key: cord-330411-hg1cxcs7
authors: Keddie, S.; Ziff, O.; Chou, M.K.L.; Taylor, R.L.; Heslegrave, A.; Garr, E.; Lakdawala, N.; Church, A.; Ludwig, D.; Manson, J.; Scully, M.; Nastouli, E.; Chapman, M.D.; Hart, M.; Lunn, M.P.
title: Laboratory biomarkers associated with COVID-19 severity and management
date: 2020-10-22
journal: Clin Immunol
DOI: 10.1016/j.clim.2020.108614
sha: 
doc_id: 330411
cord_uid: hg1cxcs7

The heterogeneous disease course of COVID-19 is unpredictable, ranging from mild self-limiting symptoms to cytokine storms, acute respiratory distress syndrome (ARDS), multi-organ failure and death. Identification of high-risk cases will enable appropriate intervention and escalation. This study investigates the routine laboratory tests and cytokines implicated in COVID-19 for their potential application as biomarkers of disease severity, respiratory failure and need of higher-level care. From analysis of 203 samples, CRP, IL-6, IL-10 and LDH were most strongly correlated with the WHO ordinal scale of illness severity, the fraction of inspired oxygen delivery, radiological evidence of ARDS and level of respiratory support (p ≤ 0.001). IL-6 levels of ≥3.27 pg/ml provide a sensitivity of 0.87 and specificity of 0.64 for a requirement of ventilation, and a CRP of ≥37 mg/L of 0.91 and 0.66. Reliable stratification of high-risk cases has significant implications on patient triage, resource management and potentially the initiation of novel therapies in severe patients.

International efforts to 'flatten the curve' of Coronavirus Disease-2019 (COVID-19) through social isolation restrictions have allowed most health systems to cope with huge demands on healthcare facilities, in particular providing time to acquire ventilators and intensive care beds.

As governments loosen lockdown processes and with the unknown influences of seasonal variation and acquired immunity, front line services need to prepare for a second wave of infection. Lessons learnt regarding COVID-19 disease biology, transmission, risk factors, complications and treatments will be used to adapt and improve clinical services and improve outcomes.

The heterogeneous disease course of COVID-19 is unpredictable with most patients experiencing mild self-limiting symptoms. However up to 30% require hospitalisation, and up to 17% of these require intensive care support for acute respiratory distress syndrome (ARDS), hyperinflammation and multiorgan failure. [1] [2] [3] A cytokine storm in patients with severe disease was identified in the early reports of Wuhan patients and is intrinsic to disease pathology. In this cohort, elevated plasma interleukin (IL)-2, IL-7, IL-10, Granulocyte colony stimulating factor (GCSF), interferon γ-induced protein 10 (IP10, monocyte chemoattractant protein-1 (MCP1), macrophage inflammatory protein 1-alpha (MIP1A) and tumor necrosis factor -alpha (TNF-α) levels in ICU patients were identified. 4 Some subsequent studies have implicated IL-6 as a valuable predictor of adverse clinical outcome and a potential therapeutic target. 5, 6 One or more clinical and wet biomarkers may enable early identification of high-risk cases, assisting disease stratification and effective use of limited specialist resources.

This study comprehensively explored clinical disease features and routine laboratory tests against specialist cytokine biomarkers associated with COVID-19 disease and its complications, to address their association with disease severity, respiratory intervention and outcome.

One hundred consecutive hospital in-patients with COVID-19 infection, whose sera were sent for cytokine testing, were investigated as part of standard care and routine practice at University College London Hospital NHS Trust. All cases were confirmed as COVID-19 by polymerase chain reaction (PCR) analysis (86%) or deemed highly likely based on internationally recognised diagnostic criteria including recent exposure, clinical, radiological and laboratory features. 7 Laboratory biomarkers including cytokines were collated with baseline demographics, risk factors, disease features, treatments and outcomes. Level of clinical care (ward based, high-dependency or intensive treatment), oxygen requirements and vital signs were associated with the biomarkers at the time of sample collection, at peak illness and on most recent assessment. Serial samples were analysed to establish whether biomarkers correlated with, or could predict disease course. Severity of COVID-19 illness was determined utilising the World Health Organisation (WHO) COVID-19 ordinal severity scale which broadly measures level of respiratory compromise and requirement of intervention, with a score of 1 defined by no limitation of activities, to 7 requiring ventilation and additional organ support. 8 Other features of organ system compromise such as neurological involvement, renal failure and requirement for dialysis were analysed for association with laboratory biomarkers.

Bloods for cytokine analysis were centrifuged within 4 hours of collection, separated and sera frozen at -80 0 C for up to 24 hours before being analysed. Immunoassays were performed according to manufacturer's instructions and reads were fully automated. Routine laboratory data from automated analysers including C-reactive protein (CRP), D-dimer, lactate dehydrogenase (LDH), lymphocyte count, ferritin, fibrinogen and platelets were collated with the clinical and cytokine data. 

Chi-square was used for the comparison of categorical variables. Biomarker levels comparing dichotomous variables were evaluated using Welch's unpaired t-test. Comparison of multiple group outcomes was performed using one-way ANOVA. Biomarker and continuous clinical variables were correlated using Pearson correlation coefficient. The predictive values of biomarkers were calculated by identifying Youden's index on receiver operator characteristic curves. P-value ≤0.05 was considered statistically significant. All statistical analyses were made using R statistical package 3.6.2.

Samples from 100 patients were collected from the 6 th April to the 18 th May from patients hospitalised at University College London Hospital (UCLH) sites, 78 from the main hospital and 22 from the National Hospital for Neurology and Neurosurgery (including neuro-intensive care). Eighty-six patients were COVID-19 PCR positive and 14 deemed highly likely based on diagnostic criteria. 7 Seventy-four cases were male, with a median age of 59 years (20-92). One hundred and forty samples were collected from ward environments versus 63 on intensive care.

The WHO COVID severity score ranged from 3 (hospitalised, no oxygen therapy) to 7 (severe J o u r n a l P r e -p r o o f disease, ventilation and additional organ support) with a median of 6 (severe disease, intubated and ventilated). Background characteristics associated with more severe infection and intervention included older age (p≤0.001), diabetes (p≤0.05) and hypertension (p≤0.05).

Clinical features and grouped biomarkers are displayed in Table 1 IL-1β was the only biomarker which significantly differentiated by gender, with higher levels in males (p≤0.01) who have a higher mortality. 9 Biomarkers of COVID-19 severity did not differ by number of comorbidities or smoking status.

All the biomarkers in figure 1 were analysed for their correlations with WHO COVID-19 severity scores and arranged into a heat map (See Figure 1 ). Biomarkers that significantly correlated with the WHO severity score included CRP (R=0.53, p=2.6e -12 ), IL-6 (R 0.49, p = 1.3e -13 ), LDH (R 0.38, p < 2.1e -5 ) and IL-10 (R 0.35, p = 2.5e -7 ). IL-6 and CRP were most significantly indicative of the level of respiratory support (Figure 2a and 2b) with IL-6 being marginally better at differentiating respiratory requirement from none. Il-6 and CRP were both closely associated with increased fraction of inspired oxygen delivery (FiO2) requirements (R=0.54 and 0.58 respectively) and radiological evidence of acute respiratory distress syndrome (ARDS) (IL-6≤0.00001, CRP ≤0.005). A cut-off of 3.27pg/ml for IL-6 in this assay gives a sensitivity 0.87 and specificity 0.64 for requirement of intubation. Similarly, a 37mg/l cut-off for CRP has a sensitivity of 0.91 and specificity of 0.66. TNF-α also correlated with WHO severity (p≤0.05) and level of respiratory support (p≤0.01) but to a lesser degree than IL-6 and CRP.

Renal failure leading to a requirement for renal dialysis was determined by higher levels of TNF-α (mean 5.87 vs 4.7pg/ml, p<0.01), CRP (162 vs 82mg/L, p<0.01), IL-6 (1.1 vs 0.6pg/ml, J o u r n a l P r e -p r o o f Journal Pre-proof p<0.01) and IL-10 (3.9 vs 2.44pg/ml, p<0.05). IL-1β did not correlate with disease severity measures, presence of ARDS or level of respiratory support.

Neurological presentations with COVID-19 were also common, with cases presenting with delirium (12), ischemic stroke (5), intracerebral haemorrhage (2) to the proinflammatory cytokine milieu, inherent to severe disease and reflect a functioning immune system. 16 The binding of COVID-19 to the Toll Like Receptor (TLR) causes release of pro-IL-1β which is cleaved by caspase-1, followed by inflammasome activation and production of active IL-1β which is a mediator of lung inflammation, fever and fibrosis. IL-1β has been shown to be raised in COVID-19 infection compared to healthy controls by some. 4 This study however does not provide any indication that IL-1β levels reflect severity of disease. 4, 17 Despite this, initial trials of the IL-1β antagonist Anakinra appear to reduce the need for mechanical ventilation and mortality in severe cases. 18 However no studies have studied the influence of the drug on pre-and post-treatment levels of IL-1β, or stratified entry to Anakinra on IL-1β levels. Our study would indicate that non-IL-1β biomarkers might be of more practical utility.

The association of TNF-α, CRP and IL-10 with dialysis requirement without any association with IL-6 possibly supports a role for sepsis driving renal failure. In post-mortem studies microthrombi are seen in both the kidney and alveolar blood vessels, but the primary insult on those two tissues may be different.

Neurological cases had lower levels of COVID-19 disease severity markers, which may be accounted for in part by selection bias, where patients were hospitalised due to neurological disease with concurrent, less severe COVID-19 infection compared to those admitted primarily due to COVID-19 respiratory failure. The primary targets for SARS-CoV-2 COV-2 are respiratory epithelial cells and alveolar macrophages. Following infection of respiratory cells, rapid production and release of cytokines and chemokines occur. Macrophages are in turn activated, as are other key components of the innate immune system, such as dendritic cells, which leads to a more extensive immune response initiating the cytokine storm. 19 It is therefore conceivable from our findings that only severe pulmonary infection and infiltration results in cytokine activation and proliferation, and neurological features are either a consequence or coincidence J o u r n a l P r e -p r o o f of COVID-19 infection in this cohort. However more research is required to understand the neurological manifestations of COVID-19 and the associated cytokine response to more definitively explain such findings.

C-reactive protein is unsurprisingly very strongly correlated with levels of IL-6, as IL-6 drives the production of CRP primarily in the liver. CRP may potentially suffice in isolation to predict high risk cases requiring more aggressive intervention, negating the requirement of more specialist cytokine tests. Similarly, LDH, released from multiple tissues on cell death and activated through T-cell proliferation may be a practical and easily measured surrogate biomarker although its utility here is lower than the CRP. Levels of ferritin were raised across the cohort likely reflecting heightened acute phase response to infection as opposed to suprahigh levels indicative of sHLH. Absolute lymphocyte count and subsets have been shown to inversely correlate with severity in COVID-19, 16, 20, 21 although not demonstrated in this study.

Lymphocyte counts may be a less reliable marker due to the variety of factors which cause lymphocytosis such as sepsis from concomitant infections and the stress response. Although select routine laboratory markers correlate with disease severity, cytokine analysis provides additional predictive support for prognosis and likely interventions in the COVID-19 cytokine storm as well as reassurance that when low, little escalation of intervention may be required.

Note that a limitation of the study is that 16 cases were deemed not appropriate for intensive care, therefore escalation of respiratory support would be precluded regardless of clinical severity. We conducted a sensitivity analysis and inclusion/exclusion of these cases had no effect on biomarker correlations with severity.

Biomarker analysis of CRP, LDH and the cytokines IL-6, IL-10 and TNFα, alongside thorough clinical assessment of COVID-19 patients, enables more accurate stratification of high from low risk cases and the need for intensive care support. Such stratification enhances management not Heatmap shows biochemical signatures for each patient, arranged by WHO COVID-19 severity score. Visualisation was performed using the pheatmap R package. Biomarkers values are log10 transformed, centred and scaled. No patients had WHO severity scores 0, 1 or 2 as these are non-hospitalised patients. A score of 3 is hospitalised patients requiring no oxygen therapy; a score of 4 requires oxygen therapy; 5 requires non-invasive ventilation (NIV); of 6 requires intubation and mechanical ventilation; and 7 requires ventilation and additional organ support including vasopressors, renal replacement therapy and ECMO. A score of 8 is death. Higher levels of CRP, IL-6, IL-10 LDH and TNF-α are associated with higher WHO COVID-19 severity scores. 

The Role of Cytokines including Interleukin-6 in COVID-19 induced Pneumonia and Macrophage Activation Syndrome-Like Disease

Report on COVID-19 in Critical Care

Rapid Risk Assessment -Coronavirus Disease 2019 COVID-19 in the EU/EEA and the UK -Eighth Update

Clinical features of patients infected with 2019 novel coronavirus in Wuhan

Level of IL-6 predicts respiratory failure in hospitalized symptomatic COVID-19 patients. medRxiv

Interleukin-6 in COVID-19: A Systematic Review and Meta-Analysis. medRxiv

European center for disease prevention and control. Case Definition for Coronavirus Disease 2019 (COVID-19), as of 29

World Health Organisation. Blueprint Novel Coronavirus COVID-19 Therapeutic Trial Synopsis

OpenSAFELY: factors associated with COVID-19 death in 17 million patients

Prognostic value of interleukin-6, C-reactive protein, and procalcitonin in patients with COVID-19

Detectable serum SARS-CoV-2 viral load (RNAaemia) is closely J o u r n a l P r e -p r o o f Journal Pre-proof correlated with drastically elevated interleukin 6 (IL-6) level in critically ill COVID-19 patients

High IL-6/IFN-γ ratio could be associated with severe disease in COVID-19 patients

Treatment of severely ill COVID-19 patients with anti-interleukin drugs (COV-AID): A structured summary of a study protocol for a randomised controlled trial

A prospective, randomised, double blind placebo-controlled trial to evaluate the efficacy and safety of tocilizumab in patients with severe COVID-19 pneumonia (TOC-COVID): A structured summary of a study protocol for a randomised controlled trial

Relationships among Lymphocyte Subsets, Cytokines, and the Pulmonary Inflammation Index in Coronavirus (COVID-19) Infected Patients

Analysis of Clinical Features of 29 Patients With

Anakinra for severe forms of COVID-19: a cohort study

May the Central Nervous System Be Fogged by the Cytokine Storm in COVID-19?: an Appraisal

Lymphocyte Subset Counts in COVID-19 Patients: A Meta-Analysis

The clinical course and its correlated immune status in COVID