This cross-sectional study consecutively recruited 142 healthcare workers with laboratory confirmed SARS-CoV-2 infection in the University Hospital of Salamanca, and who volunteered for the study. Among them, 106 healthcare workers tested positive for SARS-CoV-2 by reverse-transcriptase polymerase chain reaction (RT-PCR) from nasopharyngeal swab between March 13 and April 25, 2020; and 36 healthcare workers were diagnosed after testing positive for anti-SARS-CoV-2-immunoglobulin G (IgG) antibodies by ELISA between April 10 and May 22, 2020, as part of a major hospital campaign. The purpose of this second group was to also provide data from participants with past SARS-CoV-2 infection, who are more likely to have mild symptoms of viral infection, and because population-based SARS-CoV-2 seroprevalence studies are becoming more established.6,7 Study enrolment began on May 25 and finished on June 12, 2020.

Institutional approval (2020/05/490) for the study was provided by the University Hospital of Salamanca Ethics Committee, and all participants provided written informed consent. The study is registered with ClinicalTrials.gov NCT04413071. The responsibility for the study design, data collection and data interpretation lay solely with the study investigators. An internal adjudication monitoring board reviewed all cardiac study findings and adjudicated study outcomes. The authors had full access to all the data and elaborated all materials to submit for publication.

All participants underwent clinical evaluation, electrocardiography, laboratory tests and CMR imaging at the same visit. After obtaining written informed consent, trained interviewers used a structured questionnaire to collect baseline data in face-to-face interviews. A cardiologist took a complete medical history, performed a physical examination and reviewed the completeness of the questionnaire in a separate room, where an electrocardiogram was performed, and blood samples were drawn immediately before the CMR. Electrocardiograms were interpreted in consensus by 2 experienced readers, who were blinded to participant identification, clinical history, symptoms, physical examination, and other findings.

CMR was performed using a clinical 1.5 whole-body magnetic resonance scanner in the cardiac imaging laboratory of the University Hospital of Salamanca.8 The imaging acquisition protocol is described in detail in Methods of the supplementary data. CMR images were globally and regionally analyzed using dedicated software, in consensus by 2 experienced readers, who were blinded in a similar manner to the electrocardiogram protocol. T2 and T1-based markers of myocardial inflammation were analyzed in each of the 16 segments of the 17-segment model of the American Heart Association (the true apex was excluded), where only positive segment concordances from the different T2 and T1-based markers were considered. Because myocarditis was diagnosed according to these T2 and T1-based CMR markers and an adequate selection of normal reference values is fundamental, we used a population-based control CMR imaging group from 50 sex- and aged-matched individuals without cardiac disease.9 The prevalences of cardiovascular risk factors (hypertension, diabetes mellitus, dislipemia, current smoking) in the control cohort were similar to those in the study population.

Immunophenotypic analysis of (> 250) immune cell populations was performed in peripheral blood samples collected in K3-ethylenediaminetetra-acetic acid (EDTA, 10mL/sample, and stained with the EuroFlow lymphocyte screening tube and the cluster of differentiation 4 T cell (TCD4), natural killer (NK)/TCD8, beta-lactoglobulin hydrolysates (BIgH) and monocyte-derived dendritic cell (MoDC) immune monitoring tubes by flow cytometry (FACSCANTO II and LSR-Fortessa, respectively; Becton/Dickinson Biosciences, United States) using a dual-platform assay previously described in detail.10 Reference values for the individual immune cell subsets investigated in blood by flow cytometry were defined based on a population-based control group of 463 age-matched adults (median age 52 [IQR 47-61] years) studied prior to the SARS-CoV-2 pandemic. Anti-SARS-CoV-2-IgM (AnshLabs, Webster, United States), IgG and IgA (Mikrogen Diagnostik, Neuried, Germany) antibody levels were measured in parallel in plasma from the same blood samples using commercially available in vitro diagnostic medical device approved (semiquantitative) ELISA kits, strictly as instructed by the manufacturers.

Study outcome measures were the prevalence of clinical pericarditis and of myocarditis, and the characterization of the immune cell compartments in blood, and the virus-specific humoral immune response. Clinically suspected pericarditis was diagnosed if at least 2 of the following criteria were present, following current guidelines3: pericarditic chest pain, pericardial rub on auscultation, widespread ST-elevation or PR segment depression on electrocardiogram, and evidence of pericardial effusion on CMR. Elevation of inflammation markers, C-reactive protein, and evidence of pericardial inflammation on CMR were used as additional supporting findings. The diagnosis of clinically suspected myocarditis was based on CMR criteria11; we considered as main CMR criteria positive edema-sensitive T2-based markers (T2-weighted images or T2-mapping) or positive T1-based tissue characterization markers (abnormal T1-relaxation time or extracellular volume or late gadolinium enhancement), and as supportive CMR criteria either pericardial effusion, or evidence of pericardial inflammation on CMR, or systolic left ventricular wall motion abnormalities. Considering that participants were being examined after the acute phase of SARS-CoV-2 infection, in this study clinically suspected myocarditis was defined as the presence of a combination of at least 2 T2 or T1-based CMR main criteria or the presence of combination of only 1 T2 or T1-based main criterion with 1 additional CMR supportive criterion.

As we were aware that pericarditis and myocarditis occur together in clinical practice, we therefore defined as clinically suspected myopericarditis those cases of pericarditis with associated myocarditis on CMR but without left ventricular wall motion abnormalities, and as clinically suspected perimyocarditis those cases where left ventricular wall motion abnormalities were present.12

Descriptive statistics were used to summarized the data. Results are presented as the proportion (%) of valid cases for categorical variables and as the median [IQR] for continuous variables. Differences between groups were analyzed by the Fisher exact test for categorical variables and by the nonparametric Mann-Whitney or Kruskal-Wallis tests for continuous data. Comparisons between immune cell counts in the blood of patients and controls were adjusted for age and sex (covariates) using the 1-way ANCOVA (analysis of covariance) univariate general linear modeling test (SPSS statistical software package v25.0, IBM, Armonk, United States). We compared characteristics of participants and examinations, all tables, according to the final clinical diagnosis (nonpericardial and myocardial manifestations vs pericarditis vs myopericarditis vs myocarditis). For 2-dimensional visualization of flow cytometry data, multivariate canonical analysis with multidimensional reduction of data via linear discriminant analysis, and the t-distributed stochastic neighbor embedding machine-learning algorithm visualization tools were used (Infinicyt software, Cytognos, Universidad de Salamanca).13

Figure 1 depicts the flowchart for participant selection among healthcare workers. Of the 142 recruited healthcare workers who signed informed consent, 1 participant did not complete the CMR due to claustrophobia. Two additional participants were excluded due to a history of severe hypertrophic myocardiopathy in 1 case, and inherited immune deficiency in the other. Thus, a total of 139 participants completed the clinical assessment, electrocardiography, laboratory tests and CMR. Of these, 103 (74.1%) had been diagnosed by RT-PCR and 36 (25.9%) by serology. No participants were clinically diagnosed with post-COVID-19 cardiac involvement at the time of their index presentation, nor were those who were previously hospitalized.

Figure 1.

Flowchart for participant selection among healthcare workers. CMR, cardiac magnetic resonance; IgG, Inmunoglobulin G, RT-PCR, reverse-transcriptase-polymerase-chain-reaction; UHS, University Hospital of Salamanca.

(0.28MB).

All participant characteristics are shown in table 1. By professional categories, 49 (35.3%) were nurses, 35 (25.2%) physicians, and the remaining 55 (39.6%) included different profiles such as auxiliary nurses and other hospital staff. A total of 67 (48.2%) healthcare workers were infected while directly attending COVID-19 hospitalization wards.

Among the overall study population, 106 (76.3%) had at least 1 comorbidity and 8 (5.8%) healthcare workers had a history of cardiovascular disease: 1 with chronic ischemia with stent revascularization, 3 with paroxysmal atrial fibrillation, 2 with intranodal supraventricular tachycardias treated with ablation, and 2 with an episode of acute pericarditis several years previously.

Most (137 [98.6%]) healthcare workers experienced a viral prodrome during SARS-CoV-2 infection and cardiac symptoms with shortness of breath, chest pain, palpitations or dizziness were reported by 86 (61.9%) participants. A total of 27 (19.4%) healthcare workers were previously diagnosed with COVID-19 pneumonia and 23 (16.5%) required hospitalization (none of these were diagnosed with pericarditis or myocarditis during this hospitalization index).

Chronic drug therapy and treatment during SARS-CoV-2 infection are shown in table 1 of the supplementary data. Overall, the drug therapy aimed at ameliorating the disease was heterogeneous: hydroxychloroquine was given in 33 (23.7%) participants, lopinavir-ritonavir in 17 (12.2%), oral glucocorticoids in 9 (6.5%), high-dose intravenous bolus of methylprednisolone in 15 (10.8%), and interleukin inhibitors in 18 (12.9%).

The study examinations (table 2) were performed 10.4 (9.3-11.0) weeks after the start of symptoms of infection. All participants had vital and exploratory signs of hemodynamic stability on examination. A total of 91 (65.5%) healthcare workers still had symptoms, which were cardiac-related in 58 (41.7%).

Of the 139 electrocardiograms, electrocardiographic abnormalities were reported in 69 (49.6%) cases (table 2 of the supplementary data). A total of 33 (23.7%) electrocardiograms met the criteria for pericarditis-like changes (figure 1 of the supplementary data). As the Occupational Health Service of the hospital provide healthcare workers a baseline medical evaluation before starting their jobs, we were able to recover and review 53 (76.8%) prior baseline electrocardiograms from the 69 patients with electrocardiographic changes at the study examination. Of these 53 electrocardiographic comparisons, 67.9% electrocardiographic changes at the study examination were not present previously and consisted mainly of pericardial like changes and ST-segment depression or T-wave inversion (table 3 of the supplementary data).

Cardiac-specific and inflammatory biomarkers were within the normal range in most participants (table 2). CMR abnormalities were observed in 84 (60.4%) participants (table 3 and tables 4 and 5 of the supplementary data, figure 2). Two (1.4%) participants showed increased myocardial T2-relaxation time, 5 (3.6%) edema on T2-weighted images, 40 (28.8%) increased native myocardial T1-relaxation time, 27 (19.4%) increased T1-extracellular volume, 10 (7.2%) T1-late gadolinium enhancement, 42 (30.2%) pericardial effusion, 1 (0.7%) a pericardial thickness of 3mm and 7 (5.0%) systolic left ventricular wall motion abnormalities, global or regional.

Figure 2.

Cardiac magnetic resonance imaging composition from a participant with pericardial and myocardial involvement. The main findings are pericardial effusion on the inferior wall (*), subtle subepicardial late gadolinium enhancement (red arrows) and increased T1-native relaxation time on the inferolateral segment, with nonsignificant increased T2 relaxation time on this segment. All images are short-axis views at papillary muscles level. Image A: end-diastolic cine image (Steady State Free Precession, SSFP). Image B: phase-sensitive inversion-recovery late gadolinium enhancement. Image C: T1-native mapping (Modified Look-Locker Imaging, MOLLI). Image D: T2 mapping (Gradient and Spin-Echo, GraSE).

(0.27MB).

A total of 43 (30.9%) participants fulfilled the criteria for either clinically suspected pericarditis or myocarditis. Clinically suspected isolated pericarditis was diagnosed in 8 (5.8%) participants, isolated myocarditis in 24 (17.3%), and myopericarditis in 11 (7.9%). These were no cases of perimyocarditis. Descriptions of criteria combinations are provided in figure 3 and baseline and examination characteristics for each diagnostic group are detailed in table 1 and table 2.

Figure 3.

Description of pericarditis clinical criteria and cardiac magnetic resonance criteria combinations in participants diagnosed with pericarditis, myopericarditis, or myocarditis. CRP, C-reactive protein; ECG, electrocardiogram; ECV, increase in T1-extracellular volume; dffusion, pericardial effusion assessed by CMR; LGE, T1-late gadolinium enhancement; LV, left ventricular wall motion abnormalities; T1 map, increase in native myocardial T1-relaxation time; T2 map, increase in myocardial T2-relaxation time; T2W, increase in T2-weighted hyperintensity; thickened, pericardial thickness≥ 3mm

(0.23MB).

A higher percentage of participants with clinically suspected pericarditis, myopericarditis or myocarditis had cardiac symptoms during SARS-CoV-2 infection (32 [74.4%] vs 54 [56.3%]; P=.058) and on study examination (26 [60.5%] vs 32 [33.3%]; P=.005) than participants without these manifestations. Chronic drug therapy with statins was more frequent in participants without pericardial or myocardial manifestations than in those with clinically suspected pericarditis, myopericarditis or myocarditis (16 [16.7%] vs 1 [2.3%]; P=.022).

Among the 36 participants diagnosed with past infection through anti-SARS-CoV-2-IgG detection (data for this group, compared with RT-PCR participants, are shown in table 6 of the supplementary data), 34 (94.4%) reported at least 1 of the 18 collected symptoms of COVID-19 infection (ie, fever, persistent cough and/or anosmia, whose presence, according to current guidelines would require isolation and testing was present in 27 (75.0%) of these 36 participants); 28 (77.8%) had previously tested negative by RT-PCR after developing SARS-CoV-2 symptoms and 8 (22.2%) were never RT-PCR tested.

A lower percentage of participants diagnosed through positive serology still had symptoms on examination compared with RT-PCR participants (18 [50.0%] vs 73 [70.9%]; P=.027); nonetheless, the prevalence of clinically suspected pericarditis, myopericarditis or myocarditis was high in both groups (figure 4).

Figure 4.

Prevalence of pericardial and myocardial manifestations in participants with SARS-CoV-2 infection diagnosed through RT-PCR or through serology. RT-PCR, reverse-transcriptase polymerase chain reaction.

(0.12MB).

Most study participants (101 [73.2%]) displayed altered cell counts in blood for at least 1 major immune cell population, as illustrated in table 4 and figure 5, and described in more detail in tables 7 and 8 of the supplementary data. The most frequent alterations consisted of eosinopenia (38 [27.3%] cases; P <.001) and increased T cells (specially for CD4-CD8-/lo (cytotoxic) T cell (24 [17.3%] cases; P <.001); and, to a lesser extent, also B cell counts (16 [11.5%] cases; P <.001). In addition, compared with age- and sex-matched healthy donors, participants had higher median counts in blood of basophils (38 vs 47 cells/μL, respectively; P=.007) and monocytes (317 vs 405 cells/μL; P <.001); in contrast, participants showed decreased median numbers of circulating blood neutrophils (3723 vs 3430 cells/μL; P=.010), eosinophils (157 vs 74 cells/μL; P <.001), and circulating plasma cells (1.7 vs 0.8 cells/μL; P <.001).

Figure 5.

Altered immune cell profiles and antibody serum levels in healthcare workers according to the presence vs absence of pericarditis, myopericarditis, or myocarditis. A: t-SNE graphical representation of the distribution of the major immune cell populations in blood of a healthy donor (plot on the left) and a participant diagnosed with pericarditis (middle plot) showing both increased Tab+CD4-CD8-/lo T cell and decreased eosinophil counts in blood. B: 2-dimensional graphical representation of multivariate canonical (linear discriminant) analysis plots showing the presence of overall different immune cell profiles in blood of the participants with past SARS-CoV-2 infection (n=139; colored circles) vs age-matched healthy donors (n=463; grey squares) (2 plots on the left); distinct immune cell profiles were also observed between participants diagnosed with pericarditis (blue circles) and myocarditis (orange circles); most myopericarditis cases (red circles) shared an altered immune cell profile in blood similar to that of myocarditis. C: frequency and amount of anti-SARS-CoV-2 antibodies measured in plasma of 123/139 participants. CA, canonical (correlation) analysis; IgG, immunoglobulin G; NK, natural killer; T-SNE, t-distributed stochastic neighbor embedding algorithm (abbreviated in the graphical representations as TS1); SCC, sideward light scatter (ie, cell granularity).

(0.79MB).

Compared with healthy donors, participants with clinically suspected pericarditis showed the highest median counts in blood of CD4-CD8-/lo (cytotoxic) T cells (46 vs 104 cells/μL; P=.003) and displayed particularly decreased numbers of eosinophils (157 vs 48 cells/μL; P=.049), as well as (similar to participants with clinically suspected myopericarditis and those with myocarditis) decreased blood counts of circulating plasma cells (1.7 vs 0.8, 0.4 and 0.7 cells/μL; P=.075, P=.018 and P=.004, respectively). In turn, patients with clinically suspected myocarditis more frequently had decreased neutrophil counts in blood (7 [29.2%]; P=.011), together with decreased NK-cell numbers (260 vs 156 cells/μL; P=.021). Finally, participants diagnosed with clinically suspected myopericarditis showed mixed immune profiles, which were preferentially shared with those of myocarditis (7 of the 11 cases; 63.6%) rather than pericarditis (4 of the 11 cases; 36.4%) (figure 5B).

Overall, no major differences were observed among participants with or without pericardial and myocardial involvement in the frequency and levels of anti-SARS-CoV-2- IgM, IgG and IgA antibodies in plasma (figure 5C). Importantly, overlapping immune profiles were detected between participants diagnosed by RT-PCR compared with those diagnosed by serology.

The major limitation of this study is that clinically suspected myocarditis was not confirmed via endomyocardial biopsy. CMR T1 and T2 measures, although significant, were small between the participants with SARS-CoV-2 infection and the control group. The study analysis was limited to healthcare workers and therefore has limited external generalizability to other nonhealthcare settings. However, the strength of this study is the addition of nonhospitalized participants, as well as the inclusion of participants diagnosed with past SARS-CoV-2 infection through serology, who also had a high prevalence of pericardial and myocardial involvement.