Prone Position in COVID-19 Studied with 3-Dimensional Trans-esophageal Echocardiography
“In that day there's a moment when it all goes away …”
-The Tallest Man on Earth
A recent case series reporting experience with 3-dimensional trans-esophageal echocardiography in moderate-to-severe COVID-19 associated lung injury was published in Intensive Care Medicine. With only 9 patients described the findings are, nevertheless, noteworthy because of their implications for prone position, in general, and because 1 patient had worsening hemodynamic status following pronation.
What They Did
Ventilated patients with confirmed COVID-19 ARDS were included. All patients were ‘hemodynamically stable’ prior to pronation and none required vasoactive medications and there was no other organ failure.
3-dimensional trans-esophageal [3DTEE] assessments were performed immediately before prone, 1 hour after pronation, and 16 hours later when returned supine. Data acquisition was over 6 consecutive heartbeats during end-expiratory breath-hold lasting 3-4 seconds. The echocardiographer included the entire right ventricular cavity and myocardium, and then the entire left ventricular cavity and endocardium within the pyramidal scan volume. Two to three 3D acquisitions were obtained during each patient position change. Data were then exported in a TOMTEC® and Philips QLAB® systems for 3D measurement of RV and LV volumes, respectively.
What They Found
At baseline, the median driving pressure was 20 cmH2O with inter-quartile range between 14 and 21 cm H2O. The median PaO2:FiO2 ratio was 77. The patients appeared to have mild pulmonary hypertension at baseline with a median Vmax tricuspid regurgitant jet of 3.3 m/s [ULN roughly 2.9] and a relatively low pulmonary acceleration time of 75 milliseconds. The right ventricular end-diastolic volume to left ventricular end-diastolic volume [RVEDV/LVEDV] ratio was not heightened and the left ventricular systolic, but not diastolic, eccentricity index was elevated at baseline.
Upon prone position, biventricular EDV fell, while ESV was generally sustained such that biventricular ejection fraction was diminished. VmaxTR slightly dropped to 3.2 and acceleration time trended downwards from 75 ms to 60 ms. LV eccentricity improved. There was a non-significant reduction in left ventricular outflow tract velocity time integral that was largely compensated for by a rise in heart rate.
Importantly, one patient had a poor response to prone position. This patient had moderate acute cor pulmonale with preserved LV systolic function at baseline. On prone position, the patient's right ventricular end-diastolic volume [RVEDV] fell and paradoxical septal motion disappeared. Yet, despite normal mean blood pressure, the patient’s cardiac index, LVEF and LV strain rate all decreased with unchanged, but relatively low, calculated systemic vascular resistance. The patient also demonstrated superior vena cava collapse on pronation.
The hemodynamic impact of prone position can be conflicting and hard-to-predict. During a lecture at The Big Sick in February, I pitched prone position as akin to non-pharmacological ‘hydralazine and nitrates’ for the right ventricle. As we think of these drugs as both pre- and afterload reducers for the left ventricle, we anticipate beneficial effects when the ventricle is dilated and poorly-performing, but detrimental effects in the setting of normo- or hypovolemia and normal ventricular function.
The preload-reducing effect of prone position is particularly prominent when the abdomen is allowed to hang freely under chest and hip supports. Ostensibly, this is because abdominal venous vasculature does not decrease its capacitance as much as if the abdomen is pressed firmly against the bed. The means of pronation in the case series is not explicit. The afterload-reducing effect of prone position is most pronounced when there is dependent lung recruitment. This is, partly, because the airway pressure falls relative to the pleural pressure; in other words, the trans-pulmonary pressure falls.
Accordingly, in an overloaded patient with high right ventricular afterload from severe ARDS we expect that reducing both RV preload and afterload shrinks the size of the right ventricle but maintains [or improves] its output. Further, ‘decompressing’ the left ventricle improves its compliance while receiving the same [or more] preload across the pulmonary circuit. By contrast, in a healthy, euvolemic patient placed in prone with a freely-hanging abdomen, the right ventricular outflow tract can obstruct!
In general, the patients with moderate-to-severe COVID-19 in this case series tended to shrink RV cavity size and stroke volume. It isn’t specifically reported, but the authors note that cardiac index is maintained by increasing heart rate, and the LVOT VTI [i.e. SV] certainly does not rise upon pronation. As well, biventricular EDV falls in the face of stable biventricular ESV; further the pulmonary acceleration time also falls on pronation, which points away from reduced RV afterload and towards diminished stroke volume by preload reduction.
In the patient who deteriorates, we only are told that pronation diminished RVEDV, improved left ventricular eccentricity, and that there is a significant fall in LV ejection fraction and strain rate in the face of low, absolute calculated systemic vascular resistance. Unfortunately, from this data we are not able to definitively infer the cause. Note that both ejection fraction and strain rate are preload dependent. Mathematically, ejection fraction is 1 – [ESV/EDV]. Thus, falling EDV, by itself, attenuates ejection fraction. Similarly, strain rate characterizes early ventricular force contraction, consequently, diminished Frank-Starling mechanism also reduces strain rate. In line with a preload mechanism, we are also told that the deteriorating patient has significant SVC collapsibility upon pronation. If the cause of poor cardiac output were primarily LV volume and pressure overload, we anticipate high central venous blood volume and depressed SVC collapsibility, though pronation could raise intra-thoracic pressure and diminish SVC trans-mural pressure.
How might this be put together? Perhaps the patient who did poorly had ‘L-type’ CARDS, so on pronation there was little lung to recruit and no significant afterload attenuation. As well, maybe this patient was euvolemic or hypovolemic to begin with. Therefore, the baseline acute cor pulmonale physiology was more like a pulmonary embolism. Perhaps pronation occurred with the abdomen hanging freely, or maybe this patient had a body habitus that did not pressurize the abdomen [i.e. scaphoid abdomen, low abdominal adiposity, etc.]. Thus, reduced RV size on prone was not driven by improved afterload but rather diminished preload. This normalized LV eccentricity, but with significantly-diminished LVEDV; one observes falling ejection fraction and strain rate. Then again, the results we are given might also be secondary to overt LV failure on pronation. But here’s the rub, we can’t know.