Left ventricular (LV) hypertrophy (LVH) is a consequence of an underlying genetic or acquired condition and it is accompanied by alterations in cardiac function and haemodynamics. The differential diagnosis between LVH due to physiological adaptation or underlying pathology can be challenging.

Non-physiological left ventricular hypertrophy (LVH) regardless to the underlying cause leads to important cardiovascular complications such as atrial fibrillation, diastolic and systolic heart failure [12], and it is associated with increased risk for all-cause morbidity and mortality [13].

Dilated phenotypes are a heterogenous group characterised by large LV cavities with eccentric remodelling or hypertrophy and impaired contractility. Such phenotypes can be a response to abnormal loading conditions typically in valvular disease or hypertension, severe coronary or congenital disease [39] or predominantly confined to heart muscle like in inherited or acquired cardiomyopathies. Transthoracic echocardiography is used as a first line imaging tool for identifying and description of the phenotype. It typically shows global left or biventricular hypokinesis with or without regional wall motion abnormalities. Ventricular and atrial dilatation, intracardiac thrombi and functional mitral regurgitation due to annular dilatation might also be noted. Doppler parameters can assist in quantifying valvular abnormalities and the severity of diastolic dysfunction [40]. CMR is superior to echocardiography, it provides accurate assessment of ventricular volumes, wall thickness and contractile function, as well as tissue characterisation [41]. Typical patterns of subendocardial, mesocardial and subepicardial LGE distribution reflect underlying pathophysiological processes and reveal aetiology with such a level of confidence that in many cases myocardial biopsy may be omitted. CMR relaxometry can indicate myocardial necrosis, scarring, focal and diffuse replacement interstitial fibrosis, whereas high signal T2 intensity or increased T2 mapping values suggest myocardial oedema and inflammation [42]. The 4D-Flow technology allows the assessment of flow-based forces and their altered haemodynamic effects on the myocardial wall with a potential to become a maker of progressive adverse cardiac remodelling [43].

The longitudinal trajectories of LV ejection fraction vary between the two conditions in a very sensitive way. When adequate follow-up is available, repeat echocardiograms in HFpEF can detect a maximum of 2–5% fell in EF over 5 years, with a larger fall in the presence of coexisting coronary artery disease [61, 62]. A similar analysis, conducted over a follow-up period up to 15 years in HFrEF patients, observed an inverted U-shaped relation whereby EF indeed increased over the initial 10 years but slowly declined afterward [63]. Comparable data should be available when looking at systolic deformation, particularly in the circumferential or radial direction, either assessed using CMR or echocardiography [64]. Thus, clear discrepancies in LV systolic performance between HFpEF vs. HFrEF populations long-term trajectories make unlikely a mechanistic continuum between the two conditions, which can be represented as distinct HF phenotypes [65].

More similarities, than discrepancies, instead, can be detected when looking at diastolic dysfunction in HFpEF vs. HFrEF patients. Diastolic dysfunction contributes to exercise intolerance, both in systolic and primary diastolic dysfunction. In both conditions, in fact, diastolic impairment limits exercise tolerance before resulting in symptoms at rest [66].

In the absence of mitral stenosis, the two major factors that determine the early diastolic mitral valve pressure gradient and the rate of LV filling are the rate of LV relaxation and the LA pressure at the time of mitral valve opening [67]. With exercise, normally, there is a fall in early diastolic LV pressure that results in an increased early diastolic pressure gradient, without an increase in LA pressure to abnormal levels. This increment in gradient, obtained with no increment in LA cavity pressure, produces an improvement in the rate of early diastolic LV filling [68]. When HF ensues, however, peak early diastolic filling rate and diastolic mitral valve pressure gradient increase as during normal exercise but with a different mechanism. Relaxation, in fact, which is slower at rest in HF, is further slowed down during exercise, while early diastolic LV pressure increases [69]. Thus, the increased rate of early diastolic LV filling and the mitral valve pressure gradient in HF result from an abnormal increase in LA pressure. Non-invasive methods to identify LV and LA pressure have recently been discussed and even the traditional echo parameters considered markers of elevated LV filling pressure such as reduced mitral deceleration time (DT) and isovolumic relaxation time (IVRT) blunted A wave and high E/e′ ratio have been debated. In theory, the occurrence of the above-mentioned picture is typical of a restrictive pattern and along with increased LA volume and increased pulmonary pressure, reflects high filling pressure in either HFrEF and HFpEF [70]. However, only a few studies compared Doppler and CMR parameters with direct haemodynamic measurement. Most of these studies analysed only E/e′ ratio in relation to wedge pressure, but a complete measurement of left chambers pressure is lacking [71, 72]. Therefore, correlation between E/e′ ratio and LVEDP in a recent metanalysis appears modest and patients with atrial fibrillation (AF) were excluded [73]. All these concerns put some doubts and limitation in evaluation of isolated diastolic dysfunction and consequently in the diagnosis of HFpEF. Thus, TDI and PW Doppler parameters may be integrated with careful analysis of cardiac structure and LV remodelling. Accordingly, elevated LA pressure leads to LA dysfunction and remodelling being commonly observed in both phenotypes. These findings, at a difference from systolic properties, would suggest a mechanistic continuum, as far as diastole is concerned, between the two conditions. It must be recognised, however, that LA remodelling in HFpEF and HFrEF, has been reported as being slightly different, with more dilation and systolic dysfunction in HFrEF and with increased stiffness, pulsatility and predilection for atrial fibrillation (AF) in HFpEF [74]. We think that this distinction is partly artefactual. The extent of LA pulsatility, that is a direct function of the stiffness of the cavity, may contribute to the pulsatile major component of right ventricular afterload [75]. Thus, from this perspective, diastolic LA dysfunction can be seen as an active contributor to symptoms and to disease progression for both phenotypes.

Diastolic dysfunction is a dominant feature in many HF patients. LV diastolic dysfunction causes LA dilatation, which can lead to AF [76]. Despite of being advance in management and treatment, AF remains a source of considerable morbidity and mortality worldwide. For the patients in sinus rhythm LV, filling pressures and diastolic function grade can be determined reliably by a few simple echocardiographic or CMR parameters with a high feasibility [77]. However, for the patients in AF regardless whether AF is a reason or consequence, assessment of diastolic function is challenging. Assessment in this condition in limited by cycle length variability, absence on an organised atrial activity and frequent occurrence of atrial enlargement regardless of filling pressures [78]. Recently, some new indexes using speckle tracking echocardiography have shown promising results showing the association between LV systolic and diastolic strain, LA strain and LV diastolic function in AF patients [79]. Indexes potentially capable of describing the delicate atrio-ventricular relation should convey the strongest pathophysiological information and be less influenced by R-R variations, if they can be comprehensively acquired in 1 single beat.

Recently, in characterising the governing role of the four-chamber (near) constant-volume pump physiology, wherein the atrial and ventricular volumes simultaneously reciprocate throughout the cardiac cycle, CMR has elucidated and characterised LA and LV phasic function, thereby quantifying the conduit contribution to ventricular filling as the integral of net, diastolic, instantaneous difference between synchronised atrial and ventricular volume curves [80]. Because cardiac CMR availability is limited, 3D echocardiography can be employed to acquire complete and simultaneous LA and LV full-volume datasets to characterise the volume of both left-sided cardiac chambers at each time point during the cardiac cycle in order to quantify conduit [81] (Fig. 2).

Using single-beat simultaneous left atrial and ventricular full-volume 3D dataset, it was demonstrated that the atrial conduit contribution to ventricular filling has a direct relationship with the degree of underlying ventricular diastolic impairment in HF patients [82]. More recently, it was shown that conduit quantitation is also able to predict 1-month AF recurrence in a population of persistent AF patients imaged immediately after and/or before electrical cardioversion [83]. These findings support the concept that conduit (independently of the imaging technique used to quantify it), reflects intrinsic atrial pathology that cannot be sufficiently explained by ventricular pathology only and thus it could be proposed as a clinically effective tool for exploring the link between AF and diastolic dysfunction, in excess of ventricular derangement.

In addition to LA and LV volume curves, CMR can provide further information on myocardial tissue characteristics. Myocardial fibrosis have been implicated in the pathophysiology of HFpEF by promoting adverse ventricular remodelling, increasing myocardial stiffness, and in turn, causing diastolic dysfunction [84]. Diffuse interstitial fibrosis, a precursor for replacement fibrosis, is not detected by LGE but correlates with T1 mapping, which allows a quantitative assessment of diffuse cardiac fibrosis and estimations of the extracellular matrix volume (ECV) (Fig. 3) [85, 86]. There has been accumulating evidence that increased T1 times and ECV are related to clinical outcomes [87‐89] and have potential for risk assessment in patients with AF.

Heart failure is disease with large phenotypic variations in morphology, function and natural history. Such a heterogeneity in presentation is perhaps a reason why despite different clustering approaches, many interventions in clinical trials have not shown efficacy. Cardiac imaging provides diverse insights, but the ability to distinguish between overlapping phenotypes remains a challenging proposition. The evaluation of diastolic function by echocardiography and CMR with their traditional and novel techniques deserves specific analysis and a comparison with haemodynamic measurement before to be universally accepted. Diastolic function parameters derived by CMR may be applied in routine clinical care and matched with more feasible echo analysis in order to increase our awareness in the HFpEF mechanisms and reduces the current diagnostic gap. New parameters studying both radial and circumferential relaxation together with identification of extracellular collagen volume could facilitate the diagnosis and will play a central role in the identification of underlying pathophysiological mechanisms. Comparative imaging trials should be encouraged in order to discern which technique(s) alone, or in combination, could provide additional prognostic value.