The main findings of this experimental study are that (1) on DCR imaging, the PV change rate at the LV apex was significantly correlated with LVEF in patients with HFrEF, but not in patients with HFpEF; (2) the correlations were stronger in the supine position than in the standing position; (3) the PV change rate at the LV apex was significantly lower in patients with HFrEF than in patients with HFpEF; and (4) the PV change rate at the LV apex could identify patients with HFrEF, although future large-sample studies are needed to validate its clinical application value.

Advantages and usefulness of dynamic chest radiography compared with conventional chest radiography

Conventional chest radiography is the most commonly used examination method in the cardiovascular field. Conventional chest radiography can be used to evaluate pulmonary congestion, cardiomegaly in the pulmonary vessels, cardiac shadows, alveolar oedema, and pleural effusion, amongst other signs of HF. Conventional chest radiography cannot directly assess cardiac function, and the cardiothoracic ratio is an insufficient indicator of LV dysfunction [26]. Therefore, this imaging method does not tend to be used alone for the diagnosis or substratification of HF. DCR provides a dynamic image with less radiation exposure, as well as additional information to conventional chest radiography, making it possible to derive the degree of cardiac contraction and therefore determine LVEF. Overall, although simple chest radiography may reveal the presence of HF signs, it does not provide quantitative information on cardiac function and is not used to make a diagnosis, whereas DCR provides quantitative information that could be useful to diagnose HF. Therefore, DCR augments conventional chest radiography for use in the context of HF.

To date, no attempts have been made to estimate more detailed circulatory indices from radiographs. Therefore, in a recent study [23], we attempted to evaluate haemodynamics in patients with HF using DCR images. We found that changes in the PVs of images obtained from the LV apex ROI were useful for evaluating cardiac function and haemodynamics [23]. On the basis of the findings from our previous study, we used the PV change rate in the LV apex ROI as an image parameter in the present study. In doing so, we were able to distinguish between patients with HFrEF and patients with HFpEF using the PVs obtained by DCR. This information cannot be obtained by conventional simple chest radiography. Therefore, the estimation of LVEF and the detection of LV systolic dysfunction in patients with HF can be achieved with DCR but not with conventional chest radiography, which is of great clinical value.

Mechanisms of the relationship between the pixel value change at the left ventricular apex and left ventricular ejection fraction

In our previous study [23], the rate of change in the PV at the LV apex on DCR images from patients with HF had a strong positive correlation with cardiac output and cardiac index measured by right heart catheterisation. Furthermore, the correlations between the image and haemodynamic parameters were stronger in the supine position than in the standing position. The results of the present study are consistent with our previous study [23]. LV contractility, represented by LVEF, showed the same trend as these haemodynamic parameters, and hearts with better LV contractility tend to have a higher cardiac output and cardiac index. Although the exact mechanism by which the rate of change in the LV apex PVs is correlated with LVEF remains unclear, the significant association between LV contractility and cardiac output supports this mechanistic explanation.

The PV change rate at the LV apex was significantly correlated with LVEF in patients with HFrEF, but not in patients with HFpEF. This could suggest that the use of DCR in patients with undifferentiated HF (LVEF not known) in whom the actual LVEF is > 50% may be limited; however, further validation in a larger patient population is needed to better understand its clinical application potential in specific HF populations. In addition, the rate of PV change at the LV apex was significantly lower in patients with HFrEF than in patients with HFpEF. The reason for the lack of a correlation between the PV and LVEF in HFpEF is not entirely clear; however, if changes in PV reflect changes in blood volume, it is speculated that the smaller volume of the LV lumen in patients with HFpEF compared with that in patients with HFrEF may be a contributing factor. In the present study, dilated cardiomyopathy accounted for 37% of patients with LVEF < 50%, with a median LVEDD of 59 mm and LV remodelling. Patients with HFrEF (LVEF < 40% and LVEF < 30%) had a greater tendency toward LV dilation than patients with LVEF < 50%. However, in patients with HFpEF, the median LVEDD was 45 mm, which indicated a significant difference in LV size.

In the present study, no patients had moderate-to-severe aortic regurgitation, which did not significantly affect the LV lumen volume or, by extension, the rate of PV change at the LV apex. In contrast, the coexistence of moderate or severe mitral regurgitation was more common in patients with HFrEF than in patients with HFpEF, which suggested that mitral regurgitation, especially in the small LV lumen of patients with HFpEF, may further influence changes in LV blood volume. The present study did not provide an accurate measurement of the LV lumen volume or detailed quantification of valvular regurgitation; thus, further studies are needed.

Some patients with dilated cardiomyopathy or other cardiomyopathies may have a remodelled and enlarged right ventricle [27]. The LV apex ROI was set at the left edge of the cardiac shadow (i.e. touching-the-edge). The validity of this apical ROI in the evaluation of cardiac function in patients with cardiac disease was demonstrated in our previous study [23] and in a study by Okamoto et al. [28]. This ROI is not identical in all patients and may vary depending on the degree of ventricular dilatation, the aetiology of HF, and the presence of cardiomyopathy. Moreover, depending on the morphology of the heart, it may reflect predominance of the right ventricle rather than the left ventricle. Furthermore, the effect of pressure drainage to the heart via the lungs was not fully investigated in this study [29]. The influence of this anatomical perspective has not been fully investigated and is a subject for future studies.

Significance of left ventricular ejection fraction estimation and the identification of patients with heart failure with reduced ejection fraction by dynamic chest radiography

Okamoto et al. [28] showed that in 61 patients with cardiovascular disease undergoing DCR, the PV change rate at the LV apex could distinguish patients with reduced EF (i.e. LVEF < 50%). In 21 of the 61 patients with reduced LVEF, the cutoff values of the PV change rate at the LV apex that attained 95% specificity and 95% sensitivity were 7.7% and 17.4%, respectively. Miyatake et al. [19] did not specify how many patients with HF were included in their population; therefore, making a simple comparison with our results is difficult. However, our study is unique in that it is the first to investigate whether LVEF measured by DCR is useful to classify and identify patients with HF. Classifying patients with HF by LVEF matches the known classification of cardiac function by LVEF in patients with HF.

HF is classified on the basis of the LVEF at the time of echocardiography, as follows: LVEF < 40% is HFrEF; 40% ≤ LVEF < 50% is HFmrEF; and LVEF > 50% is HFpEF [3, 4]. Recent studies have reported normal or nearly normal (i.e. > 50%) LVEF in more than one-half of patients with HF. Patients with HFpEF have a similarly poor prognosis as patients with HFrEF, even though LVEF is preserved. Drug therapy for HFpEF has not yet been fully established, and fewer treatment options are available for HFpEF than for HFrEF [5]. In contrast, drug therapy and cardiac device therapy are well-established for HFrEF, and early diagnosis can improve cardiac prognosis [3, 4].

In the present study, the cutoff values for the change in PV were clearly demarcated at the border between patients with LVEF < 40% and patients with LVEF < 50%. This finding indicates that patients with LVEF < 40% can be identified by DCR. Furthermore, LVEF < 40% was consistent with the classification of HF based on current HF guidelines [3, 4]. In addition, the lower cutoff for PV change at the LV apex in patients with LVEF < 50% versus patients with LVEF < 40% (9.3% vs. 5.5%) parallelled lower LV contractility, which is reasonable. However, the fact that the cutoff value of PV change at the LV apex did not change between patients with LVEF < 40% and patients with LVEF < 30% (5.5% vs. 5.5%) may indicate a limitation in the ability of DCR to identify LVEF, although the mechanism is unclear. Further studies are needed to determine whether a more detailed LVEF classification is possible in a larger number of patients with HF.

It is important to note that LVEF and cardiac output or cardiac index are not identical, although the cardiac index is an important indicator of cardiac function and has prognostic relevance [30]. Some patients with HFrEF maintained their cardiac output, whereas others with HFpEF had reduced cardiac output. Thus, the LVEF should be understood as a separate index from cardiac output, and DCR could hopefully be used to ascertain LVEF.

In patients with HFrEF, cardiac output may be reduced in advanced HF, and intracardiac pressure is often elevated, resulting in a poor prognosis. Pharmacological and cardiac device therapies for HFrEF are better established than those for HFpEF. Therefore, the identification, early diagnosis, and treatment of patients with HFrEF by DCR have clinical significance as they may contribute to improving the prognosis of these patients.

LVEF is a widely used and well-established indicator of cardiac function, and it is easily understood and commonly used as a measure of cardiac contractility. LVEF is usually measured using the gold standard of echocardiography, which is economical, radiation-free, readily available, and fast; however, accurate and valid findings are difficult to obtain without a cardiologist or laboratory technician skilled in echocardiography. DCR imaging may therefore be useful for the preliminary assessment of LVEF by non-cardiologists who have difficulty performing echocardiography. DCR would provide information about the LVEF range, which may be helpful to determine the approach to acute management. DCR as an initial assessment could help to identify patients with HFrEF and promptly consult a cardiologist for more detailed assessment by echocardiography. Although the use of echocardiography is widespread and markers such as BNP are also available to stratify patients with HF, the ability of DCR to classify LVEF and stratify the HF patient population may also be meaningful. In particular, we suggest that DCR may be useful for the screening of patients with HFrEF before more detailed evaluation by echocardiography, which can be difficult for non-cardiologists to perform. Moreover, in many developed countries, the wait time to see a cardiologist can be quite long given how in demand they are. Therefore, DCR performed by a primary care provider may be helpful to bridge the period of time until the patient can see a cardiologist and undergo more detailed evaluation by echocardiography.

However, as the sample size of the present study on the utility of DCR was small, it is positioned as an experimental study only. Further research with larger numbers of patients would be needed to determine whether DCR has clinical application value for identifying changes in LVEF before thorough assessment by echocardiography and to determine the place of DCR in the current diagnostic pathway for HF.

Study limitations

This study has several limitations. First, the sample size was small, not all confounding factors were evaluated, and the study is considered to be a pilot feasibility (experimental) study. In addition, DCR findings were not compared between healthy controls and patients with HF. Second, the study was conducted on a population of patients with relatively mild HF classified as NYHA functional class I or II, despite including a small number of patients classified as NYHA functional class III. Therefore, we were unable to fully demonstrate the applicability of the system in patients with more severe HF. In this study, only five patients with NYHA functional class III HF and no patients with class IV were included. Therefore, further large-sample multicentre studies are needed in these groups. Third, the reliability of DCR to identify changes in LVEF in patients with HF may be limited in some patients, such as patients with plural effusion, lung consolidation, pulmonary congestion, or chronic pulmonary emboli. The same may also be the case for patients with right ventricular hypertrophy, pulmonary emphysema, or pulmonary hypertension. These patients were excluded from the present study; therefore, the utility of DCR in patients with acute HF with these comorbidities should be evaluated in the future. Moreover, extending the patient pool to include more patients with a variety of comorbidities would be beneficial to validate the utility of DCR in more complex populations. Fourth, our results showed that the LV apex PV change rate was not correlated with LVEF > 50% or < 30%, but LVEF < 40% could be detected. The LVEF cutoff for cardiac resynchronisation therapy (CRT) is 35%. Therefore, DCR may be limited in its ability to identify specific LVEF categories precisely enough to indicate patients for specific treatments, such as CRT. Nevertheless, we consider our results to be preliminary and the sample size was small. In the future, we would like to study a large patient population to validate the ability of DCR to distinguish different LVEF categories and to investigate whether DCR parameters other than the PV change rate are useful in the context of HF diagnosis and stratification. Finally, our study included only Japanese patients, which may limit the generalisability of our results to racially diverse populations.