1. Background
1.1 The unmet need for improved outcomes in HNSCC
In the primary setting, current treatment options for patients who have had a
head and neck squamous cell carcinoma (HNSCC) – though highly differentiated and
individualized in each specific case – remain limited to three main modalities:
surgery or radiotherapy with or without chemotherapy. The treatment decision is
guided by tumor stage and resectability, local surgical expertise, performance
status, and patient preference. A high amount of treatment variability and
individualization is introduced by tailoring the extent of surgery, the need for
local reconstruction, and the dosage and scope of radiation. As the third
treatment modality, classical cytoreductive chemotherapy is added in primary
radiotherapy and adjuvant radiotherapy in the high-risk setting [1]
[2]. The last decades have seen improvements and innovations across the
treatment spectrum: transoral robotic surgery (TORS) can minimize surgical
morbidity [3], intensity-modulated
radiotherapy (IMRT) reduces off-target radiation of healthy tissue [4], and innovative monoclonal antibodies
such as cetuximab may be used as an alternative in Cisplatin-ineligible cases
[5]. However, the overall improvement
in patient outcomes has been moderate and gradual [6] and was probably driven mostly by an
increase in HPV-associated oropharyngeal cancers [7].
Of course, for patients and their families, the oncologic outcome measured as
recurrence-free and overall survival times continues to be the most compelling
concern. Here, the determining prognostic factor is still the tumor stage and
location at diagnosis. Certainly, with the emergence of HPV-associated HNSCC as
a separate biological and clinical entity, there is a subgroup of patients with
good overall outcomes. However, this is driven by the underlying disease
pathophysiology and as yet unimproved treatment algorithms [7]. For non-HPV-associated cancers, except
for glottic and oral cavity disease which present early and in resectable
stages, the survival outcomes have been unsatisfactory, especially given the
highly invasive and sometimes extensive surgical procedures and intensive
high-dose chemoradiotherapy regimen patients must undergo to achieve said
outcomes. Indeed, a detailed analysis of SEER data comparing the changes in
relative survival across four decades (1976–2015) shows a marked increase in
survival only for oropharyngeal and oral tongue tumors, while the 40-year
survival changed little for other subsites such as larynx, hypopharynx,
nasopharynx, and non-tongue oral cavity when adjusting for other factors in
multivariate analysis [8]. The study also
outlines the unsatisfactory 5-year relative survival for the most recent
2006–2015 cohort of 38.4% for non-tongue oral cavity, 31.2% for hypopharyngeal,
and 35.8 for laryngeal cancer in regional disease metastasized to the local
lymph node basin. This is especially relevant as a large percentage of patients
present in said locally advanced disease stage [9].
In those patients who do achieve long-term survival, it is associated with
significant morbidity and reduced quality of life. Depending on the location of
the primary tumors, one or more of the basic physiological and daily social
functions such as breathing, speaking, swallowing, tasting, and olfaction might
be disturbed [10]. The extent of
functional deficit varies depending on the extent of the primary disease and the
chosen treatment modality but is nonetheless comparable across therapies and
patients. Voice and speech problems have been noted in two-thirds of HNSCC
patients — even 10 years following primary radiotherapy [11]. Dysphagia and reduced oral feeding
have been the most consistent concerns, with a high focus on reducing this
morbidity through intensive speech therapy and rehabilitation [12]. Xerostomia and loss of taste have been
mostly associated with primary radiotherapy and remain long-term issues with
limited treatment options [13]. Other
treatment sequelae that should be addressed by a multidisciplinary approach
during head and neck cancer survivorship care are fatigue, sexual dysfunction,
chronic pain, caries and dental issues, lymphedema, and cervical dystonia [14].
Surgery, whether in the primary or salvage setting, is associated with procedures
that may result in long-term body image issues [15]. This is especially true for amputating surgeries such as
laryngectomies, exenteratio orbitae, or ablatio nasi, as well as the need for
pedicled or free flap tissue transfer. All this accumulates, creating a
long-term reduction in quality of life in HNSCC survivors [16]
[17]
[18]. Alarmingly, reports
show a two times higher risk to die from suicide in this population compared to
other cancer types [19].
Taken together, this shows that even though considerable strides have been made
to improve both mortality and morbidity in HNSCC patients, the current
state-of-the-art therapy options do not provide satisfactory outcomes in terms
of both long-term survival and treatment sequelae and, therefore, the resulting
quality of life.
1.2 The introduction of immunotherapy as a fourth pillar of HNSCC
treatment
In the last decade, the introduction of treatment options that harness the power
of the immune system to detect and eliminate cancer cells has offered hopes for
improved outcomes to both patients and providers [20]. For decades, there has been
circumstantial clinical evidence about the role of the immune system in the
tumor-host interaction. Examples include the spontaneous regression of cancers,
at some point coinciding with febrile infections [21]; the recurrence of metastasis in
transplanted organs, such as cases of melanoma metastases transferred from donor
to kidney transplant patients [22]; and
the increased incidence of cancers in individuals with genetic or acquired
immunosuppression [23]
[24]. In addition to these clinical
observations, there were basic science findings that aligned with a hypothesized
protective effect of the immune system against cancer. These include but are not
limited to, a correlation between tumor infiltrating lymphocytes (TIL) and
prognosis [25]. The idea of a protective
effect of the immune system was underscored by more mechanical studies in
immunosuppressed mouse models that resulted in the immunosurveillance hypothesis
[26], which is discussed in more
detail below. Following a long era of skepticism, these findings were at last
translated into the clinic. First, cytokines that broadly stimulate the immune
system, such as IL-2 for the treatment of metastatic renal cell carcinoma and
melanoma [27] or alpha Interferon in
hairy-cell leukemia [28], were
FDA-approved. Other early innovative and pioneering interventions were the
ex-vivo expansion and reinfusion of TIL [29]. Further studies into the role of T cells and their interaction
with cancer cells (outlined in more detail below), especially the importance of
checkpoint receptors in said interaction, paved the way for the current era of
checkpoint inhibitors, monoclonal antibodies that block the inhibition of the
antitumor immune response by cancer cells. Here the approval of ipilimumab, an
anti-CTLA-4 antibody, for metastatic melanoma in 2011 marked the starting point
for an explosion of indications [30]. In
HNSCC, the CheckMate-141 trial demonstrated improved overall survival of
nivolumab, an anti-PD1 antibody, compared to investigator’s choice, in patients
with recurrent or metastatic disease that is refractory to platinum-based
treatments, leading it to be the current standard of care in this setting [31]. Further, the KEYNOTE-048 trial showed
an overall survival benefit for pembrolizumab, an anti-PDL1 antibody, compared
to the EXTREME (chemotherapy plus cetuximab) regimen in the first-line treatment
of recurrent-metastatic setting [32]. It
is now approved as monotherapy for patients expressing PD-L1 >1% and in
combination with chemotherapy in expressing PD-L1 negative tumors. Further
innovations in this field, which are beyond the scope of this review, are
discussed in detail elsewhere [33].
Suffice it to say, the improvements in outcome, i. e., better survival and lower
morbidity, that can be observed in the recurrent-metastatic setting underscore
the viability of this treatment modality and inspire incorporation in earlier,
curative therapy stages, such as adjuvant or neo-adjuvant settings.
1.3 Principles of immune-oncology and basics of current HNSCC
immunotherapy
1.3.1 Immune elimination, equilibrium, and escape
The basic function of the adaptive immune response is the development of a
specific reaction against a structure, the antigen, identified as foreign by
the immune system. This response differs according to cell type and context
and can consist of the production of specific antibodies by B cells or the
cytokine-driven organization of the immune response (CD4+helper cells).
Direct killing of structures recognized as foreign is the task of cytotoxic
T cells [34]. The antigens can be
detected in different contexts, namely infected somatic cells in acute or
chronic infection, autologous tissue in autoimmunity, a donor organ in a
rejection reaction, or mutated cells in anti-tumor immunity.
The idea that the immune system can recognize and eliminate cancer cells –
termed the cancer immunosurveillance hypothesis – was proposed in the early
1960s, yet due to a lack of scientific evidence and insufficient
experimental models, it was not pursued further [26]. As the research community focused
on the cancer cell and its genetic perturbations, such as oncogenes and
tumor-suppressor genes, there was a long period without progress in
immuno-oncology [35]. Interest in the
tumor-host interaction reemerged in the 2000s, building on new and
encouraging evidence from the previous decade [36]. Here, the conceptual framework of
immunosurveillance was expanded into the 3-Es of the immunoediting model,
recognizing that the interaction between cancer and host did not stop with
the destruction of clinically inapparent cancer lesions, the
immunosurveillance or “elimination” state, but its “immunogenic phenotype is
continuously shaped by the immunological forces in its environment” [36]. If the tumor is not eliminated, it
enters an “equilibrium” state in which it is not clinically apparent, but
cannot be cleared by the immune system. Following this, a third “escape”
phase emerges in which the tumor outgrows the stalemate with the immune
system and becomes clinically apparent.
1.3.2 T cell antigen recognition and activation
Due to the potential power of activated T cells, the initiation of cytotoxic
effector function is highly regulated and involves multiple steps or signals
[37]
[38]. For an immune response to arise
from recognition of a tumor antigen expressed on a cancer cell, the same
antigen must be shown to the T cell a second time by an antigen-presenting
cell, the first signal, along with co-stimulatory or co-inhibitory receptor,
the second signal, as well as modulatory cytokine secretion, the third
signal. If this process occurs for the first time in the context of an acute
immune response, such as an infection or vaccination, a naive T cell, i. e.,
one that has never been in contact with its antigen before, develops into an
effector T cell, which then divides and exerts its cytotoxic function. In
parallel, memory T cells develop to rapidly provide an immune response in
the event of a future reappearance of the antigen in the organism [40]. Depending on the location of these
cells, they are subdivided into central memory T cells in lymphoid organs or
effector memory cells in the tissue [41].
The cytotoxic effector T cell and its activation are at the center of the
so-called tumor-immunity cycle ([Fig.
1]) [39] in which
tumor-favoring and tumor-preventing influences of the immune system can
occur at each step: release of the tumor antigen, presentation of the tumor
antigen, activation of the T cell, invasion of the tumor by the T cell, T
cell-mediated recognition of the tumor, and killing of tumor cells. The
cycle serves as a good model for understanding the main components and
interactions, as well as potential therapeutic interventions. Analogous to
the interaction between the host and infections, the tumor-immunity cycle
centers on the recognition of antigens by the adaptive immune system.
Immunogenic cell death releases antigens into the extracellular milieu.
Dendritic cells (DCs) then capture these antigens and migrate to the
tumor-draining lymph nodes, where they are presented to naive T cells via
MHC molecules. The engagement of T cell receptors by MHC-antigen complexes
in the presence of co-stimulatory molecules like CD28 – the immune synapse –
leads to T-cell activation. The T cells then circle back to the tumor where
they recognize their respective tumor antigen and initiate cancer cell
killing via their cytotoxic properties. The cytotoxic effector function here
consists of the release of perforin and granzyme B granules, activation of
the Fas receptor by the Fas ligand, and activation of other immune cells by
proinflammatory cytokines, such as interferon-gamma and TNF-alpha [40]. Perforin forms pores in the target
cell membrane, allowing granzymes to enter and activate caspase-dependent
apoptosis within the cancer cell, while Fas triggers extrinsic apoptotic
pathways ([Fig. 2]). The death of the
cancer cell with the release of antigens can then lead to a feedback loop of
anti-tumor response.
Fig. 1 Therapies that could influence the cancer immunity
cycle. Source: Dietz A, Stöhr M, Zebralla V et al. Immunonkologie
bei Kopf-Hals-Tumoren. Laryngo-Rhino-Otologie 2021; 100(04):
303–321. doi:10.1055/a-1337–0882
Fig. 2 Effector functions of CD8+ cytotoxic T cells. Source:
Wagener C, Müller O, Hrsg. Molekulare Onkologie. 4., aktualisierte
und erweiterte Auflage. Stuttgart: Thieme; 2022.
doi:10.1055/b000000085
1.3.3 Cancer immune evasion in HNSCC
HNSCC evades immune recognition and destruction in a variety of ways that are
linked to the principles discussed above. Components of the antigen
processing machinery (APM) are under-expressed or mutated, leading to
reduced tumor antigen presentation and less T cell recognition [43]
[42]
[43]
[46], although not to such an extent as
to lead to activation of NK cells than eliminate cells that do not express
HLA. Indeed, a deficient APM has been linked to worse outcomes in HNSCC
[47]. ([Fig. 3])
Fig. 3 P2C way of cross presentation. Source: Wagener C,
Müller O, Hrsg. Molekulare Onkologie. 4., aktualisierte und
erweiterte Auflage. Thieme; 2022. doi:10.1055/b000000085
In some cases, low MHC expression can be upregulated by an interferon-γ (IFN-
γ) response. When IFN-γ connects with its receptor, it triggers the
phosphorylation of Janus kinase 1/2 (JAK1/2) and signal transducer and
activator of transcription 1 (STAT1), setting off the JAK/STAT signaling
pathway. STAT1 functions as a transcription factor, boosting the production
of interferon regulatory factor 1 (IRF1) and p48. This in turn leads to
increased expression of MHC I. However, interferon-γ signaling can be
reduced in HNSCC [48], leading to
impaired antigen presentation and T-cell function [49].
One clinically important way HNSCC can evade immune surveillance and
destruction is the expression of immune checkpoints. In a physiological
setting, these serve the role of limiting the overreaching immune
destruction of healthy tissue in the setting of acute infections as well as
preventing autoimmunity. Cancers can co-opt this mechanism by expressing
inhibitory receptors such as PD1, CTLA-4, LAG-3, TIGIT, or Tim-3. Due to its
clinical application, PD1 has been at the forefront of research interest. In
the interaction between programmed death-1 (PD-1) receptor and its ligands,
programmed death-ligand 1 (PD-L1), and programmed death-ligand 2 (PD-L2)
[50], which are overexpressed on
the surface of tumor cells as well as antigen-presenting cells, the PD-1
receptor transduces an inhibitory signal that attenuates T-cell activation
and effector functions, essentially dampening the immune response ([Fig. 4]). On the level of an
individual T effector cell, an increasing dysfunctional (exhaustion) state
of the T cell is induced, depending on the strength, i. e., high antigen
load, as well as on the duration of the stimulation [51]
[52]. Basic features of depleted T cells include a decrease in
effector functions (cytotoxicity, cytokine secretion), decreased
proliferation, an altered metabolic cell program, epigenetic reprogramming,
and increased expression of inhibitory checkpoint receptors [53].
Fig. 4 Principle of action of the immune checkpoint blockade.
Source: Blum H, Müller-Wieland D, Hrsg. Klinische Pathophysiologie.
11., unveränderte Auflage. Stuttgart: Thieme; 2020.
doi:10.1055/b000000121
A breakthrough that ultimately enabled the emergence of checkpoint inhibitors
as a new class of drugs for the treatment of cancer was the recognition that
T effector cells in chronic viral infections can be rejuvenated or revived
by blocking checkpoint receptors [52]
[53]. It is important to
acknowledge the heterogeneity within dysfunctional T cells, as there is a
hierarchical, graded development of T-cell exhaustion: CD127+KLRG1- effector
T cells develop into at least two subpopulations, PD1midT-bethigh Tex as
well as PD1highEOMEShigh Tex, only the former of which can be revived by
blockade of the PD1/PDL1 axis [56]
[57]. Similarly, cells
co-expressing different checkpoint receptors can be re-activated by a
combination of checkpoint inhibitors [58]. In HNSCC, it has been shown that the extent of PD1
expression is a critical aspect with high frequencies of PD1high patients
were associated with more dysfunctional T cells and worse outcomes [59].
1.3.4 The HNSCC tumor microenvironment
Separate from the immune synapse between the T cell and cancer cell, the
surrounding tumor microenvironment has a powerful influence on the potential
for immune elimination or evasion. Several immunosuppressive cell types have
been described in HNSCC.
The role of regulatory T cells (Tregs) within the tumor microenvironment of
head and neck squamous cell carcinoma (HNSCC) has been characterized,
although their definitive prognostic or therapeutic significance remains
unestablished [60]. Tregs misapply
their physiological function – regulating T-cell hyperactivity and
preventing autoimmunity – to foster an immunosuppressive milieu conducive to
tumor growth [61]. Specifically, Tregs
inhibit antitumor immunity by targeting cytotoxic T cells. Their suppressive
mechanisms encompass the maintenance of high-affinity IL-2 receptor alpha
chain expression, thereby mitigating IL-2-induced activation in effector
cells; expression of immune checkpoint molecules like CTLA-4, which interact
with co-stimulatory molecules CD80/CD86, thereby inhibiting T-cell
activation; and secretion of immunosuppressive cytokines such as IL-10 and
TGF-beta [62]. Emerging evidence from
colon cancer research suggests that the conventional classification of
CD4+FoxP3+Tregs is overly simplistic [63]
[64]. A more nuanced
stratification based on CD45RA and FoxP3 expression may be necessary,
delineating naive (CD45RA- and FoxP3low, nTreg), non-suppressive (CD45RA+and
FoxP3low, nsT-reg), and effector (CD45RA+and FoxP3high) Treg subtypes. Naive
Tregs are recruited to the tumor site, where they subsequently transition
into suppressive effector cells upon antigenic exposure. Specialized
suppressive Treg subsets, characterized by CD39+and Tim3+expression, have
been identified in HNSCC and demonstrated heightened immunosuppressive
potential [65]
[66].
Another cellular subset implicated in intratumor resistance to immunotherapy
and pro-tumor immunity are neutrophil granulocytes. These cells occupy a
dichotomous role, demonstrated to possess both tumor-promoting and
tumor-inhibiting functions [67].
Research in recent years has illuminated the significant influence of
tumor-associated neutrophils (TANs) on tumor angiogenesis and growth,
largely mediated by the secretion of specific cytokines and growth factors
[68]. Additionally, TANs
facilitate metastatic spread and attenuate anti-tumor immune responses by
creating a pre-metastatic niche [69].
Neutrophil plasticity is profoundly shaped by the tumor microenvironment,
exemplified by neutrophil polarization modulated by factors such as type I
interferons, TGF-beta, and G-CSF. Pro-tumoral neutrophils, notably those
emerging in the absence of type I interferons – as seen in IFN knockout
murine models – promote angiogenesis and tumor growth via the upregulation
of proangiogenic molecules like VEGF and MMP9. Further, these neutrophils
also exhibit extended longevity and increased chemokine secretion relative
to their anti-tumoral counterparts [69]
[68]
[69]
[70]
[73]. In HNSCC, in vivo
imaging models have shown a decreased contact between neutrophils and T
cells in interferon receptor deficient (Ifnar1-/-) mice, leading
to dampened T-cell proliferation and activation [74]. The ratio of pro-tumor to
anti-tumor neutrophils can fluctuate in line with tumor progression,
consequently altering their cumulative impact on tumor dynamics [75]. In this context, recent studies
have revealed that in HNSCC, antigen-loaded TANs migrate to lymph nodes,
where they modulate T-cell dependent anti-tumor immune responses in a
stage-dependent manner. In early phases, prior to lymphatic metastasis
(cN0), neutrophils acquire an antigen-presenting phenotype
(HLA-DR+CD80+CD86+ICAM1+PD-L1-) and activate T-cells. At later cancer
stages, lymph node metastases (cN+) produce GM-CSF, inducing the generation
of PD-L1+immunosuppressive neutrophils via STAT3 pathway activation,
subsequently leading to the suppression of T-cell responses and further
tumor progression [76].
Recently, single-cell RNA sequencing (scRNAseq) analyses of the head and neck
tumor environment have helped to resolve the heterogeneity of the head and
neck TME and identify prognostic cell types of interest. scRNAseq is a basic
science and translational research technique that has evolved from a highly
specialized niche method to a mainstream application [77]. Facilitated by various technical
developments, there has been an explosion of experimental platforms and an
associated popularity of the method in recent years [78]. In simple terms, single-cell RNA
sequencing allows the representation of the entire transcriptome of a sample
while maintaining single-cell resolution. A bioinformatic map of individual
cells as well as their mRNA content is thus generated. Depending on the
method, millions of cells with, on average, thousands of genes can be read
out in this way. In HNSCC, scRNAseq has helped to delineate critical
differences between the immune makeup of HPV-associated and
non-HPV-associated HNSCC, showing that helper CD4+T cells and B cells are
divergent between these two etiologies [79]. In addition, CD4+T follicular helper cell gene expression
signature is associated with longer progression-free survival in HNSCC
patients. Further, germinal center tumor-infiltrating B cells and tertiary
lymphoid structures show the favorable outcome associated with HNSCC [80]. Further, one can use scRNAseq
HNSCC to explore the role of non-immune cells and their interaction with the
immune system, showing the cellular heterogeneity among cancer cells,
pericytes, fibroblasts, and endothelial cells [81]. When analyzing the spectrum of
intratumor T cells analogous to studies in other entities – malignant
melanoma, colorectal carcinoma, hepatocellular carcinoma as well as
non-small cell lung carcinoma – it becomes clear that the exhaustion state
of T cells is a continuum.
2. Immunotherapy in the neoadjuvant setting
2.1 Potential risks and benefits of moving immunotherapy to the neoadjuvant
setting
Before discussing the rationale for new adjuvant immunotherapy in detail, it is
important to keep front and center in one’s mind the important differences
between checkpoint inhibition and classical to cytoreductive chemotherapies.
While immunotherapy aims to enhance the body’s own anti-tumor response,
classical chemotherapy agents that are used in HNSCC interfere with the ability
of rapidly dividing cells to replicate. Specifically, the most widely used
agent, cisplatin, functions predominantly through the formation of intrastrand
and interstrand DNA adducts [82]. This
induces conformational changes, triggering a cascade of cellular responses,
including impaired DNA repair mechanisms, cell cycle arrest, and apoptosis. In
the primary or induction setting, they are therefore applied primarily to
alleviate symptoms of large clinical disease and to debulk tumors before the
start of primary radiotherapy. In the context of combination therapy, cisplatin
serves as an effective radiosensitizer enhancing the cytotoxic effects of
ionizing radiation on cancer cells. These DNA adducts formed by cisplatin act
synergistically with radiation-induced breaks, complicating their repair and
consequently promoting apoptosis.
Given the basic science and clinical background discussed so far, there are
several reasons that support the use of immunotherapy, especially checkpoint
inhibition, in the neoadjuvant, pre-surgical setting in HNSCC.
First and foremost, given the poor outcomes of current therapies, there is a
clear unmet medical need to intensify treatment regimens. Yet, given the
morbidity and deteriorating impact on the quality of life of patients undergoing
current treatment, it is apparent that current modalities will be unable to
achieve this. Here, adding a fourth modality with a distinct mechanism of action
and adverse event spectrum to the primary setting seems common sense.
This is especially true given the known clinical efficacy and relatively good
safety profile of checkpoint inhibition. Further, extensive clinical experience
has been gathered in the last decade regarding monitoring and treating adverse
events of checkpoint inhibitors in the palliative setting across a wide variety
of primary cancers. As will be discussed in more detail below, immunotherapy has
shown a moderate but clinically significant survival advantage in the
recurrent-metastatic setting while maintaining a superior safety profile.
In fact, there is reason to believe that response rates might be higher in the
presurgical compared to the adjuvant or recurrent-metastatic setting due to a
variety of patient and tumor-related factors. First, in untreated cases, there
is more tumor tissue and therefore increased opportunity for immune interaction
with the cancer cells. This might be true for the absolute quantity of tumor
antigens as well as the variety and quality of antigens since they have not been
selected by the evolutionary pressures of long-term disease and multiple prior
therapies. Therefore, there might be a stochastically higher likelihood of
effective antitumor response at an initial stage. The same holds true in
principle for the tumor-draining lymph nodes, which in HNSCC are usually removed
during surgery or heavily radiated as part of primary radiotherapy. In the
untreated setting, priming of T cells by dendritic cells can occur here
unimpeded. On the patient side, there is a lower probability early in a
treatment course of treatment-related impairment of the immune response. The
indiscriminate cytotoxic effect of chemotherapy in the primary or adjuvant
setting can particularly dampen the patient’s ability to generate an effective
anti-tumor response. Similarly, extensive surgery with the need for cumbersome
recuperation or the strain of weeks-long radiotherapy might limit the immune
system’s capabilities.
Along the same lines, early stimulation of the immune system using checkpoint
inhibition might not only lead to a deeper primary response but also a
longer-lasting one, ideally even permanent remission. This can occur due to a
more effective memory formation. An increase in the breadth and quantity of
memory effector T cells that patrol the body after primary therapy might be
better equipped to engage and eliminate microscopic residual disease locally or
in distant micro-metastases. Clinically, this could lead to reduced local and
regional, as well as distant recurrence and therefore improve overall and
recurrence-free survival.
At the same time, new adjuvant immunotherapy might lead to a significant response
of the tumor at presentation. This could – in theory and after rigorous
evaluation in respective clinical trials that have identified reliable response
markers – lead to a reduced need for extensive resection and reconstruction or
improve the possibility of organ-sparing protocols. Similarly, size reductions
in the primary tumor or local lymph node metastasis could lead to decreasing
high-risk clinical features such as close-margin resections or extracapsular
spread, which in turn could minimize the need for adjuvant therapies and their
respective treatment sequelae. Thus, the addition of a fourth treatment modality
could offer an opportunity for treatment de-escalation or at least a reduction
in morbidity in select patient populations.
Last but not least, the period between panendoscopy and diagnosis and the
surgical primary treatment offers an aptly termed window of opportunity. This
can be used to study multiple basic science and translational research questions
[83]. These include, but are not
limited to, details of the tumor-host interaction, mechanisms of resistance, the
effects of immunosuppressive cell populations as well as biomarkers for response
or high-risk scenarios warranting adjuvant therapy. Ideally, this induces a
positive feedback loop from innovations going from bench to bedside and
back.
Given the possible advantages of early immunotherapy in the untreated setting, it
is important to consider possible disadvantages so that these can be monitored
or avoided. Given the theoretical increase in treatment response in the primary
setting, there is a similar risk of more pronounced side effects in populations
with a healthy immune system. This could mean a higher frequency as well as a
more severe extent of adverse events, especially autoimmune diseases. In the
neoadjuvant treatment context, it is important to consider that these patients
are in a potentially curable disease stage and have to live with permanent side
effects, especially autoimmune diseases such as diabetes, hypothyroidism, or
hypophyseal dysfunction, for the rest of their lives. Also, Grade 3 or 4 adverse
events might interfere with the patient’s scheduled surgery date or allow the
tumor to progress while the patient recuperates. Further, it is important to
consider the possibility of a negative trial in which all patients do not
benefit sufficiently from adding immunotherapy prior to primary treatment.
One final practical aspect, is that the multi-modal approach of integrating
neoadjuvant immunotherapy in to the standard of care may present a unique set of
challenges, particularly in its applicability outside the controlled environment
of clinical trials. It requires a collaboration among various medical
disciplines, including oncology, surgery, radiology, and pathology specialists.
Coordinating care across these diverse fields can be complex, potentially
leading to logistical and communication challenges. For patients, navigating
this treatment landscape in the real-world setting can be daunting, especially
when transitioning between different phases of therapy and dealing with multiple
healthcare providers. The complexity of scheduling, understanding the different
aspects of the treatment, and managing the side effects that may arise from such
a comprehensive approach could pose significant challenges. These considerations
highlight the importance of optimized care pathways and treating patients at
specialized centers.
In conclusion, while the potential of neoadjuvant immunotherapy in the treatment
of HNSCC is indeed promising, it is imperative to apply it in this early setting
with a measured and cautious approach. To what extent it can be implemented
depends critically on the design and execution of carefully crafted clinical
trials and correlational studies. These must be structured to not only
investigate the potential benefits but also to thoroughly assess any associated
long-term harm or rare adverse effects. The clinical trials should be tailored
to capture a broad spectrum of patient responses, ensuring that the findings are
representative and applicable to the diverse patient population affected by
HNSCC. One example of such a clinical study is the PIONEER - Window of
opportunity study of preoperative immunotherapy with
atezolizumab (Tecentriq) in local head and neck squamous cell carcinoma
(NCT04939480) - trial currently recruiting in Essen.
2.2 Lessons learned from neoadjuvant chemotherapy in HNSCC
Neoadjuvant treatment strategies have been applied to HNSCC in the curative
pre-surgical setting before the era of immunotherapy with a similar intention –
hoping to improve unsatisfying patient outcomes. Here, high-level data from a
randomized Phase-3 trial in resectable Stage III or IVA HNSCC randomized 256
patients to chemotherapy (Docetaxel, Cisplatin, Fluorouracil) plus surgery or
surgery alone, with adjuvant radiotherapy in both groups [84]. Even though this study proved that
there is good feasibility, with 91.6% of patients undergoing surgery within four
weeks of chemotherapy, and good clinical response, with an 80.6% RECIST response
rate and a 27.7% pathological response rate, the study missed its primary
endpoint, showing no benefit in overall or recurrence-free survival. Similarly,
for chemotherapy before radiotherapy – termed “induction” in this context – the
large, updated MACH-NC meta-analysis of 93 trials and 17,493 patients shows no
benefit from this approach (hazard ratio 0.96, 95% CI 0.90–1.02) [85].
Given the discrepancies between clinical response and its translation into a
survival benefit for the patient, which ultimately guides treatment decisions,
this provides a cautionary tale for current neoadjuvant trials.
2.3 Lessons learned from immunotherapy in the RM setting in HNSCC
Since the introduction of checkpoint inhibition to the recurrent/metastatic
setting in the first line as well as in platinum-refractory patients, there has
been ample opportunity to study treatment effects, safety profiles, and
biomarkers of response.
The initial trials that led to the FDA approval of nivolumab and pembrolizumab
have already shown a moderate but significant survival benefit of immunotherapy
over conventional chemotherapy in this setting. Long-term follow-up of these
cohorts as well as ongoing Phase IV studies highlighted that there is a small
but persistent subgroup of patients who experience a durable response, even cure
from their palliative disease. In a 2-year follow-up of the Checkmate-141 trial,
overall survival was 16.9% in the nivolumab versus 6.0% in the investigator’s
choice group [86]. Further, long-term,
4-year follow-up data from the KEYNOTE-048 trial suggests a plateau in overall
survival at around 20% in the pembrolizumab alone group and the
pembrolizumab-chemotherapy group (total cohort and CPS≥1) [87]. Similar results can be inferred from a
pooled analysis of the initial and the expansion cohort of the KEYNOTE-012 trial
where 71% of the patients responded to pembrolizumab, and this response lasted
more than a year [88]. This suggests that
even though a relatively small fraction of patients benefit, those who do have a
durable response are likely due to memory formation of the adaptive immune
system.
Importantly, the adverse event profile of checkpoint inhibitors proved to be
favorable in the long term with 7.2% percent of patients in the Check-Mate-141
trial experiencing serious adverse events compared to 15.3% in the conventional
therapy arm [86]. Along the same line,
long-term data from KEYNOTE-048 underlines the relatively better safety profile
of checkpoint inhibition: 17.0% of patients receiving pembrolizumab monotherapy
experiencing adverse events greater Grade 3 versus 71.7% in the
pembrolizumab-chemotherapy and 69.3% cetuximab-chemotherapy group [87].
Especially relevant and interesting in this setting is the introduction of
quality-of-life measures as exploratory endpoints. In the Checkmate-141 trial,
patient-reported outcomes (PROM) were collected using three questionnaires [89]. In the EORTC Quality of Life
Questionnaire-Core 30 (QLQ-C30), clinically meaningful deterioration was
observed in 8/15 (53%) domains in the chemotherapy group, while PROM stabilized
in the nivolumab group. A similar effect was seen in the EORTC head and neck
cancer-specific module (EORTC QLQ-H&N35). Further, the three-level European
Quality of Life-5 Dimensions (EQ-5D), an overall health measure, was also able
to show a benefit in the nivolumab group.
Trials in the current metastatic setting have been used to extract predictive
biomarkers. Expression of PDL1 is an obvious but imperfect predictor of response
to checkpoint inhibition targeting the PD1-PDL1 interaction [90]. Exploratory biomarker analysis of the
Checkmate 141 trial suggested that a tumor PD-L1 expression≥1% might have a more
pronounced effect [31]. In KEYNOTE-040, a
randomized, open-label, Phase 3 trial of pembrolizumab versus investigator’s
choice in HNSCC that progressed under platinum-based therapy, there was a clear
survival advantage depending on PDL1 staining. In the combined positive score
(CPS)≥1 versus <1, the hazard ratio was 0.74 (95% CI 0.58–0.93, p=0.0049)
versus 1.28 (95% CI 0.80–2.07, p=0.8476), respectively [91], with comparable results for a tumor
proportion score (TPS)≥50% or < 50%. Similarly, a subgroup analysis of the
KEYNOTE-048 trial, using pembrolizumab in the first-line setting, showed an
increased survival benefit with higher PD-L1 expression, measured as CPS [92].
2.4 Lessons learned from adjuvant immunotherapy in HNSCC
Seeing the fact that the low overall survival in HNSCC patients is primarily
driven by local or regional recurrence rather than patients showing distant
metastasis primarily or secondarily, there seemed to be sufficient rationale to
introduce immunotherapy to the treatment regimen in the adjuvant setting
following primary chemoradiation in the hopes of mitigating loco-regional
recurrence.
The Javelin Head and Neck 100 trial was a randomized, double-blinded,
placebo-controlled study designed to assess the superiority of the anti-PD-L1
antibody avelumab over placebo after primary chemoradiotherapy in 697 patients
[93]. The trial was stopped at a
preplanned interim analysis, showing that the primary objective,
progression-free survival as determined through RECIST criteria, was not
met.
In a similar setting and cohort, it was announced that KEYNOTE-412, a randomized,
double-blind, Phase III trial testing pembrolizumab or placebo concurrently with
primary chemoradiotherapy and as maintenance treatment, did not meet its primary
endpoint of event-free survival, though published results are not yet available.
A third randomized Phase III trial, the IMvoke010, which tests atezolizumab
versus placebo after primary chemoradiotherapy, is currently actively recruiting
without any information on results.
These negative results came unexpectedly, as adjuvant immunotherapy has proven
advantageous in other disease entities [94]. Translational studies have since offered a compelling
explanation of the missing effect of adjuvant immunotherapy in HNSCC, while at
the same time underlining its possible value in the neoadjuvant setting. After
developing a murine neck dissection model, it was shown that the surgical
removal of the tumor-draining lymph nodes inhibited the response to subsequent
anti-PD1 or anti-CTLA4 therapy. Similarly, radiation to the local lymph node
basin led to diminished anti-CTLA4 response [95]. Further, elective nodal irradiation, as it is performed as
standard-of-care in the current primary and adjuvant protocols of HNSCC, was
evaluated in another murine model. Here, reduced tumor control, systemic
immunity, and T-cell-specific immune response were shown [96]. These findings are supported by
ex-vivo analysis of systemic biomarkers of patients undergoing chemoradiotherapy
that showed an increase in immunosuppressive cell types when comparing pre- to
post-treatment samples [97]. Taken
together, these studies can help to explain why therapies targeting the local
tumor-draining lymph nodes might limit the host’s anti-tumor immune response.
Furthermore, this suggests treatment should be timed so that checkpoint
inhibition is administered prior to surgical or radiotherapeutic lymph node
ablation, as done in neoadjuvant treatment regimen.
2.5 Lessons learned from neoadjuvant immunotherapy in non-HNSCC
cancers
Considering the need for improved outcomes in diseases outside the head and neck,
it is unsurprising that immunotherapy has been incorporated in other disease
entities in the neoadjuvant setting. Indeed, there have been remarkable clinical
responses and patient outcomes even leading to FDA approval for some of these
indications.
Given the pathological and clinical similarities, trials in non-small cell lung
cancer (NSCLC) can provide an apt comparison and model for head and neck cancer
trials. Indeed, the first clinical trial investigating the preoperative use of
anti-PD1 therapy was conducted in NSCLC in which 21 patients received nivolumab
prior to surgical tumor resection [98].
Even this early and small trial provided insights that foreshadow future
investigations. Nearly half of the patients showed major pathologic response
with >10% tumor regression in their surgical specimens. Also, there was a
marked difference between radiological response as measured by RECIST criteria
and the histological findings in the tumor specimen, the former significantly
underestimating the effect of the neoadjuvant treatment. Also, this trial showed
the treatment to be safe, with no major adverse events or delays of surgery.
Consequently, a large Phase III trial, Checkmate-816, randomized 385 patients
with resectable NSCLC patients to chemotherapy with or without nivolumab for
three months followed by surgery [99].
There was no disadvantage in terms of safety, additional toxicity, or delays in
surgery in the treatment arm with added immunotherapy. Indeed, there was less
need for extensive surgery in this group. In terms of outcomes, the event-free
survival (EFS) was increased by approximately one year in the experimental
group. Analogous to these findings, the percentage of patients showing complete
pathological response increased from 2.2% in the chemotherapy alone group to
24.0% in patients receiving additional immunotherapy. This landmark study
underlines the profound potential effect of edit immunotherapy in the
pre-surgical setting.
Another disease entity in which neoadjuvant immunotherapy is now the standard of
care is triple-negative breast cancer, where it was previously only neoadjuvant
chemotherapy. Keynote 522, a Phase III trial that randomized 1174 patients to
chemotherapy plus either pembrolizumab or placebo followed by surgery, was the
first trial that ultimately led to FDA approval for a neoadjuvant checkpoint
inhibition (DOI: 10.1056/NEJMoa1910549). It was able to show an increase in
event-free survival from 76.8% in the chemotherapy group to 84.5% experimental
group, as well as an increase in pathological complete response from 56% to
63%.
One further entity that can serve as an interesting case study for the power of
identifying highly predictive biomarkers for response in neoadjuvant
immunotherapy is colon cancer with mismatch repair deficiency (dMMR). dMMR
constitutes an important molecular aberration in a small subset of colon cancers
closely associated with the microsatellite instability-high (MSI-H) phenotype.
This deficiency arises from the loss of function in key proteins involved in the
mismatch repair system – namely MLH1, MSH2, MSH6, and PMS2 – resulting in a
failure to correct base-pair mismatches during DNA replication. Consequently,
dMMR leads to an accumulation of errors in microsatellite sequences, manifesting
as MSI-H. This hypermutated state not only contributes to tumorigenesis but also
leads tumors to be more immunogenic due to a higher tumor mutational burden and
the presentation of novel neoantigens. Clinically, tumors exhibiting dMMR are
generally associated with a better prognosis in early-stage cases but
paradoxically may be less responsive to traditional chemotherapy agents like
fluoropyrimidines, which form the backbone of colon cancer treatment. This
chemoresistance necessitates alternative treatment paradigms for dMMR patients.
In a small single-arm Phase 2 study, dostarlimab, an anti-PD-1 monoclonal
antibody, was given for 6 months prior to planned chemoradiotherapy and surgery.
However, 12/12 patients (100%, 95% CI, 74–100) showed no sign of residual tumor
on endoscopic evaluation with biopsy, MRT, or PET-imaging [100], leading to none of the patients
receiving subsequent treatment. This highlights the potential curative power of
immunotherapy monotherapy in a highly targeted subgroup of patients.
2.6 Clinical results from neoadjuvant immunotherapy trials in HNSCC
Neoadjuvant immunotherapy trials that have published results are summarized in
[Fig. 5] ([Fig. 5]). One can appreciate the great
diversity of treatment regimens, with variables including the treatment drug(s),
dosing, number of cycles, immunotherapy combinations, duration of the
neoadjuvant phase, adjuvant therapy, and even combination with pre-operative
radiation or targeted therapy.
Fig. 5 Overview of published neoadjuvant immunotherapy trials in
HNSCC. Pembro=Pembrolizumab, Nivo=Nivolumab, Ipi=Ipilimumab,
Durva=Durvalumab, Treme=Tremelimumab, Camre=Camrelizumab,
Toripa=Toripalimab, pCR=pathological complete response, MPR=major
pathological response
2.6.1 Immunotherapy-only trials
Several trials have investigated presurgical treatment with immunotherapy
only. In the first published report, a multicenter Phase II trial
administered a single dose of pembrolizumab in 36 patients with
non-HPV-associated HNSCC in a 2–3-week window before surgery, followed by
risk-adapted adjuvant chemoradiotherapy. This protocol proved safe, as no
Grade 3 or 4 adverse events or delays in surgery occurred. In terms of
response, a pTR≥50% was achieved in 22% of patients and 10–49% in another
22%. There was no pCR [101]. A
single-center Phase II trial in 29 patients with oral cavity cancers
randomized to either nivolumab at Week 1 and 3 or nivolumab plus ipilimumab
at Week 1 and nivolumab at Week 3 followed by surgery within one Week. Here,
13% of patients developed Grade 3 or 4 adverse events in the nivolumab arm
and 33% in the nivolumab plus ipilimumab arm. There, again, was no delay in
surgery dates. A pTR 10–49% was observed in 38% of the patients receiving
nivolumab and 40% in the combination arm, while a pTR≥50% was seen in only
15% of patients receiving nivolumab but 33% in the nivolumab plus ipilimumab
arm, including one patient with pCR (7%) [102]. A similar protocol and patient cohort were studied in the
IMCISION trial; in the safety run-in Phase Ib part of the trial, 6 patients
were treated with nivolumab in Weeks 1 and 3, while 6 patients received
nivolumab plus ipilimumab at Week 1 followed by nivolumab at Week 3. The
trial was prolonged to a single-arm IIa extension cohort with 20 patients
receiving the latter combination treatment. Safety evaluation showed 33% of
patients in the nivolumab group and 38% of nivolumab plus ipilimumab
patients having Grade 3 or 4 adverse events, with none resulting in the
delay of surgery. Patients were classified into major pathological response
(MPR), partial pathological response (PPR), or no pathological response
(NPR) based on previously described criteria from melanoma studies [103]. In the nivolumab group, an MPR
was observed in 17% of patients, while in the combination arm, 35% of
patients had an MPR including 4% with pCR [104]. A more individualized single-arm trial was conducted in 12
patients with oral cavity squamous cell carcinomas, where, after giving
nivolumab three times biweekly, a clinical and radiographic re-evaluation
determined whether a fourth dose was given [105]. There were no Grade 3 or 4 adverse events definitely or
possibly related to neoadjuvant immunotherapy, and there was no delay in
surgery. Response was measured by comparing the surgical specimen’s maximum
tumor diameter with the single greatest tumor dimension on pretreatment
imaging, defining a partial response as a >30% reduction. Of the
patients, 33% showed stable disease, 33% showed a partial pathologic
response, 0% a complete pathological response, and 33% a disease
progression. One more recent, larger, multicenter Phase II trial evaluated
single-dose pembrolizumab one to three weeks prior to surgery in 96
patients. Partial pathological response, defined as tumor regression≥20% to
<90% was achieved in 32% of patients, while major pathological
response,≥90% tumor regression, was seen in 7% [106].
Two trials investigated the neoadjuvant immunotherapy paradigm predominantly
in the context of oropharyngeal cancer. CheckMate 358, a multi-center
multi-cohort trial contained neoadjuvant HNSCC cohorts of HPV-associated and
non-HPV-associated cancers, recruiting 52 patients who received nivolumab in
Week 1 and Week 2 followed by surgery in Week 4 [107]. Grade 3 or 4 adverse events were
observed in 19.2% of the HPV-associated and 11.5% of the non-HPV-associated
cancers, with no delays in surgery due to adverse events. Pathological
response was judged by evaluating residual viable tumor (RVT) with pCP
equaling 0% RVT, major pathologic response (MPR) ≤10% RVT, and pathologic
partial response (pPR, > 10%- 50%RVT). Out of 34 evaluable patients, 7%
of HPV-associated HNSCC, achieved MPR and 18% pPR, while non-HPV-associated
patients achieved pPR in 6% of cases. In a similar patient cohort, the CIAO
trial randomized patients to two cycles of durvalumab versus durvalumab plus
tremelimumab. Severe adverse events in Grades 3 or 4 were observed in 20% of
patients in the durvalumab group versus 7% in the durvalumab plus
tremelimumab group. Major pathologic response (MPR), defined as ≤10% viable
tumor, was achieved in 7% of the primary tumor in both arms and 50% of the
lymph nodes in the durvalumab group versus 22% in the combination group
[108].
One trial that is particularly specialized in its treatment and study
population was a Phase II trial of single-dose nivolumab and lirilumab
(anti-KIR) in 28 patients with recurrent but surgically salvageable HNSCC
[109]. There were no delays in
surgery due to adverse events. For the patients in the study, ≤10% viable
tumor cells, defined as a major pathological response (MPR), were achieved
by 14% of patients, and pathologic partial response (pPR) (≤50% tumor
viability) by 29% of the patients.
2.6.2 Immunotherapy in combination with other agents
There are multiple published studies that combine preoperative immunotherapy
with other treatment modalities, including targeted therapy, chemotherapy,
and radiotherapy.
In a two-arm, multi-institutional trial, nivolumab at Weeks 1 and 2 was
combined with daily phosphodiesterase-5 inhibitor (tadalafil) followed by
surgery at Week 4 in 45 patients. There were no Grade 3 or 4 adverse events
and no delays in surgery. Patients with pathological tumor response≥20% were
defined as responders, >0%- < 20% as minimal responders, 0% as
non-responders, and 100% as complete responders. Across both cohorts, 51%
had a response with an additional 7% experiencing a complete response. There
was no difference in terms of pathological response in patients receiving
tadalafil [110].
In another combination trial, 10 patients with oral cavity carcinomas were
treated with one dose of nivolumab at Week 2, combined with daily
Sitravatinib, an oral receptor tyrosine kinase inhibitor, followed by
surgery at Week 3 [111]. There was one
Grade 3, but no Grade 4, treatment-related adverse event. There was one
Grade 2 thrombocytopenia, which led to a two-week delay in surgery. Of the
patients, 10% achieved complete pathological tumor response (cPTR) with 0%
residual tumor cells, 20% of patients major response (mPTR) with <10%
residual viable tumor, and the other 70% incomplete response.
A Phase I trial of 20 patients with oral squamous cell carcinoma combined
camrelizumab, an anti-PD-1 monoclonal antibody, at Weeks 1, 2, and 4 with
four weeks of oral apatinib, a tyrosine kinase inhibitor that inhibits
vascular endothelial growth factor 2. There were no Grade 3 or 4 events in
the preoperative phase, one surgery was postponed by one week due to
elevation in cardiac troponin I, which recovered spontaneously. Residual
viable tumor cell content was evaluated, with 40% of patients showing major
pathologic response (<10% residual viable tumor), including 10% of
complete pathological response. Notably, 95% of patients had a tumor
response of≥50% [112].
In a Phase I trial, 14 patients were treated with 1 or 2 doses of bintrafusp
alfa, a bifunctional fusion protein composed of the TGF-β receptor II linked
to anti–PD-L1, followed by surgery. Of the patients, 7.1% developed Grade 3
adverse events. There were no complete or major pathologic responses, and
36% of patients showed a partial response (>50% regression) [113].
Several recent trials have explored the combined effect of chemotherapy and
immunotherapy in the neoadjuvant setting. A single-center, single-arm Phase
II trial evaluated the effect of paclitaxel or docetaxel plus cisplatin in
combination with camrelizumab, an anti-PD1 monoclonal antibody, for three
cycles. In terms of safety, 6.3% of patients experienced Grade 3 adverse
events, but there were no Grade 4 toxicities, delays of surgery, or trial
discontinuation. Major pathologic response, ≤10% residual viable tumor
cells, was achieved in 74.1%, including 37.0% with a complete pathological
response [114]. In a similar study of
oral cavity squamous cell carcinoma, 20 patients received two cycles of
paclitaxel, cisplatin, and toripalimab, an anti-PD1 monoclonal antibody. Of
the patients, 15% experienced Grade 3 or 4 adverse events, none of which led
to treatment discontinuation or delay of surgery. Major pathologic response
showing ≤10% residual viable tumor cells was seen in 60% of patients, with
30% achieving complete pathological response [115]. In a further study, a single-arm
Phase Ib trial, two cycles of gemcitabine and cisplatin with toripalimab
were administered to a mixed cohort of 23 patients with HNSCC. Grade 3
adverse events occurred in 13.0% of patients and Grade 4 in 8.7%, with no
treatment-related delays of surgery. Of the patients, 44.4% achieved major
pathological response, including 16.7% who achieved complete pathological
response [116]. Another single-arm,
single-center study with a mixed cohort of HNSCC investigated 2–3 cycles of
pembrolizumab with cisplatin and paclitaxel in 22 patients. There were Grade
3 toxicities in 9.2% of patients, but no Grade 4 events and no
treatment-related delays in surgery. Major pathological response was 54.5%,
including 36.4% with pathological complete response [117].
One distinct single institution Phase Ib trial investigated the role of
neoadjuvant nivolumab with added stereotactic body radiation therapy (SBRT)
before surgery in 21 patients divided into four treatment groups differing
by the amount of radiation, 40 Gy versus 24 Gy, HPV-status, and nivolumab
treatment. There were no delays in surgery due to adverse neoadjuvant
treatment effects. Across all cohorts, mPR was 86%, including 67%
pathological complete response [118].
3. Open questions and challenges in neoadjuvant immunotherapy
3.1 Safety and adverse events profile
Considering the curative intent in non-metastatic HNSCC, the feasibility and
safety of a neoadjuvant approach were the main concerns in Phase I and II
studies published thus far. As has been reviewed extensively above, adverse
events seem to be rare and manageable ([Fig.
6]). A meta-analysis of 344 patients, not including the more recent
trials of combined immunochemotherapy, calculated the rate of preoperative Grade
3 to 4 adverse events to be 8.4% [119].
Importantly, across all the studies reviewed above there were only two delays in
surgery reported, one for two weeks due to thrombocytopenia [111] and one for one week due to
self-limiting troponin increase [112].
Thus, unless large Phase III trials report rare severe or long-term adverse
events, checkpoint inhibition is to be considered safe in the neoadjuvant
setting. Even though some studies have reported surgical complications, their
impact following neoadjuvant (chemo)immunotherapy and how this impacts morbidity
and quality of life has not been explored consistently and systematically.
Fig. 6 Occurrence of various toxicities depending on the duration
of therapy with PD-1 and PD-L1 inhibitors (Daten aus [32]). Aus: https://cme.thieme.de/cme-webapp/#journals/0935–8943/a_1337_0882_toc/10.1055-a-1337–0882
3.2 Radiographic assessment of response
Similar to the observations made in lung cancer [98], there has been a marked discrepancy between imaging, measured
using RECIST criteria, and the response seen in histology. Even in the first
reported study, two of the three patients with progressive disease as measured
by RECIST had a pathological tumor response of 10–49% and > 50%, respectively
[101]. Analogous observations have
been made in other studies. The test validity criteria for MRI in detecting
major pathological response were evaluated in one study, showing a high
specificity of 100% but a low sensitivity of 29% [104]. Conversely, in the same cohort, a
metabolic tumor volume or total lesion glycolysis decrease pre- and
post-immunotherapy was identified as a potential marker to identify response
[120]. Taken together these studies
suggest that though hybrid imaging might be able to identify responders,
conventional imaging using MRT or CT is unable to differentiate stable or
progressive disease from a successful anti-tumor immune response.
3.3 Pathologic assessment of response
Due to the limitations of radiographic imaging to determine a tumor response in
the context of immunotherapy, pathological response criteria in the surgical
specimen seemed to be the optimal candidate to assess efficacy. Leaning on
experience from neoadjuvant chemotherapy from the pre-immunotherapy era,
respective criteria have been developed and shown to be prognostic markers in
several cancer types. Complete pathological response, with no viable tumor
cells, as well as major pathological response, meaning <10% viable tumor
cells, are the most consistently used and described methods in this context
[121]. Even though these metrics have
been applied to some extent in the neoadjuvant immunotherapy trials described
above, there is great variability in how they are used. Some studies created
customized criteria such as pTR 10–49% versus pTR > 50% [101] or comparing the surgical specimen
maximum tumor diameter with the single greatest tumor dimension on pretreatment
imaging [105]. Other studies reported pCR
and MPR but set individual cut-offs for partial pathological response such as
20%- 90% [106], ≤ 50% tumor viability
[109], or >20% [110]. It is important to consider that most
of these criteria were developed based on response to chemotherapy and not
immunotherapy. This has been highlighted in the context of NSCLC, where the
pathologic features might be different in the neoadjuvant immunotherapy setting,
leading to the development of separate immune-related pathologic response
criteria (irPRC) [122]. However, these
might not be applicable to non-NSCLC entities, as even within NSCLC the optimal
cutoff of percent viable tumor differed between adenocarcinoma and squamous cell
carcinoma [123]. This holds true even
though a pan-tumor pathologic scoring system was developed that included HNSCC
samples [124].
One further point of controversy is to what extent the tumor-draining lymph nodes
should be incorporated into the pathological response metric, where a divergence
between the primary and lymph node response has been described in several HNSCC
studies [108]
[115]
[125]. This highlights the importance of developing HNSCC-specific
pathologic response criteria and cut-offs, most likely using 10% increments of
residual viable tumor – possibly separately for the primary and tumor-draining
lymph nodes, which can then facilitate the comparison across multiple studies.
Alternatively, the pan-tumor scoring system needs to be further validated in
HNSCC.
3.4 Pre- and intraoperative decision-making
One aspect that has thus far not been explored in detail is the determination of
the extent of surgery. In this early phase of the neoadjuvant treatment paradigm
in HNSCC, most studies reported to have operated in the pre-therapeutic tumor
borders with some taking detailed care by tattooing tumor borders or using
pre-immunotherapy imaging and photography as guidance [115]. With larger trials, more experience,
and better data and presurgical markers of response, it might be possible to
tailor the surgery more directly to the patient’s remaining disease after
neoadjuvant treatment. This is especially relevant in the setting of disfiguring
surgeries such as laryngectomies, exenteratio orbitae, or ablatio nasi. Here,
new organ-sparing protocols could be tested.
3.5 Treatment combination and timing
Bearing in mind the large possible number of combinations using immunotherapy
agents (e. g., anti-PD1, anit-CTLA4, anti-LAG3, etc.), conventional
chemotherapy, targeted therapy, radiation, and even more experimental treatment
such as oncolytic viruses or therapeutic vaccination, finding the optimal
treatment regimen for each individual patient remains the most challenging task.
One must be careful when comparing small single-arm non-randomized trials with
varying regimens and divergent outcome measures. In terms of response data,
complete and major pathological responses have been the most widely and
consistently reported measures. Given the data we have thus far, there seems to
be an added benefit of adding targeted therapy or chemotherapy to checkpoint
inhibition. Here, the reported complete response rates have increased from
nivolumab (0%) [101], nivolumab plus
ipilimumab (7%) [102], or 4% [104] to 10% in two reports combining
immunotherapy with targeted therapy [111]
[112]. This increased
further in the setting of combination with chemotherapy to 37.0% [114], 30% [115], 16.7% [116], and 36.4%
[117], an assessment which is
supported by a recent metanalysis of ORR comparing neoadjuvant immunotherapy to
immunochemotherapy [126].
