In the beginning, there was heart failure (HF). Patients had fatigue; dyspnea; lung and dependent-limb edema; and big, weak hearts. When hearts were first imaged in the 1930s, we learned they normally eject about two-thirds of their filling volume, and ejection fraction (EF) was born. As HF patients had increased end diastolic volume (EDV) but normal or reduced stroke volume (SV), EF was lower and became a marker for HF. While there were also undoubtedly patients with HF symptoms but nondilated hearts and normal or even high EFs, we mostly ascribed that to something else. But in the 1960s, the idea caught on that HF could occur even in patients without cardiac dilation, and they often had EFs that were not reduced. Many had ischemic or hypertrophic–hypertensive heart disease and were quite elderly. As systolic function seemed intact but diastole impaired, the syndrome soon became "diastolic heart failure." Then we learned a lot more was going on and that diastolic dysfunction was common in many elderly adults with similar comorbidities, but this did not mean that they had HF. We needed a better name for this syndrome and after some sparring, in 2005 we settled on HF with preserved EF (HFpEF).
We called the syndrome HFpEF not because this implied a pathophysiology but because it simply described 2 main features: patients have symptoms of HF and an EF that is not reduced. We now know HF symptoms cannot all be blamed on the heart, as HFpEF also involves pulmonary, renal, skeletal muscle, vascular, metabolic (obesity), and other system dysfunction. That the first HFpEF therapy to successfully reduce cardiovascular death/hospitalization turned out to be a diabetes drug (SGLT2 [sodium-glucose cotransporter-2] inhibitor) not previously known to modulate the heart says a lot. Once we settled on HFpEF, the original HF (EF<35%) needed a new name, so it became HF with reduced EF (HFrEF). Patients in the donut hole got a name, too: HFmrEF ("mr" for midrange). But remember, HFpEF was never meant to suggest anything more than it's not HFrEF.
It is with this background that I find recent efforts to turn EF into a HFpEF biomarker puzzling. In this analysis, HFpEF patients are further divided into those with more (>60–65%) versus less (50–60%) preserved EF, suggesting they have different heart diseases in need of different precision-guided approaches. This seems odd as EF is a rather imprecise assessor of heart function, being sensitive to contractility but also to vascular resistance (declines at higher afterload), heart rate (lower with faster rates), and preload. The preload dependence in HF stems from the fact that while SV declines with EDV (Starling law), SV gets to zero first. Furthermore, cardiac output and heart rate both scale to body size and are neuro-hormonally regulated to achieve a mean arterial pressure of ≈90 mm Hg at the base of the brain and diastolic pressures of >75 to 80 mm Hg to provide adequate coronary perfusion. This constrains them, and we see this in our patients with HF who often have resting SV close to normal until severe failure sets in. Reserve cardiac output is another story, but here we are discussing resting parameters. If SV is more conserved, then the primary determinant of EF becomes EDV. This is certainly true for HFrEF, as a low EF mostly reflects chamber dilation and not always muscle function. For example, it falls in hearts with large infarct scars but otherwise normal residual muscle just as with diffuse cardiomyopathy. I am not saying EF only reflects heart size or is useless to distinguish depressed versus nondepressed function; however, I would not go too much further.
Given the impact of heart size on EF, it is not surprising that HFpEF patients with higher EF have smaller hearts and are more often women (women have smaller hearts). HFpEF patients with EFs closer to 50% have larger hearts. This also influences cardiac mechanics and must be kept in mind when interpreting such data. For example, pressure–volume analysis shows higher end-systolic and end-diastolic elastances suggesting higher contractility and diastolic stiffness in those with higher EFs and predicts different preload and afterload responses.[1,2] However, both elastances vary inversely with chamber volume so they will be higher in the EF >60% HFpEF group with smaller EDV. That does not necessarily reflect major pathobiologic differences; a perfectly normal smaller heart would also have higher values. The majority of HFpEF patients are now obese and this imposes a volume load so having a smaller ventricle is probably a disadvantage. However, I would suggest that this is best solved by reducing obesity and not enlarging the heart.
There are HFpEF patients with truly supranormal EF >75%, but they are a different group and generally have more hypertension, hypertrophic disease, and left ventricle cavity obliteration. This was once a more common phenotype, and in our 2003 study, we reported on such patients that had systolic pressures of nearly 170 mm Hg, higher EFs and end-systolic and end-diastolic elastances, and load-induced cardiac reserve limitations. This phenotype is much less prevalent today—taken over by obesity—but still, the same hemodynamic principles apply. According to the American Heart Association, a normal EF is between 50% to 70%, so I would pause before claiming those with EFs >60% have a particularly maladaptive form of HFpEF. Also, metrics unaffected by chamber size, such as the slope of a stroke work–EDV relationship for contractile function or diastolic stress–strain ratio, could help dissect out the role of chamber size.
Taking a step back, it is worth remembering why we called the syndrome "HFpEF" in the first place. It was not because EF was considered so insightful but because it indicated that you did not have reduced EF. Having preserved EF does not even mean your myocyte contractile function is "preserved" as we demonstrated in obese HFpEF patients with EFs >60% who we also found had very depressed calcium-dependent tension in skinned myocytes from their ventricular septum. While HFrEF and HFpEF certainly exhibit differences in their transcriptome, they also have many similarities, and more pertinent to this commentary, EF does not identify transcriptomic or sarcomere function subgroups among HFpEF patients.[4,5] In ongoing work, we do not see EF distinguishing metabolic differences either. However, I believe molecular/cellular features—rather than EF—are more likely to lead to effective therapeutic targets. In this respect, I would make a plug for more data from relevant tissues and not just blood. Biomarkers in blood are like testing for coronavirus disease 2019 (COVID-19) in wastewater: you know something is going on but not necessarily where or why.
The goal of a name is to capture the essence of something. In this respect, HFpEF has issues, since we now know that EF is not so preserved on exertion, that many factors contribute to HF symptoms, and even that profound myocyte dysfunction can coexist despite preserved EF. Subdividing HFpEF by EF ranges is mostly telling us what is predicted by cardiovascular models, but not too much about the underlying biology. I believe we need more insight into the latter to discover therapies that will work. Perhaps we also need a less distracting name. Before HFpEF became accepted, I had borrowed from Steven Spielberg, referring to HF of the first kind (obviously systolic), second kind (obviously diastolic), or third kind (not obviously either, ie, HFpEF). Admittedly, this may be a tough name to convey to our patients, but it would solve the EF problem.
Circulation. 2022;146(18):1327-1328. © 2022 American Heart Association, Inc.