Precision immunotherapy in oesophageal squamous cell carcinoma: Molecular pathogenesis and checkpoint inhibitor response prediction

Abstract

Oesophageal squamous cell carcinoma (ESCC) remains a significant global health challenge due to its aggressive nature and poor prognosis. Recent advances in immunotherapy, particularly immune checkpoint inhibitors (ICIs), have revolutionized treatment paradigms. This review comprehensively examines the molecular pathogenesis of ESCC, highlighting key oncogenic pathways and immune evasion mechanisms. Furthermore, it explores the evolving landscape of biomarkers and predictive models for ICI response, emphasizing the critical role of precision immunotherapy in tailoring treatment strategies. A deeper understanding of these molecular and immunological underpinnings is essential for optimizing patient selection, overcoming resistance, and ultimately enhancing therapeutic efficacy in ESCC.

Key Words: Oesophageal squamous cell carcinoma; Precision immunotherapy; Immune checkpoint inhibitors; Molecular pathogenesis; Predictive biomarkers; Tumour microenvironment

Core Tip: Precision immunotherapy is transforming the management of oesophageal squamous cell carcinoma (ESCC). This review synthesizes current knowledge on the molecular drivers of ESCC and the complex tumour-immune interactions that dictate response to immune checkpoint inhibitors (ICIs). We detail established and emerging biomarkers-such as programmed death-ligand 1 expression, tumour mutational burden, and circulating tumour DNA-for predicting ICI efficacy. The discussion extends to mechanisms of resistance and future directions, including novel immune targets and multi-omics integration for personalized treatment strategies, aiming to guide clinicians and researchers toward more effective, tailored immunotherapeutic interventions.

INTRODUCTION

Oesophageal cancer ranks as the seventh most common cancer globally and represents the sixth leading cause of cancer-related mortality, posing a substantial burden on healthcare systems worldwide[1]. Oesophageal squamous cell carcinoma (ESCC) constitutes the predominant histological subtype, accounting for approximately 90% of oesophageal cancer cases in high-incidence regions such as East Asia and parts of Africa[2]. Characterized by late-stage presentation and a proclivity for early metastasis, ESCC is associated with a dismal 5-year overall survival rate of less than 20% for advanced disease, underscoring the limitations of conventional therapies including, surgery, chemotherapy, and radiotherapy[3].

The advent of immunotherapy, particularly monoclonal antibodies targeting immune checkpoints like programmed death-1 (PD-1) and programmed death-ligand 1 (PD-L1), has introduced a transformative therapeutic avenue. Landmark trials have demonstrated that immune checkpoint inhibitors (ICIs) can elicit durable responses and extend survival in a subset of patients with advanced disease[4,5]. However, objective response rates remain variable and generally modest, observed in only 15%-30% of unselected patients[6]. This heterogeneity in treatment efficacy starkly highlights the urgent need for precision medicine approaches. Such strategies must be grounded in a robust understanding of the tumour’s intrinsic molecular drivers and the dynamic, often immunosuppressive, microenvironment in which it thrives. This review synthesizes current knowledge on the molecular pathogenesis of ESCC and the pivotal biomarkers used to predict response to ICIs, aiming to chart a course toward more personalized and effective immunotherapeutic interventions.

MOLECULAR PATHOGENESIS OF ESCC

Genetic and epigenetic alterations

The pathogenesis of ESCC is driven by an accumulation of somatic genetic and epigenetic alterations that disrupt normal cellular homeostasis. Whole-exome and whole-genome sequencing studies have revealed a landscape dominated by mutations in tumour suppressor genes. Inactivation of TP53 is a near-ubiquitous event, occurring in over 70%-90% of cases, which compromises genomic integrity and facilitates malignant progression[7]. Frequent mutations and copy number alterations also affect genes involved in cell cycle regulation (CCND1, CDKN2A), squamous differentiation (NOTCH1, NOTCH3), and PI3K signalling (PIK3CA)[8,9]. Beyond genetic changes, epigenetic dysregulation plays a co-conspiratorial role. Promoter hypermethylation of key tumour suppressor genes (e.g., CDKN2A, APC) leads to their transcriptional silencing, while global hypomethylation contributes to genomic instability[10]. Furthermore, alterations in histone modification enzymes and non-coding RNAs, including microRNAs and long non-coding RNAs, create a permissive epigenetic landscape that sustains oncogenic signalling and promotes therapeutic resistance[11].

Key signalling pathways

Dysregulation of core intracellular signalling pathways forms the functional backbone of ESCC tumourigenesis. The PI3K/AKT/mTOR pathway is frequently activated through PIK3CA mutations or PTEN loss, driving uncontrolled cell proliferation, survival, and metabolic reprogramming. Recent mechanistic studies have identified SLC39A14 as a novel upstream activator of this pathway, with its knockdown significantly inhibiting ESCC cell proliferation, migration, and invasion both in vitro and in vivo, and high SLC39A14 expression correlating with poorer overall survival in ESCC patients[12]. Concurrently, aberrant activation of the Wnt/β-catenin pathway promotes epithelial-mesenchymal transition, stemness, and metastasis. Government Medical College Kollam. Recent evidence demonstrates that lactylation of H2BC9 at lysine 44 enhances Wnt7b transcription, activating downstream Wnt/β-catenin signaling and contributing to an immunosuppressive tumour microenvironment (TME) in ESCC[13]. The pro-inflammatory NF-κB pathway is another critical player, activated by cytokines within the TME or oncogenic signals, which enhances tumour proliferation, inhibits apoptosis, and fosters an immunosuppressive milieu[14]. Importantly, these pathways do not operate in isolation; extensive crosstalk between PI3K, Wnt, and NF-κB signalling creates a robust, interconnected network that drives tumour progression and complicates therapeutic targeting.

TME and immune evasion

The ESCC TME is a complex ecosystem that actively shapes tumour behaviour and treatment response. The immunosuppressive landscape of the treatment-naive ESCC TME has been characterized by high-dimensional analyses, including single-cell transcriptomic profiling[15]. This landscape is often populated by immunosuppressive components such as regulatory T cells and myeloid-derived suppressor cells. These cells secrete anti-inflammatory cytokines (e.g., interleukin-10, transforming growth factor-beta) and express co-inhibitory ligands, collectively dampening cytotoxic T lymphocyte (CTL) function[16]. A cardinal mechanism of immune evasion is the hijacking of physiological checkpoint pathways. Tumour cells and associated immune cells frequently over-express PD-L1, which engages PD-1 on tumour-infiltrating lymphocytes (TILs) to induce T cell exhaustion and apoptosis. Recent spatial proteomic profiling of ESCC patients receiving neoadjuvant PD-1 blockade has revealed that the spatial arrangement of immune cells-particularly the proximity of CD8+ T cells and B cells within tertiary lymphoid structures-is critically associated with therapeutic response, highlighting that immune cell organization within the TME is as important as checkpoint expression itself[17].

This adaptive resistance mechanism allows the tumour to circumvent immune surveillance. However, the relationship is not absolute. Significant spatial and temporal heterogeneity in PD-L1 expression, coupled with the influence of other immunosuppressive elements within the TME, can lead to a decoupling of PD-L1 status from ICI response. Consequently, a subset of patients with high PD-L1 expression [e.g., combined positive score (CPS) ≥ 10] may still exhibit primary resistance to immunotherapy, highlighting the multifactorial nature of immune evasion. Recent single-cell analyses have demonstrated that the composition of immune cell subsets within the tumor microenvironment-particularly the balance between exhausted CD8+ T cells and immunosuppressive populations-may be more critical than PD-L1 expression alone in determining ICI response. Chen et al[18] found that patients with a greater than 3.35% increase in TIM3+ CD8+ T cells following immunotherapy experienced significantly shorter progression-free survival, suggesting that dynamic changes in T-cell exhaustion markers can decouple from PD-L1 status to drive resistance. The density and spatial distribution of CD8+ T cells relative to immunosuppressive elements are increasingly recognized as critical determinants of the “immune contexture”, which predicts both natural history and response to immunotherapy[19].

ICI IN ESCC

Clinical evidence

The integration of ICIs into the ESCC treatment arsenal marks a paradigm shift. For the second-line treatment of advanced disease, nivolumab (anti-PD-1) demonstrated a superior overall survival benefit compared to chemotherapy in the global, phase III ATTRACTION-3 trial, leading to its approval[4]. Similarly, pembrolizumab (anti-PD-1) gained approval based on the KEYNOTE-181 trial, which showed a survival advantage in patients with PD-L1 CPS ≥ 10[20]. More recently, first-line combination strategies have set a new standard of care. The CHECKMATE-648 trial established nivolumab combined with chemotherapy as a first-line option for advanced ESCC[21], and the KEYNOTE-590 study showed significant survival benefits for pembrolizumab plus chemotherapy vs chemotherapy alone, particularly in patients with PD-L1 CPS ≥ 10[22]. These successes have been mirrored in regional trials, such as ORIENT-15, which confirmed the efficacy of sintilimab (anti-PD-1) plus chemotherapy[23].

Mechanisms of action and rationale for combinations

ICIs function by blocking inhibitory receptors on T cells or their ligands on tumour/immune cells, thereby releasing pre-existing anti-tumour immune responses from suppression. PD-1/PD-L1 blockade primarily reverses T cell exhaustion within the tumour bed, rejuvenating CTL-mediated killing[24]. CTLA-4 inhibition, in contrast, acts early on in the immune cycle, augmenting T cell priming and activation in lymph nodes and thereby promoting T cell diversity[25]. The rationale for combining ICIs with chemotherapy or radiotherapy extends beyond simple additive effects. These conventional modalities can induce immunogenic cell death, releasing tumour antigens and danger signals that enhance tumour immunogenicity and potentially turn “cold” tumours “hot” thereby priming the TME for more effective checkpoint blockade[26,27]. Combinations of anti-PD-1 and anti-CTLA-4 antibodies (e.g., nivolumab + ipilimumab) are also being actively explored to co-target distinct phases of the anti-tumour immune response[28].

PREDICTIVE BIOMARKERS FOR ICI RESPONSE

PD-L1 expression

PD-L1 protein expression, assessed by immunohistochemistry (IHC), remains the most widely adopted biomarker for patient selection. Scoring systems like the CPS are used in clinical trials and practice. While higher PD-L1 expression generally correlates with increased likelihood of response, it is an imperfect predictor. Significant intra- and inter-tumoural heterogeneity, lack of standardization across different assay platforms and antibodies, and dynamic changes in expression under therapeutic pressure all limit its standalone utility[29]. Notably, a proportion of PD-L1-negative patients still derive clinical benefit, underscoring the necessity for complementary biomarkers[30].

Tumour mutational burden

Tumour mutational burden (TMB), defined as the total number of somatic mutations per megabase of DNA, serves as a proxy for tumour neoantigen load. High TMB is associated with enhanced immune recognition and has been validated as a predictive biomarker for ICI response in several cancer types[31]. ESCC typically exhibits a moderate TMB, influenced by etiological factors like tobacco and alcohol use.

Emerging evidence suggests that ESCC patients with high TMB may experience improved outcomes with immunotherapy. For instance, a TMB cut-off of ≥ 10 mutations/Mb, often used in pan-cancer analyses, has shown preliminary associations with better response in some ESCC cohorts[32]. However, it is critical to note that this threshold lacks prospective validation in ESCC-specific trials. The optimal cut-off may vary depending on the sequencing panel used and the specific ICI agent. Therefore, while TMB is a promising biomarker, its routine clinical application in ESCC awaits standardization and confirmation from larger, dedicated studies.

Microsatellite instability and mismatch repair deficiency

Microsatellite instability-high (MSI-H) and mismatch repair deficiency (dMMR) status are robust pan-cancer biomarkers for ICI response, as these tumours harbour hundreds to thousands of frame-shift mutations generating highly immunogenic neoantigens[33]. While MSI-H/dMMR is rare in ESCC (estimated < 2%), its identification is crucial as it confers high sensitivity to PD-1 blockade, making screening imperative, especially in non-endemic or atypical cases[34].

Gene expression profiles and immune signatures

Transcriptomic profiling offers a holistic view of the TME. Signatures reflecting a pre-existing but suppressed adaptive immune response, such as a T cell-inflamed gene expression profile or interferon-gamma signalling, consistently correlate with better ICI outcomes across cancers[35]. In ESCC, signatures encompassing CTL markers, antigen presentation machinery, and chemokine expression are under investigation as potential predictive tools that may integrate multiple biological dimensions beyond single biomarkers[36].

Circulating biomarkers

Liquid biopsies provide a dynamic, minimally invasive window into tumour biology. Changes in circulating tumour DNA (ctDNA) levels early during ICI treatment can serve as a sensitive measure of tumour burden and molecular response. A rapid decline in ctDNA variant allele frequency after treatment initiation has been associated with favourable clinical outcomes in ESCC[37]. Additionally, profiling peripheral immune cells, such as the baseline neutrophil-to-lymphocyte ratio or changes in monotype subsets, may offer insights into systemic immune status and predict treatment efficacy or immune-related adverse events[38]. The main biomarkers are summarized in Table 1.

Table 1 Summary of predictive biomarkers for immune checkpoint inhibitor response in esophageal squamous cell carcinoma.

Biomarker category	Biological basis/what is measured	General predictive value for ICI response	Status in ESCC clinical practice	Key challenges and limitations
PD-L1 expression[29]	Protein expression of programmed death-ligand 1 on tumour and/or immune cells via IHC	Higher expression (e.g., CPS ≥ 10) correlates with increased likelihood of response	Clinically established and mandated for first-line pembrolizumab-based therapy selection. Widely used in trials and practice	Significant intra-/inter-tumoural heterogeneity; lack of standardization across assays/platforms; dynamic expression; responses occur in some PD-L1 negative patients
Tumour mutational burden[31,32]	Total number of somatic mutations per megabase of tumour DNA, a proxy for neoantigen load	High TMB (≥ 10 mut/Mb) is associated with improved response and survival across multiple cancers	Emerging biomarker. Retrospective data suggests predictive value; not yet a standard for therapy selection in ESCC	Lack of standardized cutoff for ESCC; lack of prospective ESSC-specific validation for optimal cut-off; variability across sequencing panels; requires next-generation sequencing
Microsatellite instability/mismatch repair deficiency[33,34]	Genomic hypermutability due to defective DNA mismatch repair, leading to high frame-shift neoantigen burden	MSI-H/dMMR is a highly predictive pan-cancer biomarker for robust ICI response	Standard of care to test, but very rare (< 2% prevalence). Identification is critical due to high efficacy	Extremely low prevalence in ESCC limits utility for most patients; requires IHC for MMR proteins or PCR/NGS for MSI testing
Gene expression profiles/immune signatures[35,36]	Transcriptomic quantification of immune-related genes (e.g., IFN-γ signaling, cytotoxic T cell markers, antigen presentation)	A pre-existing, “T-cell inflamed” microenvironment signature correlates with better ICI outcomes	Investigational. Not used clinically but a focus of research to better define the immune contexture beyond single markers	Lack of standardized, validated assay/platform for clinical use; complexity of data interpretation; tumour heterogeneity
Circulating biomarkers (e.g., ctDNA)[37,38]	Dynamic, non-invasive measurement of tumour-derived signals in blood (e.g., ctDNA level, NLR)	Early clearance of ctDNA predicts favorable outcome. High baseline NLR may correlate with poorer outcomes	Investigational/for monitoring. ctDNA shows promise for real-time response monitoring. Not yet validated for prospective treatment decisions	Requires validation in large prospective trials; optimal timing and thresholds not defined; detection rates are significantly lower in early-stage vs. metastatic disease, limiting its utility for screening or monitoring in non-advanced settings; confounding factors for NLR

ESCC: Oesophageal squamous cell carcinoma; ICI: Immune checkpoint inhibitors; PD-L1: Programmed death-ligand 1; IHC: Immunohistochemistry; CPS: Combined positive score; TMB: Tumour mutational burden; MSI-H/dMMR: Microsatellite instability-high/Mismatch repair deficiency; PCR/NGS: Polymerase chain reaction/next-generation sequencing; NLR: Neutrophil-to-lymphocyte ratio.

CHALLENGES AND FUTURE DIRECTIONS

Overcoming resistance

Both primary (innate) and acquired resistance limit the curative potential of ICIs. Primary resistance, where tumours fail to respond from the outset, can be driven by a state of immune ignorance or tolerance. Mechanisms are multi-factorial and include: Defects in antigen presentation (e.g., β2-microglobulin mutations), oncogenic signalling that suppresses immune infiltration (e.g., WNT/β-catenin activation), the presence of exclusionary TME features (e.g., dense fibrosis, abnormal vasculature), and the expansion of alternative immunosuppressive pathways (e.g., LAG-3, TIM-3) upon PD-1 blockade[39,40]. Overcoming resistance requires multifaceted strategies, including rational combinations of ICIs with targeted therapies against oncogenic pathways, anti-angiogenic agents to normalize vasculature, or novel agents targeting other immune checkpoints or TME components. Single-cell immunogenomic profiling has now revealed the complex regulatory networks among immune cells during neoadjuvant chemo-immunotherapy, including the critical balance between GZMK+ effector memory T cells and CXCL13+ exhausted T cells, and the role of TNFRSF13B+ memory B cells in antigen presentation-providing a high-resolution map of the cancer-immunity cycle in ESCC[41].

Personalized immunotherapy and integrating multi-omics

The future of precision immunotherapy lies in integrating multi-dimensional data. Combining genomic (mutations, TMB), transcriptomic (immune signatures), proteomic (phospho-signalling), and spatial (multiplex IHC) profiles with clinical data can create comprehensive predictive and prognostic models[42]. Artificial intelligence and machine learning algorithms are poised to decipher these complex datasets, identifying novel biomarkers, predicting patient-specific response probabilities, and suggesting optimal therapeutic combinations[43].

Novel immune targets

Beyond the PD-1 and CTLA-4 axes, next-generation immune-therapies are targeting other inhibitory receptors (e.g., LAG-3, TIM-3, TIGIT), stimulatory co-receptors (e.g., OX40, GITR), and metabolic enzymes (e.g., IDO1, ARG1) within the TME. In the context of ESCC, single-cell analyses have revealed that LAG-3 is preferentially co-expressed with PD-1 on a distinct subset of TILs associated with severe exhaustion, providing a strong rationale for dual blockade[18]. Using single-cell RNA sequencing, Chen et al[18] demonstrated that tumour-infiltrating CD8+ T cells frequently co-express multiple exhaustion markers including TIM3, PD-1, and CD39, with TIM3+ CD8+ T cells particularly enriched in the TME compared to adjacent tissues. Similarly, TIM-3 expression has been shown to be upregulated on TILs in ESCC metastatic lesions compared to primary tumours, suggesting a potential role in acquired resistance and disease progression[18]. Notably, longitudinal assessment of peripheral blood revealed that patients with a greater than 3.35% increase in TIM3+ CD8+ T cells following immunotherapy experienced significantly shorter progression-free survival (5.0 months vs 8.5 months, P = 0.024), implicating TIM-3 dynamics in treatment resistance. Preclinical models of ESCC have also implicated the TIGIT pathway in suppressing NK cell function within the TME. Early-phase clinical trials are evaluating these agents, often in combination with PD-1 blockade, with the goal of broadening efficacy and overcoming specific resistance mechanisms in ESCC[44].

Towards a multi-modal predictive framework and unanswered questions

The current paradigm of relying on single biomarkers like PD-L1 is insufficient for precision immunotherapy in ESCC. A critical unanswered question is how to optimally integrate these biomarkers. For instance, does the combination of a T-cell-inflamed gene expression profiles with high TMB identify a distinct population with superior outcomes compared to either marker alone? Future research must prioritize prospective trials designed to validate composite biomarker scores. Furthermore, the etiological heterogeneity of ESCC (e.g., tobacco/alcohol-driven vs others) likely shapes the TME and ICI response, yet this is rarely considered in predictive models. A key future direction is the integration of multi-omics data-genomics, transcriptomics, and radiomics-stratified by patient demographics and etiology, to define novel ESCC immuno-subtypes. This would allow for truly personalized approaches, moving beyond ‘one-size-fits-all’ combinations to rationally designed therapies targeting the specific biology of an individual’s tumour.

CONCLUSION

Precision immunotherapy represents a transformative frontier in ESCC management, yet its full potential remains untapped. The field must move beyond reductionist biomarkers towards integrated, dynamic models that capture the complex dialogue between a tumour’s unique genomic drivers and the spatially organised immune ecosystem it inhabits. While PD-L1, TMB, and MSI have provided an initial framework, their inherent limitations underscore the imperative for innovation. The immediate challenge is not simply to identify more biomarkers, but to rationally combine them-integrating ctDNA kinetics with baseline transcriptomic signatures and spatial pathology-to create a composite, real-time picture of the cancer-immunity cycle. By steadfastly focusing on these multi-dimensional approaches and rigorously testing them in hypothesis-driven clinical trials, we can aspire to convert the current modest responses into durable remissions for a broader population of patients battling this aggressive malignancy.