The transition and outcomes of perioperative low ejection fraction status in cardiac surgical patients

Department of Nursing, Zhongshan Hospital, Fudan University, 200032 Shanghai, China Department of Critical Care Medicine, Zhongshan Hospital, Fudan University, 200032 Shanghai, China Department of Surgery, Massachusetts General Hospital, Harvard Medical School, Boston, MA 02114, USA Center of Clinical Epidemiology and EBM of Fudan University, 200032 Shanghai, China Department of Nutrition, Zhongshan Hospital, Fudan University, 200032 Shanghai, China Department of Critical Care Medicine, Xiamen Branch, Zhongshan Hospital, Fudan University, 361015 Xiamen, Fujian, China Shanghai Key Lab of Pulmonary Inflammation and Injury, 200032 Shanghai, China

Low left ventricular ejection fraction (LVEF) was always considered a high-risk factor for surgery. A growing number of patients with preoperative low LVEF have undergone cardiac surgery in recent years. The transition of postoperative LVEF and its correlation with shortterm outcomes is not yet clear. We retrospectively collected the clinical data of cardiac surgery patients with low preoperative LVEF (≤40%). LVEF measurements were collected preoperatively and at least twice postoperatively. The primary endpoint was the composite endpoint of hospital mortality or length of intensive care unit (ICU) stay ≥7 days. Univariate logistic regression was used to evaluate the association of each indicator with the outcomes, including calculation of the area under the receiver operating characteristic (ROC) curve. A two-piecewise linear regression model was applied to examine the threshold effect of the LVEF on the composite endpoint using a smoothing function. From 1 January to 31 December 2018, a total of 123 patients had low LVEF preoperatively, of whom 35 (28.5%) met the composite endpoint. LVEF was 35% [interquartile range (IQR) 30%-42%] at first measurement and increased to 40% (IQR 35%-45%) at final measurement during their hospitalization. There was a linear relationship between composite endpoint and lowest level of postoperative LVEF. The base e logarithm of odds ratio [Ln(OR)] of composite endpoint decreased with increasing LVEF (OR = 0.83, 95% confidence interval 0.76-0.91, p < 0.01). Most patients with low preoperative LVEF will benefit from cardiac surgery. The lowest measurement of postoperative LVEF can be used to evaluate the shortterm outcome of patients after cardiac surgery.

Background
Left ventricular ejection fraction (LVEF) is the most frequently used indicator of cardiac function [1,2]. Patients with low preoperative LVEF are at high risk in cardiac surgery, with associations with substantially elevated postoperative morbidity and mortality [3,4]. With improvements to surgical procedures and the use of various circulatory support devices and drugs, outcomes after cardiac surgery have improved over time, leading to a decrease in surgical contraindications and broad adjustment in the range of preoperative risk assessment [5,6]. The safety of cardiac surgery among patients with preoperative cardiac insufficiency has been well established by several studies [7][8][9].
A large number of studies have verified the correlation between preoperative LVEF and patient prognosis, but little attention has been paid to the transition of postoperative LVEF. LVEF is usually monitored on an ongoing basis to evaluate the outcome of the surgery and the patients' recovery of cardiac function, sometimes as reference for transfer out of the intensive care unit (ICU) or discharge [10,11]. Patients with shorter-term changes in LVEF after surgery are of concern to clinicians, especially those who already had low LVEF before surgery. Studies have shown that the LVEF of some patients does not improve from surgery but decreases in the early postoperative period, confusing clinicians about surgical and patient outcomes [12,13]. Further studies have revealed that preoperative LVEF may be affected by the structural compensation of the heart itself or other factors that cannot fully reflect actual cardiac function [14,15]. The prognostic value of LVEF and its short-term changes among cardiac surgery patients remains understudied in the real world. Questions concerning low LVEF in patients need answering to guide treatment decisions.
The aim of this study was to investigate the change in LVEF and prognosis of patients with low preoperative LVEF after cardiac surgery. At the same time, we explored the predictive value of postoperative LVEF for patient prognosis.

Patients
This retrospective cohort study enrolled patients with preoperative low LVEF who underwent cardiac surgery between 1 January and 31 December 2018 in Zhongshan Hospital, Shanghai, China. Patients younger than 18 years of age and pregnant patients were excluded from the study.
All patients underwent formal two-dimensional echocardiography less than 2 weeks prior to surgery and at least twice after surgery during hospitalization. Patients received standard anesthetic care and monitoring during the procedure and were transferred to the intensive care unit (ICU) after surgery. Standard care was provided by intensivists, respiratory therapists and ICU nurses. Standard therapy included fluid management, airway management, vasoactive medications, and inotropic and mechanical circulatory support according to hemodynamic status. The decision to discharge from the ICU was made by intensivists, based on the patient's general condition. Data were collected from the electronic medical records during hospitalization.

Data collection
Preoperative characteristics, including patient demographics and comorbidities were collected retrospectively. The latest measurement of preoperative LVEF (LVEF_Pre), first measurement of postoperative LVEF (LVEF_First), lowest level of postoperative LVEF (LVEF_Low), and last measurement of postoperative LVEF (LVEF_Last) were recorded. Laboratory measurements including creatinine, cardiac troponin T (cTnT), and N-terminal pro-brain natriuretic peptide (NT-proBNP) and were measured preoperatively and within 24 h after the surgery. In addition, we collected information about patients' procedures, supportive therapies (such as intra-aortic balloon pump, renal replacement therapy and inhaled nitric oxide), and clinical outcomes during hospitalization. All data were obtained from the patients' electronic medical record system.
The LVEF and tricuspid annular plane systolic excursion (TAPSE) values were measured using echocardiography, and the LVEF was evaluated using Simpson's biplane method. Comorbid conditions were evaluated using the Charlson Comorbidity Index. EuroSCORE was used to calculate the predicted operative mortality for patients undergoing cardiac surgery. APACHE II was used to evaluate the severity of the patient's condition after patients were transferred to ICU following surgery. These scores were evaluated by the intensivists.

Definition
Preoperative low LVEF was defined as the latest measurement of preoperative echocardiography showing LVEF of ≤40% and was calculated using Simpson's biplane method [16]. The primary endpoint was the composite endpoint of hospital mortality or length of ICU stay ≥7 days.

Statistical analysis
Discrete variables were expressed as numbers and percentages and compared using the Fisher's exact test. According to the normality of the data, continuous variables were expressed as mean ± standard deviation or median (interquartile range) and compared using the student's t test or Wilcoxon rank-sum test. Logistic regression models were used to investigate the relationship between indicators and composite endpoints in univariable analyses. The area under the receiver operating characteristic curve (AUROC), with 95% confidence interval (CI), was used to compare the prediction of composite endpoints by each indicator.
A two-piecewise linear regression model was applied to examine the threshold effect of the LVEF on the composite endpoint using a smoothing function. The threshold level (i.e., turning point) was determined by trial and error, including by selection of turning points along a pre-defined interval and then choosing the turning point that gave the maximum model likelihood. A log likelihood ratio test was conducted to compare the one-line linear regression model with a twopiecewise linear model. Statistical tests were two-tailed, and a value of p < 0.05 indicated statistical significance. Statistical analyses were performed using R, version 3.6.2 (R Foundation for Statistical Computing, Vienna, Austria).

Clinical outcomes and related factors
Of 123 patients, four patients died, giving a mortality rate of 3.3%. The overall length of ICU stay was 4 days (IQR 2-7 days) and length of hospital stay was 17 days (IQR 13-24 days). A total of 35 patients (28.5%) met the composite endpoint of death or ICU ≥7 days. Pre-and postoperative level of creatinine, cTnT, NT-proBNP, Charlson comorbidity index and APACHE II score were significantly higher in patients who met the composite endpoint. Patients who met the composite endpoint also had lower pre-and postoperative LVEF, longer mechanical ventilation, longer ICU stay, and more frequently had IABP (Table 1).
Univariate logistic regression analysis showed that higher preoperative cTnT (per 0.1 ng/mL, OR = 3.52, 95% CI 1.75-8.51), APACHE II score (OR = 2.01, 95% CI 1.35-3.16), postoperative cTnT (per 0.1 ng/mL, OR = 1.78, 95% CI 1.24-2.84), and postoperative NT-proBNP (per 10000 pg/mL, OR = 2.78, 95% CI 1.76-4.93) were risk factors for the composite endpoint of death or ICU ≥7 days. Lower measurement of first and lowest postoperative LVEF were also risk factors. To avoid overfitting, only three parameters were used for multivariate analysis. The results are presented in Appendix Table 3 to demonstrate the correlation between LVEF and composite outcome. The ROC curve analysis is summarized by the area under the curve (AUC) at 95% CI and p-values in Fig. 2. NT-proBNP had the highest AUC value among the analyzed variables.

Association between left ventricular ejection fraction and composite outcome
Two smooth curves were fitted, and the relationship between LVEF_Pre and LVEF_Low with composite endpoint was assessed. Adjusted smoothed plot shows a non-linear relationship between composite endpoint and LVEF_Pre (p for likelihood ratio test (LRT) <0.05). The relationship between Ln(OR) of composite endpoint and LVEF_Pre was not significant up to the turning point (LVEF = 34%; p > 0.05). After the turning point, the Ln(OR) of the composite endpoint decreased with increasing LVEF (OR = 0.58, 95% CI 0.42-0.80, p = 0.001). However, there was a linear relationship between the composite endpoint and LVEF_Low (p for LRT = 0.12). The Ln(OR) of the composite endpoint decreased with increasing LVEF (OR = 0.83, 95% CI 0.76-0.91, p < 0.01) (Table 2, Fig. 3).

Discussion
Patients with preoperative low LVEF had increased mortality and postoperative morbidity after cardiac surgery. In our study, the overall mortality rate was 3.3% and a total of 35 patients (28.5%) met the composite endpoint, which was similar to reports in the literature. In one cohort study, 781 patients underwent cardiac surgery with a preoperative LVEF ≤40%, with a perioperative mortality rate of 5.6% [3]. Another study included a total of 588 cardiac patients with a preoperative LVEF ≤25%, and the overall operative mortality was 7.8% [7]. A study among off-pump coronary artery bypass patients found that 30-day mortality was 2.9% in 137 pa-   tients with ejection fraction (EF) ≤35% [8]. Differing definitions of low LVEF, condition of patients and operative methods can be considered as leading to discrepancies in patient mortality among studies.
Owing to the relatively small sample size and few events, only univariate analyses were conducted. Some factors, including pre-and postoperative cTnT, NT-proBNP, APACHE II, and LVEF differed significantly between the groups. These factors might be correlated with poor progno- sis. On further analysis of the data, postoperative indicators had larger OR values and larger areas under the ROC curve, which seemed to suggest that postoperative indicators were more closely related to patient outcomes than preoperative indicators [17,18]. In this cursory statistical study, we found a trend that LVEF decreased after surgery and then gradually increased. Few studies have focused on the short-term changes in ventricular performance and LVEF before and after cardiac surgery. Sugimura et al. [19] reviewed the perioperative and 1-year follow-up data of 436 patients with primary mitral regurgitation to analyze patients' postoperative evolution of LVEF and its factors. The result showed that overall mean LVEF slightly decreased at 1-year follow-up (mean change of LVEF: -2.63% ± 9.00%). Newman et al. [20] measured scintigraphic LVEF 16 weeks after coronary bypass surgery. No change in resting ejection indices were detected and enddiastolic volume increased slightly only in those patients with postoperative symptoms. In a study of patients undergoing coronary artery bypass surgery, Mintz et al. [21] found a decrease in LVEF at 1 week when compared with preoperative measurements. Ejection indices returned to preoperative levels by 2 months and remained stable over the year follow-up. In our study, we found that LVEF did change with surgical intervention, and this change was more often a transient deterioration, which was also consistent with laboratory indicators. With postoperative treatment, the LVEF of most patients could be improved. Such changes are acceptable and prove that the majority of patients with low LVEF can benefit from cardiac surgery.
With further exploration of the relationship between LVEF and prognosis, we found that an increased log odds ratio (OR) of composite endpoint was linearly related to the decreased postoperative LVEF, but not significantly linearly related to preoperative LVEF. Multiple studies have explored the association between LVEF and prognosis in cardiac surgery patients and have found that preoperative LVEF was an independent predictor affecting long-term prognosis. In a study by Furer et al. [22], patients who underwent transcatheter aortic valve replacement with preoperative LVEF <50% were included for analysis. The results showed that preoperative LVEF was associated with increased 2-year risk of both cardiovascular mortality and allcause mortality. Dahl et al. [23] conducted a retrospective analysis of the data of patients with severe aortic stenosis and found that preoperative LVEF was a powerful predictor of outcome in patients with severe aortic stenosis undergoing aortic valve replacement. However, these two studies did not include postoperative LVEF in the analysis, nor did they analyze the linear correlation between LVEF and prognosis. There were also some differences in the characteristics of the subjects compared with those in our study.
We believe that postoperative LVEF can better predict the short-term outcomes of patients undergoing cardiac surgery. On the one hand, there is still controversy over whether LVEF reflects true cardiac function. For example, in patients with mitral regurgitation, preoperative LVEF measured by routine echocardiography may underestimate myocardial systolic dysfunction due to structural changes [24]. On the other hand, cardiac surgery is an important factor affecting the prognosis of patients. Postoperative LVEF not only reflects the preoperative cardiac function of patients, but also reflects the effect of cardiac surgery on patients. From further analysis of postoperative LVEF in this study, we recommend that short-term prognosis of patients with preoperative low LVEF should be evaluated according to postoperative LVEF.

Limitations
There were some limitations that should be addressed in this study. First, this was a retrospective cohort study with a small sample size. The number of patients with low LVEF before cardiac surgery was limited, even though they were enrolled from a large clinical cohort of over 4000 patients. Further investigation of a larger population is required, to increase statistical power and make the current results more convincing. Second, we only focused on patients' shortterm outcomes during their hospitalization. As the longterm follow-up of these patients had not yet started, data on changes to long-term LVEF and prognosis are missing. A larger sample size and more long-term follow-up data would have made our conclusions more comprehensive and meaningful. Third, only Chinese people were included in this study. Whether this conclusion is also applicable to patients of other races needs further verification.

Conclusions
The mortality rate of patients with low preoperative LVEF was very low at a large-volume cardiovascular center. Most patients experience a short decline in LVEF after surgery and then an increase during hospitalization. The lowest postoperative LVEF was associated with short-term prognosis of patients after cardiac surgery.

Ethics approval and consent to participate
The study was conducted in accordance with the Declaration of Helsinki, and the protocol was approved by the Ethics Committee of Zhongshan Hospital, Fudan University (B2019-075R). Individual consent for this retrospective analysis was waived.