Echodynamics or pulmonary artery catheter dynamics? Should they be mutually exclusive?

The use of pulmonary artery catheters (PACs) remains considered as the gold standard for haemodynamic assessment to classify the haemodynamic profiles of cardiogenic shock (CS) and define the treatment accordingly. Recent registry-based studies demonstrate that the use of PAC, particularly when including a comprehensive haemodynamic assessment with the clinical contextualization of the parameters, is associated with improved survival, likely by allowing a thoughtful escalation of care, tailoring therapy, and allocation of resources.^1,2 Despite the above, the rate use of PAC remains widely variable,³ requires appropriate expertise for the correct interpretation and implementation into accurate decision making, and carries the associated limitations of being an invasive procedure with inherent, yet small, risks. Echocardiography is universally used in the assessment of patients with heart failure (HF), and serial echo evaluation is recommended in the management of acute HF/CS.⁴ However, it has been historically questioned as opposite to PAC in the evaluation of cardiac function and in the assessment of congestion and perfusion showing appropriate correlation for certain parameters, such as cardiac output estimation and systolic pulmonary pressure (sPAP), while reporting limited correlation for the estimation of filling pressures.⁵

On the current issue, Frea et al. report the result of an important single-centre study comparing invasive and echocardiographic haemodynamic (echodynamic) assessment in an acute decompensated HF-related-CS (HF-CS) population including patients from Society of cardiovascular angiography and intervention (SCAI) B (34%) to SCAI D (1%) shock stages.⁶ The authors must be commended for having run a difficult study with a strong methodological and pathophysiological approach, evaluating common and sophisticated haemodynamic parameters. These included load adaptation index, pulmonary artery pulsatility index (PAPi), right ventricle contraction pressure index (RVCPi) as surrogate of stroke work (RVSWi), and aortic pulsatility index, which were all performed as part of a comprehensive echocardiographic evaluation within 30 min of invasive measurements. Additionally, the authors provided an interesting approach aiming to test the accuracy of phenotyping CS by using echodynamic parameters. The results of the study reported an excellent correlation between echocardiographic and invasive estimation of cardiac index (r = 0.88, P < 0.001), sPAP (r = 0.91, P < 0.001), right atrial pressure (RAP) (r = 0.86, P < 0.001), and cardiac power output (r = 0.82, P < 0.001). Interestingly, RVCPi showed good correlation with RVSWi, as well as invasive and non-invasive PAPi, especially in those with RV dysfunction and a PAPi ≤ 1.85. The physiological basis for PAPi as an indicator of right heart function is based on sPAP as an indirect indicator of RV contractile function against a given afterload and a high RAP as a sign of failing RV leading to backflow and congestion.⁷ Pulmonary artery pulsatility index has gained significant attention in clinical practice for the assessment of the right heart function^2,8 and as a prognostic marker⁹ although, as pointed out by the authors, cut-off values vary across cohorts or the specific condition studied. The results in the present study are encouraging by supporting the use of echo PAPi to assess right-side performance.

Conversely, the echo-derived pulmonary capillary wedge pressure (PCWP), although using multiple calculations, failed again to show satisfactory relation with invasive measurement. While disappointing, this is unsurprising, as it correlates with the results of previous studies addressing the same question, although in different clinical contexts.⁵ Such limitation in accurately estimating left ventricular (LV) filling pressures led to a misclassification of the CS phenotypes in the study population.⁶ Although the importance of a detailed phenotypization at the first medical contact is considered paramount, current evidence shows that one single evaluation of a haemodynamic parameter may be less informative than continuous or close monitoring to define the patients’ trajectory.^10–12

The discontinuity of the echo evaluation has always been considered one of its major limitations. A recent technology development of a wearable cardiac ultrasound device, able to provide comprehensive and continuous frame-by-frame acquisitions of cardiac images, even during intensive physical activity,¹³ may definitely overcome such limitation. This device has an orthogonal configuration to eliminate the need for manual rotation and high stretchability (up to 110%) and withstands various deformations to maintain intimate contact with the skin over a large contact and during different situations. Notably, the difference between wearable and ‘standard’ cardiac ultrasound in terms of image quality was negligible. This wearable device can evaluate the LV performance by dividing the LV according to the 17-segment model in the parasternal axis, based on the magnitude of its motion in the radial direction during contraction (thickening) and relaxation (thinning), and by recording the displacement waveforms of the myocardial boundaries. In addition, M-mode analysis computes structural features (e.g. cavity diameters and valvular function) incorporating electromechanical synchrony assessed by overimposed electrocardiogram during different phases of the cardiac cycle. Furthermore, a deep learning neural network was applied resulting in actionable information by automatically and continuously outputting curves of critical cardiac metrics, such as myocardial displacement, stroke volume, ejection fraction, and cardiac output. The wearable device was tested during a stress ergometric bicycle test, and it demonstrated uninterrupted tracking of the LV activities, including the corresponding M-mode echocardiography and synchronized heart rate waveform. This novel technology and encouraging results underscore the possibility of selecting different wearable devices to continuously monitor haemodynamic parameters in different environments according to the patients’ clinical feature and acuity (from smartwatches to implanted loop recorder and wearable ultrasound tapes) and acquire data to be further integrated with artificial intelligent and machine learning algorithms aimed to augment diagnostic and prognostic accuracy, representing a step forward in the realm of the personalized medicine and with the potential of also being applied to the critically ill cardiac patient.

As technology improves and we gather a better understanding of invasive and non-invasive data in patients with CS, the study from Frea et al. may represent a step forward towards improving echodynamic assessment, as highlighted by the authors, with a few strengths and limitations that should be outlined. Remarkably, all the echodynamic was performed using 2D echo and standard Doppler techniques, which increase consistently the reproducibility and feasibility. Recently, a good correlation was demonstrated between RV end-diastolic pressure (RVEDP) and PCWP in predominantly left (r = 0.67) and biventricular (r = 0.67) CS phenotypes irrespective of the use of positive pressure ventilation (PPV) or mechanical circulatory support.¹⁴ Although RVEDP requires a reliable pulmonary regurgitation trace, which was present only in 78% of the Frea et al. population, this may be eventually with the integration of E/e′, a good solution to test in the attempt to identify a reliable non-invasive approximation of PCWP. Understandably, patients requiring orotracheal intubation and on mechanical circulatory support were excluded from the study, as they will need dedicated validation. Non invasive RAP was calculated, as previously described in patients with LV assist device,¹⁵ by integrating inferior vena cava (IVC) diameter and collapse, hepatic venous flow pattern and the tricuspid E/e′ ratio. However, all these indices may vary consistently during PPV. Additionally, the IVC measurement in patients with non-invasive PPV was taken by reducing the positive end-expiratory pressure (PEEP) to 0. This is artefactual as the PPV affects the venous return, heart–lung interaction, and right atrial and abdominal pressure. Nevertheless, the PEEP and PPV haemodynamic effects are constantly neglected in all the studies exploring the role of congestion, and the significance of congestive parameters in HF-CS despite the rate of patients requiring any kind of ventilation is high.

As outlined by the authors, the lack of reliable estimation of increased PCWP with echocardiography may be partially compensated by the other bedside imaging modalities that are strong and reliable and identify the direct consequences of increased pulmonary congestion, such as lung ultrasound, and other ultrasound techniques that can assess organ perfusion (arterial Doppler) and congestion (venous Doppler), which are supported by a growing body of evidence.¹⁶ Whether the integration of echo phenotype may impact the SCAI stratification also in terms of prognosis is one of the many lights to be shed in future studies.

The last, but not least, important question pertains to the clinical consequences of gathering data. Would a patient with invasively measured PCWP of 25 and PAPi of 0.3 be treated differently to a patient with echo evaluation of E/e′ 25, PAPi 0.3, severe mitral regurgitation, homogeneous distribution of B lines, and mono-phasic renal venous flow? Would a patient with an invasively measured PCWP of 25 and echo visualization of severe MR, multiple B lines, and ultrasound pattern of organ congestion be treated more effectively and timely? While we wait for the clinical implementation in the critical care arena of the wearable continuous monitoring devices, the next step in clinical trial may test the clinical implication of the complementary and synergistic use of both invasive and non-invasive haemodynamics to properly phenotype and monitor the altered haemodynamic parameters and guide clinical decision-making.

Funding

None Declared.

Data availability

No new data were generated or analysed in support of this research.

References

Author notes