Is Pulmonary Artery Wedge Pressure A Fib in A-Fib?

This article refers to ‘Atrial fibrillation modifies the association between pulmonary artery wedge pressure and left ventricular end-diastolic pressure’ by M.G. Dickinson et al., published in this issue on pages 1483–1490. What we observe is not nature itself, but nature exposed to our method of questioning. Werner Heisenberg. Heisenberg's uncertainty principle states that there is a fundamental limit to the precision which we can obtain knowledge about the physical properties of a quantum particle, specifically stating one cannot precisely know the position and momentum of a quantum particle.1 In the study by Dickinson et al.2 in this issue of the Journal, the authors provide their own version of a haemodynamic uncertainty principle by studying the effects that atrial fibrillation (AF) has on the relationship between measured pulmonary artery wedge pressure (PAWP) and left ventricular end-diastolic pressure (LVEDP). The question of the accuracy of PAWP as a surrogate for left ventricular (LV) filling pressure is not one without important clinical implications. A common reason to perform a haemodynamic study is to establish the diagnosis and aetiology of pulmonary hypertension (PH). Per the European Society of Cardiology/European Respiratory Society and American College of Cardiology/American Heart Association guidelines, the presence of elevated LV filling pressures (>15 mmHg) establishes the diagnosis of post-capillary PH (or PH due to left heart disease).3, 4 Conversely, the absence of elevated LV filling pressures establishes the presence of pre-capillary PH. The distinction between these two different (though sometimes overlapping) physiologies drives therapeutic decision-making. However, the accuracy of PAWP as a LVEDP surrogate in the setting of AF has not been robustly assessed, and Dickinson and colleagues commendably set out to answer that important question. After exclusions, 123 patients without severe mitral valvular disease who underwent simultaneous left and right heart catheterization at their centre were included. Thirty-four patients (28% of the cohort) were in AF at the time of haemodynamic assessment. The AF cohort was older, had more left heart disease, had a lower cardiac output, and had lower systolic pulmonary artery pressures but higher diastolic and mean pulmonary artery pressures. As with any haemodynamic study, the methods of pressure assessment and measurement deserve consideration. First, recordings were initially made using automated pressure measurements via electronic means averaged over the respiratory cycle. For patients in sinus rhythm, the automated analysis incorporated both the A and V wave contributions. For patients in AF, the A wave analysis was manually disabled, and we may assume the V wave contribution had more relative impact on the generated mean. In AF patients, the LVEDP was measured at the Z-point (the pressure 80–120 msec after the QRS complex), rather than just prior to the pressure inflection point signalling the start of systole. Importantly, a second analysis of PAWP and LVEDP was performed during 3–5 manually-selected beats at end-expiration as per current guidelines.3 While the authors found that there was no difference between mean PAWP and LVEDP within the whole cohort, the relationship did vary between PAWP and LVEDP in those with AF or sinus rhythm. In sinus rhythm PAWP was lower than LVEDP (mean difference –2.96 mmHg, a value similar to that of Halpern and Taichman5 who reported a mean difference of –2.9 mmHg), and in AF mean PAWP was higher than LVEDP (mean difference 4.76 mmHg). Utilization of LVEDP instead of PAWP resulted in re-classification of PH aetiology (pre- or post-capillary) in 35% of patients in AF and 31% of patients in sinus rhythm when non-end-expiratory pressures were used. The authors report that in patients with PH due to left heart disease, the use of end-expiratory measurements improved agreement between PAWP and LVEDP in patients with sinus rhythm only, confirming the findings of Ryan et al.6 As an explanation for the findings, the authors first postulate that the loss of an atrial contractile kick may lead to an ‘uncoupling’ of left atrial (LA) and LV pressures—essentially a state of ‘pseudo’ mitral stenosis. They also cite the known loss of atrial compliance in AF patients and propose that a non-compliant atrium may further ‘uncouple’ the PAWP and LVEDP. Based on a model of the pulmonary–LA–LV circulation, an alternate explanation for the findings is also feasible. If we are concerned with estimating LV filling pressures, we should consider the circuit in end-diastole, when the mitral valve is open. Viewed this way, one can appreciate that in the valve-less pulmonary circulation (presuming no mitral stenosis), the compliance of the left atrium should bear little consequence on the pressure relationships between pulmonary capillaries and the left ventricle (Figure 1). Replacing a compliant left atrium with a non-compliant left atrium should equally raise PAWP and LV pressure in diastole. In keeping, Braunwald and Frahm7 demonstrated in 1961 that LA and LV end-diastolic pressure were equal in patients both in sinus rhythm and AF. Thus, we propose that the described difference between PAWP and LVEDP in sinus rhythm and AF patients in the current study must involve differences in the systolic component of pressure measurements. In considering the measurement methods of Dickinson et al., we can see how this might be the case. During PAWP measurement, the authors incorporated both A and V waves and calculated the mean from both of these values. While this method may best estimate the pressure ‘felt’ by the pulmonary circulation throughout the cardiac cycle, it will overestimate LVEDP because the V wave is a systolic phenomenon.8, 9 The authors note that they excluded patients with severe mitral regurgitation to avoid those with high V waves. However, low atrial compliance is a well-known cause of augmented V waves in the absence of mitral valve disease. As the authors acknowledge, patients with AF have lower atrial compliance, and thus we would expect them to have augmented V-wave pressures leading to a mean PAWP that is higher than end-diastolic ventricular pressure (Figure 1). One could also postulate LA–LV dyssynchrony in AF might also contribute to this phenomenon if the mitral valve had incompletely closed at the onset of systole. Lastly, it is worth mentioning that PAWP measures were not confirmed by saturations, and it is possible that the presence of AF makes it more difficult to obtain complete pulmonary arterial occlusion.10 Thus, we are led to question whether the observed discrepancy described between PAWP and LVEDP in AF is a phenomenon of ‘nature itself’ or of ‘our method of questioning’. To prove the former, prospectively performed high fidelity measurements within the same individual (while in AF and then sinus rhythm) would be required as pointed out by the authors. Even though we would propose these findings may be related to the ‘method of questioning’, the study still provides invaluable insight. First, we believe that this study shows the importance of incorporating only diastolic venous pressures in the measurement of PAWP if the goal is to find a true surrogate for LVEDP. In patients in sinus rhythm, this is accomplished by taking the mean of the A wave pressure, the final pressure prior to the onset of systole. However, in AF, the absence of an A wave provides a special challenge. The PAWP is best measured by averaging end-expiratory pressures over several beats at the pre-C wave or Z-point, the point 80–120 ms after the QRS on electrocardiogram on the PAWP tracing.11, 12 Dickinson et al. note that they employed the Z-point to time measurement of LVEDP, but not of PAWP, in AF patients. Additionally, the implications of this study are relevant to the debate of how best to haemodynamically sub-classify patients with post-capillary PH. The diastolic pulmonary gradient (DPG, calculated as the difference between the diastolic pulmonary pressure and PAWP) ≥7 mmHg has been proposed as a marker of pre-capillary involvement in post-capillary PH. Consideration based on ‘nature itself’ argues the soundness of this reasoning—diastolic pressure gradients should be less affected by left heart failure-induced changes in vascular compliance and thus should provide a better representation of the intrinsic state of the pulmonary vasculature.13 However, several studies have shown that, in real-world practice, DPG fails to perform well as a risk predictor. In fact, up to 48% of patients undergoing right heart catheterization have a negative DPG, a state which seems like a physiologic impossibility—and this is at least partially related to the presence of V waves and the method of PAWP measurement.14-17 The uncertainty surrounding ‘our method of questioning’ in this case may obviate the utility of a measurement that relies on a differentiation of pressures on the order of 1 mmHg. We believe the current study highlights a similar limitation, magnified by the difficulties in measurement added by the presence of AF. It is notable that despite having higher pulmonary vascular resistance, the average DPG of the AF cohort was 0 mmHg compared with 5 mmHg in the sinus rhythm cohort—a finding that is seemingly physiologically implausible. Perhaps, this study best highlights the increased degree of uncertainty introduced in measurements of PAWP in AF patients and the added care that must be taken to incorporate only end-diastolic venous pressures in measurements of PAWP. It reminds us that our research and clinical communities and even existing practice guidelines lack a well-defined methodology by which to determine PAWP. It also calls into question what the primary physiologic goal of measuring PAWP is in the first place. If our goal is to estimate LVEDP, we should not be using the mean PAWP value. However, if it is to reflect the total pressure to which the pulmonary vasculature is exposed throughout the cardiac cycle, perhaps the mean PAWP is the correct value. One would hypothesize that the former should be used to define the pre-capillary component of PH in left heart disease. Lastly, by recognizing the intrinsically higher degree of uncertainty, this study argues that, for patients with AF and PAWP measurements close to a haemodynamic threshold where accurate measurement will impact clinical decisions, LVEDP might be a more measurable, less uncertain, value on which to base those decisions. We applaud Dickinson and co-authors for exposing the lack of standardization and significant knowledge gaps that exist in proper haemodynamic assessment. Conflict of interest: none declared.