Exercise Echocardiography to Diagnose Heart Failure with Preserved Ejection Fraction: Are two Measures Better than One?

This article refers to ‘Diagnostic value of reduced left atrial compliance during ergometry exercise in heart failure with preserved ejection fraction’ by T. Harada et al., published in this issue on pages 1293–1303. Despite advances in echocardiography, the non-invasive diagnosis of heart failure with preserved ejection fraction (HFpEF) remains challenging. In contrast to heart failure with reduced ejection fraction where a low ejection fraction is diagnostic for disease, there is no single echo measurement that is pathognomonic for HFpEF in a symptomatic patient. The diagnosis of HFpEF therefore relies on objective evidence of ‘pump failure’, where systemic perfusion during exertion can only occur in the presence of pathological increases in left atrial (LA) pressure with associated pulmonary congestion,1 dyspnoea2 and exercise intolerance.3 The most accurate way to measure LA pressure during exercise is by measurement of pulmonary capillary wedge pressure (PCWP) with exercise right heart catheterization. Exercise echocardiography using traditional Doppler measurements with E/e' as an estimate of exercise PCWP is appealing to allow non-invasive diagnosis, but data on its diagnostic accuracy are conflicting. In the first study to evaluate the accuracy of exercise E/e' to diagnose HFpEF compared to the gold standard, there was incremental diagnostic value at the expense of worse positive predictive value, and a sizeable minority of patients with invasively-proven HFpEF had E/e' <15 at peak exercise.4 The only other study to compare accuracy to the gold standard, found that exercise E/e' had no diagnostic value, with a 95% confidence interval for the area under the curve (AUC) that included 0.5 indicating no significant advantage when compared to coin flip for diagnosis.5 There remains no multicentre study validating the accuracy of exercise E/e' compared to the gold standard with exercise PCWP, and with the totality of evidence suggesting only modest and inconsistent correlations between E/e' and PCWP,6 the real world utility of stress E/e' to diagnose HFpEF remains unclear, especially outside of the setting of laboratories with expertise in these techniques. Given the limitations of traditional Doppler techniques, novel echo measures such as LA reservoir strain have emerged with suggested superiority to E/e' even from resting measures to predict dynamic increase in exercise PCWP due to HFpEF.7, 8 LA reservoir strain assesses the deformation of the left atrium during filling and correlates with distensibility of the atrium, so indexing this to E/e' ratio (as a non-invasive measure of PCWP) provides an estimation of LA compliance, which is abnormal in HFpEF7 and worsens with progressive LA myopathy.9 The ratio of LA reservoir strain/E/e' measured at rest had the greatest diagnostic value to identify HFpEF during exercise in one study,7 but there are currently insufficient data to determine the incremental diagnostic value of LA strain and compliance with exercise echocardiography. To address this unmet need, in this issue of the Journal Harada et al.10 present the results of the first study addressing the incremental value of LA compliance during exercise echocardiography to diagnose HFpEF. They utilized two complementary cohorts to test their hypothesis. Cohort 1 included 487 patients who underwent exercise echocardiography where HFpEF was diagnosed non-invasively in 225/487 (46%) using the HFA-PEFF diagnostic algorithm (HFA-PEFF score ≥5) incorporating resting data, along with potential addition of 2–3 points from stress echo data depending on exercise E/e' and tricuspid regurgitation velocity. Cohort 2 included 67 patients who underwent exercise right heart catheterization due to diagnostic uncertainty after stress echocardiography, where HFpEF was diagnosed in 49/67 (73%) by the gold standard of elevated PCWP during rest or exercise. The demographics of HFpEF patients included were generally consistent across the two cohorts, representing an older hypertensive population with low prevalence of obesity reflective of the typical Japanese HFpEF patient. LA reservoir strain and LA compliance were lower in HFpEF than controls both at rest and exercise. Exercise LA compliance demonstrated the highest AUC of 0.87 to identify HFpEF, with incremental value to resting LA compliance, exercise E/e' and exercise LA reservoir strain in cohort 1. In the smaller cohort 2, exercise LA compliance again demonstrated the highest AUC, with statistically significant superiority compared to exercise E/e' and numerically superior discrimination compared to resting LA compliance and exercise LA reservoir strain alone, but this was not statistically significant. Notably, exercise LA reservoir strain alone did not appear to add incremental diagnostic value to resting LA reservoir strain, suggesting that it was necessary to adjust exercise LA reservoir strain for E/e' (exercise LA compliance) to add diagnostic value with exercise echo. The results were generally consistent in multiple sensitivity analyses, although it did appear that there was less incremental value of exercise LA compliance in patients in atrial fibrillation, although this subgroup analysis should be viewed with caution due to the small number of atrial fibrillation patients. This is a carefully conceived and well-executed analysis from a high quality exercise echocardiography laboratory addressing an important unanswered question about the role of exercise LA compliance in the diagnostic evaluation of HFpEF. The results have strong biological plausibility with emerging data that LA myopathy with abnormal LA compliance exists across the spectrum of HFpEF and contributes to the elevation in PCWP during exercise, characteristic of the disease.9 There are a few caveats to consider however. Although cohort 1 included patients with a high probability of HFpEF based on the HFA-PEFF algorithm, this algorithm has only modest discriminatory value. Indeed, in a multicentre study using the invasive gold standard of exercise haemodynamics, 55% of low probability HFA-PEFF scores and 76% of intermediate probability HFA-PEFF scores actually had HFpEF when invasively evaluated.11 Since ‘controls’ in the present study were defined in this way, it remains unclear how many of those patients truly did not have HFpEF if they would have been invasively evaluated, though the analysis from cohort 2 is very reassuring in this regard. Additionally, an elevated exercise E/e' is counted towards the HFA-PEFF score (and the non-invasive diagnosis of HFpEF in cohort 1), which could inflate diagnostic performance of exercise LA compliance (which incorporates exercise E/e') in cohort 1. Again, here the inclusion of cohort 2 was critical to validate the results which were generally replicated. The inclusion of a cohort with gold standard diagnosis with exercise catheterization to validate the findings is essential to critically evaluate diagnostic accuracy and should be the standard for investigation of the non-invasive diagnosis of HFpEF. As the authors appropriately suggest, there is a need for further validation of these promising results before clinical use. In particular, the need for three independent measurements with E velocity, e' velocity and LA strain during exercise may impact feasibility and contribute to measurement variability outside highly experienced research centres. The ∼90% feasibility reported by the authors is promising but will need to be replicated in real-world settings. Notably, the lack of obese HFpEF patients in this series (the most common form of HFpEF in the Western world) may overestimate diagnostic performance and feasibility since image quality may be more challenging during exercise in these patients. Although these results are promising, it is important to evaluate the totality of evidence to determine the objective accuracy of non-invasive diagnosis with stress echocardiography in HFpEF. As summarized in Table 1, across the now three available study cohorts with stress echocardiography and exercise right heart catheterization,4, 5, 10 there are inconsistencies in both the diagnostic value and role of exercise E/e' and LA reservoir strain to diagnose HFpEF. In contrast to the original study,4 both Backhaus and now Harada et al. demonstrate that exercise E/e' did not add incremental value to resting E/e'.5, 10 Both Backhaus and Harada et al. also failed to identify incremental value to exercise echo LA reservoir strain over resting LA reservoir strain.5, 10 The novel results by Harada et al. now suggest that indexing exercise LA reservoir strain for exercise E/e' may provide the best diagnostic value, but the conflicting results above across independent cohorts for exercise E/e' emphasize the need for validation of the current exercise LA compliance data for diagnosis. HFpEF 14 (10, 18) Control 8 (7, 10) p < 0.001 HFpEF 12.5 (9.7, 13,3) Control 9.2 (7.5, 10.7) p < 0.001 AUC 0.80 [0.68–0.91] HFpEF 13.7 ± 5.2 Control 9.7 ± 3.5 p = 0.004 AUC 0.75 [0.59–0.86] HFpEF 20 (15, 22) Control 8 (7, 12) p < 0.001 AUC 0.80 [0.68–0.89] HFpEF 13.8 (10.8, 15.9) Control 11.0 (10.0, 14.0) p = 0.12 AUC 0.61 [0.45–0.76] HFpEF 18.1 ± 6.7 Control 11.6 ± 3.4 p = 0.004 AUC 0.77 [0.62–086] HFpEF 22.8 (12.4, 30.1) Control 31.3 (26.4, 34.8) p < 0.001 AUC 0.79 [0.67–0.91] HFpEF 17.6 ± 9.1 Control 24.4 ± 5.6 p = 0.004 AUC 0.77 [0.64–0.86] HFpEF 21.4 (12.9, 25.7) Control 30.4 (27.5, 40.8) p < 0.001 AUC 0.82 [0.69–0.94] HFpEF 24.6 ± 12.3 Control 41.8 ± 14.8 p < 0.0001 AUC 0.82 [0.67–0.91] HFpEF 1.5 ± 1.1 Control 2.8 ± 1.0 p < 0.0001 AUC 0.81 [0.67–0.90] HFpEF 1.7 ± 1.5 Control 3.9 ± 2.1 p < 0.001 AUC 0.87 [0.76–0.94] With no single pathognomonic echo finding, the current diagnostic approach for HFpEF in a dyspnoeic patient involves a Bayesian approach combining clinical characteristics with echo measures to determine pre-test probability for HFpEF,12 and guide the need for confirmatory exercise right heart catheterization in a shared decision-making approach with patients. With this complexity in diagnosis, most patients with HFpEF in the community remain undiagnosed13 or even misdiagnosed with deconditioning, obesity, or having atrial fibrillation considered as a cause of symptoms, when in fact many actually have objective evidence of heart failure when appropriately evaluated.12, 14, 15 Given the emergence of treatments such as the sodium–glucose cotransporter 2 inhibitors that reduce heart failure hospitalizations and improve quality of life and effort tolerance in HFpEF, better evidence-based non-invasive techniques are urgently needed to facilitate diagnosis. The conflicting evidence for exercise echocardiography to diagnose HFpEF (Table 1) raises questions about its application for routine care at the present time,4, 5 but the study by Harada et al.10 represents an important step in the right direction, raising the priority for a multicentre study to assess the real-world accuracy of exercise LA reservoir strain and compliance across a broad population of HFpEF patients compared to the gold standard of exercise right heart catheterization. Conflict of interest: Y.N.V.R. is supported by National Heart, Lung, and Blood Institute of the National Institutes of Health (NHLBI) Award Number K23HL164901-01A1, grants from Sleep Number, Bayer, United Pharmaceuticals, and the Earl Wood Career Development Award from Mayo Clinic. B.A.B. receives research support from the NHLBI (R01 HL128526 and U01 HL160226) and the United States Department of Defense (W81XWH2210245), as well as research grant funding from AstraZeneca, Axon, GlaxoSmithKline, Medtronic, Mesoblast, Novo Nordisk, Rivus, and Tenax Therapeutics; has served as a consultant for Actelion, Amgen, Aria, Axon Therapies, BD, Boehringer Ingelheim, Cytokinetics, Edwards Lifesciences, Eli Lilly, Imbria, Janssen, Merck, Novo Nordisk, NGM, NXT, and VADovations, and is named inventor (US Patent no. 10,307,179) for the tools and approach for a minimally invasive pericardial modification procedure to treat heart failure.