Heart failure (HF) remains a major healthcare burden, and remains the major cause of readmission to hospital globally.1 The European Society of Cardiology conventionally classifies HF based on left ventricular ejection fraction (LVEF) measurements, whereby those with an LVEF ≤40% are described as having HF with reduced ejection fraction (HFrEF), an LVEF 41–49% as having HR with mildly reduced ejection fraction and an LVEF ≥50% as having HF with preserved ejection fraction (HFpEF).1 It is now felt that HFpEF contributes to half of all cases of HF worldwide.2 Despite lower rates of mortality, patients with HFpEF have equal rates of hospitalisation to HFrEF patients.2,3 What used to be of concern was that effective treatment for managing HFpEF remains scarce, unlike HFrEF, where there is evidence-based, guideline-directed medical therapy in existence.
Fortunately, there has been an exponential growth in research and development of therapeutic options to manage HFpEF over the past decade, with the advent of pharmacological treatments, such as sodium–glucose cotransporter 2 inhibitors, non-steroidal mineralocorticoid antagonist and glucagon-like peptide-1 receptor agonists, which have demonstrated improved clinical outcomes in patients living with HFpEF.2–4
However, a major challenge remains in the early detection and screening for the condition, as the majority of cases remain underdiagnosed owing to the lack of awareness among clinicians regarding the condition.2–4 Furthermore, dyspnoea, cough and reduced effort tolerance – common clinical presentations in HFpEF – are often associated with a myriad of other clinical conditions, some of which can exist concurrently with HFpEF.2,4–6 One such condition is chronic obstructive pulmonary disease (COPD). As both COPD and HFpEF share common risk factors, including smoking, coronary artery disease, diabetes and hypertension, it is often difficult to distinguish and appreciate their co-existence.5–7 Nevertheless, it remains important that HFpEF is considered as a differential diagnosis when evaluating patients with COPD because the treatment options vary greatly and HFpEF, if diagnosed early, is a very manageable condition.
We conducted a pilot study to identify the prevalence of HFpEF based on the Heart Failure Association Pre-test assessment, Echocardiography and natriuretic peptide, Functional testing, Final aetiology (HFA-PEFF) score among patients living with COPD, which has never been truly delineated prior to this.6 Our pilot study also aims to explore the various predictive factors for HFpEF within this cohort of patients.
Methods
A prospective, cross-sectional pilot study was performed in a single tertiary centre in Malaysia. The pilot study was conducted between 1 June 2024 and 28 February 2025. Patients recruited in the study included those aged ≥18 years with a confirmed diagnosis of COPD made via spirometry. Each patient’s FEV1 and forced vital capacity values through formal lung spirometry at the time of initial diagnosis was obtained. Classification was based on the Global Initiative for Chronic Obstructive Lung Disease ABE assessment tool.8
Patients with a known diagnosis of HFrEF and HR with mildly reduced ejection fraction, and with a recent history (i.e. 6 months) of admission for any acute illness that would preclude them from performing lung spirometry and diastolic exercise stress test (DEST) if indicated, were excluded. Others who were excluded were patients who were pregnant or 6 months postpartum, with active malignancy and/or those with life expectancy felt to be <12 months. This study was designed as a pilot study and adopted a flat rule of thumb for two-group studies in determining its sample size.9 A sample size of 30 participants per arm (i.e. high probability of having HFpEF versus those with low to intermediate probabilities of having HFpEF based on their HFA-PEFF score) was recommended to ensure sufficient data for preliminary analysis and feasibility assessment.
Patients were recruited via universal sampling, and ethical approval was obtained from the local Research and Ethics Committee. Information surrounding sociodemographic background, comorbidities and smoking history was obtained. Additionally, baseline anthropometric measurements, Roche Diagnostics N-terminal pro-brain natriuretic peptide (NT-proBNP) assay levels at the time of study entry, and both the modified Medical Research Council scores and COPD Assessment Test questionnaire scores at the time of study entry were gathered.
All patients eligible for the pilot study underwent comprehensive 2D transthoracic echocardiography (TTE) testing, which included standard measurements of parameters recommended by the American Society of Echocardiography and European Association of Cardiovascular Imaging, alongside additional assessment for global longitudinal strain (GLS) imaging of the left ventricle, as well as left atrial strain (LAS) imaging.10,11 All TTEs were performed by a single, experienced operator using a Philips EPIC CVx ultrasound machine and Philips X5–1 xMATRIX transducer.
Patients with LVEF <50% on TTE were further excluded from analysis. Then, following their TTE and baseline assessment, patients were categorised into three groups based on their HFA-PEFF score to determine the likelihood of HFpEF.2,6 Patients scoring 0–1 point were classified as having a low probability of having HFpEF, whereas patients with 5–6 points were classified as having a high probability of having HFpEF. Patients with a score of 2–4 points were classified as having an intermediate probability and were all offered further investigations in the form of a DEST, as recommended by guidelines.2,6,10
The DEST was performed using the same TTE setup, but with the patients being positioned on a GE Healthcare eBike EL ergometer 240V alongside continuous ECG, oxygen saturation and blood pressure monitoring. DEST was performed using a standardised protocol where patients were asked to cycle in a semi-supine position under a workload of 25 W throughout a 3-minute interval before a further increase in workload by increments of 25 W every 3 minutes. The DEST was terminated upon successful demonstration of impaired diastolic function with exercise, including a peak early diastolic mitral inflow velocity to average peak early diastolic mitral annulus tissue velocity ratio (E/e´)>14 and/or peak tricuspid regurgitation velocity of >2.8 m/s. Other reasons for cessation include intolerable symptoms developing during the test or haemodynamic instability, similar to any other conventional exercise stress test.
Following a positive DEST, patients with an intermediate probability were reclassified as having a high probability of having HFpEF, whereas those with a negative test were reclassified as not having the condition. For patients with intermediate probability of HFpEF who were unable to attempt or complete their DEST, the patients were grouped together with those having a low probability of HFpEF for the purpose of the study, as pragmatically such patients are not subjected to therapeutic options dedicated to treating the condition.
Categorical responses are presented as frequencies (percentages), and numerical responses as the mean (SD) or median (interquartile range; IQR). All variables were assessed for normality. The prevalence of HFpEF within the patient cohort was assessed using a frequency test. Statistical analysis was performed using SPSS version 29.0 (IBM) to compare between respondents who had a high probability of having HFpEF based on their HFA-PEFF score versus those with low to intermediate probabilities of having HFpEF. The χ2 test and Mann–Whitney U-test were performed for categorical variables, and both the paired t-test and analysis of variance for continuous variables. All tests were two-sided, with a significance level of p-value <0.05. Univariable logistic regression analysis was used to assess the association between various variables and risk of being diagnosed with a high probability of developing HFpEF. Unfortunately, multivariable regression analysis was not performed due to sample size limitations.
Results
A total of 62 patients were screened throughout the study duration, of which 21 patients were excluded due to having an LVEF of <50% on TTE (Table 1 ). The remaining 41 patients underwent further review and stratification based on TTE and baseline investigations, through calculation of their HFA-PEFF score. The mean age of the participants was 70.5 years (± 7.9), and the majority were men (90.2%) and of Malay ethnicity (90.2%). A large number of patients were either smokers (29.3%) or ex-smokers (61.50%), with an average pack-years of 31 years (± 23.7). Common comorbidities within this cohort include hypertension (75.6%) and dyslipidaemia (63.4%). When comparing between respondents diagnosed with a high probability of HFpEF versus those with low to intermediate probabilities, only diabetes showed a statistically significant difference (66.7 versus 20.0%; p=0.017). The average COPD Assessment Test score was 18.3 points (±8.3), and the majority of patients had a modified Medical Research Council score of 2 (36.6%) and 3 points (31.7%).
Six patients (14.6%) were found to have scores suggestive of a high probability of HFpEF, based on their HFA-PEFF score. The remaining 26 (63.4%) and 12 patients (29.3%) had intermediate and low HFA-PEFF probability scores, respectively. All 26 patients with a HFA-PEFF score of 2–4 were counselled for DEST. However, only three patients underwent additional testing with DEST, all of whom had unremarkable results and were later reclassified as having a low probability of having HFpEF. The remaining 23 patients were unable to pursue DEST for various reasons, including severe knee osteoarthritis (6 patients), significant clinical frailty and presumed sarcopenia (13 patients), and declining further testing through DEST over fears of safety (4 patients).
TTE parameters were also obtained for the overall study population (Table 2 ). The median LVEF was 57.0% (IQR 54.0–60.0) and median left atrial volume index (LAVI) was 27.6 ml/m2 (IQR 18.5–34.5). Of the echocardiographic parameters obtained, LAVI (37.4 versus 26.0 ml/m2; p<0.001), peak early diastolic lateral mitral annulus tissue velocity (lateral e´; 4.7 ml/m2 versus 8.6 cm/s; p<0.001), E/e´ (20.0 versus 9.0; p=0.014), GLS (−15.3 versus −19.5%; p<0.001), left atrial reservoir strain (LAS reservoir; 14.5 versus 36.0%; p<0.001), left atrial pump strain (LAS pump; −6.0 versus −18.0%; p<0.001) and left atrial conduit strain (LAS conduit; −4.3 versus −24.0%; p<0.001) were statistically significant. The median baseline NT-proBNP level in the cohort was 113.0 pg/nl (IQR 47.0–250.1), and was significantly different between both cohorts (1,648.5 versus 101.0 pg/nl; p<0.001).
Following univariate logistic regression analysis, it was shown that presence of diabetes in our COPD cohort was significantly associated with high probability of developing HFpEF (OR 8.0, 1.2–52.2; p=0.03) (Table 3). Other comorbidities, including AF (OR 0.26; 95% CI [0.035–1.89]; p=0.182), chronic kidney disease (OR 0.25; 95% CI [0.041–1.52]; p=0.132), obesity (OR 2.61; 95% CI [0.27–24.94]; p=0.405) and ischaemic heart disease (OR 0.59; 95% CI [0.09–3.85]; p=0.584), were not shown to be significant. Univariate regression analysis also revealed that symptom burden, measured through conventional COPD assessment tools, including FEV1 (OR 1.02; 95% CI [0.97–1.07]; p=0.548), COPD Assessment Test (OR 1.03; 95% CI [0.93–1.15]; p=0.549) and modified Medical Research Council scores (OR 1.02; 95% CI [0.96–1.07]; p=0.283), respectively, was not predictive of a high probability of developing HFpEF.
Unsurprisingly, the majority of the conventional echocardiographic parameters assessed, including low lateral e´ (OR 0.39; 95% CI [0.19–0.79]; p=0.009), high E/e´ (OR 1.49; 95% CI [1.08–2.06]; p=0.015) and high LAVI (OR 1.45; 95% CI [1.04–2.03]; p=0.027), were highly associated with a high probability of developing HFpEF in our COPD cohort (Table 2). However, peak early diastolic septal mitral annulus tissue velocity (OR 0.59; 95% CI [0.33–1.07]; p=0.082), pulmonary artery systolic pressure (OR 1.06; 95% CI [0.99–1.15]; p=0.075) and peak tricuspid regurgitation velocity (OR 4.67; 95% CI [0.77–28.46]; p=0.095) were shown to be non-significant. Interestingly, novel strain imaging parameters, including low GLS (OR 0.581; 95% CI [0.37–0.91]; p=0.018), low LAS reservoir (OR 0.621; 95% CI [0.39–0.92]; p=0.044), low LAS pump (OR 0.36; 95% CI [0.14–0.93]; p=0.034) and low LAS conduit (OR 0.58; 95% CI [0.36–0.92]; p=0.022), were shown to be predictive of having a high probability of a HFpEF diagnosis in our patient cohort. Similarly, raised NT-proBNP levels was also shown to be significantly associated with having a high probability of HFpEF (OR 1.01, 1.00–1.01; p=0.035).
Discussion
Our pilot study demonstrates a significant prevalence of patients with a high probability of developing HFpEF, diagnosed within a cohort of COPD patients (14.6%). This is comparable to existing literature, where the estimated prevalence of HFpEF in COPD ranged between 14% and 30%.12 However, of note was the substantial number of patients who fell within the intermediate probability group, which was more than half of the pilot study’s cohort. As per guidelines, such patients require further testing, normally in the form of a non-invasive DEST or invasive haemodynamics, with or without exercise – the latter of which is neither pragmatic nor accessible in most clinical settings.2,6
Unfortunately, through this pilot study, we were able to highlight how non-invasive testing through DEST posed its own challenges, especially in a population with significant symptom burden and poor exercise tolerance, such as those afflicted by COPD. Of the 26 patients with intermediate HFA-PEFF scores, only three patients could undergo additional testing, in the form of DEST. Interestingly, 13 of those patients in the intermediate probability group were felt to have significant clinical frailty and presumed sarcopenia (although not formally proven during the study), which are commonly associated with HFpEF, raising concerns over missed diagnosis of HFpEF among those with intermediate probability for the disease. This calls into question the practicality of existing guidelines in addressing patients within the intermediate probability group and the need to consider alternative means of diagnosing HFpEF, especially for future, larger prospective studies.
Despite promising data in the early years leading to its current widespread use in diagnosing HFpEF, conventional echocardiographic parameters used at rest in the assessment of diastolic dysfunction, such as peak early diastolic mitral inflow velocity (E) and E/e´ ratio, have been shown to correlate poorly with progression into HFpEF in a number of studies.13–16 To circumvent this, incorporating exercise into the testing of diastolic dysfunction has since been recommended, which has been shown to improve diagnostic yield.14,16,17 However, as mentioned before, exercise intolerance is a common clinical manifestation among patients living with COPD, which limits the utility of DEST in this cohort and possibly contributes to grave underdiagnosis of the condition. This is further compounded by the diagnostic accuracy of certain diastolic dysfunction parameters in the presence of COPD.
As demonstrated in our study, peak early diastolic septal mitral annulus tissue velocity was not particularly useful in distinguishing between COPD patients with HFpEF versus those without. In respiratory conditions that significantly affect right heart pressure, such as COPD and pulmonary hypertension, e´ values – specifically that of the septal wall – may be reduced irrespective of known cardiac dysfunction disease, which likely explains the non-significant results shown.18,19 Furthermore, this is corroborated by our study’s findings surrounding peak tricuspid regurgitation velocity and pulmonary artery systolic pressure. In our pilot study, it is shown that both parameters were not predictive of a high probability of HFpEF, possibly because respiratory conditions, such as COPD, can impair right ventricular haemodynamics, regardless of concomitant cardiac dysfunction and, therefore, may not be ideal parameters to use to screen for HFpEF in such patient groups.
An intriguing finding from this pilot study was surrounding the role of strain imaging in predicting the presence of a high probability of HFpEF within our COPD cohort. Strain imaging uses speckled tracking methods to assess the degree of deformation of the myocardium during contraction and relaxation.11,17,18 The novelty in strain imaging is in its ability to detect early, subclinical myocardial disease or damage following insult to the left ventricle (through GLS) or to the left atrium (via LAS reservoir, LAS pump or LAS conduit).11,17,18 At present, only GLS is recognised as a possible marker of HFpEF, where abnormal results have been incorporated as a minor criterion within the HFA-PEFF scoring algorithm.6
Through this pilot study, we were able to demonstrate that that abnormalities seen in these strain-based parameters likely indicate intrinsic left-sided myocardial disease, which was unlikely related or linked to COPD, a predominantly respiratory-centric disease with right-sided myocardial sequelae. Furthermore, these strain-based parameters appear equally reliable in predicting a high probability of HFpEF, in contrast to conventional diastolic dysfunction parameters.11,17,18,20,21 Understandably, clinical and subclinical injury to the left atrium, which will manifest as abnormal LAS values, would likely appear to supersede changes in left atrium volume (i.e. LAVI), which would only become abnormal following prolonged exposure to raised filling pressures from the left ventricle to the left atrium.11,20,22 Although novel and promising, measuring LAS does require further exploration to ensure that it remains relatively pragmatic for clinical practice, as it can often be challenging to perform in patients living with pulmonary disease and poor acoustic windows.
As to why COPD patients may develop concomitant HFpEF, this is likely to be due to systemic inflammation, which has been shown to be a shared pathophysiological process in both COPD and HFpEF.23,24 Exercise intolerance in HFpEF is closely linked to a number of derangements within the pulmonary system, including pulmonary vascular congestion, pulmonary hypertension, impaired alveolar gas exchange and increased ventilation/perfusion mismatch. This is coupled with respiratory musculoskeletal wasting following underlying metabolic and structural changes in skeletal muscle, which are also common features in COPD.25 This may also explain why diabetes was shown to be a significant risk factor associated with HFpEF development in our study population. Diabetes is commonly associated with interstitial fibrosis and myocardial stiffening (consequent from hyperglycaemia and advanced glycated end-products production), and free fatty acid usage within the myocardium, all of which lead to higher oxidative stress, and release of proinflammatory and profibrotic cytokines.15 Similarly, COPD has been described as a low-grade inflammatory state, where a key pathophysiological pathway in the disease includes inflammation in proximal and distal airways, parenchymal tissues and pulmonary vasculature.25–28
Surprisingly, other comorbidities that have been linked to inflammation or a proinflammatory state, such as AF and obesity, were not shown to be equally significant.26,29 Although this may be attributed to our small sample size (and, thus, such associations should be further evaluated through future studies that incorporate a much larger cohort), an intriguing hypothesis would be that the patients with HFpEF in our cohort were unique. Unlike conventional descriptions of HFpEF affecting mainly women and in those with hypertension, our cohort was predominantly men, with no statistically significant difference seen in rates of hypertension between the two groups, identifying a potentially novel phenotype of HFpEF; that is, a HFpEF-COPD cohort in our region, where its prevalence and significance may be worth exploring.
Raised NT-proBNP levels were shown to be associated with a higher risk of having HFpEF, with every 1.0 pg/nl increase of NT-proBNP increasing the odds of HFpEF by 0.8% (p=0.035). NT-proBNP remains an excellent marker of ‘heart stress’, a term commonly used to denote abnormal myocardial wall stretching following either volume and/or pressure overload.30,31 It should be noted that there is growing appreciation regarding how right-sided pressure overload, which may occur with pre-existing COPD in the absence of HF, can also lead to raised NT-proBNP, potentially leading to misdiagnosis of HF; and that HFpEF has been linked to lower levels (and sometimes normal levels) of NT-proBNP, which may in turn lead to underdiagnosis of the condition.30–32 However, those remain the exception and not the rule, and any elevated levels of NT-proBNP warrants interrogation of left ventricular systolic and diastolic function, as to not miss a diagnosis of HF, and normal levels of NT-proBNP despite highly suggestive clinical symptoms for HFpEF warrants further diagnostic work-up. If anything, our pilot study demonstrates a potential to develop a risk prediction tool using both NT-proBNP levels and novel strain-based imaging parameters to improve the detection accuracy for HFpEF, specifically in complex groups of patients, such as in those living with COPD, although this remains beyond the scope of the current study.
Limitations
Our pilot study identifies an important gap in the guidelines with regard to triaging patients with intermediate risk of HFpEF – where current recommended diagnostic procedures and protocols can be difficult to pursue by a significant proportion of the population. This raises the need for more practical approaches, using both biomarker-based and imaging-based tools to help with the diagnostic pathways. As it stands, our study is unable to accurately stratify patients within the intermediate-risk group, which could possibly underreport the true prevalence of HFpEF among our COPD cohort.
Another notable limitation in our pilot study includes the small sample size, which restricts our ability to perform multivariate logistic regression to assess independent predictors and controlling confounders. Additionally, our pilot study was not designed to be an outcomes trial. Thus, the relationship between significant predictive parameters for a high probability of HFpEF, and confirmation of the diagnosis using gold standard invasive haemodynamics alongside long-term clinical outcomes, such as mortality and hospitalisation, could not be established. Opportunities to include such analysis in subsequent studies would be beneficial, where a more prolonged study duration with extended follow-up could be incorporated. Multivariable regression analysis was, also, not performed due to sample size limitations in this study, which restricts insight into the correlation between variables, especially in exploring complex, multicausal data relationships that often requires multivariable analysis.
Conclusion
To our knowledge, our pilot study is the first of its kind to delineate the prevalence of HFpEF within a COPD population regionally. This remains an important effort, as HFpEF remains poorly recognised and underdiagnosed, especially among those living with chronic respiratory illnesses, and identification of HFpEF has the potential to change disease trajectory now that effective, evidence-based therapeutics in HFpEF are on the rise. Our study highlights the significant prevalence of HFpEF among patients living with COPD in our population, which may warrant screening for the disease. However, screening for HFpEF using current diagnostic pathways can be challenging, especially in a COPD cohort, underlining the need for a more tailored approach that combines both biomarkers and imaging-based parameters. Our pilot study provides a platform that future studies incorporating much larger cohorts could be built on, which would help further validate our study’s results.
Clinical Perspective
- Heart failure with preserved ejection fraction remains underrecognised among patients living with chronic obstructive pulmonary disease.
- Our study identifies an important gap in the guidelines with regard to triaging patients with intermediate risk of heart failure with preserved ejection fraction.
- A tailored and pragmatic approach combining both biomarkers and imaging is needed in the detection of heart failure with preserved ejection fraction in patients with chronic obstructive pulmonary disease.