Incremental diagnostic value of post-exercise lung congestion in heart failure with preserved ejection fraction

Abstract

Aims

Lung ultrasound (LUS) may unmask occult heart failure with preserved ejection fraction (HFpEF) by demonstrating an increase in extravascular lung water (EVLW) during exercise. Here, we sought to examine the dynamic changes in ultrasound B-lines during exercise to identify the optimal timeframe for HFpEF diagnosis.

Methods and results

Patients with HFpEF (n = 134) and those without HF (controls, n = 121) underwent a combination of exercise stress echocardiography and LUS with simultaneous expired gas analysis to identify exercise EVLW. Exercise EVLW was defined by B-lines that were newly developed or increased during exercise. The E/e′ ratio peaked during maximal exercise and immediately decreased during the recovery period in patients with HFpEF. Exercise EVLW was most prominent during the recovery period in patients with HFpEF, while its prevalence did not increase from peak exercise to the recovery period in controls. Exercise EVLW was associated with a higher E/e′ ratio and pulmonary artery pressure, lower right ventricular systolic function, and elevated minute ventilation to carbon dioxide production (V_E vs. VCO₂) slope during peak exercise. Increases in B-lines from rest to the recovery period provided an incremental diagnostic value to identify HFpEF over the H₂FPEF score and resting left atrial reservoir strain.

Conclusion

Exercise EVLW was most prominent early during the recovery period; this may be the optimal timeframe for imaging ultrasound B-lines. Exercise stress echocardiography with assessments of recovery EVLW may enhance the diagnosis of HFpEF.

Graphical Abstract

Open in new tabDownload slide

See the editorial comment for this article ‘Of high water, stress echo, and diastole’, by R. Sicari, https://doi.org/10.1093/ehjci/jead029.

Introduction

Heart failure (HF) with preserved ejection fraction (HFpEF) accounts for more than half of all HF cases and is an important public health concern.¹ Patients with HFpEF and a lower degree of congestion often have normal left ventricular (LV) filling pressures at rest but develop abnormal elevation during physical stress such as exercise.^2–5 This elevation in LV filling pressure causes extravascular lung water (EVLW) or lung congestion during exercise, which has the potential to lead to exertional dyspnoea and poor clinical outcomes.^6,7

Lung ultrasound (LUS) can identify EVLW by demonstrating ultrasound B-lines.^8–10 Ultrasound B-lines have been demonstrated to reflect dynamic changes in EVLW that develop in response to elevation in pulmonary capillary wedge pressure (PCWP) during exercise in HFpEF.^7,10–14 Recent interest has focused on a combination of diastolic stress echocardiography with LUS to identify occult HFpEF,^10,15 but the optimal timeframe to image ultrasound B-lines is unknown. A time lag may exist between the elevation in PCWP and the manifestation of EVLW during exercise, which requires fluid leakage from pulmonary capillaries to extravascular spaces.^7,12,16 We hypothesized that the B-lines would increase throughout exercise and would be most pronounced early during the recovery period in patients with HFpEF.

Improved understanding of the dynamics of ultrasound B-lines during exercise may enhance the application of this technique to the diagnostic evaluation of HFpEF. Accordingly, we sought to determine the dynamic changes in B-lines during exercise in patients with HFpEF compared with those in participants without HF.

Methods

Study population

Consecutive patients referred for bicycle exercise stress echocardiography for the evaluation of exertional dyspnoea at Gunma University Hospital between November 2019 and April 2022 were retrospectively enrolled. The diagnosis of HFpEF was defined by the established the Heart Failure Association Pre-test assessment, Echocardiography & natriuretic peptide, Functional testing, Final etiology (HFA-PEFF) algorithm proposed by HFA of the European Society of Cardiology.¹⁷ In brief, the HFA-PEFF score was calculated as the sum of echocardiographic functional [age-specific cut-offs for early diastolic mitral annular velocity (e′) velocity, early transmitral flow velocity (E)/e′ ratio, tricuspid regurgitation (TR) velocity (TRV), and longitudinal strain (LS): maximum 2 points], morphological [rhythm-specific left atrial (LA) volume, relative wall thickness, and sex-specific measures of LV mass: maximum 2 points], and natriuretic peptide (NP; maximum 2 points) domains. Thirty-one patients had no NP data, and therefore, no points were added to the NP domain for these patients. Subsequently, 2 or 3 points were added depending on the E/e′ ratio and TRV during exercise stress echocardiography. A diagnosis of HFpEF was confirmed if the total score was ≥5 points. Patients with elevated LV filling pressures (PCWP > 15 mmHg at rest and/or ≥ 25 mmHg during exercise) on exercise right heart catheterization (RHC) were classified as having HFpEF. Patients with EF < 50%, significant left-sided valvular heart disease (>moderate regurgitation, > mild stenosis), infiltrative, restrictive, or hypertrophic cardiomyopathy, non-Group II pulmonary artery hypertension, exercise-induced pulmonary hypertension without elevation in E/e′ ratio [pulmonary artery mean pressure (mPAP) with exercise >30 mmHg with a total pulmonary resistance (i.e. mPAP/cardiac output) of >3 mmHg･min/L], and interstitial lung diseases were excluded.^18–20 Patients who did not meet the HFpEF criteria were categorized as controls.

The H₂FPEF score, which estimates the probability of HFpEF, was also calculated based on four clinical parameters [body mass index (BMI) > 30 kg/m² (2 points), treatment with two or more antihypertensive medicines (1 point), atrial fibrillation (AF; 3 points), and age >60 years (1 point)], as well as two echocardiographic variables [E/e′ ratio > 9 (1 point) and pulmonary artery systolic pressure (PASP) > 35 mmHg (1 point)].²¹ The study was approved by our institutional review board with a waiver of consent. The data underlying this article cannot be shared publicly due to the privacy of individuals that participated in the study.

Exercise stress echocardiography

Two-dimensional and Doppler echocardiography were performed by experienced sonographers using a commercially available ultrasound system (Vivid E95; GE Healthcare, Horten, Norway). LV systolic function was assessed based on ejection fraction (EF), systolic mitral annular tissue velocity at the septal annulus (mitral s′), and LV longitudinal strain (LVLS). The LV volume, EF, and LVLS were determined using apical four-chamber views, and the LV diastolic function was assessed using the septal E/e′ ratio. The right ventricular (RV) basal, mid-ventricular, and longitudinal dimensions were measured at end-diastole using RV-focused views. RV systolic function was assessed using tricuspid annular plane systolic excursion (TAPSE) and systolic tissue velocities at the lateral tricuspid annulus (TV s′). The right atrial (RA) pressure (RAP) was estimated from the diameter of the inferior vena cava and its respiratory change. The PASP was calculated as 4 × (peak TRV)² + estimated RAP. The LA reservoir and booster pump strains were measured from apical four-chamber views. Myocardial deformation analyses were performed at rest using a commercially available software (EchoPAC PC, GE, Milwaukee, WI, USA), and strain measurements are presented as absolute values. All measurements represent the mean of two beats in sinus rhythm and ≥3 beats in AF.

All participants underwent supine ergometry exercise echocardiography, starting at 20 W for 5 min, increasing in 20 W increments in 3 min stages until the subject reported exhaustion.^15,20,22 Echocardiographic images were obtained at baseline, during all stages of exercise, and during the recovery phase. In the majority of participants (n = 192), expired gas analysis was performed simultaneously with echocardiography at rest and throughout the exercise to assess breath-by-breath oxygen consumption (VO₂), carbon dioxide production (VCO₂), tidal volume (V_T), respiratory rate, and minute ventilation (V_E = V_T × respiratory rate). The ventilation efficiency was estimated using the V_E vs. VCO₂ slope.

LUS imaging

LUS was performed 1 min after the initiation of each stage to assess the EVLW. We focused on changes in B-lines throughout the exercise, and four intercostal spaces in the right hemithorax (right third and fourth spaces along the mid-axillary and mid-clavicular lines) were scanned because of the time constraints for imaging.^7,11,23 The sonographic signs of EVLW are ultrasound B-lines, which present as vertical, hyperechoic lines that originate from the pleural line and extend to the bottom of the ultrasound screen while moving synchronously with respirophasic motion.^10,11 The number of B-lines was assessed offline by a single observer (K.K.) who was unaware of the patients’ information, and the total number of B-lines in the four predefined zones was used for the primary analyses. We have demonstrated high inter-observer and intra-observer reproducibility for this measurement in our laboratory.²⁰ Exercise EVLW was considered to be present if B-lines appeared only during exercise or if the number of B-lines increased during exercise as compared with rest, as described previously.⁷

Exercise right heart catheterization

A subset of participants (n = 29) underwent clinically indicated RHC at rest and during supine ergometry exercises as a confirmatory test. The RAP, PA pressure, and PCWP were measured at end-expiration (mean of ≥3 beats) using a 6 Fr fluid-filled catheter from the right internal jugular vein, as previously described.^3,4,24 Following the acquisition of baseline data, haemodynamic assessments were performed during supine ergometry exercise, starting at 20 W for 5 min and increasing in 20 W increments in 3 min stages until volitional exhaustion.

Statistical analysis

Data are reported as the mean (SD), median (interquartile range), or number (%) unless otherwise specified. Between-group differences were compared using the unpaired t-test, Wilcoxon rank-sum test, or χ² test, as appropriate. Within-group comparisons at four time points were performed using one-way repeated analysis of variance. Within-group differences in prevalence were compared by the McNemar test. Receiver-operating curves were constructed to evaluate the diagnostic performance of B-lines in distinguishing HFpEF from controls. The incremental diagnostic value of B-lines over the H₂FPEF score was evaluated using sequential logistic regression analysis with nested models. The change in the overall −2 log-likelihood ratios of the models was used to assess the increase in the diagnostic information. All tests were two-sided, with a significance level of P < 0.05. All statistical analyses were performed using JMP 15.2.0 (SAS Institute, Cary, NC, USA).

Results

Baseline clinical characteristics

A total of 255 consecutive participants (134 patients with HFpEF and 121 controls) met the study criteria. Measurements of B-lines were feasible in all participants at rest, during 20 W and peak exercises, and during the recovery period. Of the 134 patients with HFpEF, 29 were diagnosed based on exercise RHC; the haemodynamic data are presented in Supplementary material online, Table S1. As expected, PCWP and RAP markedly increased during peak exercise, with secondary increases in PA pressure.

Compared with the controls, the patients with HFpEF were older and had a higher prevalence of comorbidities, including diabetes, systemic hypertension, AF, and coronary artery disease (Table 1). The sex and BMI were similar between the groups. Patients with HFpEF were treated with neurohormonal blockers and diuretics more frequently and had higher NP levels and HFA-PEFF scores than the control participants. The LV and RV sizes were similar between groups, while the LV mass index and LA and RA volume index were larger in the patients with HFpEF than in the controls.

Resting echocardiographic markers and EVLW

At rest, the heart rate, blood pressure, oxygen saturation, and LVEF were similar between patients with HFpEF and controls (Table 2). Compared with control participants, the patients with HFpEF had more impaired LV diastolic function (lower e′ velocity and higher E/e′ ratio), lower s′ tissue velocity and LVLS, reduced LA reservoir and booster pump strain, higher estimated PASP and RAP, and lower TAPSE, while the TV s′ was similar between the groups. At least one ultrasound B-line was present in 25.4% of controls and 38.4% of patients with HFpEF (P = 0.03). Moreover, the NT-pro BNP levels were higher in patients with HFpEF with B-lines at rest than in those without [1015 (336–2270) pg/mL vs. 404 (134–830) pg/mL; P = 0.0009], while there was no such difference in the control group [95 (65–201) pg/mL vs. 89 (56–130) pg/mL; P = 0.62].

Echocardiographic markers and EVLW during exercise

Compared with the controls, the peak exercise workload was lower and exercise duration was shorter in patients with HFpEF (Table 3). During peak exercise, the heart rate and systolic BP were lower in the patients with HFpEF than in the controls, while the oxygen saturation was similar between the groups. Compared with the controls, the patients with HFpEF displayed lower LVEF, higher mitral E and A velocities, lower mitral e′ and s′ velocities, higher E/e′ ratio, estimated PASP and RAP, and lower TAPSE and TV s′ during peak exercise. Figure 1 shows the prevalence of exercise EVLW, defined by B-lines that appeared only during exercise or increased during exercise compared with rest. The prevalence of exercise EVLW increased from 20 W to peak exercise to a greater extent in patients with HFpEF than that in the controls (Figure 1). The number of B-lines increased from rest to peak exercise was correlated with peak E/e′ ratio, but its strength was weak (r = 0.17, P = 0.007).

Echocardiographic measures, haemodynamics, and EVLW during the recovery period

In patients with HFpEF, the E/e′ during the recovery period was significantly lower than that during peak exercise (P = 0.01; Figure 2A). This observation was confirmed by an immediate decrease in invasively measured PCWP after exercise termination in a subset of participants with exercise RHC (P < 0.001; Figure 2B). Despite the reduction in PCWP, the prevalence of patients with HFpEF with exercise EVLW was further increased from 50.0% during peak exercise to 67.9% during the recovery period (P < 0.0001, Figure 1). In contrast, the prevalence of exercise EVLW in controls remained almost unchanged (27.3% during peak exercise and 29.8% during the recovery period, P = 0.55; Figure 1). Results were similar when exercise EVLW was defined as ≥2 B-lines appearing only during exercise or if the number of B-lines increased by ≥2 during exercise compared with rest (see Supplementary material online, Figure S1). The number of B-lines increased from rest to recovery period was mildly correlated with E/e′ ratio during peak exercise (r = 0.34, P < 0.0001).

Participants with exercise EVLW from rest to the recovery period (n = 121) showed a lower LA reservoir strain at rest, higher E/e′ ratio and estimated PASP, lower TV s′, and an elevated V_E vs. VCO₂ slope and V_D/V_T during peak exercise than those without (n = 134; Figure 3). The differences in E/e′ ratio and V_E vs. VCO₂ slope remained significant when assessed in patients with HFpEF only (see Supplementary material online, Figure S2). When assessing the controls only, the prevalence of elevated NP levels was higher (80 vs. 52%, P = 0.02), the LA volume index was larger [28 (22–33) vs. 22 (17–29) mL/m², P = 0.007], and the LA reservoir strain was lower (31 ± 12% vs. 37 ± 13%, P = 0.03) in those with exercise EVLW (n = 36) than in those without (n = 85).

Diagnostic ability of increased B-lines during exercise to identify HFpEF

The number of B-lines increased from rest to the recovery period demonstrated moderate diagnostic accuracy, with an area under the curve of 0.74 (P < 0.0001, Figure 4A). The H₂FPEF score distinguished HFpEF from controls (χ² 26, P < 0.0001; Figure 4B), and the addition of resting LA reservoir strain improved the diagnostic value over the H₂FPEF score (χ²: 45.6 vs. 19.1, P < 0.0001). The diagnostic value was further improved by adding the number of B-lines increased from rest to the recovery period (χ²: 72.9 vs. 45.6, P < 0.0001). The addition of B-lines increased from rest to peak exercise to the H₂FPEF score and LA reservoir strain also increased the model performance, but the improvement was modest (54.2 vs. 45.6, P = 0.003).

Discussion

To the best of our knowledge, this is the first study to examine the dynamic changes in ultrasound B-lines throughout exercise in patients with HFpEF compared with non-HF controls. We found a difference in dynamic changes between LV filling pressure and exercise EVLW: E/e′ ratio and invasively measured PCWP peaked at maximal exercise, with an immediate decrease after the exercise, while exercise EVLW was most prominent during the recovery period. During the recovery period, EVLW was newly developed or increased in two-thirds of the patients with HFpEF, and the presence of exercise EVLW was associated with increased LV filling pressures, RV dysfunction, PA-RV uncoupling, and ventilatory inefficiency during exercise. Increases in EVLW from the baseline to the recovery period demonstrated incremental diagnostic value in identifying HFpEF among patients with dyspnoea over the established H₂FPEF score and LA reservoir strain. These data highlight the potential of exercise EVLW assessments in the diagnostic evaluation of HFpEF and provide important insights into the incorporation of LUS into exercise stress echocardiography.

Lung water cascade during exercise in HFpEF

Previous studies have demonstrated that ultrasound B-lines increase during exercise in patients with HFpEF, and that the severity of exercise B-lines was correlated with PCWP.^7,10–14 These observations suggest that B-lines may reflect increases in EVLW secondary to elevation in LV filling pressure in HFpEF. Reddy et al.⁷ reported an interesting observation in that EVLW developed in only half of the patients with HFpEF during submaximal exercise despite an elevation in invasively measured LV filling pressure. The authors speculated that, in addition to high LA pressure (or pulmonary capillary hydrostatic pressure), high right-sided filling pressure was necessary for the development of lung congestion by impairing lymphatic clearance of lung water. In the current study, we found different dynamics between PCWP and EVLW throughout the exercise period. The E/e′ ratio and invasively measured PCWP reached peak values at maximal exercise in patients with HFpEF, with an immediate reduction in recovery period. In contrast, the EVLW was most apparent during the recovery period in patients with HFpEF. Our findings and those of others suggest that the discrepancy between PCWP and EVLW may be also explained by the time required for fluid to shift from the pulmonary capillaries to the extravascular space following an acute increase in PCWP during exercise.^12,16 The development of EVLW during exercise is attributed not only to pulmonary capillary pressure (hydrostatic pressure) but also to colloid osmotic pressure and vascular permeability.²⁵ Thus, further investigation is warranted to understand the mechanisms of exercise EVLW in patients with HFpEF.

Association of EVLW with exercise haemodynamics and ventilatory efficiency

We demonstrated that patients with exercise EVLW had a lower resting LA reservoir strain and higher E/e′ ratio during peak exercise than those without exercise. This suggests that reduced LA compliance may worsen pulmonary congestion through elevation in LA and thus pulmonary capillary pressure during exercise.²⁶ We also demonstrated the association between exercise EVLW and RV dysfunction and abnormal RV-PA coupling. This finding is in agreement with those of previous studies and could be explained by impaired lymphatic clearance to remove excess fluid from the extravascular space in the lungs to the central venous system in the setting of high systemic venous pressure.^7,27 Moreover, we extended upon previous studies and found that patients with HFpEF with exercise EVLW displayed a higher V_E vs. VCO₂ slope and worse dead space ventilation compared with those without. This finding suggests that lung congestion itself may worsen ventilatory efficiency by impairing physiological reduction in alveolar dead space ventilation during exercise.^3,28 Collectively, these data suggest that therapies aimed at reducing lung water may help to mitigate or prevent abnormal RV-PA coupling or ventilatory inefficiency in patients with HFpEF.

Notably, the prevalence of EVLW was slightly increased even in controls, which is in agreement with the findings of previous studies.¹² Our controls had symptoms of dyspnoea and multiple comorbidities. We also observed that controls with exercise EVLW had elevated NP levels, a larger LA volume index, and a lower LA reservoir strain compared with those without; this could be considered a pre-HFpEF during the disease process. However, further studies are warranted to determine the pathophysiological and prognostic role of EVLW in patients without overt HF.

Exercise EVLW in the diagnostic evaluation of occult HFpEF

Patients with HFpEF and a lower degree of congestion often have normal LV filling pressure at rest, with abnormal elevation only observed during exercise.^2,3 This elevation in LV filling pressure causes EVLW, which is readily detectable by LUS as ultrasound B-lines.^8,10 LUS is sensitive, highly feasible, and reproducible even in patients with poor echocardiographic images during exercise.^8,10,11 The assessment of EVLW is gaining attention for the diagnosis of HFpEF, and the current recommendation proposes its use during exercise stress echocardiography.^15,29 From a diagnostic perspective, it is important to determine the optimal time frame to assess exercise EVLW. In this study, we evaluated EVLW at rest, 20 W and peak exercise, and 1 min during the recovery period, and found that exercise EVLW was most prominent during the recovery period in patients with HFpEF, which is in agreement with results of a previous study.¹² These findings, together with those of previous studies, suggest that early recovery may be the optimal time frame to assess EVLW.⁷ Notably, this will allow time to focus on the acquisition of diastolic echocardiographic parameters during peak exercise. We also found that exercise EVLW provided an incremental diagnostic value over the H₂FPEF score and LA reservoir strain, both of which are promising diagnostic tools for HFpEF.^21,30 Collectively, the current data provide important insights into the development of exercise stress echocardiography with LUS for the evaluation of HFpEF. This may enhance the diagnosis of HFpEF.

Limitations

This single-centre study was conducted at a tertiary referral centre, which inevitably introduced selection and referral bias. Although the patients with HFpEF were carefully identified, we cannot exclude the possibility that some patients might have been missed because only a few patients underwent exercise RHC, which is the gold standard for the diagnosis of HFpEF.² Moreover, patients with interstitial lung disease might have been included because the ultrasound B-lines observed at rest were not increased in some participants. B-lines might detect early interstitial lung lesions that could not be identified by computed tomography or physical signs.²⁰ The control population was not truly normal in that they had prevalent comorbidities and were referred to exercise stress echocardiography for evaluation of exertional dyspnoea. From a diagnostic perspective, it is necessary to include a control group that complains of dyspnoea. There is little question that individuals with no symptoms of HF do not have HF, and most would agree that this is not the cohort in which diagnostic evaluation is required. The fact that the control population was more diseased than a truly normal healthy control population only biases our data towards the null. In the current study, LUS was performed in four lung sites on the right chest because of the time constraints during exercise. This may have reduced the sensitivity of assessment for EVLW. Further studies using more lung-site scans are needed to confirm the diagnostic value of EVLW.

Conclusions

In patients with HFpEF, exercise EVLW was most prominent during the early recovery period. Imaging ultrasound B-lines during the recovery period may allow time to obtain diastolic echocardiographic parameters during submaximal and peak exercises. Exercise stress echocardiography with assessments of recovery EVLW may enhance the diagnosis of HFpEF.

Supplementary material

Supplementary materials are available at European Heart Journal – Cardiovascular Imaging online.

Acknowledgements

M.O. received research grants from the Fukuda Foundation for Medical Technology; the Mochida Memorial Foundation for Medical and Pharmaceutical Research, Nippon Shinyaku; the Takeda Science Foundation; the Japanese Circulation Society; the Japanese College of Cardiology; and the JSPS KAKENHI 21K16078.

References

Author notes