論文ID: CJ-15-0425
Background: Heart failure (HF) with preserved (HFpEF) left ventricular ejection fraction (LVEF) is a syndrome with complex pathophysiology. Little is known about changes in LVEF that occur over time in HFpEF patients. A fundamental clinical question about HFpEF is whether HFpEF is an early manifestation of HF with reduced LVEF (HFrEF). If so, which patients with HFpEF are likely to show a decline in LVEF to less than 50%? The aim of the present study was to examine longitudinal changes in LVEF in patients with HFpEF.
Methods and Results: Among 279 consecutive HFpEF patients admitted as emergencies, we examined 100 who underwent echocardiography at least 1 year after discharge. EF >50% was used as the definition of HFpEF. During a mean duration from hospitalization to follow-up echocardiography of 31.5 months, 11% of patients had LVEF ≤50% (mildly reduced LVEF), known as mildly reduced (HFmrEF). The utility of LVEF during hospitalization to predict HFmrEF was assessed with receiver-operating characteristic curve analysis. A cutoff value of 55% had sensitivity of 90.9% and specificity of 97.7%. Logistic regression analysis indicated that LVEF ≤55% and ischemic etiology were strong predictors of progression from HFpEF to HFmrEF (odds ratio [OR] 435, 95% confidence interval [CI] 52.65–10,614, P<0.0001 and OR 10.9, 95% CI 2.60–74.80, P=0.0007, respectively).
Conclusions: The present study suggests that HFpEF patients with LVEF ≤55% may progress to HFmrEF in the future.
Heart failure (HF) is an important public health issue worldwide. Until now, most large clinical studies have targeted HF with reduced (HFrEF) left ventricular ejection fraction (LVEF).1–4 However, HF with preserved LVEF (HFpEF) has recently gained attention because many large clinical studies have demonstrated that half of HF patients have HFpEF5–7 and they have a similar poor prognosis as those with HFrEF,8–11 even though various lines of evidence suggest that the pathophysiology of HFpEF is different from that of HFrEF.
HFpEF is a complex syndrome, of which the molecular mechanisms and clinical characteristics remain unclear. Recently, some studies12,13 have reported changes in LVEF that occur over time in patients with HFpEF; a substantial number of patients with HFpEF showed a decline to LVEF <50%. However, it is unclear which patients with HFpEF are more likely to show such a decline. In this context, we performed a longitudinal assessment of LVEF based on echocardiography in patients with acute decompensated HF (ADHF) in the Nara Registry and Analyses for Heart Failure 2 (NARA-HF 2 Study) cohort study.
The NARA-HF 2 Study recruited 611 consecutive patients admitted as emergencies with documented ADHF (either acute new-onset or acute-on-chronic HF) between January 2007 and December 2012.14–16 The diagnosis of HF was based on the Framingham criteria.17 The study population included both HFrEF and HFpEF patients, but patients with acute myocardial infarction (AMI), acute myocarditis, and acute HF with acute pulmonary embolism were excluded.
The NARA-HF Study 2 included 279 patients with LVEF >50%. We analyzed data from 100 patients who underwent follow-up with echocardiography at least 1 year after discharge. The remaining 179 patients were not enrolled in the present investigation: 15 patients died in the hospital during the emergency admission, 55 patients died within 1 year of discharge, 7 patients were lost to follow-up, and 102 patients were not able to undergo follow-up echocardiography in at the study hospital. None of the 100 patients had severe valvular disease (aortic or mitral stenosis or regurgitation) or developed new-onset AMI during the follow-up period. For each patient, baseline data included age, sex, body mass index (BMI), HF etiology, medical history, as well as vital signs, laboratory data, medications, and echocardiography results during hospitalization and at follow-up.
The study was approved by the Ethics Committee of Nara Medical University, and written informed consent was given by all patients according to the Declaration of Helsinki Ethical Principles for Medical Research Involving Human Subjects.
DefinitionsUsing echocardiography, we measured LVEF at admission and at follow-up at least 1 year after discharge. We adopted the generally accepted criteria of LVEF >50%6,12,18 as the definition for HFpEF in this study. Receiver-operating characteristic (ROC) curve analysis was performed on LVEF data obtained during hospitalization to define a cutoff for predicting LVEF ≤50% at follow-up.
EchocardiographyAll echocardiography was performed at Nara Medical University Hospital. For each patient, echocardiograms obtained during hospitalization and at follow-up (at least 1 year after discharge) included measurements of LV end-diastolic dimension (LVEDD), LV end-systolic dimension (LVESD), LV end-diastolic volume (LVEDV), LV end-systolic volume (LVESV), left atrial dimension (LAD), interventricular septal (IVS) and LV posterior wall (LVPW) thickness by 2D echocardiography or M-mode. LVEF assessment was based on 2D echocardiography using the quantitative 2D biplane volumetric Simpson method from 4- and 2-chamber views. LV hypertrophy (LVH) was defined as IVS and LVPW thicknesses >12 mm. If there echocardiography was performed multiple times during the hospitalization, we used the data from the examination performed closest to discharge, because data immediately after admission might be incorrect because of tachycardia or inadequate positioning. All measurements were calculated separately by 1 echocardiologist and 1 expert sonographer. The variation in measurements between the 2 investigators was 3.1% in the present study.
Statistical AnalysisContinuous variables are expressed as mean±standard deviation or median (interquartile range [IQR]), and between-group differences were compared using Student’s t-test. Categorical variables were summarized as percentages and analyzed using the chi-square test. To evaluate the progression from HFpEF to HFrEF, results are reported as odds ratio (OR), 95% confidence interval (CI), and P values using logistic regression. JMP version 10 for Windows (SAS Institute Inc, Cary, NC, USA) was used for all statistical analyses. P<0.05 was considered statistically significant.
The mean duration between echocardiography during hospitalization for ADHF and follow-up echocardiography was 31.5 months. During this interval, LVEF fell to <50% in 11.0% (n=11) of patients. The mean age at hospital admission was 70.3±12.1 years, and 48.0% of the patients were women. Regarding the etiology of HF, 35.0 % of patients had ischemic causes, 15.0% had valvular causes, 10.0% had hypertensive heart disease, and 6.0% had hypertrophic cardiomyopathy. The New York Heart Association (NYHA) function class on admission was III or IV in 78.0% of patients. The median (IQR) plasma B-type natriuretic peptide concentration at discharge was 191 (131–348) pg/ml (Table 1).
| Total (n=100) |
50%<LVEF≤55% (n=13) |
LVEF >55% (n=87) |
P value | |
|---|---|---|---|---|
| Demographic | ||||
| Age, years | 70.3±12.1 | 69.2±12.8 | 70.5±12.0 | 0.8056 |
| Female, % | 48.0 | 38.5 | 49.4 | 0.4605 |
| Body mass index, kg/m2 | 24.2±4.0 | 25.5±4.3 | 24.0±3.9 | 0.2699 |
| Etiology of HF, % | ||||
| Ischemic | 35.0 | 84.6 | 27.6 | <0.0001 |
| Valvular | 15.0 | 7.7 | 16.1 | 0.4289 |
| Hypertensive | 10.0 | 0.0 | 11.5 | 0.1976 |
| Hypertrophic cardiomyopathy | 6.0 | 0.0 | 6.9 | 0.3288 |
| Medical history, % | ||||
| Hypertension | 85.0 | 84.6 | 85.1 | 0.9668 |
| Diabetes mellitus | 53.0 | 61.5 | 51.7 | 0.5084 |
| Dyslipidemia | 40.0 | 38.5 | 40.2 | 0.8528 |
| Old myocardial infarction | 19.0 | 53.9 | 13.8 | 0.0006 |
| Atrial fibrillation | 33.0 | 23.1 | 34.5 | 0.4146 |
| Procedures, % | ||||
| PCI | 23.0 | 53.9 | 18.4 | 0.0046 |
| CABG | 3.0 | 0.0 | 3.5 | 0.4966 |
| NYHA class on admission, % | ||||
| III or IV | 78.0 | 76.9 | 78.2 | 0.9200 |
| Vital signs at discharge | ||||
| SBP, mmHg | 121.5±17.0 | 117.1±11.2 | 122.2±17.7 | 0.3563 |
| Heart rate, beats/min | 68.9±9.4 | 71.2±5.9 | 68.6±9.8 | 0.2435 |
| Laboratory data at discharge | ||||
| Hemoglobin, g/dl | 11.0±1.9 | 10.9±1.4 | 11.1±2.0 | 0.8922 |
| eGFR, ml/min/1.73 m2* | 32.5 (12.4–58.3) | 25.6 (11.0–46.2) | 35.4 (12.4–58.4) | 0.3822 |
| Sodium, mEq/L | 138.9±3.4 | 139.1±4.9 | 139.8±3.5 | 0.5005 |
| Plasma BNP, pg/ml* | 191 (131–348) | 347 (206–536) | 184 (122–324) | 0.0524 |
| Medication at discharge, % | ||||
| ACE inhibitor or ARB | 80.0 | 69.2 | 81.6 | 0.2980 |
| β-blocker | 39.0 | 46.2 | 37.9 | 0.5707 |
| MR blocker | 20.0 | 15.4 | 20.7 | 0.6466 |
| Diuretic | 78.0 | 76.9 | 78.2 | 0.9203 |
*Data are shown as percentage, mean±standard deviation, or median (interquartile range). ACE, angiotensin-converting enzyme; ARB, angiotensin-receptor blocker; BNP, B-type natriuretic peptide; CABG, coronary artery bypass grafting; eGFR, estimated glomerular filtration rate; HF, heart failure; LVEF, left ventricular ejection fraction; MR, mineralocorticoid receptor; NARA-HF 2 Study, the Nara Registry and Analyses for Heart Failure 2; NYHA, New York Heart Association; PCI, percutaneous coronary intervention; SBP, systolic blood pressure.
The mean LVEF was 67.0±9.2% during hospitalization and 67.4±11.1% at follow-up. During the follow-up period, LVEF decreased in 50.0% of patients (n=50), increased in 45.0% (n=45), and did not change in 5.0% (n=5). The median annual change in LVEF was –0.1%, with 25% and 75% percentiles of –1.9% and +2.6%, respectively. Among patients with a decline in LVEF from hospitalization to follow-up, LVEF decreased to below 50% in 11 patients. Based on ROC curve analysis for LVEF ≤50% at follow-up, the area under the ROC curve was 0.9893. The LVEF cutoff value was 55%, with sensitivity of 90.9% and specificity of 97.7% (Figure 1).
Receiver-operating characteristic curve analysis for progress to heart failure with mildly reduced ejection fraction (HFmrEF). At the optimal cutoff of left ventricular EF 55%, sensitivity was 90.9% and specificity was 97.7%. The area under the curve (AUC) was 0.9893.
As shown by the distribution of LVEF during hospitalization and follow-up (Figure 2), 10 of 11 patients with LVEF <50% at follow-up had LVEF between 50% and ≤55% during hospitalization. Consequently, the proportion of patients with 50%<LVEF≤55% decreased dramatically, from 13.0% during hospitalization to 4.0% at follow-up. Only 1 of 87 patients with LVEF>55% during hospitalization had a follow-up LVEF <50%.
Distribution of left ventricular ejection fraction (LVEF) (A) during hospitalization for acute decompensated heart failure and (B) at follow-up. Red represents patients with LVEF <50% at follow-up and the identical subjects in (A) and (B).
To identify other clinical predictors of LVEF <50% during follow-up, we compared the baseline clinical characteristics of patients with 50%<LVEF≤55% with those with LVEF >55% (Table 1). Age, BMI, and the proportion of females were similar in both groups. With regards to HF etiology, the proportion of patients with ischemic causes was significantly higher in patients with 50%<LVEF≤55% compared with patients with LVEF>55%. The prevalence of old MI was significantly higher in patients with 50%<LVEF≤55% than in patients with LVEF>55%. There were no significant differences in the prevalence of comorbidities other than old MI between the 2 groups. NYHA functional class was similar. Systolic blood pressure and heart rate at discharge were similar in both groups. There were also no significant differences in laboratory findings or medications at discharge.
Table 2 shows the echocardiographic parameters. The mean follow-up duration in both groups was similar. There was a significant difference in the annual change in LVEF between patients with 50%<LVEF≤55% and LVEF >55%. LVEDD and LVESD were significantly higher in patients with 50%<LVEF≤55% than in patients with LVEF >55% at both measurement points. Regarding LV volume, both LVEDV and LVESV were significantly larger in patients with 50%<LVEF≤55% than in patients with LVEF >55% during hospitalization as well as at follow-up. In patients with LVEF >55%, LVEDV and LVESV were unchanged during hospitalization to follow-up, but LVEDV increased by 10.1% and LVESV by 28.6% in patients with 50%<LVEF≤55%. LAD and the prevalence of LVH were similar between the 2 groups (data not shown).
| Echocardiographic parameter | Total (n=100) |
50%<LVEF≤55% (n=13) |
LVEF >55% (n=87) |
P value |
|---|---|---|---|---|
| Time to follow-up echocardiography, months | 31.5±17.0 | 37.3±16.6 | 30.6±17.0 | 0.1426 |
| LVEF during hosp, % | 67.0±9.2 | 51.9±1.9 | 69.2±7.5 | <0.0001 |
| LVEF at follow-up, % | 67.4±11.1 | 46.0±4.1 | 70.6±7.7 | <0.0001 |
| LVEF change per year, %* | −0.1 (−1.9 to +2.6) | −4.3 (−6.0 to −1.5) | +0.5 (−1.4 to +2.7) | <0.0001 |
| LVEDD during hosp, mm | 49.6±7.7 | 55.4±6.1 | 48.8±7.5 | 0.0031 |
| LVEDD at follow-up, mm | 49.4±6.5 | 57.3±6.3 | 48.3±5.7 | <0.0001 |
| LVEDD change per year, ml | 0.0 (−1.4 to +1.6) | +0.3 (−0.4 to +2.9) | 0.0 (−1.5 to +1.6) | 0.1987 |
| LVESD during hosp, mm | 33.1±7.2 | 40.4±5.6 | 32.0±6.7 | <0.0001 |
| LVESD at follow-up, mm | 32.4±6.6 | 42.8±5.6 | 30.9±5.2 | <0.0001 |
| LVESD change per year, ml | 0.0 (−1.6 to +1.2) | 0.0 (−1.1 to +2.2) | 0.0 (−1.7 to +1.0) | 0.2500 |
| LVEDV during hosp, ml | 71.9±31.4 | 100.8±30.7 | 67.5±29.2 | 0.0006 |
| LVEDV at follow-up, ml | 70.3±34.4 | 111.4±48.9 | 64.2±27.2 | <0.0001 |
| LVEDV change per year, ml | −0.5 (−5.6 to +8.0) | +3.0 (−7.0 to +10.4) | −0.5 (−5.4 to +8.0) | 0.5610 |
| LVESV during hosp, ml | 24.9±15.3 | 49.3±16.1 | 21.2±11.3 | <0.0001 |
| LVESV at follow-up, ml | 24.9±20.3 | 62.1±28.4 | 19.3±11.0 | <0.0001 |
| LVESV change per year, ml | +0.1 (−2.7 to +2.9) | +3.4 (−2.0 to +8.6) | 0.0 (−2.8 to +2.5) | 0.0614 |
*Data are shown as percentage, mean±standard deviation or median (interquartile range). LVEF/LVEDV/LVESV change=change between hosp and follow-up. EDD, end-diastolic dimension; EDV, end-diastolic volume; EF, ejection fraction; ESD, end-systolic dimension; ESV, end-systolic volume; hosp, hospitalization; LV, left ventricular.
Next, we examined which factors were associated with the transition of LVEF from >55% to ≤55%. As shown in Table 3, 50%<LVEF≤55% during hospitalization and ischemic etiology were strong predictive factors (OR 435, 95% CI 52.65–10,614, P<0.0001 and OR 10.9, 95% CI 2.60–74.80, P=0.0007, respectively). Other than these 2 factors, LVEDD, LVESD, LVEDV and LVESV were significantly associated with progression to HF with mildly reduced EF (HFmrEF). Regarding the change in LV volume from baseline to follow-up, the annual change in LVESV was a predictor (OR 1.12, 95% CI 1.02–1.26, P=0.0232) but the change in LVEDV was not. In contrast, none of age, sex and medications was associated with progression to HFmrEF (Table 3).
| OR | 95% CI | P value | |
|---|---|---|---|
| 50%<LVEF≤55% | 435.0 | 52.65–10,614 | <0.0001 |
| Age, years | 0.98 | 0.93–1.03 | 0.3696 |
| Female sex | 0.89 | 0.24–3.17 | 0.8577 |
| HF of ischemic etiology | 10.9 | 2.60–74.80 | 0.0007 |
| LVEDD during hosp, mm | 1.14 | 1.04–1.28 | 0.0066 |
| LVESD during hosp, mm | 1.15 | 1.05–1.28 | 0.0018 |
| LVEDV during hosp, ml | 1.04 | 1.01–1.06 | 0.0007 |
| LVEDV change per year, ml | 1.02 | 0.98–1.06 | 0.4224 |
| LVESV during hosp, ml | 1.16 | 1.09–1.28 | <0.0001 |
| LVESV change per year, ml | 1.12 | 1.02–1.26 | 0.0232 |
| ACE inhibitor or ARB at discharge | 0.63 | 0.16–3.10 | 0.5368 |
| β-blocker at discharge | 1.35 | 0.36–4.81 | 0.6442 |
| MR blocker at discharge | 0.88 | 0.13–3.79 | 0.8717 |
| Diuretic at discharge | 1.35 | 0.36–4.81 | 0.6442 |
LVEDV/LVESV change=change between hosp and follow-up. CI, confidence interval; OR, odds ratio. Other abbreviations as in Tables 1,2.
HF is classified simply by LVEF into 2 (HFrEF and HFpEF) or 3 (HFrEF, HF-borderline EF, and HFpEF) categories.3,12,19,20 As for HFpEF, both the European Society of Cardiology (ESC) and the American College of Cardiology Foundation (ACCF)/American Heart Association (AHA) guidelines state that HFpEF is defined as LVEF >50%,18,21 but large clinical trials on HFpEF have enrolled patients with LVEF >40% or 45%. Therefore, the definition of HFpEF is not still strictly fixed, so we used LVEF >50% as the cutoff for HFpEF in the present study. The present study results indicated that approximately 10% of patients with HFpEF at baseline had a decline in LVEF to less than 50% but above 40% after a mean follow-up of 31.5 months. Thus, approximately 10% of patients change from HFpEF to HFmrEF, or HF-borderline EF. It is unclear from the present study whether these patients will further progress to HFrEF over a longer period of time.
The present study found LVEF of 55% as a cutoff for the transition from HFpEF to HFmrEF with high sensitivity and specificity based on ROC curve analysis. Although HFpEF is commonly thought to represent diastolic dysfunction with normal systolic function, through a more sensitive method, LV strain, subtle impairment of LV systolic contractility was recently already demonstrated in some patients with HFpEF.22,23 However, given that normal LVEF as measured is 64.9±4.9%24 by echocardiography and 61% in women and 55% in men by MRI,25 systolic function with LVEF<55% on echocardiography is moderately reduced rather than normal. The ESC guidelines propose that patients with LVEF in the range of 35–50% are in a “grey area” and most likely have primary mild systolic dysfunction.18 However, this “grey area” might be wider.
The clinical syndrome of acute HF diagnosed by Framingham criteria occurs in patients with any level of LVEF. Earlier studies have demonstrated that there is a bimodal distribution of LVEF among patients with acute HF, with a lower proportion of patients with 40%<LVEF≤55%.12,26 Because the present study enrolled only patients with LVEF >50%, LVEF at baseline did not show a bimodal distribution, but in the overall NARA-HF Study 2 there was a similar a bimodal distribution (Figure S1).
The clinical characteristics of patients with 50%≤LVEF<55% were different from those with LVEF >55%. Consistent with prior studies,6,12,13 there was a much higher proportion of patients with ischemic etiology among patients with 50%<LVEF≤55%. Ischemic etiology was a strong predictor for transition from HFpEF to HFmrEF in the present study, as reported previously.12,13 In fact, the rate of LVEF decline was much higher among patients with ischemic etiology than in those with non-ischemic etiology (Figure 3). In addition, in patients with 50%<LVEF≤55%, LVEDV and LVESV during hospitalization were larger than in patients with LVEF >55%, and the percent increment of LVESV between the 2 echocardiography examinations was much greater than that of LVEDV. Thus, decline of LVEF in patients with 50%<LVEF≤55% was probably related to the increase in LVESV. These findings all suggest that there are qualitative differences in the pathophysiology and time course of LV dysfunction between patients with LVEF >55% and those with LVEF ≤55%.
Change in left ventricular ejection fraction (LVEF) from hospitalization to follow-up. The median change (interquartile range) was −1.40% (−3.03 to +2.23) in patients with ischemia and +0.90% (−1.31 to +2.65) in patients with heart failure of non-ischemic etiology (P=0.0174).
Patients whose LVEF had fallen to below 50% at follow-up were not confirmed as having a clinical episode of ischemic disease during follow-up. Moreover, the proportion of readmission for worsening of HF during follow-up was similar in patients with LVEF <50% at follow-up and those with LVEF ≥50% at follow-up (45% and 36%, respectively, P=0.5427). Also, in the univariate logistic regression analysis, readmission for worsening of HF was not a predictor of the decline in LVEF. Therefore, it is unlikely that additional ischemic events or worsening of HF during follow-up was the cause of the decline in LVEF in this study.
Recently, some large randomized clinical trials in HFpEF patients with various therapeutic agents such as angiotensin-receptor blockers (CHARM-preserved, I-preserved),7,27 and mineralocorticoid receptor blocker (TOPCAT),28 failed to show beneficial effects of these drugs in HFpEF, although these agents have been proven to effectively reduce cardiovascular events in HFrEF. Of note, the inclusion criteria was LVEF >40% for the CHARM-preserved study and LVEF >45% for the I-preserved and TOPCAT studies; because a substantial number of patients with “grey area” LVEF were included, further analyses or subanalyses should be conducted with consideration of this.
Study LimitationsThe major limitations are that the sample size was small, the study was retrospective in nature, and based at a single center. Approximately half of potentially eligible subjects were excluded for lack of echocardiography at follow-up, which might be a potential source of bias. Furthermore, we did not collect information on medications after discharge that can potentially affect LVEF. These factors underscore the need for future prospective studies of greater power, ideally controlled for medication regimens, that could further elucidate the natural history of HFpEF.
The present study showed that HFpEF patients with LVEF ≤55% were more likely to progress to HFmrEF in the future than those with LVEF >55%. This finding provides insights to the pathophysiology of HFpEF and suggests that patients with ischemic disease, who show 50%≤LVEF<55%, may actually have HFrEF and not HFpEF. A large-scale prospective study is necessary to confirm this hypothesis.
This work was supported in part by grants-in-aid from the Ministry of Health, Labor, and Welfare of Japan, and Takeda Science Foundation.
Y.S. has conflicts of interest to disclose as follows.
Honoraria: MSD Co, Ltd, Mitsubishi Tanabe Pharma Corporation, Takeda Pharmaceutical Co, Daiichi Sankyo Company Ltd, Otsuka Pharmaceutical Co, Ltd, Pfizer Japan Inc.
Research funding: Japan Heart Foundation, The Naito Foundation Subsidies or Donations: MSD Co, Ltd, Mitsubishi Tanabe Pharma Corporation, Daiichi Sankyo Company Ltd, Takeda Pharmaceutical Co, Ltd, Novartis Pharma K.K., Shionogi & Co, Ltd, Astellas Pharma Inc, AstraZeneca K.K., Ltd, Otsuka Pharmaceutical Co, Ltd, St. Jude Medical Japan Co, Ltd, Kyowa Hakko Kirin Co Ltd.
Endowed departments by commercial entities: MSD Co, Ltd.
Other authors have no financial conflicts of interest to disclose.
Supplementary File 1
Figure S1. Distribution of left ventricular ejection fraction (LVEF).
Please find supplementary file(s);
http://dx.doi.org/10.1253/circj.CJ-15-0425
