Biomarkers predictive of COVID-19 prognosis identified in Bangladeshi patients and validated in Japanese cohorts

Topics

COVID-19 patients in Bangladesh

Patients clinically suspected of SARS-CoV-2 infection who visited Evercare Hospital, Dhaka between December 25, 2021 and September 21, 2022 were considered for this study. The study was approved by the Ethical Practice Committee of Evercare Hospital, Dhaka (ERC approval number 33/2022-01) and the Research Ethics Committee of the Institute of Microbial Diseases Research from Osaka University, Japan (No. 2021-3).

Among the considered patients, 129 patients confirmed with COVID-19 within 7 days of onset were included. Most of these cases in Bangladesh coincide with the sixth wave and part of the seventh wave in Japan. Patients were recruited at the first hospital visit, blood serum was collected, and the attending physician classified the findings as mild, moderate, or severe according to criteria established by the World Health Organization (WHO) .⁴⁷ (Table 1), at the time of discharge. Severity of illness was not determined during patients' first visit, but rather on their case sheets as they were discharged from the hospital. Laboratory-confirmed Mild cases of COVID-19 were those with one or more symptoms (e.g., fever, cough, runny nose, fatigue, headache, nausea, vomiting, diarrhea, chest pain, abdominal pain, and loss of taste or of smell), but lacked shortness of breath, dyspnea on exertion and abnormal radiological findings. Laboratory-confirmed moderate cases of COVID-19 were those with pneumonia, oxygen saturation >93%, and may have required low oxygen support. Severe cases developed COVID-19 pneumonia and required hospitalization; patients had dyspnea, respiratory rate ≥ 30 breaths/min, blood oxygen saturation ≤ 93% on room air, pulmonary infiltrates > 50% and may have required mechanical ventilation and/or supportive care intensive.

All patients went to the hospital with mild symptoms; Subsequently, 64 remained in mild condition, 46 declined to moderate condition, and 19 deteriorated to severe stage. The mean time to hospital visit from symptom onset was 2.3 ± 0.12 days for mild patients, 2.3 ± 0.14 days for moderate patients and 3.0 ± 0 .37 days (mean ± SD) for severe patients. COVID-19 was diagnosed based on PCR testing and the date of onset and vaccination status were recorded by the doctor during the patient interview. Most of the patients' laboratory results have been carefully reported (ref.⁴⁸).

COVID-19 patients in Japan

This sample consisted of 197 patients with clinical suspicion of SARS-CoV-2 infection who were admitted to Habikino Hospital and Tokushukai Hospital from late June 2020 to mid-June 2022.^49.50. All patients provided blood samples at their first visit and written informed consent and the study was approved by the Ethics Committee of Osaka Habikino Medical Center (Approved ID: 150-7), Hospital Tokushukai (TGE01547) and the Louis Pasteur Center for Medical Research (LPC). .29). This study followed the principles of the Declaration of Helsinki and was approved by the Institutional Review Board of Osaka University Hospital (No-885). Data on healthy subjects were obtained from the Louis Pasteur Medical Research Center (LPC.8 and LPC.25).

In Japan, patients' illness severity was determined upon hospital admission according to the guidelines for medical treatment of COVID-19 (https://www-mhlw-go-jp/content/000785119 -pdf). COVID-19 is classified as mild, moderate I, moderate II and severe. However, for comparison with Bangladesh, moderate II and severe cases were combined and considered severe. The severity of the illness here refers to the final course of the patient and not to their condition at the time of admission to hospital. Patient distribution, severity classification and age distribution are shown in Table 1. As shown in Table 1, 197 untreated COVID-19 patients who visited Habikino and Tokushukai Hospitals within 7 days of onset were included. The date of onset was determined by the physician based on PCR test results and patient interviews. The mean number of days to onset was mild: 3.08 ± 0.32, moderate: 3.24 ± 0.45, and severe: 3.71 ± 0.21 days. Ninety-one healthy Japanese subjects, with a mean age of 63.6 ± 1.9 years, were also included.

Soluble cytokine/chemokine/receptor assay

Cytokines, chemokines, and soluble receptors were quantified using the Bio-Plex 200, a multiplex cytokine array system (Bio-Rad Laboratories, CA, USA) according to the manufacturer's instructions. Blood sera from healthy subjects and COVID-19 patients were collected and centrifuged at 1600 g for 10 min. Serum samples were frozen at −80°C until analysis. We simultaneously quantified cytokines, chemokines and soluble receptors. The Bio-Plex Human Cytokine 48-Plex panel and the Inflammation panel (Bio-Rad Laboratories, CA, USA) were used to simultaneously quantify 78 elements: CTACK, Eotaxin, FGF basic, G-CSF, GM-CSF, GRO- α, HGF, IFN-α2, IFN-γ, IL-1α, IL-1β, IL-1ra, IL-2, IL-2Rα, IL-3, IL-4, IL-5, IL-6, IL- 7, IL-8, IL-9, IL-10, IL-12(p40), IL-12(p70), IL-13, IL-15, IL-16, IL-17, IL-18, IP- 10, LIF, MCP-1(MCAF), MCP-3, M-CSF, MIF, MIG, MIP-1α, MIP-1β, β-NGF, PDGF-BB, RANTES, SCF, SCGF-β, SDF-1α , TNF -α, TNF-β, TRAIL, VEGF) and inflammation panel (37 plex: APRIL, BAFF, CD30, CD163, Chitinase, sgp130, IFN-α 2, IFN-β, IFN-γ, sIL-6Ra, IL -10, IL-11, IL-12 (p40), IL-12 (p70), IL-19, IL-20, IL-22, IL-26, IL-27, IL-28A, IL-29, IL -32, IL-34, IL-35, LIGHT, MMP-1, MMP-2, MMP-3, Osteocalcin, Osteopontin, Pentraxin-3, sTNF-R1, sTNF-R2, TSLP, TWEAK The following have been excluded panel for data analysis due to overlap with 48-plex: IFN-α2, IFN-γ, IL-2, IL-8, IL-10, IL-12 (p40) and IL- 12 (p70). were quantified several times over a 2-year period. As there were batch-to-batch and measurement-to-measurement errors, these data were corrected based on values ​​from healthy subjects. The Bangladeshi samples were measured simultaneously with healthy Japanese subjects as a control, and the values ​​from the healthy subjects were used as a reference to correct batch-to-batch and between-measurement errors; this was then used to make comparisons with serum samples from patients in Japan. Biomarkers with significant inter-kit errors and low measurement sensitivity were excluded from the analysis: IL-19, IL-20, IL-26, IL-28A, IL-29, IL-35, LIGHT.

statistical analyzes

The distribution of soluble cytokine/chemokine/receptor values ​​in healthy controls was analyzed to determine whether raw values ​​or log-transformed values ​​were more normally distributed. All parameters had log-transformed values ​​that were more normally distributed (data not shown), and these were therefore used in our analysis.⁴⁵. The t-test results for the data used in Bangladesh and Japan, categorized by mild, moderate, and severe disease, are shown in Figure 2. To correct for error between measurements, Japanese healthy controls were used as reference and adjusted accordingly. To allow multiple comparisons, the p-the values ​​were corrected by Holm using the p.adjust function of the R language stats package.

ANOVA was performed and quantitative data were presented as mean ± SEM. The significance of the difference between groups was evaluated using the Dunnett test with a value of p< 0.05 considered significant. All statistical analyzes were performed with JMP 20.0 statistical software (JMP Statistical Discovery LLC, NC, USA).

The aim of the study was to use soluble cytokine/chemokine/receptor data, collected within 7 days of COVID onset, to predict whether a patient would subsequently deteriorate to a moderate disease state or worse and would require hospitalization. To achieve this, we used a binary logistic regression model. To refine the predictors, we used Least Absolute Shrinkage and Selection Operator (LASSO) regression to select relevant cytokines as candidate markers.²². Selection of the optimal number of variables was guided by Leave-One-Out cross-validation (LOO CV). LOO-CV is known to be overtrained, however, in this case correct generalization performance could be assessed using validation data, without the need to take into account variations due to partitioning (such as factor k).

Due to the disproportionate distribution of disease severities among Bangladeshi COVID-19 patients, we applied weights to the LASSO logistic regression model to minimize potential bias in the data. In accordance with the WHO severity classification, groups that did not require hospitalization were defined as those with mild illness and recovered, while groups that were hospitalized and required medical treatment were defined as moderate and severe. Additionally, healthy individuals were added to the “no hospitalization group” so that the model could correctly predict that hospitalization was not necessary. Here, the number of people in each group is clinically known. Markers characteristic of the smallest group, particularly the most seriously ill patients, may be buried. Therefore, we determined the ratio of the number of people in each distribution and weighted the objective variable by its inverse (the glmnet function from the glmnet package has a weight argument, which can be used to assign weights to the objective variable) .

The performance of the logistic regression model was evaluated using training and validation datasets. This included determining the area under the curve (AUC) from the receiver operating characteristic (ROC) curve and calculating performance measures such as specificity and sensitivity. All analyzes were performed using the R language v4.2 (https://www.r-project.org/), with the glmnet 4.1-7 package supporting variable selection and analysis of logistic regression. Version 1.18.4 of the pROC package was used to evaluate the model performance via ROC curve and AUC.