Ability Of Pulse Pressure Variation To Predict Fluid Responsiveness In Patients With Extrapulmonary ARDS

Objectives: Mechanical ventilation-induced changes in stroke volume(SV),quantified by pulse pressure variation(ΔPP), have been proposed to predict fluid responsiveness(FR). Nevertheless, cardiopulmonary interactions are complex in case of acute respiratory distress syndrome(ARDS). The aims of the current study were(1)to assess the ability of ΔPP to predict FR by using both the ROC and the gray zone approach in extrapulmonary ARDS population;(2) to determine how the magnitude and predictive performance of this ventilation-induced changes in pulse pressure(ΔPP) are influenced by respiratory variables(airway driving pressure [ΔPaw], Tidal volume [Vt], respiratory changes transpulmonary pressure[ΔPtp]), and identify factors responsible for false-negative and false–positive cases of ΔPP;(3) To evaluate whether or not ΔPP adjusted by respiratory variables could improve the prediction of FR.Methods: In this prospective observational study, 84 consecutive patients with ARDSexp admitted to the ICU were analysed. Fluid challenge(FC) was performed with a 500-m L saline bolus infused over 20 minutes. Before FC, esophageal pressure(Pes) was measured at the end-inspiratory(Pes,eio) and end-expiratory(Pes,eeo)occlusions. Change in pleural pressure(ΔPpl) was calculated as the difference between Pes,eio and pes,eeo. ΔPaw, ΔPtp, chest wall elastance(Ecw), respiratory system elastance(Ers) and lung elastance(EL) were all calculated using standard Equations. Hemodynamic measurements, including stroke volume(SV), cardiac output(CO) and ΔPP, were obtained before and after the FC.Results:The fluids increased CO by greater than 15% in 47 patients(responders).ΔPP >14.5% predicted FR with a sensitivity of 76.6%, a specificity of 97.3% and an areas under the ROC curve(AUC) of 0.945(95% CI, 0.873-0.983). On Multivariate linear regression, the magnitude of ΔPP in both responders(ΔPP = 0.32 + 2.196 ×ΔPpl + 0.499 × ΔPtp, R2=0.52) and non-responders(ΔPP =1.923 + 0.414 × ΔPpl +0.488 × ΔPtp, R2=0.45) were determined by two independent respiratory parameters(ΔPpl and ΔPtp), suggesting that a low ΔPpl and a high ΔPtp values may be responsible for false-negative(ΔPP <12% in responders) and false positive(ΔPP>12% in non-responders) ΔPP results, respectively. This assertion agrees with our results in that ΔPtp, but not ΔPpl, is significantly higher in 7 nonresponders with aΔPP greater than12%(false positive) than in 30 other nonresponders with a ΔPP less than 12%(true negative). In responders(n=47), ΔPP was also positively correlated with Vt and ΔPaw, such that normalization of ΔPP by Vt markedly improved the prediction of FR, avoiding some of the FNs observed using ΔPP alone. In nonresponders(n=37), ΔPP was positively correlated with ΔPaw,(ΔPP= 1.884+0.458×ΔPaw, R2=0.48,P<0.001), such that normalizing ΔPP by using ΔPP/ΔPaw was able to avoid some of the false positives observed by using ΔPP(12%) alone. For normal fluid policy(cost ratio = 1), the gray zone approach identified a range of ΔPP values, between 11% and 15%, for which FR cannot be reliably predicted. Around25% of the patients were within this inconclusive zone. By contrast, ΔPP/ΔPpl,ΔPP/Vt and ΔPP/ΔPaw had narrow gray zones for normal fluid policy(cost ratio = 1)that only included 2.4%, 16.7% and 11.9% of the patients, respectively.Conclusions: 1. Our findings confirm our hypothesis according to which, owing to the high value of ΔPpl resulting from the high Ecw, ΔPP can predict FR with relatively high accuracy in this population despite a small Vt(<8 ml/kg). 2. A lowΔPpl is mainly responsible for false-negative(ΔPP <12% in responders) cases of ΔPP, whereas a high ΔPtp values may be responsible for false positive(ΔPP>12% in non-responders) ΔPP results. 3. In patients with ARDSexp, Vt is positively related to the performance of ΔPP in predicting FR. 4. In nonresponders, the significant correlation of ΔPaw with ΔPP(R2=0.48,P<0.001) suggests that a ‘too high’ΔPaw may be a cause for false-positive results of ΔPP in patients with ARDSexp, such that normalizing ΔPP by using ΔPP/ΔPaw was able to avoid some of the false positives observed by using ΔPP(12%) alone. 5.Despite the high AUC of ΔPP in predicting FR,around 25% of patients had ΔPP values within the gray zone(11%-15%) for which FR cannot be reliably predicted, potentially limiting its clinical application to ARDSexp patients. By contrast, ΔPP/ΔPpl and ΔPP/ΔPaw had narrow gray zones for normal fluid policy(Cost ratio = 1), and should be encouraged to use in an attempt to make more rational and informed decisions on fluid management for ARDSexp patients. 6. More importantly, the gray zones of ΔPP and adjusted ΔPP(i.e.,ΔPP/ΔPpl, ΔPP/ΔPaw) change depending on whether the clinician aims at a“restrictive,” “normal,” or “liberal” fluid policy, which is of major clinical importance.Clinicians should consider different thresholds, depending on whether they want to avoid unnecessary volume loading(Cost ratio >1, “restrictive” fluid management) or nonoptimal CO maximization(Cost ratio <1, “liberal” fluid strategy).