We analyzed all patients admitted to Nihon University Itabashi Hospital diagnosed with COVID-19; this comprised a total of 584 patients hospitalized at our hospital from January 2020 until September 2021. During their hospital stay, we informed all 584 COVID-19 patients to visit our cardiology outpatient clinic if they had any CV long COVID symptoms (dyspnea, chest pain, and/or palpitations) after discharge. As a result, 9% (52/584) visited the cardiology outpatient clinic because of CV long COVID symptoms after discharge and were enrolled in the study (Figure 1). Patients with persistent organizing pneumonia were excluded from this study. This single center prospective observational study was approved by the Nihon University Itabashi Hospital institutional review board (IRB) (RK-210209-1) and informed consent was obtained from patients as appropriate.

At the cardiology outpatient clinic, screening tests, such as an electrocardiogram (ECG), chest X-ray, and echocardiography, were performed. The ECGs were evaluated by an experienced cardiologist according to the Sokolow-Lyon voltage criterion (10, 11). If left ventricular hypertrophy, abnormal left ventricular repolarization, non-specific ST-T changes, and left atrial enlargement were detected, the patients were defined as being suspected of having left atrial and/or ventricular involvement (LAVI). However, if right ventricular hypertrophy and right atrial enlargement were detected, the patients were defined as being suspected of having right atrial and/or ventricular involvement (RAVI). The chest X-rays were evaluated by an experienced cardiologist. If left 3rd arch and left 4th arch enlargement were detected, the patients were defined as being suspected of having LAVI. However, if right 2nd arch and left 2nd arch enlargement were detected, the patients were defined as being suspected of having right RAVI. Echocardiography was performed by an experienced sonographer and the analysis of the results was performed by two expert cardiologists (YS and TY) according to the American Society of Echocardiography guidelines (12, 13). If a low ejection fraction [<55%], wall motion abnormalities, and elevated E/e’ [>14] were detected, the patients were defined as being suspected of having LAVI. However, if a high tricuspid regurgitant pressure gradient [TRPG > 30 mmHg] and low tricuspid annular plane systolic excursion [TAPSE < 17 mm] were detected, the patients were defined as being suspected of having right RAVI. Patients suspected of having LAVI underwent cardiac magnetic resonance (CMR) imaging, those suspected of having RAVI underwent lung perfusion single-photon emission computed tomography/computed tomography (SPECT/CT), and those suspected of having LAVI/RAVI underwent both CMR and SPECT/CT (Figure 1). CMR was not performed in cases with contraindications (severe renal dysfunction with an estimated glomerular filtration rate [eGFR] of less than 30 ml/min/1.73 m², pacemaker implantation, or allergy to gadolinium contrast media). CMR and SPECT/CT were chosen for the detailed examinations of this protocol because myocardial injury (14, 15) and pulmonary embolisms (PEs) (16) are common cardiovascular complications of COVID-19, according to various reports since the COVID-19 outbreak.

The CMR data acquisition and post processing have been described in a previous study at our hospital (17). MR images were acquired using a 1.5-T scanner (Ingenia; Philips Healthcare, Eindhoven, Netherlands) with retrospective electrocardiographic gating. The CMR protocol consisted of standard steady-state free precession cine MRI and LGE-MRI. LGE-MRI was performed with a T1-weighted inversion-recovery gradient echo sequence (slice thickness, 10 mm; slice gap, -5 mm; TR/TE, 4.6/2.2 ms; flip angle, 15°; field of view 360 mm² × 360 mm²; acquisition matrix size, 266 × 180; reconstruction matrix size, 384 × 384; SENSE factor 2; and inversion time, 200–300 ms) 15 min after contrast agent administration (0.15 mmol/kg, Gd-BTDO3A, Gadovist; Bayer Japan, Tokyo, Japan) in three long-axis slices (two-, three-, and four-chamber) and a stack of short-axis slices completely encompassing the left ventricle. The CMR images were interpreted by experienced observers using commercial post processing software (Circle CVI 42; Circle Cardiovascular Imaging Inc., Calgary, Canada). Myocardial LGE was independently identified by 3 observers (AY, HF, and NH). Discrepancies were resolved through a consensus. The diagnosis of myocardial injury was based on confirmation of the presence of a CMR-late gadolinium enhancement (LGE) by multiple physicians in accordance with previous studies (18). The ratio of the LGE volume and the total LV myocardium volume in the LGE-positive patients was calculated using a 5 SD technique. For the pulmonary blood flow scintigraphy, after an intravenous injection of 260 MBq of ^99mTc-macro-aggregated albumin, SPECT imaging with low-dose CT (GCA9300A; Canon Medical Systems, Tokyo, Japan) was performed with the patient in the supine position. Planar imaging was performed in multiple projections (anterior and posterior, right and left lateral, right anterior and posterior oblique, and left anterior and posterior oblique views). The images were reconstructed in the transaxial, coronal, and sagittal views and reviewed for perfusion defects. The mismatched perfusion defects in this study were based on mismatches between the CT and scintigraphy images. The PE diagnosis was made by multiple physicians using the imaging findings of segmental or subsegmental perfusion defects in areas where no abnormalities were observed on CT images, in accordance with a previous study (19).

We reviewed the medical records of the patients and extracted data on the age, sex, body mass index (BMI), Clinical Frailty Scale (CFS), history of the illness, smoking, and drinking. After the initial outpatient visit, follow-up visits were ordered every 1–3 months for up to one year. Follow-up visits to assess the duration of the symptoms were conducted in the outpatient cardiology clinic of Nihon University Itabashi Hospital. All participants were interviewed face-to-face by a trained cardiologist, who assessed the presence or absence of chest pain, palpitations, and dyspnea, which have a high cardiovascular specificity among the long COVID symptoms reported in previous studies (20), using standardized questions. The in-hospital cardiac events (heart failure, arrythmia), disease severity (requiring ICU admission or ventilator management), treatment, blood tests during hospitalization, long COVID symptom clusters and duration, and blood tests after discharge were similarly recorded. A severe condition during the COVID-19 hospitalization was defined as an SpO2 of less than 93% requiring ventilator management or an ICU admission for an equivalent respiratory management according to the criteria of the Japanese Ministry of Health, Labour, and Welfare.

Continuous variables were reported as the mean ± standard deviation (SD) or median (interquartile range [IQR]). Categorical variables were reported as the percentage and number of patients. Differences in the continuous variables between the two groups (presence/absence of cardiovascular disorders) were evaluated using the unpaired Student’s t-test or Mann–Whitney U-test. Between-group differences in the categorical variables were evaluated using the chi-square test. Comparisons of continuous and categorical variables among the three groups (population with cardiovascular disorders, myocardial injury, PE, or both) were performed using the Kruskal–Wallis test and chi-square test, respectively. A logistic regression analysis was performed to evaluate the odds ratio (OR) of the risk factors and associated long COVID symptoms. Due to the small number of cases, a multivariate analysis was performed using a propensity score based on the age, BMI, and prevalence of atrial fibrillation. A Spearman’s rank correlation was used to assess the association between the follow-up period and duration of the long COVID symptoms. All statistical analyses were performed using SPSS Statistics 24 software (IBM Corp., Armonk, NY, USA), and a p-value <0.05 was considered significant.

Of the 52 patients who visited the cardiology outpatient clinic, a screening examination with an ECG, Chest X-ray, and echocardiography identified RAVI in 9, LAVI in 7, and RAVI/LAVI in 23 patients. SPECT/CT was performed in all 9 patients with RAVI and 4 patients who had RAVI/LAVI but contraindications to CMR. CMR and SPECT/CT were performed in the remaining 19 RAVI/LAVI patients; CMR was performed in 6 of the 7 patients with LAVI (1 was excluded due to a contraindication to CMR). Screening examinations identified the presence of cardiovascular disorders in 27% (14/52), i.e., 15% (8/52) had myocardial injury, 8% (4/52) PEs, and 4% (2/52) both (Figure 1). There were no specific associations between each of the screening examinations and myocardial injury and PEs, but an elevated TRPG identified PEs in 29% (2/7) and right ventricular enlargement in 50% (1/2), respectively, and ST depression identified myocardial injury in 40% (2/5), negative T waves in 40% (2/5), left 4th arch enlargement in 75% (3/4), anterior wall asynergy in 50% (1/2), left atrial P waves in 33% (1/3), and complete left bundle branch block in 50% (1/2), respectively. The site of the LGE lesion was the mid-myocardium (3/10 patients), endomyocardium (1/10 patients), and/or sub-epicardium (6/10 patients), respectively. The inferior and inferolateral wall was the most frequently involved area (8/10 patients). The median of the LGE/myocardium ratio was 7.2% (IQR, 4.2–10.9%). Images of representative cases of myocardial injury and PEs are shown in Figure 2.

The baseline characteristics are summarized in Table 1. Patients with cardiovascular disorders tended to be older (62 ± 10 years vs. 56 ± 14 years, p = 0.07) and had a higher BMI (27.1 ± 6.0 vs 24.7 ± 5.4, p = 0.08) and higher prevalence of atrial fibrillation (AF) (21% [3/14] vs. 5% [2/38], p = 0.08). There were no differences in the sex, CFS, smoking and drinking history, prevalence of diabetes mellitus, dyslipidemia, hypertension, hyperuricemia, chronic kidney disease, history of coronary artery disease, heart failure, strokes, chronic obstructive lung disease/asthma, and malignant tumors between the two groups. The status and therapy profiles are shown in Table 1. The incidence of a severe condition (36% [5/14] vs. 8% [3/38], p = 0.014), and in-hospital cardiac events (71% [10/14] vs. 24% [9/38], p = 0.002) was significantly higher in patients with cardiovascular disorders. There was no significant difference between the two groups regarding drug treatment, such as with steroids, antiviral drugs, antibody cocktails, antirheumatic drugs, and anticoagulants. In terms of the peak biomarker values during the hospitalization, there were no differences in the other biomarkers, such as the troponin I, N-terminal pro-brain natriuretic peptide (NTproBNP), and D-dimer. In patients with cardiovascular disorders, when qualitatively comparing the subgroups with myocardial injury, PEs, and both, the patients with PEs were more frequently female. The incidence of using oxygen therapy, in-hospital cardiac events, and the use of anti-rheumatic drugs were higher in the patients with myocardial injury or both myocardial injury and PEs than in those with a PE alone.

Univariate and multivariate logistic regression analyses evaluated the factors associated with cardiovascular disorders (Table 2). A severe condition (OR, 5.789; 95% CI 1.442–45.220; p = 0.017) and in-hospital cardiac events (OR, 8.079; 95% CI 1.306–25.657; p = 0.021) were independent risk factors of cardiovascular disorders in long COVID patients.

The time from the onset to hospitalization in this cohort of patients was a median of 5 days (IQR, 3–7 days), the length of the hospitalization was a median of 13 days (IQR, 10–24 days), and the time from discharge to the initial outpatient visit was a median of 21 days (IQR, 15–29 days), respectively. The aforementioned total time from the onset to the initial outpatient visit for long COVID symptoms was a median of 40 days (IQR, 35–52 days). The long COVID symptoms are shown in Table 1. Palpitations tended to be more frequent (36% [5/14] vs. 13% [5/38], p = 0.07), and the duration of the long COVID symptoms from the onset of the illness tended to be longer (11 weeks [IQR, 4–20weeks] vs. 6 weeks [IQR, 4–12 weeks], p = 0.09) in the group with cardiovascular disorders (Figure 3). A closer look at the myocardial injury and PEs revealed that PE patients more often presented with chest pain. The median follow-up period was 163 days (IQR, 60–300 days). There was no correlation between the follow-up period and duration of long COVID symptoms (r = 0.33, p = 0.07). No deaths or hospitalizations due to cardiovascular events occurred during the observation period.

In our registry, approximately 10% of the patients had CV long COVID symptoms (dyspnea, palpitations, and/or chest pain) and visited the cardiology outpatient clinic after discharge. In a meta-analysis regarding long COVID, the median (IQR) proportion of COVID-19 survivors experiencing at least one long COVID episode at 1 month was 54.0% (45.0–69.0%; 13 studies), at 2–5 months, 55.0% (34.8–65.5%; 38 studies), and at 6 or more months, 54.0% (31.0–67.0%; 9 studies) (21). Dyspnea is a non-specific symptom; however, chest pain and palpitations are common cardiovascular manifestations of long COVID in survivors of COVID-19 (21). The median (IQR) frequency of chest pain and palpitations is 13.3% (8.8–17.8%; 14 studies) and 9.3% (6.0–10.8%; 5 studies), respectively (21), which was similar to the proportion of patients presenting to the cardiology outpatient clinic with CV long COVID symptoms in this study. Although there are few previous reports of outpatient consultation and treatment specifically for cardiovascular symptoms of long COVID, approximately 10% of our patients presented with cardiovascular symptoms, suggesting that cardiologists should participate in the management of long COVID.

In CV long COVID patients, 15% had a myocardial injury. Several studies on echocardiography abnormalities have been conducted since the beginning of the COVID-19 epidemic (22, 23). While, Puntmann et al. reported that CMR revealed cardiac involvement in 78% of patients and ongoing myocardial inflammation in 60% of patients in post-COVID-19 German patients (14). Brito et al. reported that approximately 30% of previously healthy college athletes in the U.S. recovering from COVID-19 showed imaging features of a resolving pericardial inflammation (15). The frequency of myocardial injury caused by COVID-19 varied in previous studies and was relatively lower in this study. This may be because the patients in this study were mainly those who presented to the cardiology outpatient clinic with CV long COVID symptoms and they were diagnosed by our specific screening examination. A study by Clark et al. with a similar patient population reported that among soldiers with cardiopulmonary symptoms in the late convalescent phase of recovery from SARS-CoV-2, only 8% of the patients had a myocardial injury as detected by CMR (9). Several mechanisms have been proposed to account for myocardial injury, including direct cytotoxic injury (1), dysregulation of the renin-angiotensin-aldosterone system (24), endotheliitis and thromboinflammation (25), and a dysregulated immune response with cytokine release (26). In addition, findings of ST depression, negative T waves, left 4th arch enlargement, asynergy, left atrial P waves, and complete left bundle branch block in the screening tests in this study may be detected in patients with myocardial injury. When these findings are detected in an outpatient setting, CMR and other techniques may be used to make an appropriate diagnosis of myocardial injury. Of the patients with CV long COVID, 8% were diagnosed with peripheral PEs based on the lung perfusion SPECT/CT findings. There have been various reports on the association between COVID-19, thromboses, and abnormal coagulation fibrinolysis since the early stages of the epidemic. It is widely known that an elevated blood D-dimer is often observed, especially in the acute phase of the disease (27). An elevated D-dimer level during a COVID-19 hospitalization is an independent risk factor for an acute thrombosis and death (28, 29). The factors that cause thrombi in various organs during COVID-19 are thought to be an infection of the vascular endothelial cells expressing ACE2 by SARS-CoV2, subsequent damage to vascular endothelial cells, and cytokine storms (30). Some case reports have mentioned PEs as a symptom of long COVID (31); however, most long COVID cases have large PEs that can be detected by contrast-enhanced CT, and there have been few reports involving peripheral PEs and long COVID that can only be detected by lung perfusion SPECT/CT. Among the screening tests in this study, findings such as an elevated TRPG and right ventricular enlargement were characteristic of patients diagnosed with PEs. If these findings are detected in an outpatient setting, SPECT or other procedures may be useful in making an appropriate diagnosis of the PE. In the present study, an in-hospital severe condition and cardiac events were independent risk factors of cardiovascular disorders in CV long COVID patients. Although not statistically significant, patients with cardiovascular disorders tended to be older and had a higher BMI, which are risk factors for the COVID-19 severity, and they also tended to have more palpitations and a longer duration of symptoms. Among the patients with cardiovascular-specific symptoms, a history of severe or cardiac events during hospitalization and a long duration of palpitations and symptoms suggested a high probability of the presence of a cardiovascular disorder and may require special attention in this patient group.

During a median follow-up of 163 days (IQR, 60–300 days), no deaths or hospitalizations due to cardiovascular events were observed among the patients. The duration of symptoms was 11 weeks (IQR, 4–20 weeks) in patients with cardiovascular disorders and 6 weeks (IQR, 4–12 weeks) in those without cardiovascular disorders. In this study, even those who visited the cardiology outpatient clinic with CV long COVID symptoms improved within approximately 12 weeks on average, and no major adverse events occurred during the observation period. Weber et al. reported that among the COVID-19 patients who survived their index hospitalization, the incremental mortality through 1 year was low, even among troponin-positive patients (32). However, few previous studies have reported the prognoses for groups of patients with long COVID symptoms. This may be because long COVID itself is unlikely to have organic cardiovascular disease, and even if it does, it is a very mild condition. However, patients with long COVID with cardiovascular complications tended to have longer-lasting symptoms of long COVID, however, the prognosis was not worse. Early detection of cardiovascular problems in cardiology for symptomatic long COVID patients may inform patients of the duration of symptoms and allow symptoms to be shortened through appropriate therapeutic intervention.

This study had several limitations. First, as it was conducted in a prospective observational fashion, causal relationships could not be established. Because of the selection bias that existed in this cohort due to the presence of those who did not accept the physician’s suggestion of an outpatient visit, the small proportion of women and patients in this study with a history of heart failure and coronary artery disease with little or no left ventricular systolic dysfunction, unmeasured confounders should be considered when interpreting the results. In particular, our results were based on the data obtained through specific screening examinations, which have a different sensitivity and specificity in identifying a cardiac pathology at the cardiology outpatient clinic; therefore, because symptomatic potential pulmonary disease and asymptomatic potential cardiac disease might be missed, our results regarding the use of imaging modalities and clinical outcomes between facilities should be considered descriptive. Second, due to the low number of clinical events, this study included a small number of patients, and we could not perform a statistical analysis of the factors associated with myocardial injury and PEs. In addition, the number of clinical events may not have been large enough to obtain statistical significance. Third, in the CMR analysis of this study, mainly a qualitative evaluation of the LGE was performed, but T1mapping, T2mapping, and a strain analysis were not performed. The purpose of this study was to clarify the incidence of overt myocardial injury in patients with long COVID. Therefore, subtle left ventricular and right ventricular dysfunction might have been missed. Finally, the clinical data before the COVID-19 hospitalization was lacking; hence, whether cardiovascular disorders existed before the hospitalization could not be determined.

The suspicion of cardiovascular involvement in CV long COVID patients was approximately 30%. A severe condition during hospitalization and in-hospital cardiac events were risk factors of a cardiovascular sequelae in CV long COVID patients.