The pandemic has raised awareness of the cytokine storm and exposed the potential value of diagnostic tools for assessing the host cytokine immune response. These tools could make a difference not only in COVID-19 but also in other areas of healthcare.

One of the most notable clinical findings during the early pandemic was that hospitalised patients with COVID-19 had elevated levels of cytokine immune messengers, in particular interleukin 6 (IL-6), in their blood.

Along with other blood tests, this suggested that these patients could be experiencing a “cytokine storm” – a clinical syndrome that arises when immune-signalling cytokine levels go off the charts, giving rise to an out-of-control and destructive immune response.

This raised the hope that drugs could be used to calm the cytokine storm. A further question was whether the storm might be predicted in order to reduce the severity of COVID-19 and deaths.

The clinical world was intrigued. Trials were organised to attempt to blockade the ill effects of IL-6 and other inflammation-promoting, or “pro-inflammatory”, cytokines. Diagnostics companies have developed assays, and clinicians have come away with the expectation that cytokine measurement and targeted drugs will become available to help them care for critically ill patients. Indeed, an IL-6 receptor antibody (tocilizumab) is now recommended by the NHS and the NIH for COVID-19 therapy whilst other anti-IL-6 therapies and a selection of anti-inflammatory agents are being trialled [1].

Yet, a broader look at what we know about cytokines and the host immune response suggests that detailed clinical insight in this area is unlikely to come from single cytokine measurement alone.

Measurement of multiple cytokines simultaneously, whether from blood or other biological samples and implemented in a tool for rapid diagnostic testing, could have wider clinical application, not only in COVID-19 patients but many others, for example in sepsis, which contributes to about half of all hospital deaths [2], and in the growing field of cell therapy for cancer.

What are cytokines?

Cytokines are a broad range of immune-modulating cell signalling molecules.  There are a large number of different cytokines and they can be peptides, small proteins and glycoprotein. They are secreted by nearly all cells, but primarily by immune cells and endothelial cells and act as messengers of both the innate (first line of defence) and adaptive (acquired) immune system. They are usually found at low picomolar concentrations in the circulation but may rise a thousand-fold or more in response to infection.

Release of cytokines can be stimulated by physiological processes or as a result of immune insults such as attack by viral or microbial pathogens. Cytokines then influence local or systemic events, with some acting on the same cells that they are secreted from, whilst others are secreted into the blood and affect cells and organs at a distant site.  

The cytokine signalling network is exceedingly complex: a single cytokine can exert different cellular effects and have multiple biological activities. Multiple cytokines can exert a similar effect, and cytokines can also work together synergistically in inducing a reaction, or antagonistically to inhibit one cytokine’s effects by another as part of the fine-tuning of inflammatory responses. 

There are several groups of cytokines and they are named according to their functional properties and site of production. The aforementioned interleukins (IL) are a group of cytokines largely expressed from leukocytes and play important roles in the regulation of cell differentiation of immune cells. Other examples include the interferons (denoted by INF), which mediate response to viral attacks, tumour necrosis factor (TNF) with prominent roles in cellular differentiation, and chemokines (CC, CXC) which are vital in cell migration by attracting cells by chemotaxis.

What is a cytokine storm?

Whilst the immune system is there to protect us, imbalances in the cytokine response can also activate processes that are harmful to the body [3]. Sometimes cascading activation of immune cells drives an overproduction of cytokines, with some pro-inflammatory cytokines known to reach several 1000-fold of healthy concentrations in the blood.

When this happens, the immune system may start to attack the body itself. Common features are immune cells engulfing blood cells and attacking healthy tissues, blood vessels becoming leaky, blood clotting cascades being activated, and such a cytokine storm can ultimately lead to multi-organ failure and death.

In the context of immunotherapy this phenomenon is also known as cytokine release syndrome (CRS). This adverse side effect of immunotherapy was first described a decade ago and the management of CRS is complex and remains a significant clinical challenge [4].

Cytokine release syndrome has been recognised in a range of conditions. Notably, sepsis has long been understood as a systemic inflammatory response syndrome. Cytokine release can also occur following trauma, and other cytokine storm syndromes are triggered by rheumatic conditions, blood cancers or infections.

In terms of clinical definition, currently there is no unified definition of cytokine storms and these syndromes are largely diagnosed by looking for clinical criteria and blood measurements of generic inflammation markers. Indeed, early in the pandemic clinicians attempted to apply existing criteria to diagnose the ‘cytokine storm’ in COVID-19 – without success [5].

While the focus is often on the extreme pro-inflammatory response, it has long been known that cytokine storm syndromes involve a range of cytokines, including “anti-inflammatory” cytokines, and display considerable interindividual variation, which complicates any clinical interpretation of one-off measurements of single cytokines [6, 7].

Accordingly, a longstanding question is whether, instead of measuring a single biomarker, combinations of multiple cytokines may be more effective as diagnostic tools – by gauging the effect of a combination of mediators that amplify the storm.

What are the current sensing technologies?

Testing for cytokines is generally carried out in centralised clinical laboratories by means of immunoassays, several of which have gained emergency use authorisation (EUA) during the pandemic, such as Roche’s Elecsys IL-6 test or Beckman Coulter’s Access IL-6 [8, 9]. 

This approach leads to turn-around times of hours to days, while some cytokines can reach high systemic values within hours. This means that results are not typically available in a manner that would allow for near-time clinical decision making and offer a picture of disease progression in the clinic. As a result of these technological limitations, most studies have focused on intermittent sampling of a small number of cytokines in the systemic circulation.

During the pandemic, some multi-cytokine measurement systems have been pressed into service in the hope of identifying cytokine imbalances that may be targeted with drugs as well as predictive signatures and a growing number of commercial laboratories now offer cytokine research panels with up to dozens of different cytokines [10].

The majority of existing cytokine diagnostics measure cytokines from processed blood – plasma or serum – bringing with it the challenges of blood collection and processing. However, cytokines have been detected in biological fluids other than blood (such as sweat, in the eye and interstitial fluid), which may ultimately provide a route towards less invasive and more continual detection approaches. Already, we are seeing a gradual focal shift in point-of-care testing with an increasing number of inflammatory markers on the test menu and integrating non-invasive measurement would help accelerate this [11].

Syndromes associated with cytokine storm


A common clinical threat of cytokine storm exists in sepsis, where bacterial and viral infection can lead to an uncontrolled immune response that extends to the entire body and eventually escalates into the heterogeneous clinical syndrome “sepsis”. As the host response evolves, the immune system may enter a state of persistent immuno-paralysis. This is now thought to be the overriding immune dysfunction responsible for morbidity and mortality, in particular by opening the door to secondary and opportunistic infections. 

While attempts to target the hyperinflammatory response with drugs to improve sepsis outcomes have overwhelmingly yielded disappointing results, clinical studies are now showing promise in predicting sepsis through a combination of multiple cytokine measurements and machine learning [12, 13]. Efforts to demonstrate near-time multiplex cytokine monitoring are also underway, and this could help to achieve the early diagnosis of sepsis in the clinic by monitoring critically ill patients [14, 15]. Both approaches could aid earlier antibiotic treatment, which currently is the main effective treatment.

Recent transcriptomics research suggests that the clinical syndrome is constituted by several endotypes (subtypes of conditions). The focus now is on defining distinct host response subgroups and establishing endotype-driven, targeted treatment strategies, facilitated by multiplex cytokine monitoring at the point of care [16, 17].

Cytokine Intext


Early in the pandemic, the available testing pointed to IL-6 and pro-inflammatory cytokines as the key mediators of the cytokine storm. However, recent research, through measurement of a range of cytokines, has shown that the immune response to COVID-19 infection is heterogeneous and complex, with many patients entering a state of severe, infection-induced immuno-suppression [18].

In fact, only a minority of COVID-19 patients display very high levels of IL-6 and a sustained storm of pro-inflammatory cytokines. This is a pattern also observed in some cases of sepsis [19].

Multiplex cytokine measurement studies reveal that the host immune response clusters into distinct early response types with cytokine profiles that can act as early predictive markers for disease severity and outcome. Better understanding of the molecular landscape brings promise for building information for more targeted treatment [20, 21].

Interestingly, endotypes observed in sepsis also appear to be recapitulated in COVID-19 – perhaps making it a “virus-induced sepsis” or suggesting that host response syndromes cluster in common ways across disease aetiologies, which may allow development of common treatment strategies [22].


CRS is also a common and serious side effect of immunotherapy, such as monoclonal antibody therapeutics and CAR-T cell therapy. In this context where the timing of the immune insult is known, cytokine blockade shows some success, and an IL-6 receptor antagonist (tocilizumab, approved for rheumatoid arthritis over a decade ago) is used to manage severe CRS. 

However, clinicians would like to be able to predict which patients are likely to develop severe and potentially life-threatening CRS in the days to weeks after immunotherapy. Standard laboratory markers, such as CRP, and even IL-6, only peak after patients become unwell.

A further challenge for clinicians is to distinguish between patients with sepsis due to an infection and those with CRS following CAR-T cell therapy. This is important because treatment for CRS can mask some of the signs of infection and thereby delay diagnosis and appropriate treatment. 

Initial studies suggest that both these clinical challenges are addressable through prediction and improved diagnosis, respectively, on the basis of multiplex cytokine measurement [23, 24].

What could we do with multiplex cytokine measurement?

The experience with COVID-19 has put the spotlight firmly on the diagnostic utility of multiplex cytokine measurement. Indeed, “the lack of diagnostic tools to evaluate immune function” has been a key reason why it has taken time to understand the immune response in severe COVID-19 [21]. 

In the clinical syndromes discussed above, research indicates that signatures with a relatively small number of cytokines could be sufficient to predict severe immune insults and identify distinct host response profiles. This would be an improvement over the inflammatory markers that are currently commonly used in the clinic, which do not show the same discriminatory power or which peak too late to be clinically useful [12,18,24]. 

Beyond diagnosis of the cytokine storm, rapid multiplex cytokine profiling could also facilitate host immune response monitoring. This is seen as promising in conditions like sepsis and COVID-19, where ill-timed immuno-suppressant treatment appears to further compromise the immune status of patients. We could expect this to help define the window of opportunity during which blockade of specific cytokines has the best chance of success. 

Additionally, cytokine profiling can yield broader insights, for example about the nature of the causative agent, and could thus offer direct host response diagnostics [25, 26, 27].

In the longer term, rapid multiplex cytokine measurement could be used to target particular cytokines that may spike early in a disease, or support the development of therapies that target combinations of cytokines. 

The number of FDA-approved drugs aimed at circulating proteins has doubled to 34 in recent years. Cytokines constitute both the lion’s share of the targets of approved drugs (11 cytokines) and of agents in development (31 cytokines) in this area. Still, seeing that there are in the order of 100 cytokines, there remains scope to expand the cytokine-specific medicine chest [28].  

We can envision that rapid multiplex cytokine measurement could enable identification of cytokine storms and a precision medicine-like approach to the assessment and treatment of clinical syndromes, not entirely unlike what has already happened with biomarker-guided site-agnostic therapy for cancer.


01. Tocilizumab in patients admitted to hospital with COVID-19 (RECOVERY): a randomised, controlled, open-label, platform trial. The Lancet. Vol. 397, pp. 1637-1645 (2021)

02. Hospital Deaths in Patients With Sepsis From 2 Independent Cohorts. JAMA. Vol. 312, pp. 90-92 (2014)

03. Cytokine Dysregulation and Acute Graft-Versus-Host Disease. Blood. Vol. 80, pp. 2964-2968 (1992)

04. IN VIVO CELL ACTIVATION FOLLOWING OKT3 ADMINISTRATION.Transplantation. Vol. 49, pp. 697-702 (1990)

05. Preliminary predictive criteria for COVID-19 cytokine storm. Annals of Rheumatic Diseases. Vol. 80, pp. 88-95 (2020)

06. Sepsis biomarkers: a review. Critical Care. Vol. 14, R15 (2010)

07. Role of Cytokines as a Double-edged Sword in Sepsis. In Vivo. Vol. 27, pp. 669-684 (2013)

08. Roche Elecsys® IL-6

09. Beckman Coulter Access IL-6 Assay

010. Protein Simple Ella

011. Eurolyser Cube-S

012. Presymptomatic Prediction of Sepsis in Intensive Care Unit Patients. Clinical and Vaccine Immunology. Vol. 15, 1089-1094 (2008)

013. Combining Biomarkers with EMR Data to Identify Patients in Different Phases of Sepsis. Scientific Reports. Vol. 7, 10800 (2017)

014. Multiplexed cytokine detection using electrochemical point-of-care sensing device towards rapid sepsis endotyping. Biosensors and Bioelectronics. Vol. 171, 112726 (2021)

015. Machine learning-based cytokine microarray digital immunoassay analysis. Biosensors and Bioelectronics. Vol. 180, 113088 (2021)

016. Multiplexed cytokine detection using electrochemical point-of-care sensing device towards rapid sepsis endotyping. Biosensors and Bioelectronics. Vol. 171, 112726 (2021)
Precision Immunotherapy for Sepsis. Frontiers in Immunology. Vol. 9, p. 1926 (2018)

017. Severe immunosuppression and not a cytokine storm characterizes COVID-19 infections. JCI Insight. Vol. 5, e140329 (2020)

018. Distinct inflammatory profiles distinguish COVID-19 from influenza with limited contributions from cytokine storm. Science Advances. Vol. 6, no. 50, eabe3024 (2020)

019. Discrimination of COVID-19 from inflammation-induced cytokine storm syndromes by disease-related blood biomarkers. Arthritis & Rheumatology. (First published 20 April, 2021)

020. Longitudinal analyses reveal immunological misfiring in severe COVID-19. Nature. Vol. 584, pp 463–469 (2020)

021. Validation of Inflammopathic, Adaptive, and Coagulopathic Sepsis Endotypes in Coronavirus Disease 2019. Critical Care Medicine. Vol. 49, pp. e170-e178 (2021)

022. Identification of Predictive Biomarkers for Cytokine Release Syndrome after Chimeric Antigen Receptor T-cell Therapy for Acute Lymphoblastic Leukemia. Cancer Discovery. Vol. 6, pp. 664–679 (2016)

023. Identification of Predictive Biomarkers for Cytokine Release Syndrome after Chimeric Antigen Receptor T-cell Therapy for Acute Lymphoblastic Leukemia. Cancer Discovery. Vol. 6, pp. 664–679 (2016)

024. Diagnostic biomarkers to differentiate sepsis from cytokine release syndrome in critically ill children. Immunobiology and Immunotherapy. Vol. 4, pp. 5174-5183 (2020)

025. Inflammatory Mediator Profiles Differ in Sepsis Patients With and Without Bacteremia. Frontiers in Immunology. Vol. 9, p. 691 (2018)

026. Cytokine profile in severe gram-positive and gram-negative abdominal sepsis. Scientific Reports. Vol. 5, 11355 (2015)

027. Soluble ligands as drug targets. Nature Reviews Drug Discovery. Vol. 19, pp 695–710 (2020)

028. Screening and identification of a six-cytokine biosignature for detecting TB infection and discriminating active from latent TB. Journal of Translational Medicine. Vol. 16, 206 (2018)

Keiko Yata