10 Cytokine Storm

The n e w e ng l a n d j o u r na l of m e dic i n e
Review Article
Dan L. Longo, M.D., Editor
Cytokine Storm
David C. Fajgenbaum, M.D., and Carl H. June, M.D.
T
he severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) From the Department of Medicine, Divi-
pandemic has reminded us of the critical role of an effective host immune sion of Translational Medicine and Human
Genetics, Center for Cytokine Storm Treat-
response and the devastating effect of immune dysregulation. This year ment and Laboratory (D.C.F.), and the
marks 10 years since the first description of a cytokine storm that developed after Center for Cellular Immunotherapies and
chimeric antigen receptor (CAR) T-cell therapy1 and 27 years since the term was the Parker Institute for Cancer Immuno-
therapy (C.H.J.), Perelman School of Med-
first used in the literature to describe the engraftment syndrome of acute graft- icine, University of Pennsylvania, Phila-
versus-host disease after allogeneic hematopoietic stem-cell transplantation.2 The delphia. Address reprint requests to Dr.
term “cytokine release syndrome” was coined to describe a similar syndrome after Fajgenbaum at davidfa@pennmedicine
.
upenn .
edu or to Dr. June at cjune@
infusion of muromonab-CD3 (OKT3).3 Cytokine storm and cytokine release syn- upenn.edu.
drome are life-threatening systemic inflammatory syndromes involving elevated
N Engl J Med 2020;383:2255-73.
levels of circulating cytokines and immune-cell hyperactivation that can be trig- DOI: 10.1056/NEJMra2026131
gered by various therapies, pathogens, cancers, autoimmune conditions, and Copyright © 2020 Massachusetts Medical Society.
monogenic disorders.
From a historical perspective, cytokine storm was previously referred to as an
influenza-like syndrome that occurred after systemic infections such as sepsis and
after immunotherapies such as Coley’s toxins.4 Yersinia pestis infection (i.e., the
plague) has led to major pandemics (e.g., the Black Death) and triggers alveolar
macrophages to produce excessive amounts of cytokines, resulting in cytokine
storm.5 An exaggerated immune response was suspected to contribute to the lethal-
ity of the 1918–1919 influenza pandemic. In fact, a reconstructed H1N1 virus
isolated from the 1918 pandemic, as compared with common reference strains of
the virus that causes influenza A, triggered marked pulmonary inflammation in
mice.6 Recognition that the immune response to the pathogen, but not the patho-
gen itself, can contribute to multiorgan dysfunction and that similar cytokine
storm syndromes could occur with no obvious infection led to the investigation of
immunomodulators and cytokine-directed therapies. One of the earliest targeted
therapies for abrogation of a cytokine storm was the anti–interleukin-6 receptor
monoclonal antibody tocilizumab, which was developed for the treatment of idio-
pathic multicentric Castleman’s disease in the 1990s. A host of other disorders
have been described as causes of cytokine storm and targeted with immune-di-
rected therapies, such as sepsis, primary and secondary hemophagocytic lympho-
histiocytosis (HLH), autoinflammatory disorders, and coronavirus disease 2019
(Covid-19).
No single definition of cytokine storm or the cytokine release syndrome is
widely accepted, and there is disagreement about how these disorders differ from
an appropriate inflammatory response. The National Cancer Institute’s definition,
based on the Common Terminology Criteria for Adverse Events (CTCAE), is too
broad, since the criteria for an inflammatory syndrome can also apply to other
physiological states, and the definition of the American Society for Transplanta-
tion and Cellular Therapy is based on criteria that focus too specifically on iatro-
genic causes of cytokine storm alone.7 Although cytokine storm is easy to identify
in disorders with elevated cytokine levels in the absence of pathogens, the line
between a normal and a dysregulated response to a severe infection is blurry, es-
n engl j med 383;23 nejm.org December 3, 2020 2255

The New England Journal of Medicine
Downloaded from nejm.org by MARIO FRANCISCO GUERRERO on March 30, 2022. For personal use only. No other uses without permission.
Copyright © 2020 Massachusetts Medical Society. All rights reserved.
pecially considering that certain cytokines may In severe cases of cytokine storm, renal fail-
be both helpful in controlling an infection and ure, acute liver injury or cholestasis, and a stress-
harmful to the host. The interdependence of related or takotsubo-like cardiomyopathy can
these inflammatory mediators further compli- also develop.9 The combination of renal dysfunc-
cates the distinction between a normal and a tion, endothelial-cell death, and acute-phase hypo
dysregulated response. albuminemia can lead to capillary leak syn-
It is important for the clinician to recognize drome and anasarca — changes that are similar
cytokine storm because it has prognostic and to those observed in patients with cancer who
therapeutic implications. In this review, we pro- are treated with high-dose interleukin-2.10 Neu-
pose a unifying definition of cytokine storm; rologic toxicity associated with T-cell immuno-
discuss the pathophysiological features, clinical therapy is referred to as immune effector cell–
presentation, and management of the syndrome; associated neurotoxicity syndrome or cytokine
and provide an overview of iatrogenic, pathogen- release syndrome–associated encephalopathy.7
induced, neoplasia-induced, and monogenic The neurologic toxic effects are often delayed,
causes. Our goal is to provide physicians with a developing several days after the onset of the
conceptual framework, a unifying definition, cytokine storm.
and essential staging, assessment, and therapeu- The laboratory findings in cytokine storm are
tic tools to manage cytokine storm. variable and influenced by the underlying cause.
Nonspecific markers of inflammation such as
C-reactive protein (CRP) are universally elevated
Cl inic a l Fe at ur e s a nd
L a bor at or y A bnor m a l i t ie s and correlate with severity.11 Many patients have
hypertriglyceridemia and various blood-count
Cytokine storm is an umbrella term encompass- abnormalities, such as leukocytosis, leukopenia,
ing several disorders of immune dysregulation anemia, thrombocytopenia, and elevated ferritin
characterized by constitutional symptoms, sys- and d-dimer levels. Changes in circulating cell
temic inflammation, and multiorgan dysfunc- counts are most likely due to a complex interplay
tion that can lead to multiorgan failure if inad- among cytokine-induced changes in production
equately treated (Fig. 1). The onset and duration and mobilization of cells from the bone marrow,
of cytokine storm vary, depending on the cause immune-mediated destruction, and chemokine-
and treatments administered.7 Although the ini- induced migration. Prominent elevations in serum
tial drivers may differ, late-stage clinical mani- inflammatory cytokine levels, such as inter
festations of cytokine storm converge and often feron-γ (or CXCL9 and CXCL10, chemokines in-
overlap. Nearly all patients with cytokine storm duced by interferon-γ), interleukin-6, interleu-
are febrile, and the fever may be high grade in kin-10, and soluble interleukin-2 receptor alpha,
severe cases.8 In addition, patients may have fa- a marker of T-cell activation, are usually present.
tigue, anorexia, headache, rash, diarrhea, arthral- Highly elevated serum interleukin-6 levels are
gia, myalgia, and neuropsychiatric findings. found in CAR T-cell therapy–induced cytokine
These symptoms may be due directly to cytokine- storm and several other cytokine storm disorders.8
induced tissue damage or acute-phase physio- The approach to evaluating a patient with
logical changes or may result from immune- cytokine storm should accomplish the following
cell–mediated responses. Cases can progress three main goals: identifying the underlying dis-
rapidly to disseminated intravascular coagulation order (and ruling out disorders that may mimic
with either vascular occlusion or catastrophic cytokine storm), establishing severity, and deter-
hemorrhages, dyspnea, hypoxemia, hypotension, mining the clinical trajectory. A complete workup
hemostatic imbalance, vasodilatory shock, and for infection, as well as laboratory assessment of
death. Many patients have respiratory symptoms, kidney and liver function, should be performed
including cough and tachypnea, that can prog- in all suspected cases of cytokine storm. Mea-
ress to acute respiratory distress syndrome surements of inflammatory acute-phase bio-
(ARDS), with hypoxemia that may require me- markers, such as CRP and ferritin, and blood
chanical ventilation. The combination of hyper- counts should be obtained, since they correlate
inflammation, coagulopathy, and low platelet with disease activity. Arterial blood-gas measure-
counts places patients with cytokine storm at ment should be performed if the respiratory evalu-
high risk for spontaneous hemorrhage. ation warrants it. Cytokine profiles may be helpful
2256 n engl j med 383;23 nejm.org December 3, 2020

Cytokine Storm
Lungs Nervous system

• Pneumonitis • Confusion
• Pulmonary edema • Delirium
• Dyspnea, hypoxemia • Aphasia
• ARDS • Seizures
Liver Constitutional symptoms

• Hepatomegaly • Fever
• Elevated liver enzymes • Anorexia
• Increased hepcidin • Fatigue
• Hypoalbuminemia
• Liver injury
• Cholestasis Heart
• Liver failure
• Hypotension
• Tachycardia
Kidneys • Cardiomyopathy
• Acute renal dysfunction or injury

• Renal failure
Rheumatologic system
• Vasculitis
Vascular and lymphatic systems • Arthritis, arthralgia
• Cytopenia, anemia, leukocytosis
• Coagulopathy
Gastrointestinal system
• Hyperferritinemia, increase in other
acute-phase reactants (e.g., CRP, • Nausea
D-dimer)
• Vomiting
• Elevated cytokines (e.g., interleukin-1, • Diarrhea
interleukin-6, interferon-γ) and growth
factors (e.g., VEGF) • Ascites
• Endothelial damage and vascular
permeability
• Capillary leak syndrome Skin
• Vasodilatory shock • Rash
• Spontaneous hemorrhage • Edema
• Lymphadenopathy
Figure 1. Clinical Presentation of Cytokine Storm.

A wide range of clinical and laboratory abnormalities can be observed in cytokine storm. However, all cases involve elevated circulating
cytokine levels, acute systemic inflammatory symptoms, and secondary organ dysfunction (often renal, hepatic, or pulmonary). ARDS
denotes acute respiratory distress syndrome, CRP C-reactive protein, and VEGF vascular endothelial growth factor.
in determining the trend from baseline values, tients with septicemia. Unfortunately, it is diffi-
although these findings are typically not avail- cult to distinguish cytokine storm due to sepsis
able soon enough to include as part of the im- from cytokine storm due to CAR T-cell therapy
mediate workup or to guide treatment decisions. on the basis of clinical features alone. Levels of
Establishing the disorder underlying the cyto- serum cytokines — most prominently, inter-
kine storm can be challenging. Cytokine storm feron-γ — are often more elevated in patients
is not a diagnosis of exclusion, and it can en- with cytokine storm due to CAR T-cell therapy
compass many disorders. For example, patients than in patients with sepsis-induced cytokine
may have both sepsis and cytokine storm. How- storm, who often have higher levels of circulat-
ever, it is important to distinguish between cyto- ing interleukin-1β, procalcitonin, and markers
kine storm due to an iatrogenic cause such as of endothelial damage.12 Thus, combinations of
CAR T-cell therapy and cytokine storm due to assays to rule out infection and measure serum
systemic infection, since immunosuppressive cytokines can help to identify the cause of the
treatments could be detrimental if used in pa- cytokine storm. However, CAR T-cell therapy

and other noninfectious causes can also occur Figure 2 (facing page). Pathophysiological Features
with infections, and infections can develop dur- of Cytokine Storm.
ing the course of therapy, so continued monitor- Cytokine storm can occur as a result of inappropriate
ing for infections is warranted. Disorders that recognition (e.g., in hypersensitivity) or ineffective recog-
should be ruled out in considering cytokine nition with immune evasion (e.g., in Epstein–Barr virus
storm include anaphylaxis and physiological re- [EBV]–associated hemophagocytic lymphohistiocytosis
[HLH]), an inappropriate response with an exaggerated
sponses to microbial infections. effector response and cytokine production (e.g., in chi-
The grading systems used to predict and as- meric antigen receptor [CAR] T-cell therapy) or an ineffec-
sess the severity of cytokine storm differ accord- tive response due to immune evasion (e.g., in sepsis),
ing to the cause. Serum biomarkers, including or failure to terminate homeostasis or return to homeo-
stasis (e.g., in HLH). Examples of drugs that can inhibit
glycoprotein 130 (gp130), interferon-γ, and inter-
signaling pathways are shown in boxes. Covid-19 denotes
leukin-1–receptor antagonist (IL1RA), can be used coronavirus disease 2019, CS cytokine storm, IL1RA inter-
to predict the severity of cytokine storm induced leukin-1–receptor antagonist, IP-10 interferon-inducible
by CAR T-cell therapy,13 with a separate grading protein 10, JAK-STAT3 Janus kinase–signal transducer
scale used to assess the current severity.7 HScore and activator of transcription 3, MAPK mitogen-activated
protein kinase, MCP-1 monocyte chemotactic protein 1,
and MS score are used for classifying HLH-asso- MIP-1α macrophage inflammatory protein 1α, mTOR
ciated cytokine storm, and HLH-2004 guides mammalian target of rapamycin, NF-κB nuclear factor κB,
treatment. For the grading of cytokine storm due TNF tumor necrosis factor, and Tregs regulatory T cells.
to other causes, the immune systems disorders
section of CTCAE is used (https://ctep.cancer.gov/
protocolDevelopment/electronic_applications/docs/ Immune hyperactivation in cytokine storm
CTCAE_v5_Quick_Reference_5x7.pdf). can occur as a result of inappropriate triggering
or danger sensing, with a response initiated in
the absence of a pathogen (e.g., in genetic disor-
Pathoph ysiol o gic a l Fe at ur e s
of C y t ok ine S t or m ders involving inappropriate inflammasome ac-
tivation or idiopathic multicentric Castleman’s
Inflammation involves a set of biologic mecha- disease); an inappropriate or ineffective ampli-
nisms that evolved in multicellular organisms to tude of response, involving excessive effector
contain invasive pathogens and resolve injuries immune-cell activation (e.g., in cytokine storm
by activating innate and adaptive immune re- due to CAR T-cell therapy), an overwhelming
sponses. The immune system is expected to pathogen burden (e.g., in sepsis), or uncontrolled
recognize foreign invaders, respond proportion- infections and prolonged immune activation
ally to the pathogen burden, and then return to (e.g., in HLH associated with Epstein–Barr virus
homeostasis. This response requires a balance [EBV]); or failure to resolve the immune re-
between sufficient cytokine production to elimi- sponse and return to homeostasis (e.g., in pri-
nate the pathogen and avoidance of a hyperin- mary HLH) (Fig. 2). In each of these states, there
flammatory response in which an overabundance is a failure of negative feedback mechanisms that
of cytokines causes clinically significant collat- are meant to prevent hyperinflammation and
eral damage. Cytokines play a key role in coor- the overproduction of inflammatory cytokines
dinating antimicrobial effector cells and provid- and soluble mediators. The excessive cytokine
ing regulatory signals that direct, amplify, and production leads to hyperinflammation and mul-
resolve the immune response. Cytokines have tiorgan failure. Regulatory cell types, decoy re-
short half-lives, which normally prevents them ceptors for proinflammatory cytokines such as
from having effects outside lymphoid tissue and IL1RA, and antiinflammatory cytokines such as
sites of inflammation. Although typically con- interleukin-10 are important for antagonizing
sidered to be pathologic, sustained production of inflammatory-cell populations and preventing
cytokines that leads to elevated circulating levels immune hyperactivity.
may be necessary to appropriately control some Given the lack of a unifying definition for
disseminated infections. At increased levels, cyto- cytokine storm14 and disagreement about the
kines can have systemic effects and cause col- distinction between cytokine storm and a physi-
lateral damage to vital organ systems. ologic inflammatory response, we propose the

Cytokine Storm
Driver cells
• Adaptive T cells (CD4+, CD8+)
Iatrogenic causes Pathogen-induced triggers • Innate antigen-presenting cells
(e.g., CAR T-cell therapy, (e.g., bacterial sepsis, (macrophages, dendritic cells)
blinatumomab, other T-cell engaging EBV-associated HLH, Covid-19)
immunotherapies, gene therapies).
Monogenic and
autoimmune disorders Cancer
(e.g., autoinflammatory disorders,
primary or secondary HLH)
Negative regulators Immune hyperactivation

• Tregs, mesenchymal stem cells Inadequate • Inappropriate triggering or danger sensing
• Decoy cytokine receptors (IL1RA) negative
• Inappropriate or ineffective amplitude
• Antiinflammatory regulation
of response
cytokines (interleukin-10) • Failure to resolve inflammation
and return to homeostasis
Cytokine storm
Chemokines Interleukin-18 Interleukin-1β Complement
(e.g., MCP-1, MIP-1α) activation
(e.g., sepsis-induced CS) Tocilizumab
TNF
(anti–interleukin-6
Emapalumab Interferon-γ receptor antibody)
1 Pathologically Interleukin-6 Interleukin-17
sustained cytokine (e.g., CAR T-cell Infliximab
production leading to therapy–induced CS)
excessive circulating Anakinra Siltuximab
cytokine levels
IP-10 (anti–interleukin-1
anti–interleukin-1
receptor antagonist)
Prolonged activation of signaling pathways
MAPK NF-κB JAK-STAT3 mTOR
Glucocorticoids JAK inhibitors Sirolimus
2 Acute systemic inflammatory effects
Cytokine-induced 3 Secondary organ dysfunction (e.g., hepatic, renal, pulmonary)

“collateral damage” a Dysfunction due to inflammation beyond normal response to pathogen
(if pathogen present)
b Cytokine-driven dysfunction (if no pathogen present)
Multiorgan failure
Death

T-cell effector Th1 Th2 Th9 Th17 Cytotoxic

subgroups CCR1 CCR3 CCR3 CCR6 T lymphocyte
CCR5 CCR4 CCR6 CXCR4 CXCR3 T-bet

T-bet GATA-3 PU.1 RORγT
CD4 Eomes
CXCR3 CCR6 CXCR3 CD8
CD4 CD4 CD4
Cell products Interleukin-4 Interleukin-17 Interferon-γ

(cytokines, Interleukin-2 Interleukin-9
Interleukin-5 Interleukin-21 Perforin
chemokines, Interferon-γ Interleukin-10
enzymes) Interleukin-13 Interleukin-22 Granzyme
Recruitment Recruitment Recruitment Recruitment Cytotoxicity
Macrophages Eosinophils
Cell targets Mast cells Neutrophils Virally infected
Dendritic cells Basophils cells
Figure 3. T-Cell Effector Subgroups Involved in Cytokine Storm.

The master transcription factors (T-bet, GATA-3, PU.1, RORγT, and eomesodermin [eomes]), effector molecules, and cell targets are shown
for the following T-cell subgroups: types 1, 2, 9, and 17 helper T (Th1, Th2, Th9, and Th17, respectively) cells and cytotoxic T lymphocytes.
following three criteria for identifying cytokine may be appropriate for controlling the infection
storm: elevated circulating cytokine levels, acute if excessive secondary organ dysfunction does
systemic inflammatory symptoms, and either not occur, whereas similarly high levels of cyto-
secondary organ dysfunction (often renal, hepat- kines in cancer-associated HLH or idiopathic
ic, or pulmonary) due to inflammation beyond multicentric Castleman’s disease would be con-
that which could be attributed to a normal re- sidered a pathologic state of cytokine storm be-
sponse to a pathogen (if a pathogen is present), cause no pathogen requiring an immune re-
or any cytokine-driven organ dysfunction (if no sponse is involved and patients benefit from
pathogen is present). Improvement in outcomes treatment with cytokine neutralization and other
with cytokine neutralization or antiinflamma- antiinflammatory agents. Circulating cytokine
tory agents further supports the pathologic role levels can be difficult to measure because cyto-
of excessive cytokines and the classification of a kines have short half-lives, circulating levels may
condition as a cytokine storm. However, lack of not accurately reflect local tissue levels, and
a treatment response does not necessarily rule measurements may not be easily obtained world-
out cytokine storm, because underlying condi- wide. We do not propose a specific threshold for
tions are likely to play a part, a different cyto- elevations in cytokine levels above the normal
kine may be the disease driver, or the timing of range, and we do not recommend specific cyto-
treatment may have been poor. kine panels or list particular cytokines whose
In short, cytokine storm involves an immune levels must be elevated, given the lack of avail-
response that causes collateral damage, which able evidence. However, we believe that this is an
may be greater than the immediate benefit of important area for future research and could
the immune response. Thus, an exuberant in- benefit from systematic assessment by a multi-
flammatory response to a large pathogen burden disciplinary consortium.

Cytokine Storm
Cell Types Involved in Cytokine Storm 17 helper T (Th17) cells recruit neutrophils.18 An
The cells of the innate immune system are the exaggerated Th1-type inflammatory response
first line of defense against pathogens. Neutro- often occurs during cytokine storm. Th1 cells
phils, monocytes, and macrophages recognize produce large quantities of interferon-γ, induce
pathogens, produce cytokines, and engulf patho- delayed hypersensitivity reactions, activate mac-
gens and cells by phagocytosis. There are many rophages, and are essential for defense against
other innate immune cells, such as dendritic intracellular pathogens.19 Iatrogenic causes of
cells, gamma–delta T cells, and natural killer cytokine storm involving excessive T-cell activa-
(NK) cells.15 Innate immune cells use pattern- tion, such as CAR T-cell and anti-CD28 antibody
recognition receptors, which are not specific for therapy, point to the ability of activated T cells
any particular antigen, to recognize and respond to initiate cytokine storm. Impaired granule-
to a wide variety of microbial invaders by pro- mediated killing of infected cells or tumor cells
ducing cytokines that activate cells of the adap- by CTLs is a key aspect of some forms of cyto-
tive immune system. kine storm.20 Data from mouse models of HLH
Innate cells that are most often implicated in and patients with cytokine storm indicate that
the pathogenesis of cytokine storm include neu- the inability of CTLs to kill efficiently leads to
trophils, macrophages, and NK cells. Neutro- prolonged activation of T cells, triggering a cas-
phils can produce neutrophil extracellular traps, cade of inflammatory tissue damage.21-23 Th17
a network of fibers that contribute to thrombi cells have a major role in host defense, particu-
formation and amplify cytokine production dur- larly antifungal protection, and abnormal Th17-
ing cytokine storm. Macrophages, which are cell function can lead to autoimmunity.24 An
tissue-resident cells that are often derived from experimental model of macrophage activation
circulating monocytes, do not divide; they have syndrome (a form of secondary HLH) provides
diverse functions, from the removal of senescent evidence that Th17 cells can be drivers of a cyto-
cells by engulfment, to tissue repair and immu- kine storm that is independent of interferon-γ.25
noregulation, to antigen presentation. In many B cells are not often associated with the patho-
forms of cytokine storm, macrophages become genesis of cytokine storm. However, the effec-
activated and secrete excessive amounts of cyto- tiveness of B-cell depletion in treating some cyto-
kines, ultimately causing severe tissue damage kine storm disorders, such as human herpesvirus
that can lead to organ failure. Hemophagocytic 8 (HHV-8)–associated multicentric Castleman’s
macrophages are often observed in bone mar- disease, suggests that these cells are capable of
row biopsy specimens from patients with cyto- initiating or propagating cytokine storm, particu-
kine storm. Interferon-γ can induce hemophago- larly when virally infected.
cytosis by macrophages, and this may contribute
to the cytopenias commonly observed in patients Cytokines
with cytokine storm.16 The cytolytic function of As noted above, the recognition of cytokine
NK cells is diminished in some forms of cyto- storm as an entity is relatively recent. The advent
kine storm, which can lead to prolonged anti- of molecular cloning technologies led to the
genic stimulation and difficulty resolving inflam- discovery of the panoply of cytokines and chemo-
mation.17 Excess interleukin-6 may mediate the kines involved in cytokine storm (Table 1); the
impairment in NK-cell function by lowering per- realization that diverse entities can cause cyto-
forin and granzyme production. kine storm (Table 2) also contributed to its rec-
The adaptive immune system is composed of ognition. The administration of recombinant
B cells and T cells. T cells differentiate into a cytokines (e.g., interleukin-1, interleukin-6, inter-
number of subsets with distinct effector-cell leukin-12, interleukin-18, tumor necrosis factor
functions potentially involved in cytokine storm [TNF], and interferon-γ) in animal models and
(Fig. 3). Type 1 helper T (Th1) cells and cyto- for cancer treatment in humans induces severe
toxic T lymphocytes (CTLs) are primarily respon- toxic effects or lethality consistent with the cen-
sible for the host defense against viral infec- tral role of cytokines as mediators of hyperin-
tions. Th1 cells regulate the recruitment of flammation in cytokine storm.27-29 Conversely,
macrophages, whereas type 2 helper T (Th2) reduction in symptoms and improvement in or-
cells recruit eosinophils and basophils, type 9 gan function with neutralization of specific cyto-
helper T (Th9) cells recruit mast cells, and type kines with monoclonal antibodies also reveal

Table 1. Soluble Mediators in Cytokine Storm.*
Mediator Main Cell Source Type and Function

Cytokines and growth
factors
Interleukin-1 Macrophages, epithelial cells; pyroptotic cells Proinflammatory alarmin cytokine; pyrogenic function, macrophage
and Th17 cell activation
Interleukin-2 T cells Effector T-cell and regulatory T-cell growth factor
Interleukin-6 Macrophages, T cells, endothelial cells Proinflammatory cytokine; pyrogenic function, increased antibody
production, induction of acute-phase reactants
Interleukin-9 Th9 cells Protection from helminth infections, activation of mast cells,
association with type I interferon in Covid-1926
Interleukin-10 Regulatory T cells, Th9 cells Antiinflammatory cytokine; inhibition of Th1 cells and cytokine
release
Interleukin-12 Dendritic cells, macrophages Activation of the Th1 pathway; induction of interferon-γ from Th1
cells, CTLs, and NK cells; acting in synergy with interleukin-18
Interleukin-17 Th17 cells, NK cells, group 3 innate lymphoid Promoting neutrophilic inflammation, protection from bacterial and
cells fungal infections
Interleukin-18 Monocytes, macrophages, dendritic cells Proinflammatory alarmin cytokine; activation of Th1 pathway, acting
in synergy with interleukin-12
Interleukin-33 Macrophages, dendritic cells, mast cells, Proinflammatory alarmin cytokine; amplification of Th1 and Th2
epithelial cells cells, activation of NK cells, CTLs, and mast cells
Interferon-γ Th1 cells, CTLs, group 1 innate lymphoid Proinflammatory cytokine; activation of macrophages
cells, and NK cells
Tumor necrosis factor Macrophages, T cells, NK cells, mast cells Increasing vascular permeability; pyrogenic function
GM-CSF Th17 cells Proinflammatory cytokine
VEGF Macrophages Angiogenesis
Chemokines
Interleukin-8 (CXCL8) Macrophages, epithelial cells Recruitment of neutrophils
MIG (CXCL9) Monocytes, endothelial cells, keratinocytes Interferon-inducible chemokine; recruitment of Th1 cells, NK cells,
plasmacytoid dendritic cells
IP-10 (CXCL10) Monocytes, endothelial cells, keratinocytes Interferon-inducible chemokine; recruitment of macrophages, Th1
cells, NK cells
MCP-1 (CCL2) Macrophages, dendritic cells, cardiac Recruitment of Th2 cells, monocytes, dendritic cells, basophils
myocytes
MIP-1α (CCL3) Monocytes, neutrophils, dendritic cells, NK Recruitment of macrophages, Th1 cells, NK cells, eosinophils,
cells, mast cells dendritic cells; pyrogenic function
MIP-1β (CCL4) Macrophages, neutrophils, endothelium Recruitment of macrophages, Th1 cells, NK cells,
dendritic cells
BLC (CXCL13) B cells, follicular dendritic cells Recruitment of B cells, CD4 T cells, dendritic cells†
Plasma proteins
CRP Hepatocytes Monomeric CRP increases interleukin-8 and MCP-1 secretion;
interleukin-6 increases CRP expression
Complement Hepatocytes, other cells Complement activation contributes to tissue damage in cytokine
storm; complement inhibition can reduce immunopathologic
effects of cytokine storm
Ferritin Ubiquitous Primary site of iron storage in cells
* BLC denotes B-lymphocyte chemoattractant; Covid-19 coronavirus disease 2019; CRP C-reactive protein; CTLs cytotoxic T lymphocytes;
CXCL C-X-C motif chemokine ligand; GM-CSF granulocyte–macrophage colony-stimulating factor; IP-10 interferon-inducible protein 10;
MCP-1 monocyte chemoattractant protein 1; MIG monokine induced by interferon-γ; MIP-1α and MIP-1β macrophage inflammatory pro-
tein 1α and 1β, respectively; NK natural killer; Th1, Th2, Th9, and Th17 cells types 1, 2, 9, and 17 helper T cells, respectively; and VEGF
vascular endothelial growth factor.
† In idiopathic multicentric Castleman’s disease, the levels of CXCL13 are the most elevated of all the cytokines or chemokines.

Cytokine Storm
Table 2. Clinical Causes of Cytokine Storm, Pathologic Drivers, and Therapeutic Approaches.*
Type of Cytokine Storm Pathologic Cellular or Cytokine

and Trigger Cause Driver Common Therapeutic Approaches
Iatrogenic
CAR T-cell therapy Infusion of CAR T cells Macrophages, CAR T cells, inter- Anti–interleukin-6 antibody, gluco-
leukin-6, interleukin-1β corticoids
Blinatumomab Infusion of CD19- and CD3-specific Activated T cells, macrophages, Anti–interleukin-6 antibody, gluco-
T-cell receptor–engaging antibody interleukin-6 corticoids
Pathogen-induced
Bacterial sepsis Hematogenous bacterial infection Heterogeneous and multifacto- Intravenous antibiotics
rial drivers
EBV-associated HLH EBV infection in patient with genetic Interferon-γ, TNF, CD8+ T cells B-cell–depleting therapy, glucocor-
susceptibility ticoids
HHV-8–associated MCD HHV-8 infection in patient with HIV Viral interleukin-6, interleukin-6 B-cell–depleting therapy
coinfection, genetic susceptibility,
or both
Covid-19 SARS-CoV-2 infection, potentially in a Unknown driver Glucocorticoids
susceptible person
Monogenic and auto
immune
Primary HLH Germline mutation in genes regulating CD8+ T cells, interferon-γ T-cell inhibition or ablation, inter
granule-mediated cytotoxicity feron-γ inhibitor, glucocorticoids
Secondary HLH, or MAS Viral cause (EBV or CMV), autoim- CD8+ T cells, interferon-γ, Treatment of the underlying cause,
mune disorder (rheumatoid arthri- interleukin-1β, myeloid-cell in addition to T-cell inhibition or
tis or adult-onset Still’s disease), or autoinflammation ablation, interleukin-1β inhibitor,
neoplastic disorder in patient with JAK1 and JAK2 inhibitors, gluco-
genetic susceptibility (lymphoma) corticoids
Autoinflammatory Germline mutations in genes regulat- Innate cells, TNF, interleukin-1β Anti-TNF antibody, anti–interleukin-1
disorders ing the innate immune system and antibody
inflammasome activation
Idiopathic MCD Unknown cause Interleukin-6, activated T cells, Anti–interleukin-6 antibody, siroli-
mTOR mus, cyclosporine, cytotoxic
chemotherapy, glucocorticoids
* CAR denotes chimeric antigen receptor, CMV cytomegalovirus, Covid-19 coronavirus disease 2019, EBV Epstein–Barr virus, HHV-8 human
herpesvirus 8, HIV human immunodeficiency virus, HLH hemophagocytic lymphohistiocytosis, JAK1 Janus kinase 1, JAK2 Janus kinase 2,
MAS macrophage activation syndrome, MCD multicentric Castleman’s disease, mTOR mammalian target of rapamycin, and SARS-CoV-2
severe acute respiratory syndrome coronavirus 2.
that excessive levels of certain cytokines play a ed by activated T cells and NK cells and is a
critical role in a number of cytokine storm dis- potent activator of macrophages. Clinically,
orders. interferon-γ causes fever, chills, headache, dizzi-
A complex, interconnected network of cell ness, and fatigue.31 Emapalumab, a monoclonal
types, signaling pathways, and cytokines is in- antibody that binds interferon-γ, was recently
volved in cytokine storm disorders. Interferon-γ, approved for the treatment of cytokine storm in
interleukin-1, interleukin-6, TNF, and interleu- patients with primary HLH.32 This agent may
kin-18 are key cytokines that often have elevated also be useful in other cytokine storm disorders,
levels in cytokine storm and are thought to have such as macrophage activation syndrome or CAR
central immunopathologic roles. The pattern of T-cell–associated cytokine storm, although in the
cytokine elevations varies on the basis of such latter case, it may diminish antitumor effects.
factors as the microbiome, genetic features, and Fever, a clinical hallmark of cytokine storm,
underlying disorders.30 The specific immune cells can be elicited by interleukin-1, interleukin-6, or
that secrete the various cytokines are not fully TNF through distinct mechanisms. Interleukin-1
understood and most likely vary among cytokine is encoded by two genes (IL1A and IL1B), both of
storm disorders. Interferon-γ is primarily secret- which bind to the same interleukin-1 receptor,

activating a cascade of intracellular signaling thelial cells. This results in systemic hyperin-
pathways, including nuclear factor κB (NF-κB). flammation involving secretion of monocyte
The interleukin-1–receptor antagonist anakinra chemoattractant protein 1 (MCP-1), interleukin-8,
is effective as a single agent and in combination and additional interleukin-6, as well as increased
with other agents for the treatment of some vascular endothelial growth factor (VEGF) and
forms of cytokine storm.33,34 reduced E-cadherin expression on endothelial
Levels of interleukin-6, an important mediator cells, which contribute to vascular hyperperme-
of the acute inflammatory response and patho- ability, leakiness, hypotension, and pulmonary
physiological features of cytokine storm, are high- dysfunction.38
ly elevated across various underlying immuno- TNF is a potent, multifunctional, proinflam-
pathologic disorders35,36 and in mouse models of matory cytokine that belongs to the TNF–TNF
cytokine storm.37 Both tocilizumab, a monoclo- receptor superfamily. In addition to inducing
nal antibody directed at the interleukin-6 recep- fever, augmenting systemic inflammation, and
tor (interleukin-6R), and siltuximab, which neu- activating antimicrobial responses such as inter-
tralizes interleukin-6 directly, have been shown leukin-6, TNF can induce cellular apoptosis and
to be effective in a number of cytokine storm regulate immunity. TNF and other cytokines in
disorders, including HLH, idiopathic multicentric the TNF–TNF receptor superfamily are potent
Castleman’s disease, and CAR T-cell–induced inducers of NF-κB, leading to the expression of
cytokine storm.38 multiple proinflammatory genes. In mouse mod-
Interleukin-6 is one of the more complex cyto- els of toxic shock, TNF is the cytokine driver of
kines, since it is produced by and acts on immune superantigen-driven cytokine storm.39 The effec-
and nonimmune cells across multiple organ sys- tiveness of anti-TNF therapies in certain auto-
tems. It can signal through two main pathways, inflammatory-driven cytokine storm conditions
referred to as classic cis signaling and trans points to their potential role in the treatment of
signaling.38 The membrane-bound interleukin-6R cytokine storm, but the limitations and dangers
does not possess intracellular signaling domains of anti-TNF therapies in patients with sepsis in-
but signals instead through interaction with dicate that more work is needed.
membrane-bound gp130. In cis signaling, soluble Interleukin-18 is a member of the large inter-
interleukin-6 binds to membrane-bound interleu- leukin-1 family40 that has recently been associated
kin-6R, forming an interleukin-6–interleukin-6R with cytokine storm disorders. Interleukin-18 and
complex that binds to gp130, which then initi- interleukin-1β are activated from precursors by
ates signaling through its intracellular domain. inflammasomes. The inflammasome is a multi-
Downstream signal transduction is mediated by molecular cytosolic sensor that detects patho-
JAKs (Janus kinases) and STAT3 (signal transducer genic microorganisms and sterile stressors and
and activator of transcription 3), as well as by Akt– activates caspase-1 during the process of pyrop-
mTOR (mammalian target of rapamycin) and tosis, which, in turn, causes the inactive precur-
MAPK–ERK (mitogen-activated protein kinase– sor forms of interleukin-1β and interleukin-18 to
extracellular signal-regulated kinase) pathways. become the active forms.41,42 Macrophages and
Membrane-bound gp130 is ubiquitously expressed, dendritic cells are the primary sources of bioac-
whereas expression of membrane-bound inter- tive interleukin-18, which has many proinflam-
leukin-6R is restricted largely to immune cells. matory effects. Most important, it synergizes with
Activation of cis signaling results in pleiotropic interleukin-12 or interleukin-15 to stimulate se-
effects on the immune system, which can con- cretion of interferon-γ from T cells and NK cells,
tribute to cytokine storm.38 In the presence of and thus promotes Th1-type inflammatory re-
high circulating levels of interleukin-6, which can sponses. The interleukin-18 receptor is constitu-
be present in cytokine storm, trans signaling tively expressed on NK cells and induced on
occurs through the binding of interleukin-6 to activation in most T cells. Interleukin-1β and
the soluble form of interleukin-6R, forming a interleukin-18 are also potent inducers of inter-
complex with a gp130 dimer on potentially all leukin-6 secretion from macrophages.43
cell surfaces. The resultant interleukin-6–soluble Patients with cytokine storm due to macro-
interleukin-6R–gp130–JAK-STAT3 signaling is phage activation syndrome have high levels of
then activated in cells that do not express the interleukin-18 in serum,44 and interleukin-18 is a
membrane-bound interleukin-6R, such as endo- biomarker of severity that correlates with hyper-

Cytokine Storm
ferritinemia, elevated aminotransferase levels, cytokine storm. Although some studies suggest
and disease flare.45 The proinflammatory effects that the driver cytokines are released by CAR T
of interleukin-18 are normally kept in check by cells, resulting in a positive feedback loop of T-cell
the interleukin-18–binding protein (IL18BP), which activation and inflammatory cytokine release,51
prevents the binding of interleukin-18 to its re- recent studies in mice suggest that the cytokines
ceptor.46 The ratio of free interleukin-18 to bound and factors mediating the severity of cytokine
interleukin-18–IL18BP complexes in serum is an storm are produced not by the CAR T cells but by
important indicator of the severity of the macro- macrophages and can be reversed by interleukin-6
phage activation syndrome.44,47 Tadekinig alfa is and interleukin-1 blockade.52-54 Tumor lysis most
a recombinant IL18BP currently under investiga- likely also contributes to the cytokine storm
tion as a treatment for hyperinflammation. through the induction of pyroptosis in target
Chemokines are a class of cytokines that con- cells.55 Since interleukin-6 blockade is highly ef-
tribute to a variety of immune-cell functions, fective at reversing symptoms and organ dys-
including leukocyte recruitment and trafficking. function in most patients, it is the likely cyto-
Dysregulated trafficking during inflammation kine driver of cytokine storm induced by CAR
may have a role in hyperinflammation. Numerous T-cell therapy. Glucocorticoids and interleukin-1
regulatory cytokines such as interleukin-10 and inhibition can also be effective in the treatment
natural cytokine antagonists such as IL1RA serve of this type of cytokine storm.
as buffers to limit systemic off-target effects. Cytokine storm can be observed with other
Interleukin-10 inhibits the production of TNF, T-cell–engaging immunotherapies as well, such as
interleukin-1, interleukin-6, and interleukin-12 blinatumomab, a bispecific antibody that binds
and down-regulates antigen presentation. Further- to CD19+ and CD3+ T cells.56 Like CAR T cells,
more, in mice lacking interleukin-10, infection activated T cells initiate the cytokine storm, and
leads to cytokine storm.48 Though interleukin-10 macrophage activation propagates blinatumomab-
and IL1RA are often elevated in cytokine storm, induced cytokine storm, which also responds to
this finding most likely reflects a secondary, anti–interleukin-6 antibody therapy.36 The unfor-
albeit insufficient, counterregulatory response tunate consequences of another T-cell–activating
to the proinflammatory cytokines. Anakinra is a treatment with the anti-CD28 superagonist
therapeutic agent that mimics the endogenous TGN1412 show that rapid activation of large
immunoregulatory effects of IL1RA. numbers of T cells can result in severe cytokine
Plasma proteins such as complement proteins storm within minutes after infusion.57 However,
and other inflammatory mediators can contrib- cytokine storm does not develop in all patients
ute to the pathogenesis of cytokine storm. These treated with CAR T cells or blinatumomab, so
soluble proteins recognize pathogens, amplify additional factors, such as CAR structure and
cellular responses, and provide feedback on cyto- design,51 disease burden,58 and host genomic
kine signaling. In fact, cytokines can enhance background,59 are likely to play a part. In a re-
the production of complement proteins, which in cent study of NK-cell CAR therapy, there were no
turn can enhance or inhibit cytokine production. reported cases of cytokine storm or even elevated
Thus, complement can be highly effective in interleukin-6 levels,60 possibly because of lower
eliminating microbes but can also cause collat- interleukin-6 production by NK cells than by T cells
eral damage if excessive. Hypocomplementemia, and different cross-talk with myeloid cells. Addi-
resulting from increased consumption by im- tional iatrogenic causes of cytokine storm include
mune complexes, can be observed in cytokine rituximab,35 gene therapies, immune checkpoint
storm.49 Complement inhibitors are under evalua- inhibitors, cardiac-bypass surgery,61 and alloge-
tion for the treatment of cytokine storm disorders. neic stem-cell transplantation, as well as bioterror-
ism agents such as staphylococcal enterotoxin B
and Francisella tularensis.
I at ro genic C y t ok ine S t or m
Infusion of CAR T cells engineered to recognize Patho gen-Induced C y t ok ine
and eliminate CD19+ lymphoma cells can induce S t or m
cytokine storm, with supraphysiologic levels of
interferon-γ and interleukin-6.50 The highly acti- Cytokine storm can also result from naturally
vated CAR T cells are clearly the initiators of the occurring microbial infections. Though data on

relative frequencies are limited, infections are cytokine storm occurs in these patients from
most likely the most common trigger of cyto- defective perforin-mediated cytolysis that leads
kine storm. Distinguishing between appropriate to prolonged engagement between lymphocytes
cytokine production for controlling a wide- and antigen-presenting cells and defective clear-
spread infection and excessive cytokine produc- ance of antigen-bearing dendritic cells, resulting
tion is challenging. Disseminated bacterial in- in continuous activation and proliferation of
fections causing sepsis induce the production of T cells and macrophages, hemophagocytosis, and
many cytokines that can lead to fever, cell death, an autocrine loop of proinflammatory cyto-
coagulopathies, and multiorgan dysfunction. The kines.21,65-67 Furthermore, retrospective analyses
collateral damage caused by the immune re- of data from persons who died from coagulopa-
sponse as it attempts to clear the pathogen can thies and hemophagocytosis during the H1N1
be more deadly than the pathogen itself. Certain influenza pandemic of 2009 revealed germline
bacteria, including streptococcus species and mutations previously associated with HLH-asso-
Staphylococcus aureus, can produce superantigens ciated cytokine storm.30 Thus, the pathogen ini-
that cross-link the major histocompatibility com- tiates and T-cell activation propagates cytokine
plex and T-cell receptors, leading to polyclonal storm in patients with a genetic susceptibility.
activation of T cells, cytokine production, and Cyclosporine and anti–interleukin-6 receptor
toxic shock syndrome. Superantigens are the monoclonal antibody therapy can be effective in
most powerful T-cell mitogens, and bacterial some virus-driven forms of HLH-associated cyto-
superantigen concentrations of less than 0.1 pg kine storm, indicating the critical role of T-cell
per milliliter are sufficient to stimulate T cells in activation and interleukin-6.
an uncontrolled manner, resulting in fever, shock, Another pathogen-induced form of cytokine
and death. storm is HHV-8–associated multicentric Castle-
In sepsis-associated cytokine storm, it is un- man’s disease. In this disorder, uncontrolled
clear which immune cell types and cytokines may infection with HHV-8 (also known as Kaposi’s
be responsible for propagating the pathologic sarcoma herpesvirus) leads to a cytokine storm
hyperinflammation. Antibiotics are the main- driven primarily by excessive production of human
stay of treatment. The administration of mono- interleukin-6 and viral interleukin-6 by HHV-8–
clonal antibodies directed at specific cytokines infected plasmablasts.68 Patients with HHV-8–
and the use of apheresis or medical devices to associated multicentric Castleman’s disease are
remove cytokines from circulation have had gen- immunocompromised as a result of human im-
erally disappointing results in clinical trials.62 munodeficiency virus infection or a genetic
Although the timing of treatment in these stud- susceptibility, making it difficult to control the
ies may have contributed to the lack of benefit, HHV-8 infection, which is a common, typically
additional host or pathogen factors may be im- asymptomatic infection in the general popula-
portant, beyond the specifically elevated cyto- tion.69 A recent study showed that the effect of
kine levels. For example, reanalysis of a negative tocilizumab in patients with HHV-8–associated
trial of interleukin-1β blockade in patients with multicentric Castleman’s disease was minimal
sepsis identified a subgroup of patients with ele- and short-lived, most likely because of viral in-
vated ferritin levels who seemed to benefit from terleukin-6 signaling that was independent of
the treatment.63 the neutralized interleukin-6 receptor.70 As with
Disseminated viral infections can also induce EBV-associated HLH,71 rituximab is highly effec-
profound cytokine storm. Patients with hyperin- tive in patients with HHV-8–associated multi-
flammatory responses to microbes often have centric Castleman’s disease, since B-cell deple-
defects in pathogen detection, effector and regu- tion removes the primary reservoir for HHV-8.72
latory mechanisms, or resolution of inflamma- Many additional microbes can trigger cytokine
tion. For example, patients lacking functional storm, including other herpesviruses, such as
perforin, which is critical for resolving infec- herpes simplex virus, and other influenza viruses,
tions and inflammation, have prolonged CD8+ such as H5N1.
T-cell production of interferon-γ and TNF, and Targeted treatment is more challenging in
HLH-associated cytokine storm develops in such patients with viral infections than in patients
patients when they are infected with EBV or cyto- with bacterial infections, since fewer antiviral
megalovirus.64 Experimental models suggest that agents are available. Intravenous immune globu-

Cytokine Storm
lin and convalescent plasma are sometimes used Autoinflammatory diseases are characterized
to help control the pathogen and provide benefi- by seemingly unprovoked inflammation and cyto-
cial immunomodulation. For some viral infec- kine storm without signs of infection or autoim-
tions, treating patients with proinflammatory munity. Affected patients have germline muta-
cytokines in the early stages of infection can tions in genes regulating the innate immune
help to control the virus before detrimental ef- system and activation of the inflammasome.
fects of the immune response occur.73 Several genetic disorders are associated with al-
tered regulation of the innate immune system,
including familial Mediterranean fever (MEFV),
Mono genic or Au t oim mune
C y t ok ine S t or m TNF receptor–associated periodic syndrome
(TNFRSF1A), hyperimmunoglobulinemia D with
In rare cases, a pathogen triggers cytokine storm periodic fever syndrome (MVK), familial cold
in patients with monogenic disorders, and in autoinflammatory syndrome (NLRP3), the Muckle–
other cases, cytokine storm has autoimmune, Wells syndrome (NLRP3), neonatal-onset multi-
neoplastic, or idiopathic causes. In patients with system inflammatory disease (NLRP3), deficiency
primary HLH, various autosomal recessive mono- of ADA2 (CECR1), NLRC4 inflammasomopathies,
genic abnormalities in granule-mediated cyto- X-linked lymphoproliferative type 2 disorder
toxicity lead to cytokine storm. Common patho- (XIAP), the Takenouchi–Kosaki syndrome (CDC42),
logic mutations include those occurring in PRF1, and the Wiskott–Aldrich syndrome (CDC42). Al-
UNC13D, STXBP1, RAB27A, STX11, SH2D1A, XIAP, though all patients with these disorders have
and NLRC4.23 In patients with secondary HLH, periodic fevers, only a portion have cytokine
viral, autoimmune, or neoplastic disorders trig- storm. Given the primary genetic defects and the
ger cytokine storm, and such patients often have effective treatments that are available, innate cells
heterozygous polymorphisms in the same genes are most likely the primary cell drivers involved,
that are altered in primary HLH.65,74 Elevated and TNF, interleukin-1, interleukin-18, or a com-
levels of interferon-γ, TNF, interleukin-1, inter- bination of these cytokines probably drives
leukin-4, interleukin-6, interleukin-8, interleu- pathogenesis. Patients with genetic immunode-
kin-10, CXCL9, CXCL10, and interleukin-18 are ficiency syndromes such as chronic granuloma-
frequently associated with HLH. Anti–inter tous disease and STAT1 gain-of-function disease
feron-γ antibody therapy with emapalumab has can, paradoxically, present with cytokine storm
recently been approved for the treatment of pri- from overwhelming infections.78
mary HLH, as a bridge to allogeneic stem-cell Idiopathic multicentric Castleman’s disease is
transplantation, which is typically curative. another cytokine storm disorder that is similar
The beneficial effects of glucocorticoids, cyclo- to HHV-8–associated multicentric Castleman’s
sporine, anti–interleukin-1 antibody, JAK1 and disease, but the cause is unknown. Patients with
JAK2 inhibitors, anti–interleukin-6 antibody, and the thrombocytopenia, anasarca, fever, reticulin
cytotoxic chemotherapies in some patients with fibrosis, and organomegaly (TAFRO) subtype tend
primary or secondary HLH suggest that path- to have the most severe cytokine storm.79 Al-
ways targeted by these agents are key to patho- though the cause is unknown, interleukin-6 is
genesis. Cyclophosphamide and etoposide, which the driver of pathogenesis in a large portion of
are broadly cytotoxic but particularly effective at patients. As a result, tocilizumab, which targets
eliminating activated CD8+ T cells, are often effec- the interleukin-6 receptor, and siltuximab, which
tive in patients with primary HLH, secondary HLH targets interleukin-6 directly, were developed and
(including macrophage activation syndrome), and approved by regulatory agencies in Japan (tociliz
corresponding models.75 Etoposide also targets umab) and in the United States and dozens of
macrophages, including those involved in regu- other countries (siltuximab) for the treatment
lating inflammation, which could be harmful. of idiopathic multicentric Castleman’s disease.
Generalized T-cell and B-cell ablation with alem- Both siltuximab and tocilizumab have been
tuzumab and T-cell ablation with antithymocyte shown to resolve disease flares and sustain
globulin have been reported; ablation most likely remission in approximately one third to one half
works by depleting the pathogenic CD8+ T cells, of patients.80 However, some patients with low
among other cell types.76 Nonablative inhibition circulating interleukin-6 levels have a response
of T cells with cyclosporine can also be helpful.77 to interleukin-6 blockade, and some patients

with high systemic interleukin-6 levels do not survival.91 The relative frequencies of circulating
have a response. A seven-protein panel that can activated CD4+ and CD8+ T cells and plasma-
predict which patients with idiopathic multicen- blasts are increased in Covid-19.92 In addition to
tric Castleman’s disease are most likely to benefit the elevated systemic cytokine levels and acti-
from siltuximab was recently identified and vali- vated immune cells, several clinical and labor
dated (https://ashpublications.org/blood/article/132/ atory abnormalities, such as elevated CRP and
Supplement%201/3716/265269/Serum-Proteomics d-dimer levels, hypoalbuminemia, renal dysfunc-
-Reveals-Distinct-Subtypes?searchresult=1). tion, and effusions, are also observed in Covid-19,
Patients with idiopathic multicentric Castle- as they are in cytokine storm disorders. Labora-
man’s disease who have progressive organ dys- tory test results reflecting hyperinflammation
function and who do not have a response to and tissue damage were found to predict wors-
anti–interleukin-6 therapy are often treated with ening outcomes in Covid-19.93
combination cytotoxic chemotherapy to nonspe- Although immunologic dysregulation has been
cifically eliminate hyperinflammatory cells.81 Oth- observed in severe cases of Covid-19,26 it is not
er elevated serum cytokines and cellular signal- known whether immune hyperactivity or a fail-
ing pathways that could be considered for ure to resolve the inflammatory response because
therapeutic targeting include CXCL13, CXCL10 of ongoing viral replication or immune dysregu-
(interferon-inducible protein 10 [IP-10]), VEGF-A,82 lation underlies severe cases. The correlation
type I interferon,83 mTOR complex 1 (mTORC1),84 between the nasopharyngeal viral load and cyto-
and JAK-STAT3. These findings have led to treat- kine levels (e.g., interferon-α, interferon-γ, and
ment with the mTORC1 inhibitor sirolimus in TNF), as well as a declining viral load in moder-
patients with idiopathic multicentric Castleman’s ate but not severe cases, suggests that the im-
disease who do not have a response to anti–inter- mune response is positively associated with the
leukin-6 therapy.85 Sirolimus therapy is being viral burden.26 Alternatively, the discoveries of
evaluated in an ongoing clinical trial involving inborn errors of type I interferon immunity and
patients with active disease who do not yet have autoantibodies against type I interferons in the
fulminant cytokine storm (ClinicalTrials.gov num- most severe cases of Covid-19 suggest that an
ber, NCT03933904). inadequate antiviral response may be contribu-
tory in some patients with Covid-19.94,95 Host
immune responses and immune-related symp-
C ov id -19 –A sso ci ated C y t ok ine
S t or m toms are extremely variable between asymptom-
atic patients (who have effective control of SARS-
Covid-19, which is caused by SARS-CoV-2, is char- CoV-2) and patients with severe Covid-19 (who
acterized by heterogeneous symptoms ranging are unable to control the virus), which suggests
from mild fatigue to life-threatening pneumo- that host immune dysregulation contributes to
nia, cytokine storm, and multiorgan failure. pathogenesis in some cases. Another hypothe-
Cytokine storm was also reported in patients sized mechanism involves autoimmunity due to
with SARS and was associated with poor out- molecular mimicry between SARS-CoV-2 and a
comes.86 Although the mechanisms of lung in- self-antigen. These mechanisms may be involved
jury and multiorgan failure in Covid-19 are still in subgroups of patients, such as children with
under investigation,14 reports of hemophagocy- postinfection multisystem inflammatory syn-
tosis and elevated cytokine levels — as well as drome, a condition that seems to be ameliorated
beneficial effects of immunosuppressant agents by immunomodulatory therapies such as intra-
— in affected patients, particularly those who venous immune globulin, glucocorticoids, and
are the most severely ill, suggest that cytokine anti–interleukin-1 and anti–interleukin-6 thera-
storm may contribute to the pathogenesis of pies. Patients with multisystem inflammatory
Covid-19.87,88 syndrome very clearly meet the definition of cyto-
Serum cytokine levels that are elevated in pa- kine storm, since SARS-CoV-2 is no longer pres-
tients with Covid-19–associated cytokine storm ent; however, it is unclear whether the cytokine
include interleukin-1β, interleukin-6, IP-10, TNF, storm is a driver of Covid-19 or a secondary
interferon-γ, macrophage inflammatory protein process. Furthermore, it is now clear that pa-
(MIP) 1α and 1β, and VEGF.89,90 Higher interleu- tients with SARS-CoV-2 infection can be asymp-
kin-6 levels are strongly associated with shorter tomatic or can have acute Covid-19 with hetero-

Cytokine Storm
geneous severity, a chronic course of Covid-19, on the same platform across Covid-19–associated
or multisystem inflammatory syndrome. A criti- cytokine storm and other cytokine storm disor-
cal question concerns the factors that contribute ders, preliminary results suggest that circulating
to the severe cytokine storm–like phenotype ob- levels of several cytokines, such as interleukin-6,
served in a small fraction of patients. Coexisting as well as other inflammatory markers, such as
conditions such as hypertension, diabetes, and ferritin, are less severely elevated in Covid-19
obesity are associated with more severe cases of than in some of the other cytokine storm disor-
Covid-19, possibly because of the preexisting ders.26 Levels of inflammatory mediators in pul-
chronic inflammatory state or a lower threshold monary tissue during infection with SARS-CoV-2
for the development of organ dysfunction from remain unknown.
the immune response. Despite the many unknowns, a recent ran-
Several important differences in therapeutic con- domized, controlled trial showing that dexa-
siderations should be noted between Covid-19– methasone reduces mortality among the most
associated cytokine storm and many other cyto- severe cases of Covid-19, characterized by elevat-
kine storm disorders. First, cytokine storm ed CRP levels and supplemental oxygen require-
triggered by infection with SARS-CoV-2 may re- ments, and potentially worsens outcomes in
quire different therapies from those used for milder cases suggests that excessive, late-stage
cytokine storm due to other causes. Cytokines inflammation contributes to mortality.88 A meta-
may be both a key component of the cytokine analysis of seven randomized trials showed that
storm and an essential factor in the antimicro- 28-day all-cause mortality in critically ill patients
bial response. Thus, blocking cytokine signaling with Covid-19 was lower among those who were
may actually impair clearance of SARS-CoV-2, treated with glucocorticoids than among those
increase the risk of secondary infections, and who received usual care or placebo.100 An obser-
lead to worse outcomes, as seen with influenza vational study suggesting that patients with
virus.96 Since interleukin-6 and other cytokines Covid-19 have a good response to glucocorti-
are potentially critical for both a healthy re- coids when the CRP level is high but a poor re-
sponse to SARS-CoV-2 and a detrimental cyto- sponse when the level is low is consistent with
kine storm, it is particularly important that the these findings.101 Further support comes from
right subgroups of patients with Covid-19 are positive anecdotal reports of targeted antago-
selected for treatments at the right time. Despite nists against interleukin-1, granulocyte–macro-
positive anecdotal reports, two large, randomized, phage colony-stimulating factor, and JAK1 and
controlled trials of anti–interleukin-6 receptor JAK2 in patients with Covid-19.102-105 Likewise, the
antibody therapies did not show a survival benefit observation that proinflammatory agents such
in hospitalized patients with Covid-19.97,98 as inhaled interferon-β have a positive effect if
Second, the primary site of infection and dis- given early in the disease course is consistent
ease most likely contributes to differences in with a model in which immunostimulation that
immune responses and mechanisms underlying enhances antiviral activity is helpful early (and
the cytokine storm, which have implications for probably harmful late), whereas immunosuppres-
treatment. For example, selective elimination of sion is helpful late and harmful early. As with
the primary viral reservoir is beneficial in pa- dexamethasone, the timing of treatment and selec-
tients with HHV-8–associated multicentric Castle- tion of subgroups of patients included in studies
man’s disease but is not possible in patients with will most likely have an effect on outcomes.
Covid-19. Despite unknowns regarding the role of
Third, lymphopenia is not often observed in immune dysregulation and cytokine storm in
cytokine storm disorders, but it is a hallmark of Covid-19, hundreds of immunomodulatory drugs
severe Covid-19. It is currently unclear whether are currently under investigation.102 Many of these
the lymphopenia observed in Covid-19 is due to treatments have been used for other cytokine
tissue infiltration or destruction of lymphocytes. storm disorders. Canakinumab, an anti–inter
Fourth, clotting issues can occur across cyto- leukin-1β monoclonal antibody, and anakinra
kine storm disorders, but thromboembolic events are both being studied for Covid-19–induced
appear to be more frequent in Covid-19–associ- ARDS. Acalabrutinib, a selective inhibitor of
ated cytokine storm.99 Finally, although cytokine Bruton tyrosine kinase that regulates B-cell and
panels have not been measured simultaneously macrophage signaling and activation, may have

promise for dampening the hyperinflammatory simultaneously, an approach that may be effec-
response in Covid-19.106 JAK1 and JAK2 inhibi- tive for multiple diseases driven by different cyto-
tors, which are approved for the treatment of a kines. In addition, plasma exchange and plasma
number of autoimmune and neoplastic condi- filtration columns for the adsorption of cyto-
tions, have the potential to inhibit signaling kines are both under evaluation for cytokine
downstream of type I interferon, interleukin-6 storm disorders.
(and other gp130 family receptors), interferon-γ, It is important to consider several factors in
and interleukin-2, among other cytokines.107 Much managing cytokine storm. Neutralization of a
like anti–interleukin-6 antibody therapy, inhibi- particular cytokine whose level is elevated in the
tion of Bruton tyrosine kinase and JAK could circulation with an existing agent (anti–interleu-
prove to be damaging or unhelpful if given too kin-6, anti-TNF, anti–interferon-γ, or anti–inter
soon, when the immune response to SARS-CoV-2 leukin-1β antibody) will not always be effective,
is critical in controlling viral replication and and blocking a cytokine with a low or normal
clearance. circulating level can be effective if it is a key
component of the hyperinflammatory circuit or
if its level is potentially elevated in tissue. In ad-
Ther a peu t ic s
dition, the various therapies mentioned in this
The general treatment strategy for cytokine storm review have distinctive side-effect and risk pro-
involves supportive care to maintain critical or- files. All targeted agents have target-specific
gan function, control of the underlying disease risks, and combination therapy has more poten-
and elimination of triggers for abnormal im- tial risks than single-agent therapy. Furthermore,
mune system activation, and targeted immunopathologic hyperinflammation itself is an im-
modulation or nonspecific immunosuppression munodeficiency that can put patients at risk for
to limit the collateral damage of the activated infections, and immunosuppressive agents most
immune system. As noted throughout this review, likely increase the risk further. In this age of
a number of drugs are effective across multiple cytokine profiling and individualized medicine,
disorders under the cytokine storm umbrella patients must be monitored and given appropri-
and still more may be effective in multiple con- ate prophylaxis when treated empirically, and
ditions that have not yet been studied. randomized, controlled trials should always be
Given the growing number of new therapeu- performed to assess efficacy and safety.
tics targeting various aspects of the immune Advancing the research and treatment of cyto-
system and our ability to probe the biologic kine storm will require pooling of samples for
mechanisms of disease, further research should “omics” studies and collaboration among ex-
focus on the identification of drugs that can be perts across conditions. The introduction of an
used across cytokine storm disorders and preci- International Classification of Diseases, 10th Revision,
sion diagnostics for selecting the right drugs for code for cytokine release syndrome in 2021
the right patients, regardless of the underlying should facilitate electronic health record–based
condition.108,109 A study involving patients with research into its natural history, pathogenesis,
systemic juvenile idiopathic arthritis revealed sub- and treatments. Once sufficient scientific prog-
groups of patients with cytokine profiles in which ress has been achieved toward biomarker-guided,
interleukin-6 and interleukin-18 predominated, individualized treatment of cytokine storm, reli-
pointing toward available therapeutic approach- able, quick, and accessible assays will be needed
es.110 Likewise, biomarkers were recently shown to measure soluble mediators of inflammation
to effectively predict which patients with adult- in plasma and tissues.
onset Still’s disease would have a response to
anakinra or tocilizumab.111 The progress made Sum m a r y
in precision oncology suggests that similar ef-
forts across cytokine storm disorders are war- Mild, secondary organ dysfunction during an
ranted to identify specific therapeutic targets inflammatory response is evolutionarily accept-
and signatures of response to certain drugs that able if it allows the host to overcome the infec-
cross disease boundaries. JAK signaling is an tion and survive. If the inflammatory response
interesting target in cytokine storm, because causes excessive organ dysfunction that puts
multiple cytokine–receptor pairs can be targeted host survival and reproductive fitness at risk (in

Cytokine Storm
the absence of ventilatory support and dialysis), symptoms, and secondary organ dysfunction be-
then it is pathologic. Extensive regulatory mech- yond that which could be attributed to a normal
anisms exist that modulate the immune response response to a pathogen, if a pathogen is present.
and prevent cytokine storm. Nevertheless, the Targeted therapeutic approaches to cytokine
disorder can still occur due to iatrogenic causes, storm associated with idiopathic multicentric
pathogens, cancers, autoimmunity, and autoin- Castleman’s disease, HLH, or CAR T-cell therapy
flammatory mechanisms. Distinguishing between have turned deadly conditions into often revers-
protective inflammatory responses and patho- ible states. Given advances in “multi-omic” pro-
logic cytokine storm has important implications filing and therapeutic modulation of the im-
for treatment and is quite challenging. No unify- mune system, as well as concerted efforts to work
ing definition of cytokine storm exists, and there across the cytokine storm umbrella, we expect
is much disagreement about what the definition to see continued improvements in outcomes.
should be and whether specific conditions such
Disclosure forms provided by the authors are available at
as Covid-19 should be included in the spectrum NEJM.org.
of cytokine storm disorders. We propose a unify- We thank our colleagues Michael Jordan, Taku Kambayashi,
ing definition for cytokine storm that is based Ivan Maillard, Sheila Pierson, Ruth-Anne Langan Pai, Dan Rader,
Patricia Tsao, Frits van Rhee, Dermot Kelleher, Shanmuganthan
on the following criteria: elevated circulating Chandrakasan, Amber Cohen, Alexis Phillips, John Wherry, and
cytokine levels, acute systemic inflammatory Charles Dinarello for their critical review and feedback.
References
1. Morgan RA, Yang JC, Kitano M, Dud- 10. Schwartzentruber DJ. Guidelines for Cron RQ. The immunology of macro-
ley ME, Laurencot CM, Rosenberg SA. the safe administration of high-dose inter- phage activation syndrome. Front Immu-
Case report of a serious adverse event fol- leukin-2. J Immunother 2001;24:287-93. nol 2019;10:119.
lowing the administration of T cells trans- 11. Lee DW, Gardner R, Porter DL, et al. 21. Jordan MB, Hildeman D, Kappler J,
duced with a chimeric antigen receptor Current concepts in the diagnosis and Marrack P. An animal model of hemo-
recognizing ERBB2. Mol Ther 2010;18: management of cytokine release syndrome. phagocytic lymphohistiocytosis (HLH):
843-51. Blood 2014;124:188-95. CD8+ T cells and interferon gamma are
2. Ferrara JL, Abhyankar S, Gilliland DG. 12. Diorio C, Shaw PA, Pequignot E, et al. essential for the disorder. Blood 2004;
Cytokine storm of graft-versus-host dis- Diagnostic biomarkers to differentiate 104:735-43.
ease: a critical effector role for interleu- sepsis from cytokine release syndrome in 22. Zhang K, Jordan MB, Marsh RA, et al.
kin-1. Transplant Proc 1993;25:1216-7. critically ill children. Blood Adv 2020;4: Hypomorphic mutations in PRF1,
3. Chatenoud L, Ferran C, Bach JF. The 5174-83. MUNC13-4, and STXBP2 are associated
anti-CD3-induced syndrome: a consequence 13. Teachey DT, Lacey SF, Shaw PA, et al. with adult-onset familial HLH. Blood
of massive in vivo cell activation. Curr Top Identification of predictive biomarkers for 2011;118:5794-8.
Microbiol Immunol 1991;174:121-34. cytokine release syndrome after chimeric 23. Schulert GS, Cron RQ. The genetics of
4. Coley WB. The treatment of malig- antigen receptor T-cell therapy for acute macrophage activation syndrome. Genes
nant tumors by repeated inoculations of lymphoblastic leukemia. Cancer Discov Immun 2020;21:169-81.
erysipelas: with a report of ten original 2016;6:664-79. 24. Korn T, Bettelli E, Oukka M, Kuchroo
cases. Am J Med Sci 1893;105:487-511. 14. Sinha P, Matthay MA, Calfee CS. Is a VK. IL-17 and Th17 cells. Annu Rev Im-
5. Pechous RD, Sivaraman V, Price PA, “cytokine storm” relevant to COVID-19? munol 2009;27:485-517.
Stasulli NM, Goldman WE. Early host cell JAMA Intern Med 2020;180:1152-4. 25. Avau A, Mitera T, Put S, et al. System-
targets of Yersinia pestis during primary 15. Hashimoto D, Chow A, Noizat C, ic juvenile idiopathic arthritis-like syn-
pneumonic plague. PLoS Pathog 2013; et al. Tissue-resident macrophages self- drome in mice following stimulation of
9(10):e1003679. maintain locally throughout adult life the immune system with Freund’s com-
6. Kash JC, Tumpey TM, Proll SC, et al. with minimal contribution from circulat- plete adjuvant: regulation by interferon-γ.
Genomic analysis of increased host im- ing monocytes. Immunity 2013;38:792-804. Arthritis Rheumatol 2014;66:1340-51.
mune and cell death responses induced by 16. Zoller EE, Lykens JE, Terrell CE, et al. 26. Lucas C, Wong P, Klein J, et al. Longi-
1918 influenza virus. Nature 2006;443: Hemophagocytosis causes a consumptive tudinal analyses reveal immunological
578-81. anemia of inflammation. J Exp Med 2011; misfiring in severe COVID-19. Nature
7. Lee DW, Santomasso BD, Locke FL, 208:1203-14. 2020;584:463-9.
et al. ASTCT consensus grading for cyto- 17. Perez N, Virelizier J-L, Arenzana-Seis- 27. Doherty GM, Lange JR, Langstein
kine release syndrome and neurologic dedos F, Fischer A, Griscelli C. Impaired HN, Alexander HR, Buresh CM, Norton
toxicity associated with immune effector natural killer activity in lymphohistiocy- JA. Evidence for IFN-gamma as a media-
cells. Biol Blood Marrow Transplant 2019; tosis syndrome. J Pediatr 1984;104:569-73. tor of the lethality of endotoxin and tu-
25:625-38. 18. Sallusto F. Heterogeneity of human mor necrosis factor-alpha. J Immunol
8. Grupp SA, Kalos M, Barrett D, et al. CD4+ T cells against microbes. Annu Rev 1992;149:1666-70.
Chimeric antigen receptor-modified T cells Immunol 2016;34:317-34. 28. Cohen J. IL-12 deaths: explanation
for acute lymphoid leukemia. N Engl J 19. Mosmann TR, Coffman RL. TH1 and and a puzzle. Science 1995;270:908.
Med 2013;368:1509-18. TH2 cells: different patterns of lympho- 29. Atkins MB. Interleukin-2: clinical ap-
9. Templin C, Ghadri JR, Diekmann J, kine secretion lead to different functional plications. Semin Oncol 2002;29:Suppl 7:
et al. Clinical features and outcomes of properties. Annu Rev Immunol 1989;7:145- 12-7.
takotsubo (stress) cardiomyopathy. N Engl 73. 30. Schulert GS, Zhang M, Fall N, et al.
J Med 2015;373:929-38. 20. Crayne CB, Albeituni S, Nichols KE, Whole-exome sequencing reveals muta-

tions in genes linked to hemophagocytic Severe imbalance of IL-18/IL-18BP in pa- ogy Am Soc Hematol Educ Program 2016;
lymphohistiocytosis and macrophage ac- tients with secondary hemophagocytic 2016:567-72.
tivation syndrome in fatal cases of H1N1 syndrome. Blood 2005;106:3483-9. 59. Dahmer MK, Randolph A, Vitali S,
influenza. J Infect Dis 2016;213:1180-8. 45. Shimizu M, Yokoyama T, Yamada K, Quasney MW. Genetic polymorphisms in
31. Vadhan-Raj S, Nathan CF, Sherwin SA, et al. Distinct cytokine profiles of systemic- sepsis. Pediatr Crit Care Med 2005;6:Sup-
Oettgen HF, Krown SE. Phase I trial of onset juvenile idiopathic arthritis-associ- pl:S61-S73.
recombinant interferon gamma by 1-hour ated macrophage activation syndrome 60. Liu E, Marin D, Banerjee P, et al. Use
i.v. infusion. Cancer Treat Rep 1986;70: with particular emphasis on the role of of CAR-transduced natural killer cells in
609-14. interleukin-18 in its pathogenesis. Rheu- CD19-positive lymphoid tumors. N Engl J
32. Locatelli F, Jordan MB, Allen C, et al. matology (Oxford) 2010;49:1645-53. Med 2020;382:545-53.
Emapalumab in children with primary he- 46. Dinarello CA, Novick D, Kim S, Kaplan- 61. Nebelsiek T, Beiras-Fernandez A,
mophagocytic lymphohistiocytosis. N Engl ski G. Interleukin-18 and IL-18 binding Kilger E, Möhnle P, Weis F. Routine use
J Med 2020;382:1811-22. protein. Front Immunol 2013;4:289. of corticosteroids to prevent inflamma-
33. Eloseily EM, Weiser P, Crayne CB, 47. Novick D, Kim S, Kaplanski G, Din- tion response in cardiac surgery. Recent
et al. Benefit of anakinra in treating pedi- arello CA. Interleukin-18, more than a Pat Cardiovasc Drug Discov 2012;7:170-4.
atric secondary hemophagocytic lympho- Th1 cytokine. Semin Immunol 2013;25: 62. Fisher CJ Jr, Dhainaut JF, Opal SM,
histiocytosis. Arthritis Rheumatol 2020; 439-48. et al. Recombinant human interleukin 1
72:326-34. 48. Behrens EM, Canna SW, Slade K, et al. receptor antagonist in the treatment of
34. Durand M, Troyanov Y, Laflamme P, Repeated TLR9 stimulation results in mac- patients with sepsis syndrome: results
Gregoire G. Macrophage activation syn- rophage activation syndrome-like disease from a randomized, double-blind, placebo-
drome treated with anakinra. J Rheuma- in mice. J Clin Invest 2011;121:2264-77. controlled trial. JAMA 1994;271:1836-43.
tol 2010;37:879-80. 49. Gorelik M, Torok KS, Kietz DA, 63. Shakoory B, Carcillo JA, Chatham
35. Winkler U, Jensen M, Manzke O, Hirsch R. Hypocomplementemia associ- WW, et al. Interleukin-1 receptor block-
Schulz H, Diehl V, Engert A. Cytokine- ated with macrophage activation syn- ade is associated with reduced mortality
release syndrome in patients with B-cell drome in systemic juvenile idiopathic in sepsis patients with features of macro-
chronic lymphocytic leukemia and high arthritis and adult onset Still’s disease: phage activation syndrome: reanalysis of
lymphocyte counts after treatment with 3 cases. J Rheumatol 2011;38:396-7. a prior phase III trial. Crit Care Med 2016;
an anti-CD20 monoclonal antibody (rituxi 50. Porter DL, Hwang WT, Frey NV, et al. 44:275-81.
mab, IDEC-C2B8). Blood 1999;94:2217-24. Chimeric antigen receptor T cells persist 64. Lykens JE, Terrell CE, Zoller EE, Risma
36. Teachey DT, Rheingold SR, Maude SL, and induce sustained remissions in re- K, Jordan MB. Perforin is a critical physi-
et al. Cytokine release syndrome after lapsed refractory chronic lymphocytic leu- ologic regulator of T-cell activation. Blood
blinatumomab treatment related to ab- kemia. Sci Transl Med 2015;7:303ra139. 2011;118:618-26.
normal macrophage activation and ame- 51. Xu XJ, Tang YM. Cytokine release syn- 65. Zhang M, Bracaglia C, Prencipe G,
liorated with cytokine-directed therapy. drome in cancer immunotherapy with et al. A heterozygous RAB27A mutation
Blood 2013;121:5154-7. chimeric antigen receptor engineered T associated with delayed cytolytic granule
37. van der Stegen SJ, Davies DM, Wilkie cells. Cancer Lett 2014;343:172-8. polarization and hemophagocytic lympho-
S, et al. Preclinical in vivo modeling of 52. Singh N, Hofmann TJ, Gershenson Z, histiocytosis. J Immunol 2016;196:2492-
cytokine release syndrome induced by et al. Monocyte lineage-derived IL-6 does 503.
ErbB-retargeted human T cells: identify- not affect chimeric antigen receptor T-cell 66. Terrell CE, Jordan MB. Perforin defi-
ing a window of therapeutic opportunity? function. Cytotherapy 2017;19:867-80. ciency impairs a critical immunoregula-
J Immunol 2013;191:4589-98. 53. Giavridis T, van der Stegen SJC, Ey- tory loop involving murine CD8(+) T cells
38. Kang S, Tanaka T, Narazaki M, Kishi- quem J, Hamieh M, Piersigilli A, Sadelain and dendritic cells. Blood 2013;121:5184-
moto T. Targeting interleukin-6 signaling M. CAR T cell-induced cytokine release 91.
in clinic. Immunity 2019;50:1007-23. syndrome is mediated by macrophages 67. Pachlopnik Schmid J, Ho C-H, Chré-
39. Faulkner L, Cooper A, Fantino C, Alt- and abated by IL-1 blockade. Nat Med tien F, et al. Neutralization of IFNgamma
mann DM, Sriskandan S. The mechanism 2018;24:731-8. defeats haemophagocytosis in LCMV-
of superantigen-mediated toxic shock: not 54. Norelli M, Camisa B, Barbiera G, et al. infected perforin- and Rab27a-deficient
a simple Th1 cytokine storm. J Immunol Monocyte-derived IL-1 and IL-6 are dif- mice. EMBO Mol Med 2009;1:112-24.
2005;175:6870-7. ferentially required for cytokine-release 68. Polizzotto MN, Uldrick TS, Wang V,
40. Garlanda C, Dinarello CA, Mantovani syndrome and neurotoxicity due to CAR T et al. Human and viral interleukin-6 and
A. The interleukin-1 family: back to the cells. Nat Med 2018;24:739-48. other cytokines in Kaposi sarcoma herpes-
future. Immunity 2013;39:1003-18. 55. Liu Y, Fang Y, Chen X, et al. Gasder- virus-associated multicentric Castleman
41. Martinon F, Pétrilli V, Mayor A, Tardi- min E-mediated target cell pyroptosis by disease. Blood 2013;122:4189-98.
vel A, Tschopp J. Gout-associated uric acid CAR T cells triggers cytokine release syn- 69. Dispenzieri A, Fajgenbaum DC. Over-
crystals activate the NALP3 inflamma- drome. Sci Immunol 2020;5(43):eaax7969. view of Castleman disease. Blood 2020;
some. Nature 2006;440:237-41. 56. Topp MS, Gökbuget N, Stein AS, et al. 135:1353-64.
42. Frank D, Vince JE. Pyroptosis versus Safety and activity of blinatumomab for 70. Ramaswami R, Lurain K, Peer CJ, et al.
necroptosis: similarities, differences, and adult patients with relapsed or refractory Tocilizumab in patients with symptom-
crosstalk. Cell Death Differ 2019;26:99- B-precursor acute lymphoblastic leukae- atic Kaposi sarcoma herpesvirus-associ-
114. mia: a multicentre, single-arm, phase 2 ated multicentric Castleman disease.
43. Netea MG, Kullberg BJ, Verschueren I, study. Lancet Oncol 2015;16:57-66. Blood 2020;135:2316-9.
Van Der Meer JW. Interleukin-18 induces 57. Suntharalingam G, Perry MR, Ward S, 71. Chellapandian D, Das R, Zelley K, et al.
production of proinflammatory cytokines et al. Cytokine storm in a phase 1 trial Treatment of Epstein Barr virus-induced
in mice: no intermediate role for the cyto- of the anti-CD28 monoclonal antibody haemophagocytic lymphohistiocytosis with
kines of the tumor necrosis factor family TGN1412. N Engl J Med 2006;355:1018-28. rituximab-containing chemo-immunother-
and interleukin-1beta. Eur J Immunol 2000; 58. Frey NV, Porter DL. Cytokine release apeutic regimens. Br J Haematol 2013;
30:3057-60. syndrome with novel therapeutics for 162:376-82.
44. Mazodier K, Marin V, Novick D, et al. acute lymphoblastic leukemia. Hematol- 72. Dalla Pria A, Pinato D, Roe J, Naresh

Cytokine Storm
K, Nelson M, Bower M. Relapse of HHV8- pathogenic PI3K/AKT/mTOR signaling in COVID-19: an updated analysis. Thromb
positive multicentric Castleman disease IL-6-blockade-refractory idiopathic multi- Res 2020;191:148-50.
following rituximab-based therapy in HIV- centric Castleman disease. J Clin Invest 100. Sterne JAC, Murthy S, Diaz JV, et al.
positive patients. Blood 2017;129:2143-7. 2019;129:4451-63. Association between administration of sys-
73. Grajales-Reyes GE, Colonna M. Inter- 86. Huang K-J, Su I-J, Theron M, et al. An temic corticosteroids and mortality among
feron responses in viral pneumonias. Sci- interferon-gamma-related cytokine storm critically ill patients with COVID-19: a meta-
ence 2020;369:626-7. in SARS patients. J Med Virol 2005;75: analysis. JAMA 2020;324:1330-41.
74. Kaufman KM, Linghu B, Szustakowski 185-94. 101. Keller MJ, Kitsis EA, Arora S, et al.
JD, et al. Whole-exome sequencing reveals 87. Moore JB, June CH. Cytokine release Effect of systemic glucocorticoids on
overlap between macrophage activation syndrome in severe COVID-19. Science mortality or mechanical ventilation in pa-
syndrome in systemic juvenile idiopathic 2020;368:473-4. tients with COVID-19. J Hosp Med 2020;
arthritis and familial hemophagocytic 88. The RECOVERY Collaborative Group. 15:489-93.
lymphohistiocytosis. Arthritis Rheumatol Dexamethasone in hospitalized patients 102. Fajgenbaum DC, Khor JS, Gorzewski
2014;66:3486-95. with Covid-19 — preliminary report. N Engl A, et al. Treatments administered to the
75. Johnson TS, Terrell CE, Millen SH, J Med DOI: 10.1056/NEJMoa2021436. first 9152 reported cases of COVID-19:
Katz JD, Hildeman DA, Jordan MB. Etopo- 89. Huang C, Wang Y, Li X, et al. Clinical a systematic review. Infect Dis Ther 2020;
side selectively ablates activated T cells to features of patients infected with 2019 9:435-49.
control the immunoregulatory disorder novel coronavirus in Wuhan, China. Lan- 103. De Luca G, Cavalli G, Campochiaro
hemophagocytic lymphohistiocytosis. J Im- cet 2020;395:497-506. C, et al. GM-CSF blockade with mavrilimu
munol 2014;192:84-91. 90. Zhu Z, Cai T, Fan L, et al. Clinical value mab in severe COVID-19 pneumonia and
76. Marsh RA, Allen CE, McClain KL, et al. of immune-inflammatory parameters to systemic hyperinflammation: a single-
Salvage therapy of refractory hemophago- assess the severity of coronavirus disease centre, prospective cohort study. Lancet
cytic lymphohistiocytosis with alemtuzu 2019. Int J Infect Dis 2020;95:332-9. Rheumatol 2020;2(8):e465-e473.
mab. Pediatr Blood Cancer 2013;60:101-9. 91. Del Valle DM, Kim-Schulze S, Huang 104. Bronte V, Ugel S, Tinazzi E, et al.
77. Mouy R, Stephan JL, Pillet P, Haddad H-H, et al. An inflammatory cytokine sig- Baricitinib restrains the immune dysreg-
E, Hubert P, Prieur AM. Efficacy of cyclo- nature predicts COVID-19 severity and ulation in patients with severe COVID-19.
sporine A in the treatment of macrophage survival. Nat Med 2020;26:1636-43. J Clin Invest 2020 November 03 (Epub
activation syndrome in juvenile arthritis: 92. Mathew D, Giles JR, Baxter AE, et al. ahead of print).
report of five cases. J Pediatr 1996;129: Deep immune profiling of COVID-19 pa- 105. Rodriguez-Garcia JL, Sanchez-Nievas
750-4. tients reveals distinct immunotypes with G, Arevalo-Serrano J, Garcia-Gomez C,
78. Faitelson Y, Grunebaum E. Hemo- therapeutic implications. Science 2020; Jimenez-Vizuete JM, Martinez-Alfaro E.
phagocytic lymphohistiocytosis and pri- 369(6508):eabc8511. Baricitinib improves respiratory function
mary immune deficiency disorders. Clin 93. Caricchio R, Gallucci M, Dass C, et al. in patients treated with corticosteroids for
Immunol 2014;155:118-25. Preliminary predictive criteria for COVID-19 SARS-CoV-2 pneumonia: an observational
79. Iwaki N, Fajgenbaum DC, Nabel CS, cytokine storm. Ann Rheum Dis 2020 cohort study. Rheumatology (Oxford)
et al. Clinicopathologic analysis of TAFRO September 25 (Epub ahead of print). 2020 October 06 (Epub ahead of print).
syndrome demonstrates a distinct subtype 94. Zhang Q, Bastard P, Liu Z, et al. Inborn 106. Roschewski M, Lionakis MS, Sharman
of HHV-8-negative multicentric Castleman errors of type I IFN immunity in patients JP, et al. Inhibition of Bruton tyrosine
disease. Am J Hematol 2016;91:220-6. with life-threatening COVID-19. Science kinase in patients with severe COVID-19.
80. Nishimoto N, Kanakura Y, Aozasa K, 2020 September 24 (Epub ahead of print). Sci Immunol 2020;5(48):eabd0110.
et al. Humanized anti-interleukin-6 re- 95. Bastard P, Rosen LB, Zhang Q, et al. 107. Zhang W, Zhao Y, Zhang F, et al. The
ceptor antibody treatment of multicentric Auto-antibodies against type I IFNs in pa- use of anti-inflammatory drugs in the
Castleman disease. Blood 2005;106:2627- tients with life-threatening COVID-19. treatment of people with severe corona-
32. Science 2020 September 24 (Epub ahead virus disease 2019 (COVID-19): the per-
81. van Rhee F, Voorhees P, Dispenzieri A, of print). spectives of clinical immunologists from
et al. International, evidence-based con- 96. Lauder SN, Jones E, Smart K, et al. China. Clin Immunol 2020;214:108393.
sensus treatment guidelines for idiopath- Interleukin-6 limits inf luenza-induced 108. Behrens EM, Koretzky GA. Review:
ic multicentric Castleman disease. Blood inf lammation and protects against fatal cytokine storm syndrome: looking to-
2018;132:2115-24. lung pathology. Eur J Immunol 2013;43: ward the precision medicine era. Arthritis
82. Pierson SK, Stonestrom AJ, Shilling 2613-25. Rheumatol 2017;69:1135-43.
D, et al. Plasma proteomics identifies a 97. Hermine O, Mariette X, Tharaux PL, 109. de Jesus AA, Hou Y, Brooks S, et al.
‘chemokine storm’ in idiopathic multi- et al. Effect of tocilizumab vs usual care Distinct interferon signatures and cyto-
centric Castleman disease. Am J Hematol in adults hospitalized with COVID-19 and kine patterns define additional systemic
2018;93:902-12. moderate or severe pneumonia: a ran- autoinflammatory diseases. J Clin Invest
83. Langan Pai R-A, Sada Japp A, Gonza- domized clinical trial. JAMA Intern Med 2020;130:1669-82.
lez M, et al. Type I IFN response associ- 2020 October 20 (Epub ahead of print). 110. Shimizu M, Nakagishi Y, Yachie A.
ated with mTOR activation in the TAFRO 98. Stone JH, Frigault MJ, Serling-Boyd NJ, Distinct subsets of patients with systemic
subtype of idiopathic multicentric Castle- et al. Efficacy of tocilizumab in patients juvenile idiopathic arthritis based on their
man disease. JCI Insight 2020;5(9):e135031. hospitalized with Covid-19. N Engl J Med cytokine profiles. Cytokine 2013;61:345-8.
84. Arenas DJ, Floess K, Kobrin D, et al. 2020 October 21 DOI: 10.1056/NEJM 111. Vercruysse F, Barnetche T, Lazaro E,
Increased mTOR activation in idiopathic oa2028836. et al. Adult-onset Still’s disease biological
multicentric Castleman disease. Blood 99. Klok FA, Kruip MJHA, van der Meer treatment strategy may depend on the
2020;135:1673-84. NJM, et al. Confirmation of the high cu- phenotypic dichotomy. Arthritis Res Ther
85. Fajgenbaum DC, Langan R-A, Sada mulative incidence of thrombotic compli- 2019;21:53.
Japp A, et al. Identifying and targeting cations in critically ill ICU patients with Copyright © 2020 Massachusetts Medical Society.


10 Cytokine Storm

Uploaded by

Document Information

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

10 Cytokine Storm

Uploaded by

Copyright:

Available Formats

The n e w e ng l a n d j o u r na l of m e dic i n e

Dan L. Longo, M.D., Editor

n engl j med 383;23 nejm.org December 3, 2020 2255

2256 n engl j med 383;23 nejm.org December 3, 2020

The New England Journal of Medicine

Lungs Nervous system

Liver Constitutional symptoms

• Acute renal dysfunction or injury

Figure 1. Clinical Presentation of Cytokine Storm.

n engl j med 383;23 nejm.org December 3, 2020 2257

2258 n engl j med 383;23 nejm.org December 3, 2020

The New England Journal of Medicine

Negative regulators Immune hyperactivation

Prolonged activation of signaling pathways

MAPK NF-κB JAK-STAT3 mTOR

Glucocorticoids JAK inhibitors Sirolimus

2 Acute systemic inflammatory effects

Cytokine-induced 3 Secondary organ dysfunction (e.g., hepatic, renal, pulmonary)

n engl j med 383;23 nejm.org December 3, 2020 2259

T-cell effector Th1 Th2 Th9 Th17 Cytotoxic

CCR5 CCR4 CCR6 CXCR4 CXCR3 T-bet

Cell products Interleukin-4 Interleukin-17 Interferon-γ

Recruitment Recruitment Recruitment Recruitment Cytotoxicity

Figure 3. T-Cell Effector Subgroups Involved in Cytokine Storm.

2260 n engl j med 383;23 nejm.org December 3, 2020

The New England Journal of Medicine

n engl j med 383;23 nejm.org December 3, 2020 2261

Table 1. Soluble Mediators in Cytokine Storm.*

Mediator Main Cell Source Type and Function

2262 n engl j med 383;23 nejm.org December 3, 2020

The New England Journal of Medicine

Type of Cytokine Storm Pathologic Cellular or Cytokine

n engl j med 383;23 nejm.org December 3, 2020 2263

2264 n engl j med 383;23 nejm.org December 3, 2020

The New England Journal of Medicine

n engl j med 383;23 nejm.org December 3, 2020 2265

2266 n engl j med 383;23 nejm.org December 3, 2020

The New England Journal of Medicine

n engl j med 383;23 nejm.org December 3, 2020 2267

2268 n engl j med 383;23 nejm.org December 3, 2020

The New England Journal of Medicine

n engl j med 383;23 nejm.org December 3, 2020 2269

2270 n engl j med 383;23 nejm.org December 3, 2020

The New England Journal of Medicine

n engl j med 383;23 nejm.org December 3, 2020 2271

2272 n engl j med 383;23 nejm.org December 3, 2020

The New England Journal of Medicine

n engl j med 383;23 nejm.org December 3, 2020 2273

You might also like