Cytokine Storm among Bangladeshi adults with COVID-19: A prospective cohort study Authors: Monira Sarmin, Fhameda Akter, Anika Nawrin Islam, Mustafa Mahfuz, Subhasish Das, Sharifuzzaman, S.M. Tafsir Hasan, Taufiqur Rahman Bhuiyan, Mustafizur Rahman, Amran Gazi, Fariha Bushra Matin, Md Tariqujjaman, Lubaba Shahrin, Mofakharul Islam, Asif Mujtaba Mahmud, Sayera Banu, Mohammod Jobayer Chisti, Firdausi Qadri, John David Clemens, Tahmeed Ahmed Published: 2024 ABSTRACT ObjectiveIn COVID-19, cytokine storms (CS) result in higher mortality and morbidity. Our study evaluated the rate of cytokine storms among COVID-positive Bangladeshi adults.MethodsFrom October 2020 to March 2022, this cohort study enrolled both COVID-positive and COVID-negative healthy adults. COVID-positive cases were treated in a makeshift COVID unit of icdr,b Dhaka Hospital. CS was defined as having IL-6 >80 pg/mL or any three of the following: high CRP, ferritin, LDH, D-dimer, or low lymphocyte. Stored plasma samples were tested for the cytokines IL-6, IL-1β, and TNF-α.ResultsThis study involved 77 participants, 32 were in the severe-critical group, 30 were in the mild-moderate group, and 14 were COVID-negative. Twelve participants in the severe-critical group had CS. Thus, the rate of CS was 37.5 % (12/32). Compared to COVID-19-negative patients, COVID-19-positive patients had higher IL-6 levels, which decreased at discharge except for those dying. Among the COVID-19-positive patients, nine died. For both the mild-moderate and severe-critical patients, IL-6 increased with increasing CRP (p < 0.001).ConclusionDuring the COVID-19 pandemic, Bangladeshi adults experienced a surge in cytokine storms. The rate of cytokine storm in Bangladeshi COVID-19-positive adults was 37.5 %. Introduction COVID-19 patients with severe or critical illness experience a high mortality rate of 49.0–61.5 % [1]. This is due to an inflammatory response, with persistent fever, cytopenia, high levels of C-reactive protein, ferritin, and D-dimer, and higher levels of proinflammatory cytokines like IL-1β, IL-6, IL-10, and TNF-α [2,3]. Emerging data suggest that these clinical and laboratory abnormalities occur due to an exaggerated response of the immune system, termed cytokine storm [4,5]. Cytokine storms, which occur in graft versus host disease and allogeneic stem cell transplant, autoimmune diseases, tuberculosis, and chimeric antigen receptor T-cell therapy in malignant diseases, are responsible for these abnormalities [4]. Severe acute respiratory syndrome coronavirus 2 (SAR-CoV-2) replicates in respiratory epithelial cells, induces inflammation, and activates macrophages, monocytes, and other B cells and T cells to produce cytokines and chemokines. Some of the substances that are made are IL-6, IL-1β, TNF-α, interferon-gamma, granulocyte-colony stimulating factor, interferon-γ inducible protein 10, monocyte chemoattractant protein 1, macrophage inflammatory protein 1-α, and more [3]. Studies show that IL-6 and TNF-α increase in severe COVID-19 and can predict respiratory failure and mechanical ventilation [5]. Patients with an IL-6 level of >80 pg/mL had a 22-fold higher risk of respiratory failure than patients with lower IL-6 levels [6]. IL-1β also induces lung inflammation, fever, and lung fibrosis in COVID-19 [7]. Bangladesh also fought this deadly pandemic with its existing health infrastructure when the disease reached March 8, 2020. Over the period, the pandemic caused 2,039,911 illnesses, with 29,451 recorded deaths [8]. To our knowledge, in Bangladesh, no study has evaluated cytokine levels in COVID-19 patients and their correlation with disease severity. Therefore, we aimed to estimate the burden and frequency of “cytokine storms” among COVID-19-positive Bangladeshi adults using conventional laboratory criteria and validate these criteria against cytokine-based criteria. We planned to examine the correlation between cytokines levels and COVID-19 severity, and compare baseline levels of cytokines IL-6, IL-1β, and TNF-α in COVID-19-positive cases with a healthy population. Methodology Study design, settings, and eligibility criteria A prospective cohort study was conducted at icddr,b Dhaka Hospital focusing on COVID-19 diagnosis, management, and referral at its makeshift COVID unit, [9,10]. at the very beginning of the pandemic preparedness. The makeshift hospital has all the facilities to care for critically ill COVID-19 patients also [11]. The study included adults aged 18 years or older who tested positive for COVID-19 by reverse-transcription polymerase chain reaction (RT-PCR), as well as healthy volunteers without acute illnesses and with normal physical examinations. Participation was contingent upon obtaining consent from the individuals or their close relatives. COVID-19-positive pregnant women and patients with recent corticosteroids, immunosuppressants, or chemotherapy were excluded from the study. Data collection Clinical information After enrollment, data were collected from the patients or their caregivers and recorded on a study case record form. Variables comprised demographics, major comorbidities (hypertension, bronchial asthma/chronic obstructive airway disease (COPD), ischemic heart disease (IHD), myocardial infarction (MI), stroke, diabetes mellitus (DM), psychiatric illness, and any surgery), disease symptoms (fever, cough, dyspnea, diarrhea, headache) with duration, examination findings (pulse, blood pressure, respiratory rate, oxygen saturation in room air, axillary temperature), requirement of supplemental oxygen, mechanical ventilation or noninvasive ventilation, diagnosis, hospital course, and outcome of this admission. Laboratory information The study collected whole blood samples in tubes containing ethylenediamine tetraacetic acid from COVID-19 patients, including leftover samples from laboratories to avoid multiple punctures for the same patient. After centrifugation, the plasma was collected and stored at −80 °C. At the end of enrollment, IL-1β, IL-6, and TNF-α were quantified by a sandwich enzyme immunoassay using a Quantikine ELISA kit (R&D Systems, Minneapolis, USA) according to the manufacturer's instructions. Complete blood count (CBC), C reactive protein (CRP), D-dimer, ferritin, and lactate dehydrogenase (LDH) were performed in real-time following the protocol in the icddr,b laboratory. Other investigations were performed according to the clinical discretion of the treating clinical team. Chest X-rays in anteroposterior view with a portable machine for patients were performed at enrollment and when clinical conditions were needed. Radiological scoring was performed as per a published report on ground glass opacification by the experienced radiologist of the icddr,b Dhaka hospital [12]. For healthy non-COVID-19 participants, we collected only one blood sample for cytokine analysis, CBC, and CRP. Operational definition Cytokine Storm [[15], [16], [17],20]: There was no established definition of a COVID-19-related cytokine storm during the study. In this study, we defined cytokine storms in three ways, as follows:1.If a clinically sick COVID-19 patient had at least two of the five criteria: I) CRP >5 mg/dL, ii) LDH >250 U/L, iii) ferritin >500 ng/mL, iv) D-dimer >1000 ng/mL, and v) absolute lymphocyte count (ALC) <600 [13].2.If a clinically sick COVID-19 patient had at least three of the following five criteria: i) CRP >5 mg/dL, ii) LDH >250 U/L, iii) ferritin >500 ng/mL, iv) D-dimer >1000 ng/mL, and v) absolute lymphocyte count (ALC) <600 [13].3.If IL-6 > 80 pg/mL for a symptomatic patient [4]. 1.If a clinically sick COVID-19 patient had at least two of the five criteria: I) CRP >5 mg/dL, ii) LDH >250 U/L, iii) ferritin >500 ng/mL, iv) D-dimer >1000 ng/mL, and v) absolute lymphocyte count (ALC) <600 [13]. If a clinically sick COVID-19 patient had at least two of the five criteria: I) CRP >5 mg/dL, ii) LDH >250 U/L, iii) ferritin >500 ng/mL, iv) D-dimer >1000 ng/mL, and v) absolute lymphocyte count (ALC) <600 [13]. 2.If a clinically sick COVID-19 patient had at least three of the following five criteria: i) CRP >5 mg/dL, ii) LDH >250 U/L, iii) ferritin >500 ng/mL, iv) D-dimer >1000 ng/mL, and v) absolute lymphocyte count (ALC) <600 [13]. If a clinically sick COVID-19 patient had at least three of the following five criteria: i) CRP >5 mg/dL, ii) LDH >250 U/L, iii) ferritin >500 ng/mL, iv) D-dimer >1000 ng/mL, and v) absolute lymphocyte count (ALC) <600 [13]. 3.If IL-6 > 80 pg/mL for a symptomatic patient [4]. If IL-6 > 80 pg/mL for a symptomatic patient [4]. Clinical case definition in COVID-19 [10,14]: Mild cases: Patients with fever, cough, sore throat, malaise, headache, muscle pain without shortness of breath, or abnormal chest X-ray imaging. Moderate case: Cases having signs of pneumonia with a respiratory rate ≤30 breaths/min or peripheral arterial oxygen saturation (SpO2) in blood >93 % at room air. Severe case: Patients with respiratory distress with a respiratory rate >30 breaths/min, SpO2 ≤ 93 % in room air (later, the cutoff was ≤90 % in room air), or lung infiltrates >50 %. Critical case: Cases having respiratory failure (oxygen saturation/fraction of inspired oxygen [SpO2/FiO2] ≤ 315 mmHg) and requiring mechanical ventilation, or shock where hypotension persists despite volume resuscitation and requiring vasopressors to maintain the mean arterial pressure >60 mmHg. Acute respiratory distress syndrome (ARDS): Acute respiratory failure of unclear etiology involving the lung bilaterally with ground-glass opacities on chest imaging and impairment of oxygenation, i.e., SpO2/FiO2) ≤ 315 mmHg. Sample size This study was performed with an exploratory approach considering resource-limitation and the vulnerability of the research staff without calculating sample size following a statistical formula. We screened adult patients with SARS-CoV-2 infection at the icddr,b COVID-19 unit from October 2020 to May 2021. From them, we enrolled 30 mild-moderate and 32 severe-critical cases according to the aforementioned criteria [9,12]. Initial evidence suggested that cytokine storms were more likely to occur in severe and critical cases than in mild-moderate cases; hence, we stratified our samples into two groups. Fourteen -negative healthy volunteers negative for COVID-19 by RT-PCR were also enrolled to compare their cytokines with COVID-19-positive cases. Statistical analysis We presented descriptive statistics for categorical variables in percentages with frequencies, mean with standard deviation and median with interquartile range for quantitative variables, and used chi-square, t, and Mann-Whitney tests to analyze associations between background and clinical characteristics and COVID-19. The strength of association was determined by calculating the odds ratio (OR) and their 95 % confidence intervals (CI) with a significant P less than 0.05. ANOVA and the Kruskal‒Wallis test were performed to determine the mean and median differences, respectively, of the three groups. We used the receiver operating characteristic (ROC) curve to analyze cytokine storm criteria and their agreement, considering IL-6 levels above 80 pg/mL in a patient with symptoms as a cytokine storm. As both the outcome and explanatory variables were not distributed normally, we log transformed the variables to normalize the distribution. Then, we used linear regression to examine IL-6, IL-1β and TNF-α′s relationship with biomarkers like CRP, LDH, ferritin, D-dimer, and lymphocytes. The regression analysis estimates were back-transformed and presented as the exponential of the coefficient, exp(β), with 95 % CI. The statistical software SPSS version 20 (IBM Corp, New York, USA) was used for data entry. All statistical analyses were performed using the statistical software Stata, version 15.0. Ethical consideration The study (protocol number: PR-20084), approved by the Research Review Committee and the Ethical Review Committee of the icddr,b, obtained written informed consent from the participants or attending caregivers as appropriate. Results Clinical and laboratory characteristics of the study participants Severe-critical COVID-19-positive patients were elderly, comorbid with chronic respiratory illness, hypertension, diabetes, and heart diseases, and often presented with fever, cough, and dyspnea compared to mild-moderate COVID-19-positive patients (Table 1). When compared to mild to moderate COVID-19 cases, they had lower hemoglobin, red blood cells, platelet, and lymphocyte count but higher neutrophil (%), CRP, D-dimer, LDH, and ferritin levels (Supplementary Table 1).Table 1Baseline clinical characteristics of the study participants.Table 1CharacteristicsCOVID-19 negative healthy volunteer (n = 14), n (%)Mild-moderate COVID-19 cases (n = 30) n (%)Severe-critical COVID-19 cases (n = 32), n (%)p valueAge (years) (median, IQR)31.5 (28.0, 44.0)39.0 (31.0, 52.0)57.0 (51.5, 66.5)<0.001¶Sex male5/14 (35.7)22/30 (73.3)18/32 (56.3)0.055Chronic respiratory illness–5/30 (16.7)8/32 (25.0)0.115Hypertension1/14 (7.1)12/30 (40.0)23/32 (71.9)<0.001Ischemic heart disease–2/30 (6.7)10/32 (31.3)0.006Diabetes mellitus–10/30 (33.3)17/32 (53.1)0.002Received antibiotics at home–18/30 (60.0)15/31 (48.4)0.363COVID-19 positive relatives at home6/14 (42.9)18/28 (64.3)12/26 (46.2)0.287Fever–22/30 (73.3)29/32 (90.6)<0.001Duration of fever, (mean ± SD)–6.1 (2.8)5.6 (3.4)0.564¥Cough–23/30 (76.7)26/32 (81.3)<0.001Duration of cough, (mean ± SD)–5.9 (3.3)6.1 (3.8)0.810¥Headache–7/30 (23.3)4/32 (12.5)0.264Diarrhea–2/30 (6.7)7/32 (21.8)0.149€Dyspnea–6/30 (20.0)18/32 (56.3)<0.001€Chest pain–3/30 (10.0)1/32 (3.1)0.402Respiratory rate/min, (mean ± SD)18.7 (2.7)23.1 (5.9)33.1 (7.8)<0.001∑Pulse rate/min, (mean ± SD)81 (9.7)91.4 (12.9)94.6 (13.4)0.007∑Systolic blood pressure (mm Hg), (mean ± SD)116.7 (11.5)121.2 (18.6)127.6 (19.5)0.159∑Diastolic blood pressure (mm Hg), (mean ± SD)76.3 (9.8)82.3 (12.0)75.3 (11.6)0.050∑Temperature (°C)36.6 (0.5)36.2 (0.6)37.1 (1.0)<0.001∑Random blood sugar (m mol/L)–6.4 (5.3, 11)8.4 (6.8, 10.3)0.140Alert14/14 (100)30/30 (100)30/32 (93.8)0.663€P values were generated by ¶Kruskal‒Wallis test; ¥t-test; €Fisher's exact test; £Mann‒Whitney test; ∑Analysis of variance; otherwise chi-square test. Table 1Baseline clinical characteristics of the study participants.Table 1CharacteristicsCOVID-19 negative healthy volunteer (n = 14), n (%)Mild-moderate COVID-19 cases (n = 30) n (%)Severe-critical COVID-19 cases (n = 32), n (%)p valueAge (years) (median, IQR)31.5 (28.0, 44.0)39.0 (31.0, 52.0)57.0 (51.5, 66.5)<0.001¶Sex male5/14 (35.7)22/30 (73.3)18/32 (56.3)0.055Chronic respiratory illness–5/30 (16.7)8/32 (25.0)0.115Hypertension1/14 (7.1)12/30 (40.0)23/32 (71.9)<0.001Ischemic heart disease–2/30 (6.7)10/32 (31.3)0.006Diabetes mellitus–10/30 (33.3)17/32 (53.1)0.002Received antibiotics at home–18/30 (60.0)15/31 (48.4)0.363COVID-19 positive relatives at home6/14 (42.9)18/28 (64.3)12/26 (46.2)0.287Fever–22/30 (73.3)29/32 (90.6)<0.001Duration of fever, (mean ± SD)–6.1 (2.8)5.6 (3.4)0.564¥Cough–23/30 (76.7)26/32 (81.3)<0.001Duration of cough, (mean ± SD)–5.9 (3.3)6.1 (3.8)0.810¥Headache–7/30 (23.3)4/32 (12.5)0.264Diarrhea–2/30 (6.7)7/32 (21.8)0.149€Dyspnea–6/30 (20.0)18/32 (56.3)<0.001€Chest pain–3/30 (10.0)1/32 (3.1)0.402Respiratory rate/min, (mean ± SD)18.7 (2.7)23.1 (5.9)33.1 (7.8)<0.001∑Pulse rate/min, (mean ± SD)81 (9.7)91.4 (12.9)94.6 (13.4)0.007∑Systolic blood pressure (mm Hg), (mean ± SD)116.7 (11.5)121.2 (18.6)127.6 (19.5)0.159∑Diastolic blood pressure (mm Hg), (mean ± SD)76.3 (9.8)82.3 (12.0)75.3 (11.6)0.050∑Temperature (°C)36.6 (0.5)36.2 (0.6)37.1 (1.0)<0.001∑Random blood sugar (m mol/L)–6.4 (5.3, 11)8.4 (6.8, 10.3)0.140Alert14/14 (100)30/30 (100)30/32 (93.8)0.663€P values were generated by ¶Kruskal‒Wallis test; ¥t-test; €Fisher's exact test; £Mann‒Whitney test; ∑Analysis of variance; otherwise chi-square test. Baseline clinical characteristics of the study participants. P values were generated by ¶Kruskal‒Wallis test; ¥t-test; €Fisher's exact test; £Mann‒Whitney test; ∑Analysis of variance; otherwise chi-square test. Distribution of cytokine storm among the COVID-19-positive cases Considering IL-6 >80 pg/mL as CS, 12 participants from the severe-critical group had CS, and none had CS from the mild-moderate COVID-19 group. Therefore, 37.5 % (12/32) of participants in the severe-critical group had CS, while 19.3 % (12/62) of the overall population had CS. Considering at least two criteria out of five (CRP >5 gm/dl, ALC <600, D-dimer >1000 ng/mL, LDH >250 U/L, ferritin >500 ng/mL), cytokine storms were present in five participants from the mild-moderate and 24 participants from severe-critical groups, with a prevalence of 75 % (24/32) in the severe-critical group and 46.7 % (29/62) overall. Considering at least three criteria, one and 10 participants from the mild-moderate and severe-critical groups, respectively, had CS. Therefore, 31.3 % (10/32) of participants in the severe-critical group had CS, while the overall prevalence was 17.7 % (11/62) (Fig. 1).Fig. 1Distribution of cytokine storms among the enrolled COVID-19-positive patients.Fig. 1 Distribution of cytokine storms among the enrolled COVID-19-positive patients. ROC to determine the area under the curve for different conventional five laboratory markers and their combinations Considering IL-6 >80 pg/mL as a standard criterion for cytokine storm, we observed that CRP >5 gm/dl had an AUC of 0.78 with a sensitivity of 71.4 % and a specificity of 73.2 %. D-Dimer >1000 ng/mL had an AUC of 0.77 with a sensitivity of 57.1 % and a specificity of 97.6 %. The combination of any two laboratory markers out of five (CRP >5 mg/dL, LDH >250 U/L, ferritin >500 ng/mL, D-dimer >1000 ng/mL, and absolute lymphocyte count <600) had an AUC of 0.80 with a sensitivity of 100 % and a specificity of 61 %. The combination of any three laboratory markers out of five (CRP >5 mg/dL, LDH >250 U/L, ferritin >500 ng/mL, D-dimer >1000 ng/mL, and absolute lymphocyte count <600) had an AUC of 0.78 with a sensitivity of 71.4 % and a specificity of 85.4 % (Fig. 2). The overall p-value was <0.001, indicating that there was statistically significant difference among the values of the AUROC curves.Fig. 2Receiver operating characteristic (ROC) curve demonstrating the area under the curve for different laboratory markers and their combination.Fig. 2 Receiver operating characteristic (ROC) curve demonstrating the area under the curve for different laboratory markers and their combination. Association of IL-6, IL-1β and TNF-α with different categories of COVID-19 The log of IL-6 decreased for all COVID-19-positive cases from enrollment to discharge, except in death cases where the value increased. At enrollment, COVID-19-positive patients had significantly higher logIL-6 levels than COVID-19-negative patients (all p < 0.05). Severe-critical and deceased cases had significantly higher logIL-6 levels than mild and moderate cases (p < 0.001), while severe-critical and deceased cases showed no difference in logIL-6 (p = 0.077). The study found that logIL-6 levels in deceased cases significantly differed from those in mild-moderate and severe-critical cases at discharge (death vs. mild-moderate and death vs. severe-critical, p < 0.001), while no significant difference was observed between mild-moderate and severe-critical cases (mild-moderate vs. severe-critical, p = 0.424) (Fig. 3, Supplementary Table 2). There were no statistically significant differences between groups for log IL-1β at enrollment or discharge (Supplementary Fig. 1). For the logTNF-α, at enrollment, all COVID-19-positive cases significantly differed from COVID-19-negative cases (p < 0.05). However, no significant difference was found among COVID-19-positive cases. At discharge, severe-critical cases significantly differed from mild-moderate (severe-critical vs. mild-moderate, P = 0.005) and death (severe-critical vs. death, p = 0.002) cases, where both the mild-moderate and death cases had higher TNF-α levels than severe-critical cases (Supplementary Fig. 2).Fig. 3LogIL-6 and COVID-19. Distribution of LogIL-6 among COVID-19-positive and COVID-19-negative cases both at enrollment and discharge.At enrollment, COVID-19-positive cases had a significantly higher log of IL-6 than COVID-19-negative cases (all p < 0.05); death vs mild-moderate and severe-critical vs mild-moderate, p < 0.001; severe-critical vs death cases, p = 0.077. At discharge, death vs mild-moderate and death vs severe-critical, p < 0.001; mild-moderate vs severe-critical, p = 0.424.Fig. 3 LogIL-6 and COVID-19. Distribution of LogIL-6 among COVID-19-positive and COVID-19-negative cases both at enrollment and discharge. At enrollment, COVID-19-positive cases had a significantly higher log of IL-6 than COVID-19-negative cases (all p < 0.05); death vs mild-moderate and severe-critical vs mild-moderate, p < 0.001; severe-critical vs death cases, p = 0.077. At discharge, death vs mild-moderate and death vs severe-critical, p < 0.001; mild-moderate vs severe-critical, p = 0.424. Association of IL-6, IL-1β, and TNF-α with conventional laboratory markers The study found that IL-6 levels increased with increasing CRP in both mild-moderate and severe-critical patients (exp (β): 1.56; 95 % CI: 1.31, 1.85; p < 0.001 for mild-moderate and exp (β): 1.56; 95 % CI: 1.12, 2.17; p < 0.001 for severe-critical cases). However, after controlling for confounders such as patients' age, sex, chronic respiratory illness, hypertension, ischemic heart diseases, diabetes mellitus, and steroid, we observed CRP increased with IL6, but only statistically significant for mild-moderate cases. IL-6 increased with increasing D-dimer (exp (β): 2.35; 95 % CI: 1.40, 3.93; p = 0.002) only for severe-critical cases even after controlling for the confounder. However, IL-6 levels decreased with increased lymphocyte count, indicating a potential immune response to inflammation (exp (β): 0.29; 95 % CI: 0.12, 0.69; p = 0.006 in mild-moderate and exp (β): 0.53; 95 % CI: 0.31, 0.91; p < 0.022 for severe-critical cases) (Table 2). We found that IL-1β decreased with increased biomarkers except lymphocytes in severe-critical cases. However, we did not find any statistically significant association between IL-1β and laboratory biomarkers (Supplementary Table 3) or TNF-α and laboratory biomarkers (Supplementary Table 4).Table 2Association of IL-6 with conventional laboratory markers by simple linear regression both simple and multivariable linear regression.Table 2Laboratory markersMild-moderate COVID-19 cases (n = 30)Severe-critical COVID-19 cases (n = 32)exp (β), 95 % CIP valueaAdj exp (β), 95 % CIp valueexp (β), 95 % CIp valueaAdj exp (β), 95 % CIp valueCRP1.56 (1.31, 1.85)<0.0011.42 (1.18, 1/72)0.0011.56 (1.12, 2.17)0.0101.26 (0.86–1.83)0.216LDH2.30 (0.83, 6.36)0.1043.12 (1.27, 7.68)0.0160.66 (0.19, 2.24)0.4871.66 (0.39, 7.0)0.466Ferritin1.32 (0.92, 1.89)0.1241.37 (0.94, 2.0)0.0940.99 (0.60, 1.66)0.9981.10 (0.61–2.01)0.729D-Dimer1.88 (0.92, 3.83)0.0811.10 (0.48, 2.48)0.8162.35 (1.40, 3.93)0.0022.32 (1.23, 4.39)0.012Lymphocyte0.29 (0.12, 0.69)0.0060.39 (0.15, 1.01)0.0520.53 (0.31, 0.91)0.0220.58 (0.34, 1.00)0.051aMultivariable linear regression model adjusting for patients' age, sex, chronic respiratory illness, hypertension, ischemic heart diseases, diabetes mellitus, and steroid. Table 2Association of IL-6 with conventional laboratory markers by simple linear regression both simple and multivariable linear regression.Table 2Laboratory markersMild-moderate COVID-19 cases (n = 30)Severe-critical COVID-19 cases (n = 32)exp (β), 95 % CIP valueaAdj exp (β), 95 % CIp valueexp (β), 95 % CIp valueaAdj exp (β), 95 % CIp valueCRP1.56 (1.31, 1.85)<0.0011.42 (1.18, 1/72)0.0011.56 (1.12, 2.17)0.0101.26 (0.86–1.83)0.216LDH2.30 (0.83, 6.36)0.1043.12 (1.27, 7.68)0.0160.66 (0.19, 2.24)0.4871.66 (0.39, 7.0)0.466Ferritin1.32 (0.92, 1.89)0.1241.37 (0.94, 2.0)0.0940.99 (0.60, 1.66)0.9981.10 (0.61–2.01)0.729D-Dimer1.88 (0.92, 3.83)0.0811.10 (0.48, 2.48)0.8162.35 (1.40, 3.93)0.0022.32 (1.23, 4.39)0.012Lymphocyte0.29 (0.12, 0.69)0.0060.39 (0.15, 1.01)0.0520.53 (0.31, 0.91)0.0220.58 (0.34, 1.00)0.051aMultivariable linear regression model adjusting for patients' age, sex, chronic respiratory illness, hypertension, ischemic heart diseases, diabetes mellitus, and steroid. Association of IL-6 with conventional laboratory markers by simple linear regression both simple and multivariable linear regression. Multivariable linear regression model adjusting for patients' age, sex, chronic respiratory illness, hypertension, ischemic heart diseases, diabetes mellitus, and steroid. Discussion In this study, we analyzed COVID-19 patients′ clinical and immunological responses and found that cytokine storms were present in 37.5 % of severe-critical cases with IL-6 levels >80 pg/mL. 58.3 % (7/12) of those who were in cytokine storms died. Using IL-6 >80 pg/mL as a cutoff for CS, the three-criteria-based diagnosis had a high sensitivity and specificity. Patients with higher IL-6 levels had a higher risk of death with CRP being a useful surrogate for IL-6 quantification in COVID-19-positive cases. COVID-infected cells activate monocytes, macrophages, B cells, and T cells, releasing inflammatory cytokines like IL-6, IL-1β, TNF-α, interferon-gamma, and granulocyte colony-stimulating factor, leading to respiratory distress syndrome, clot formation in blood vessels, multiorgan failure, and death [3,15]. Our findings from the makeshift COVID-19 unit also resonated with the same pictures. As expected, we observed that severe to critically ill COVID-19 patients had higher IL-6 and TNF-α levels than mild-moderate cases. seven out of nine deaths had cytokine storms, suggesting an association between COVID-19 induced cytokine storms and mortality. Unchecked inflammatory cascades with abnormal and uncontrolled release of inflammatory cytokines impair immune systems, leading to severe COVID-19 associated with cytokine storms, T-cell depletion, and impaired lung function [16,17]. Such phenomena were also observed during other viral infections, such as severe acute respiratory syndrome (SARS) and Middle East respiratory syndrome (MERS) [18]. Patients with SARS and MERS experience cytokine storms, characterized by elevated levels of IL-1β, IL-8, IL-6, and chemokine ligands, leading to increased mortality rates of 10 % and 36 %, respectively [19]. We examined three cytokines, namely, IL-6, IL-1β, and TNF-α. During cytokine storms, studies showed that there were high levels of TNF-α, IL-6, IFN-γ, and IL-1β [20]. IL-6 was linked to disease severity and ventilatory support requirements [6,21,22]. We observed baseline IL-6 was significantly higher in COVID-19-positive cases than in COVID-19-negative cases, which declined with clinical improvement except among patients who died. In death cases, the IL-6 increased further, which also signifies that they were in a cytokine storm. Pedersen and colleagues observed that increased levels of IL-6, along with TNF-α and IL-10, were significantly associated with reduced recovery chances requiring intensive care admission [23]. CRP, a marker of inflammation, is produced by hepatocytes after IL-6 stimulation and partly by IL-1β [24,25]. Early COVID-19 studies showed a positive association between CRP levels, lung lesions, and disease severity [24,26]. Age, sex, and physical condition do not affect CRP levels [27]. In our studied patients, we found that CRP and IL-6 levels in COVID-19-positive patients changed in the same direction, indicating that CRP quantifies IL-6 levels. However, for severe-critical illness, the relation was not significant in a multivariable model. TNF-α, a key cytokine in COVID-19 increases inflammation and induces fever, phagocytosis by macrophages, and an acute phase response by the liver [28]. In COVID-19, it is associated with severe illness and ICU admission [2,29], in our cohort, we observed that TNF-α declined from the baseline value at discharge or death but only increased in mild-moderate cases. We have no ready explanation; probably, clinical improvement and decline in TNF-α did not occur simultaneously. IL-1β, a leukocytic pyrogen, is produced by activated macrophages in COVID-19, causing lung inflammation, fever, and fibrosis [7]. Studies show that increased levels predicted the requirement of ICU admission in COVID-19 [4]. However, we did not observe any association of IL-1β in our COVID-19 cohort with the illness severity. Furthermore, we classified cytokine storms in different ways. Cytokines storms were diagnosed using an IL-6-based criterion if IL-6 levels >80 pg/mL. The cytokine IL-6 has been studied extensively in relation to COVID-19. We adopted a definition where two or three conventional laboratory markers out of five in sick COVID-19 patients were used to diagnose cytokine storms [11]. Cytokines quantification is a complex and time-consuming process that relies on various methods, reagents, and the expertise of lab personnel. Laboratory markers can identify patients at risk for cytokine storms, with two positive markers having higher sensitivity and three positive markers having higher specificity, and CRP alone being a better predictor. The combination of several conventional biomarkers for the diagnosis and management of cytokine storms plays a crucial role in pandemic situations, where movement restrictions and resource allocation challenges make patient care more difficult. We have several limitations. This was a single-center study with a small sample size. However, the inclusion of healthy volunteers was a strength. We included healthy volunteers and tested their samples for IL-6, TNF-α, and IL-1β at enrollment. We used the leftover sample for nine cases before commencement of the main study, as protocol development and IRB approval needed time, and we were going through the peak of COVID-19 cases at that time. Therefore, all planned baseline investigations were not available for them, which might underestimate the number of cytokine storms. We estimated the cytokines at enrollment and discharge, not serially, so we were unable to identify the trend appropriately. As we tested the cytokines at the end of the study for valid reasons, we were unable to use the results in planning the management of our critically ill COVID-19 patients. Considering resources and the vulnerability of the research staff, we conducted this exploratory study at the midst of COVID-19 with a limited sample size. A study with sufficient power and an appropriate sample size would provide a robust answer in clarifying the mechanisms behind the cytokine storm. Finally, we were able to test only three cytokines, other cytokines can cause cytokine storms in different ways. So, it is difficult to ascertain whether IL-6 is the only offending candidate for cytokine storms in COVID-19. Conclusion The rate of cytokine storm in Bangladeshi COVID-19-positive adults was 37.5 % with a mortality rate of 58.3 % (7/12). IL-6 levels increased in patients, with CRP potentially serving as a surrogate. During the pandemic, this type of study could help clinicians identify patients needing specialist care. Moreover, the correlation with a conventional biomarker, such as CRP, is a key finding. Early identification of such surrogate biomarkers would enable clinicians to test, diagnose, and monitor more patients, even in resource-limited settings. CRediT authorship contribution statement Monira Sarmin: Writing – review & editing, Writing – original draft, Visualization, Validation, Project administration, Methodology, Investigation, Formal analysis, Data curation, Conceptualization. Fhameda Akter: Writing – review & editing, Writing – original draft, Visualization, Resources, Methodology, Data curation, Conceptualization. Anika Nawrin Islam: Writing – review & editing, Writing – original draft, Resources, Methodology, Data curation, Conceptualization. Mustafa Mahfuz: Writing – review & editing, Visualization, Resources, Formal analysis, Conceptualization. Subhasish Das: Writing – review & editing, Validation, Methodology, Formal analysis, Conceptualization. Sharifuzzaman: Writing – review & editing, Resources, Methodology, Investigation, Data curation, Conceptualization. S.M. Tafsir Hasan: Writing – review & editing, Software, Methodology, Formal analysis. Taufiqur Rahman Bhuiyan: Writing – review & editing, Writing – original draft, Software, Project administration, Methodology, Data curation, Conceptualization. Mustafizur Rahman: Writing – review & editing, Project administration, Methodology, Data curation, Conceptualization. Amran Gazi: Writing – review & editing, Project administration, Methodology, Investigation, Formal analysis, Conceptualization. Fariha Bushra Matin: Writing – review & editing, Project administration, Methodology, Investigation, Conceptualization. Md Tariqujjaman: Writing – review & editing, Visualization, Software, Investigation, Formal analysis, Conceptualization. Lubaba Shahrin: Writing – review & editing, Project administration, Methodology, Investigation, Formal analysis, Data curation. Mofakharul Islam: Writing – review & editing, Project administration, Methodology, Conceptualization. Asif Mujtaba Mahmud: Writing – review & editing, Visualization, Methodology, Formal analysis, Conceptualization. Sayera Banu: Writing – review & editing, Supervision, Methodology, Investigation, Conceptualization. Mohammod Jobayer Chisti: Writing – review & editing, Visualization, Validation, Supervision, Project administration, Methodology, Investigation, Formal analysis, Data curation, Conceptualization. Firdausi Qadri: Writing – review & editing, Visualization, Supervision, Formal analysis, Conceptualization. John David Clemens: Writing – review & editing, Visualization, Supervision, Formal analysis, Conceptualization. Tahmeed Ahmed: Writing – review & editing, Validation, Supervision, Methodology, Funding acquisition, Formal analysis, Conceptualization. Ethical approval The study (Protocol number: 20084) was approved by the Research Review Committee and the Ethics Review Committee of the International Centre for Diarrhoeal Disease Research, Bangladesh. Data and code availability statement icddr,b adheres to standard guidelines for sharing data with the research scholar community. Upon a reasonable request, individuals may contact Ms Shiblee Sayeed (shiblee_s@icddrb.org), Senior Manager of Research Administration, for access to data. Funding This research did not receive any specific grant from funding agencies in the public, commercial, or not-for-profit sectors. Declaration of competing interest The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.