Patients suffering from acute myocardial infarction (AMI) are at high risk for recurrent major adverse cardiovascular events (MACE) despite optimized treatment and patient management (1). Many efforts to identify new parameters with greater prognostic benefits for improved risk stratification have been made over the past several years. In this context, cardiovascular magnetic resonance (CMR) imaging has evolved as a key imaging modality providing important information with incremental prognostic value (2). Particularly, CMR-feature tracking (CMR-FT)-derived left ventricular (LV) global longitudinal strain (GLS) and left atrial (LA) reservoir strain (Es) have been demonstrated to possess decisive diagnostic and prognostic capabilities, making them increasingly important for comprehensive cardiac performance analyses and optimized risk assessment (3–6). Importantly, besides their role as potent and superior prognostic parameters, various cutoff values for these strain measurements have been identified previously, enabling substantially improved risk stratification in patients suffering from AMI (5–7).

However, although the results of preceding studies are promising, a successful clinical application of these parameters and novel risk prediction models require independent validation (8). Only if previously identified imaging biomarkers and developed models can be utilized and proved to possess similar prognostic capabilities in external patient populations, a transfer of the findings to a widespread clinical practice or even guideline recommendations can be considered (9).

Whether previously determined cutoff values and associated prognostic benefits of atrial and ventricular strain values are applicable and can be reconfirmed in an independent cohort of STEMI patients is unknown. Therefore, the aim of this study is to reassess and revalidate the prognostic significance of CMR-FT-derived atrial and ventricular strain analyses in a large cohort of STEMI patients.

This CMR imaging study aimed to assess and validate the prognostic importance and predefined cutoff values of comprehensive CMR-FT-derived strain analyses in an independent large cohort of STEMI patients. Several notable findings should be considered: (1) LV GLS, GCS, and GRS as well as LA Es, Ee, and Ea were significantly reduced in patients experiencing MACE during a 12-month follow-up. (2) LA Es emerged as the strongest predictor for MACE, outperforming all LV strain values (3) Applying predefined cutoff values of LA and LV strain measurements enabled accurate risk stratification, with especially LA strain values identifying additional high-risk patients beyond commonly used parameters.

Myocardial performance analyses are essential for optimal risk stratification and patient management after AMI (21). Based on a standard CMR imaging protocol not requiring additional scanning time, CMR-FT-derived strain values represent robust imaging biomarkers. Beyond myocardial volumetric evaluations, these strain measurements allow making conclusions not only on global cardiac function but also on regional levels (22) and can additionally provide insights into diastolic capacities (23). Importantly, comprehensive myocardial strain measurements have been repeatedly proved to be sensitive parameters for myocardial dysfunction, enabling improved risk assessment outperforming most commonly used LVEF (5, 24). However, after the identification of prognostic powerful strain parameters and development of novel risk prediction models in preceding studies, external and independent validations of the findings are a prerequisite for implementing these imaging biomarkers and their cutoff values as useful tools in clinical practice and for potential further therapeutic decision-making.

In line with previous findings, our study demonstrated all strain parameters to be significantly reduced in patients with MACE compared to those without MACE. Further, both LA and LV strain values were predictors of MACE in univariable regression calculations. However, after adjusting for univariably significant parameters in different models such as cardiac risk factors, LVEF, or IS characteristics, our study could not reconfirm LV strain parameters to be independent predictors of MACE when simultaneously including LVEF in multivariable regression models. These results are in contrast to several previous works including the derivation cohort for the cut-off values, that have shown a superior role for LV strain in patients following AMI (4, 5). Of note, these studies included larger AMI cohorts with >1,000 patients in their analyses. In particular, LV GLS has been demonstrated to possess an eminent and independent prognostic role not only above commonly used parameters such as LVEF but also among all strain measurements in a spectrum of various different cardiovascular diseases (3, 5, 7, 25–28). It remains unclear, if an absent superior prognostic role of LV strain values and especially LV GLS, might be caused by a smaller study population with a relatively too low event rate of 41 MACE compared with some of the previous studies, which might influence and hamper the validation and reconfirmation of earlier findings. It is noteworthy that in contrast to other AMI study cohorts at least parts a proportion of the current study population underwent RIC and PostC or only PostC in addition to conventional PCI. Although there were no significant differences between the conditioning subgroups and controls, one cannot exclude an additional influence on strain alterations after the coapplication of RIC or PostC. As a consequence, this could reduce the prognostic power of GLS values measured shortly after the occurrence of the acute index event. At present, there is a contrary debate in the literature pertaining to whether and to what extent the coapplication of RIC influences myocardial damage and outcome after AMI. While some studies have documented positive effects of RIC on myocardial salvage and outcome (10, 11, 29), others have documented neutral effects of RIC on CMR-derived IS or outcome (30). However, especially improved GLS values have been repeatedly shown after the coapplication of RIC (31, 32) and, therefore, similar positive effects of RIC resulting in substantial GLS improvement during follow-up might have influenced the results of our study. Consequently, these effects of an improved GLS during follow-up might at least partly distort the superior prognostic power of GLS in our study compared with the findings in other study cohorts. Moreover, one might speculate whether a lower prognostic power of GLS compared with previous findings might be biased by differing baseline characteristics or varying infarct properties and localizations (33). Nevertheless, relatively similar distribution patterns of the diseased vessels as well as infarct characteristics can be observed when comparing these parameters with those of preceding studies, which most likely does not explain the failed validation as an independent predictor of MACE in our work. Remarkably, there were no significant differences of cardiovascular risk factors between patients with an adverse event and those without MACE in our study cohort, which is in contrast to other study populations, and might be another component of explanation for a less powerful prognostic relevance of LV GLS in this study. However, previous studies evaluating large STEMI registries or pooled data from various AMI trials documented equal or even worse clinical outcomes in patients without cardiovascular risk factors compared with patients presenting with at least one cardiovascular risk factor (34, 35). Whether this risk factor distribution pattern may have influenced the predictive power of GLS analyses cannot be answered on the basis of our data. Lastly, whether a missing 3-CV long axis orientation within the CMR scan protocol might influence the prognostic value of GLS remains speculative.

In contrast, LA Es emerged as the only independently associated MACE predictor in a multivariable regression model including all LV strain parameters and, moreover, increased the diagnostic accuracy of MACE prediction in AUC analyses. These findings are in line with previous study findings (16, 36) and underline the important role of atrial deformation analyses in AMI patients. Although LA performance is inevitably associated with and influenced by LV function, LA deformation has the potential to compensate ventricular failure after AMI (37). Consequently, reduced LA mechanics (especially LA Es) are sensitive markers of atrial compliance and compensatory capacity with a significant association with cardiovascular risk. In this context, intrinsic atrial failure is increasingly discussed and considered as an atrial cardiomyopathy beyond being merely a consequence of LV failure (38). It is important to note, that LA Es also represent the cumulative adverse impact of impaired LA relaxation and pulmonary venous congestion, which allows risk prediction beyond LV systolic function assessment (39). Furthermore, it has been shown, that LA strain deterioration precedes ventricular strain alterations (40) and LA strain assessment even enables a more precise detection of diastolic dysfunction than invasive pressure measurements (41). Importantly, beyond a successful validation of predetermined LA and LV strain cutoff values for risk prediction in the present study cohort, LA strain cutoff values allowed a further identification of patients at higher risk, yielding substantial additional value for prognostic characterization beyond the established clinical markers of MACE.

Analyzing strain performance to identify high-risk patients following AMI entails several important clinical considerations. Beyond a more precise detection of mechanical dysfunction compared with the commonly used LVEF, the prognostic implications could enable important improvements in patient management. It is noteworthy that a large number of patients suffering sudden cardiac death after AMI were shown to have an LVEF above 35% (42). Therefore, since current guideline recommendations for implantable cardioverter-defibrillator indications almost entirely rely on LVEF assessment, future clinical decision-making might include impaired atrial strain measurements when considering patients for primary prevention device implantation despite a relatively preserved LVEF. Likewise, patients at higher risk according to LA strain deteriorations might face a more aggressive or tailored treatment for heart failure including pharmacotherapy and/or device therapy. It is noteworthy that with the approval of SGLT-2 inhibitors in the recent past with potential positive effects on diastolic and subsequently atrial function, the prognostic attributes, especially of atrial strain values, might change in the future.

Initial efforts including 3-dimensional analyses or using neuronal networks for automated and facilitated deformation assessment have been made with promising results (43, 44), and it is interesting to speculate whether future software refinements might further improve the diagnostic and prognostic accuracy of strain parameters in patients following AMI. Of note, vendor- or technique (e.g., myocardial tagging or SENC)-dependent differences in strain values need to be considered when interpreting strain parameters. Although similar reproducibility and prognostic values of strain parameters have been demonstrated using different postprocessing software or techniques (5, 24, 45, 46), the application of different software packages or techniques might hamper direct comparability of the parameters (47). In this context, further standardization of image acquisition and analysis is required to reduce systematic intervendor differences and to harmonize comparability (46, 48).

Despite these considerations, further analyses and larger validation studies in additional STEMI cohorts (undergoing additional RIC/PostC) with different cardiovascular risk factor distribution patterns and a larger number of events are required to establish these imaging biomarkers for widespread clinical use in AMI patients.

This study aimed to validate and reconfirm the incremental prognostic value of both CMR-FT-derived LV and LA strain parameters for risk stratification in an external population of patients following AMI. In line with the literature, all strain values were significantly reduced in patients with MACE. While LV strain assessment did not enable improved risk prediction compared with the commonly used LVEF, LA Es emerged as an independent and superior imaging marker providing important prognostic information beyond traditionally used clinical parameters. Larger validation studies are needed for an unlimited and widespread application of comprehensive strain analyses in clinical routine.