First, Morosanu explicitly enrolled only patients with a ‘safe’ and, potentially, ‘effective’ profile as the 40 included had a decongested portal vein and were fluid responsive based upon a PLR. However, in their exclusion flow chart, there were initially 64 patients who had good echocardiographic windows and a low PVPI (i.e., fluid tolerant); of these, 21 were fluid unresponsive. That is to say, 33% of patients who were fluid tolerant were also fluid unresponsive. Within the Doppler Starling framework, we have previously found that 33% of patients in ‘Quadrant 3’ were fluid unresponsive [9] (see Fig. 1A below); this profile has been termed ‘dynamic fluid intolerance [4]’ because VC is expressed only with a dynamic maneuver like a PLR. Morosanu and colleagues did not record the change in PVPI in these patients but, in theory, VC would be likely. The clinical relevance of this finding is that giving fluids based only upon a baseline ‘low preload’ or ‘fluid tolerant’ profile risks giving ineffective IV fluids in a clinically significant proportion of acutely-ill patients; this is the ultrasonographic equivalent of giving fluids for a central venous pressure of less than 8 mmHg [10].

The Doppler Starling curve. The 4 hemodynamic phenotypes (1–4) are generated by combinations of normal and low stroke volume on the y-axis and normal or congested venous measures on the x-axis. A.) shows adapted data from Morosanu et.al. Fig. 2 for all patients comparing the passive leg raise (PLR) portion of the study (between T1 and T2) and the change between return to baseline and 2 min following the Lactated Ringer’s (LR) infusion (T3 and T4). T5 is excluded for clarity. The red arrow accounts for the 21 patients who were fluid tolerant but unresponsive. B.) shows the change recorded for PLR and the LR infusion when patients were split into those who did not have early congestion following LR (grey) and those who did have early congestion following LR (blue). This is adapted from Morosanu et al. Fig. 3. Based on mean values, some patients (i.e., early VC) moved from Quadrant 3 to 2 and this progression was predicted by the PLR (i.e., the change from T1 to T2). Advanced echocardiographers might argue that these patients did show signs of fluid intolerance without a PLR given their impaired, baseline right ventricular (RV) function–when the RV is no longer operating as an unstressed chamber [11]

Second, Morosanu and colleagues show that patients can begin with a ‘safe’ and potentially ‘effective’ profile but, nevertheless, display another kind of ‘dynamic fluid intolerance’–moving both ‘up’ the Doppler Starling curve (i.e., on the y-axis), but also ‘out’ (i.e., along the x-axis). Based upon averages (see Fig. 1B below), these patients move from quadrant 3 to 2; the evolution of VC was predicted accurately by PLR, before IV fluids. While the slopes of the curves between those who developed early VC (blue curves) and those who did not (grey curves) were found to be statistically the same, this framework implies that with a greater range of measured values, perhaps there were subtle slope differences; this cannot be known given this pilot data. Research on the ‘slope’ of the Doppler Starling curve is underway; more specifically, whether the ratio between the LVOT VTI and VExUS + 1 (to prevent zero in the denominator) can predict patient outcome in the ICU.

In summary, Morosanu and colleagues are to be congratulated for their important pilot investigation. We should continue to anticipate divergence between venous measures and fluid responsiveness, especially with impaired cardiac function. Doppler phenotyping in this manner is an exciting avenue of active investigation.