Cardiac biomarkers of prognostic importance in chronic obstructive pulmonary disease

Chronic Obstructive Pulmonary Disease (COPD) is a common but under-diagnosed disease [1, 2]. In COPD, there is an increased risk for concomitant cardiovascular disease [3] and cardiovascular comorbidity is associated with increased mortality [4]. Still, patients with COPD may be less likely to receive cardiovascular preventative therapies [5], albeit most guidelines for diagnosis and treatment of COPD [6] include recommendations regarding evaluation of cardiovascular risk factors and the presence of cardiovascular disease. However, no biomarker has been shown to identify subclinical cardiovascular disease in a representative population-based cohort of individuals with COPD, even though ischemic abnormalities on electrocardiogram (ECG) are suggested to be associated with worse prognosis [7].

Cardiac troponins are specific markers of myocardial injury and are used universally for the diagnosis of myocardial infarction [8]. High-sensitivity cardiac troponin I (hs-cTnI) assays have limits of detection 10 to 100-fold lower than contemporary assays and can detect troponin in the majority of healthy individuals [9]. Hence, these assays may identify patients with subclinical cardiac disease [10], using risk stratification thresholds well below those used to diagnose myocardial infarction [11, 12]. Recently published data suggest that a risk stratification threshold of < 5 ng/L can identify patients with chest pain at low risk of having cardiac events [11] and, further, is able stratify patients with stable COPD and concomitant cardiovascular risk factors into low and high-risk groups [13]. Increased troponin levels are also associated with poor prognosis in patients with acute exacerbations of COPD [14, 15]. One recent study, including stable patients with mainly moderate and severe COPD, also showed a correlation between hs-cTnI ≥6 ng/L and poor prognosis [16]. However, this has rarely been evaluated in population-based COPD-studies. Based on these observations, we hypothesise that hs-cTnI concentrations reflect subclinical heart disease and may help to guide prognosis among individuals with COPD in the population. If this hypothesis is confirmed, hs-cTnI testing could be used more widely to identify individuals with COPD at increased risk.

The aim of this study was to evaluate hs-cTnI concentrations among individuals with COPD and individuals with normal lung function to determine the prognostic value of hs-cTnI. A secondary aim was to include ischemic ECG abnormalities in the risk assessment.

The baseline characteristics of individuals with COPD (n = 601) and those with normal lung function (n = 755) are presented in Table 1. Among those with COPD, 41.6, 51.1 and 7.3% were classified as GOLD grades 1, 2 and 3–4 respectively. The prevalence of reported ischemic heart disease was higher among individuals with COPD (17.1%) compared to individuals with normal lung function (12.5%).

Troponin concentrations were above the limit of detection in 580 (96.5%) individuals with COPD and 708 (93.8%) individuals with normal lung function and were above the upper reference limit in 20 individuals with COPD and 24 individuals with normal lung function (3.3 and 3.2%, respectively). The median hs-cTnI concentration and proportion of individuals with hs-cTnI > 5 ng/L were higher in COPD than in normal lung function (Table 1).

In both individuals with COPD and those with normal lung function, hs-cTnI concentrations > 5 ng/L were associated with higher age, male sex, and higher proportions of diabetes mellitus and ischemic heart disease (see Supplementary Table 1, Additional File 1). Analyses in an adjusted model confirmed that age and sex remained significantly associated with hs-cTnI ≥5 ng/L in both COPD and NLF, together with ischemic heart disease in the COPD group and ischemic ECG abnormalities in the NLF group (see Supplementary Table 2, Additional File 1).

The prevalence of ischemic ECG abnormalities was similar among individuals with COPD and those with normal lung function, 14.8 and 13.4% respectively (Table 1). The proportion with ischemic ECG abnormalities was higher among those with hs-cTnI > 5 ng/L than among those with hs-cTnI < 5 ng/L in both groups (24.6% vs. 10.4, and 26.6% vs. 9.0% respectively; p < 0.001 for both, see Supplementary Table 1, Additional File 1). Among individuals with COPD, 23.5% had hs-cTnI > 5 ng/L alone, 7.2% had ischemic ECG abnormalities alone and 7.7% had both hs-cTnI > 5 ng/L and ischemic ECG abnormalities, while the corresponding prevalence among those with normal lung function was 18.3, 6.8 and 6.6% respectively (Fig. 1 a-b).

In this study, a history of ischemic heart disease, myocardial injury and myocardial ischemia was common and more frequent among individuals with COPD than among individuals with normal lung function. As expected, mortality was higher among individuals with COPD than among those with normal lung function, and high-sensitivity cardiac troponin was associated with increased mortality in both groups. An elevated hs-cTnI concentration with or without signs of myocardial ischemia on ECG increased the risk for death three- to fourfold among individuals with COPD, independent of age, sex, smoking habits, diabetes mellitus and disease severity. Importantly, these cardiac biomarkers were associated with an increased risk for death of similar magnitude also among individuals with COPD without known ischemic heart disease.

Even though individuals with COPD comprise a high-risk population, ischemic heart disease is often-overlooked, but the converse is also true, as COPD is often undiagnosed in patients with ischemic heart disease [23]. Most current guidelines for the diagnosis and treatment of COPD highlight the need for greater awareness of cardiovascular risk and comorbidities among patients with COPD, in order to initiate preventive measures and optimise outcomes. However, there are no specific recommendations for the approach to risk assessment of ischemic heart disease among individuals with COPD.

In a large-scale multi-center study including > 16,000 patients with COPD and cardiovascular disease or risk factors for cardiovascular disease, acute exacerbations of COPD (AECOPD) were associated with an increased risk for new cardiovascular events, including cardiovascular death [24]. Cardiac troponin has over time been proven to be an independent risk factor for all-cause mortality during acute exacerbation of COPD [14]. Troponin, as well as NT-proBNP, a marker for chronic heart failure, have also been associated with an increased risk for death among patients hospitalized for AECOPD without previously known ischemic heart disease [25]. During recent years, there are a few reports based on selected patient populations with stable COPD, indicating that troponin is increased [26] and associated with mortality [16, 27] among COPD patients without known heart disease. In recent years, the development and clinical use of high-sensitivity cardiac troponin assays have permitted the detection of myocardial injury more widely [28]. In a recent study of 88 patients hospitalized for AECOPD with elevated hs-cTnI, coronary angiography detected ischemic heart disease in two-thirds, and half of these patients underwent percutaneous coronary intervention [28], supporting the assumption that, at least in this setting, elevated hs-cTnI is associated with subclinical ischemic heart disease.

The diagnostic threshold for myocardial infarction using our high-sensitivity cTnI assay is 34 ng/L in men and 16 ng/L in women [13]. However, hs-cTnI is increasingly being used to evaluate cardiovascular risk at concentrations well below the diagnostic threshold. The optimal threshold for risk stratification continues to be debated, but the largest study performed to date, demonstrated that individuals in the Emergency Department with hs-cTnI concentrations < 5 ng/L are at very low risk of both short and long-term cardiac events [11]. Moreover, also in stable COPD patients participating in a randomized controlled trial, the same threshold identified patients at low or high risk of cardiovascular death [13]. However, cardiovascular disease or risk factors were inclusion criteria in this study, and the study population was limited to patients with moderate COPD (FEV₁ ≥ 50 and ≤ 70% of predicted) having dyspnea corresponding to mMRC ≥2 and a smoking history of ≥10 pack-years. Still to date, this approach to risk stratification has not been evaluated in a less selected group of individuals with COPD and, the generalizability of this approach to population-based COPD cohorts has not been known.

Restricting our analyses to individuals with COPD without reported ischemic heart disease supported our hypothesis; the increased risk of death associated with an elevated hs-cTnI concentration alone or in combination with ischemic ECG abnormalities persisted. Our observations imply that these cardiac biomarkers may identify individuals with unrecognized cardiac disease of prognostic importance.

Both ischemic ECG abnormalities and troponin are independently known to be risk markers for mortality, both among individuals with COPD [7, 16, 29] and in the general population [30, 31]. However, the impact of the combination of these markers has rarely been investigated. This study shows a difference in the risk profile between individuals with COPD and NLF. Therefore, we hypothesize that I-ECG represents a past event, that increases long-term risk, while circulating hs-cTnI indicates an ongoing subclinical myocardial disease that, in combination with a chronic airway obstruction, negatively affects the survival at a five-year perspective. Yet, the mechanisms behind the impact of hs-cTnI in COPD are still mainly unknown [32]. Future research is required to understand the relationship between these biomarkers and coronary or structural heart disease among individuals with COPD. Henceforth, this may contribute to the development of an algorithm to identify individuals with COPD in the general population at increased risk in whom specific treatment and preventive measures for cardiovascular disease could be introduced. Whilst the numbers of deceased in each of the biomarker groups in the current study were limited, still around 70% of the deaths among those with COPD were cardiovascular, compared with 53% among those without any biomarkers. Yet, death certificates are to a great extent based on clinical diagnosis and, thus, potential subclinical cardiovascular disease, assumed to associate with elevated biomarkers under study, will therefore remain undetected [33].

The strength of the current study is the large population-based COPD cohort identified by post-bronchodilator spirometry according to the GOLD guidelines [6], and with a distribution of disease severity dominated by mild to moderate COPD, as in other population-based studies [34, 35]. The cohort is thus considered representative for COPD in the population and under-diagnosis of COPD [2, 6] is not expected to affect the results. Further strengths are the use of well-validated methods; a structured interview following a validated questionnaire [36] Minnesota coding of ECG [18] and a well validated high-sensitivity cTnI assay [12].

However, there are some limitations that merit discussion. The prevalence of ischemic heart disease and diabetes mellitus was based on interview data and not on medical records, and we do not have data on other known risk factors, such as blood pressure and total cholesterol. Still, there is known to be fairly good agreement between self-reported data and medical records on the comorbid conditions evaluated in the current study, cardiovascular disease and diabetes mellitus [37‐39]. Whilst information on hypertension would also be of importance, self-reported data on hypertension have a fairly low validity [40] and were thus not used. The blood samples were collected during epidemiological fieldwork and stored at -20 °C, but we do not anticipate that troponin concentrations will have changed during storage and have previously reported that hs-cTnI levels predicted long-term cardiovascular events from samples stored for 20 years [10]. Impaired renal function may affect troponin levels, but data on renal function were not available in this study. Yet, in a large population-based study, there were no differences in unadjusted comparisons and, also in adjusted analyses, those with mild airway obstruction had similar mean glomerular filtration ratio as those with normal lung function (89.1 vs. 89.6 ml/min/1.73 m², p = 0.619), while those with moderate/severe/very severe airway obstruction had slightly lower mean value (87.6 ml/min/1.73 m², p = 0.015) [41]. Our study included mainly mild to moderate COPD and we believe that adjustment for renal function would hardly affect the results. The OLIN COPD study was designed shortly after the shift of the millennium, when the fixed ratio criterion was generally accepted to define airway obstruction in COPD. It is recognized that the fixed ratio may overestimate COPD among elderly non-smoker [42], and the lower limit of normal (LLN) criterion for COPD is nowadays recommended in epidemiological studies. Still, most clinical guidelines use the fixed ratio criterion for COPD [6], thus the results are highly clinically relevant and applicable in every-day care for COPD.

In this study, elevated hs-cTnI concentrations in combination with signs of myocardial ischemia on the electrocardiogram were associated with a more than fourfold increased risk of death in a population-based COPD-cohort, independent of disease severity. This increased risk was of similar magnitude also among those with COPD, but without known ischemic heart disease. Thus, these cardiac biomarkers may indicate the presence of undiagnosed and prognostically important cardiovascular disease among individuals with COPD and should be further evaluated for screening and risk modeling at a population level.