Consecutive patients with moderate IMR and scheduled for mitral valve repair and simultaneous CABG between January 2010 to December 2017 were identified from medical records. The inclusion criteria were: (1) previous myocardial infarction (MI) indicated by regional wall motion abnormalities revealed by echocardiography, or as detected by electrocardiogram; (2) sinus rhythm; and (3) no structural mitral valve abnormalities. The exclusion criteria were: (1) echocardiographic evidence and/or clinical manifestations of other structural heart diseases; (2) organic mitral apparatus abnormalities; (3) unstable global clinical status; (4) atrial fibrillation that was not appropriate for this study because it was reported to cause atrioventricular valve regurgitation [14]; (5) concomitant tricuspid annuloplasty; and (6) emergency surgery.

Within 3 days before surgery, routine transthoracic echocardiography was performed to assess the severity and mechanism of MR. The IMR level was classified as mild (narrow central jet area less than 20% of left atrium (LA) and vena contracta less than 3.0 mm under Doppler), moderate (regurgitant volume less than 30 mL, effective regurgitant orifice area (EROA) less than 20 mm${}^2$, and regurgitant fraction less than 50%), or severe (regurgitant volume over 30 mL, EROA over 20 mm${}^2$, and regurgitant fraction over 50%). MR suggested by echocardiography was evaluated by two independent professional readers, with a third reader used when discrepancies arose. Examinations were performed according to current guidelines [15].

All procedures began with a midline sternotomy. The detailed protocol for on-pump CABG was described in a previous study [16]. After grafting, the quality of anastomosis was evaluated during the operation using a transit-time flow probe. Technical details regarding our institutional approach to mitral valve repair were described previously [17]. After weaning off the bypass, intraoperative transesophageal echocardiography was performed immediately to determine the quality of mitral valve repair. If moderate or more residual MR was observed, a repeat procedure was executed immediately.

This single-center, retrospective, observational study received approval from the medical ethics committee of Zhongshan Hospital, Fudan University (No. B2022-024R), and followed the principles of the Declaration of Helsinki. Similar to a previous report [8] in which an EF of 40% was used as the cut-off point to discriminate depressed LV function, EF 40% was used in the present study to define depressed LV function. All patients included in this study were assigned to either the EF 40% group or the EF $\ge$40% group. Baseline characteristics and surgical data were extracted for all patients, and in-hospital and follow-up outcomes were compared between groups. An electronic in-hospital database was used to access baseline characteristics, as well as in-hospital outcomes. A standard case report form was used to record the data. Enrolled patients were routinely followed up postoperatively at three and six months, and afterward at 6-month intervals. Telephone interviews and clinical visits were used to obtain follow-up data. A clinical visit was scheduled if recrudescent symptoms of coronary artery disease or questionable symptoms of MR appeared during the follow-up.

In-hospital outcomes consisted of surgical mortality and major postoperative morbidity. Death during the same hospitalization or within 30 days after surgery would be recognized as surgical mortality [18]. Major postoperative morbidities were CABG-associated MI, prolonged ventilation, low cardiac output, new-onset stroke, acute kidney injury requiring hemodialysis, redo for bleeding, and deep sternal wound infection (DSWI). CABG-associated MI was diagnosed as elevation of cardiac troponin T (cTnT) values to $>$10 times the 99th percentile of the upper reference range using one or more of the following methods: (1) new onset pathological Q wave or left bundle branch block recorded with electrocardiography; (2) new occlusion of graft or native coronary artery documented by angiography; (3) new viable myocardium loss or abnormal regional wall motion manifested by imaging [19]. Low cardiac output was recorded when an intra-aortic balloon pump (IABP) and/or positive inotropic agent support became necessary for difficulty in weaning off the cardiopulmonary bypass, or when longer than 30 mins was required to maintain the cardiac index $>$2.2 L/min/m${}^2$ with systolic blood pressure $>$90 mmHg after the patient returned to intensive care unit (ICU) [20]. Postoperative prolonged ventilation was identified as when the mechanical ventilation exceeded 48 hours, or when re-intubation was required after surgery. New-onset stroke was considered to be a new onset of global or focal brain dysfunction occurring over 24 hours, or permanent neurological damage persisting until either discharge or death [21]. The definition of DSWI was the same as in our previous study [22]. The duration of the postoperative hospital stay and of the ICU stay were counted.

Follow-up outcomes included all-cause death, reoperation (including repeat mitral valve procedure and repeat revascularization), moderate or severe MR, and NYHA functional class. All-cause mortality is the most unbiased and robust index and was therefore selected rather than cardiac mortality. This helped to avoid misinterpretation of the cause of death due to unreliable medical records. The minimum follow-up time in this study was 30 months. Follow-up information obtained at 30 months after surgery was used for the analysis of NYHA classification and residual MR.

The normal distribution of variables was determined using the Shapiro-Wilks test. An independent-samples t-test was used while comparing normally distributed continuous variables between groups, which were exhibited as the mean $\pm$ standard deviation (SD). The Wilcoxon rank sum test was performed on non-normally distributed continuous variables, which were exhibited as the median and interquartile range (IQR). Categorical variables were exhibited as frequencies and percentages, which were compared using the Chi-square test between groups, or using the Fisher’s exact test when the expected frequency was 5.

The propensity score (PS) was generated for each patient using a multivariable logistic regression model to control potential confounders in the dataset. The PS was performed according to 17 independent variables, with LV function-based grouping (EF 40% vs. EF $\ge$40%) used as a binary dependent variable. Demographics, complications, cardiac status (except for EF and EuroSCORE), and EROA were included in the 17 variables. The model was verified with the Hosmer-Lemeshow goodness-of-fit method. The greedy-matching algorithm was used with a caliper width of 0.2 of the SD of the logit of the PS, thus implementing a 1:1 nearest-neighbor. Other details of the PS matching could refer to our former study [16]. After matching, paired t-test was used for normally distributed continuous variates. The Wilcoxon signed-rank test was used for non-normally distributed continuous variates. McNemar’s test was used for categorical variates. Conditional mixed-effects logistic regression analysis was applied to evaluate the effect of grouping as independent risk factors. The Kaplan-Meier method was applied to estimate the overall survival and survival free from reoperation with the stratified log-rank test used to compare survival curves in the PS-matched population. Between the two matched groups, the Cox regression model was utilized to estimate the PS-adjusted hazard ratio (HR) and 95% confidence interval (CI) of midterm mortality. Statistical analysis was performed using SPSS version 22.0 (SPSS Inc., Chicago, IL, USA), with a two-sided p-value 0.05 considered to represent statistical significance.

A total of 5336 consecutive patients in our department received surgical revascularization with or without other concomitant cardiac surgery between January 2010 to December 2017. Amongst these, 322 patients were eligible according to the inclusion criteria. During patient enrollment, as shown in Fig. 1, 50 patients were ruled out, leaving 272 patients for data analysis. Of these, 90 patients were enrolled in the EF 40% group, and 182 patients in the EF $\ge$40% group.

Fig. 1.

Flow chart for the patient enrollment. IMR, ischemic mitral regurgitation; CABG, coronary artery bypass grafting; pts, patients; EF, ejection fraction.

Baseline characteristics of the patients are shown in Table 1. Patients in the EF 40% group had a larger LV endo-diastolic diameter (64.5 $\pm$ 7.8 mm vs. 59.2 $\pm$ 7.1 mm, p 0.001), a higher proportion of patients in NYHA class III–IV (p = 0.023), and a higher additive EuroSCORE (p = 0.012) compared with patients in the EF $\ge$40% group. Baseline characteristics were otherwise similar between groups.

Surgical data are shown in Table 2. No significant difference was discovered between groups in terms of cardiopulmonary bypass time or aortic cross-clamping time. The number of grafts was also similar between the two groups (p = 0.653). Annuloplasty was performed on 232 patients using a downsized complete rigid ring for 84.4% of the EF 40% group and 85.7% of the EF $\ge$40% group (p = 0.781). No significant difference was apparent for the type of repair techniques used in the two groups (p = 0.667). Transesophageal echocardiography examination revealed that moderate or severe MR was not found in either group immediately after weaning off the bypass.

To compare baseline characteristics between the two groups, we performed bivariate analyses. The propensity score was calculated based on 17 predefined variables. The model’s Hosmer-Lemeshow goodness of fit was 4.65 (p = 0.793). Furthermore, good discrimination power was achieved with an area under the curve of 0.78 (95% CI, 0.65–0.84, p = 0.019) of the receiver operating curve. Ultimately, 82 patient pairs were matched at a 1:1 ratio. After matching, Fig. 2 shows that the absolute standardized differences were all 10%, which indicates adequate balance. Except for EF, the matched cohorts were comparable for baseline characteristics (Table 1). In addition, no significant difference was found with regard to surgical characteristics between the two matched groups.

Fig. 2.

Pre-match and post-match absolute standardized differences for baseline characteristics. NYHA, New York heart assessment functional classification.

The in-hospital outcomes are shown in Table 3. Patients in the EF 40% group had slightly higher surgical mortality, but this reached no statistical significance (p = 0.076). More patients in the EF 40% group developed low cardiac output (8.9% vs. 2.7%, p = 0.034) and received intra-aortic balloon pump support (10.0% vs. 3.8%, p = 0.042) compared to patients in the EF $\ge$40% group. More patients in the EF 40% group also developed prolonged ventilation (13.3% vs. 5.5%, p = 0.026). Major postoperative morbidity was otherwise similar between the two groups, including CABG-associated MI, redo for bleeding, new-onset stroke, DSWI, and acute kidney injury requiring hemodialysis. The matched groups of patients showed similar major postoperative morbidity, with the exception of prolonged ventilation. Patients in the EF 40% group had longer ICU stays and longer postoperative hospital stays compared with those in the EF $\ge$40% group, both before and after PS matching.

The effects of grouping (matched EF 40% group vs. matched EF $\ge$40% group) on major postoperative morbidity and surgical mortality are shown in Table 4. Conditional mixed-effects logistic regression analysis revealed that EF 40% had an independent impact on postoperative prolonged ventilation (OR = 2.814, 95% CI 1.321–6.151, p = 0.031). However, EF 40% was not an independent risk factor for surgical mortality (OR = 2.967, 95% CI 0.712–7.245, p = 0.138), nor was it an independent risk factor for other major postoperative morbidities.

Follow-up visits were completed with 243 patients in total. The median follow-up time was 42 months (IQR, 34–50), with the shortest follow-up time being 30 months. At the 30-month follow-up, the incidence of moderate or more MR and the proportion of NYHA class III and IV did not differ either before or after matching (Table 3). As shown in Fig. 3, similar cumulative survival was shown both before and after PS matching (log-rank p = 0.278, stratified log-rank p = 0.832, respectively). No significant difference in cumulative survival free from reoperation was found between the two groups, either before or after matching (log-rank p = 0.425, stratified log-rank p = 0.729, respectively) (Fig. 4). Finally, Cox regression analysis was utilized to estimate the follow-up death in the matched cohorts. As shown in Fig. 5, grouping based on EF (EF 40% vs. EF $\ge$40%) was not related to midterm mortality (PS-adjusted HR 1.151, 95% CI 0.763–1.952, p = 0.281).

Fig. 3.

Kaplan-Meier curves for overall survival. (A) Kaplan-Meier curves in the unmatched cohorts. (B) Kaplan-Meier curves in the matched cohorts. EF, ejection fraction.

Fig. 4.

Kaplan-Meier curves for survival free from reoperation. (A) Kaplan-Meier curves in the unmatched cohorts. (B) Kaplan-Meier curves in the matched cohorts. EF, ejection fraction.

Fig. 5.

PS-adjusted Cox regression analysis in the matched population. PS, propensity score; HR, hazard ratio; CI, confidence interval; EF, ejection fraction.