-
PDF
- Split View
-
Views
-
Cite
Cite
Laura Sánchez-de Prada, Adrián García-Concejo, Álvaro Tamayo-Velasco, Marta Martín-Fernández, Hugo Gonzalo-Benito, Óscar Gorgojo-Galindo, A Montero-Jodra, María Teresa Peláez, Iciar Martínez Almeida, Miguel Bardají-Carrillo, Rocío López-Herrero, Patricia Román-García, José María Eiros, Iván Sanz-Muñoz, Teresa Aydillo, María Ángeles Jiménez-Sousa, Amanda Fernández-Rodríguez, Salvador Resino, María Heredia-Rodríguez, David Bernardo, Ester Gómez-Sánchez, Eduardo Tamayo, miRNome Profiling of Extracellular Vesicles in Patients With Severe COVID-19 and Identification of Predictors of Mortality, The Journal of Infectious Diseases, 2024;, jiae310, https://doi.org/10.1093/infdis/jiae310
- Share Icon Share
Abstract
Extracellular vesicles (EVs), containing microRNAs (miRNAs) and other molecules, play a central role in intercellular communication, especially in viral infections caused by SARS-CoV-2. This study explores the miRNA profiles in plasma-derived EVs from patients with severe COVID-19 vs controls, identifying potential mortality predictors.
This prospective study included 36 patients with severe COVID-19 and 33 controls without COVID-19. EV-derived miRNAs were sequenced, and bioinformatics and differential expression analysis between groups were performed. The plasma miRNA profile of an additional cohort of patients with severe COVID-19 (n = 32) and controls (n = 12) was used to compare with our data. Survival analysis identified potential mortality predictors among the significantly differentially expressed (SDE) miRNAs in EVs.
Patients with severe COVID-19 showed 50 SDE miRNAs in plasma-derived EVs. These miRNAs were associated with pathways related to inflammation and cell adhesion. Fifteen of these plasma-derived EV miRNAs were SDE in the plasma of severe cases vs controls. Two miRNAs, hsa-miR-1469 and hsa-miR-6124, were identified as strong mortality predictors with an area under the receiver operating characteristic curve of 0.938.
This research provides insights into the role of miRNAs within EVs in severe COVID-19 and their potential as clinical biomarkers for mortality.
The COVID-19 pandemic caused by SARS-CoV-2 has profoundly affected health care and economic systems worldwide [1]. Clinical presentation is varied and can range from asymptomatic to severe disease and death. Severe disease presents with interstitial pneumonia that rapidly leads to acute respiratory distress syndrome or septic shock [2]. Patients with severe cases have been described to present a dysregulated immune response, including a “cytokine storm” and immunosuppression [3]. Mortality in COVID-19 has evolved through time and variants since the emergence of the pandemic [1].
Extracellular vesicles (EVs) are lipid-bilayer membrane vesicles of 30 to 150 nm secreted into the extracellular environment carrying different cargoes, such as DNAs, microRNAs (miRNAs), and mRNAs, among others, which mediate cell-to-cell communication [4]. Many studies have demonstrated that miRNAs delivered by EVs are an essential mechanism of intercellular communication [5, 6]. Actually, EV miRNAs have been identified as critical players in the modulation of numerous biological processes [7].
Recent findings suggest that they might play a key role in infection and inflammatory responses, as in SARS-CoV-2 infection, by modulating the immune response to the virus or interfering with viral replication and transmission [8]. In addition, EV miRNAs have been associated with immune responses in convalescence and long COVID symptoms, as well as with influencing the response to SARS-CoV-2 vaccination in patients with cancer [9]. Additionally, ACE2-positive exosomes contain specific miRNAs that are increased in patients with COVID-19 [10]. But the literature on the miRNA profile of COVID-19 is limited, with a lack of extensive characterization of EV miRNAs. Thus, we hereby aim to investigate the expression profile of plasma-derived EV miRNAs in patients with severe COVID-19 as compared with a control population (COVID-19 negative). Furthermore, we compared this plasma EV miRNA profile with the plasmatic miRNA profile in a previously described cohort with severe COVID-19 [11]. We also integrated our results into a survival analysis to identify potential mortality predictors among those differentially expressed miRNAs.
METHODS
Design and Study Population
A prospective study was performed among 36 patients with a diagnosis of severe or critical COVID-19, which was defined as admission to the intensive care unit (ICU), intubation or mechanical ventilation, or death, based on the Centers for Disease Control and Prevention [12]. Patients were recruited from the Hospital Clínico Universitario de Valladolid (Spain) between 24 March and 11 April 2020. All patients had a positive result for SARS-CoV-2 infection confirmed by reverse transcription polymerase chain reaction in nasopharyngeal swabs. Patients presenting other acute diseases, infections, or chronic terminal illnesses were excluded from the study. Additionally, 33 volunteers without COVID-19 were recruited as controls. These patients were recruited upon admission for scheduled minor surgery and had a negative reverse transcription polymerase chain reaction COVID-19 test result at the moment of the sample collection. The hospital's Clinical Ethics Committee approved the study protocol, and informed written consent was obtained from patients or legal representatives before recruitment (PI-20-1717). This research was performed according to the Declaration of Helsinki’ ethics code.
A second cohort was used to compare the plasma EV miRNA profile with the plasma miRNA profile from a multicenter observational study performed at 3 public Spanish hospitals [11]: Hospital Universitario Infanta Leonor, Hospital Universitario Del Tajo, and Hospital Universitario Príncipe de Asturias. These patients were enrolled from March to August 2020, where 32 with severe COVID-19 and 12 controls were selected for reanalysis and comparison with our results. The Ethics Committee of the Instituto de Salud Carlos III (PI 33_2020-v3) and the ethics committee of each hospital approved the study protocol, and informed written consent was obtained from patients or legal representatives before recruitment. This research was also performed according to the Declaration of Helsinki’s ethics code.
Clinical Data and Sample Collection
Epidemiologic and clinical data were collected from medical records. Plasma samples were collected at 9 Am after admission to avoid fluctuations due to circadian rhythms and before treatment with specific therapies for COVID-19, such as immunotherapy (tocilizumab, interferon β, corticoids, or ribavirin, among others). Samples were drawn into 3.2% sodium citrate tubes, which have been recommended as the optimal anticoagulant for the study of microvesicles. The samples were centrifuged at 2000g for 20 minutes at room temperature. The resulting plasma was aliquoted and frozen at −80 °C until used. Plasma samples from the second cohort were collected and processed as previously described [11].
Analysis of miRNome by High-Throughput Sequencing
EVs were purified and total RNA was isolated, including small RNAs, from 500 μL of plasma with the exoRNeasy Midi Kit (Qiagen) following the manufacturer's instructions. The quality and quantity of RNA were evaluated by the Bioanalyzer 2100 with the Agilent RNA 6000 Nano Kit. Small RNA libraries were constructed with the NEXTFLEX Small RNA-Seq Kit version 4 for Illumina Platforms (PerkinElmer). Next, small RNAs were sequenced in NovaSeq 6000 (Illumina) by ADM-BIOPOLIS GENOMICS at Parc Científic Universitat of Valencia (Spain), estimating >10 million reads per sample. Sequences from EV-derived miRNA samples can be accessed through BioStudies with accession number S-BSST1192, while sequences from plasma-derived miRNAs can be accessed through accession number E-MTAB-10562.
Data-Processing Pipeline: Bioinformatics Analysis
Raw sequence data from BCL files were subjected to translation into the FASTA format via the bcl2fastq tool (Illumina). Demultiplexing was performed to assign individual reads to their respective samples. To ensure data quality, reads failing to meet the Illumina chastity filter were eliminated due to ambiguous base calls. The remaining reads underwent quality assessment through FastQC (version 0.11.3), and adapter sequences were removed with cutadapt (version 1.13). miRDeep2 (version 0.0.7) was used for reading alignment against the reference human genome (GRCh38) by the Bowtie1-based mapper.pl module. Retention criteria included zero mismatches in the seed region and mapping to <5 genomic loci. miRNA quantification was performed via the quantifier.pl module, involving alignment against predefined mature miRNA sequences and quantification of reads within a specific interval around these sequences, which determines the expression of the corresponding known miRNAs by using the predefined precursors in miRBase (version 2.0), the public repository for all published miRNA sequences [13]. Reads falling into an interval of 2 nucleotides upstream and 5 nucleotides downstream of the mature miRNA sequences were determined for quantification.
At least 10 million reads per sample were obtained. A total of 2880 miRNAs from miRBase were used in the alignment, with 1300 plasma EV-derived miRNAs successfully prefiltered with sufficiently large counts for the differential expression analysis. The prefilter step was performed with the recommended parameters provided by the DESeq2 R package documentation.
Statistical Analysis
Statistical analyses were carried out with the R statistical package (version 4.2.3). Categorical data underwent chi-square testing, while continuous variables were subjected to the Wilcoxon test. Normalization of miRNA counts was achieved via the DESeq2 method (version 1.38.3). Differential expression analysis relied on a generalized linear model with negative binomial distribution, adjusting for age and sex. Significance was determined by fold change (FC) ≥1.5 and Q value ≤ 0.05 (false discovery rate–corrected P value). miRNA expression data were subjected to partial least squares discriminant analysis (PLS-DA) with the mixOmics R package (version 6.22.0). PLS-DA was employed to identify patterns and distinctions in miRNA expression among different groups within the study cohorts. A multivariate Cox regression model with LASSO (least absolute shrinkage and selection operator) incorporated variance stabilized transformation normalized miRNA counts for predictive value selection. Mortality analysis from the admission and sampling up to 90 days was conducted via a Cox proportional hazards regression model, adjusting for age and sex. This analysis was carried out via the R software packages survival (version 3.5.5), pROC (version 1.18.2), and survminer (version 0.4.9). This analysis involved the generation of Kaplan-Meier and receiver operating characteristic curves for each significantly differentially expressed (SDE) miRNA, considering an area under the curve ≥0.85 as excellent.
miRNA-Based Target Prediction and Pathway Enrichment and Cellular Process Analysis
SDE miRNAs underwent miRNA target interaction analysis with the miRNAtap and ClusterProfiler R packages. Gene targets with ≥3 validated interactions and Q value ≤ 0.05 were selected for functional enrichment analysis via the Kyoto Encyclopedia of Genes and Genomes (KEGG) and Gene Ontology (GO) cellular processes databases. Significance was established at Q value ≤ 0.05.
RESULTS
Patient Characteristics
A total of 36 patients with severe COVID-19 and 33 controls were included for the miRNA differential expression analysis in plasma-derived EVs. Epidemiologic and clinical characteristics are described in Table 1.
. | COVID-19 . | Controls . | P Valuea . |
---|---|---|---|
No. of patients | 36 | 33 | |
Age, y, median (IQR) | 73.69 (18.80) | 65.01 (14.56) | .008 |
Sex: male, No. (%) | 21 (58.3) | 21 (63.6) | .652 |
Comorbidities, No. (%) | |||
Smoker | 2 (5.6) | 5 (15.5) | .187 |
Alcoholism | 1 (2.8) | 2 (6.1) | .504 |
Cardiopathy | 9 (25.0) | 7 (21.2) | .710 |
Diabetes | 8 (22.2) | 4 (12.1) | .269 |
Arterial hypertension | 22 (61.1) | 16 (48.5) | .292 |
Obesity | 7 (19.4) | 2 (6.1) | .099 |
Chronic obstructive pulmonary disease | 5 (13.9) | 3 (9.1) | .534 |
Asthma | 1 (2.8) | 1 (3.0) | .950 |
Chronic kidney disease | 0 (0.0) | 3 (9.1) | .064 |
Biochemistry, median (IQR) | |||
Creatinine, mg/dL | 0.85 (0.432) | 0.645 (0.389) | .060 |
Leucocytes/μL | 8030 (3233) | 7231 (1818) | .465 |
Linfocytes/μL | 1200 (872) | 2321 (1035) | <.001 |
Neutrophils/μL | 6430 (3209) | 4020 (1617) | <.001 |
Platelets/μL | 229 030 (109 932) | 241 406 (60 384) | .267 |
Triglycerids, mg/dL | 135.20 (48.06) | 106.47 (48.9) | .018 |
Cholesterol, mg/dL | 135.8 (37.96) | 173.13 (30.8) | <.001 |
Glucose, mg/dL | 128.8 (60.17) | 101.9 (15.3) | .272 |
Creatinine, mg/dL | 1.21 (1.096) | 1.0 (0.6) | .633 |
Bilirubin, mg/dL | 0.60 (0.50) | 0.60 (0.30) | .751 |
Hospitalization variables | |||
Temperature, °C, median (IQR) | 36.71 (0.81) | 36.67 (0.75) | .953 |
Exitus, No. (%) | 12 (33.3) | 0 (0.0) | <.001 |
Length of hospital stay, d, median (IQR) | 9.5 (18) | 0 (0) | <.001 |
Therapy, No. (%) | |||
Thyroid hormones | 0 (0.0) | 4 (12.1) | .031 |
Statins | 7 (19.4) | 11 (33.3) | .189 |
β-blockers | 1 (2.8) | 2 (6.1) | .504 |
Immunosupressants | 1 (2.8) | 0 (0.0) | .338 |
. | COVID-19 . | Controls . | P Valuea . |
---|---|---|---|
No. of patients | 36 | 33 | |
Age, y, median (IQR) | 73.69 (18.80) | 65.01 (14.56) | .008 |
Sex: male, No. (%) | 21 (58.3) | 21 (63.6) | .652 |
Comorbidities, No. (%) | |||
Smoker | 2 (5.6) | 5 (15.5) | .187 |
Alcoholism | 1 (2.8) | 2 (6.1) | .504 |
Cardiopathy | 9 (25.0) | 7 (21.2) | .710 |
Diabetes | 8 (22.2) | 4 (12.1) | .269 |
Arterial hypertension | 22 (61.1) | 16 (48.5) | .292 |
Obesity | 7 (19.4) | 2 (6.1) | .099 |
Chronic obstructive pulmonary disease | 5 (13.9) | 3 (9.1) | .534 |
Asthma | 1 (2.8) | 1 (3.0) | .950 |
Chronic kidney disease | 0 (0.0) | 3 (9.1) | .064 |
Biochemistry, median (IQR) | |||
Creatinine, mg/dL | 0.85 (0.432) | 0.645 (0.389) | .060 |
Leucocytes/μL | 8030 (3233) | 7231 (1818) | .465 |
Linfocytes/μL | 1200 (872) | 2321 (1035) | <.001 |
Neutrophils/μL | 6430 (3209) | 4020 (1617) | <.001 |
Platelets/μL | 229 030 (109 932) | 241 406 (60 384) | .267 |
Triglycerids, mg/dL | 135.20 (48.06) | 106.47 (48.9) | .018 |
Cholesterol, mg/dL | 135.8 (37.96) | 173.13 (30.8) | <.001 |
Glucose, mg/dL | 128.8 (60.17) | 101.9 (15.3) | .272 |
Creatinine, mg/dL | 1.21 (1.096) | 1.0 (0.6) | .633 |
Bilirubin, mg/dL | 0.60 (0.50) | 0.60 (0.30) | .751 |
Hospitalization variables | |||
Temperature, °C, median (IQR) | 36.71 (0.81) | 36.67 (0.75) | .953 |
Exitus, No. (%) | 12 (33.3) | 0 (0.0) | <.001 |
Length of hospital stay, d, median (IQR) | 9.5 (18) | 0 (0) | <.001 |
Therapy, No. (%) | |||
Thyroid hormones | 0 (0.0) | 4 (12.1) | .031 |
Statins | 7 (19.4) | 11 (33.3) | .189 |
β-blockers | 1 (2.8) | 2 (6.1) | .504 |
Immunosupressants | 1 (2.8) | 0 (0.0) | .338 |
aBold indicates P < .05.
. | COVID-19 . | Controls . | P Valuea . |
---|---|---|---|
No. of patients | 36 | 33 | |
Age, y, median (IQR) | 73.69 (18.80) | 65.01 (14.56) | .008 |
Sex: male, No. (%) | 21 (58.3) | 21 (63.6) | .652 |
Comorbidities, No. (%) | |||
Smoker | 2 (5.6) | 5 (15.5) | .187 |
Alcoholism | 1 (2.8) | 2 (6.1) | .504 |
Cardiopathy | 9 (25.0) | 7 (21.2) | .710 |
Diabetes | 8 (22.2) | 4 (12.1) | .269 |
Arterial hypertension | 22 (61.1) | 16 (48.5) | .292 |
Obesity | 7 (19.4) | 2 (6.1) | .099 |
Chronic obstructive pulmonary disease | 5 (13.9) | 3 (9.1) | .534 |
Asthma | 1 (2.8) | 1 (3.0) | .950 |
Chronic kidney disease | 0 (0.0) | 3 (9.1) | .064 |
Biochemistry, median (IQR) | |||
Creatinine, mg/dL | 0.85 (0.432) | 0.645 (0.389) | .060 |
Leucocytes/μL | 8030 (3233) | 7231 (1818) | .465 |
Linfocytes/μL | 1200 (872) | 2321 (1035) | <.001 |
Neutrophils/μL | 6430 (3209) | 4020 (1617) | <.001 |
Platelets/μL | 229 030 (109 932) | 241 406 (60 384) | .267 |
Triglycerids, mg/dL | 135.20 (48.06) | 106.47 (48.9) | .018 |
Cholesterol, mg/dL | 135.8 (37.96) | 173.13 (30.8) | <.001 |
Glucose, mg/dL | 128.8 (60.17) | 101.9 (15.3) | .272 |
Creatinine, mg/dL | 1.21 (1.096) | 1.0 (0.6) | .633 |
Bilirubin, mg/dL | 0.60 (0.50) | 0.60 (0.30) | .751 |
Hospitalization variables | |||
Temperature, °C, median (IQR) | 36.71 (0.81) | 36.67 (0.75) | .953 |
Exitus, No. (%) | 12 (33.3) | 0 (0.0) | <.001 |
Length of hospital stay, d, median (IQR) | 9.5 (18) | 0 (0) | <.001 |
Therapy, No. (%) | |||
Thyroid hormones | 0 (0.0) | 4 (12.1) | .031 |
Statins | 7 (19.4) | 11 (33.3) | .189 |
β-blockers | 1 (2.8) | 2 (6.1) | .504 |
Immunosupressants | 1 (2.8) | 0 (0.0) | .338 |
. | COVID-19 . | Controls . | P Valuea . |
---|---|---|---|
No. of patients | 36 | 33 | |
Age, y, median (IQR) | 73.69 (18.80) | 65.01 (14.56) | .008 |
Sex: male, No. (%) | 21 (58.3) | 21 (63.6) | .652 |
Comorbidities, No. (%) | |||
Smoker | 2 (5.6) | 5 (15.5) | .187 |
Alcoholism | 1 (2.8) | 2 (6.1) | .504 |
Cardiopathy | 9 (25.0) | 7 (21.2) | .710 |
Diabetes | 8 (22.2) | 4 (12.1) | .269 |
Arterial hypertension | 22 (61.1) | 16 (48.5) | .292 |
Obesity | 7 (19.4) | 2 (6.1) | .099 |
Chronic obstructive pulmonary disease | 5 (13.9) | 3 (9.1) | .534 |
Asthma | 1 (2.8) | 1 (3.0) | .950 |
Chronic kidney disease | 0 (0.0) | 3 (9.1) | .064 |
Biochemistry, median (IQR) | |||
Creatinine, mg/dL | 0.85 (0.432) | 0.645 (0.389) | .060 |
Leucocytes/μL | 8030 (3233) | 7231 (1818) | .465 |
Linfocytes/μL | 1200 (872) | 2321 (1035) | <.001 |
Neutrophils/μL | 6430 (3209) | 4020 (1617) | <.001 |
Platelets/μL | 229 030 (109 932) | 241 406 (60 384) | .267 |
Triglycerids, mg/dL | 135.20 (48.06) | 106.47 (48.9) | .018 |
Cholesterol, mg/dL | 135.8 (37.96) | 173.13 (30.8) | <.001 |
Glucose, mg/dL | 128.8 (60.17) | 101.9 (15.3) | .272 |
Creatinine, mg/dL | 1.21 (1.096) | 1.0 (0.6) | .633 |
Bilirubin, mg/dL | 0.60 (0.50) | 0.60 (0.30) | .751 |
Hospitalization variables | |||
Temperature, °C, median (IQR) | 36.71 (0.81) | 36.67 (0.75) | .953 |
Exitus, No. (%) | 12 (33.3) | 0 (0.0) | <.001 |
Length of hospital stay, d, median (IQR) | 9.5 (18) | 0 (0) | <.001 |
Therapy, No. (%) | |||
Thyroid hormones | 0 (0.0) | 4 (12.1) | .031 |
Statins | 7 (19.4) | 11 (33.3) | .189 |
β-blockers | 1 (2.8) | 2 (6.1) | .504 |
Immunosupressants | 1 (2.8) | 0 (0.0) | .338 |
aBold indicates P < .05.
Plasma-Derived EV miRNA Profiles Between Patients With Severe COVID-19 and Controls
The PLS-DA revealed a clear classification between the COVID-19 and control groups based on their miRNome expression (Supplementary Figure 1). Furthermore, DESeq2 analysis identified 50 SDE miRNAs (Figure 1A): 29 were upregulated and 21 were downregulated. The strongest upregulated miRNAs in these patients were hsa-miR-658 and hsa-miR-4800-5p, which showed log2 FCs of 3.54 and 3.10, respectively. On the contrary, hsa-miR-5192 was the strongest downregulated miRNA (log2 FC, −22.67). The complete list of SDE miRNAs is presented in Supplementary Table 1, and the heat map for the SDE genes is represented in Figure 1B.
![A and B, Volcano plots and heat map of the SDE miRNAs for COVID-19 vs controls in exosomes. C and D, Gene Ontology enrichment for cellular processes and KEGG molecular pathways for the SDE miRNAs found in exosomes. In the volcano plot, vertical dashed lines indicate the threshold value for absolute fold change ≥1.5, and the horizontal dashed line indicates the threshold value for the false discovery rate–adjusted P value ≤.05. Gray dots represent the miRNAs that are below the threshold, and red and green dots indicate the miRNAs that are above the threshold. In the heat map, study participants are represented in columns and SDE miRNAs in rows, with clustering dendrograms on the left for miRNAs and at the top for samples. The color scale shows the relative expression level of SDE miRNAs. In the pathway enrichment figures, the dots represent the number of genes for each pathway or cellular process, while the color depends on the q value. The P value and the Q value for the analysis were set to 0.05. SDE, statistically differential expressed.](https://cdn.statically.io/img/oup.silverchair-cdn.com/oup/backfile/Content_public/Journal/jid/PAP/10.1093_infdis_jiae310/2/m_jiae310f1.jpeg?Expires=1724833507&Signature=ZdzLEEXhouO-SmdMG8mA51bea1pUt-aadVqRlyIi4xE5IlRHALoha~3fSbjbofP60CSVw1dTmw~Ik9276OhMQ32jgqJLBPHZ2WIvrsRudhYmwYpPsau9PzsujVKiFbsyy4eS4WqLQd1RGuPzsqH2QRXWky67Z55m23HkYXTy3oCrm3uYiKd5dGafeaQqInqDcJFa1-oEojkQRRuhw3dwC25hxZLX0D5HPucFZhdPFDUAJ0iC15UyahyP82Pg9QfG6O-h5LdLDZQN3SvlP~147rHnkScfJUZamyd3VujCof5jMmReQsYhYvKzZ5XfM2m8h-tIoiqx421~eZwkUq4qWA__&Key-Pair-Id=APKAIE5G5CRDK6RD3PGA)
A and B, Volcano plots and heat map of the SDE miRNAs for COVID-19 vs controls in exosomes. C and D, Gene Ontology enrichment for cellular processes and KEGG molecular pathways for the SDE miRNAs found in exosomes. In the volcano plot, vertical dashed lines indicate the threshold value for absolute fold change ≥1.5, and the horizontal dashed line indicates the threshold value for the false discovery rate–adjusted P value ≤.05. Gray dots represent the miRNAs that are below the threshold, and red and green dots indicate the miRNAs that are above the threshold. In the heat map, study participants are represented in columns and SDE miRNAs in rows, with clustering dendrograms on the left for miRNAs and at the top for samples. The color scale shows the relative expression level of SDE miRNAs. In the pathway enrichment figures, the dots represent the number of genes for each pathway or cellular process, while the color depends on the q value. The P value and the Q value for the analysis were set to 0.05. SDE, statistically differential expressed.
The functional analysis of SDE miRNAs in plasma-derived exosomes reported 44 KEGG biological pathways and 1354 GO cellular processes. The top 10 pathways are represented in Figure 1C and 1D. The most relevant GO cellular processes in the EVs of COVID-19 cases were cellular component biogenesis, cell adhesion, cellular component disassembly, and viral process. The most relevant pathways for the miRNA targets showed enrichment for MAPK, ErbB, T-cell receptor, neurotrophin signaling pathways, and focal adhesion.
Plasma miRNA Profile Between Patients With Severe COVID-19 and Controls
We analyzed the miRNA profile of patients with severe COVID-19 vs the control group, as this analysis has not been addressed previously [11]. A total of 115 SDE miRNAs were identified in COVID-19 cases: 96 upregulated and 19 downregulated (Figure 2A). Of them, hsa-miR-20b-3p and hsa-miR-1275 were the strongest up- and downregulated miRNAs, respectively. The values of the SDE miRNAs in plasma are shown in Supplementary Table 2, and the heat map for the SDE genes is shown in Figure 2B. The functional analysis of SDE miRNAs in plasma reported 1354 GO cellular processes and 44 KEGG biological pathways. The top 10 pathways are shown in Figure 2D and 2E. The most enriched GO cellular processes for the plasma miRNAs were cellular component biogenesis, cell adhesion, and histone modification. The most relevant pathways for the miRNA targets showed enrichment for MAPK, ErbB, T-cell receptor, neurotrophin signaling pathways, and focal adhesion.
![Volcano plots and heat map of the SDE miRNAs for COVID-19 vs healthy controls in plasma A and B. Venn diagram for the common SDE miRNAs in exosomes and plasma in patients with severe COVID-19 C. Gene Ontology enrichment for cellular processes and KEGG molecular pathways for the SDE miRNAs in plasma (D and E) and common to both (F and G). In the volcano plot, vertical dashed lines indicate the threshold value for absolute fold change ≥1.5, and the horizontal dashed line indicates the threshold value for false discovery rate–adjusted P value ≤.05. Gray dots represent the miRNAs that are below the threshold, and red and green dots indicate the miRNAs that are above the threshold. In the heat map, study participants are represented in columns and SDE miRNAs in rows, with clustering dendrograms on the left for miRNAs and at the top for samples. The color scale shows the relative expression level of SDE miRNAs. In the pathway enrichment figures, the dots represent the number of genes represented for each pathway or cellular process, while the color depends on the q value. The P value and the Q value for the analysis were set to 0.05. EV, extracellular vesicle; SDE, statistically differential expressed.](https://cdn.statically.io/img/oup.silverchair-cdn.com/oup/backfile/Content_public/Journal/jid/PAP/10.1093_infdis_jiae310/2/m_jiae310f2.jpeg?Expires=1724833507&Signature=x0b8jovEfxuoW5pa7Upw26u3Yzu3VrFDKssWehmvKjd75U8kMFJo9StmvSTQA78xUbV6d3~w5--OnzgJx6jOWlMpLbaZN~yIzEDu-VXoSq6YnI-rc2EQkKBwCrwYS3bM8JdgVPekX6auyVS~gP73Khm7R0T8~sCAxR2XX3yN5ZSO8RsgClG1B5-OfVclwgxq81VtXbxEbDOeu~MJZZye69uiYlE4twGkEF9kl38vW6NZl4GZIC0J7jP-cnIfg3WNQbYAXGPpaBwtjV1Rhl74V45rwJXsbLdokIRfhQdy4QsYIrpkkc78j9VeflP8ca9jQjsc~BOGD3HudF~v7938zA__&Key-Pair-Id=APKAIE5G5CRDK6RD3PGA)
Volcano plots and heat map of the SDE miRNAs for COVID-19 vs healthy controls in plasma A and B. Venn diagram for the common SDE miRNAs in exosomes and plasma in patients with severe COVID-19 C. Gene Ontology enrichment for cellular processes and KEGG molecular pathways for the SDE miRNAs in plasma (D and E) and common to both (F and G). In the volcano plot, vertical dashed lines indicate the threshold value for absolute fold change ≥1.5, and the horizontal dashed line indicates the threshold value for false discovery rate–adjusted P value ≤.05. Gray dots represent the miRNAs that are below the threshold, and red and green dots indicate the miRNAs that are above the threshold. In the heat map, study participants are represented in columns and SDE miRNAs in rows, with clustering dendrograms on the left for miRNAs and at the top for samples. The color scale shows the relative expression level of SDE miRNAs. In the pathway enrichment figures, the dots represent the number of genes represented for each pathway or cellular process, while the color depends on the q value. The P value and the Q value for the analysis were set to 0.05. EV, extracellular vesicle; SDE, statistically differential expressed.
Plasma-Derived EVs vs Plasma miRNA Profiles
Next, we investigated the common SDE miRNAs shared between the 50 SDE plasma-derived EV miRNAs and the 115 SDE plasma miRNAs identified between patients with severe COVID-19 and controls. A total of 15 SDE miRNAs were in common: hsa-miR-7-5p, hsa-miR-22-3p, hsa-miR-99b-5p, hsa-miR-139-5p, hsa-miR-150-5p, hsa-221-3p, hsa-miR-320b, hsa-miR-320c, hsa-miR-320d, hsa-miR-363-5p, hsa-miR-423-5p, hsa-miR-425-3p, hsa-miR-486-5p, hsa-miR-574-5p, and hsa-miR-1246 (Venn diagram, Figure 2C).
The functional analysis of these 15 SDE common miRNAs reported 21 KEGG biological pathways and 545 GO cellular processes. The most relevant enriched GO processes for these miRNAs were related to catabolic processes such as ubiquitination, neuron death/apoptotic process, and 1-κB kinase/NF-κB signaling process. The KEGG pathways were related to pathways in cancer, MAPK, ErbB, and mTOR signaling. The top 10 pathways are shown in Figure 2F and 2G.
Mortality at 90 Days and Survival Analysis
Patient characteristics according to mortality at 90 days are shown in Supplementary Table 3. We did not find epidemiologic and clinical differences between alive and deceased groups. Next, we explored the predictive capability of the miRNAs from the plasma-derived EVs by using LASSO Cox, which identified 10 miRNAs as putative predictors of mortality at 90 days when adjusted by age and gender (Supplementary Figure 2): hsa-miR-450a-2-3p, hsa-miR-4730, hsa-miR-6855-5p, hsa-miR-6890-5p, hsa-miR-199b-5p, hsa-miR-18a-3p, hsa-miR-6124, hsa-miR-1469, hsa-miR-3613-5p, and hsa-miR-320a-3p. Among those, we selected hsa-miR-1469 and hsa-miR-6124 due to their higher predictive power as calculated by areas under the receiver operating characteristic curve (>0.85). All the results obtained are summarized in Table 2.
. | ß Coefficient . | HR . | P Value . | AUCa . |
---|---|---|---|---|
hsa-miR-4730 | 3.44 | 31.10 | .026 | 0.50 |
hsa-miR-18a-3p | 0.24 | 1.28 | .006 | 0.66 |
hsa-miR-199b-5p | 0.11 | 1.12 | .010 | 0.69 |
hsa-miR-450a-2-3p | 0.29 | 1.34 | .018 | 0.75 |
hsa-miR-6855-5p | 0.18 | 1.19 | .104 | 0.75 |
hsa-miR-6890-5p | 0.74 | 2.10 | .001 | 0.75 |
hsa-miR-3613-5p | −0.02 | 0.98 | .116 | 0.75 |
hsa-miR-320a-3p | 0.01 | 1.00 | .101 | 0.81 |
hsa-miR-6124 | 0.05 | 1.05 | .004 | 0.88 |
hsa-miR-1469 | 0.14 | 1.15 | .016 | 0.91 |
. | ß Coefficient . | HR . | P Value . | AUCa . |
---|---|---|---|---|
hsa-miR-4730 | 3.44 | 31.10 | .026 | 0.50 |
hsa-miR-18a-3p | 0.24 | 1.28 | .006 | 0.66 |
hsa-miR-199b-5p | 0.11 | 1.12 | .010 | 0.69 |
hsa-miR-450a-2-3p | 0.29 | 1.34 | .018 | 0.75 |
hsa-miR-6855-5p | 0.18 | 1.19 | .104 | 0.75 |
hsa-miR-6890-5p | 0.74 | 2.10 | .001 | 0.75 |
hsa-miR-3613-5p | −0.02 | 0.98 | .116 | 0.75 |
hsa-miR-320a-3p | 0.01 | 1.00 | .101 | 0.81 |
hsa-miR-6124 | 0.05 | 1.05 | .004 | 0.88 |
hsa-miR-1469 | 0.14 | 1.15 | .016 | 0.91 |
Abbreviations: AUC, area under the curve; HR, hazard ratio; LASSO, least absolute shrinkage and selection operator; miRNA, microRNA.
aAs obtained from the receiver operating characteristic curve.
. | ß Coefficient . | HR . | P Value . | AUCa . |
---|---|---|---|---|
hsa-miR-4730 | 3.44 | 31.10 | .026 | 0.50 |
hsa-miR-18a-3p | 0.24 | 1.28 | .006 | 0.66 |
hsa-miR-199b-5p | 0.11 | 1.12 | .010 | 0.69 |
hsa-miR-450a-2-3p | 0.29 | 1.34 | .018 | 0.75 |
hsa-miR-6855-5p | 0.18 | 1.19 | .104 | 0.75 |
hsa-miR-6890-5p | 0.74 | 2.10 | .001 | 0.75 |
hsa-miR-3613-5p | −0.02 | 0.98 | .116 | 0.75 |
hsa-miR-320a-3p | 0.01 | 1.00 | .101 | 0.81 |
hsa-miR-6124 | 0.05 | 1.05 | .004 | 0.88 |
hsa-miR-1469 | 0.14 | 1.15 | .016 | 0.91 |
. | ß Coefficient . | HR . | P Value . | AUCa . |
---|---|---|---|---|
hsa-miR-4730 | 3.44 | 31.10 | .026 | 0.50 |
hsa-miR-18a-3p | 0.24 | 1.28 | .006 | 0.66 |
hsa-miR-199b-5p | 0.11 | 1.12 | .010 | 0.69 |
hsa-miR-450a-2-3p | 0.29 | 1.34 | .018 | 0.75 |
hsa-miR-6855-5p | 0.18 | 1.19 | .104 | 0.75 |
hsa-miR-6890-5p | 0.74 | 2.10 | .001 | 0.75 |
hsa-miR-3613-5p | −0.02 | 0.98 | .116 | 0.75 |
hsa-miR-320a-3p | 0.01 | 1.00 | .101 | 0.81 |
hsa-miR-6124 | 0.05 | 1.05 | .004 | 0.88 |
hsa-miR-1469 | 0.14 | 1.15 | .016 | 0.91 |
Abbreviations: AUC, area under the curve; HR, hazard ratio; LASSO, least absolute shrinkage and selection operator; miRNA, microRNA.
aAs obtained from the receiver operating characteristic curve.
The expression levels of the miRNAs were classified as high and low based on their expression, as calculated by the surv_cutpoint function of the survminer R package. Using the Kaplan-Meier estimator, we observed that high expression levels of the 2 miRNAs (above the calculated threshold for each) were associated with a lower probability of survival at 90 days (represented as Kaplan-Meier curves, Figure 3A and 3B). The hazard ratio and statistical significance of the dichotomous variable high/low for each miRNA, as well as age and gender, are presented in Figure 3C. In addition, we evaluated the predictive power of miR-1469 and miR-6124 by combining their expression values with age and gender, reaching an area under the curve of 0.938, more than the predictive power of the individual miRNAs whether adjusted by gender and age or not (Supplementary Table 4). Neither miR-1469 nor miR-6124 was SDE in plasma.
![A and B, Kaplan-Meier curves for miR-1496 and miR-6124 in the first 90 days from admission and sampling. C, Hazard ratio forest plot for the multivariate analysis. These results show the prognosis of COVID-19 associated with the expression levels of both miRNAs, each of which is classified between high and low, as calculated by survminer R package.](https://cdn.statically.io/img/oup.silverchair-cdn.com/oup/backfile/Content_public/Journal/jid/PAP/10.1093_infdis_jiae310/2/m_jiae310f3.jpeg?Expires=1724833507&Signature=i5LVri~TUTv7OtkAARWrpXujTMXrACQS~aAvSOQLJZXHf-G17KVO7P8BJUuYKC94jS9SaQzl40qiGm6y3W5hwjexOByIgx2C6mXqiG8KP3~9efZr-~-CKP6g-kWgn3g1xfuQZXs3xCw5ugIbwB6xseXqiG4BfL0M-HAo5-5kKwgu30gC6muAis8hmdeDCYMYw4rPZmtSGtp-9enN73wcl3kYdVwyviSDTChpQUx4c5oATvgwjibXs18HvtxP35FBYBWtJQyphOdsdFebW28qPN6aBMFt9A3Z-LLIC~vPXOiYm3Ua5cTVuPYj7Xa4CCQ~T7ZKgFVol4baWMkA8UM5Tw__&Key-Pair-Id=APKAIE5G5CRDK6RD3PGA)
A and B, Kaplan-Meier curves for miR-1496 and miR-6124 in the first 90 days from admission and sampling. C, Hazard ratio forest plot for the multivariate analysis. These results show the prognosis of COVID-19 associated with the expression levels of both miRNAs, each of which is classified between high and low, as calculated by survminer R package.
DISCUSSION
In this study, we found that SARS-CoV-2 infection alters the miRNA profile in plasma-derived EVs of patients with severe COVID-19, revealing 50 SDE miRNAs. Indeed, 15 were in the plasma of a cohort with severe COVID-19. We also found that the combination of the miRNA expression levels in plasma-derived EVs of hsa-miR-1469 and hsa-miR-6124, adjusted by age and gender, predicted mortality with excellent accuracy.
The characterization of the miRNome in plasma-derived EVs of patients with severe COVID-19 showed 50 miRNAs to be SDE with respect to controls. We found hsa-miR-658, hsa-miR-4800-5p, and hsa-1249-5p to be the most upregulated miRNAs, with a FC >24 as compared with controls. hsa-miR-658 is upregulated in a range of diseases and upregulates the hepatocyte growth factor (MET) in lung cancer [14]. Other upregulated miRNAs have been related to other viral infections, such as 4800-5p and hsa-1249-5p. miR-4800-5p is upregulated in the plasma of patients with hepatic fibrosis after hepatitis B infection [15] and with sepsis secondary to pneumonia [16]. hsa-miR-4800-5p has been associated with the inflammatory response and dysfunction caused by influenza A virus pneumonia [17]. On the contrary, hsa-miR-5192 was highly downregulated. This miRNA has been associated with idiopathic pulmonary fibrosis [18] and exacerbation of asthma [19]. The strong changes observed in all these miRNAs should be studied further since their specific involvement in viral infections is unknown.
The targets of the SDE miRNAs in the plasma-derived EVs of patients with severe COVID-19 are involved in cellular processes related to the viral process and the positive regulation of cell adhesion and cellular component biogenesis. We should keep in mind that viruses hijack the exosomal cell machinery and that the surface adhesion molecules of exosomes play a vital role in their functional end point by acting as effectors or by affecting cellular uptake, milieu, and cell interaction, as well as immune recognition [20]. Previous findings have correlated blood levels of cell adhesion molecules with severity in COVID-19 disease [21]. Additionally, virus-induced cellular component disassembly has been described in these patients as potentially contributing to the dysregulation of proinflammatory molecules observed in severe cases [22]. Accordingly, the most relevant biological pathways dysregulated in these patients involved focal adhesion and enrichment of MAPK, ErbB, neurotrophin, and T-cell receptor signaling pathways. These pathways are interrelated and regulate cellular proliferation, migration, differentiation, inflammation, and survival. Also, they can be activated by cytokines typically found in SARS-CoV-2 infection or after stress signals such as hypoxia [23].
Plasma-derived EV miRNAs are more stable and reliable as biomarkers, as they are protected from endogenous RNase degradation by a lipid bilayer membrane [24]. Plasma also contains free miRNAs, but few studies address the similarity in miRNA profile between plasma and plasma-derived EVs. In healthy people, a previous study found that the miRNA content in plasma-derived EVs and plasma was similar [25]. However, Endzeliņs et al reported differences when comparing the diagnostic potential of miRNAs in plasma-derived EVs and plasma in patients with cancer. They described that only a small fraction of the plasma miRNAs was recovered from the plasma-derived EVs, concluding that, for some miRNAs, EVs provide a more consistent source of RNA than plasma, while for others, a better diagnostic performance is obtained when they are tested in plasma [26]. Therefore, due to the limited data available, we decided to compare the SDE miRNA profile in both compartments in patients with severe COVID-19. Thus, comparing both compartments could provide insights into the miRNA-specific compartmentalization during SARS-CoV-2 infection. In this regard, we found 15 SDE miRNAs in common between plasma (115 SDE) and plasma-derived exosomes (50 SDE) of patients with severe COVID-19: 11 upregulated, 2 downregulated, and 2 discordant. One of the discordant miRNAs is hsa-miR-574-5p (upregulated in plasma-derived EVs but downregulated in plasma), which is upregulated in mice with acute respiratory distress syndrome [27]. It has also been described as a putative regulator of ACE2 in male patients with COVID-19 [28]. hsa-miR-99b-5p, however (downregulated in plasma-derived EVs and upregulated in plasma), is upregulated in Mycobacterium-infected cells in murine models, being a putative modulator of the immune response [29]. Although all EV miRNAs are present in plasma samples, the fact of being more diluted in plasma may cause them not to be identified as SDE. Several upregulated miRNAs in plasma-derived EVs and plasma in this study have been previously related to dysregulation of the inflammatory response and COVID-19 severity. This is the case of the hsa-miR-320 family, hsa-miR-7-5p, hsa-miR-486-5p, and hsa-miR-221-3p, which have been correlated with modulation of the inflammatory response and tissue injury biomarkers [30–33]. hsa-miR-1246 has been identified as a regulator of ACE2 [34], and it has been defined as a biomarker for severity and pneumonia-related sepsis [30, 31]. hsa-miR-221-3p is also involved in ACE2-dependent shedding through the ADAM17 receptor, key in COVID-19 endothelial injury and coagulopathy in lung tissues [32, 33]. Interestingly, increased expression of hsa-miR-423-5p and hsa-miR-221-3p correlates with heart failure and cardiac injury in viral myocarditis [35, 36]. This miRNA has been described as a biomarker of SARS-CoV-2 infection [37] and a predictor of COVID-19 mortality [11]. In contrast, we found 2 downregulated miRNAs in plasma-derived EVs and plasma: hsa-miR-150-5p and hsa-miR-139-5p. The first inhibits expression of the viral structural protein Nsp10, and its downregulation has been suggested to promote viral replication and disease severity [38]. In addition, hsa-miR-150-5p is a key regulator of the immune response in influenza infections [39], modulates B-cell and NK development [40], and is associated with a longer ICU stay in COVID-19 cases [41]. However, hsa-miR-139-5p negatively regulates endothelial CXCR4 (C-X-C motif chemokine receptor 4), which plays a key role in coordinating innate and adaptative immune responses, functioning as a T-cell costimulatory. CXCL12 (C-X-C motif chemokine ligand 12), the sole ligand of CXCR4, is highly expressed in inflamed tissues and exacerbates inflammation and immune responses [42]. Severe COVID-19 cases have been described to present increased CXCL12 levels [43].
In this regard, the functional enrichment results of the predicted targets of the common SDE miRNAs between the groups suggest strong involvement of these miRNAs in biological processes such as ubiquitination and neuronal death. Ubiquitination has been described to play a key role in the innate immune response to SARS-CoV-2 infection by inducing antiviral defenses but also facilitating virus replication [44]. Neuronal death has been described to occur in patients with COVID-19 due to glial-mediated neuroinflammation after oxidative stress [45]. Since all cell types produce EVs, the analysis of plasma-derived EVs could provide helpful insight into the whole organism's molecular state, which could help to correlate the different responses of organs to a specific disease. Finally, the most relevant KEGG molecular pathways involved MAPK signaling pathways and cancer-related pathways, which have been described to play a key role in COVID-19 severity of infection and immune dysregulation [46].
COVID-19 mortality depends on many factors [47]. Our study found that the combination of 2 miRNAs (hsa-miR-1469 and hsa-miR-6124) within plasma-derived EV miRNAs can be used as mortality predictors. Of note, these miRNAs were not detected in plasma. This could be related to a different distribution of miRNAs in plasma and EVs, as plasma miRNAs are mainly derived from blood cells, while EV composition varies depending on the cell status and tissue from which it is produced. Regarding previous literature, hsa-miR-1469 has been associated with many cancer types, and particularly high levels have been associated with cell apoptosis in lung cancer [48]. Moreover, hsa-miR-6124 has been proposed as a biomarker of stroke and lung cancer [49, 50]. Combining the expressions of both miRNAs could be used as an early mortality risk predictor to improve the treatment of patients with higher death risk at ICU admission.
Finally, several considerations should be considered to interpret our data correctly. First, this was an observational study. Second, the sample size was low. Third, the miRNAs identified as mortality predictors have not been detected in plasma samples; therefore, their role in these EVs should be further evaluated.
CONCLUSIONS
Our study has identified significant changes in the miRNA profile within plasma-derived EVs of patients with severe COVID-19, revealing potential insights into disease mechanisms. Of them, 15 miRNAs were SDE in plasma and plasma-derived EVs. In addition, the combination of the miRNA expression levels in plasma-derived EVs of hsa-miR-1469 and hsa-miR-6124, with age and gender, had excellent accuracy in predicting mortality. These findings illuminate the role of miRNAs in patients with severe COVID-19 and warrant further investigation into their mechanistic involvement and clinical applications.
Supplementary Data
Supplementary materials are available at The Journal of Infectious Diseases online (http://jid.oxfordjournals.org/). Supplementary materials consist of data provided by the author that are published to benefit the reader. The posted materials are not copyedited. The contents of all supplementary data are the sole responsibility of the authors. Questions or messages regarding errors should be addressed to the author.
Notes
Acknowledgments. This study was possible thanks to the collaboration of all the patients and their relatives, as well as the medical and nursery staff of the participating clinical service for their continuous support. We thank the data managers who took part in the project.
Author contributions. Funding acquisition: H. G.-B., M. H.-R., E. G.-S., E. T. Conceptualization: A. G.-C., L. S.-D. P., A. T.-V., M. A. J.-S., A. F.-R., S. R., H. G.-B., D. B., J. M. E., I. S.-M., E. T. Patient selection and clinical data acquisition: A. T.-V., M. T. P., I. M. A., M. B.-C., P. R.-G., R. L.-H., M. H.-R., E. G.-S., E. T. Sample processing and RNA extraction: O. G.-G., A. M.-J., A. G.-C. Formal analysis: A. G.-C., L. S.-D. P., M. A. J.-S., A. F.-R. Writing–original draft preparation: L. S.-D. P., A. G.-C., A. T.-V., M. M.-F., T. A., M. A. J.-S., A. F.-R. Writing–review and editing: L. S.-D. P., A. G.-C., A. T.-V., M. M.-F., O. G.-G., A. M.-J., T. A., M. A. J.-S., S. R., A. F.-R., D. B., I. S.-M., J. M. E., E. G.-S., M. T. P., I. M. A., M. B.-C., P. R.-G., R. L.-H., M. H.-R., H. G.-B., E. T. Supervision and visualization: L. S.-D. P., A. G.-C., A. T.-V., M. M.-F., M. A. J.-S., A. F.-R., I. S.-M., J. M. E., S. R., H. G.-B., D. B., M. H.-R., E. G.-S., E. T.
Availability of data and materials. Sequences from EV-derived miRNA samples can be accessed through BioStudies with accession number S-BSST1192, while sequences from plasma-derived miRNAs can be accessed through accession number E-MTAB-10562.
Ethics approval and consent to participate. The study protocol was approved by the hospital’s Clinical Ethics Committee, and informed written consent was obtained from patients or legal representatives before recruitment (PI-20-1717). This research was performed according to the Declaration of Helsinki’s ethics code.
Financial support. This work was supported by the Instituto de Salud Carlos III (COV20/00491, PI18/01238, and COV20/1144); Centro de Investigación Biomédica en Red en Enfermedades Infecciosas, Instituto deSalud Carlos III (CB21/13/00051 and CB21/13/00044); Junta de Castilla y León (VA321P18, GRS 1922/A/19, GRS 2057/A/19, CCVC8485); Consejería de Educación de Castilla y León (VA256P20); Fundación Ramón Areces (CIVP19A5953); ISCIII (CP17CIII/00007 to M. A. J.-S. as a Miguel Servet researcher); and CSIC's Global Health Platform (PTI Salud Global; D. B.).
References
World Health Organisation. WHO coronavirus (COVID-19) dashboard. 2023. Available at: https://covid19.who.int/. Accessed 10 May 2023.
Centers for Disease Control and Prevention. Underlying medical conditions associated with higher risk for severe COVID-19: information for healthcare professionals. 2024. Available at: https://www.cdc.gov/coronavirus/2019-ncov/hcp/clinical-care/underlyingconditions.html. Accessed 1 May 2024.
Author notes
L. S.-D. P., A. G.-C., E. G.-S., and E. T. contributed equally to the study.
Potential conflicts of interest. All authors: No reported conflicts.
All authors have submitted the ICMJE Form for Disclosure of Potential Conflicts of Interest. Conflicts that the editors consider relevant to the content of the manuscript have been disclosed.