2019 Volume 83 Issue 8 Pages 1718-1725
Background: The cardiopulmonary exercise test (CPX) is a tool for evaluating disease severity and limitations in activities of daily living in patients with cardiac disorders. However, few studies have evaluated the association between exercise oscillatory ventilation (EOV) severity and prognosis in heart failure (HF) patients with EOV. EOV severity can be evaluated by detecting endtidal CO2 pressure (PETCO2, an indicator of the arterial partial pressure of CO2 (PaCO2)) and minute ventilation, which is a reflection of the respiratory response to elevated CO2. We hypothesized that the magnitude of EOV severity can predict the severity and prognosis of cardiac disorders and aimed to validate this hypothesis.
Methods and Results: In total, 2,043 patients who underwent symptom-limited maximal CPX between 2010 and 2016 were evaluated. We enrolled 70 patients who had HF with reduced ejection fraction (HFrEF) and EOV. The endpoint was cardiovascular death. During a median follow-up of 4.3 years, 34 participants died (48%). Those who died showed significantly larger EOV loop size and lower hemoglobin (Hb) levels than those who survived (17.3±7.0 cm2 vs. 12.8±6.1 cm2, P<0.001; 12.2±1.2 g/dL vs. 13.2±2.9 g/dL, P=0.004). Cox regression analyses revealed Hb levels and EOV loop size as independent predictors of cardiovascular death in HFrEF patients with EOV.
Conclusions: EOV loop size was associated with cardiovascular death of HFrEF patients with EOV.
Parameters obtained from the cardiopulmonary exercise test (CPX) are known to reflect the severity of heart failure (HF) in cardiac patients.1 One of the key parameters of the CPX that makes it a useful clinical tool for evaluating disease severity and limitations of activities of daily living in patients with cardiac disorders is exercise oscillatory ventilation (EOV), which is cyclic fluctuation of minute ventilation (V̇E) and expired gas kinetics during exercise. Kato et al reported that the cycle length of EOV is closely related to impaired cardiac reserve during exercise in patients with cardiac disorders.2 A meta-analysis of studies reporting hazard ratios (HRs) for cardiovascular events demonstrated that HF patients with EOV have a 4-fold higher risk for adverse events than HF patients without EOV.3 EOV is now considered a marker of disease severity and worse prognosis in HF.3,4 Despite the clear association between EOV and outcomes in HF, there are limited data regarding the mechanistic basis for EOV.5 Moreover, only a few studies have evaluated the association between EOV severity and prognosis of HF patients with EOV.6
EOV occurs because of delayed responses to elevated partial pressure of arterial carbon dioxide (PaCO2) and exaggerated ventilatory responses.6 Hypoperfusion of the lung caused by reduced pulmonary blood flow in patients with left ventricular (LV) dysfunction exacerbates the V/Q mismatch, leading to progressively lower endtidal CO2 pressures (PETCO2).7 Thus, EOV severity can be evaluated by detecting PETCO2 and V̇E, which indicate reduced pulmonary blood flow in patients with LV dysfunction, exacerbating the V/Q mismatch. During CPX, EOV appears as a loop (EOV loop) formed by graphing the 2 parameters, PETCO2 and V̇E. Hence, we hypothesized that the size of this loop during CPX would be larger in patients with severe cardiac disorders and would be associated with prognosis. Consequently, we investigated the correlation between area of the EOV loop and patient prognosis.
This cohort was retrieved from the Gunma Prefectural Cardiovascular Centre Database, established in 2002 for patients who newly visited the hospital. A total of 2,043 patients who underwent symptom-limited maximal CPX between 2010 and 2016 were analyzed. Inclusion criteria were: previous or present HF symptoms (NYHA functional class II–III, ACC/AHA classification stage C), documentation of LV systolic dysfunction (LV ejection fraction (LVEF) <40%), stable clinical condition with unchanged medications for ≥3 months, able to undergo CPX, no major cardiovascular treatment or intervention scheduled, and presence of EOV, defined as ≥3 consecutive cyclic fluctuations of ventilation during CPX, as described by Sun et al.8 To be defined as positive for EOV, the amplitude of EOV must exceed 30% of concurrent mean ventilation with a complete oscillatory cycle within 40–140 s. Oscillations of similar frequency must also be visible in ≥3 of the following variables: oxygen pulse, oxygen uptake (V̇O2), carbon dioxide output (V̇CO2), V̇E/V̇CO2, respiratory exchange ratio (R), or endtidal O2 pressures (PETO2) and endtidal CO2 pressures (PETCO2). The exclusion criteria were history of pulmonary embolism, moderate to severe aortic and mitral stenoses, pericardial disease, severe obstructive lung disease, exercise-induced angina and significant ECG alterations, or the presence of any clinical comorbidity interfering with exercise performance. Based on these criteria, we enrolled 70 patients with HF with reduced EF (HFrEF) and EOV.
History and the results of the following examinations were recorded for all patients. (1) Blood sampling performed after a 12-h overnight fast to analyze total cholesterol, hemoglobin (Hb), percentage of lymphocytes, serum sodium, uric acid (UA), and B-type natriuretic peptide (BNP). Estimated glomerular filtration rate (eGFR) levels were calculated using the Japanese Society of Nephrology formulae (194×serum creatinine−1.094×age−0.287 for men and 194×serum creatinine−1.094×age−0.287×0.739 for women).9 (2) Blood pressure was measured after a 10-min rest while seated and expressed as the average of 3 consecutive measurements of each arm using the conventional cuff method. (3) Echocardiography.
The endpoint was cardiovascular death using the following definition: any fatal event related to coronary artery disease, cerebrovascular disease and other heart or vascular diseases, specifically fatal myocardial infarction, acute coronary syndrome, stroke, transient ischemic attack, HF and arrhythmia.10 The time interval from the date of CPX to cardiovascular death was defined as the duration of follow-up. Endpoints were censored in October 2017. The follow-up rate was 100%.
CPXThe anaerobic threshold (AT) and peak V̇O2 were evaluated using symptom-limited CPX on an upright cycle ergometer (StrengthErgo8, Mitsubishi Electric Engineering, Tokyo, Japan) with an ECG machine (ML-9000, Fukuda Denshi, Ltd., Tokyo, Japan). CPX was performed 2–4 h after a light meal. The tests began as per recommendations by Buchfuhrer et al11 with 3 min of rest and a 3-min warm-up period at 0 W, followed by a continuous increase in the work rate (WR) by 1 W/6 s until exhaustion. The criteria for halting exercise testing in this study are outlined in the American College of Sports Medicine guidelines.12 The increases in WR were based on the ability of patients to perform the exercises in a period between 8 and 15 min.11 We measured V̇O2, V̇CO2, and V̇E on a breath-by-breath basis using a gas analyzer (MINATO 300S, Minato Science Co., Ltd., Osaka, Japan). Peak V̇O2 was determined at the highest WR during exercise. The AT was measured by V-slope method.13 The respiratory compensation point (RCP) was recorded when an increase in the V̇E/V̇CO2 and decrease in PETCO2 occurred simultaneously.14,15 The predicted peak V̇O2 (% peak V̇O2) and AT (% AT) were based on a normal Japanese population.16
Evaluation of EOV Loop SizeWe measured the area of the largest loop produced by V̇E and PETCO2, using the following elliptic formula: long side length (V̇E; 0.14 cm/L/min)×short side length (PETCO2; cm/5 mmHg)×π (Figure 1A).
(A) Representative EOV loop. The graph shows counterclockwise rotation during CPX in an elliptical shape (Left). This can be mimicked (e.g., Right). The area of the largest loop was calculated. (B) ROC curve of EOV loop size as a predictor of cardiovascular death. The blue line represents the EOV loop size, and the red line represents the reference. CPX, cardiopulmonary exercise test; EOV, exercise oscillatory ventilation; PETCO2, endtidal CO2 pressure; ROC, receiver-operating characteristic; VE, ventilatory equivalent.
The cycle length of the EOV’s V̇E was calculated as the interval from the peak to the following peak of the EOV’s V̇E for each of the cycles noted at rest, and then expressed as the mean value.2 The amplitude of the EOV’s V̇E was calculated as the difference between the peak and nadir of the EOV’s V̇E for each of the cycles noted at rest, and then expressed as a mean value.2 The product of the EOV’s V̇E cycle length and amplitude was calculated from the EOV’s V̇E cycle length and amplitude for each of the cycles noted at rest, and then expressed as a mean value.
Statistical AnalysisParticipants’ demographics (i.e., age, body mass index (BMI)), and other cardiovascular risk factors (i.e., frequency of lymphocytes, Hb, Na, UA, total cholesterol), biomarker (BNP) levels, and CPX parameters (i.e., EOV loop size, V̇O2, AT, peak V̇O2, V̇E vs. V̇CO2 slope, load, PETCO2, and peak O2 pulse) were compared between those who survived the follow-up period and those who died. The distribution of each continuous variable was tested for normality using the Shapiro-Wilk test and expressed as the mean±standard deviation (SD). Variables with a skewed distribution are expressed as median value [interquartile range]. Statistical analysis was performed using the unpaired t-test or Mann-Whitney test for continuous variables and the chi-square test for categorical variables to compare variables between 2 groups (patients who died or survived). To reveal an association with cardiovascular death, univariate and multivariate Cox regression analyses were applied to examine the potential factors reported previously, including Na, eGFR, % peak V̇O2, V̇E vs. V̇CO2 slope, EF,17,18 and BNP.19 We did not use heart rate at peak work rate in the multivariate analyses, even though there was a significant difference between the 2 groups, because we chose our explanatory variables based on experience in specialized fields and previous literature.17–19
A receiver-operating characteristic (ROC curve) was used to define the sensitivity and specificity of EOV loop size for predicting cardiovascular death, and the area under the curve was calculated. Survival analyses using Kaplan-Meier models were also performed using the Statistical Package for Social Sciences 21.0J for Mac (IBM Corp., Armonk, NY, USA). Statistical significance was set at a two-sided P-value <0.05.
Ethics StatementThis study was approved by the Ethics Committee of Gunma Prefectural Cardiovascular Centre (approval no. 29025) and conducted in accordance with the Declaration of Helsinki. We obtained informed consent from all registered patients after being provided a description of the study requirements.
Table 1 shows the baseline characteristics of the 2 groups of patients: cardiovascular death and survival during follow-up (median 4.3 years). Cardiovascular death occurred in 34 participants (48%): 26 (37%) died of HF, and 8 (11%) sudden deaths caused by arrhythmia. The mean age and BMI were not significantly different between groups. Participants who died had lower Hb and lymphocyte count and higher BNP levels (12.2±1.2 g/dL vs. 13.2±2.9 g/dL, P=0.004; 23.9±8.1% vs. 30.4±8.5%, P=0.04; 763.0 [682.5, 1,091.3] pg/mL vs. 314.5 [218.5, 599.8] pg/mL, P<0.001, respectively); in contrast, there were no significant differences in the levels of Na, UA or total cholesterol.
| Death (n=34) |
Survival (n=36) |
P value | |
|---|---|---|---|
| Clinical characteristics | |||
| Sex (M/F) | 32/2 | 33/3 | 0.70 |
| Age (years) | 71.5±7.1 | 67.6±10.2 | 0.31 |
| Height (cm) | 164.7±8.9 | 164.9±8.1 | 0.91 |
| Weight (kg) | 60.1±15.1 | 61.8±12.7 | 0.45 |
| BMI (kg/m2) | 21.8±3.9 | 22.7±3.7 | 0.29 |
| LVEF (%) | 34.2±12.8 | 37.0±15.6 | 0.42 |
| Hb level (g/dL) | 12.2±1.2 | 13.2±2.9 | 0.004 |
| Lymphocyte count (%) | 23.9±8.1 | 30.4±8.5 | 0.04 |
| Sodium (mEq/L) | 137.6±4.3 | 137.9±2.5 | 0.75 |
| Uric acid (mg/dL) | 5.7±1.6 | 6.2±2.3 | 0.88 |
| Total cholesterol (mg/dL) | 167.0 [159.5, 181.0] | 158.5 [156.0, 164.8] | 0.15 |
| eGFR (mL/min/1.73 m2) | 40.4±17.1 | 54.9±16.6 | 0.09 |
| BNP level (pg/mL) | 763.0 [682.5, 1,091.3] | 314.5 [218.5, 599.8] | <0.001 |
| NYHA class II/III | 2/32 | 5/31 | 0.42 |
| Etiology of HF | |||
| Ischemic/nonischemic | 14/20 | 11/25 | 0.45 |
| Medical history | |||
| Hypertension | 8 (23%) | 11 (30%) | 0.59 |
| Diabetes mellitus | 6 (17%) | 9 (25%) | 0.56 |
| Medical therapy | |||
| β-blocker | 42 (79%) | 28 (80%) | 0.93 |
| ACEI/ARB | 36 (67%) | 25 (71%) | 0.81 |
| Spironolactone | 21 (40%) | 15 (42%) | 0.75 |
Data are mean±SD or n (%). ACEI, angiotensin-converting enzyme inhibitor; ARB, angiotensin-receptor blocker; BMI, body mass index; BNP, B-type natriuretic peptide; eGFR, estimated glomerular filtration rate; Hb, hemoglobin; HF, heart failure; LVEF, left ventricular ejection fraction; NYHA, New York Heart Association; SD, standard deviation.
There were no significant differences between groups for NYHA classification, prevalence of hypertension and diabetes, or etiology of HF. Regarding the echocardiography findings, there was no significant difference in LVEF between groups. There were no differences between groups in the proportion of patients administered cardioprotective drugs such as β-blockers.
Comparison of CPX ParametersThe CPX parameters of each group are shown in Table 2. There were no significant differences in systolic blood pressure at rest and peak, or in heart rate at rest during CPX. In contrast, heart rate at peak exercise was significantly lower in the nonsurvivors group (80.5 [77.8, 97.0] vs. 108.0 [91.0, 117.3] beats/min, P<0.001).
| Death (n=34) |
Survival (n=36) |
P value | |
|---|---|---|---|
| Systolic BP at rest (mmHg) | 106.5±19.8 | 103.6±19.8 | 0.73 |
| Systolic BP at peak (mmHg) | 115.7±22.9 | 128.6±20.7 | 0.11 |
| Heart rate at rest (beats/min) | 72.0 [69.3, 78.3] | 70.5 [64.0, 77.2] | 0.36 |
| Heart rate at peak (beats/min) | 80.5 [77.8, 97.0] | 108.0 [91.5, 117.3] | <0.001 |
| Rest V̇O2 (mL/min/kg) | 3.5±0.7 | 3.7±0.5 | 0.51 |
| AT (mL/min/kg) | 8.0±1.7 | 8.9±1.4 | 0.002 |
| Peak V̇O2 (mL/min/kg) | 10.3±1.8 | 12.3±1.9 | <0.001 |
| % AT (%) | 49.9±10.3 | 56.4±8.6 | 0.001 |
| % Peak (%) | 45.9±10.8 | 54.7±13.2 | 0.002 |
| V̇E vs. V̇CO2 slope | 50.2 [44.1, 72.9] | 40.8 [37.0, 50.0] | <0.001 |
| Nadir V̇E/V̇CO2 | 54.4 [44.9, 56.3] | 44.5 [39.7, 51.3] | <0.001 |
| Peak LOAD (watts) | 57.0 [39.3, 61.0] | 58.0 [45.0, 71.8] | 0.004 |
| Maximum PETCO2 (mmHg) | 32.0±3.8 | 32.6±4.5 | 0.29 |
| Peak O2 pulse (mL/beat) | 7.2±2.1 | 7.3±2.1 | 0.82 |
| EOV loop size (cm2) | 8.1±4.0 | 2.2±1.8 | <0.001 |
| EOV V̇E cycle length (min) | 1.5±0.4 | 1.3±0.3 | 0.02 |
| EOV V̇E amplitude (L/min) | 17.3±7.0 | 12.8±6.1 | 0.007 |
Data are mean±SD. AT, anaerobic threshold; BP, blood pressure; EOV, exercise oscillatory ventilation; O2 pulse, oxygen pulse; Peak V̇O2, oxygen uptake at peak work rate; PETCO2, endtidal carbon dioxide pressure; Rest V̇O2, oxygen uptake at peak work rate; RCP, respiratory compensation point; SD, standard deviation; V̇E, minute ventilation; V̇CO2, carbon dioxide production.
There were no significant differences in V̇O2 at rest; however, significant differences in AT and peak V̇O2 were observed. Parameters indicating cardiac function during exercise such as V̇E vs. V̇CO2 slope were significantly different between groups (50.2 [44.1, 72.9] vs. 40.8 [37.0, 50.0], P<0.001). Finally, the EOV loop size, V̇E cycle length and V̇E amplitude were larger in the nonsurvivors group than in the survivors group (8.1±4.0 cm2 vs. 2.2±1.8 cm2, P<0.001, 1.5±0.4 min vs. 1.3±0.3 min, P=0.02, 17.3±7.0 L/min vs. 12.6±6.1 L/min, P=0.007).
Predictors of Cardiovascular DeathWe performed Cox regression analysis, examining death and previously reported risk factors for cardiovascular death.17–19 In the univariate analysis, several factors were significantly predictive of cardiovascular death: % peak V̇O2 (HR 0.954; 95% confidence interval (CI) 0.926–0.984; P=0.002), Hb (HR 0.707; 95% CI 0.582–0.859; P<0.001), V̇E vs. V̇CO2 slope (HR 1.037; 95% CI 1.021–1.053; P<0.001), and EOV loop size (HR 1.24; 95% CI 1.152–1.342; P<0.001). The other risk factors were not predictive of cardiovascular death in this study population (Table 3).
| Variable | Univariate | Multivariate | ||||
|---|---|---|---|---|---|---|
| HR | 95% CI | P value | HR | 95% CI | P value | |
| % peak V̇O2 (%) | 0.954 | 0.926–0.984 | 0.002 | |||
| Hb level (g/dL) | 0.707 | 0.582–0.858 | <0.001 | 0.622 | 0.455–0.849 | 0.002 |
| V̇E vs. V̇CO2 slope | 1.037 | 1.021–1.053 | <0.001 | |||
| Sodium (mEq/L) | 0.968 | 0.899–1.054 | 0.450 | |||
| LVEF (%) | 0.997 | 0.969–1.026 | 0.860 | |||
| eGFR (mL/min/1.73 m2) | 0.976 | 0.953–1.001 | 0.059 | |||
| BNP (pg/mL) | 0.990 | 0.989–1.012 | 0.630 | |||
| EOV loop size (cm2) | 1.243 | 1.152–1.342 | <0.001 | 1.227 | 1.135–1.327 | <0.001 |
CI, confidence interval; HR; hazard ratio; % peak V̇O2, predicted value of V̇O2. Other abbreviations as in Tables 1,2.
We confirmed the onset of cardiovascular death using multivariate analysis. Multivariate analysis revealed that EOV loop size (HR 1.227; 95% CI 1.135–1.327; P<0.001) and Hb (HR 0.622; 95% CI 0.455–0.849; P=0.002) were independent predictors of cardiovascular death (Table 3).
We also performed univariate and multivariate Cox regression analyses that changed the explanatory variables from EOV loop size to V̇E cycle length, V̇E amplitude, and the product of V̇E cycle length and amplitude. However, none of these values were independent predictors of cardiovascular death (Tables 4–6).
| Variable | Univariate | Multivariate | ||||
|---|---|---|---|---|---|---|
| HR | 95% CI | P value | HR | 95% CI | P value | |
| % Peak V̇O2 (%) | 0.954 | 0.926–0.984 | 0.002 | |||
| Hb level (g/dL) | 0.707 | 0.582–0.858 | <0.001 | 0.669 | 0.498–0.899 | 0.007 |
| V̇E vs. V̇CO2 slope | 1.037 | 1.021–1.053 | <0.001 | 1.04 | 1.001–1.08 | 0.014 |
| Sodium (mEq/L) | 0.968 | 0.899–1.054 | 0.450 | |||
| LVEF (%) | 0.997 | 0.969–1.026 | 0.860 | |||
| eGFR (mL/min/1.73 m2) | 0.976 | 0.953–1.001 | 0.059 | |||
| BNP (pg/mL) | 0.990 | 0.989–1.012 | 0.630 | |||
| EOV V̇E cycle length (min) | 2.890 | 1.12–7.01 | 0.018 | |||
Abbreviations as in Tables 1–3.
| Variable | Univariate | Multivariate | ||||
|---|---|---|---|---|---|---|
| HR | 95% CI | P value | HR | 95% CI | P value | |
| % Peak V̇O2 (%) | 0.954 | 0.926–0.984 | 0.002 | |||
| Hb level (g/dL) | 0.707 | 0.582–0.858 | <0.001 | 0.644 | 0.479–0.864 | 0.003 |
| V̇E vs. V̇CO2 slope | 1.037 | 1.021–1.053 | <0.001 | 1.028 | 1.001–1.049 | 0.014 |
| Sodium (mEq/L) | 0.968 | 0.899–1.054 | 0.450 | |||
| LVEF (%) | 0.997 | 0.969–1.026 | 0.860 | |||
| eGFR (mL/min/1.73 m2) | 0.976 | 0.953–1.001 | 0.059 | |||
| BNP (pg/mL) | 0.992 | 0.989–1.012 | 0.630 | |||
| EOV V̇E amplitude (L/min) | 1.070 | 1.024–1.118 | 0.002 | |||
Abbreviations as in Tables 1–3.
| Variable | Univariate | Multivariate | ||||
|---|---|---|---|---|---|---|
| HR | 95% CI | P value | HR | 95% CI | P value | |
| % Peak V̇O2 (%) | 0.954 | 0.926–0.984 | 0.002 | |||
| Hb level (g/dL) | 0.707 | 0.582–0.858 | <0.001 | 0.643 | 0.479–0.864 | 0.003 |
| V̇E vs. V̇CO2 slope | 1.037 | 1.021–1.053 | <0.001 | |||
| Sodium (mEq/L) | 0.968 | 0.899–1.054 | 0.450 | |||
| LVEF (%) | 0.997 | 0.969–1.026 | 0.860 | |||
| eGFR (mL/min/1.73 m2) | 0.976 | 0.953–1.001 | 0.059 | |||
| BNP (pg/mL) | 0.992 | 0.989–1.012 | 0.630 | |||
| Product of V̇E cycle length and amplitude | 1.070 | 1.025–1.117 | 0.002 | |||
Abbreviations as in Tables 1–3.
Based on ROC curve analysis, the optimal cutoff for EOV loop size to predict cardiovascular death was ≥3.47 cm2, with a specificity and sensitivity of 87.1% and 86.7%, respectively (area under the curve=0.915; 95% CI 0.881–0.950) (Figure 1B).
Long-Term Follow-up and New-Onset Cardiovascular DeathThe event-free survival curve was also evaluated using the optimal EOV loop size value (3.47 cm2). Results indicated that individuals with larger (≥3.47 cm2) EOV loop size had a significantly higher risk for cardiovascular death than individuals with smaller EOV loop size (<3.47; P<0.001) (Figure 2).
Kaplan-Meier survival curve of cardiovascular death by EOV loop size. The red line represents EOV loop ≥3.47, and the blue line represents EOV loop <3.47 cm2. EOV, exercise oscillatory ventilation.
The present study found that patients with a larger EOV loop size had a worse prognosis and greater severity of HF. In addition, Hb was also an important factor for HF prognosis. As far as we know, this is the first study to use the EOV loop size as a prognostic parameter. There is limited information on the mechanistic basis for EOV. Putative mechanisms of EOV have been largely extrapolated from studies of periodic breathing (PB) at rest and during sleep,20,21 despite the limited overlap between EOV and PB during sleep.22
Ventilation is regulated through a feedback loop between pulmonary gas-exchanging capillaries and chemoreceptors in the carotid bodies (peripheral) or the medulla (central).23,24 Ventilatory instability and ventilation oscillations may arise from (1) delay in information transfer (i.e., increased circulation time because of reduced cardiac index),25 (2) increase in controller gain such as increased chemosensitivity to PaCO2 and PaO2,26,27 or (3) reduction in system damping, such as baroreflex impairment.26
In the current study, we measured the largest loop size produced by V̇E and PETCO2 during CPX. As previously noted, one of the mechanisms causing EOV is increased carotid and aortic chemoreceptor sensitivity to minimal changes in arterial O2 and CO2.26,27 These changes may contribute to sympathetic nerve function during exercise, which, in turn, leads to excessive and irregular ventilation during exercise. Enhanced hypoxic and central hyperpneic chemosensitivity may cause an increased ventilation response (V̇E/V̇CO2) to exercise in HF patients. Here, the loop size, but neither the amplitude nor the cycle length, was used to determine the severity of EOV.
CPX is routinely used in the prognostic evaluation of patients with HFrEF, in whom the prognostic value of peak V̇O2, V̇E vs. V̇CO2 slope and the presence of EOV is powerful and well established.28–30
However, in the current study, only EOV loop size was found to be an independent factor for cardiovascular death among these 3 valuables. This is because this study was limited to patients with EOV. Because patients with EOV have a poor prognosis, as described earlier, the patients included in this study already had a poor prognosis. As we mentioned, peak V̇O2 is a prognostic indicator in HFrEF patients. There were significant differences between the survivor and nonsurvivor groups; however, both groups had low values. This would explain why the peak V̇O2 was not selected as a prognostic predictor by our statistical analysis and we can make the same conclusion regarding V̇E vs. V̇CO2 slope. In addition, it is known that in patients with EOV, the V̇E vs. V̇CO2 drawn slope consists of distorted lines rather than straight lines, so it cannot be measured accurately. For these reasons, peak V̇O2 and V̇E vs. V̇CO2 slope are not useful for predicting the prognosis of patients with EOV.
In a previous study, Hb was also an important factor in predicting a poor prognosis for HF patients.31 The presence of anemia has been reported as an independent risk factor for higher all-cause mortality and hospitalization rates.31–33 Thus, it is not surprising that our study confirmed these established results.
There were only 70 (3.4%) HF patients with EOV among 2,043 patients who underwent CPX. Thus, the present result applied only to the small number of HF patients presenting with EOV. There is very little data on assessing the severity of EOV, and this study showed it could be useful to assess not only prognosis but also as an evaluation of HF treatment. For example, one could determine the dose of cardioprotective drugs such as β-blockers.
Study LimitationsFirst, our sample size and number of outcomes were small. Second, it was a retrospective, single-center study. Therefore, the possibility of unintentional bias in the selection of patients cannot be fully excluded. Third, the EOV loop does not close completely in a single beat but conforms to a sequence of consecutive open loops. During exercise testing, we found that the biggest loop mimicked an ellipse. Finally, calculation of the EOV loop size is time consuming. Despite the difficulty in measuring the EOV loop size at the present time, it would be possible to develop a computing system to measure it accurately and quickly. In the future, we would like to focus on developing such a system.
It is important to perform CPX before treating HF patients. Among patients with EOV, the loop size may be associated with a worse prognosis.
We thank our lecturer, Haruyasu Fujita (Department of Public Health, Gunma University, Gunma, Japan) for his valuable statistical advice during the preparation of this report. We also thank Editage (www.editage.jp) for English language editing.
T.N. led the study as the principal investigator, wrote the statistical analysis plan, drafted, and revised the paper. H.A. wrote the statistical analysis plan and cleaned and analyzed the data. M.M. and S.O. collected data. All authors approved the final version of the manuscript.
This research received no grant from any funding agency in the public, commercial, or not-for-profit sectors.
The authors declare no conflicts of interest.
In addition to research expenses, instruments, and medicines, there was no support for helping to smoothly carry out the research described.
