment” and exacerbation risk. To describe the data as closely as possible, the observed correlations between the administration of OCS and ICS, as well as between OCS and diuretics, were included in the model; they are omitted for the sake of clarity, as they do not provide additional information.

      4. Discussion

      The present study addresses the role of acid-base balance in stable COPD, particularly its association with exacerbation risk and lung function, including kidney function via the estimated glomerular filtration rate. The results of regression analyses were implemented into a more intricate model allowing for the description of direct and indirect effects on exacerbation risk (according to GOLD) and lung function. BE was not only linked to PaCO2 and pH, but also directly to lung function impairment. Overall, the changes seemed compatible with (over-)compensated respiratory acidosis. The indirect effects of BE on exacerbation risk were mediated via lung function impairment and CaO2. eGFR was directly linked to exacerbation risk, as well as indirectly via BE, PaCO2 and CaO2. The results did not critically depend on the inclusion of potentially relevant medication in terms of diuretics, OCS and ICS. The major finding was that even in patients with stable COPD and largely normal values of acid-base indices, acid-base status had an impact on exacerbation risk. The effects including those of kidney function were largely mediated via BE which probably provides an integrative long-term marker of compensatory demands from the respiratory disorder and comorbidities in stable COPD.
      Previous work on acid-base balance as outcome predictor often dealt with severe COPD, e.g. patients with noninvasive home ventilation [
      • Budweiser S.
      • Jorres R.A.
      • Heinemann F.
      • et al.
      [Prognostic factors for COPD patients with chronic hypercapnic respiratory failure and home ventilation].
      ,
      • Budweiser S.
      • Jorres R.A.
      • Riedl T.
      • et al.
      Predictors of survival in COPD patients with chronic hypercapnic respiratory failure receiving noninvasive home ventilation.
      ,
      • Budweiser S.
      • Hitzl A.P.
      • Jorres R.A.
      • et al.
      Impact of noninvasive home ventilation on long-term survival in chronic hypercapnic COPD: a prospective observational study.
      ]. Our study population comprised all COPD grades, the majority of patients being in grades 2 and 3. Patients who would have been categorized as the former GOLD grade 0 (at risk) [
      • Pauwels R.A.
      • Buist A.S.
      • Calverley P.M.
      • et al.
      Global strategy for the diagnosis, management, and prevention of chronic obstructive pulmonary disease. NHLBI/WHO Global Initiative for Chronic Obstructive Lung Disease (GOLD) Workshop summary.
      ] were omitted from the analysis. Our findings obtained in stable COPD were consistent with previous results regarding the role of acid-base balance in acute exacerbations [
      • Ucgun I.
      • Oztuna F.
      • Dagli C.E.
      • et al.
      Relationship of metabolic alkalosis, azotemia and morbidity in patients with chronic obstructive pulmonary disease and hypercapnia.
      ,
      • Schiavo A.
      • Renis M.
      • Polverino M.
      • et al.
      Acid-base balance, serum electrolytes and need for non-invasive ventilation in patients with hypercapnic acute exacerbation of chronic obstructive pulmonary disease admitted to an internal medicine ward.
      ,
      • Terzano C.
      • Di Stefano F.
      • Conti V.
      • et al.
      Mixed acid-base disorders, hydroelectrolyte imbalance and lactate production in hypercapnic respiratory failure: the role of noninvasive ventilation.
      ]. Even though in the majority of patients acid-base balance indices were close to normal, except for base excess (BE), we found significant associations with exacerbation risk. Exacerbation risk was also directly linked to kidney function in terms of eGFR, in addition to lung function. Another important determinant was the oxygen content of the arterial blood (CaO2) as derived from PaO2, hemoglobin and oxygen saturation. Oxygen content was linked to PaCO2 as well as DLCO, which seems very plausible. PaCO2 itself was closely associated with BE. BE was additionally linked to lung function, while dependent on eGFR. Taken together, the combined effect of eGFR and CaO2 on exacerbation risk was about half of that arising from lung function. This can be seen in Table 3 by multiplying and summing up the standardized coefficients of the respective pathways. It suggests that even among patients with largely normal or near-to-normal values of both kidney function and acid-base balance, these still contribute to exacerbation risk.
      The key comorbidity, which we addressed in our study, was an impairment in kidney function. This, however, was minor, and eGFR values were mostly above the cutoff value of 60 mL/min/1.73 m2 that is commonly used for the definition of chronic kidney disease [
      • Inker L.A.
      • Astor B.C.
      • Fox C.H.
      • et al.
      KDOQI US commentary on the 2012 KDIGO clinical practice guideline for the evaluation and management of CKD.
      ]. A study by Polverino et al. delineated cigarette smoke-induced systemic endothelial injury as a pathological mechanism for both, the development of emphysema and renal injury. The authors suggested that albuminuria could identify patients with COPD in whom angiotensin-converting enzyme inhibitor therapy might improve renal and lung function by reducing endothelial injury [
      • Polverino F.
      • Laucho-Contreras M.E.
      • Petersen H.
      • et al.
      A pilot study linking endothelial injury in lungs and kidneys in chronic obstructive pulmonary disease.
      ].
      COPD patients with renal failure may suffer from lower compensatory efficacy in case of respiratory acidosis, with impairments in ammoniagenesis, production of titratable acidity and reduction of serum bicarbonate, leading to further deterioration of acidosis [
      • Bruno C.M.
      • Valenti M.
      Acid-base disorders in patients with chronic obstructive pulmonary disease: a pathophysiological review.
      ,
      • Hamm L.L.
      • Nakhoul N.
      • Hering-Smith K.S.
      Acid-base homeostasis.
      ]. It is therefore conceivable that kidney function may have direct effects in terms of enhancing exacerbations. The major biomarker describing the result of regulatory efficiency is BE, and previous studies have demonstrated the close relationship between BE and kidney function in patients with and without COPD [
      • Ucgun I.
      • Oztuna F.
      • Dagli C.E.
      • et al.
      Relationship of metabolic alkalosis, azotemia and morbidity in patients with chronic obstructive pulmonary disease and hypercapnia.
      ].
      Chronic kidney disease in COPD is correlated with other, especially cardiovascular, comorbidities [
      • Fedeli U.
      • De Giorgi A.
      • Gennaro N.
      • et al.
      Lung and kidney: a dangerous liaison? A population-based cohort study in COPD patients in Italy.
      ], while a broad panel of comorbidities has been shown to be associated with exacerbation risk [
      • Kahnert K.
      • Alter P.
      • Young D.
      • et al.
      The revised GOLD 2017 COPD categorization in relation to comorbidities.
      ]. Metabolic state and compensatory capacity including kidney function are to a significant extent reflected in BE. In our study population, 46.4% of patients showed BE values above 1 mmol/L, which is usually considered as indicating metabolic alkalosis [
      • Magder S.
      • Emami A.
      Practical approach to physical-chemical acid-base management. Stewart at the bedside.
      ]. Hypokalemia, hypochloremia and aldosterone excess are known to maintain metabolic alkalosis [
      • Khanna A.
      • Kurtzman N.A.
      Metabolic alkalosis.
      ], and major comorbidities relevant for this are gastrointestinal diseases, diabetes, arterial hypertension, heart failure and further cardiac comorbidities requiring the administration of diuretics, in addition to the mineralocorticoid effect of oral steroids [
      • Luke R.G.
      • Galla J.H.
      It is chloride depletion alkalosis, not contraction alkalosis.
      ,
      • Yasuda K.
      • Hayashi M.
      • Murayama M.
      • et al.
      Acidosis-induced hypochloremic alkalosis in diabetic ketoacidosis confirmed by the modified base excess method.
      ,
      • Galla J.H.
      Metabolic alkalosis.
      ]. A common finding is compensatory metabolic alkalosis due to alveolar hypoventilation in COPD, but also due to comorbid conditions such as obstructive sleep apnea [
      • Verbraecken J.
      • McNicholas W.T.
      Respiratory mechanics and ventilatory control in overlap syndrome and obesity hypoventilation.
      ].
      In the present analysis we did not include other comorbidities than kidney dysfunction, and we considered BE as a biomarker integrating over the acid-base effects of various comorbidities. When tentatively including coronary heart disease, diabetes and hyperlipidemia into the model of Fig. 2, none of these was related to the variables of the model, consistent with the assumption that their effect was covered by BE. This observation probably also reflected the fact that all variables were adjusted for major common risk factors.
      BE was linked to PaCO2 as a consequence of the equations governing acid-base balance via bicarbonate; the latter was therefore omitted as redundant. Moreover, initial analyses had revealed that BE was superior to bicarbonate regarding its statistical explanatory power. pH only appeared as dependent variable without direct effects on the other variables. This low information content corresponded to the finding that pH values were very close to normal, in accordance with the fact that we investigated patients with stable COPD.
      We aimed at a parsimonious model rendering interpretation easy and fitting current pathophysiological knowledge. Following this aim, oxygen saturation, PaO2 and hemoglobin content were combined into oxygen content, CaO2, which turned out to be superior to the explicit introduction of the single variables. It was also found to be directly dependent on DLCO, underlining the adequacy of our model and estimates. In addition to this specific link, DLCO could be summarized together with FEV1 and ITGV into the construct “lung function impairment”, which had the strongest influence on exacerbation risk. In comparison, the influence of CaO2 was only about one third as strong (see standardized estimates in Table 3), while the direct effect of kidney function was about one fifth of the combined effect of lung function.
      PaCO2 was mostly within the normal range, and there was only a slight tendency towards increased values with increasing COPD grade (Table 2). In patients with stable COPD and OSA, an association between nocturnal hypoventilation, i.e. elevated nocturnal PaCO2, and exacerbation risk has been found, and treatment with continuous positive airway pressure was associated with improved survival and decreased hospitalization rate [
      • Marin J.M.
      • Soriano J.B.
      • Carrizo S.J.
      • et al.
      Outcomes in patients with chronic obstructive pulmonary disease and obstructive sleep apnea: the overlap syndrome.
      ]. One might speculate that the metabolic overcompensation as indicated by BE was a result of nocturnal impairments in PaCO2. Such overcompensation might, however, have a price, as alkalemia is associated with myopathy, cardiac arrhythmias, increased airway resistance and compensatory hypoventilation [
      • Galla J.H.
      Metabolic alkalosis.
      ,
      • Brijker F.
      • van den Elshout F.J.
      • Heijdra Y.F.
      • et al.
      Effect of acute metabolic acid/base shifts on the human airway calibre.
      ] which per se negatively affect exacerbation risk. In accordance with this, a reduction of BE by noninvasive home ventilation in severe COPD was observed as an independent predictor of improved survival [
      • Budweiser S.
      • Jorres R.A.
      • Riedl T.
      • et al.
      Predictors of survival in COPD patients with chronic hypercapnic respiratory failure receiving noninvasive home ventilation.
      ].

      5. Limitations

      The present analysis has the obvious limitations of cross-sectional analyses, and proper causal inferences can not be drawn. The SEM has directed arrows but these do not necessarily indicate a causative direction, similarly to ordinary regression analysis, in which also dependent and independent variables may be interchanged. Despite this, based on the complexity of relationships within the model, there can be strong hints towards causality, if the single arrow cannot be inverted without losing statistical significance. There was no statistically significant link from the exacerbation risk to eGFR. It should also be emphasized that we deal with the exacerbation risk but not with acute exacerbations, where it is especially reasonable to expect the reverse effect. In contrast to a potential first impression, our model also does not suggest that eGFR explains everything, as the other, intermediate variables are all measured and of course only partially explained by the direct and indirect effects arising from eGFR. The direction of the arrow indicates that under stable conditions there is an additional effect of eGFR on exacerbation risk (i.e. frequency and/or severity) that is not mediated via the other variables in the model, particularly not those of the acid-base balance which should be closely linked to kidney dysfunction. One might speculate that these unknown pathways are inflammatory pathways but when we tentatively included, e.g., the serum levels of IL-8, TNF or CRP into the analyses, using properly adjusted values, there were no statistically siginificant links that could explain the direct arrow.
      Microalbuminuria as a possible link of early kidney injury, lung function decline and exacerbations was not available in our cohort. Moreover, serum creatinine and thus the estimated glomerular filtration rate has known limitations as a biomarker of renal function, and the great majority of values was normal or close to normal. Despite this, we observed meaningful associations with eGFR. The presence of LTOT did not play a role for the relationship between variables, and blood gas analysis had to be performed without oxygen flow. Moreover, a sensitivity analysis introducing an additional adjustment for the presence of LTOT did not alter the SEM.

      6. Conclusions

      In patients with stable COPD, acid-base metabolism had a major impact on the COPD outcome measure exacerbation risk, similar as known for acute COPD exacerbations, and in addition to lung function. In accordance with the known link between acid-base balance and kidney function, we found even slight functional impairments to be linked to exacerbation risk. The key integrative role was played by base excess, probably as a long-term marker of compensatory status and demands arising from the respiratory disease and comorbidities.

      Conflicts of interest

      Dr. Trudzinski received personal fees from Novartis and Berlin-Chemie.
      Dr. Vogelmeier reports personal fees from Almirall, grants and personal fees from AstraZeneca, grants and personal fees from Boehringer Ingelheim, grants and personal fees from Chiesi, grants and personal fees from GlaxoSmithKline, grants and personal fees from Grifols, grants and personal fees from Mundipharma, grants and personal fees from Novartis, grants and personal fees from Takeda, personal fees from Cipla, personal fees from Berlin Chemie/Menarini, personal fees from CSL Behring, personal fees from Teva, grants from German Federal Ministry of Education and Research ( BMBF ) Competence Network Asthma and COPD (ASCONET), grants from Bayer Schering Pharma AG, grants from MSD, grants from Pfizer, outside the submitted work;.
      Dr. Alter reports grants from German Federal Ministry of Education and Research ( BMBF ) Competence Network Asthma and COPD (ASCONET), grants from AstraZeneca GmbH, grants and non-financial support from Bayer Schering Pharma AG, grants, personal fees and non-financial support from Boehringer Ingelheim Pharma GmbH & Co. KG, grants and non-financial support from Chiesi GmbH, grants from GlaxoSmithKline, grants from Grifols Deutschland GmbH, grants from MSD Sharp & Dohme GmbH, grants and personal fees from Mundipharma GmbH, grants, personal fees and non-financial support from Novartis Deutschland GmbH, grants from Pfizer Pharma GmbH, grants from Takeda Pharma Vertrieb GmbH & Co. KG, outside the submitted work.Dr. Alter reports grants from German Federal Ministry of Education and Research ( BMBF ) Competence Network Asthma and COPD (ASCONET), grants from AstraZeneca GmbH, grants and non-financial support from Bayer Schering Pharma AG, grants, personal fees and non-financial support from Boehringer Ingelheim Pharma GmbH & Co. KG, grants and non-financial support from Chiesi GmbH, grants from GlaxoSmithKline, grants from Grifols Deutschland GmbH, grants from MSD Sharp & Dohme GmbH, grants and personal fees from Mundipharma GmbH, grants, personal fees and non-financial support from Novartis Deutschland GmbH, grants from Pfizer Pharma GmbH, grants from Takeda Pharma Vertrieb GmbH & Co. KG, outside the submitted work.
      Dr. Seiler received personal fees from Fisher & Paykel, Getinge, and Novartis.
      Dr. Watz reports personal fees from AstraZeneca, personal fees from Boehringer Ingelheim, personal fees from GlaxoSmithKline, personal fees from BerlinChemie, personal fees from Chiesi, personal fees from Novartis, personal fees from Roche, outside the submitted work;.
      Dr. Welte reports grants from Astra Zeneca, Bayer, Boehringer, Berlin-Chemie, Chiesi, Grifols, GSK, MSD, Mundipharma, Novartis, Pfizer, Takeda, Teva, grants from Ministry for Research and Education, during the conduct of the study; personal fees from Astra Zeneca, Bayer, Boehringer, Berlin-Chemie, Chiesi, Grifols, GSK, Novartis, outside the submitted work;.
      Dr. Kauczor reports grants, personal fees and non-financial support from Siemens, personal fees from Boehringer Ingelheim, non-financial support from Bayer, personal fees from Bracco, grants and personal fees from Philips, outside the submitted work;.
      Dr. Bals reports grants and personal fees from AstraZeneca, grants and personal fees from Boehringer Ingelheim, personal fees from GlaxoSmithKline, personal fees from Grifols, grants and personal fees from Novartis, personal fees from CSL Behring, grants from German Federal Ministry of Education and Research ( BMBF ) Competence Network Asthma and COPD (ASCONET), grants from Sander Stiftung, grants from Schwiete Stiftung, grants from Krebshilfe, grants from Mukoviszidose eV, outside the submitted work;.
      Dr. Kahnert, Dr. Biertz, Dr. Jörres, Dr. Fähndrich, Dr. Speer and Dr. Zewinger have nothing to disclose.

      Financial support

      This work was supported by the German Federal Ministry of Education andResearch (BMBF) Competence Network Asthma and COPD (ASCONET) and performed in collaboration with the German Centre for Lung Research(DZL) . The project is funded by the BMBF with grant number 01 GI 0881 , and is funded by unrestricted grants from AstraZeneca GmbH, BayerSchering Pharma AG, Boehringer Ingelheim Pharma GmbH & Co. KG, ChiesiGmbH, GlaxoSmithKline, Grifols Deutschland GmbH, MSD Sharp & DohmeGmbH, Mundipharma GmbH, Novartis Deutschland GmbH, Pfizer PharmaGmbH, Takeda Pharma Vertrieb GmbH & Co. KG, Teva GmbH for patient investigations and laboratory measurements.

      Authors’ contributions

      FCT, RB, RAJ and KK contributed to design of the study, to data analysis and interpretation, and drafted the manuscript. CV, TW, HW, SF, RAJ, H-UK and RB contributed to data collection and interpretation and revised the manuscript critically for intellectual content. FB, PA, FS, TS, SZ contributed to data interpretation and revised the manuscript critically for intellectual content. All authors approved the final version of the manuscript.

      Acknowledgments

      We are grateful to the COSYCONET study group and study centers who contributed in patient recruitment and data collection, as well as to all patients participating in this study.

      Appendix A. Supplementary data

      The following are the Supplementary data to this article:

      References

        • Inker L.A.
        • Astor B.C.
        • Fox C.H.
        • et al.
        KDOQI US commentary on the 2012 KDIGO clinical practice guideline for the evaluation and management of CKD.
        Am. J. Kidney Dis. 2014; 63: 713-735
        • Saran R.
        • Robinson B.
        • Abbott K.C.
        • et al.
        US renal data system 2016 annual data report: Epidemiology of kidney disease in the United States.
        Am. J. Kidney Dis. 2017; 69: A7-A8
        • Vogelmeier C.F.
        • Criner G.J.
        • Martinez F.J.
        • et al.
        Global strategy for the diagnosis, management, and prevention of chronic obstructive lung disease 2017 report: GOLD executive summary.
        Eur. Respir. J. 2017; 49
        • Haroun M.K.
        • Jaar B.G.
        • Hoffman S.C.
        • et al.
        Risk factors for chronic kidney disease: a prospective study of 23,534 men and women in Washington County, Maryland.
        J. Am. Soc. Nephrol. 2003; 14: 2934-2941
        • Fox C.S.
        • Larson M.G.
        • Leip E.P.
        • et al.
        Predictors of new-onset kidney disease in a community-based population.
        JAMA, J. Am. Med. Assoc. 2004; 291: 844-850
      1. WHO, Chronic obstructive pulmonary disease. Secondary WHO, Chronic obstructive pulmonary disease.

        • Baty F.
        • Putora P.M.
        • Isenring B.
        • et al.
        Comorbidities and burden of COPD: a population based case-control study.
        PLoS One. 2013; 8e63285
        • Gjerde B.
        • Bakke P.S.
        • Ueland T.
        • et al.
        The prevalence of undiagnosed renal failure in a cohort of COPD patients in western Norway.
        Respir. Med. 2012; 106: 361-366
        • Yoshizawa T.
        • Okada K.
        • Furuichi S.
        • et al.
        Prevalence of chronic kidney diseases in patients with chronic obstructive pulmonary disease: assessment based on glomerular filtration rate estimated from creatinine and cystatin C levels.
        Int. J. Chronic Obstr. Pulm. Dis. 2015; 10: 1283-1289
        • Wedzicha J.A.
        • Singh R.
        • Mackay A.J.
        Acute COPD exacerbations.
        Clin. Chest Med. 2014; 35: 157-163
        • Seiler F.
        • Trudzinski F.C.
        • Kredel M.
        • et al.
        [Update: acute hypercapnic respiratory failure].
        Med Klin Intensivmed Notfmed. 2017; 114: 234-239
        • Ucgun I.
        • Oztuna F.
        • Dagli C.E.
        • et al.
        Relationship of metabolic alkalosis, azotemia and morbidity in patients with chronic obstructive pulmonary disease and hypercapnia.
        Respiration. 2008; 76: 270-274
        • Hopkins E.
        • Sharma S.
        Physiology, Acid Base Balance.
        StatPearls. Treasure Island (FL), 2018
        • Kraut J.A.
        • Kurtz I.
        Metabolic acidosis of CKD: diagnosis, clinical characteristics, and treatment.
        Am. J. Kidney Dis. 2005; 45: 978-993
        • Hodgkin J.E.
        • Soeprono F.F.
        • Chan D.M.
        Incidence of metabolic alkalemia in hospitalized patients.
        Crit. Care Med. 1980; 8: 725-728
        • Babitt J.L.
        • Lin H.Y.
        Mechanisms of anemia in CKD.
        J. Am. Soc. Nephrol. 2012; 23: 1631-1634
        • Luke R.G.
        • Galla J.H.
        It is chloride depletion alkalosis, not contraction alkalosis.
        J. Am. Soc. Nephrol. 2012; 23: 204-207
        • Westerik J.A.
        • Metting E.I.
        • van Boven J.F.
        • et al.
        Associations between chronic comorbidity and exacerbation risk in primary care patients with COPD.
        Respir. Res. 2017; 18: 31
        • Kahnert K.
        • Alter P.
        • Young D.
        • et al.
        The revised GOLD 2017 COPD categorization in relation to comorbidities.
        Respir. Med. 2018; 134: 79-85
        • Karch A.
        • Vogelmeier C.
        • Welte T.
        • et al.
        The German COPD cohort COSYCONET: aims, methods and descriptive analysis of the study population at baseline.
        Respir. Med. 2016; 114: 27-37
        • Miller M.R.
        • Hankinson J.
        • Brusasco V.
        • et al.
        Standardisation of spirometry.
        Eur. Respir. J. 2005; 26: 319-338
        • Criee C.P.
        • Berdel D.
        • Heise D.
        • et al.
        [Recommendations on spirometry by deutsche atemwegsliga].
        Pneumologie. 2006; 60: 576-584
        • Criee C.P.
        • Sorichter S.
        • Smith H.J.
        • et al.
        Body plethysmography--its principles and clinical use.
        Respir. Med. 2011; 105: 959-971
        • Wanger J.
        • Clausen J.L.
        • Coates A.
        • et al.
        Standardisation of the measurement of lung volumes.
        Eur. Respir. J. 2005; 26: 511-522
        • Macintyre N.
        • Crapo R.O.
        • Viegi G.
        • et al.
        Standardisation of the single-breath determination of carbon monoxide uptake in the lung.
        Eur. Respir. J. 2005; 26: 720-735
        • Mahler D.A.
        • Wells C.K.
        Evaluation of clinical methods for rating dyspnea.
        Chest. 1988; 93: 580-586
        • Siggaard-Andersen O.
        • Fogh-Andersen N.
        Base excess or buffer base (strong ion difference) as measure of a non-respiratory acid-base disturbance.
        Acta Anaesthesiol. Scand. Suppl. 1995; 107: 123-128
        • Siggaard-Andersen O.
        • Gothgen I.H.
        • Wimberley P.D.
        • et al.
        The oxygen status of the arterial blood revised: relevant oxygen parameters for monitoring the arterial oxygen availability.
        Scand. J. Clin. Lab. Invest. Suppl. 1990; 203: 17-28
        • Duke J.W.
        • Davis J.T.
        • Ryan B.J.
        • et al.
        Decreased arterial PO2, not O2 content, increases blood flow through intrapulmonary arteriovenous anastomoses at rest.
        J. Physiol. 2016; 594: 4981-4996
        • Levey A.S.
        • Stevens L.A.
        • Schmid C.H.
        • et al.
        A new equation to estimate glomerular filtration rate.
        Ann. Intern. Med. 2009; 150: 604-612
        • Lucke T.
        • Herrera R.
        • Wacker M.
        • et al.
        Systematic analysis of self-reported comorbidities in large cohort studies - a novel stepwise approach by evaluation of medication.
        PLoS One. 2016; 11e0163408
        • Kahnert K.
        • Lucke T.
        • Huber R.M.
        • et al.
        Relationship of hyperlipidemia to comorbidities and lung function in COPD: results of the COSYCONET cohort.
        PLoS One. 2017; 12e0177501
        • Alter P.
        • Watz H.
        • Kahnert K.
        • et al.
        Airway obstruction and lung hyperinflation in COPD are linked to an impaired left ventricular diastolic filling.
        Respir. Med. 2018; 137: 14-22
        • Kahnert K.
        • Alter P.
        • Welte T.
        • et al.
        Uric acid, lung function, physical capacity and exacerbation frequency in patients with COPD: a multi-dimensional approach.
        Respir. Res. 2018; 19: 110
        • Budweiser S.
        • Jorres R.A.
        • Heinemann F.
        • et al.
        [Prognostic factors for COPD patients with chronic hypercapnic respiratory failure and home ventilation].
        Pneumologie. 2009; 63: 484-491
        • Budweiser S.
        • Jorres R.A.
        • Riedl T.
        • et al.
        Predictors of survival in COPD patients with chronic hypercapnic respiratory failure receiving noninvasive home ventilation.
        Chest. 2007; 131: 1650-1658
        • Budweiser S.
        • Hitzl A.P.
        • Jorres R.A.
        • et al.
        Impact of noninvasive home ventilation on long-term survival in chronic hypercapnic COPD: a prospective observational study.
        Int. J. Clin. Pract. 2007; 61: 1516-1522
        • Pauwels R.A.
        • Buist A.S.
        • Calverley P.M.
        • et al.
        Global strategy for the diagnosis, management, and prevention of chronic obstructive pulmonary disease. NHLBI/WHO Global Initiative for Chronic Obstructive Lung Disease (GOLD) Workshop summary.
        Am. J. Respir. Crit. Care Med. 2001; 163: 1256-1276
        • Schiavo A.
        • Renis M.
        • Polverino M.
        • et al.
        Acid-base balance, serum electrolytes and need for non-invasive ventilation in patients with hypercapnic acute exacerbation of chronic obstructive pulmonary disease admitted to an internal medicine ward.
        Multidiscip Respir Med. 2016; 11: 23
        • Terzano C.
        • Di Stefano F.
        • Conti V.
        • et al.
        Mixed acid-base disorders, hydroelectrolyte imbalance and lactate production in hypercapnic respiratory failure: the role of noninvasive ventilation.
        PLoS One. 2012; 7e35245
        • Polverino F.
        • Laucho-Contreras M.E.
        • Petersen H.
        • et al.
        A pilot study linking endothelial injury in lungs and kidneys in chronic obstructive pulmonary disease.
        Am. J. Respir. Crit. Care Med. 2017; 195: 1464-1476
        • Bruno C.M.
        • Valenti M.
        Acid-base disorders in patients with chronic obstructive pulmonary disease: a pathophysiological review.
        J. Biomed. Biotechnol. 2012; 2012: 915150
        • Hamm L.L.
        • Nakhoul N.
        • Hering-Smith K.S.
        Acid-base homeostasis.
        Clin. J. Am. Soc. Nephrol. 2015; 10: 2232-2242
        • Fedeli U.
        • De Giorgi A.
        • Gennaro N.
        • et al.
        Lung and kidney: a dangerous liaison? A population-based cohort study in COPD patients in Italy.
        Int. J. Chronic Obstr. Pulm. Dis. 2017; 12: 443-450
        • Kahnert K.
        • Alter P.
        • Young D.
        • et al.
        The revised GOLD 2017 COPD categorization in relation to comorbidities.
        Respir. Med. 2018; 134: 79-85
        • Magder S.
        • Emami A.
        Practical approach to physical-chemical acid-base management. Stewart at the bedside.
        Ann Am Thorac Soc. 2015; 12: 111-117
        • Khanna A.
        • Kurtzman N.A.
        Metabolic alkalosis.
        J. Nephrol. 2006; 19: S86-S96
        • Yasuda K.
        • Hayashi M.
        • Murayama M.
        • et al.
        Acidosis-induced hypochloremic alkalosis in diabetic ketoacidosis confirmed by the modified base excess method.
        J. Clin. Endocrinol. Metab. 2016; 101: 2390-2395
        • Galla J.H.
        Metabolic alkalosis.
        J. Am. Soc. Nephrol. 2000; 11: 369-375
        • Verbraecken J.
        • McNicholas W.T.
        Respiratory mechanics and ventilatory control in overlap syndrome and obesity hypoventilation.
        Respir. Res. 2013; 14: 132
        • Marin J.M.
        • Soriano J.B.
        • Carrizo S.J.
        • et al.
        Outcomes in patients with chronic obstructive pulmonary disease and obstructive sleep apnea: the overlap syndrome.
        Am. J. Respir. Crit. Care Med. 2010; 182: 325-331
        • Brijker F.
        • van den Elshout F.J.
        • Heijdra Y.F.
        • et al.
        Effect of acute metabolic acid/base shifts on the human airway calibre.
        Respir. Physiol. 2001; 124: 151-158