Left ventricular ejection fraction (LVEF)[1] has persisted as the primary measure of left ventricular systolic function despite flaws in this approach. Patients with heart failure are relieved when the LVEF is reported as normal, and clinicians may use the report of a depressed LVEF to persuade patients of the need for treatment. Therefore, patients often use any change in LVEF to track progress or deterioration in their condition. Although misguided, many patients will assume a report indicating a normal LVEF indicates recovery.
The best method to define a normal LVEF, or any diagnostic test result, is unclear. Investigators traditionally have evaluated populations who are assumed to be healthy and determined the distribution of the test result. They then picked a boundary of the population distribution (upper or lower 5% or 1%) to determine "normal." A recent meta-analysis of such population studies that included adults without clinical cardiovascular or renal disease, hypertension, or diabetes found that the threshold for a normal LVEF (defined as the lower 5th percentile of the distribution) varied by race or ethnicity and sex (European men 50%, women 51%; East Asian men 56%, women 57%; South Asian men 52%, women 53%; comparable data were not available for other races or ethnicities).[2] However, this population-based threshold for a normal LVEF may not indicate a "healthy" LVEF. In the ARIC study (Atherosclerosis Risk in Communities), incident heart failure risk (and NT-proBNP [N-terminal pro-B-type natriuretic peptide] levels) increased progressively with lower LVEFs, beginning with LVEF percentages in the low 60s.[3] Thus, whereas LVEF percentages in the high 50s may be viewed as normal from a population distribution, they may also indicate an increased health risk.
Do the LVEF thresholds for prognosis differ in people with heart failure? The relationship between LVEF and outcome for those with heart failure is explored in this issue by Kondo and colleagues.[4] The authors merged data from 33 699 patients from 6 randomized controlled heart failure trials. They determined how patient characteristics and several important outcomes varied by LVEF. They found that people with a higher LVEF (≥50%) were more likely to be female, to be older, and to have atrial fibrillation and diabetes, and were less likely to have ischemic heart disease.
The authors noted that the incidence of cardiovascular outcomes increased with decreasing LVEF, although there were different LVEF thresholds for different outcomes. Inflection points were at an LVEF of 50% for cardiovascular death, 40% for pump failure death, 35% for heart failure hospitalization, and no inflection point for sudden death.
A strength of this study is the use of multiple clinical trials of patients with heart failure across the LVEF spectrum. This allowed detailed clinical data on each individual with adjudicated outcomes, which are rarely available from observational studies. On the other hand, the clinical trial population may differ from the community heart failure population in that those eligible (and choosing) to enroll in a trial are likely better treated and may have better noncardiovascular health than those who are ineligible or refuse to enroll in a trial.
The authors did not find a J-shaped relationship between LVEF and noncardiovascular or all-cause death, which contrasts with previous studies[5] that found increased risk of death with an LVEF >65% for both those with and without heart failure. One potential reason for the different results is that clinical trials (combined for the current study) typically exclude patients with severe systemic disease (eg, infection). In such patients, the left ventricle may respond to systemic disease with a higher stroke volume (and higher LVEF). In addition, the earlier cited study had a much larger sample size (and greater power to detect small differences) given its use of electronic health records, including >13 000 patients with an LVEF >70% compared with 1018 in the current study of clinical trial participants.[5]
As noted previously, the authors have shown that there is no single LVEF for predicting outcome. In addition, there are different LVEF thresholds for treatment benefit. Most treatments for heart failure show reduced benefit as the LVEF increases. SGLT2 inhibitors (particularly dapagliflozin) have been promoted as an exception to this rule on the basis of evidence of a similar relative benefit (risk ratio) across different LVEF categories. However, the more patient-relevant metric is absolute benefit. The study by Kondo et al.[4] demonstrates that the absolute risk of cardiovascular death and heart failure hospitalization decreases with increasing LVEF. Thus, the relative benefit of treatment, when multiplied by the lower absolute risk, leads to less absolute benefit with SGLT2 inhibitors at a higher LVEF. This is consistent with cost-effectiveness studies of the SGLT2 inhibitor dapagliflozin showing that the cost per quality-adjusted life-year gained was much better in those with reduced LVEF.[6] This raises an additional LVEF threshold to consider: the threshold below which treatment is a good value and deserves coverage (with minimal copayments) through any insurance program.
The study provides several more useful insights. Concerns have been raised by some that many patients with preserved LVEF in clinical trials may have unrecognized amyloidosis. In the current study, among those with echocardiographic data, there were no structural differences in patients with a high-normal LVEF that suggested amyloidosis was commonly undiagnosed. As noted in other studies, women had a higher LVEF, were older, and had more atrial fibrillation and less ischemic heart disease than men. Despite this, there was no significant interaction among sex, LVEF, and survival, whether classified as cardiovascular or noncardiovascular death.
Some of the findings may be surprising, including the fact that those with the lowest LVEF had the best health status as identified by the Kansas City Cardiomyopathy Questionnaire. This paralleled the findings of more edema, orthopnea, and rales in those with higher LVEF. It appeared that those with lower LVEFs were more compensated than those with higher LVEFs from the patient's point of view. In contrast, there was a U-shaped relationship with New York Heart Association class and LVEF, with more patients with New York Heart Association class III or IV among those with an LVEF <20% or with higher LVEFs. The difference between patient-reported and clinician-reported outcomes is not new. Other studies have found that patient-reported outcomes (eg, Kansas City Cardiomyopathy Questionnaire) and clinician-reported outcomes (eg, New York Heart Association class) provide independent prognostic information with slightly greater prognostic value for the patient-reported outcome.
This study raises several important questions. Do the LVEF–outcome relationships vary by other patient groups (eg, ancestry/race/ethnicity), social determinants of health, and treatment status? Will community-based data show similar relationships to those seen with these clinical trial participants? Perhaps the most important question is how these results should be incorporated into clinical guidelines and ultimately practice.
Guidelines combine evidence of treatment effects from clinical trials with our understanding of physiology to make clinical recommendations and definitions of LVEF categories. As noted by the authors, the major clinical guidelines now include a mildly depressed LVEF category of 41% to 49%. The authors conclude that such a threshold is biologically rational on the basis of their finding of thresholds for outcome effects near 50%. Although it is encouraging that the prognostic thresholds are consistent with the treatment thresholds, guidelines should not focus on prognosis when determining LVEF thresholds for treatment. These LVEF categorizations should be on the basis of enrollment criteria from clinical trials showing treatment benefit. Whereas clinicians can use the data from this study to better communicate prognosis, they should continue to focus on randomized trial results for treatment decisions.
The study by Kondo et al.[4] shows that the LVEF–outcome relationship is complex, with different thresholds for increased risk depending on the outcome. Thus, there is no clear range at which we can say the LVEF is normal (optimal health). For those enrolled in clinical trials (those without severe systemic disease), it appears that the higher the LVEF, the better the outcome. Despite its limitations, LVEF continues to provide important information for prognosis and treatment benefit.
Circulation. 2023;148(9):750-752. © 2023 American Heart Association, Inc.
