By Michael H. Crawford, MD, Editor
SYNOPSIS: A large population study has shown a U-shaped relationship between left ventricular ejection fraction (EF) and mortality, with a nadir at 60-65%. The authors identified a new group at high risk for death: those with an EF ≥ 70%.
SOURCE: Wehner GJ, Jing L, Haggerty CM, et al. Routinely reported ejection fraction and mortality in clinical practice: Where does the nadir of risk lie? Eur Heart J 2020;41:1249-1257.
In heart failure (HF) patients, there is a well-known exponential relationship between left ventricular (LV) ejection fraction (EF) and mortality. However, the relationship of EF to mortality in large populations of patients who undergo echocardiography for a variety of clinical indications is less clear.
Wehner et al evaluated records from Geisinger Health System, which serves patients in New Jersey and Pennsylvania. Specifically, the authors investigated information collected from 1998 to 2018 regarding patients who underwent an echocardiogram. Researchers also studied similar records from the Waitemata District Health Board (New Zealand) for the sake of creating a validation cohort. As about three-quarters of the Pennsylvania echo labs recorded EF in the 4% to 5% range, all studies were categorized in 5% intervals, where the lowest and highest intervals were ≤ 20% and ≥ 70%. The derivation of the EF was described as qualitative in 59% of cases; 8% were quantitative, mostly from the biplane method. In the remaining 33%, no method was given. The primary endpoint was all-cause mortality. The authors studied several subgroups, looking for features potentially associated with increased EF or reduced EF. Also, they performed several sensitivity analyses to assess clinical factors that could affect the results, such as multiple echoes during the observation period.
Overall, 596,503 echoes from 271,201 patients were discovered in the U.S. cohort. Because of missing data, such as EF, 192,526 echoes were excluded, leaving 403,977 echoes in 203,135 patients who met inclusion criteria. Their mean age was 64 years, and 13% had a diagnosis of HF. During a median follow-up of four years (range, 1.3-8.4), 23% of patients died. Adjusted hazard ratios (HR) for death exhibited a U-shaped relationship with EF, with the risk nadir at 60-65%. When EF was ≥ 70%, the HR was 1.71 (95% confidence interval [CI], 1.64-1.77), which was similar to the HR when EF was 35-40% (1.73, 95% CI, 1.66-1.80).
Similar results were observed in the New Zealand cohort. Also, the results were not changed significantly when restricted by age, sex, and HF as well as when adjusted for conditions associated with increased EF, such as mitral regurgitation, LV hypertrophy, and anemia. The authors concluded any deviation in EF from the 60-65% range is associated with increased adjusted all-cause mortality. These results uncovered a new high-risk group: those with EF ≥ 70%.
The most provocative finding in this study was that patients with EF ≥ 70% died at the same rate as patients with EF in the 35% to 40% range. Most HF studies have shown that once EF is > 45%, differences in mortality are not significant. However, prior HF studies usually only included patients with EF < 50%. The authors of studies of HF with preserved (p) EF used various EF cutpoints, but > 50% seems to be most strongly associated with the futility of standard HF medications. The Wehner et al study revealed increases in mortality at > 65%. This raises the issue of how cardiologists should define HFpEF.
Other studies have shown similar results. For example, the authors of the MESA population study reported an EF nadir at 60%.1 The authors of the GRACE study of women with acute coronary syndrome reported a mortality nadir at 65% EF.2 When the Wehner et al study was subdivided into inpatients and outpatients, the EF nadir for mortality remained at 60-65% for inpatients. However, in outpatients, the rate was 55-60%. Thus, the precise nadir may vary with the population studied.
Another interesting finding in the Wehner et al study is that end-systolic volume index < 10 mL/m2 also was associated with increased mortality, suggesting that a small LV cavity size may be part of the adverse physiology of EF ≥ 70%. An analysis of the influence of body mass index (BMI) on mortality is interesting, too. BMI showed a U-shaped relationship to mortality, with a nadir at 30-35 kg/m2, which confirms the obesity paradox — at least at the moderately obese level. However, the highest HR for mortality was at BMI < 18.5 kg/m2, which suggests one can be too thin, especially if he or she has a small, hyperdynamic left ventricle.
There were limitations to the Wehner et al study other than its retrospective, observational design. Foremost is the almost exclusive use of visual estimation of EF on two-dimensional echo, although studies show that with experienced readers these estimates are reasonably accurate. Also, the authors reported no data on cause of death (but the fact the patients studied were referred for an echo suggests the authors may have suspected cardiac disease). In addition, both populations studied were largely of European ancestry.
The new information from this study is the patient with a small hyperdynamic left ventricle is at particularly high risk for mortality for reasons that are not yet clear.