Introduction

The analysis of left ventricular (LV) function in patients with symptoms of heart failure (HF) is usually performed by transthoracic echocardiography (TTE) as the primary imaging technique in clinical cardiology [1, 2]. Many HF patients with high normal or elevated natriuretic peptides present with a preserved or normal LV ejection fraction (LVEF) above 50% and are currently classified as “heart failure with preserved LVEF” (HFpEF) [3,4,5,6,7]. However, HFpEF should not be considered as a diagnosis, but rather as a syndrome [8]. Multiple pathologic entities may present with the same phenotype “HFpEF”, but with different comorbidities and varying outcomes [9]. Thus, the real diagnostic challenge of TTE in these patients is not to merely measure LVEF, but to characterize the underlying cardiac pathologies and to precisely obtain a specific diagnosis.

“Diastolic HF” was initially defined as “an increased resistance to filling in one or both ventricles, leading to symptoms of pulmonary congestion due to an inappropriate upward shift of the diastolic pressure–volume relation” [3]. Later it was replaced by the term “HFpEF” [5, 6]. The incidence and prevalence of “HFpEF” symptoms approximately constitute half of all HF patients [2, 7]. Recently, diagnostic algorithms have been introduced to characterize patients with preserved LVEF and HF symptoms by the hypernym “HFpEF” as a common diagnosis despite different underlying pathologies and treatment options, as well as varied prognosis [7, 9, 10]. The “HFpEF” algorithms are based on clinical complaints, laboratory findings, echocardiographic data and/or invasive hemodynamic measurements [4, 7], which often do not lead to the specific underlying diagnosis [11,12,13,14,15,16,17,18]. If cardiac diagnoses like heart valve disease, significant coronary artery disease, pericardial constriction, et cetera are detected, these specific diagnoses are labeled as “HFpEF masqueraders” and must be excluded [7]. Arrhythmias—most often symptoms due to specific cardiac diagnoses—are also described as “HFpEF” masqueraders, except for atrial fibrillation (AF) which is very frequently observed in patients with “HFpEF” symptoms [19, 20]. To be fair, the exclusion of “HFpEF masqueraders” at best results in the diagnosis “HFpEF” with unknown origin. However, there are several other cardiologic diagnoses like myocarditis, myocardial infarction with non-obstructive coronary artery disease, restrictive and infiltrative cardiomyopathies contributing to the dilemma of multiple components of etiological variety in patients with “HFpEF” symptoms [21,22,23,24,25,26]. Furthermore, non-cardiac diseases—especially pulmonary diseases, anemia, diabetes, systemic infections, consuming neoplasms, obesity, and frailty—complicate the multifactorial hodgepodge of “HFpEF” as a diagnosis [27,28,29,30,31,32].

Thus, the present expert proposal focuses on a diagnostic TTE workflow to overcome diagnostic challenges and to increase options for the detection of specific cardiac entities in patients with HF symptoms and preserved LVEF (Fig. 1). With respect to conventional and modern echocardiographic tools the authors try to highlight, why, when, and how to implement special echocardiographic features into clinical practice. In addition, the proposal underlines new aspects, which are presumably worth to be included into the workflow.

Fig. 1
figure 1

Scheme illustrating the echocardiographic workflow to characterize patients with “HFpEF” symptoms due to cardiac phenotypes

Full size image

TTE guiding to diagnose specific cardiac pathologies commonly reflects the initial contact of patients undergoing diagnostic cardiological procedures prior to laboratory findings, prior to invasive measurements, and sometimes even prior to interpretation of the electrocardiogram. Thus, the focus of a detailed morphological and functional characterization of cardiac entities by TTE plays a key role of cardiac diagnostics. With respect to morphological and functional alterations of cardiac chambers, the following cardiac phenotypes are outlined (Fig. 2):

  1. 1.

    The normal cardiac phenotype which can be associated with increased LV stiffness in patients with HFpEF symptoms,

  2. 2.

    The LV (= left ventricular) phenotype which is associated with morphological or functional abnormalities of the LV wall,

  3. 3.

    The LA (left atrial) phenotype which is solely associated with an increase of LA size and consecutive impairment of LA function,

  4. 4.

    The RA (= right atrial) and RV (= right ventricular) phenotype which is solely associated with morphological and functional abnormalities of right cardiac chambers,

  5. 5.

    The combination of LA and LV phenotype,

  6. 6.

    The combination of all three phenotypes and

  7. 7.

    Rare combinations of LA and RA/(RV) phenotype or LV and (RA)/RV phenotype.

Fig. 2
figure 2

Illustration of cardiac phenotypes in patients with “HFpEF” symptoms by echocardiography. The rows illustrate representative parasternal long and short axis views (pLAX, pSAX) and apical long axis views (aLAX) or apical 4-chamber views (4ChV) of the respective phenotypes. In A the normal cardiac phenotype is shown, in B the isolated LV phenotype (in compensated hypertensive heart disease—short history), in C the isolated LA phenotype (in paroxysmal lone atrial fibrillation), in D the isolated RA/RV phenotype (in chronic pulmonary hypertension due to repetitive pulmonary thromboembolism), in E the combined LA and LV phenotype (in compensated hypertensive heart disease—long history), and in F the combined LA, LV, and LA/RV phenotype (in amyloidosis)

Full size image

The introduction of isolated echocardiographic phenotypes is proposed with respect to diagnostic systematics to focus on the predominant morphological or functional cardiac abnormality. The classification of patients with “HFpEF” symptoms according to these morphological and functional cardiac entities is still challenging due to the complex interactions of pathophysiological factors on the heart. Hence, majority of patients with “HFpEF” symptoms have combinations of these echocardiographic phenotypes.

The characteristic echocardiographic phenotypes and their respective combinations are depicted with respect to the representative diseases in the central illustration (central illustration). In addition, the cardiac pathology and pathophysiology as well as typical representative diseases are highlighted to demonstrate the diagnostic pathway depending on echocardiographic morphological classification.

Targets of echocardiography in patients with heart failure and preserved ejection fraction

A proper LVEF assessment is the first diagnostic step to identify HF patients with preserved LVEF. The recommended quantitative LVEF assessment is mostly performed by biplane planimetry using the modified Simpson method [33]. The two-dimensional approach from the apical four- and two-chamber view often results in underestimation of LV cavity size by apical foreshortening. Nowadays, standardization of apical LV views can be ensured using triplane or multidimensional TTE [33,34,35]. Thus, these modern modalities should be generally preferred, especially in the presence of regional wall motion abnormalities and pathological LV geometry. According to recent recommendations the parameter LVEF is still predominantly used to differentiate between normal, preserved, mildly, moderately, or severely reduced LV function [2]. Total LV stroke volume (LVSVtot) measured by 2D planimetry or 3D volumetry reflects effective (systemic) LV stroke volume (LVSVeff) if patients with significant aortic and/or mitral regurgitation are excluded. Stroke volume can also be estimated using Doppler echocardiography in the LV and right ventricular (RV) outflow tract (LVOT, RVOT). LVSVtot and LVSVeff should be in the same range if no relevant valvular heart disease is present and can be counterchecked to verify hemodynamic plausibility of the measurements [33,34,35]. In case of limited acoustic windows, optimization of LV endocardial border detection by contrast echocardiography should be considered for LV volume assessment [33, 35, 36].

The first integral component of HF diagnostics by TTE in patients with “HFpEF” symptoms (predominantly dyspnea or disability) is to rule-out or to rule-in HF as the reason for clinical symptoms [2]. Cardiac morphology and function should be classified in relation to the cardiac phenotypes and their combinations to structure the echocardiographic workflow cardiac in patients with “HFpEF” symptoms (Fig. 2). If no abnormalities are detected by conventional echocardiography, a “normal cardiac phenotype” refers to a non-cardiac entity of the “HFpEF” symptoms (Figs. 3, 4). The pathophysiological reason for “HFpEF” symptoms in “normal cardiac phenotype” patients can be explained by the mismatch between oxygen supply and oxygen demand due to high output stages, chronotropic incompetence, and/or subclinical stages of unknown cardiac diseases. Non-cardiac disease-related “HFpEF” symptoms might be caused by a high cardiac output (CO) due anemia, hyperthyroidism, liver cirrhosis, infectious diseases, and/or obesity due to different pathophysiologies. Notably, the measurement of CO is mandatorily in patients with dyspnea and normal cardiac phenotype if LV hypercontractility and brady- or tachycardia is observed. The problem of the echocardiographic “normal cardiac phenotype” is to miss early echocardiographic signs related to LV and left atrial (LA) wall thickness, stiffness, and function. Thus, this scenario describes the diagnostic challenge to detect early or preclinical disease stages with subtle echocardiographic signs of pathological cardiac phenotypes.

Fig. 3
figure 3

Scheme of echocardiographic workflow describing the tool to rule out cardiac manifestations as the primary cause of “HFpEF” symptoms

Full size image
Fig. 4
figure 4

Scheme of the ratio between patients with noncardiac and cardiac HF symptoms (left)—scheme of the interactions/combinations between cardiac phenotypes in patients with “HFpEF” symptoms

Full size image

The proposed pathological echocardiographic phenotypes can be subdivided into a LV phenotype, a LA phenotype, and a combined right atrial (RA) and RV phenotype (Figs. 1, 2). The diagnostic aspects of these pathological phenotypes in patients with “HFpEF” symptoms are described in detail in the following paragraphs. They address the second step of TTE after the rule-in of HF as the cause of the symptoms. We believe that this proposed classification of echocardiographic phenotypes will help to improve the identification of a specific underlying diagnosis for HF. The third task of echocardiography in HF patients is to monitor specific treatment effects during follow up and to estimate individual patient’s prognosis.

The echocardiographic characterization of patient with “HFpEF” symptoms using recent “HFpEF” algorithms: what is useful?

“New” paradigms—e. g. the identification of a systemic proinflammatory state induced by comorbidities as the cause of myocardial structural and functional alterations—have been proposed to explain the genesis of “HFpEF” symptoms [10, 11] and new scores have been introduced to overcome the challenges of diagnostics in patients with “HFpEF” symptoms [13,14,15,16,17,18].

After LVEF determination the second pivotal finding by TTE in patients with “HFpEF” symptoms is the increased ratio of E/E´ (E = early mitral flow velocity, E´ = early tissue Doppler lengthening velocity of the myocardium near to the mitral annulus) [4, 7]. Interestingly, increased NT-proBNP- and BNP levels (NT-proBNP = N-terminal-pro brain natriuretic peptide, BNP = brain natriuretic peptide) are not necessarily measured in patients characterized by these echocardiographic criteria. Invasive hemodynamic measurements of PCWP, LVEDP (PCWP = pulmonary capillary wedge pressure, LVEDP = left ventricular end diastolic pressure), the time constant of ventricular relaxation τ and LV chamber stiffness are proposed as an alternative to prove cardiac cause of “HFpEF” symptoms, but often not performed [2, 4, 7]. Thus, echocardiography should implement additional parameters beside the documentation of E´, E/E´, and systolic pulmonary artery pressure (sPAP) to characterize diastolic dysfunction (DD) in patients with “HFpEF” symptoms [37,38,39,40,41,42].

According to recent recommendations, the second target in patients with “HFpEF” symptoms is the detection of the “masqueraders” (Fig. 5), as described above [4, 7]. The list of “masqueraders” might be complemented by all specific diagnoses which can be detected by a comprehensive echocardiography to minimize the usage of “HFpEF” with unknown origin as a final diagnosis. In addition, the underlying diagnoses in the presence of arrhythmias—including AF—should also be clarified in patients with HF symptoms and preserved LVEF.

Fig. 5
figure 5

Scheme of the echocardiographic workflow in the presence of “HFpEF masqueraders”. “HFpEF masqueraders” are specific detected diagnoses in patients with “HFpEF” symptoms

Full size image

After ruling out patients with non-cardiac causes and excluding those with a known specific cardiac diagnosis the remaining cohort of patients is composed of patients with hypertensive heart diseases, infiltrative/storage diseases, and unknown specific diagnoses. Because clinical and echocardiographic abnormalities can be observed in subgroups of this patient cohort several phenotypes have also been introduced in the literature, i. e. phenotypes associated with LA abnormalities, with pulmonary hypertension and pulmonary disease in the presence of RV dysfunction, with obesity combined with cardiac abnormalities, and with the ischemia/microvascular dysfunction [39, 43,44,45,46]. However, several further entities like aging, myocardial LA and LV stiffness, systemic hypertension, high output stages, and lone AF can be added illustrating the diversity within this remaining cohort of patients with HF symptoms and preserved LVEF.

Thus, we like to propose the following phenotypes, solely described by echocardiographic findings: the LV phenotype, the LA phenotype, and the combined RA and RV phenotype (Fig. 2).

The LV phenotype: abnormalities of LV volumes, LV function, LV geometry, and LV mass

Conventional echocardiographic analyses of LV wall thickness, LV volumes and LVEF

This phenotype addresses LV abnormalities characterized by TTE. Despite LV planimetry or LV volumetry is possible in native 2D- and 3D echocardiography with adequate image quality endocardial contour delineation can often be improved by LV opacification using contrast to avoid underestimation of LV volumes [33,34,35,36, 47]. Modern modalities of triplane and 3D echocardiography are strongly recommended to avoid foreshortening which leads to smaller LV volumes and underestimation of LV function. Triplane and 3D echocardiography modalities are particularly useful in patients with regional wall motion abnormalities and pathological LV geometry.

The conventional echocardiographic M-Mode and 2D parameters characterizing LV dimensions, LV wall thickness, relative wall thickness (RWT), LV mass are most useful and reproducible in normal LV geometry and during normal cardiac conditions [33, 48]. If LV abnormalities are detectable and if the acoustic window is appropriate, LV wall thickness, LV volumes, LV wall stress and LV remodeling index can also be assessed by 3D echocardiography to distinguish between normal conditions, LV remodeling, concentric and eccentric LV hypertrophy [33,34,35,36, 47]. Standardization of M-Mode measurements can be achieved by anatomical M-Mode in biplane image acquisition to provide appropriate sectional planes through the LV center rather than LV secants, and LV measurements perpendicular to the LV long axis at the level of the mitral valve (MV) leaflet tips. A high spatial resolution using the best acoustic window usually enables the exclusion of right and left ventricular trabecula for appropriate delineation of LV wall diameters.

Strain analyses of GLS and myocardial work

Specific echocardiographic modalities should be used to demarcate subclinical and/or subtle findings as a screening tool in this predefined patient cohort. Global longitudinal strain (GLS) is nowadays an established parameter to characterize LV function as well as prognosis [49, 50]. Global circumferential and global radial strain (GCS, GRS) are also detectable by speckle tracking echocardiography, but are still not generally used. Regional myocardial dysfunction due to local ischemia, scar, hypocontractility or dyssynchrony can be visualized and quantified by the segmental 2D strain curves, which are often showing disease specific deformation patterns. When compared to the conventional LVEF assessment, the quantification of GLS is much more sensitive for the detection of subtle impairment of LV function like in early stages of infiltrative/storage diseases—especially in patients with amyloidosis—as many pathologies tend to primarily affect the longitudinal LV deformation [51, 52]. The conventional evaluation of LV longitudinal deformation by M-mode measurement of mitral annular plane systolic excursion (MAPSE) is a valid surrogate marker for GLS, but without regional/segmental information, thus GLS by 2DS should be preferred [53]. The different strain distribution patterns are associated with different underlying pathologies. The presence of subclinical fibrosis, e.g. in hypertensive heart disease, toxic, immune, inflammatory, infiltrative, metabolic, genetic and endomyocardial myocardial pathologies, typically affects more the basal and mid LV segments, [7, 49, 54], while the “apical sparing” pattern with a preserved regional strain only in the apical segments is typically seen in patients with cardiac amyloidosis. Predominant alterations of regional circumferential and radial strain are observed in patients with myocardial edema due to myocarditis or cardiotoxic agents [55,56,57]. Thus, the analysis of the systolic LV contraction patterns should include the analysis of all interacting compounds of LV deformation.

LV function must be considered in relation to LV contraction, which is strongly dependent on pre- and afterload. Myocardial work can be non-invasively analyzed by pressure–strain loops to (partially) overcome the load-dependent limitations of LVEF and GLS. The parameter global work index (GWI), global constructive work (GCW), global wasted work (GWW), and global work efficiency (GWE) can be determined, if blood pressure is added to the strain analysis measured simultaneous to the TTE [58, 59]. The pressure–strain loop area reflects myocardial metabolic demand and oxygen consumption providing information about myocardial energetics [58, 59]. In patients with “HFpEF” symptoms and normal cardiac phenotype or LV phenotype differentiation between hypertensive and coronary heart disease is facilitated because myocardial work incorporates deformation and loading conditions into the analysis [58,59,60,61]. Thus, strain patterns with regional decreased longitudinal strain correspond to myocardial work patterns, whereas normalization of myocardial work patterns can be observed in hypertensive heart disease.

Analysis of LV stiffness, LV elastance, and arterial elastance

Doppler and tissue Doppler echocardiographic parameters have been evaluated in comparison to invasive measurements to estimate diastolic function and LV filling pressures favoring the echocardiographic parameter of the lateral E´ [62]. However, the estimation of LV filing pressures solely by echocardiographic parameters has several limitations, especially in patients with regional LV wall motion abnormalities [63, 64]. In patients with “HFpEF” symptoms, a further approach to detect subtle alterations of the LV myocardium might be the assessment of diastolic stiffness, myocardial relaxation, and resulting LV contractility by the analysis of LV end-diastolic volume (LVEDV) in relation to LVEDP [65,66,67]. The invasive LV pressure–volume relationship is recognized as the gold standard to analyze these parameters. It requires the generation of several pressure–volume loops in different preload states. The approximately linear relationship of their end-systolic pressure–volume points represents the end-systolic pressure volume relationship (ESPVR). The slope of the ESPVR defines the end-systolic elastance (Ees), which serves as a measure of myocardial contractility (Fig. 6). Compared to LVEF or GLS, Ees exhibits a significant correlation to increased global myocardial afterload characterized by arterial elastance (Ea). The increased steepness and left-shift of ESPVR with increasing inotropy represents the myocardial homeometric autoregulation known as Anrep effect [68,69,70]. Graphically, Ea is reflected by the negative slope between the end-systolic pressure–volume point and the end-diastolic volume at a LV pressure of 0 mmHg (Fig. 6). Ea incorporates arterial compliance and vascular resistance but is also affected by heart rate. In healthy individuals, Ea and Ees are well connected to provide optimal mechanical ventricular-arterial coupling with an Ea/Ees ratio between 0.5 and 1.0 [71]. The indirect measurement of Ees has been developed using a single-beat non-invasive approach to calculate Ees and Ea from pressure and volume measurements during one cardiac cycle [72, 73]. Specifically, the pressure–volume loop is computed from LVEDV and LV end-systolic volume (LVESV), measured by TTE, and blood pressure, measured non-invasively. For Ees calculation, systolic time intervals, defined by pre-ejection period and systolic ejection time, are determined from pulsed-waved Doppler in the LVOT with simultaneous ECG recording. Non-invasive single beat pressure–volume analysis has been applied to study the pathophysiology, prognosis [65,66,67] as well as the effect of therapeutic HF interventions [74,75,76]. However, due to the complexity of non-invasive Ea and Ees determination this method is not commonly used and needs further evaluation.

Fig. 6
figure 6

Scheme of a non-invasive LV pressure volume loop to explain the calculations of the end-systolic pressure volume relationship (ESPVR) by its slope which defines the end-systolic elastance (Ees), and by the slope between the end-systolic pressure–volume point and the end-diastolic volume at a LV pressure of 0 mmHg, which defines arterial elastance (Ea) representing LV afterload. EDP, end-diastolic pressure; EDPVR, end-diastolic pressure volume relationship; EDV, end-diastolic volume; ESV, end-systolic volume; V0, x-axis intercept of Ees

Full size image

In HF patients with reduced ejection fraction (HFrEF), Ees is reduced due to impaired myocardial contraction, while Ea might be elevated despite normal blood pressure. Hence, Ea/Ees increases in HFrEF patients, indicating ineffective arterial-ventricular coupling (Fig. 7) [66, 71]. An important hemodynamic characteristic in patients with normal LV size and normal or increased LV wall thickness and “HFpEF” symptoms (normal cardiac phenotype and LV phenotype) and in HFrEF patients is the response in LV ejection to changes in afterload. In HFrEF patients, LV stroke volume is more sensitive for afterload changes than in the non-failing heart or in this defined phenotype of patients with “HFpEF” symptoms—which can be labelled as an increased LV stiffness phenotype—because the slope of ESPVR (Ees) is flat (Fig. 7). Thus, afterload reduction, for instance using vasodilators, reduces Ea, which is frequently elevated in HFrEF patients, but lowers blood pressure only slightly. Thus, LVSVtot increases. In increased LV stiffness phenotype Ees is high or normal because LV contractility is preserved (Fig. 7). Vasodilators reduce Ea and blood pressure significantly but LVSVtot would remain relatively constant [77]. Both patients with hypertensive heart disease and with “HFpEF” symptoms show increased Ea values, and Ees increases to compensate for elevated Ea. Thus, ventricular-arterial coupling remains in the normal range. When HF symptoms occur, patients with “HFpEF” symptoms are characterized by increased myocardial stiffness and reduced myocardial contractility—both impact Ees—keeping Ees at a high level. Hence, Ees does not specifically characterize patients with “HFpEF” symptoms nor does it predict outcome [65].

Fig. 7
figure 7

Characteristic pressure–volume loops in patients with heart failure with reduced (HFpEF; top)and preserved (HFpEF; bottom) ejection fraction. In HFrEF patients, end-systolic elastance (Ees) is reduced and LV volumes are increased, thus LV pressure–volume loop is right-ward shifted. In patients with “HFpEF” symptoms and increased LV stiffness, LV end-diastolic volume is reduced due to concentric remodeling, which leads to a leftward shifted end-diastolic pressure volume relationship (EDPVR), i. e. increased LV filling. Ea is frequently elevated, Ees normal or increased. The pressure–volume curves of a non-failing heart with its EDPVR, Ees and Ea is drawn with dotted lines. The schemes explain the increase of LV stroke volume (LVSVtot) in HFrEF patients and the minor changes of LVSVtot in patients with “HFpEF” symptoms with afterload reduction

Full size image

The single-beat method to assess Ees and Ea allows also to calculate the end-diastolic pressure–volume relationship (EDPVR) (Fig. 6) [78]. EDPVR is built from echocardiographic measurement of LVEDVs and the calculation of LVEDP, derived from Doppler and tissue Doppler (E/E’) parameters [78]. Thus, the increased LV stiffness phenotype is also characterized by increased diastolic stiffness and impaired myocardial relaxation, leading to reduced LVEDVs at similar LVEDPs compared to patients with hypertension without HF symptoms and healthy controls [79]. In consequence, the EDPVR curve is leftward-shifted in patients with “HFpEF” symptoms; in contrast, EDPVR is rightward-shifted in HFrEF patients compared to non-heart failure patients (Fig. 7). Both alterations of EDPVR have prognostic implications [67, 80].

In summary, non-invasive single-beat analysis of the LV pressure–volume relationship is based on proper measurements by TTE in combination with simultaneous non-invasive blood pressure measurements [72, 73, 78]. It allows to evaluate LV systolic and diastolic function, LV afterload, and ventricular–arterial coupling. The pressure–volume analysis of the LV provides important insights into the specific pathophysiology and hemodynamics in patients with “HFpEF” symptoms defining a LV stiffness sub-phenotype. The analysis of LV stiffness might be helpful to detect subclinical stages of cardiac diseases, but still needs validation in further clinical studies.

Table 1 summarizes the echocardiographic parameters, which are important to differentiate between the normal cardiac phenotype and the LV phenotype. The parameters are listed including normal ranges and cut offs [33, 72, 81,82,83,84,85,86,87,88], methodological aspects of TTE determination, their importance to be determined, and their reasons, why it is worth to determine the respective parameter in clinical routine. The intention of the tables is to list these important echocardiographic parameters for the sake of completeness as an easy reference for the user. Thus, the authors do not mandatory propose to measure of all these parameters.

Table 1 Echocardiographic parameters characterizing patients with heart failure (HF) symptoms and normal or preserved left ventricular ejection fraction (LVEF) in normal cardiac phenotype and the left ventricular (LV) phenotype
Full size table

Figure 8 summarizes the echocardiographic workflow in patients with “HFpEF” symptoms presenting normal cardiac phenotype or LV phenotype.

Fig. 8
figure 8

Scheme of the echocardiographic workflow in patients with “HFpEF” symptoms to characterize the echocardiographic LV phenotype and to exclude normal cardiac phenotype

Full size image

The LA phenotype: abnormalities of left atrial (LA) volumes and LA function (diastolic dysfunction and increased LV filling pressures)

Increased stiffness of the LV wall impedes LV filling and increases LA pressure. This leads to the distinct morphological and functional LA alterations: the “LA phenotype” (Fig. 2).

The chronic increase in LA pressure results in progressive LA dilatation with increased volumes. Traditionally, LA size is measured in the anterior–posterior dimension from parasternal views. However, in pathologic states the LA frequently enlarges predominantly in the long-axis direction (base-apex). Therefore, the assessment of LA volume during pathological conditions should at least be performed by biplane planimetry from apical views. Volumetry of the complete LA by 3D TTE is preferred because centering of the longitudinal LA dimensions in standardized apical LV views may lead to foreshortening and systematic over- or underestimation of LA volumes when compared to 3D data sets (Fig. 9).

Fig. 9
figure 9

Illustration about the necessity for 3D-LA assessment. A presents the enface view from the LA cavity to the MV and TV. In B the apical 3D views (4ChV, 2ChV—apical 4- and 2-chamber view; aLAX—apical long axis view) within the 3D data set are centered to the LV illustrating the axis deviation of the LA mainly in the aLAX. In C the LA long axis is centered to document the deviation of LV an LA center line. In D a multisclice presentation of the 3D data set illustrate the complete acquisition of the LA volume. In E the LA alignment for LA volume determination is shown. In F the diastolic LA volume, in G the systolic LA volume is presented by a 3D volume calculation

Full size image

Biplane LA volume measurement indexed to body surface area (BSA) is the currently recommended standard. The obtained LA volume index (LAVI) has been associated with cardiovascular outcomes [33]. A maximum LAVI > 34 ml/m2 documents an increased LA volume [2, 4, 7, 33, 35, 48]. Different echocardiographic parameters of LA function can be described by conventional LA volume measurements: the maximum LA volume (LAVmax) measured at LV end systole (onset of MV opening), the minimum LA volume (LAVmin) at LV end diastole (prior to MV closure), and the LA volume prior to LA contraction at the onset of P wave (LAVpreA). The echocardiographic parameters characterizing LA function and the respective calculations are listed in Table 2 [53, 89].

Table 2 Echocardiographic parameters characterizing LA function and their calculations by conventional LA volume and LA strain determination
Full size table

Of all LA volume measurements, maximum LAVI (LAVImax) has the most important prognostic value [90,91,92]. However, minimum LAVI (LAVImin) might be better to characterize LA function during sinus rhythm because it may better reflect elevated LVEDP and PCWP than LAVImax [93, 94]. The main practical challenge to assess LAVImin is the higher error-proneness in comparison to LAVImax due to inappropriate LA centering in the respective sectional 2D planes [91, 95,96,97]. The geometric assumptions of the LA size by 2D echocardiography usually cause an underestimation of LA volumes in comparison to LA volume measurements by 3D echocardiography [53, 91, 97, 98] underlining the necessity of 3D echocardiography for a better LA volume assessment. With respect to 3D echocardiography a higher threshold for the maximum LAVI above 40 ml/m2 has been proposed (instead of 34 ml/m2 by 2D) [53, 91, 97, 98]. In chronic AF threshold for maximum LAVI are higher than a LAVImax of 40 ml/m2 in the literature [2, 91].

LA function can presumably be more objectively assessed by analysis of longitudinal LA strain. However, atrial strain is to a large extent determined by ventricular strains and modulated by the atrial and ventricular volumes. LA strain (LAS) is analyzed using apical 2D four-chamber views by triggering the beginning of the QRS complex [53]. Global average LA reservoir (LASr), LA conduit (LASr minus average LAS value at the onset of the p-wave), and LA contraction strain (LASct = LASr − LA conduit strain) are extracted from the respective strain curves. In patients with atrial fibrillation, LAS analysis was limited to the investigation of LASr and LAScd, as proposed by the recent EACVI recommendations [53, 85, 89, 90].

LAS provides valuable additive information for the differential diagnosis between infiltrative/storage diseases like amyloidosis, hypertrophic cardiomyopathy and other types of LA and LV wall thickening including edema [99,100,101] (Fig. 10). Two distinct mechanisms can theoretically lead to LA dysfunction in infiltrative/storage diseases: firstly, the increased LA pressure induced by LV dysfunction and secondly, the direct accumulation of pathological deposits in the LA wall. Thus, hypertrophic cardiomyopathy as a primary LV disease affects LA dysfunction to a lesser degree than e.g. infiltrative/storage diseases. The diagnostic strength of LAS has been shown to be better than classical LV strain features like “apical sparing” [99,100,101]. Recent data suggests that LA dysfunction plays an essential role in the clinical course of patients with “HFpEF” symptoms due to early stages of infiltrative/storage diseases followed by DD and AF [99,100,101].

Fig. 10
figure 10

Illustration about advanced strain assessment in patients with “HFpEF” symptoms. In A, a quad screen of 4ChV (4-chamber view) documents LV strain analysis by presenting the tracking LV area, the respective strain graphs of the regional LV strain, the corresponding maximum strain values, and the color-coded curved strain M-Mode in a patient with hypertensive heart disease. Reduced mid-basal strain and normal mid-apical strain values are documented. In B, a comparable quad screen of the 4ChV for LA strain analysis is shown documenting LA strain values higher than 16%. In C, a comparable quad screen of 4ChV for LV strain analysis in a patient with Fabry’s disease illustrates reduced reginal LV strain values in all LV segments. In D, a comparable quad screen of 4ChV for LA strain analysis in the Fabry patient disease illustrates the possibility of early detection of LA dysfunction by the reduced LA strain values

Full size image

The clinical symptoms and the impaired exercise capacity in patients with “HFpEF” symptoms can mainly be explained by DD due to increased LVEDP at rest and/or during exercise. Historically, DD has initially been graded mainly according to the pw transmitral inflow E/A ratio. A high E/A ratio ≥ 2 with a short E-wave deceleration time (Edt) < 140 ms indicates severely impaired diastolic function due to increased LVEDP, however rarely documented at compensated stages. The frequently observed E/A ratio < 0.8–1 with a normal or prolonged Edt can be classified as abnormal relaxation and may indicate a slightly elevated LVEDP with normal LA pressure and no evidence for pulmonary congestion at rest [102,103,104]. However, the diagnostic value of the E/A ratio is severely limited by its bimodal behavior with the inability to distinguish between a normal and elevated filling pressure if the E/A is between 0.8 and 2. This limitation may be overcome by integrating pw Doppler information from pulmonary venous inflow. The duration of atrial reversal flow in the pulmonary vein is longer than the duration of forward transmitral atrial flow (A-wave) in the presence of elevated LVEDP [102,103,104]. Elevated LV filling pressures will also progressively shorten isovolumic relaxation time (IVRT) < 60–80 ms, which is typically obtained by a cw Doppler placed in-between the transmitral inflow and LV outflow [105]. The recommended and more practical alternative to estimate LVEDP can be performed by tissue Doppler imaging of the E´ wave in the basal septal and lateral LV segments. An average E/E’ ratio > 15 calculated with the mean E´ velocity of the basal septal and lateral LV segments indicates a significantly elevated LVEDP and is a major criterion in the HFA-PEFF score. However, several coexisting conditions such as the presence of left bundle branch block, valvular regurgitation, regional wall motion abnormalities affect the E/E’ ratio. If E/E´ratio is measured within the grey zone between 9 to 14, further echocardiographic parameters can be determined to estimate the degree of DD [4, 34, 41, 55, 84, 102,103,104,105,106,107]. The elevated LVEDP is transmitted through the pulmonary vasculature and will chronically lead to an increase in pulmonary pressure with mild to moderate postcapillary pulmonary hypertension. This can be identified by an increase in the tricuspid peak regurgitant (TR) velocity > 2.8 m/s reflecting an estimated systolic pulmonary artery pressure (sPAP) > 35 mmHg. These hemodynamic results may importantly vary with changing volume conditions.

All echocardiographic parameters defining DD are changing with age. An age-related decrease in the E/A ratio, mitral annular velocities, septal and lateral E’ has been observed in many population-based studies [108,109,110]. Also, mitral annular velocities decrease significantly during a lifetime. Elevated LV filling pressures and DD are not synonymous but refer to abnormal mechanical LV properties. Thus, patients with “HFpEF” symptoms usually have a higher sPAP, E/E’ ratio, and LAVI than patients with LV wall thickening presumably due to arterial hypertension without HF signs [111].

Patients with “HFpEF” symptoms and sole LA abnormalities are rare.

Table 3 summarizes the echocardiographic parameters, which are important to characterize the LA phenotype. The parameters are listed including normal ranges and cut offs [84, 85, 89, 102, 103, 110,111,112,113], methodological aspects of TTE determination, their importance to be determined, and their reasons, why it is worth to determine the respective parameter in clinical routine. Specific parameters can be helpful to detect functional LA abnormalities—especially in subclinical stages of possible cardiac diseases.

Table 3 Echocardiographic parameters characterizing patients with heart failure (HF) symptoms and normal or preserved left ventricular ejection fraction (LVEF) in left atrial (LA) phenotype
Full size table

Figure 11 summarizes the echocardiographic workflow in patients with “HFpEF” symptoms presenting the LA phenotype.

Fig. 11
figure 11

Scheme of the echocardiographic workflow in patients with “HFpEF” symptoms to characterize the echocardiographic LA phenotype

Full size image

The RA/RV phenotype: abnormalities of right atrial (RA) volumes and RA function as well as of right ventricular (RV) volumes, RV function, and RV geometry

In patients with “HFpEF” symptoms abnormalities of RA and RV morphology and function can be induced by several factors. The constellation of pathological echocardiographic findings and its interpretation encompasses RV wall thickening, RA and RV volume, RV contractility and deformation as well as tricuspid and pulmonary valvular function to differentiate between RV volume and RV pressure overload [114, 115]. Increased RV contractility combined with increased RV volume primarily refers to compensated staged with increased RV volume load, increased RV contractility combined with RV wall thickening and normal RV volumes primarily refers to compensated stages with RV pressure load. Thus, these echocardiographic features can lead to diagnoses like tricuspid regurgitation or left–right shunts due to increased RV volume load or to diagnoses like primary or secondary pulmonary hypertension due to increased RV pressure load [114, 115]. The interpretation of the echocardiographic findings is more challenging and difficult at decompensated stages.

Due to secondary pulmonary hypertension or infiltrative/storage diseases the RA/RV phenotype is often linked to the LA phenotype and the LV phenotype. The isolated RA/RV phenotype in patients with “HFpEF” symptoms normally characterizes primary pulmonary hypertension due to specific lung diseases or rare diseases like arrhythmogenic RV dysplasia. Especially in this scenario, the non-invasive determination of pulmonary vascular resistance by echocardiography is recommended (Table 4). The combination of the RA/RV phenotype with the other echocardiographic phenotypes is observed in almost all other cardiac diseases.

Table 4 Echocardiographic parameters characterizing patients with heart failure (HF) symptoms and normal or preserved left ventricular ejection fraction (LVEF) in the combined right atrial (RA) and right ventricular (RV) phenotype
Full size table

The determination of the echocardiographic RV parameters presented in Table 4 should be considered if RV abnormalities are observed in patients with “HFpEF” symptoms [114, 115]. RV wall thickness, the tricuspid annular plane systolic excursion (TAPSE), sPAP, and RV strain are relatively robust parameters [33, 102, 103, 114, 115] whereas estimations of RV volumes resulting on 2D measurements are often error prone due to problems to standardize the sectional planes through the right ventricle [116,117,118]. Thus, 3D RV volume assessment is always recommended if possible—and 2D distance measurements to characterize RV volume should be avoided [33, 118]. Measurements of the RV fractional area change (FAC) determined by 2D planimetry of the end diastolic and end systolic RV area in an apical four chamber view focused to the right ventricle are often misleading due to methodology of standardization [116,117,118]. Thus, estimation of RV function should be performed by the assessment of RV ejection fraction (RVEF) by 3D echocardiography or by RV strain analysis (Figs. 12, 13).

Fig. 12
figure 12

Illustration about the necessity for 3D-RV assessment—normal conditions. A presents the enface view from the LV cavity to the MV and TV. In B, the apical 3D views (4ChV, 2ChV—apical 4- and 2-chamber view; aLAX—apical long axis view) within the 3D data set are centered to the LV illustrating the standardized 4ChV by representative angular distance of 60° between aLAX, 2ChV, and 4ChV. In C, oblique 4ChV are shown to illustrate that even during normal conditions the deviation of non-standardized 4ChVs for 2D measurements is within a Δ of 25°. In D, a multisclice presentation of the 3D data set illustrate the complete acquisition of the RV volume. In E, the RV alignment for RV volume determination is shown. In F the diastolic RV volume, in G the systolic RV volume is presented by a 3D volume calculation

Full size image
Fig. 13
figure 13

Illustration about the necessity for 3D-RV assessment -pathological conditions. A presents the enface view from the LV cavity to the MV and TV. In B the apical 3D views (4ChV, 2ChV—apical 4- and 2-chamber view; aLAX—apical long axis view) within the 3D data set are centered to the LV illustrating the standardized 4ChV by representative angular distance of 60° between aLAX, 2ChV, and 4ChV. In C oblique 4ChV are shown to illustrate that during pathological RV conditions the deviation of non-standardized 4ChVs for 2D measurements is within a Δ of 65°. In D a multisclice presentation of the 3D data set illustrate the complete acquisition of the RV volume. In E the RV alignment for RV volume determination is shown. In F the diastolic RV volume, in G the systolic RV volume is presented by a 3D volume calculation

Full size image

RV strain should be measured in a RV-focused apical four chamber visualizing the RV free wall in the center of the scanning sector. Sector size should be minimized to increase frame rate for tissue Doppler measurements of the free RV wall [49, 50, 119]. Speckle tracking of the RV wall also like of the LV wall requires frame rates between 40 and 80/min. Then, RV strain can be analyzed in the basal, medial, and apical RV segments. In addition to RV strain, RA strain can be analyzed semi-automatically after manual determination of the endocardial borders. Global RA reservoir strain was assessed by analyzing the maximum excursion of the respective strain curve representing average peak strain values of all RA segments [50, 119, 120]. In addition, RA strain during passive RV filling (RA conduit strain) and during peak atrial contraction (RA active strain) can be determined. Comparable to LA strain RA conduit strain can be calculated by the difference of RA global strain and average RA strain value at the onset of the p-wave, whereas RA active strain can be calculated by the difference of RA global strain and RA reservoir strain. Reference values of mean RA reservoir strain, RA conduit and contraction strain are 44.9 ± 11.6%, 27.1 ± 9.5%, and 17.0 ± 5.9%, respectively [121, 122].

Table 4 summarizes the echocardiographic parameters, which are important to characterize the RA/RV phenotype. The parameters are listed including normal ranges and cut offs [33, 85, 87, 91, 113, 117, 118, 126,127,128], methodological aspects of TTE determination, their importance to be determined, and their reasons, why it is worth to determine the respective parameter in clinical routine. Specific parameters can be helpful to detect functional RA and RV abnormalities—especially in combination with pathological LA and LV findings.

Figure 14 summarizes the echocardiographic workflow in patients with “HFpEF” symptoms presenting the RA/RV phenotype.

Fig. 14
figure 14

Scheme of the echocardiographic workflow in patients with “HFpEF” symptoms to characterize the echocardiographic RA/RV phenotype

Full size image

The role of diastolic stress echocardiography in patients with “HFpEF” symptoms

Sensitivity of resting echocardiographic parameters is limited to detect DD as the cause of “HFpEF” symptoms. Only 34%–60% of DD patients characterized by TTE have been confirmed with invasively proven DD in patients with preserved LVEF [62, 126]. Thus, artificial intelligence is proposed to predict HF in asymptomatic patients with normal cardiac phenotype, changes of diastolic filling properties and diastolic abnormalities with structural cardiac findings [127]. Thus, asymptomatic patients at rest might provide “HFpEF” symptoms due to DD during stress. The detection of DD development during physical diastolic stress echocardiography [2, 7] requires a proper standardization of the acquisition of Doppler spectra with respect to comparable breathing state and comparable sample volume positions at the respective stress levels. Besides these methodological challenges, increasing stress levels are normally limited due to the fusion of E- and A-wave with increasing heart rate. In consequence, DD detection by diastolic stress echocardiography should be confirmation by invasive hemodynamic exercise testing [7, 128, 129]. However, invasive hemodynamic studies are rarely performed to confirm an elevated PCWP in patients with “HFpEF” symptoms in clinical practice. Adding the criterion “exercise E/E’ > 14” to detect DD by diastolic stress echocardiography improved the sensitivity to 90%, but decreased specificity from 92 to 71% [126]. The hemodynamic response to exercise in patients with “HFpEF” symptoms was analyzed documenting that E/E’ at rest was higher in HF patients with small LV size and normal LA size group than in healthy controls, but that this difference did not reflect a difference in PCWP [128]. Furthermore, E/E’ did not significantly increase during stress in both groups and did not correlate with invasively measured PCWP at peak stress. It was concluded that increased arterial stiffness is presumably one of the key mechanisms of “HFpEF” symptoms [128, 129]. Stress testing in cardiology, however, is mainly used to detect coronary artery disease by stress-induced LV wall motion abnormalities, but it might obviously contribute to distinguish between cardiac and non-cardiac causes of “HFpEF” symptoms [2, 7].

Summary

The present proposal provides an echocardiographic focused algorithm for patients with HF symptoms and preserved LVEF. In line with recent recommendations [2, 7], distinct cardiac phenotypes characterized by TTE are introduced as predominant gatekeepers for further diagnostic steps to clarify the specific underlying diagnoses. The algorithm starts with ruling out patients with non-cardiac causes for HF symptoms because edema has a low sensitivity of 20% and dyspnea on exertion a low specificity of 17% to be caused by cardiac failure [1]. In addition, a comparison of invasively measured PCWP with clinical HF classification according to the New York Heart Association (NYHA) showed in 61% (NYHA II) and in 14% (NYHA III) a normal PCWP < 16 mmHg, and biomarkers like NT-pro BNP showed no correlation to PCWP at rest [130]. Thus, symptoms suggestive of “HFpEF” may in some patients represent non-HF comorbidities” [131, 132]. The morphological classification of patients with “HFpEF” symptoms into the normal cardiac phenotype, the LV phenotype, the LA phenotype, and the combined RA/RV phenotype is the main challenge of the algorithm to detect the underlying pathologies responsible for “HFpEF” symptoms. The systematic echocardiographic approach emphasizes conventional, but uncommon parameters like Ees and Ea, but also modern parameters like RV volume, LA strain and myocardial work—determined by 3D echocardiography and deformation imaging—to improve echocardiographic detection of specific cardiac diagnosis in patients with “HFpEF” symptoms. The assessment of LA and LV stiffness might be particularly useful to detect subclinical pathologic entities in an apparently normal cardiac phenotype with borderline morphological abnormalities. For example, LA dysfunction might precede LA remodeling which is better predicted by LASr enables to distinguish asymptomatic patients with DD from those with HF symptoms [130]. Presumably, additional diastolic stress testing by dynamic stress echocardiography can further improve the diagnostic accuracy. In summary, the proposed diagnostic TTE workflow highlights the necessity to detect a specific diagnosis of non-cardiac and cardiac diseases. The common use of “HFpEF” as a diagnosis should be avoided.