Left ventricular-arterial coupling is associated with prolonged mechanical ventilation in severe post-cardiac surgery patients: an observational study

We performed a retrospective single-center study of severe cardiac patients who were admitted to the ICU after cardiac surgery with continuous PiCCO monitors (Pulsio Cath PV 2015 L20: Pulsion Medical Systems: Munich, Germany) for unstable hemodynamics from January 2015 to December 2017 at the Department of Critical Care Medicine of Peking Union Medical College Hospital. These patients had a preoperative class III or IV New York Heart Association (NYHA) classification. The exclusion criteria include 1. Discharged within 48 h; 2. Discharged against medical advice or died in the first 6 days; 3. Acquired pneumonia or serious cerebrovascular accidents in the first 6 days of ICU admission.

A standardized weaning protocol has been applied for all included patients. We divided patients into two groups according to the duration of MV (PMV group: prolonged MV group, defined as MV > 6 days; non-PMV group: non-prolonged MV group, defined as MV ≤6 days). Medical records were reviewed to obtain information about gender, age, concomitant diseases, types of surgery, the 1st day SOFA, the duration of extracorporeal circulation (ECC), vasopressor and inotrope treatment, and the duration of MV and ICU stay. We also collected hemodynamics and tissue perfusion data at two time points (T0: admission; T1: 48 h after admission; both time points were approximate, within +/− 3 h) from the Critical Care Monitor System of the Department of Critical Care Medicine at Peking Union Medical College Hospital, which recorded real-time clinical data from bedside equipment. This system is maintained by Donghua software cooperation through DtHealth system. We also collected the PaO2/FiO2 (P/F) data on the first and second day in the ICU.

A descriptive analysis was performed. Continuous variables are expressed as the mean ± standard deviation, and categorical variables are expressed as absolute values and percentages. For continuous variables, the data were analyzed using Student’s t-test, the Mann-Whitney U test or the Kruskal-Wallis test depending on data distribution and the number of variables. Categorical variables were analyzed using the chi-square test or Fisher’s test.

Variables were introduced into a multivariable binary logistic regression model if they were significantly associated with prolonged MV in the univariate analysis (p value < 0.2). General demographics were also used in the model. Discrimination of values was performed using the receiver operating characteristic (ROC) analysis. All comparisons were two-tailed, and P < 0.05 was required to exclude the null hypothesis. The statistical analysis was performed using IBM SPSS Statistics, Version 20.0 (Armonk, NY: IBM Corp).

There were no significant differences in gender, age, types of surgery (CABG (coronary artery bypass grafting), pericardiectomy, valve surgery, ventricular septal defect repair, atrial neoplasm resection, aorta replacement) or concomitant diseases (hypertension, diabetes, CAC, CKD, CHF) between the two groups. The PMV group had a significantly higher SOFA score than the non-PMV group (PMV group vs non-PMV group: 12.56 ± 2.71 vs 11.09 ± 2.01, P = 0.003). No significant differences were found in the use of inotropes between the two groups during the three periods. The demographics and clinical characteristics of all the patients are presented in Table 1.

There were no differences in HR_T0 (heart rate), SBP_T0 (systolic blood pressure), CVP_T0 (central venous pressure), P/F_T0 (PaO₂/FiO₂), lactate_T0, CI_T0 (cardiac index), SVI_T0 (stroke volume index), SVRI_T0 (systemic vascular resistance index), Ea_,T0, Ees_,T0 or Ea/Ees_,T0 between the two groups (Table 2). Compared to the non-PMV group, the PMV group had significantly higher values for HR_T1 (P = 0.018), lactate_T1 (P < 0.001), and Ea/Ees_,T1 (4.05 ± 1.20 VS 5.93 ± 1.81, P < 0.001) and lower values for Ees_,T1 (P = 0.022), SVI_T1 (P = 0.049), SBP_T1 (P = 0.014).

There was a significant correlation between Ea/Ees,_T1and the duration of MV (Fig. 3); the correlation coefficient was 0.512 (Ea/Ees,_T1,P < 0.01). There were no correlations between SVI_T1 or CI_T1 and the duration of MV.

The multivariable logistic regression analysis of the data at T1 demonstrated that SBP_T1 (OR 0.921; 95% CI 0.853–0.994, P < 0.05), Ea/Ees_,T1 (OR 6.840; 95% CI 1.031–45.359, P < 0.05) and lactate_T1 (OR 15.269; 95% CI 1.231–189.427, P < 0.05) were independent predictors of PMV in severe post-cardiac surgery patients (Table 3).

ROC curves were drawn to compare the predictive values of three different variables for PMV (Fig. 4). The AUC demonstrated that the predictive values of SOFA, Ea/Ees_,T1 and lactate_T1 were 0.766 (95% CI:0.680–0.952), 0.801 (95% CI:0.664–0.938), 0.816 (95% CI:0.680–0.952), respectively (Table 4). The cutoff values for SOFA (1st day in the ICU), Ea/Ees_,T1 and lactate_T1 were 12.5 (sensitivity: 70%; specificity: 80%), 5.12 (sensitivity: 65%; specificity: 90%) and 1.60 (sensitivity: 75%; specificity: 85%), respectively, based on the maximum Youden index.

When weaning begins, the cardiac preload and afterload of the LV increase dramatically, which requires the LV to increase its SW. If a patient’s heart has a greater SW reserve before weaning, which indicates better VAC, it will be much easier to achieve adequate SW and wean the patient earlier. In several previous studies, SW before weaning was considered a predictive factor for PMV [7]. Actually, a heart can maintain adequate SW by sacrificing its reserve. Under this circumstance, CO or SV may be not altered, but the VAC will be changed. Therefore, it is the SW reserve and not the absolute SW that is important for early weaning. Our study confirmed this with the finding of no correlations between CI or SVI and the duration of MV in severe post-cardiac surgery patients.

Several studies [20, 21] have recommended VAC reserve as a significant tool for assessing cardiovascular function and prognosis. Tonino [20] et al. found that the change in VAC during stress echocardiography can be used as to predict adverse events in patients with negative stress echocardiography. In their study, they defined the change as the VAC reserve (VAC reserve = VAC_before-VAC_after), which can reflect the cardiac compensatory function capability. Kensuke [21] et al. also proved that a good VAC reserve was an important determinant of cardiovascular outcome for patients with dilated cardiomyopathy. In the present study, we found that an increased Ea/Ees_,T1 (OR 6.840; 95% CI 1.031–45.359, P < 0.05) was a risk factor for prolonged MV. Hence, the assessment and optimization of the left VAC might be helpful during ventilator weaning in severe post-cardiac surgery patients. Further studies are required to validate VAC-directed weaning from ventilation.

Vasodilators and inotropes are always used in severe post-cardiac surgery patients to assist in weaning from MV, which also optimizes VAC. Vasodilators, which decrease Ea by reducing the afterload, and inotropes, which augment Ees by strengthening contractility [22], are two important therapies that can improve VAC, which is crucial for ensuring a short duration of MV. Several studies have shown that vasodilators and inotropes can facilitate the weaning process [9, 23]. Routsi [9] et al. used a nitroglycerin infusion to control systolic blood pressure, and it greatly improved the weaning process of difficult-to-wean patients. Despite concerns about inotrope use, the researchers verified that levosimendan and dobutamine can facilitate weaning in difficult-to-wean COPD patients [24]. Moreover, fluid management also be favorable for VAC. Teboul [5] et al. suggested fluid removal as a reasonable option to help difficult-to-wean patients. Removing excessive fluid can reduce left ventricular end-diastolic pressure [25], which can increase coronary perfusion pressure and ameliorate myocardial perfusion [26].

Several limitations should be acknowledged. First, leaving V₀ out of the calculation formula and using GEDV/4 as LVEDV might be an arbitrary decision. However, we had several reasons for making this decision: (a) we considered the heart as a whole, because PMV is not only a problem of poor left ventricular function but also a thing of right ventricle [27]. It might be more proper to combine right and left ventricle together, so we used GEDV/4, which can take both cardiac chambers into consideration, to address this problem. (b) the value of GEDV as a reflection of the preload of the LV has been confirmed in many studies [28], and (c) GEF, calculated by 4SV/GEDV, has a good correlation with LVEF, calculated by SV/LVEDV [29]. Most importantly, in 2015, our department had done a research about sepsis and this “revised VAC” previously, and confirmed its significance in managing sepsis. Further studies are required to validate these choices; however, to some extent, they were reasonable. Second, this was a single-center study with a small sample size. Further investigations with large sample sizes need to be completed to confirm our results. Third, VAC is influenced by many factors: gender, age, concomitant diseases, etiologies, etc. [30‐32]. We analyzed these possible variables and found no differences between the two groups. In addition, there were no differences in VAC at admission. We believe that the significance of the differences in VAC was not caused by these irrelevant factors. Fourth, in this study, we aimed to determine what we should do before weaning and tried to recognize patients at risk for prolonged MV patients early in their stay; consequently, we chose the first 48 h as our study period [33, 34]. Fifth, existing research leaves no doubt that VAC is a measure of cardiac work efficiency. Considering our thorough descriptions of VAC and its role in prolonged MV, there is adequate support for our views and conclusions. Further studies must be performed to confirm our perspectives in practice.