Developing a Prediction Model for Persistent Systolic Anterior Motion After Mitral Valve Repair

Abstract

Background

We hypothesized that the distance between the anterior leaflet tip and the interventricular septum is a novel predictor for systolic anterior motion (SAM) after mitral valve repair.

Methods

In this case-control study, we included 139 patients who underwent mitral valve repair for degenerative mitral regurgitation between November 2014 and August 2022. We conducted multivariable logistic regression analysis to investigate the impact of the predictors associated with persistent SAM. A prediction model was developed and assessed for discrimination and calibration, and its clinical implications were evaluated by decision curve analysis.

Results

The overall incidence of persistent SAM was 5.8% (8/139). The distance between the anterior leaflet tip and the interventricular septum exhibited the most significant association with persistent SAM. The prediction model constructed with this index combined with the body surface area yielded a concordance index of 0.934. The calibration curve displayed good visual alignment with the ideal 45-degree line. Decision curve analysis revealed that the net benefit offered by the model consistently outperformed that of the “all or none” intervention strategies.

Conclusions

The distance between the anterior leaflet tip and the interventricular septum, in conjunction with the body surface area, can be used to establish a prediction model for persistent SAM.

Visual Abstract

In Short.

• ▪The distance between the anterior leaflet tip and the interventricular septum is a good predictor for persistent systolic anterior motion after mitral valve repair.
• ▪By combining this indicator with the body surface area, a reliable prediction model can be developed.

Systolic anterior motion (SAM) after mitral valve (MV) repair is one of the challenging complications postoperatively. Although numerous echocardiographic predictors of post-repair SAM have been reported,1, 2, 3 their ability to accurately predict the occurrence of post-repair SAM remains insufficient. We hypothesized that the distance between the tip of the anterior leaflet and the interventricular septum (T-sept) serves as a novel predictor of post-repair SAM.

Patients and Methods

This case-control study received approval from our institutional review board (2022-76, approved on October 25, 2022). Informed consent was waived by employing an opt-out approach for disclosing the study. We retrospectively reviewed the medical records of patients who underwent MV repair for degenerative regurgitation between November 2014 and August 2022. Patients were excluded if they underwent preemptive prophylactic measures to prevent post-repair SAM, required conversion to MV replacement, or experienced severe heart failure after operation leading to death.

The distances or angles were measured from the stored images of preoperative transesophageal echocardiography (TEE; Figure 1A), in accordance with the guidelines of the American Society of Echocardiography.⁴ In this study, we introduced the novel concept of T-sept, which was defined as the distance between the anterior leaflet tip and the interventricular septum in a midesophageal horizontal 4- or 5-chamber view during TEE (Figure 1B).

Figure 1.

Examples of conventional echocardiographic predictors. (A) C-sept indicates the distance between the coaptation point and the interventricular septum. (AL, anterior leaflet; PL, posterior leaflet.) (B) T-sept (distance between the anterior leaflet tip and the interventricular septum) measured in the midesophageal horizontal view.

A single surgeon performed the surgeries primarily using a midline sternotomy approach. MV repairs involved the implantation of artificial chordae, either alone or in combination with a small triangular resection. All cases involved the implantation of an artificial semirigid ring (Physio II or Physio Flex ring; Edwards Lifesciences) or a flexible ring (Tailor flexible ring or band; Abbott).

Post-repair SAM was characterized by a phenomenon observed on TEE whereby any portion of the MV leaflets protruded into the left ventricular outflow tract immediately after the cessation of cardiopulmonary bypass (CPB), leading to definite mitral regurgitation during systole. Transient SAM was defined as post-repair SAM that was resolved with conservative therapy, including inotropic drug cessation, intravenous fluid loading, and infusion of a beta blocker. Persistent SAM was described as post-repair SAM that continued and necessitated repeated CPB or as a condition in which mild SAM with mild regurgitation persisted at the end of surgery.

Statistical Analysis

Continuous variables are presented as medians and interquartile ranges, whereas categorical variables are expressed as numbers and percentages. The Mann-Whitney U test was used to compare continuous variables; Fisher exact test was used for categorical variables. The Bland-Altman analysis assessed the agreement between the measurements of the T-sept obtained by 2 observers who were blinded to the outcomes.

The association between independent variables and persistent SAM development was assessed by multivariable logistic regression analysis. Basic demographic information, including age, sex, and body surface area (BSA), along with the conventional echocardiographic risk factors presented in the American Society of Echocardiography guidelines were included as independent variables.

No data were missing, and the significance level was set at P < .05. Statistical analyses were conducted primarily with EZR software (Saitama Medical Center, Jichi Medical University), a graphical user interface for R version 4.2.1 (R Foundation for Statistical Computing) with additional packages integrated into R Studio.

Prediction Model Development and Evaluation

A logistic regression model was constructed to predict persistent SAM after MV repair using T-sept combined with BSA. The model’s performance was assessed by the concordance index (C-index) and calibration. Internal validation was conducted through 1000 rounds of bootstrap resampling. Decision curve analysis was employed to assess the clinical implications.

Results

Of 146 consecutive patients, 5 who underwent preemptive prophylactic measures to prevent post-repair SAM, 1 who died of profound heart failure that developed after surgery, and 1 whose procedure was converted to MV replacement during the initial CPB were excluded. Of the 14 patients who had post-repair SAM, conservative therapy alleviated SAM in 6 patients. The overall incidence of persistent SAM was 5.8% (8/139). These 8 patients required resumption of CPB (repeated CPB, n = 7) or still had residual mild SAM on CPB weaning (n = 1). All 7 patients who needed repeated CPB underwent additive modifications. One of them required conversion to MV replacement during the third CPB procedure (Supplemental Figure 1).

The baseline characteristics and operative procedures are described in Table 1. Preoperative echocardiographic measurements are described in the Supplemental Table. The mean difference between T-septs measured by the 2 observers was −0.62 mm (95% CI, −1.27 to 0.04 mm). The limits of agreement ranged from −8.25 to 7.02 mm (Supplemental Figure 2).

Table 1.

Baseline Characteristics

Variables	All	No or Transient SAM	Persistent SAM	P
No. of patients	139	131 (94.2)	8 (5.8)
Demographic parameters
Age, y	65 (56.5-73)	65 (56.0-73.0)	65 (62.3-66.5)	.82
Female sex	51 (36.7)	46 (35.1)	5 (62.5)	.14
BSA, m2	1.64 (1.5-1.79)	1.64 (1.5-1.79)	1.57 (1.52-1.67)	.23
Coexisting disease
Hypertension	78 (56.1)	75 (57.3)	3 (37.5)	.30
Diabetes mellitus	12 (8.6)	11 (8.4)	1 (12.5)	.52
Ischemic heart disease	14 (10.1)	14 (10.7)	0	1
COPD	5 (3.6)	5 (3.8)	0	1
Atrial fibrillation	51 (36.7)	50 (38.2)	1 (12.5)	.26
Stroke history	4 (2.9)	4 (3.1)	0	1
Chronic kidney disease	4 (2.9)	4 (3.1)	0	1
Isolated posterior leaflet	81 (58.3)	73 (55.7)	8 (100)	.021
Repair procedure
Artificial chordae only	121 (87.1)	116 (88.5)	5 (62.5)	.07
Artificial ring size, mm	29 (29-31)	29 (28-31)	31 (30.8-33)	.048
Flexible ring/band	102 (73.4)	94 (71.8)	8 (100)	.11
Partial band	96 (69.1)	91 (69.5)	5 (62.5)	.70

Categorical variables are presented as number (percentage). Continuous variables are presented as median (interquartile range).

BSA, body surface area; COPD, chronic obstructive pulmonary disease; SAM, systolic anterior motion.

In the multivariable analysis, only T-sept (odds ratio, 0.48; 95% CI, 0.27-0.85) and BSA (multiplied by 100; odds ratio, 1.14; 95% CI, 1.01-1.28) retained statistical significance (Table 2). The logistic function as the prediction model of persistent SAM development with these 2 predictors is as follows:

$$\text{probability}\text{of}\text{persistent}\text{SAM}=\frac{1}{1+e^{(0.502+0.569\times \mathrm{T}-\text{sept}-5.918\times \text{BSA})}}$$

where e equals 2.718 as the Euler number. The variance inflation factor for these predictors was 1.60.

Table 2.

Logistic Regression Analysis

Predictor	Univariable Result		Multivariable Result
	Odds Ratio (95% CI)	P	Odds Ratio (95% CI)	P
Age	1.01 (0.944-1.07)	.87	0.897 (0.765-1.05)	.18
Sex	0.325 (0.0742-1.42)	.14	0.057 (0.0005-6.61)	.24
BSA (×100)	1.00 (0.964-1.03)	.95	1.14 (1.01-1.28)	.034
LVDS	0.822 (0.698-0.967)	.018	0.717 (0.456-1.13)	.15
A-M angle	1.0 (0.932-1.07)	.994	1.17 (0.926-1.47)	.19
AL/PL ratio	0.161 (0.0311-0.838)	.030	0.094 (0.002-4.58)	.23
C-sept	0.776 (0.658-0.916)	.003	1.20 (0.821-1.75)	.35
T-sept	0.674 (0.552-0.822)	<.001	0.476 (0.267-0.847)	.012

AL/PL, anterior leaflet height/posterior leaflet height; A-M, aorto-mitral; BSA, body surface area; C-sept, distance between the coaptation point and the interventricular septum; LVDS, left ventricular systolic dimension; T-sept, distance between the anterior leaflet tip and the interventricular septum.

The C-index, equal to the area under the receiver operating characteristic curve (Figure 2A), of the model was 0.934 (95% CI, 0.821-1). The calibration curves depicted a close alignment with the ideal 45-degree line (Figure 2B). The bias-corrected calibration intercept and slope were 0.0755 and 0.917, respectively. The bias-corrected C-index was 0.936.

Figure 2.

(A) Receiver operating characteristic curve of the prediction model. (B) The apparent and bias-corrected calibration curves. (C, D) Decision curve analysis of the prediction model. The net benefit when patients are opted in (C) or out (D) using the model is presented.

Decision curve analysis shows that the net benefit yielded by the model is consistently higher than that of both all-intervention and no-intervention strategies. Figure 2C describes the decision curve analysis when patients over the threshold probability are selected for intervention. On the contrary, Figure 2D describes the decision curve analysis when patients below the threshold probability are selected for no intervention. In addition, in comparison to Denti’s model,³ our model consistently yielded a higher net benefit.

For bedside use, logistic curves for 7 distinct BSA levels are described in Supplemental Figure 3.

Comment

The study has 3 main findings. T-sept and BSA emerged as significant predictors of persistent SAM. The model developed for predicting persistent SAM exhibited favorable discrimination and calibration. This new prediction model holds substantial clinical implications.

Of all the variables evaluated through logistic regression analysis, T-sept showed the most significant association with persistent SAM development. One plausible reason for T-sept standing out among all echocardiographic variables is its inherent simplicity. BSA was also identified as a significant risk factor for persistent SAM. In our opinion, BSA should be considered an important predictor of persistent SAM development. Given these conditions, we concluded that both T-sept and BSA were appropriate independent variables for predicting persistent SAM. Although the potential for overfitting of the model was a concern, internal validation indicated that the model’s optimism was adequately controlled. In addition, the discriminative ability of the model expressed by the C-index was sufficiently high.

The results of the decision curve analysis suggest the practical usefulness of the model in clinical settings. To the best of our knowledge, a prediction model for SAM has not been proposed, apart from Denti’s model³ constructed with 6 variables. The decision curve analysis indicates the superiority of our model in clinical settings.

Logistic curves obtained by discretely setting different BSA values (Supplemental Figure 3) allow us to conclude that the predicted probability consistently exceeds 0.5 when T-sept is below 13 mm, regardless of BSA. Conversely, when T-sept measurement is above 24 mm, the probability falls below 0.1. We speculate that this BSA-based discrepancy may arise from a difference in cardiac output, which was not directly measured.

Since the report by Maslow and coworkers¹ in 1999, echocardiographic predictors of post-repair SAM have been recognized. Nevertheless, these indices have not always led surgeons to modify MV repair procedures. Even when conventional echocardiographic predictors suggest a high risk of post-repair SAM, it remains uncertain whether to perform more complex procedures.5, 6, 7 Our prediction model will offer valuable guidance to all surgeons planning mitral repair for degenerative regurgitation.

Limitations

This study has some limitations. First, this was a retrospective study, which may introduce inherent biases. Second, the relatively small number of patients who had persistent SAM may have led to an overly optimistic estimation of the reliability and applicability of this prediction model. Finally, the results were obtained from surgical procedures performed by a single surgeon at a single institution, potentially limiting the generalizability of the findings to other settings with different surgical practices. To address this limitation, further investigations including external validation using new data sets in different settings are needed.

Conclusion

T-sept measurement is the most significant predictor of post-repair persistent SAM. By combining T-sept measurements with BSA, a reliable prediction model can be developed.

Supplementary Data

Supplementary Figure 1.

Supplementary Figure 2.

Supplementary Figure 3.