Prognostic values of coronary artery calcium score and 123I-BMIPP SPECT in patients with non-ischemic heart failure with preserved ejection fraction

This study aimed to determine whether coronary artery calcium score (CACS) can be a prognostic indicator for the development of major adverse cardiac events (MACEs) and compare the value of CACS with that of the 123I-betamethyl-p-iodophenyl-pentadecanoic acid (123I-BMIPP) defect score (BDS) in patients with non-ischemic heart failure with preserved ejection fraction (NIHFpEF). Among 643 consecutive patients hospitalized due to acute heart failure, 108 (74 ± 13y) were identified to have NIHFpEF on non-contrast regular chest computed tomography and 123I-BMIPP single-photon emission computed tomography (SPECT). We evaluated whether CACS and BDS were associated with MACEs using multivariate Cox models. Thirty-two MACEs developed at a mean follow-up period of 2.4 years. CACS > 0 (hazard ratio [HR] 2.38, 95% confidence interval [CI] 1.02–5.54) and higher BDS (HR 16.00, 95% CI 5.88–43.49) were significantly associated with the development of MACEs. The proportion of patients who experienced MACEs was significantly higher in the CACS > 0 and high BDS group than in the CACS = 0 and low BDS group (3% vs. 75%, p < 0.001). CACS, as well as BDS, could serve as potential prognostic indicators in patients with NIHFpEF.


Introduction
One in nine people die of heart failure (HF), and half the patients diagnosed as having HF die within 5 years [1]. Approximately half the patients with HF have preserved ejection fraction (HFpEF) [2]. Although considerable research efforts and funds have been devoted to improve mortality and reduce hospitalization, there has been no evidence-based treatment developed, to date, in patients with HFpEF [3].
The coronary artery calcium score (CACS), calculated using non-contrast regular chest computed tomography (CT), has recently been recognized as a prognostic marker for the assessment of cardiovascular event risk among asymptomatic patients [4]. The most common cause of HFpEF is diastolic HF, and the primary cause of diastolic HF is hypertensive heart disease [5]. Experimental and clinical data have demonstrated serological and morphometric evidence of increased myocardial fibrosis in patients with hypertensive heart disease [6,7]. Therefore, HFpEF could be related to the progression of arteriosclerosis. However, the prognostic utility of CACS in patients with nonischemic HFpEF (NIHFpEF) has not been well examined. On the other hand, 123 I-betamethyl-p-iodophenyl-pentadecanoic acid single-photon emission computed tomography ( 123 I-BMIPP SPECT) plays an important role in the assessment of cardiovascular event risk in patients with coronary artery disease [8]. Although, in the clinical setting, the relationship between 123 I-BMIPP defect score (BDS) and CACS of the patients with NIHFpEF, and clinical importance of them were unclear. Our previous study showed that 123 I-BMIPP SPECT could be a useful modality for identifying high-risk patients with NIHFpEF [9]. In the present study, we aimed to investigate whether CACS is associated with the occurrence of major cardiac events (MACEs) and to compare the prognostic value of CACS with that of BDS in patients with NIHFpEF.

Patient population
Among 643 consecutive patients who were admitted to our hospital for congestive HF and underwent non-contrast regular chest CT to evaluate the presence of complication of pneumonia or quantity of pulmonary congestion and pleural effusion, 123 I-BMIPP SPECT imaging to evaluate the presence of ischemic myocardial injury, echocardiography, and coronary artery angiography within 60 days between January 2010 and January 2014. We finally enrolled 108 patients with no obstructive coronary artery disease (< 50% stenosis) on coronary artery angiography and had preserved ejection fraction (≥ 50%) on echocardiography (Fig. 1). The patients' clinical characteristics, including age, sex, coronary risk factors, blood biochemical data, echocardiography data, and medications were assessed. The institutional review board approved this retrospective study and the requirement to obtain informed consent was waived (M20136). This study was performed in accordance with the ethical standards of the Declaration of Helsinki.

Invasive coronary angiography
Invasive coronary angiography was performed for all participants. According to the American College of Cardiology/ American Heart Association guidelines [10], quantitative coronary angiography was performed by experienced interventionalists. Coronary stenosis was defined as > 50% diameter stenosis on invasive coronary angiography.

Echocardiographic imaging
Echocardiography from the parasternal window was performed to evaluate left ventricular function (Vivid E9device, GE Vingmed, Horten, Norway). The Teichholz formula was used to calculate the left ventricular ejection fraction (LVEF) [11]. A LVEF > 50% was defined as a preserved EF.

Non-contrast non-cardiac chest CT
Baseline CT scans were conducted with a 64-slice multidetector CT scanner (SOMATOM Definition Flash, Siemens, Munich, Germany). The scanning protocol consisted of the following parameters were applied: 28 mm × 0.6 mm beam collimation, pitch 1.2, caudocranial scan direction, and smallest field of view to include the outer rib margins. No electrocardiographic triggering was performed, and no contrast agent was administered. Exposure settings were applied based on body weight: 125mAs at a tube voltage of 120 kVp. All scans were reconstructed as 5.0 mm thick slices with an increment of 5.0 mm. Coronary artery calcium measurements were performed on a workstation (SYNAPSE VIN-CENT; FUJIFILM Medical Co., Ltd, Tokyo, Japan), and CACS was calculated using the Agatston scoring method [12]. 123 I-BMIPP (111 MBq) was injected with the patients at rest. Twenty minutes later, 123 I-BMIPP SPECT was performed over 360° in 72 steps of 37.5 s each in a 64 × 64 matrix, using a triple-head gamma camera (Prism IRIX; Philips, Amsterdam, Netherlands) equipped with low-energy general-purpose collimators. A Butterworth filter (order 8.0, cutoff value 0.25-0.30 cycle/pixel) and filtered back projection were used to process and reconstruct the images, respectively.

I-BMIPP imaging
BDS was calculated according to a previously reported method [9], using an automated program for myocardial SPECT (Heart Risk View-S software; Nihon Medi-Physics Co Ltd, Tokyo, Japan). A hybrid 2-part sampling method [13], i.e., an algorithm for generating a polar map, was used to generate count profiles from a 3-dimensional sampling scheme of short-axis slices by operating in a 3-dimensional space and using short axis images. Then, polar maps were generated from 123 I-BMIPP SPECT data divided into 17 segments to calculate the mean count in each segment, as recommended by the American Society of Nuclear Cardiology guidelines [14]. The mean counts in these segments were compared with references from the 123 I-BMIPP database for Japanese patients which was developed by the Japanese Society of Nuclear Medicine working group [15]. The mean % uptake in each segment was derived and converted to scores using a five-point grading system (0, normal; 1, mildly reduced; 2, moderately reduced; 3, severely reduced; 4, absent). The BDS value in each segment was summed to obtain the total defect score in each patient.

Assessment of clinical outcomes
The endpoint was defined as the occurrence of MACEs, including cardiac death, which was defined as death caused by HF, acute myocardial infarction, or other definitive cardiac disorders, cardiovascular events (acute myocardial infarction, unstable angina), or severe HF requiring hospitalization. Standard laboratory, electrocardiogram, or examination criteria were used for the diagnosis of acute myocardial infarction and unstable angina. HF exacerbation was defined as dyspnea accompanied by pulmonary edema or congestion on chest radiography requiring hospitalization. The first event was included in the counting of clinical outcomes during the follow-up period. The event data were retrospectively gathered from the patients' records, including in-hospital or out-of-hospital reports.

Statistical analysis
Data are expressed as average ± standard deviation of continuous variables. Continuous variables in patients with and without events were compared using the Mann-Whitney U test, and categorical data were analyzed using the chisquare test. Age, sex, and factors with a significance level of P < 0.05, were included in a univariate Cox regression model. Variables with significant probable values were included in a multivariate Cox regression model to determine whether the future occurrence of MACE was associated with echocardiographic imaging, CACS, or BDS. To evaluate the clinical importance of CACS and BDS, all patients were divided into two groups based on their CACS and BDS. The cutoff value of CACS was determined to be 0, 1 to 100, or ≥ 100, comparing between MACEs group and no MACEs. The cutoff value of BDS was determined using the area under the curve (AUC) from a receiver operating characteristic (ROC) analysis performed based on the occurrence of MACEs. The proportion of event-free patients was estimated using the Kaplan-Meier method and compared between the high and low CACS and BDS groups using the log-rank test. Subsequently, to evaluate the clinical importance of the combined use of CACS and BDS, all patients were divided into four groups and assessed using the Kaplan-Meier method and log-rank test. A P-value of < 0.05 was considered statistically significant. All statistical analyses were performed using StatMate V software version 5.01 (Advanced Technology for Medicine and Science, Tokyo, Japan).
Overall, 32 patients (30%) experienced MACEs over a mean follow-up period of 2.4 ± 1.6 years. Cardiac deaths occurred in four patients (acute myocardial infarction in two patients, deterioration of HF in two patients), nonfatal acute myocardial infarction in one patient, and severe HF requiring hospitalization in 27 patients, respectively. Table 1 shows that BNP, left atrial dimension (LAD), left ventricular end-diastolic volume index (LVEDVI), left ventricular mass index (LVMI), CACS, and BDS were significantly higher in patients with MACEs. From the ROC analysis, the cutoff value for high BDS was 4 (AUC = 0.87). To determine the cutoff value of CACS, the patients categorized as CACS 0, 1 to 100, and ≥ 101, and were compared between the MACE and non-MACE MACEs was significantly higher in the CACS > 0 group than in the CACS = 0 group (Fig. 2). Of the 32 events, 26 occurred in the high BDS group. The proportion of patients who experienced MACEs was significantly higher in the high BDS group than in the low BDS group (Fig. 3). The Kaplan-Meier curves for MACEs in the combined groups of CACS and BDS are shown in Fig. 4. The proportion of patients who experienced MACEs was significantly  Figure 5 shows a typical patient in the CACS > 0 and high BDS group. This 72-year-old man had HF due to hypertensive heart disease (NYHA: class II, Nohria-Stevenson classification: wet and warm). He had a history of hypertension, diabetes mellitus, and hyperlipidemia. He also had a smoking habit. He underwent non-contrast non-cardiac chest CT for HF evaluation, and 123 I-BMIPP SPECT, coronary artery angiography, and echocardiography owing to a suspicion of coronary heart disease. He had no coronary artery disease on coronary artery angiography, and his LVEF was 56% on echocardiography. His CACS was 1150.4, and his BDS was 7. In this case, the patient was admitted to the hospital because of HF deterioration at 210 days after non-contrast non-cardiac chest CT.

Discussion
The findings of the present study demonstrated that CACS calculated using non-contrast regular chest CT was associated with an increase in MACEs and 123 I-BMIPP SPECT findings, and that the evaluation of CACS in addition to BDS had a predictive value for the identification of future MACEs in patients with NIHFpEF.

HFpEF
As patients with HFpEF are in a high-risk situation, it is essential that future cardiac risk be predicted using noninvasive imaging modalities. Echocardiography [16] and cardiac magnetic resonance imaging [17] have been shown to be useful in determining the outcomes of patients with HFpEF. In general, left ventricular diastolic dysfunction is recognized as a major pathophysiological abnormality [18]. Mechanisms underlying the development of HFpEF include Fig. 4 Kaplan-Meier curve in reference to MACEs stratified by combination of CACS and BDS values The y-axis represents the cumulative event-free rate; the rate in the CACS = 0 and low BDS group was significantly higher than in the CACS > 0 and high BDS group (P < 0.001) Abbreviations as in Fig. 2 and 3

Fig. 5
Non-contrast non-cardiac chest CT image, polar map and heart risk map of 123 I-BMIPP SPECT images of a patient in the CACS > 0 and higher BDS group The CACS was 1150.4 and BDS was 7. This patient was admitted to hospital due to deterioration of HF 210 days after the examination endothelial dysfunction, myocardial hypertrophy, and myocardial fibrosis [19]. Endothelial dysfunction is reported to play an important role in the initial stage of atherosclerosis development [20,21]. Atherosclerosis has been shown to be involved in the development of hypertension, accumulation of extracellular matrix material, and fibrosis within myocardial tissue [22]. More specifically, endothelial dysfunction initially causes atherosclerosis, leading to the development of hypertension, increase of extracellular matrix material, and fibrosis within myocardial tissue, which in turn results in left ventricular diastolic dysfunction. Gaudieri et al. reported that coronary vascular dysfunction assessed using 82 Rb positron emission tomography imaging was associated with MACEs in patients with resistant hypertension. Resistant hypertension was associated with an impaired myocardial perfusion reserve. The results from these studies indicate that myocardial perfusion reserve assessment might identify an earlier stage of coronary dysfunction in the evolution of the atherosclerosis process [23,24]. The present study demonstrated the potential of CACS to reflect the status of atherosclerosis, and BDS aside from a previous study showed the potential for reflecting the amount of extracellular matrix material and fibrosis within the myocardial tissue might reflect coronary dysfunction, which leads to the step of atherosclerotic progression.

Non-contrast non-cardiac chest CT
The prognostic value of CACS calculated using cardiac CT has been evident across various population groups [25,26]. Previous reports have shown that CACS by using noncontrast non-cardiac chest CT is a valid modality for determining CACS with an accuracy of up to 90%, compared with dedicated cardiac CT [27]. Hughes-Austin JM et al. reported that the CACS calculated using non-contrast noncardiac chest CT is highly correlated with the CACS calculated using cardiac CT, and that the two scores are similarly associated with the mortality risk [28]. While the prognostic potential of CACS calculated using non-contrast non-cardiac chest CT in patients with HFpEF remains unexplored, this is the first study on the relationship between the prognostic value of CACS calculated using non-contrast non-cardiac chest CT and NIHFpEF. Our findings suggest that the evaluation of atherosclerosis could predict the grade of LV diastolic dysfunction and that could be able to predict the prognosis of HFpEF excepting for predicting of ischemic heart disease.

Prognostic value of CACS in addition to BDS in patients with NIHFpEF
In the current study, NIHFpEF patients with CACS > 0 and high BDS had a poor prognosis, and CACS > 0 and high BDS were independent predictors of MACEs in multivariate analysis. In a previous report, the prognostic value of CACS and myocardial perfusion reserve obtained with 82 Rb PET/CT in patients with suspected coronary artery disease was investigated, and CACS and coronary vascular dysfunction were independently associated with an increased risk of MACEs [29,30]. Dikic et al. demonstrated that both CACS and coronary flow reserve obtained using CT and echocardiography provided an independent prognosis in asymptomatic diabetic patients [31]. These findings suggest that mere anatomical information obtained using angiography is not sufficient to explain the physiologic severity and prognosis of HF. The mechanism, including the effect of atherosclerosis on vasculature resulted in hypertension, coronary arteries resulted in microcirculatory dysfunction, and myocardium resulted in left ventricular diastolic dysfunction, may also contribute to cardiovascular risk in patients with NIHFpEF.
In the present study, CACS calculated using non-contrast non-cardiac chest CT could detect not only coronary calcification but also left ventricular diastolic dysfunction and hypertension due to heart failure. Moreover, BDS calculated using 123 I-BMIPP SPECT could detect impaired not only myocardial metabolism caused by microcirculatory dysfunction but also left ventricular diastolic function due to atherosclerosis. The combined use of these two parameters could improve risk stratification and help identify high-risk patients with NIHFpEF who require aggressive treatment.

Study limitations
There are several limitations to this study. First, the relatively small sample size and short follow-up period precluded the statistical reliability of the study. It was also difficult to determine the cutoff values of CACS and BDS using timedependent ROC. However, our results showed higher CACS and BDS to be significantly associated with the development of MACEs. Second, this study did not provide data on echocardiographic parameters, such as E/e', which may have an important implication for determining diastolic function. Instead, other diastolic parameters, such as LVEDVI and LVMI, were calculated as substitutes for diastolic parameters. Third, the study population might have included patients with ischemia and no obstructive coronary artery disease or myocardial infarction with non-obstructive coronary arteries. Therefore, patients with ischemic heart disease could not be completely excluded from this study. Fourth, we did not compare the accuracy between CACS obtained without using electrocardiogram (ECG)-gated CT and CACS obtained using ECG-gated CT, because we had no data on CACS obtained using ECG-gated CT. However, a previous study reported that CACS obtained without using ECG-gated CT is highly correlated with CACS obtained using ECG-gated CT. It is necessary to confirm the accuracy of CACS obtained without using ECG-gated CT in patients with NIHFpEF. Finally, this study was limited by the retrospective nature of the data obtained from non-contrast regular chest CT, 123 I-BMIPP SPECT, echocardiography, and coronary artery angiography, as well as the outcomes of patients with NIHFpEF. Because patients with a severe grade of HF or kidney function for undergoing examinations did not participate in this study, the timing of the imaging studies varied from patient to patient, and patient outcomes, which were reviewed based on the medical records, might have been incomplete. Further prospective studies with a larger number of participants are warranted to confirm the prognostic values of CACS and BDS in patients with NIHFpEF.

Conclusion
In this study, CACS demonstrated a high prognostic value for MACEs in patients with NIHFpEF. The evaluation of CACS in addition of BDS could have a predictive value for the identification of future MACEs in patients with NIHFpEF.