UCSF
UC San Francisco Previously Published Works
Title
Is autoimmunity the Achilles' heel of cancer immunotherapy?
Permalink
https://escholarship.org/uc/item/8m89q9bm
Journal
Nature medicine, 23(5)
ISSN
1078-8956
Authors
June, Carl H
Warshauer, Jeremy T
Bluestone, Jeffrey A
Publication Date
2017-05-01
DOI
10.1038/nm.4321
Peer reviewed
eScholarship.org Powered by the California Digital Library
University of California
5 4 0 VOLUME 23 | NUMBER 5 | MAY 2017 nature medicine
p e r s p e c t i v e
The emergence of immuno-oncology as the first broadly
successful strategy for metastatic cancer will require clinicians
to integrate this new pillar of medicine with chemotherapy,
radiation, and targeted small-molecule compounds. Of equal
importance is gaining an understanding of the limitations
and toxicities of immunotherapy. Immunotherapy was initially
perceived to be a relatively less toxic approach to cancer treatment
than other available therapies—and surely it is, when compared
to those. However, as the use of immunotherapy becomes
more common, especially as first- and second-line treatments,
immunotoxicity and autoimmunity are emerging as the Achilles’
heel of immunotherapy. In this Perspective, we discuss evidence
that the occurrence of immunotoxicity bodes well for the
patient, and describe mechanisms that might be related to the
induction of autoimmunity. We then explore approaches to limit
immunotoxicity, and discuss the future directions of research and
reporting that are needed to diminish it.
Immuno-oncology drug development presently encompasses a broad
range of agents, including antibodies, peptides, proteins, small mol-
ecules, adjuvants, cytokines, oncolytic viruses, bispecific molecules,
and cellular therapies
1
. A survey of recent literature indicates that
adverse events affecting nearly every organ system have been reported
in association with cancer immunotherapy (Fig. 1). However, one
baseline assumption is that, at present, the frequency of autoimmune
complications following cancer immunotherapy is probably underes-
timated, in part because most cancer trials follow patients for only a
brief time after enrollment (typically 6 months), and some symptoms,
such as lethargy, might have an unclear etiology. The true incidence
is probably further underestimated still, because the numerator does
not include patients who died from their cancer. Related to this, the
incidence of certain immune-related adverse events (irAEs) is corre-
lated with an increased probability of prolonged survival
2
. However,
these associations might be related to lead-time bias, because patients
whose tumors progress succumb to their disease, whereas those who
respond to immunotherapies have longer treatment duration and
more time to develop autoimmune toxicities. In addition, immuno-
toxicity, as presently defined
3
, can have a delayed onset. For instance,
some cases of graft-versus-host disease after hematopoietic stem cell
transplantation (HSCT) can take more than a year to manifest
4
. Onset
of thyroiditis has been reported as late as 3 years after the initiation of
therapy with the cytotoxic T-lymphocyte-associated protein 4 (CTLA-4)
antagonist ipilimumab
5
. Adding more complexity, the natural his-
tory of certain autoimmune diseases, such as type 1 diabetes (T1D), is
unpredictable; the onset of clinical disease manifestations can vary from
weeks to decades after the appearance of islet autoantibodies
6
.
The existence of paraneoplastic syndromes in oncology has
long been known to clinicians, antedating immune-checkpoint
blockade, and it provides the clearest example of naturally occur-
ring tumor immunity and autoimmunity in humans. Patients with
occult or advanced tumors may develop a wide variety of syndromes,
varying from myasthenia gravis to cerebellar degeneration, owing
to acquired cellular and/or humoral immunity to antigens expressed
by the tumor
7
. In small-cell lung cancer, the occurrence of paraneo-
plastic syndromes can portend prolonged survival
8
. These observa-
tions raise the question of whether deliberate attempts to provoke
paraneoplastic syndrome should be attempted, and, on a related note,
whether the occurrence of these syndromes will increase with increas-
ing numbers of cancer patients exposed to immune-checkpoint-
blockade administration.
Previous studies have described the evolving spectrum of immu-
notherapy-associated irAEs
3,9,10
. Here we provide further perspective
on the emerging clinical syndromes of immunotoxicity and autoim-
munity in cancer therapy.
Barriers to immune recognition of tumors
Immune tolerance is crucial for preventing autoimmunity, and now,
we suspect that it might also be highly relevant to cancer immunity.
Immune tolerance is defined as a lack of lymphocyte reactivity to self-
antigens or foreign-tissue antigens in an organ graft, achieved with-
out the need for long-term immunosuppression and while retaining
immune competence and reactivity to all other foreign antigens
11
.
Immune tolerance starts in the thymus, where the diverse T cell recep-
tor (TCR) repertoire is created through random somatic recombination
events. Autoreactive T cell progenitors expressing TCRs that bind with
high affinity to self-peptide–major histocompatibility complex (MHC)
complexes are deleted through a process called negative selection. The
transcription factor autoimmune regulator (AIRE), which is selectively
expressed by a subset of CD80-expressing medullary thymic epithelial
cells (TECs), drives the expression of tissue-specific self-antigens to
ensure selective removal of thymocytes bearing TCRs that recognize
these antigens
12
. Thymocytes that have low affinity for self-peptide-
MHC complexes are positively selected to progress to the periphery. In
addition, some thymocytes expressing TCRs that bind with high affin-
ity to self-antigen peptide–MHC complexes differentiate into Forkhead
box protein 3 (FOXP3)-expressing regulatory T (T
reg
) cells
13
.
1
Parker Institute for Cancer Immunotherapy, San Francisco, California, USA.
2
Center for Cellular Immunotherapies, University of Pennsylvania, Philadelphia,
Pennsylvania, USA.
3
Endocrine Division, Department of Medicine, University
of California, San Francisco, San Francisco, California, USA.
4
Diabetes Center,
University of California, San Francisco, San Francisco, California, USA.
Correspondence should be addressed to C.H.J. (cjune@upenn.edu).
Received 24 January; accepted 13 March; published online 5 May 2017;
corrected after print 5 May 2017; doi:10.1038/nm.4321
Is autoimmunity the Achilles’ heel of cancer immunotherapy?
Carl H June
1,2
, Jeremy T Warshauer
3
& Jeffrey A Bluestone
1,4
© 2017 Nature America, Inc., part of Springer Nature. All rights reserved.
p e r s p e c t i v e
nature medicine VOLUME 23 | NUMBER 5 | MAY 2017 5 4 1
As T cells exit the thymus, additional peripheral tolerance mecha-
nisms prevent autoreactive T cells that escaped negative selection
from reacting to or attacking self-antigen-expressing healthy cells.
These include T cell–intrinsic mechanisms, such as immunological
ignorance, anergy, exhaustion, phenotypic skewing, and apoptosis, as
well as extrinsic-cell-based mechanisms
14
. Among the latter, T
reg
cells
stand out as unique because they are comprised both of ‘central’ T
reg
cells—formed from high-affinity interactions with self-peptide–MHC
complexes in the thymus, as mentioned above—and ‘peripheral’ T
reg
cells, formed from T cells that engage in prolonged interactions with
low-affinity self-antigens and non-self-antigens, such as allergens,
food and commensal microbiota. Together, these pathways help to
maintain peripheral tolerance to self-antigens and certain foreign
antigens
13,15
. Therefore, although tumors express tumor-specific
neoantigens and overexpress self-antigens that can potentially initiate
a potent anti-tumor immune response
16
, the immune system has
developed a complex, redundant, and robust combination of cells
and molecules all designed to keep the immune system in check and
avoid unwanted inflammation and tissue damage
17
.
Several specific cases exemplify the degree to which tumors can take
advantage of immune-tolerance mechanisms to disrupt anti-tumor
immunity. Some tumor-associated antigens, such as tyrosinase-related
protein 1 (TYRP1, also known as TRP1), are expressed in medullary
thymic epithelial cells
18
. In addition, tumors can escape anti-tumor
T cell responses by decreasing their expression of tumor-associated
antigens and/or MHC molecules; by secreting immunosuppres-
sive soluble factors (vascular endothelial growth factor, stromal-
cell-derived factor, interleukin 10, interleukin 6, transforming
growth factor-β, adenosine, and prostaglandins); and/or by engaging
immune checkpoints (CTLA-4, PD-1, Tim-3, and LAG-3) that sup-
press anti-tumor activity. Immunosuppressive cells, including mye-
loid-derived suppressor cells (MDSCs), T
reg
cells, tumor-associated
macrophages, regulatory B cells, and regulatory dendritic cells
19,20
,
present within the tumor microenvironment can also suppress
anti-tumor responses
21
.
Therapeutic induction of anti-tumor responses
Clinical development and approval of immunomodulators, also
known as immune-checkpoint inhibitors, have transformed the
treatment of certain tumors, such as melanoma, non-small-cell lung
cancer, and bladder cancer. Checkpoint inhibitors act by blocking
interactions that normally suppress T cell responses. The binding
of CTLA-4 on naive T cells in the lymph nodes to B7 on antigen-
presenting cells (APCs) produces an inhibitory signal during the
primary phase of T cell activation (Fig. 2). CTLA-4 can also strip B7
molecules—which are ligands for CD28 costimulatory molecules on
T cells—from the APCs through a process of transendocytosis, which
further impairs T cell activation
22
. Thus, the blockade of CTLA-4
leads to a more robust costimulatory signal, and this boosted signal
may enable otherwise naive T cells with weak affinity to respond to
overexpressed and mutant tumor antigens. CTLA-4 is also expressed
on CD4
+
CD25
hi
FOXP3
+
T
reg
cells, and its engagement leads to
enhanced IL-35, IL-10, TGF-β, and indoleamine 2,3-dioxygenase
(IDO) production, T
reg
cell proliferation, and decreased effector
T cell activation and proliferation
23
.
By contrast, PD-1 binding to PD-L1 and PD-L2 regulates those
already activated T cells later in the immune response in the periph-
eral tissues. PD-1 engagement inhibits T cell proliferation, IFN-γ,
TNF-α, and IL-2 production
24
, although some data suggest that it
may also be involved in early T cell activation
25
, central tolerance,
and regulation of negative selection
26
. In addition to the direct
effect of checkpoints on effector T cells, blockade of the molecular
interactions can also affect the tumor microenvironment (TME).
For example, altering IDO expression and CTLA-4 engagement on
APCs, and decreasing T
reg
cells in the TME through mechanisms
such as STING/IFN-αβ signaling and myeloid-derived suppressor
cells improves cancer immunity
27
.
Autoimmune consequences
Considering their diverse mechanisms of action, it is perhaps not
surprising that these immunomodulators induce multiple immune-
mediated adverse events that lead to antigen-specific autoimmune
manifestations. In humans, autoimmune manifestations caused by
drugs targeting the CTLA-4 and PD-1 pathways seem to be dependent
on the pathway(s) targeted
5,10,28,29
. For example, the most commonly
reported endocrine irAE following therapy with the CTLA-4-blocking
antibody ipilimumab is hypophysitis, an event that is rarely observed
after PD-1-antibody therapy. Ectopic expression of CTLA-4 in the
pituitary gland may be responsible for this effect
30
, and antibody-
dependent, cell-mediated cytotoxicity (ADCC) with activation of
complement might be involved in the destruction of the hypophy-
sis. By contrast, the most commonly reported endocrine-related
toxicity after PD-1-antibody therapy is hypothyroidism, a syndrome
that is rarely observed in patients treated with ipilimumab. These
autoimmune syndromes could be a consequence of revealing pre-
existing conditions in these patients. However, with one excep-
tion
31
, analysis of pre-existing autoantibodies and single-nucleotide
polymorphisms (SNPs) associated with autoimmune disease has not
proved useful in the identification of patients at risk for irAE.
Given that the function of CTLA-4 is to primarily affect CD4
+
T cells at an early stage in lymphatic tissue, its blockade might be
expected to have broader and more nonspecific consequences than
Nervous system
Guillain–Barré syndrome
Myasthenia gravis
Encephalitis
Pituitary
Hypophysitis
Lungs
Pneumonitis
Adrenal
Insufficiency
Pancreas
T1D
Rheumatologic
Vasculitis
Arthritis
Kim Caesar/Springer Nature
Thyroid
Hypothyroid
Hyperthyroid
Heart
Myocarditis
Gastrointestinal
Colitis
Autoimmune hepatitis
Skin
Vitiligo
Psoriasis
Stevens–Johnson syndrome
DRESS syndrome
Figure 1 Examples of autoimmune and other immune-related adverse
effects associated with cancer immunotherapy. See Supplementary Note
for references describing each of these autoimmune and immune-related
adverse events. T1D, type 1 diabetes.
© 2017 Nature America, Inc., part of Springer Nature. All rights reserved.
p e r s p e c t i v e
5 4 2 VOLUME 23 | NUMBER 5 | MAY 2017 nature medicine
the blockade of PD-1, which interacts with its ligand primarily within
the peripheral tissue and tumor microenvironment
32
. Moreover,
CTLA-4 inhibition lowers the threshold required for T cell activa-
tion, which results in increased expansion and diversification of
circulating, low-avidity T cells
33
; it also causes an Fc-γR-mediated
depletion of T
reg
cells
34
. In agreement with this, treatment with ipili-
mumab was found to broaden the TCR repertoire more robustly,
within 2 weeks, in those experiencing irAEs than in those without
irAEs, and treatment response improved along with the increase in
TCR diversity. This further underscores cancer immunotherapy as a
double-edged sword in which patients and clinicians must weigh the
risk of immunotoxicity against the benefit of tumor destruction
35
.
By comparison, PD-1 blockade is likely to reinvigorate a previously
overactive immune system
32
; studies of T cell exhaustion in chronic
infection show that PD-1 functions to limit effector T cell–mediated
inflammatory injury
36
. Studies of individuals with cancer have not
yet compared T cell exhaustion profiles from patients treated with
PD-1 with those from patients receiving CTLA-4 blockade as treat-
ment. Some effects, however, might be common to CTLA-4 and PD-1
blockade. For example, local T
reg
cells in the TME of human tumors
also show upregulated expression of PD-L1 and PD-L2 (ref. 37), and
therefore, might be affected by PD-1 blockade; if this is the case, then
alterations in T
reg
cell activity might contribute to both PD-1- and
CTLA-4-induced autoimmunity.
An anti-tumor immune response can kill tumor cells, and host
APCs can then pick up the antigens, which, in a form of antigen
presentation referred to as cross-presentation, leads to the priming
of secondary immune responses. There is increasing evidence that
cross-presentation of neoantigens or shared antigens might induce
a loss of tolerance and subsequent autoimmunity in patients treated
with checkpoint blockade. In two patients with fulminant myocarditis
resulting from a combination treatment of ipilimumab and nivolumab,
an analysis of T cells infiltrating the skeletal muscle, myocardium, and
tumor revealed an increase in the most abundant TCR type in one
of the patients; tumors in both patients expressed abundant muscle-
specific antigens, including desmin and troponin
38
. This led to
the suggestion that an epitope shared by tumor and healthy tissue
contributed to the myocarditis, but it is also possible that this find-
ing was coincidental, because no predominant TCR clonotype was
detected in the other patient
38
.
Indeed, autoimmune manifestations may be indirect and due to
epitope spreading (ES) caused by immunotherapy-induced inflam-
mation and tumor lysis. ES refers to the recruitment of additional
T cells and the development of an immune response to epitopes dis-
tinct from and non-cross-reactive with the primary epitope recognized
by the original effector T cells
39
. Recognition of multiple epitopes
might enhance anti-tumor responses, for example, by promoting addi-
tional help from CD4-mediated T cells; through linked recognition;
or through direct tumor destruction, if it is a CD8 T cell–activating
epitope. Linked recognition is a process of CTL priming whereby CD4
helper cells recognize antigens on the same APC that cross-presents
the CTL epitope
40
. ES might result through unknown mechanisms
during the initiation of T cell responses that target self-antigens (Fig. 3)
shared by normal tissue, which thereby leads to loss of tolerance and
autoimmunity. ES has been documented in patients receiving tumor
vaccines
41
and in patients who have undergone adoptive transfer
(Box 1) of CTLs and CAR T cells
42,43
. Checkpoint therapy with ipili-
mumab can induce ES
44
, and it is likely that neoantigen-directed
immune responses against tumors, when followed by ES, can trigger
autoimmunity against nonmalignant tissues in which the neoantigens
are absent. Although many efforts are under way to exploit ES using
several immune and nonimmune (for example, radiation and chemo-
therapy) treatment modalities, in certain circumstances, the conse
-
quences of ES might need to be curtailed through the development
and administration of selective inhibitors of cross-presentation
45
,
perhaps by manipulating proteasome activity or regulatory T cell
function (Fig. 3).
Another mechanism that might influence autoimmune side effects
is T cell functional flexibility and plasticity. Changes in epigenetic-
control mechanisms may allow for switching between exhausted
and activated T cell states. In addition, T
reg
cells can be converted
into T
H
17 cells in the presence of IL-6 and TGF-β
46,47
, which might
underlie certain forms of autoimmunity, such as autoimmune hepa-
titis and psoriasis
48,49
, and CXCL11 promotes T
reg
cell differentia-
tion into CXCR3
+
CD4
+
effector T cells, suggesting that it might be
involved in the development of autoimmune encephalitis
50
. Control
of T
reg
cell plasticity is further regulated by EZH2, a target for mul-
tiple small-molecule inhibitors in cancer trials
51
. EZH2 is a histone
modifier that functions as the catalytic component of the polycomb
repressive complex 2 (PRC2). As a lysine methyltransferase, EZH2
promotes the addition of the repressive mark histone H3K27me3 to
target chromatin, which thereby induces chromatin compaction and
transcriptional repression by restricting access to transcriptional reg-
ulators such as RNA polymerase II and other transcription-associated
factors. Hyperactivation of or mutations in EZH2 are found in a
variety of malignancies
52
. However, EZH2 activity in tumor cells can
shape the immune microenvironment of tumors by controlling the
expression of chemokines
53
. Thus, in cancer, small molecules targeting
proteins that are mutated in tumor cells could have dual effects: they
Lymph node
Naive T cell
CD4
+
FoxP3
+
T
reg
cell
MHC
TCR
CD28
CTLA-4
PD-1
PD-L1
PD-1
PD-1
CTLA-4
B7
APC
Tremelimumab
Ipilimumab
Effector
T cell
Exhausted
T cell
CAR T cell
IFN-γ
Pembrolizumab (anti-PD-1)
Nivolumab
Atezolizumab (anti-PD-L1)
Durvalumab
Avelumab
ScFv
CD28
CD3ζ
Tumor
cell
Cell surface
tumor neoantigen
Kim Caesar/Springer Nature
Figure 2 CTLA-4 and PD-1 checkpoint blockade affects T cells at
different stages of differentiation and at different anatomical locations.
In lymphoid tissues, CTLA-4 expression is induced in naive T cells.
After the TCR is triggered by an antigen–MHC encounter, CTLA-4 is
expressed on the cell surface. CTLA-4-blocking therapies suppress
negative signals delivered by CTLA-4, which permits sustained T cell
activation and proliferation. The major role of the PD-1 pathway is at a
later stage of T cell activation. In peripheral tissues (including tumors),
activated T cells upregulate PD-1 expression. Inflammatory signals in
the tissues induce the expression of PD-1 ligands, which downregulate
the activity of T cells through binding to PD-1 and CD80, a feedback
mechanism to limit collateral tissue damage. T cell exhaustion, a state of
terminal T differentiation, is induced by prolonged exposure to high levels
of antigen. Anti-PD-1 or PD-L1 therapies prevent this negative regulation
and may reinvigorate exhausted T cells or delay T cell exhaustion in
response to chronic antigen exposure. Alternative new therapies with
CAR T cells that are highly specific for tumor antigens cause destruction
directly within the tumor microenvironment.
© 2017 Nature America, Inc., part of Springer Nature. All rights reserved.
p e r s p e c t i v e
nature medicine VOLUME 23 | NUMBER 5 | MAY 2017 5 4 3
might directly suppress proliferation or survival of cancer cells, and at
the same time, they might modulate the anti-tumor immune response
at the level of T
reg
cells. Understanding the complex network of tran-
scription factors involved in the regulation of T cell plasticity will be
crucial to the development of targeted therapies that can effectively
treat cancer without also introducing autoimmune risk.
Efforts to predict and understand toxicity
Understanding and manipulating the mechanisms and factors that
determine a patient’s risk of developing immune toxicity during or
after checkpoint blockade will require basic and preclinical research,
as well as changes in current clinical-reporting practice. With regard
to the latter, although existing observations of immune-related toxicity
are helpful, most literature reporting autoimmune-associated disease
consists of case reports
54,55
, in which, for example, autoantibodies are
often not described. Moreover, frequently, autoantibodies are negative
when reported and autoimmune disease presents rapidly, as seen with
diabetic ketoacidosis and T1D; because subsequent longitudinal data
are often not reported, scientists have generally been unable to assess
whether autoantibodies, such as GAD-65, ICA-512, and ZnT8, become
detectable over time. At present, it remains unclear whether the autoim-
mune manifestations seen after immunotherapy are clearly connected
to the ‘classical’ autoimmune diseases that they symptomatically repre-
sent, because simple correlations with factors such as MHC haplotypes,
autoantibodies, and antigen-specific T cell identification are mostly
absent from current analyses. Moreover, organ involvement in an
autoimmune response is unpredictable from patient to patient, perhaps
owing to different genetics, epigenetics, or microbiota environment,
and/or because of polymorphisms in the checkpoints themselves
56
.
For example, in terms of genetic associations, CTLA4 polymor-
phisms have been linked to an increased risk of autoimmune diseases
such as T1D, and preclinical models have shown that anti-CTLA-4
can increase the risk of autoimmune diabetes
57
. In summary, large
patient databases containing more metadata will be required to deter-
mine whether similar genetic predispositions and biologic pathways
are shared between classic autoimmune diseases and the autoimmune
syndromes induced by various cancer immunotherapies.
Patients with a history of or ongoing autoimmune disease are
currently excluded from clinical trials, but whether this is justified
remains unclear, given that the underlying mechanisms of immu-
notoxicity from checkpoint inhibition may be distinct and have no
substantial impact on a patient’s existing autoimmune disease. Crucial
to clinicians’ understanding of and ability to predict immunotoxicity
from these drugs will be enhanced pharmacovigilance. The estab-
lishment of a national database to track long-term outcome would
enable a deeper understanding of irAEs. One effort in this area is an
initiative by the US Food and Drug Administration (FDA) Center
for Biologics Evaluation and Research (CBER) that used the High-
Performance Integrated Virtual Environment (HIVE) to establish a
database on the safety of engineered T cells. Other efforts, such as
the Cancer Moonshot launched as part of the 21
st
Century Cures
Act, also highlight the need for a more complete understanding of
the consequences of these new therapies as well as the importance
of collating large data sets and maintaining robust clinical and basic
research efforts. Together, these academic, industry, philanthropic,
and government efforts will help us to shape and take advantage of
these exciting new therapeutic opportunities.
Some insights into the mechanisms by which cancer immuno-
therapy promulgates autoimmunity might be gained by analysis of
T cell responses in other settings. For instance, CD8
+
T cell exhaustion
is frequently observed in chronic viral infections
58
. Moreover, PD-1
blockade seems to reactivate effector T cells through the targeting of
certain transcriptional factors (i.e., nuclear factor kappa-light-chain-
enhancer of activated B cells, interferon regulatory factors 1 and 2,
orphan nuclear receptor NR4A1, and B-lymphocyte-induced matura-
tion protein 1) that cause a reengagement of the effector mechanisms
in the epigenome of exhausted T cells
59
. CTLA-4 has also been found
to have a clear role in multiple chronic infections, including HBV, HCV,
and HIV, and its inhibition can increase the function of pathogen-spe-
cific T cells
60,61
. Epigenetic and transcriptional mechanisms determine
the functional plasticity of T cells to switch between their exhausted
and effector states
62
, and so altering the transcriptional landscape—for
example, by manipulating transcription-factor and gene-enhancer
expression—and epigenetic landscape—for example, by manipulat-
ing the activity of histone-modifying methylation or demethylation
enzymes, histone acetylases, and DNA demethylases—of adoptively
transferred T cells may yield a more specific anti-tumor response than
generalized PD-1 and CTLA-4 blockade. Genomic editing of chimeric
antigen receptor (CAR) T cells to render them resistant to exhaustion
has been proposed
63,64
, and Sen and colleagues suggested that map-
ping state-specific enhancers in exhausted T cells could enable more
precise genome editing for adoptive T cell therapy
62
.
Improved clinical reporting and new basic and preclinical research
is especially important because more and more patients are receiving
checkpoint-blockade treatment. Indeed, the emerging standard of care
for patients with many forms of disseminated cancer is therapy with
a checkpoint antagonist. Moreover, for patients in clinical trials, the
trend is to combine one or several of the increasing ‘toolbox of thera-
pies,’ including vaccines, CAR T cells, oncolytic viruses, radiation,
TCR
MHCI
TCR
TCR
CD28
MHCI
CD80/86
TCR
MHCII
MHCII
Tumor
Tumor cell death
B cell DC/MΦ
BCR
Tumor mutant
antigens
Epitope
spreading
(self-antigens)
Continued tumor
destruction
AutoAbs and
Auto Rx T cells
CAR-T cell
TILs
TCR-T cell
CD8 T cell CD4 T cell CD8 T cell CD4 T cell
Endolysosome
Kim Caesar/Springer Nature
Figure 3 Potential mechanism of epitope spreading leading to
autoimmunity. Top, T cells recognizing antigens (for example, neoantigens
or antigens overexpressed on tumor cells) on tumor cells induce
cytokine secretion and/or cytotoxic killing of tumor cells. This can
also result in the death of nontransformed bystander cells. The antigens
released by all cells in the environment are ingested by APCs
(for example, dendritic cells, macrophages, and B cells), which migrate
to the lymph nodes. Bottom, in the lymph nodes, these activated
APCs can present tumor antigens as well as antigens from bystander cells,
thereby priming a second wave of T cells that can re-enter the tissue and
cause additional tumor destruction and off-target destruction of normal
tissue. This, in turn, leads to autoimmunity. AutoAbs, autoantibodies.
Auto RX, autoreactive.
© 2017 Nature America, Inc., part of Springer Nature. All rights reserved.