We screened 303 consecutive STEMI patients with LVA who were admitted to our department between March 2014 and June 2022. Of these, transthoracic echocardiography (TTE) identified MT in 133 patients, leading to their classification as MTLVA cases. Two independent experts confirmed the diagnosis of MT through imaging review. These 133 patients were included in this retrospective clinical cohort study for further analysis. All patients received standard antithrombotic therapy in accordance with the relevant guidelines issued by the European Society of Cardiology (post-LVA) [8, 13]. Smoking status was defined as former or current smoking based on self-reported history. Alcohol intake was defined as an average daily intake exceeding 40 g for men or 20 g for women, sustained for at least five years. Patients were followed for a median duration of 25 months, during which clinical visit and echocardiographic examinations were conducted at each follow-up visit.

Detection of MT and LVA was accomplished using TTE (Philips EPIQ-7C, Philips Healthcare, Amsterdam, The Netherlands). MT appeared as an echodense mass adjacent to the left ventricular (LV) wall, with confirmation of LVA achieved through akinetic or dyskinetic LV wall motion, which bulged outward during systole. To differentiate MT from the underlying myocardium, a distinct thrombus-blood interface was necessary, and MT had to be observable on at least two views throughout the cardiac cycle. Quantitative evaluation of LVA and MT characteristics was performed offline, involving measurements of the maximal long and short diameters of LVA (LVA_LD and LVA_SD), the area of LVA (LVA_area), the maximal long and short diameters of MT (MT_LD and MT_SD), and the area of MT (MT_area). All measurements were based on the mean of three assessments.

Patients were divided into two groups based on the resolution status of MTLVA. The MTLVA resolution group was characterized by the complete disappearance of MTLVA following antithrombotic treatment, as confirmed by echocardiographic examinations during follow-up. Conversely, the MTLVA non-resolution group comprised patients in whom MTLVA persisted or recurred despite antithrombotic treatment, as observed through echocardiographic assessment during follow-up.

The primary endpoint was defined as the occurrence of adverse events during follow-up, encompassing stroke, re-revascularization, major bleeding, systemic embolism, or cardiac death.

Continuous variables are expressed as mean $\pm$ standard deviation or median (interquartile range, IQR), and categorical variables as number (percent). Group differences were compared using the independent samples t-test or Mann-Whitney U test, as appropriate. For categorical data, comparisons across groups were made using the Chi-square test or Fisher’s exact test. Dynamic changes in left ventricular ejection fraction (LVEF) over time were analyzed using one-way repeated measures analysis of variance (ANOVA) with the General Linear Model. Diagnostic performance of LVA and MT for predicting MTLVA non-resolution was assessed using receiver operating characteristic (ROC) curves, with a comparison among the areas under curves (AUC). Optimal cut-offs for observed variables such as laboratory parameters and echocardiographic parameters in this analysis were determined using the Youden Index method derived from ROC estimation. To identify independent risk factors associated with clinical outcomes, univariable and multivariable Cox proportional hazards regression analysis were performed. Variables with p-value 0.05 in the univariable Cox analysis were included in the multivariable models, which were constructed using “backward stepwise” selection method. Collinearity was assessed using the correlation coefficient, with a threshold of $>$0.70 indicating the presence of significant collinearity between variables. To prevent multicollinearity, which could distort the estimation of regression coefficients and affect model stability, variables exhibiting collinearity were excluded from the multivariable models. Only independent variables with no significant collinearity were retained for inclusion. Results were reported as hazard ratios (HR) with corresponding 95% confidence intervals (CI). A p value 0.05 (two-tailed test) was considered statistically significant. The statistical analysis was performed using the SPSS statistical software, version 28.0 (IBM SPSS Statistics, Chicago, IL, USA).

All patients underwent serial echocardiography follow-up for MTLVA assessment, with a median of 4 (3–5) evaluations. Over a median follow-up of 25 (14–39) months, MTLVA resolution was observed in 44.4% (59 out of 133) of patients with STEMI, while 55.6% (74 out of 133) exhibited non-resolution including two cases of MTLVA recurrence. Specifically, MTLVA disappeared in 42 patients at the second echocardiography [median period of 46 (21–98) days], in 52 patients at the third echocardiography [median period of 6.5 (2.8–18.2) months], and in 59 patients at the fourth echocardiography [median period of 12.0 (8.5–23.5) months].

Table 1 details the baseline characteristics of STEMI patients with MTLVA resolution and non-resolution. The mean age was 62 $\pm$ 11 years, 107 (80.5%) patients were male. In comparison to patients with MTLVA resolution, those with MTLVA non-resolution showed a higher proportion of excessive alcohol drinkers (25.7% vs. 11.9%, p = 0.046) and elevated D-Dimer levels (median 720 vs. 435 ng/mL, p = 0.002).

All patients received low molecular weight heparin at admission. Post-discharge, 57.1% (n = 76) were on vitamin K antagonists (VKAs), 33.1% (n = 44) on rivaroxaban, and 7.5% (n = 10) on dabigatran. Anticoagulation was not applied to three patients due to high bleeding risk. Concomitant antiplatelet therapy was common (87.2%, n = 116), with 26.2% (n = 35) on anticoagulation + single antiplatelet and 60.9% (n = 81) on anticoagulation + dual antiplatelet therapy (Table 1). 12.8% patients (n = 17) received anticoagulation alone due to high bleeding risk. Anticoagulant and antiplatelet medication usage showed no significant difference between MTLVA resolution and non-resolution groups (Table 1).

Among the patients, 73.7% (n = 98) underwent PCI, 19.5% (n = 26) underwent coronary artery bypass grafting (CABG) or a combination of PCI and CABG, and 6.8% (n = 9) received conservative treatment. The rates of coronary revascularization did not differ significantly between the resolution and non-resolution groups (Table 1). However, STEMI involving the left circumflex artery (LCX) was more common in the non-resolution group compared to the resolution group (45.9% vs. 27.1%, p = 0.026).

LVEF was significantly lower in the MTLVA non-resolution group compared to the MTLVA resolution group (mean 41.7% vs. 45.7%, p = 0.008). Additionally, the size of the LVA and MT was significantly larger in the MTLVA non-resolution group than in the MTLVA resolution group (all p 0.001, Table 1).

Larger LVA_area and MT_area demonstrated excellent diagnostic performance for predicting MTLVA non-resolution, with an AUC of 0.875 (95% CI 0.817–0.933, p 0.001; Fig. 1A) and an AUC of 0.825 (95% CI 0.755–0.895, p 0.001; Fig. 1B), respectively. Correlation analysis revealed a positive association between MT_area and LVA_area with MTLVA non-resolution (Spearman’s rho, r = 0.736, p 0.001; R² = 0.602, p 0.001; Fig. 2A). LVEF exhibited a negative correlation with MT_area (Spearman’s rho, r = –0.254, p = 0.003; R² = 0.053, p = 0.008; Fig. 2B).

Fig. 1.

Diagnostic performance of LVA and MT in predicting MTLVA non-resolution. (A) ROC curve showing the diagnostic performance of LVA_area for predicting MTLVA non-resolution, with an AUC of 0.875. (B) ROC curve demonstrating the diagnostic ability of MT_area, with an AUC of 0.825, both indicating strong predictive value. CI, confidence interval; MT, mural thrombus; ROC, receiver operating characteristic; AUC, areas under curves.

Fig. 2.

Correlation analysis of LVA_area, MT_area, and LVEF in relation to MTLVA non-resolution. (A) Scatter plot showing a strong positive correlation between MT_area and LVA_area, suggesting that larger thrombus and aneurysm areas are linked to MTLVA non-resolution. (B) Scatter plot showing a negative correlation between LVEF and MT_area, indicating that lower ejection fraction is associated with larger mural thrombus areas.

Univariable Cox regression analysis identified several potential risk factors for non-resolution of MTLVA, including hypercholesterolemia, LCX-related STEMI, D-Dimer $>$780 ng/mL, lipoprotein (a) [Lp(a)] $>$270 mg/L, baseline LVEF 40%, LVA_area $>$4.5 cm², and MT_area $>$2.2 cm² (Table 2).

Given the significant collinearity between LVA_area and MT_area (Spearman’s rho = 0.736, p 0.001; Fig. 2A), two separate multivariable models were constructed to address this issue and ensure the stability of the analysis. Model A included LVA_area, while Model B incorporated MT_area.

In Model A, which adjusted for age, sex, hypercholesterolemia, LCX-related STEMI, D-Dimer $>$780 ng/mL, Lp(a) $>$270 mg/L, baseline LVEF 40%, and LVA_area $>$4.5 cm², hypercholesterolemia (HR 2.135, p = 0.006), LCX-related STEMI (HR 2.228, p = 0.005), Lp(a) $>$270 mg/L (HR 2.270, p = 0.003), and LVA_area $>$4.5 cm² (HR 4.038, p 0.001) were independently associated with non-resolution of MTLVA. Similarly, in Model B, adjusting for the same covariates, hypercholesterolemia (HR 1.715, p = 0.043), LCX-related STEMI (HR 1.875, p = 0.021), Lp(a) $>$270 mg/L (HR 2.264, p = 0.003), and MT_area $>$2.2 cm² (HR 2.398, p = 0.002) remained significant predictors of non-resolution.

As depicted in Fig. 3, throughout the follow-up periods, the mean LVEF was consistently lower in the non-resolution group, while in the resolution group, LVEF demonstrated a trend of increase over time (p = 0.006, 0.001, and 0.001 at the 1st, 2nd, and 3rd echocardiography examination, respectively).

Fig. 3.

Dynamic changes in LVEF associated with non-resolution of MTLVA. The plot shows that, throughout the follow-up period, the mean LVEF remained significantly lower in the non-resolution group compared to the resolution group. In contrast, the resolution group exhibited a consistent increase in LVEF over time. The differences between the two groups were statistically significant at the 1st, 2nd, and 3rd echocardiographic evaluations (p = 0.006, 0.001, and 0.001, respectively).

Non-resolution of MTLVA was associated with significantly worse outcomes compared to resolution (Table 1 and Fig. 4). Cardiovascular mortality (Fig. 4A) was higher in the non-resolution group (8.1% vs. 1.7%), though not statistically significant (p = 0.066). Major bleeding events (Fig. 4B) showed a similar trend, occurring more frequently in the non-resolution group (8.1% vs. 1.7%, p = 0.059). Stroke (Fig. 4C) was significantly more common in the non-resolution group (14.9% vs. 3.4%, p = 0.025). Systemic embolism (Fig. 4D) occurred in 5.4% of the non-resolution group but was absent in the resolution group (p = 0.034). The composite of adverse events (Fig. 4E), including stroke, re-revascularization, major bleeding, systemic embolism, and cardiac death, was significantly higher in the non-resolution group than in the resolution group (28.4% vs. 8.5%, p = 0.003).

Fig. 4.

Clinical outcomes associated with non-resolution of MTLVA. Non-resolution of MTLVA was associated with worse clinical outcomes. Cardiovascular mortality (A) and major bleeding events (B) showed higher rates in the non-resolution group (p = 0.066 and p = 0.059, respectively). Stroke (C) was significantly more frequent in the non-resolution group (14.9% vs. 3.4%, p = 0.025). Systemic embolism (D) occurred in 5.4% of the non-resolution group but was absent in the resolution group (p = 0.034). The composite adverse event rate (E), including cardiovascular death, major bleeding, stroke, and systemic embolism, was significantly higher in the non-resolution group (28.4% vs. 8.5%, p = 0.003). LVT, left ventricular thrombus.

Present study identified 6 potential prognostic markers associated with the occurrence of adverse events in this cohort, including N-terminal pro-brain natriuretic peptide (NT-proBNP), serum creatinine, total cholesterol (TC), LVEF, LVA_area, and MTLVA non-resolution (Table 3). Multivariable Cox regression analysis revealed that NT-proBNP, TC, and MTLVA non-resolution remained independently associated with adverse events, while the correlation with adverse events for serum creatinine level, LVEF, and LVA_area was no longer significant after multivariable adjustment (Table 3).

Our study highlights the ongoing challenge of MTLVA non-resolution despite contemporary antithrombotic therapies. In our cohort, resolution occurred in only 45.4% of STEMI patients with MTLVA over a median follow-up of 25 months. Similarly, Lattuca et al. [5] reported a resolution rate of 62.3% within 103 days, while Zhou et al. [4] observed non-resolution in 32.6% of patients during a median follow-up of 1.2 years, with 24.3% experiencing recurrence.

Importantly, our study identified independent predictors of MTLVA non-resolution, including elevated Lp(a) levels ($>$270 mg/L), larger LVA area ($>$4.5 cm²), and greater MT area ($>$2.2 cm²). These findings underline the clinical importance of risk stratification. Furthermore, patients with non-resolution consistently exhibited lower LVEF than those with resolution throughout follow-up, while the latter group showed progressive LVEF improvement, suggesting a relationship between thrombus resolution and myocardial recovery.

We observed that a higher proportion of patients in the resolution group were on VKAs compared to the non-resolution group (66.1% vs. 50.0%, p = 0.062). In contrast, the non-resolution group had a greater proportion of patients on non-vitamin K oral anticoagulants (NOACs, 50.0% vs. 28.8%, p = 0.013). This suggests that VKAs may be more effective in facilitating thrombus resolution compared to NOACs in this patient cohort with MT in the LVA. However, previous studies, although based on small sample sizes, have generally reported similar LV thrombus resolution rates between VKAs and NOACs [14]. Therefore, the comparative efficacy of VKAs versus NOACs in thrombus resolution should be explored in larger, prospective studies to better understand their roles in this clinical context.

Our analysis revealed that a baseline LVEF 40% was a potential risk factor associated with non-resolution of MTLVA in this cohort. Previous studies have indicated that patients with LVEF 40% are at an increased risk of developing LVT due to blood stasis in the ventricular cavity and endocardial injury with associated inflammation [15]. Allard et al. [16] demonstrated that lower LVEF was predictive of LVT. Our study cohort further showed that LVEF was significantly lower in the MTLVA non-resolution group compared to the MTLVA resolution group (mean 41.7% vs. 45.7%, p = 0.008). Additionally, patients with persistently lower LVEF over time also exhibited an increased risk of MTLVA non-resolution.

Lp(a) is a genetically determined lipoprotein with pro-thrombotic and anti-fibrinolytic properties [17], recognized as a causal risk factor for atherosclerotic cardiovascular disease and associated with vascular inflammation and thrombus burden [18, 19, 20, 21, 22, 23]. While Celik et al. [24] reported no association between Lp(a) and LVT risk in acute myocardial infarction, our findings suggest that Lp(a) $>$270 mg/L is independently associated with an increased risk of MTLVA non-resolution. This association may reflect the pro-thrombotic role of Lp(a) in MTLVA pathogenesis, but further studies are needed to clarify this relationship.

D-dimer, a marker of coagulation and fibrinolysis activation in response to the body’s hypercoagulable state [25, 26], has been associated with thrombus formation in conditions like dilated cardiomyopathy and post-myocardial infarction (MI) LV dysfunction [27, 28]. Although we observed a potential link between elevated D-dimer levels ($>$780 ng/mL) and MTLVA non-resolution, this association lost significance after adjustment. Future studies are required to determine whether D-dimer could reliably predict MTLVA non-resolution.

Patients with an LVA_area $>$4.5 cm² had a nearly 4-fold higher risk of MTLVA non-resolution compared to those with an LVA_area $\le$4.5 cm². Similarly, patients with an MT_area $>$2.2 cm² faced approximately a 2-fold increased risk of MTLVA non-resolution compared to those with an MT_area $\le$2.2 cm.

The association between larger LVA_area and MT_area with thrombus non-resolution remains unclear. Several factors may contribute to this relationship: (1) In larger LVA areas, blood flow could be slower, potentially resulting in a more stable thrombus that is less susceptible to resolution; (2) Larger thrombus areas might be more resistant to effective resolution by antithrombotic therapy, possibly due to insufficient drug penetration or thrombus composition. Given these considerations, it may be reasonable to extend the duration of antithrombotic therapy in patients with larger LVA_area and MT_area to improve the chances of thrombus resolution.

Severe myocardial injury, especially in large infarctions, creates a pro-thrombotic environment by promoting persistent endothelial dysfunction, local inflammation, and microvascular damage [29]. These pathological processes facilitate MT formation and hinder its resolution. In our study, patients with larger LVA areas (LVA_area $>$4.5 cm²) and MT areas (MT_area $>$2.2 cm²) were at significantly higher risk of thrombus non-resolution. These findings underscore the role of extensive myocardial damage as a sustained substrate for thrombus persistence.

Reperfusion injury adds another layer of complexity. The restoration of blood flow to ischemic myocardium, while critical for salvaging viable tissue, can intensify damage through microvascular obstruction, oxidative stress, and inflammation. This cascade is often accompanied by intramyocardial hemorrhage (IMH), a result of microvascular rupture during reperfusion [30, 31]. IMH introduces blood products into the myocardium, forming a nidus for thrombus development and further complicating its resolution. The recent Canadian Society of Cardiology definition of tissue injury severity emphasizes the impact of extensive myocardial injury, including IMH, on adverse clinical outcomes [31], and highlights the interplay between severe myocardial injury and reperfusion injury in driving thrombus persistence. Together, these mechanisms suggest that even under standard antithrombotic therapy, severe tissue damage and associated complications like IMH play a pivotal role in the non-resolution of MT. Future studies utilizing advanced imaging modalities, such as cardiac magnetic resonance imaging, could provide a deeper understanding of these mechanisms. Such insights may help identify high-risk patients and inform tailored therapeutic strategies to improve thrombus resolution and clinical outcomes.

Our results confirm that MTLVA non-resolution is associated with worse clinical outcomes. Non-resolution patients in our study had a higher incidence of adverse events compared to resolution patients. Consistent with Zhou et al. [4], who reported higher rates of MACE, embolic events, and stroke among patients with LVT recurrence, these findings underscore the importance of achieving complete thrombus resolution to reduce adverse cardiovascular and cerebrovascular outcomes.

Defining independent risk factors of non-resolution of MTLVA might add the risk stratification of patients with MTLVA. Patients with higher risk of non-resolution MTLVA might benefit from more guideline-adherent antithrombotic treatment.

This study has several limitations. First, its retrospective design and single-center setting introduce the possibility of selection bias, which may limit the generalizability of our findings. Larger and prospective studies are needed to confirm these results and provide more robust evidence. Second, the diagnosis of MTLVA was primarily based on transthoracic echocardiography, which, while a standard and widely accessible imaging modality, may lack the sensitivity and specificity of advanced techniques such as contrast-enhanced ultrasound or cardiac magnetic resonance imaging. Future studies incorporating these modalities could enhance diagnostic accuracy and provide additional insights. Third, we did not capture reperfusion metrics such as symptom-to-balloon or door-to-balloon times, which are critical determinants of STEMI outcomes and may have influenced our findings. Finally, the relatively small sample size, although representative of this patient population, may have reduced the statistical power to detect certain associations. Thus, our findings should be interpreted as exploratory and hypothesis-generating, warranting validation in larger, multicenter cohorts.