Background
Prone positioning is recommended in mechanically ventilated patients with acute respiratory distress syndrome (ARDS) and a ratio of partial pressure of arterial oxygen (PaO
2) to the fraction of inspired oxygen (FiO
2) < 150 mmHg [
1‐
3], for at least 16 h [
1] and even more for some patients [
4]. During the Coronavirus disease 2019 (COVID-19) pandemic, prone positioning has been widely applied in patients with ARDS [
5‐
9]. Patients with ARDS, especially those with COVID-19, are characterized by lung edema mainly related to increased pulmonary endothelial permeability [
10,
11]. Fluid management in such patients is challenging [
12,
13]. On the one hand, ARDS patients could experience shock that might require fluid therapy and vasopressors [
14,
15]. The objective of fluid therapy is to restore adequate organ perfusion in patients in the case of fluid responsiveness, a phenomenon that is generally present in 50% of patients [
16]. On the other hand, fluid therapy may worsen lung edema due to the altered permeability of pulmonary microvessels [
17]. Therefore, the prediction of fluid responsiveness is important to test in patients with ARDS to prevent fluid administration in those who are fluid unresponsive and for whom harmful consequences of fluid therapy would be maximal [
18]. Pulse pressure variation (PPV), passive leg raising (PLR), and end-expiratory occlusion (EEO) are dynamic variables or tests that are commonly used to predict fluid responsiveness in mechanically ventilated patients [
19,
20]. Nevertheless, PLR and EEO test require real-time cardiac output measurements, while PPV, which can be measured non-invasively [
21,
22] or invasively by a simple arterial catheter, is less reliable in patients mechanically ventilated with a low tidal volume [
23‐
25]. The tidal volume challenge (TVC) was suggested to compensate for the limitation of PPV during low tidal volume ventilation [
26], though limited and controversial evidence exists in patients under prone position [
27,
28]. The primary objective of our study was to investigate whether a one-minute TVC could assess preload responsiveness in patients with ARDS under prone position. The secondary objective was to investigate the predictive performance of EEO test at the tidal volume of 6 mL/kg predicted body weight (PBW) under prone position.
Discussion
Our study showed that in patients with ARDS under prone position, an increase in PPV ≥ 3.5% during one-minute TVC could reliably assess preload responsiveness and its predictive value was better than that of PPV alone.
The PPV is one of the most utilized dynamic indices to predict fluid responsiveness [
44]. One of the reasons is that it can be easily obtained [
21,
22]. However, interpretation of PPV is limited in many circumstances such as low tidal volume ventilation [
25], arrhythmia, low respiratory compliance, and spontaneous breathing activity [
24]. In patients who receive low tidal volume ventilation, respiratory changes in intrathoracic pressure might be insufficient to produce significant changes in preload and therefore PPV may lack sensitivity to predict fluid responsiveness [
45]. This was illustrated by studies performed in patients with or without ARDS and who were mechanically ventilated with a tidal volume of ≤ 8 mL/kg [
25,
46]. To compensate for the limitations of PPV at low tidal volume, it has been suggested to evaluate the response of PPV to a TVC [
26]. An increase in PPV > 3.5% during a TVC was shown to predict fluid responsiveness reliably in patients who were ventilated with 6 mL/kg tidal volume in supine position (AUROC curve of 0.99) [
26]. Another study [
47] and a recent meta-analysis [
20] have confirmed such excellent results in supine patients ventilated with low tidal volume. The findings of our present study confirmed that the TVC was still valid in patients with ARDS who underwent prone position with a threshold value (3.5%) which is the same as that previously reported [
26]. The least significant change of PPV according to De Courson et al. was 8.9% [
48]. Thus, for a given value of PPV, for example, 7 (which is the mean baseline value of PPV in our study), the smallest change in absolute value that can be trusted as a real PPV change would be small (less than 1 in absolute value). In our study, the cutoff value of absolute PPV change for TVC (+ 3.5) is thus higher than the least significant change value found in the previous literature [
48].
Our study also confirmed that ΔPPV TVC6–8 could perform better than PPVbase to assess preload responsiveness. Importantly, there were many cases (56/84 cases) where PPV was inconclusive (between 4 and 11%) and where the ΔPPV TVC6–8 still reliably assessed preload responsiveness. The plateau pressure and hence the driving pressure increased significantly during the one-minute TVC, although there was no difference regarding respiratory compliance. Since the TVC is quite short, the effects on the driving pressure are expected to be transient and reversible. Nevertheless, caution should be taken in using this test in patients with markedly increased driving pressure.
Until now, only one study addressed the issue of assessment of preload responsiveness in patients with ARDS under prone position. Yonis et al. found that an increase in cardiac output greater than 8% during a Trendelenburg maneuver well assessed fluid responsiveness in a series of 33 patients under prone position and ventilated with 6 mL/kg tidal volume [
28]. They also found that the changes in PPV during TVC did not well predict fluid responsiveness, which is in disagreement with our present results. It has to be noted that in this study [
28], PPV and its changes in response to TVC were assessed in only 19/33 patients since 14 patients with cardiac arrhythmia were excluded from the analysis. There is no clear reason to substantiate the argument that heart–lung interactions cannot apply to patients in prone position as they apply to patients in the supine position. Therefore, there is no clear argument to support the fact that the hemodynamic effects of TVC are undermined during prone position. In this regard, in patients under prone position for neurosurgery and who received 6 mL/kg tidal volume, the response of PPV to a TVC was shown to predict fluid responsiveness with excellent accuracy (AUROC of 0.96; sensitivity: 95%; specificity: 95%) [
27]. It has to be noted that the Crs was normal (around 65 mL/cmH
2O on average) in the latter study, whereas it was low in the Yonis et al. study (around 30 mL/cmH
2O on average), and this might account for the discrepancies between the findings of these two studies. Nevertheless, in the study by Myatra et al. [
26], where the TVC performed very well to predict fluid responsiveness, the mean Crs was low (28 mL/cmH
2O on average), and thus comparable with the Crs values reported by Yonis et al. [
28]. In our present study, the Crs values were also low (around 30 mL/cmH
2O on average) as we included patients with severe ARDS. An important difference between our study and that by Yonis et al. is the definition of preload responsiveness. As we did not administer fluids to all our patients with ARDS as they did, we defined preload responsiveness by the positivity of two preload responsiveness tests (Trendelenburg maneuver and EEO
8). To minimize risks of uncertain interpretation, we excluded cases where one of the two tests was positive and the other one negative, a situation that occurred in 15% of cases. Therefore, in our study, we considered the presence of preload responsiveness when both ΔCI
TREND and ΔCI EEO
8 were ≥ 8% and ≥ 5%, respectively, and the presence of preload unresponsiveness when both ΔCI
TREND and ΔCI EEO
8 were < 8% and < 5%, respectively. It is noteworthy that our definition could identify the same proportion of preload responsive cases (50%)
vs. preload unresponsive cases (50%) as reported in previous studies [
16] including those where low tidal volume ventilation was used [
20]. However, this does not totally exclude that both tests could be positive—according to our definition—while the patient would be fluid unresponsive and vice versa. Nevertheless, in all the cases with preload responsiveness—according to our definition—where fluid was administered, CI increased by ≥ 15% in response to fluid infusion, suggesting that our definition was appropriate at least in terms of specificity.
The predictive performance of PPV
base in our present study was better than that reported in some previous studies performed in patients receiving low tidal volume ventilation in supine position [
25,
46]. Nevertheless, a recent meta-analysis that investigated the performance of PPV in patients under mechanical ventilation with tidal volume ≤ 8 mL/kg without arrhythmia and respiratory effort (22 studies) showed an AUROC of 0.82 (sensitivity 74% and specificity 77%) [
20]. It is noteworthy that in patients with ARDS and ventilated with 6 mL/kg, Freitas et al. showed that PPV could predict fluid responsiveness with an AUROC of 0.91 (0.82–1.0), the sensitivity of 89%, and specificity of 90% [
49]. Nevertheless, in our study, in two-thirds of cases, the PPV
base fell in a range of uncertainty (between 4 and 11%). Interestingly, the TVC might be helpful in the subgroup of cases where the intra-abdominal pressure was > 12 mmHg (
n = 34) and where PPV
base failed to predict the preload responsiveness, although further studies are warranted to confirm this finding. All the above findings limit a broad application of PPV during prone position under low tidal volume ventilation and justify performing another test such as the TVC.
Since varied results were reported regarding the predictive performance of the EEO
6 [
26,
28,
50‐
52], we chose the EEO
8 as one of the tests to definite the preload responsiveness in our current study in order to minimize uncertainty. Our results showed that the EEO test performed at 6 mL/kg tidal volume reliably assessed preload responsiveness in patients under prone position, which is consistent with our previous studies in the supine position [
32,
51,
52] but in disagreement with some other studies showing a less reliable predictive performance of the EEO test at 6 mL/kg tidal volume in supine [
19,
43] or in prone position [
20,
21]. Nevertheless, although EEO
6 seems to be reliable in prone position in our population, the cutoff value of CI change defining the preload responders was quite low (3.2%). Whereas the TVC only requires an arterial catheter to track the changes in PPV, the EEO
6 has thus the disadvantage to require a real-time cardiac output monitor with a very high precision [
31], a condition that is uncommon in resource-limited settings [
53].
Our study has some limitations. Firstly, not all our patients received the standard fluid challenge, since administering fluid is not routinely performed by attending clinicians in patients with ARDS, even when preload responsiveness is present. Nevertheless, a postural maneuver (
i.e., PLR) had been previously used to replace fluid administration in order to evaluate the validity of preload responsiveness tests [
54‐
56]. Secondly, to interpret the analysis more straightforwardly, we did not include the cases where only one of the two reference tests of preload responsiveness was positive (
i.e., either ΔCI
TREND ≥ 8% or ΔCI EEO
8 ≥ 5%). This occurred, nevertheless in only 15% of cases. Thirdly, 16 patients were included more than once. It is noteworthy that these patients were never included twice during the same day or during the same prone position session. For these 16 patients, the delay between two inclusions was five days, which could be considered long enough for patients to present different hemodynamic profiles. Nevertheless, our statistical analysis took into account the effects of the repeated measurements on the same subject.
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.