Long‐term prognostic impact of subclinical myocardial dysfunction in patients recovered from COVID‐19

Cardiovascular sequelae may occur in patients recovered from coronavirus disease 2019 (COVID‐19). Recent studies have detected a considerable incidence of subclinical myocardial dysfunction—assessed with speckle‐tracking echocardiography—and of long‐COVID symptoms in these patients. This study aimed to define the long‐term prognostic role of subclinical myocardial dysfunction and long‐COVID condition in patients recovered from COVID‐19 pneumonia.

and is associated with a higher risk of MACE at long-term follow-up. Speckle-tracking echocardiography is a promising tool to optimize the risk-stratification in patients recovered from COVID-19 pneumonia, while the definition of a long-COVID condition has no prognostic relevance.

K E Y W O R D S
advanced echocardiography, COVID-19, global longitudinal strain, long-COVID, subclinical myocardial dysfunction

INTRODUCTION
Severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2), the virus that causes coronavirus disease 2019 (COVID- 19), may affect the cardiovascular (CV) system through direct and indirect mechanisms. 1 Myocardial injury (MI) related to COVID-19 has been associated with a dismal in-hospital, 2-5 mid-and long-term prognosis. [6][7][8][9] Moreover, persisting CV abnormalities and long-COVID symptoms are determining a burdensome impact on the quality of life in many survivors of SARS-CoV-2 infection. 10,11 Several CV imaging abnormalities have been described in  patients, both during acute disease and after infection recovery, [12][13][14] but the prognostic implications of these findings are unknown yet. Similarly, there is no certainty regarding the pathophysiologic basis and the prognostic impact of long-COVID, a condition defined as the persistence of symptoms resulting in a perception of a worse health status as compared with the pre-COVID-19 one. 15  patients has been widely demonstrated. [16][17][18] Moreover, an impairment of LV-GLS has been associated with a poor in-hospital prognosis in patients hospitalized due to COVID-19. [19][20][21] Despite an increased risk for CV events in patients who recovered from COVID-19 pneumonia, there is no standardized risk-stratification strategy for the follow-up of this wide population. The aim of the present analysis is to evaluate the long-term prognostic implications of subclinical myocardial dysfunction assessed with LV-GLS and of long-COVID condition in patients hospitalized due to COVID-19 pneumonia during the first wave in April 2020. Besides, we will compare the predictive power of LV-GLS and other follow-up parameters for adverse CV prognosis, to define an effective follow-up strategy for COVID-19 survivors.

Study design and population
This is an observational prospective study conducted at an academic

Data collection
After COVID-19 hospitalization, patients underwent an echocardiographic and clinical follow-up at 7 months and a clinical follow-up at 21 months ( Figure 1). Study data were collected and managed using RED-

Baseline characteristics
Baseline clinical and laboratory data were collected during COVID-19 hospitalization and are summarized in Table 1 and Table S1. Baseline clinical features were comparable among the two groups, except for dyslipidemia and a history of coronary artery bypass grafting, which were significantly more prevalent in patients with subclinical myocardial dysfunction. The group with subclinical myocardial dys-function showed significantly higher values of baseline and peak BNP during hospitalization. Baseline and in-hospital medical therapy are summarized in Table S2.

Seven-month follow-up
Clinical, EKG, and echocardiographic 7-month follow-up were complete for all patients, while BNP and troponin values at follow-up were available, respectively, only for 85 (77%) and 62 patients (56%).
Six-month follow-up data are shown in Table 2

Twenty-one-month follow-up
The median follow-up was 650 days (622-671). Twenty-one-month clinical follow-up with NYHA class definition, EQ-5D-5L questionnaire assessment and clinical events is summarized in  A few multivariable regression analysis models including 7-month follow-up data were performed to identify independent predictors of extended MACE at 21-month follow-up (   (Table S3).
The supplementary univariate regression analysis exploring the predictive value for primary endpoint of cardiac biomarkers alterations, both during COVID-19 hospitalization and at 7-month follow-up, did not detect any statistically significant association (Table S4).

DISCUSSION
The main findings of this prospective study may be summarized as follows: 1  Among patients recovered from COVID-19, high rates of subclinical myocardial dysfunction at short-term follow-up, mostly evaluated with 2D-STE, have been described. 16,18,33 However, the prognostic role of this dysfunction remains unknown. To the best of our knowledge, our study is the first to evaluate the long-term prognostic impact of LV-GLS evaluated with 2D-STE in patients recovered from COVID-19.
LVEF is the main parameter used to define a systolic myocardial dysfunction, but shows several limitations, like inter-and intra-observer variability and low sensitivity for subclinical myocardial dysfunction detection. 34,35 Our study confirms that LV-GLS is more sensitive in detecting subclinical myocardial dysfunction than LVEF, as we identified 27% of patients with normal LVEF having LV-GLS ≥−18%.
Of note, PCI for unstable angina was the most incident major CV event in our cohort, possibly due to the acute and long-term destabilizing effect of COVID-19 on atherosclerotic CV disease with prothrombotic, inflammatory, and autoimmune mechanisms. 36 Another potential reason for the high incidence of PCI for unstable angina in our cohort may be due to chest pain being a very common symptom among patients with post-acute sequelae of SARS-CoV2 infection. Thus, the onset of typical chest pain in patients with a significant CV risk profile may have triggered a diagnostic pathway that leads to the incidental discovery of a stable coronary artery disease, finally deemed worthy of treatment with PCI.
As far as a systematic approach for follow-up of patients recovered from COVID-19 is lacking, a better identification of patients who will benefit from a closer follow-up is warranted. In this context, as suggested by our results, symptomatic burden may not be a reliable parameter since neither NYHA class nor EQ-5D-5L questionnaire nor long-COVID definition were predictive of MACE. Our data shed the light from a heterogeneously defined condition with a constellation of different symptoms, namely long-COVID, on subclinical myocardial dysfunction as real driver of long-term CV prognosis. According to our results, the combined use of two simple and feasible parameters such as NYHA class and LV-GLS would allow the identification of post-COVID-19 patients at higher risk for CV events with high accuracy.
Our study should be interpreted in light of some limitations. First, its single-center and non-randomized nature and its small sample size determine intrinsic limitations in terms of generalizability and selection bias. However, we included an unselected cohort of consecutive patients. Second, even if this study was prospectively conducted, we cannot exclude the effects of unmeasured confounding factors. Future multicenter investigations are therefore needed to confirm our study findings. Notwithstanding, our analyses focused on 7-month follow-up measurable data in order to provide pragmatical tools with an easy clinical application. Third, the absence of both baseline pre-COVID-19 and in-hospital echocardiographic data limits our possibilities to hypothesize a causal relationship between COVID-19 and subclinical myocardial dysfunction. Then, the low incidence of major CV events led us to include only a limited number of covariates in the multivariable models, in order to reduce the risk of overfitting. Moreover, the incomplete availability of BNP and troponin at follow-up limits our analysis regarding the prognostic value of cardiac biomarkers. Furthermore, our follow-up strategy did not include other imaging modalities than echocardiography, such as cardiac magnetic resonance. Finally, the normal reference values for LV-GLS might be affected by vendordependency. However, in our study LV-GLS analysis was performed with a vendor-independent software (TomTec).

CONCLUSIONS
In an unselected cohort of patients recovered from COVID-19 pneumonia, LV-GLS assessed with 2D-STE at 7-month follow-up identified a subclinical myocardial dysfunction in one out of three patients and shows the best discriminatory power in predicting CV events at longterm follow-up. In light of the significant association between LV-GLS and adverse CV events and the feasibility of 2D-STE analysis, we suggest applying this tool to improve the risk-stratification for CV events in patients recovered from COVID-19. Conversely, a long-COVID con-dition at 7-month follow-up was present in almost one half of the cohort but did not show any long-term prognostic relevance.