The Interrelationship between Ventilatory Inefficiency and Left Ventricular Ejection Fraction in Terms of Cardiovascular Outcomes in Heart Failure Outpatients

The relationship between left ventricular ejection fraction (LVEF) and cardiovascular (CV) outcome is documented in patients with low LVEF. Ventilatory inefficiency is an important prognostic predictor. We hypothesized that the presence of ventilatory inefficiency influences the prognostic predictability of LVEF in heart failure (HF) outpatients. In total, 169 HF outpatients underwent the cardiopulmonary exercise test (CPET) and were followed up for a median of 9.25 years. Subjects were divided into five groups of similar size according to baseline LVEF (≤39%, 40–58%, 59–68%, 69–74%, and ≥75%). The primary endpoints were CV mortality and first HF hospitalization. The Cox proportional hazard model was used for simple and multiple regression analyses to evaluate the interrelationship between LVEF and ventilatory inefficiency (ventilatory equivalent for carbon dioxide (VE/VCO2) at anaerobic threshold (AT) >34.3, optimized cut-point). Only LVEF and VE/VCO2 at AT were significant predictors of major CV events. The lower LVEF subgroup (LVEF ≤ 39%) was associated with an increased risk of CV events, relative to the LVEF ≥75% subgroup, except for patients with ventilatory inefficiency (p = 0.400). In conclusion, ventilatory inefficiency influenced the prognostic predictability of LVEF in reduced LVEF outpatients. Ventilatory inefficiency can be used as a therapeutic target in HF management.


Introduction
Heart failure (HF) is a leading cause of cardiovascular (CV) mortality and hospitalization. Preventing hospitalization in HF patients, such as using a multidisciplinary treatment strategy, has become a great priority for clinicians, researchers, and policymakers [1]. In addition to clinical demographic risk factors, left ventricular ejection fraction (LVEF) determined by echocardiography is the most commonly used parameter for the diagnosis and management of stable chronic HF patients [2,3]. The relationship between LVEF and CV outcome is well documented in patients with low LVEF HF [4]. However, LVEF is less useful as a prognostic indicator when it is > 45% [5,6]. Thus, reliable assessment of prognosis and risk strati cation remain a challenge in HF outpatients across the full spectrum of LVEF.
The cardiopulmonary exercise test (CPET) is a useful tool in all stages of HF patient management, from diagnosis to risk assessment [7]. In the past several decades, the peak oxygen uptake (peak VO2/kg) from CPET was considered as the best predictor of 1-to 3-year event-free survival after HF [8]. In some patients, ventilatory ine ciency during exercise may be a superior predictor of prognosis compared to peak VO2/kg [9,10].
Pulmonary abnormalities, such as impaired lung mechanics and abnormal alveolar-capillary gas exchange, may be caused by respiratory comorbidities or HF itself [11]. In stable HF outpatients, whether the relationship between LVEF and CV outcome is affected by ventilatory ine ciency remains unknown.
In this study, we hypothesized that the presence of ventilatory ine ciency in uences the prognostic predictability of LVEF in stable chronic HF patients.

Subjects
A cohort of 169 HF outpatients with exercise intolerance took the CPET at a tertiary referral center between May 2007 and July 2010. Patients with concurrent signs and symptoms of HF (New York Heart Association functional class II ~ IV) and evidence of structural heart disease (increased left atrial size or left ventricle hypertrophy) were recruited consecutively. Diagnosis was established by the attending physicians. Ischemic cardiomyopathy was de ned as HF with the presence of severe coronary artery disease or a history of myocardial infarction. Valvular cardiomyopathy was de ned as HF caused by primary disease of one of the four heart valves. Dilated cardiomyopathy was de ned as dilation and impaired left ventricle contraction, in which primary and secondary causes of heart disease (e.g., coronary artery disease and myocarditis) were excluded. Patients who had a history of HF hospitalization within 6 months or are unable to perform an exercise test were excluded from the study. The patients were followed up at a median of 9.25 years (interquartile range [IQR], 7.48 ~ 10.32 years) since the administration of CPET. LVEF was assessed by quantitative echocardiography. This study was approved by the Institutional Review Board of the Kaohsiung Chang Gung Memorial Hospital and was conducted in accordance with the Helsinki Declaration of 1975 (as revised in 1983). This study was registered at ClinicalTrials.gov (identi er: NCT04141345). Informed consent was obtained prior to CPET administration in all subjects.

CPET procedures
Patients performed an upright graded bicycle exercise using an individualized protocol. The heart rate was continuously monitored by electrocardiography at rest and during exercise. Blood pressure was measured using an electronic sphygmomanometer every 2 minutes and as needed. The minute ventilation (VE), oxygen consumption (VO2), and carbon dioxide production (VCO2) were continuously recorded every 1 minute using a respiratory mass spectrometer (Vmax Encore, VIASYS, Yorba Linda, CA, USA). Prior to each respiratory gas analysis study, the mass spectrometer was calibrated with a standard gas of known concentration. The peak VO2/kg and the peak respiratory exchange ratio (RER) were de ned as the highest 30-second average value obtained during exercise. The anaerobic threshold (AT) was determined using the V-slope method. The VE/VCO2 at AT was calculated as the average VE/VCO2 for 1 minute during AT and immediately after AT. If the AT could not be determined, the lowest VE/VCO2 was determined by averaging the three lowest consecutive 0.5-minute data points. Since the variability of VE/VCO2 at AT is slightly lower than the variability of the slope of VE versus VCO2 below the ventilatory compensatory point [12,13], this study used VE/VCO2 at AT as a marker of ventilatory e ciency. Spirometric measurements included lung vital capacity, forced vital capacity, forced expiratory volume in 1 second, and maximal voluntary ventilation.
The criteria for discontinuing the test were as follows: request by the subject, threatened arrhythmia, peak RER > 1.1, and ≥2.0 mm of horizontal or downslope ST segment depression during progressive exercise.
The CPET exams were conducted by a quali ed physical therapist under the supervision of a physician.

Outcome analysis
De ned time-dependent CV outcomes included CV mortality and rst HF hospitalization, which were the primary endpoints of the analysis. Study subjects were followed up until the end of 2018. HF hospitalization was de ned as an unplanned hospitalization due to new or worsening HF requiring the use of intravenous diuretics, inotropes, or vasodilators.

Statistical analyses
Subjects were divided into ve groups of similar size according to baseline LVEF (≤39%, 40-58%, 59-68%, 69-74%, and ≥75%) to evaluate the relationship between LVEF and CV outcomes. Comparisons between LVEF groups were analyzed using Pearson's chi-square test or Fisher's exact test for categorical variables. Continuous variables were expressed as median (IQR). Comparisons between LVEF groups were analyzed using the Kruskal-Wallis test and multiple comparisons for continuous variables. The Kolmogorov-Smirnov test was used to test for normality. For the univariate and multivariable analyses, the hazard ratio and 95% con dence interval were computed using the Cox proportional hazard model. The primary endpoint was de ned as CV mortality or the rst HF hospitalization. The various CPET parameters were evaluated as predictors of primary endpoints by performing time-dependent receiver operating characteristic curve (ROC) analyses. Optimized threshold values for VE/VCO2 at AT were identi ed via ROC analysis and the Youden index. The Cox proportional hazard model was used for simple and multiple regression analyses to evaluate the interrelationship between LVEF and ventilatory ine ciency (de ned as VE/VCO2 at AT > 34.3, optimized cutoff point). The interaction term "ventilatory ine ciency multiplied by LVEF category" was introduced to the previous model. Data were analyzed using R v3.6.1 software using "time ROC" and "survival" package and SPSS 22.0 (SPSS Inc., Chicago, IL, USA). In all analyses, a p value less than 0.05 was considered statistically signi cant.

Results
The mean LVEF in our HF outpatients was 64.0 ± 18.6%. The baseline clinical demographic and pharmacological characteristics according to LVEF are shown in Table I. Patients with higher ejection fraction (EF) were more often female and more likely to have a history of hypertension. Patients with lower EF were more likely to have a smoking history, have received coronary intervention, and have ischemic cardiomyopathy. Patients who suffered from dilated cardiomyopathy had lower EF. The incidence of diabetes, valvular heart disease, and ischemic stroke did not differ across these LVEF subgroups. The distribution of age also did not differ signi cantly across the LVEF subgroups. The proportion of patients who received beta-blockers, angiotensin-converting enzyme inhibitors (ACEIs), angiotensin-receptor blockers, loop diuretics, and mineralocorticoid receptor antagonists (MRAs) increased in the lower EF patients. In contrast, the proportion of patients who received dihydropyridine calcium channel blockers increased in the higher EF patients. The CPET parameters including peak VO2/kg, AT, and VE/VCO2 at AT had a signi cant difference across the spectrum of LVEF (Table 1). Within a median follow-up period of 9.25 years (IQR, 7.48 ~ 10.32 years), 49 patients had achieved our primary endpoints. The relationship between LVEF and the major cardiac events, including CV mortality, is shown in Table 2. The risk of major CV events and CV mortality was increased in the lower LVEF subgroups (p = 0.002 and 0.001, respectively). Table III shows that, according to the univariate Cox regression analysis, the signi cant predictors of major CV events included comorbid lung disease, diabetes, smoking, LVEF, dilated cardiomyopathy, and treatment with beta-blockers, loop diuretics, or MRAs. The CPET parameters including VE/VCO2 at AT, ΔVO2/ΔWR, peak O2 pulse, peak VO2, peak VO2/kg, peak work, and AT were signi cant predictors for major CV events based on the univariate analysis. In the multivariate Cox regression analyses, only LVEF and VE/VCO2 at AT were found to be signi cant predictors of major CV events in our cohort study (   HR: Hazard ratio, CI: con dence interval, CPET: cardiac pulmonary exercise test, ACEI: angiotensinconverting enzyme inhibitor, ARB: angiotensin receptor blocker, DHP: dihydropyridine, VE/VCO 2 at AT: ventilatory equivalent for carbon dioxide at anaerobic threshold, ΔVO2/ΔWR: the ratio of increase in oxygen uptake to increase in work rate, peak VO2: peak oxygen consumption, RER: respiratory exchange ratio, VE: minute ventilation, VO2/kg: oxygen consumption per kilogram, AT: anaerobic threshold.
As presented in Fig. 1, the relationship between LVEF and major CV events was not linear. We de ned ventilatory ine ciency as VE/VCO2 at AT > 34.3. To characterize the relationship between LVEF and the risk of CV mortality or HF hospitalization among patients with ventilatory ine ciency, subjects were divided into ve subgroups according to baseline LVEF. Figure 2 shows the relationship between LVEF and major CV events in patients with ventilatory ine ciency (VE/VCO2 at AT > 34. 3

Discussion
In chronic HF outpatients followed for a median of 9.25 years, LVEF and VE/VCO2 at AT were both found to be signi cant independent predictors of increased risk of CV mortality or HF hospitalization. LVEF was a poor predictor in patients with ventilatory ine ciency and in those with LVEF > 40%. Although our study showed that the interaction effect between LVEF and VE/VCO2 at AT was not signi cant, the prognostic predictability of LVEF was decreased in the HF with reduced LVEF (HFrEF, LVEF ≤ 39%) population in the ventilatory ine ciency group. As demonstrated in the CHARM Program [5], the relationship between LVEF and CV outcomes is not linear. We also demonstrated a similar nding in chronic HF outpatients. This relationship was further diminished in the ventilatory ine ciency group. This phenomenon revealed that HF with preserve LVEF (HFpEF, LVEF ≥ 50%) patients who had ventilatory ine ciency had similar CV outcomes as that of their HFrEF counterpart.
This study showed that the ventilation e ciency variable, in addition to LVEF, was a signi cant prognostic predictor in HF outpatients. Ventilatory ine ciency re ects the adverse effects of HF on lung mechanics and diffusion capacity [14]. An HF also augments ventilatory drive and increases hemodynamic demand associated with breathing work [15]. Ergoreceptors stimulate ventilation and activate sympathetic hormones in response to work. The ergore ex in the muscle also affects ventilatory effort. In response to carbon dioxide and pulmonary J receptors (which likely respond to congestion and alveolar stiffness), central and pulmonary chemoreceptors contribute to the ergore ex and result in excess ventilation [16]. In HF patients, a high ventilatory drive can reduce the partial pressure of CO 2 (PaCO2) [17]. Consequently, a reduced PaCO2 and increased fractional dead space cause abnormally high VE/VCO2 at AT, i.e., ventilatory ine ciency [18,19].
The mechanism of ventilatory ine ciency in uences the outcomes of HF patients differently between the HFrEF and HFpEF patients. A study analyzed the ventilatory ine ciency between 24 HFrEF patients and 33 HFpEF patients [20]. It demonstrated the loss of cardiac output augmentation related to ventilatory ine ciency regardless of LVEF; however, lung congestion parameters (echocardiographic parameter: e' and E/e') correlated with ventilatory ine ciency only in HFpEF. In another study, ventilatory ine ciency appears to be in uenced by mechanisms regulating PaCO2 in HFrEF. In contrast, dead space to tidal volume ratio (VD/VT) plays a more important role in developing ventilatory ine ciency in HFpEF [21].
HFpEF and HFrEF may be two distinct entities in terms of ventilatory response to exercise; this study provides evidence that ventilatory ine ciency plays a critical role in HFpEF.
CPET-based measurements of ventilatory ine ciency provide unique physiologic information clinically relevant to contemporary treatment for HF. Several therapeutic interventions for HF affect ventilatory abnormalities both at rest and during exercise. For example, ACEI improves pulmonary diffusion, removes interstitial uid, and improves pulmonary hemodynamic status [22]. Carvedilol, but not bisoprolol, improves ventilatory e ciency during exercise (reduction of VE/VCO2 slope and increase in maximum end-tidal CO 2 pressure) [23]. Carvedilol may have direct effects on respiratory chemoreceptor activity based on the CARNEBI trial [24]. As ventilatory ine ciency is a signi cant prognostic predictor across the spectrum of LVEF, we should consider ventilatory abnormalities during exercise as therapeutic targets and treat them accordingly. Therapeutic interventions such as rehabilitation training (isolated quadriceps training) [25], device-guided paced breathing [26], yoga mantras [27], and reduction of afferent stimuli from ergopulmonary and cardiopulmonary receptors [28,29] might all alleviate ventilatory ine ciency.
The use of CPET-derived variables to guide therapy and improve outcome deserves further investigation.
This study has some limitations. First, the sample size was relatively small compared to those in other epidemiological studies. However, our study had a longer follow-up period than those of previous works. Second, patients were only recruited from outpatient clinics, which may have caused selection bias. The ndings of this study may need further validation in other populations of patients with HF. Third, this study did not analyze other CPET variables that have been used to predict HF outcomes, e.g., oscillatory ventilation, end-tidal CO 2 pressure, VO2 kinetics during exercise, oxygen uptake e ciency slope, and heart rate recovery. Therefore, whether the predictive accuracy of these variables can be increased by combining them with VE/VCO2 at AT requires further investigation. The relationship between LVEF and CV outcomes in all patients. This relationship was not linear. The lower LVEF subgroup (LVEF 39%) was associated with a signi cantly increased risk of CV mortality or HF hospitalization relative to the LVEF 75% subgroup. (p=0.002) Abbreviation: LVEF: left ventricular ejection fraction, CV: cardiovascular, HF: heart failure