Introduction

Post-infarction ventricular septal rupture (PIVSR) is one of the most severe types of mechanical complications after acute myocardial infarction (AMI), which has high mortality and poor prognosis [1,2,3,4,5]. And if left untreated, it is almost inevitably fatal [6]. Currently, improvements in revascularization, including pharmacological, catheter-based and surgical, have resulted in improved outcomes for patients with AMI [1, 6, 7], but in-hospital mortality of this mechanical complication after AMI remained still as high as 60% and the prognosis has not improved over the past 2 decades [8,9,10], derived from the unavoidable remodeling based on a large amount of infarcted myocardium. Early diagnosis and risk stratification are crucial to improving outcomes [5]. However, studies on this disease in recent years are still limited, mainly because of the rarity for this disease [11, 12]. Moreover, most of the previous studies came from single-center studies or case reports [11, 13,14,15,16], which would limit further explorations for better clinical management of PIVSR. Crenshaw BS et al. had identified risk factors associated with increased mortality in patients who developed PIVSR [17]. Thus, identifying robust predictors of the early mortality might help clinical decision and improve the prognosis of this population, which would be a meaningful way to improve the current complex situation. Based on these, we conducted this present real-world multi-center observational study to evaluate the mortality of PIVSR and try to identify the risk predictors for early mortality, and then provide some information regarding the management of such patients [18].

Methods

This was a multi-center, retrospective registry designed to reflect the “real world” clinical practice since 2013. The primary data were extracted from the electronic medical or archived records. Data on patient demographics, clinical features, echocardiography features, and outcomes were collected for all patients. The data were independently reviewed by two researchers in the data collection process. Standardized definitions for all patient-related variables and clinical diagnoses were used. Participant data had been anonymized and had not distorted the scholarly meaning.

Data source and population

From June 2013 to December 2022, we retrospectively studied patients hospitalized for PIVSR at the Second Affiliated Hospital of Army Medical University, the First Affiliated Hospital of Army Medical University and the First Affiliated Hospital of Chongqing Medical University, Chongqing, China. The inclusion criteria for this study were as follows: (1) Age > 18 at inclusion; (2) With the evidence of left-to-right shunt in ventricular septal based on the ultrasonic cardiogram; (3) Definitely diagnosed as AMI, including ST-segment elevation myocardial infarction (STEMI) or non-ST-segment elevation myocardial infarction (NSTEMI). The diagnosis of AMI was based on typical clinical symptoms, electrocardiographic findings of ST elevation 0.1mV in more than two limb leads or > 0.2mV in two or more contiguous precordial leads, as well as cardiac enzyme elevation. The exclusion criteria for this study were as follows: (1) Ventricular septal defect caused by congenital heart disease or traumatic heart injury rather than caused by AMI; (2) Patients with malignant tumors or end-stage diseases; (3) No confirmation of the ultrasonic cardiogram; (4) Patients with the ambiguous time for AMI. Finally, 62 patients were included in the study (Fig. 1). During the retrospective screening of the medical records, the baseline and procedural characteristics of enrolled patients were collected for further analyses, as well as the relevant laboratory data. The definition of each variable was in line with the cardiovascular data standards. Acute & Sub-acute type was defined as being present PIVSR within 72 h, and late type was defined as being present PIVSR more than 72 h. The diagnosis of diabetes mellitus are based on the standards of medical care in diabetes mellitus (2019) issued by The American Diabetes Association (ADA) [19]. The 2018 ESH/ESC guidelines for the management of arterial hypertension are referred to the diagnosis of hypertension [20]. Hyperlipidemia is a metabolic disease caused by abnormal fat metabolism, which is mainly manifested by an abnormal increase of Total cholesterol (TC), Triglyceride (TG) and low-density lipoprotein (LDL) levels in the blood. Cardiogenic shock definition was according to clinical and hemodynamic criteria, including hypotension [systolic blood pressure (SBP) < 90 mmHg for 30 min or need for supportive measures to maintain the SBP of > 90 mmHg] and evidence of end-organ hypoperfusion. The diagnosis of heart failure was according to the clinical diagnosis of heart failure and ejection fraction. All enrolled patients were divided into the survival group or the non-survival group based on whether death occurred within 30 days after diagnosis of PIVSR.

Fig. 1
figure 1

Flowchart for enrolling patients with PIVSR

Full size image

Statistical analysis

Baseline characteristics and clinical outcomes were expressed as number, percentage, or mean and standard deviation (SD) as appropriate. Numerical variables would be shown as median and interquartile range (IQR) values if the data were not normal distribution. categorical variables in the survival group and the non-survival group were compared using Fisher’s exact or chi-square test. Student’s test or Wilcoxon rank-sum test was performed for analyzing continuous data as appropriate. P values were 2-tailed, and P < 0.05 was statistically significant unless otherwise indicated.

We used the multiple imputation method in the MICE R package to fill in missing data. Then the univariable logistic regression was used for screening predictors of mortality. The multivariate logistic regression model was established using these variables. The values are related to a significant difference (P < 0.10) in the univariable logistic regression model and a significant difference (P < 0.05) in the multivariable regression model. And multicollinearity was evaluated by variable inflation factors (VIF). VIF > 5.0 was interpreted as indicating multicollinearity. Variables with VIF > 5.0 were not included in the final model analysis. The AUC, sensitivity and specificity were used to evaluate the model’s performance. Finally, the nomogram was plotted using the R package “rms”. The calibration C index (bootstrap resampling 1,000 times), the calibration curve (relationship between observation probability and prediction probability), Hosmer Lemeshow goodness of fit test (HL test), and brier score were used to evaluate the degree of consistency between observed and predicted outcomes. Decision curve analysis (DCA) was used to assess the net clinical benefit [21]. All analyses were performed using R language (version 4.2.1).

Results

Baseline comparison between the two groups

Only 18 patients survived in the first month and got discharged successfully, showing high mortality of PIVSR (71.0%) (Fig. 1). Of all the patients, mean age was 70.7 ± 10.7 years (38.7% female), 25.8% of patients (16/62) had hyperlipidemia, 43.5% of patients (27/62) had hypertension, 30.6% of patients (19/62) had diabetes mellitus, and the patients who had the history of ACS accounted for 16.1%. Most (n = 47 [75.8%]) of the patients were in Killip class III or IV at the time of rupture diagnosis. All the patients enrolled were diagnosed as STEMI. The baseline between the two groups were generally consistent and showed no statistical significance. Patients in the survival group were more likely to be a higher level of LVEF, but more likely to have a lower level of Serum glucose, Cardiac troponin I and WBC count. More details about the patients with PIVSR are shown in Table 1.

Table 1 Baseline characteristics of patients with PIVSR
Full size table

Procedural characteristics of patients with PIVSR

Most of patients (n = 56, 90.3%) were medically managed which meant treated conservatively. Few of patients (n = 6, 9.7%) underwent surgical repair or device closure. Interestingly, all the patients who received surgical repair or interventional closure survived and got discharged. The patients in the survival group had a significantly longer duration from AMI to PIVSR compared with those in the non-survival group, P = 0.020; And the non-survival group had higher proportion of suffering from cardiogenic shock or heart failure (HF). There are 40.3% of patients (25/62) who underwent either percutaneous coronary intervention (PCI) or performing coronary artery bypass grafting (CABG) only. And 34 patients received coronary angiography (CAG). Among them, 64.7% of patients (22/34) were complicated with multi-vessel lesions. The survival group had a higher proportion of chronic types (72.2% vs. 31.8%; p = 0.004). There are only 2 patients (11.1%) with intra-aortic balloon pump (IABP) implantation in the survival group, and 6 patients (13.6%) in the non-survival group doing. However, there was no significance between the two groups (11.1% vs. 13.6%; p > 0.900). In this study, the size of the rupture and the ICU duration have no significance between the two groups. The details are shown in Table 2.

Table 2 Procedural characteristics of patients with PIVSR
Full size table

Univariable and multivariable analysis predicating in-hospital death

Based on univariate analysis, nine variables (P < 0.1), including Cardiogenic shock, HF, PIVSR type, Revascularization, LVEF, log NT-pro BNP, Cardiac troponin I, WBC and Serum glucose, were related with the short-term mortality among these populations. Considering sample size and test efficiency, to confirm independent risk predictors of early mortality in patients with PIVSR and to avoid overadjustment and collinearity, HF, PIVSR type, Serum glucose, NT-pro BNP, Revascularization and Cardiac troponin I were adjusted by WBC count, Cardiogenic shock, and LVEF, based on the Akaike information criterion (AIC) under multivariate analysis (Table 3).

Table 3 Univariable and multivariable analysis predicating death in hospital
Full size table

Development of a nomogram

We further prudently used the univariable and multivariable regression analysis results and chose one of the lowest AIC score models to develop the nomogram. The nomogram for PIVSR including LVEF, Cardiogenic Shock, WBC and Revascularization (Fig. 2) was used to identify patients whose prognosis were likely to be poor. The calibration curve showed a good fit during internal validation, while the HL test showed that our predicted and observed values were close (P = 0.939). Our model yielded an AUC value of 0.956 (95% CI 0.912–1.000) (Fig. 2). Meanwhile, The DCA of the nomogram was performed (Fig. 3). Our results showed that our model had a good net clinical benefit in this population.

Fig. 2
figure 2

Nomogram to predict the risk of short-term mortality in patients with PIVSR

Full size image
Fig. 3
figure 3

DCA for our model

In DCA, the red curve in the figure represents that all the patients would dead in short term, the straight green line represents that patients would not dead in short term and the blue curve represents the clinical benefit of our model

Full size image

Discussion

In this multicenter retrospective cohort study, we found that the mortality of PIVSR remained high, was 71.0% (44/62). Meanwhile, we identified that WBC, LVEF, cardiogenic shock were the independent predictors of short-term mortality. Finally, we developed the nomogram for predicting the risk of short-term mortality.

Previous studies had indicated that leukocyte played an important role in systemic inflammatory reactions, and cardiogenic shock was commonly associated with a severe inflammatory response [22,23,24]. The high levels of total WBC count and C-reactive protein (CRP) may be considered as independent prognostic factors in patients with ACS [25,26,27].

Of note, a significant difference between two groups in LVEF was observed, suggesting an association between relatively low LVEF and increased mortality, even though LVEF in the non-survival group was still within the relatively normal range [28, 29]. Due to its simplicity and ease of observation, LVEF was one of the indicators traditionally used for the early identification of high-risk patients with AMI [28].

Schlotter et al. had found that PIVSR complicating AMI frequently leads to cardiogenic shock [30]. Attia R et al. had reported that the 30-day mortality was 65% with strong correlation with cardiogenic shock [22]. Our study also confirmed higher mortality in patients who developed cardiogenic shock, emphasizing improving the hemodynamic status of patients during clinical intervention was crucial [31] and the need for early and effective hemodynamic management in this subset of patients.

Furthermore, Phan DQ et al. had reported that revascularization strategies (with either PCI or CABG) were associated with benefit for ACS and all-cause mortality [32]. Several studies had indicated that coronary revascularization combined with the closure of rupture might be helpful in improving the prognosis of AMI patients [5, 29]. Based on these, we chose the revascularization which was significant in univariate analysis as one variable of the nomogram for PIVSR. The size of the rupture and the ICU duration maybe not the predictors for the short-term mortality according to the group comparison analyses. Patients might be considered receiving procedural treatment to improve prognosis rather than consider the size of rupture much more [22, 33]. Few of therapeutic regimens took into account the predicting model for this clinical complication. It was therefore necessary to inform the best-fitting combination of variables, associated with predicting the risk of short-term mortality, for developing an easy-to-use and reliable tool to inform clinical practice. In the present study, we established a nomogram consisting of four predictors including LVEF, cardiogenic shock, WBC count and whether underwent revascularization which can complement and update others already known, such as LVEF and WBC count before PIVSR occurring. Our prediction model was in good agreement with the actual results and performed well in discrimination after internal validation of the model by using multiple indicators during the validation process, including AUC, calibration curve, HL test, AIC and DCA, although we did not conduct an external validation due to the low incidence of PIVSR.

Conclusions

This study described the current status of PIVSR and found the WBC count, cardiogenic shock, and LVEF as the independent predictive factors of short-term mortality. Moreover, the nomogram for PIVSR could provide physicians a new way to screen high-risk patients during early clinical practice, making this special patient population have net clinical benefit eventually.