Potent molecular-targeted therapies for advanced esophageal squamous cell carcinoma

Clinicopathological features of ESCC

Clinicopathological features according to histological subtypes are shown in Table 1. ESCC is still the major histological subtype of EC.^3,7 The predominant incidence rates of the ESCC subtype are generally observed in developing countries with low socioeconomic status, and risk factors include consumption of tobacco, alcohol, and hot beverages. ESCC develops from the lining of the esophageal squamous epithelium via precursor dysplastic lesions in a multi-step process when exposed to tumorigenesis.^27,28 In contrast to the declining incidence rates of ESCC due to economic gains, dietary improvements, and smoking cessation,^1,17,29 the rapidly increasing rates of EAC have been attributed to gastroesophageal reflux disease, and obesity in highly developed countries.^3,6,7,30 EAC typically originates from Barrett’s mucosa, triggered by persistent gastroesophageal reflux, in which the normal squamous epithelium of the esophagus is replaced by a metaplastic, columnar, or glandular epithelium.^6,30,31

OPEN IN VIEWER

Table 1.

Clinicopathological features of esophageal cancer.^{3 –7,17}.

Characteristics	Squamous cell carcinoma	Adenocarcinoma
Proportion	85%	15%
Incidence trend	Declining	Increasing
Geography based on the number of estimated cases	Eastern Asia, followed by South-Central Asia, Sub-Saharan Africa, Northern/Western Europe, and South America	Eastern Asia, followed by Northern America, Northern/Western Europe, and South-Central Asia
Geography based on the incidence rates according to the histological subtypes	Asia, followed by sub-Saharan Africa, Eastern/Western Europe, and South America	Northern Europe, followed by Northern America and Oceania
Socioeconomic status	Low	High
Risk factors	Tobacco, alcohol, hot beverages/foods, and regional micronutrient deficiency	Obesity, acid or bile reflux, Barrett’s esophagus, and tobacco consumption
Race	Black > Caucasian	Caucasian > Black
Median age	Younger (median 53 years)	Older (median 63 years)
Location of primary tumor	Upper two-thirds of the esophagus	Lower third of the esophagus and esophagogastric junction
Precursor lesion	Squamous dysplasia	Barrett’s esophagus
Radiosensitivity	High	Low
Survival trend	Stable	Improving

In the Surveillance, Epidemiology, and End Results (SEER) database of 93,167 patients diagnosed with EC between 1973 and 2009, ¹⁶ the median OS of EAC improved steadily. However, ESCC remained relatively stable after the 1990s (median OS of 8 months) and was one of the independent unfavorable factors associated with the OS of EC in multivariable Cox regression analyses, with adjustment for relevant clinicopathologic factors and treatment. The median OS was only 10 months for ESCC, even in a recent SEER database of 37,723 patients diagnosed with EC between 2004 and 2015 ³² . ESCC also more easily tends toward lymph node metastasis than EAC, ¹⁷ and the status of metastatic lymph nodes is characteristic of a life-threatening phenotype of ESCC. ³³ Thus, ESCC cells have aggressive traits, which results in poor prognosis.

In a phase III CROSS trial of neoadjuvant chemoradiotherapy (CRT) plus surgery versus surgery alone for EC or esophagogastric junction, ¹⁵ neoadjuvant CRT plus surgery showed a significantly favorable OS compared with surgery alone. ¹⁵ The survival benefit of neoadjuvant CRT was more clinically relevant for patients with ESCC than for those with EAC. In addition, a pathological complete response (pCR) was significantly higher in ESCC than in EAC. ¹⁵ The high rates of pCR in ESCC versus EAC were also observed in a large cohort study of 895 EC patients who underwent neoadjuvant CRT followed by surgery. ³⁴ In contrast, the treatment efficacy of neoadjuvant chemotherapy was similar for both ESCC and EAC in a randomized OEO2 trial. ³⁵ Thus, it is evident that ESCC benefits significantly from radiotherapy. ¹⁵

Aberrant signaling pathways in ESCC

Aberrant molecular profiles in ESCC

Developments in high-throughput genomic technologies have led to a better understanding of the molecular profiles of ESCC and EAC (Table 2). ESCC and EAC exhibit almost mutually exclusive alterations of driver genes. ⁹ ESCCs share multiple SMGs exclusively with SCCs of other tissues, including the head and neck, lung, cervix, and skin, but differ from that of EAC despite the anatomical overlap.^{8 –13} Although DNA methylation is much less frequent in ESCC than in EAC, ¹³ the SCC histology-driven epigenetic patterns were also similar across organs. ¹¹ In contrast, SMGs of EAC closely resemble those of the CIN molecular subtype of stomach adenocarcinoma. ¹³ Thus, two histological subtypes of EC should be considered distinct disease entities, and SCCs that originate from developmentally similar cell lineages may have common pathogenic mechanisms, regardless of their location in different parts of the body. However, ESCC remains a heterogeneous disease with diverse genetic and molecular levels.^13,45,46 The complexity of spatial and temporal heterogeneity may be associated with ESCC aggressiveness.^23,39,47 In ESCC, several aberrant molecules were categorized according to their functions and pathways (Table 2): (i) receptor tyrosine kinase (RTK)/RAS/PI3K, (ii) cell cycle, (iii) cell adhesion, (iv) cell differentiation and proliferation, (v) chromatin modification, and (vi) oxidative stress response.^{8 –10,12,13,22,24,41,48}

OPEN IN VIEWER

Table 2.

Dysregulated genetic alteration in ESCC and EAC.^{8,10,12,13,22 –26,48 –52}

		ESCC	EAC
TCGA molecular subtypes		SCC (ESCC1-3)	CIN
Molecules	Signaling pathway	Frequency of genetic aberrations (%)
RTK/RAS/MAPK and PI3K
EGFR	RTK	6–26%	7–18%
HER2	RTK	2–6%	15–32%
HER3	RTK	1–3%	3–6%
HER4	RTK	2–13%	10–12%
IGF1R	RTK	2–5%	1–10%
FGFR1	RTK	11–31%	2–4%
MET	RTK	1–5%	4–8%
KRAS	MAPK/PI3K pathways	4–15%	13–24%
PIK3CA	PI3K pathway	9–45%	3–10%
mTOR	PI3K pathway	2–10%	6–13%
PTEN	PI3K pathway	2–11%	0–8%
VEGFA	Angiogenesis	0–3%	5–28%
Cell cycle
TP53	Regulator of cell cycle	69–94%	50–80%
CDKN2A	Regulator of cell cycle	15–76%	13–76%
CCND1	Regulator of cell cycle	33–59%	10–17%
CCNE1	Regulator of cell cycle	4–7%	10–14%
CDK4	Regulator of cell cycle	0–4%	1–3%
CDK6	Regulator of cell cycle	14–23%	7–17%
RB1	Regulator of cell cycle	2–14%	0–2%
MDM2	Regulator of cell cycle	2–25%	5–6%
Cell adhesion and proliferation
AJUBA	Regulator of cell adhesion	4–7%	3%
FAT1	Regulator of cell adhesion	5–16%	9–13%
CTNNB1	Wnt/β-catenin pathway	0–1%	2–4%
AXIN1	Wnt/β-catenin pathway	0–2%	4%
APC	Wnt/β-catenin pathway	1–2%	3–19%
MYC	Transcription factor	5–44%	12–32%
SMAD4	TGFβ pathway	1–8%	13–59%
SMAD2	TGFβ pathway	0–2%	7–30%
PTCH1	Hedgehog pathway	3–10%	0–5%
Cell differentiation
SOX2/TP63	Transcription factor	18–49%	1–12%
NOTCH1	Transcription factor	7–33%	0–3%
FBXW7	Transcription factor	4–18%	0–6%
ZNF750	Transcription factor	7–15%	0–8%
GATA4	Transcription factor	0–1%	15–19%
GATA6	Transcription factor	0–3%	14–21%
YAP1	Co-transcription factor	4–6%	5%
Chromatin modification
KDM6A	Histone modifiers	6–19%	2–4%
KMT2D	Histone modifiers	12–31%	1–13%
KMT2C	Histone modifiers	4–8%	10–25%
CREBBP	Histone modifiers	2–14%	5%
EP300	Histone modifiers	8–18%	0–5%
ARID1A	SWI/SNF	1–3%	8–18%
SMARCA4	SWI/SNF	5–10%	7–17%
PBRM1	SWI/SNF	2–5%	6%
Oxidative stress response
NFE2L2	NRF2 pathway	10–36%	0–1%
KEAP1	NRF2 pathway	2–13%	5%
CUL3	NRF2 pathway	2–5%	3%
Immune check points
PD-L1 CPS ≥ 1018,21,53,54	Immune check points	31–52%	25–48%

cBioPortal for Cancer Genomics (https://www.cbioportal.org/) was also referred.

CIN, chromosomal instability; CPS, combined positive score of PD-L1; EGFR, epidermal growth factor receptor; PTEN, phosphatase and tensine homolog; TCGA, The Cancer Genome Atlas.

RTKs trigger the activation of PI3K/v-akt murine thymoma viral oncogene homolog (Akt) and the mitogen-activated protein kinase (MAPK) signaling pathway, which play crucial roles in tumorigenesis, proliferation, survival, angiogenesis, and metastasis. EAC has a wider range of genetic alterations of RTK/RAS signaling pathways, including human epidermal growth receptor 2 (HER2), KRAS, and vascular endothelial growth factor A (VEGFA), but these are rare events in ESCC.^{13,25,26,49 –51,55} In fact, the frequency of HER2 amplifications was reported in only 2–6% of ESCCs but 15–32% of EACs.^{8,10,12,13,22 –26,48 –52,56} In 1391 patients with ECs, including 215 ESCCs and 1176 EACs, HER2 overexpression was observed in 13% of EACs and 1% of ESCCs. ⁵⁵ Among RTKs, epidermal growth factor receptor (EGFR) and fibroblast growth factor receptor 1 (FGFR1) serve as potential therapeutic targets in ESCC because of their frequent gene amplifications. The PI3K signaling pathway is also activated via genetic alterations of PIK3CA and MTOR and the loss of phosphatase and tensine homolog expression occurs in approximately 60% of ESCCs. ²²

Cyclin D1 encoded by CCND1 can form a complex with cyclin-dependent kinase 4 and 6 (CDK4/6), which may lead to abrogation of both retinoblastoma protein (Rb) and p53 function, resulting in disruption of cell cycle restriction, genome instability, and tumorigenesis. ⁵⁷ Cyclin-dependent kinase inhibitor 2A (CDKN2A) blocks the binding of cyclin D1 to CDK4/6 as a tumor suppressor gene. The cell cycle pathway is dysregulated in almost all ESCCs by mutations or methylations of TP53, CDKN2A, and RB transcriptional corepressor 1 (RB1), and amplifications of CCND1, CDK4/6, and MDM2.^{10,12,13,24,48,58}

Cell adhesion is dysregulated mainly by mutations in FAT atypical cadherin 1 (FAT1) and AJUBA in ESCC.^{9,10,12,13,22,24,41,48} FAT1 is a member of a cadherin superfamily and regulates cell–cell adhesion, proliferation, migration, and actin dynamics as a tumor suppressor gene. ⁵⁹ The function of FAT1 is frequently disrupted by FAT1 mutation, accounting for 5–16% of ESCCs.^10,22,24 In ESCC cell line models, knockdown of FAT1 suppresses cell adhesive force and accelerates cell migration and invasion through activation of the Yes1-associated transcriptional regulator (YAP1) in the Hippo signaling pathway.^60,61 AJUBA functions as a scaffold that is implicated in the assembly of multiple protein complexes to regulate cell adhesion, cytoskeletal organization, migration, mitosis, microRNA maturation, and cell differentiation.^62,63 Mutation of AJUBA also leads to YAP1 activation by inhibiting large tumor suppressor kinases 1/2 (LATS1/2) in the Hippo pathway. ⁶⁴ Catenin beta 1 (CTNNB1) encodes a β-catenin protein that acts as an essential part of the Wnt signaling pathway. The β-catenin promotes adherens junction formation by binding to E-cadherin, but it can also promote cell proliferation, differentiation, and epithelial–mesenchymal transition (EMT). ⁶⁵ Although genetic aberration of CTNNB1 is rare, altered genes in the Wnt/β-catenin pathway have been reported in 26–86% of ESCCs, including AJUBA and FAT1.^24,25,48,66 In addition, this pathway is activated by hypermethylation of its negative regulators ⁶⁶ or methylation-mediated de-repression of WNT2. ⁶⁷ MYC proto-oncogene is a transcription factor acting as a master regulator of genes involved in cell cycle progression, cell proliferation, differentiation, and apoptosis, and it can be activated by several pathways, such as RTK/RAS, Wnt/β-catenin, Notch, and transforming growth factor-β (TGF-β) signaling. ⁶⁸ MYC amplification has been reported in 43% of ESCCs, ⁸ and its overexpression detected in 61% of ESCCs. ⁶⁹ In ESCC, MYC contributes not only to cell proliferation, ⁷⁰ but also to the driver of cancer stem-like cells (CSCs) ⁷¹ and intrinsic resistance to chemoradiotherapy. ⁷²

Squamous cell differentiation is controlled by lineage-specific transcription factors (TFs), including SRY-box transcription factor 2 (SOX2), tumor protein P63 (TP63), ZNF750, and Notch family members.^{8 –13,58} TP63 and SOX2 are crucial core TFs that orchestrate gene regulatory networks for chromatin accessibility, epigenetic modifications, and gene expression patterns in ESCC. ⁷³ SOX2 and TP63 genes are localized on chromosome 3q, which are coamplified in ESCC.^13,74 TP63 and SOX2 exhibited overlapping genomic occupancy and cooperatively promoted SCC tumorigenesis and progression through activation of the MAPK and PI3K/Akt signaling pathways.^{73,75 –77} ZNF750 is a gene related to keratinocyte terminal differentiation and is frequently mutated in ESCC due to APOBEC-driven mutagenesis.^{13,22 –24} ZNF750 acts as a tumor suppressor by regulating long non-coding RNA and cell cycle activator E2F transcription factor 2 (E2F2).^{78 –80} Notch signaling regulates squamous differentiation in a normal esophagus. ⁸¹ The Notch pathway was exclusively activated by recurrent mutations of Notch pathway-related genes, including NOTCH1, NOTCH3, and F-box and WD repeat domain-containing-7 (FBXW7) in ESCC but not in EAC.^{12,24,26,48,82} The Notch pathway is also activated through histone H4 arginine methylation. ⁸³ The activated Notch pathway facilitates not only EMT, but also tumor initiation and CSC phenotype in concert with TGF-β signaling in the tumor microenvironment (TME). ⁸⁴ Importantly, these TFs cooperate to dysregulate the squamous differentiation program, such as the interaction between NOTCH1 and TP63 ⁸⁵ and between TP63 and ZNF750. ⁷⁹

Dysregulation of chromatin modification affects the development of both ESCC and EAC. Histone-modifying factors, including KMT2D, lysine methyltransferase 2C (KMT2C), lysine demethylase 6A (KDM6A), CREB binding protein (CREBBP), and E1A binding protein p300 (EP300), were frequently altered in ESCC, while alterations of Switch/sucrose nonfermentable (SWI/SNF)-encoding genes, including AT-rich interaction domain 1A (ARID1A) and SWI/SNF-related, matrix-associated, actin-dependent regulator of chromatin, subfamily a, member 4 (SMARCA4), were more common in EAC.^{12,13,22,24,41}

Oxidative stress plays diverse and important roles in carcinogenesis, metabolic reprogramming, aggressive cancer phenotypes, and drug resistance. ⁸⁶ The NFE2L2 gene encodes the NRF2 protein, which is a master transcriptional regulator that induces cytoprotective proteins upon oxidative stress. ⁸⁷ The NRF2 signaling pathway is composed of NF2EL2, KEAP1, and cullin 3 (CUL3) genes, which exhibit mutually exclusive mutations in a manner specific to squamous cell lineage.^{9,12,13,24,88} Gain-of-function mutations of the NFE2L2 gene confer resistance to stressors, including antitumor therapy, by promoting tumorigenic metabolism.^8,87 Loss-of-function mutations of the KEAP1 and CUL3 genes also activate the NRF2 pathway by suppressing ubiquitination of NRF2. ⁹ Thus, ESCCs have activity in the antioxidative pathway.

The most frequently dysregulated biochemical pathways are the cell cycle pathway, followed by chromatin remodeling, Notch, PI3K, Wnt/β-catenin, NRF2, and Hippo pathways in ESCC.^{8,10 –13,22 –24,48} Dysregulation of the TGF-β pathway, MAPK pathway, E-cadherin signaling, and the SWI/SNF chromatin remodeling complexes are rare in ESCC, but not in EAC.^13,51,55

Although cancer genes and biology are dynamically changed through tumor progression, most comprehensive molecular profiles of ESCC were analyzed using the disease without distant metastases.^{8,10,12,13,22 –26,48 –52} There is a possibility of selection bias that hampers an in-depth understanding of the associations with advanced-stage ESCC. Circulating tumor DNA sequencing is likely a strict approach to reveal the genomic landscape of metastatic ESCC with spatial heterogeneity. ⁸⁹

Biological features of ESCC

Cancer cells are extremely heterogeneous in terms of their malignant potential, drug sensitivity, and potential to metastasize, and relatively rare CSCs are at the top of the cellular hierarchy. CSCs have stemness properties, such as sphere formation, self-renewal, invasion, and cell lineage differentiation, which contribute to the driving force of tumorigenesis and metastasis.^{90 –92} CSCs are resistant to conventional chemotherapies that eliminate proliferating non-CRCs, not only via activation of drug efflux, DNA damage repair (DDR), drug-resistant proteins, and certain tumorigenic signaling pathways, but also by evading apoptosis and programmed cell death. ⁹³ Furthermore, cytotoxic chemotherapies stimulate non-CRCs to gain stemness traits. ⁹⁴ In ESCC, there is accumulating evidence that the biological function of CSCs is activated by various critical signaling pathways, such as Wnt/β-catenin, TGF-β, PI3K, Notch, Hedgehog, YAP1, and Janus kinase (JAK)/signal transducer and activator of transcription 3 (STAT3).^{48,71,95 –99} In an ESCC patient-derived xenograft model, chemoresistant tumors had high proportions of CSCs with CD90+/CD271+ in line with the overexpression of stemness-associated genes, such as SOX2, OCT4, and ATP-binding cassette superfamily G member 2 transporter (ABCG2), through Wnt/β-catenin and TGF-β signaling pathways. ⁹⁵ CSCs isolated from the ESCC cell line were resistant to genotoxicity through slow-cycling status and avoidance of apoptosis or senescence. ¹⁰⁰ In addition, a single-cell transcriptome sequencing analysis of ESCC cells with high stemness traits showed high expression of genes related to DNA replication and DDR. ¹⁰¹ Thus, the elimination of CSCs is crucial for achieving long-term therapeutic efficacy in ESCC.

EMT represents a morphogenetic process of redifferentiation of epithelial cells into mesenchymal ones, which is implicated in acquiring migratory and invasive abilities, CSC traits, metastatic potential, and chemoresistance as the hallmarks of aggressive phenotype. ¹⁰² In ESCC, EMT may be promoted by the activation of TGF-β, STAT3, Wnt/β-catenin, Notch, and Hedgehog.^{84,98,103 –106} ESCC xenograft models showed that TGF-β drove Notch-mediated EMT via activation of the transcription factor Zinc finger E-Box binding homeobox 1 (ZEB1) and concurrently generated CSCs. ⁸⁴ Tumor necrosis factor-associated apoptosis-inducing ligand (TRAIL) promoted EMT via STAT3-dependent expression of programmed death-ligand 1 (PD-L1), an inhibitory immune checkpoint molecule. ¹⁰⁶ Epigenetic alterations, including regulatory non-coding RNAs (ncRNAs), interact with numerous TFs and act as crucial mechanisms of EMT. The ncRNAs have two types of long non-coding RNAs (lncRNAs) and microRNAs (miRNAs), and these execute a broad repertoire of cellular processes. Emerging studies demonstrate the effects of ncRNAs on EMT, including HOTTIP/miR-30b/Snail axis, ¹⁰⁷ lncRNA-TTN-AS1/miRNA-133b/Snail axis, ¹⁰⁸ and miRNA-31/Hippo signaling axis, ¹⁰⁹ in ESCC.

The TME is composed of diverse cell types, including cancer-associated fibroblasts (CAFs), immune cells, vascular endothelial cells, and stem cells, which play a central role in proliferation, invasion, metastasis, angiogenesis, and chemoresistance in EC. ¹⁰³ As an essential component of the TME, CAFs contribute to EMT as well as angiogenesis through TGF-β and STAT3 pathways in ESCC.^98,105 The interaction between ESCCs and CAFs dysregulates the expression of multiple genes encoding growth factors, cytokines, chemokines, matricellular proteins, EMT-related molecules, and components of inflammatory signaling pathways. ¹¹⁰ Furthermore, TME creates a CSC niche that is made up of stromal cells, blood vessels, soluble factors, and extracellular matrix proteins, which not only maintains stemness properties, but also creates immune evasion.^111,112

Collectively, the aggressive phenotypes of ESCC may be partially explained by the biological features of EMT and CSC phenotypes, establishing a preclinical rationale to target molecules that play a pivotal role in EMT and CSCs in terms of antitumor efficacy and immunity.

Cytotoxic chemotherapy

The impact of cytotoxic chemotherapy on ESCC has been evaluated in clinical trials (Table 3). Platinum-based chemotherapy remains the backbone of treatment in the first-line setting. ² However, the treatment efficacy of platinum-based doublet chemotherapy is generally modest, with reported ORR of 15–56% and median progression-free survival (PFS) of 4–7 months. The cytotoxic effect of platinum compounds has been attributed to the formation of platinum DNA adducts. Excision repair cross-complementing 1 (ERCC1) is a key rate-limiting enzyme in the nucleotide excision repair system, which can repair these adducts and is consequently associated with platinum resistance. ¹¹³ The expression of ERCC1 was higher in ESCC than in EAC. ⁵⁵ A pooled analysis demonstrated the status of ERCC1 expression as a valuable predictive biomarker for platinum-based chemotherapy in ESCC, in which the median ORR was 77.5% and 39.3% for patients with low and high expression of ERCC1, respectively. ¹¹⁴

OPEN IN VIEWER

Table 3.

Clinical trials of cytotoxic agents in advanced esophageal squamous cell carcinoma.

Regimen	Line	Phase	Histology	No. of pts (No. of ESCC)	ORR (%)	PFS (months)	OS (months)	Ref.
CDDP plus 5-FU	≥1	II	SCC	39	36	–	9.5	Iizuka et al. 115
CDDP plus 5-FU versus CDDP	1	II	SCC	88	35 versus 19	–	8.3 versus 7.0	Bleiberg et al. 116
PTX plus CDDP	1	II	SCC	39	48.6	7	13	Zhang et al. 117
PTX plus CDDP	1	II	SCC	46	56.5	5.6	17.0	Huang et al. 118
PTX plus CBDCA	1	II	SCC/AC	35 (13)	43	3.4*	9	El-Rayes et al. 119
DTX plus CDDP	1	II	SCC	39	33	5.0	8.3	Kim et al. 120
GEM plus CDDP	1	II	SCC	38	42.1	4.1	10	Huang et al. 121
CPT-11 plus CDDP	1	II	SCC	32	31.3	4.4	9.6	Lee et al. 122
CPT-11 plus CDDP	1	II	SCC	27	30.0	4.5	8.8	Kim et al. 123
PTX plus L-OHP	1	I/II	SCC/AC	26 (19)	15	4.5	12.3	Kaechele et al. 124
5-FU plus LV plus L-OHP (FOLFOX)	1	II	SCC	56	23.2	4.4	7.7	Wang et al. 125
CPT-11 plus 5FU (AIO)	1	II	SCC/AC	24 (11)	33	6.6**	13.6 (10.0 for SCC)	Wolff et al. 126
PTX plus CDDP plus 5-FU	1	II	SCC/AC	61 (31)	48	–	10.8	Ilson et al. 127
DTX plus CDDP plus 5-FU	1	I/II	SCC/OTH	55	62	5.8	11.1	Hironaka et al. 128
DTX plus CDDP plus 5-FU	1	I/II	SCC	48	62.5	6	13	Ojima et al. 129
DTX	≥1	II	SCC/AC/OTH	52 (46)	20	–	8.1	Muro et al. 130
PTX	≥2	II	SCC/AC	53 (52)	44.2	3.9**	10.4	Kato et al. 131
CPT-11	≥1	II	SCC/AC	13 (10)	22.2	3.8**	6.1	Mühr-Wilkenshoff et al. 132
CPT-11	2	II	SCC/AC	14 (7)	15.4	2	5	Burkart et al. 133

*

Median time to progression (TTP) in 27 patients with advanced disease.

**

Median TTP.

5-FU, 5-fluorouracil; AC, adenocarcinoma; CBDCA, carboplatin; CDDP, cisplatin; CPT-11, irinotecan; DTX, docetaxel; L-OHP, oxaliplatin; LV, leucovorin; ORR, overall response rate; OS, overall survival; OTH, others; PFS, progression-free survival; Pts, patients; PTX, paclitaxel; Ref, reference; SCC, squamous cell carcinoma.

Single-agent chemotherapy with taxanes or irinotecan is an option for a second- or later-line therapy. ^{2,130 –133} A retrospective study using a propensity score model showed the clinical benefit of OS in a second-line therapy compared with active symptom control alone for patients with ESCC, with a hazard ratio (HR) of 0.40 [95% confidence interval (CI), 0.24–0.69]. ¹³⁴

Immunotherapy

The immune system plays a key role in eliminating cancer cells. However, antitumor immune escape is often promoted by inhibitory immune checkpoint molecules, such as PD-1 and its ligand PD-L1, during the cancer-immunity cycle process.^{135 –137} PD-1 receptors on T cells bind to PD-L1, and the activated PD-1/PD-L1 signaling axis induces immune tolerance in the TME by disrupting the functioning of both cytotoxic and effector T cells.^136,137 ESCC exhibits higher expression of PD-L1 compared with EAC^{18,21,53,55,138} (Table 2), and high PD-L1 expression shows poor prognostic trends in a meta-analysis of 1350 ESCC patients. ¹³⁹ The treatment of ESCC has changed dramatically following several landmark trials, which have proven the benefit of immunotherapy (Table 4).

OPEN IN VIEWER

Table 4.

Clinical trials of immune checkpoint inhibitors in advanced esophageal squamous cell carcinoma.

Trials	Line	Phase	Target	Inhibitor	Agent	Treatment	No. of pts	Histology	Proportion of ESCC (%)	ORR (%)	PFS (months)	OS (months)	HR (95% CI) of PFS in ESCC	HR (95% CI) of OS in ESCC	Information	Reference
KEYNOTE-028	≥2	Ib	PD-1	Ab	Pembrolizumab (Pembro)	mono	23	SCC/AC	78	30 (28 for SCC)	1.8	7.0	–	–	Inclusion criteria: CPS ≥ 1%	Doi et al. 140
KEYNOTE-180	≥3	II				mono	121	SCC/AC	52	9.9 (14.3 for SCC)	2.0 (2.1 for SCC)	5.8 (6.8 for SCC)	–	–	ORR: 13.8% and 6.3% for CPS ≥ 10% or <10%, respectively. PFS: both 2.0 months in CPS ≥ 10% or <10%	Shah et al. 141
KEYNOTE-181	2	III				Pembro versus Chemo §	628	SCC/AC	64	13.1 versus 6.7 for ITT (16.7 versus 7.4 for SCC)	2.1 versus 3.4 for ITT (2.2 versus 3.1 for SCC)	7.1 versus 7.1 for ITT (8.2 versus 7.1 for SCC)	0.92 (0.75–1.13)	0.77 (0.63–0.96)	Primary endpoints: OS in pts with PD-L1 CPS ≥ 10, with ESCC, and in all pts	Kojima et al. 18
KEYNOTE-590	1	III				CF = pembro	749	SCC/AC	73	45.0 versus 29.3 for ITT	6.3 versus 5.8 for ITT (6.3 versus 5.8 for SCC)	12.4 versus 9.8 for ITT (12.6 versus 9.8 for SCC, and 13·9 versus 8·8 for SCC with CPS ≥ 10)	0.65 (0.54–0.78)	0.72 (0.60–0.88) for SCC and 0.57 (0.43–0.75) for SCC with CPS ≥ 10%	Primary endpoints: OS in pts with ESCC and PD-L1 CPS ≥ 10, and OS and PFS in pts with ESCC, PD-L1 CPS ≥ 10, and in all pts	Sun et al. 53
ATTRACTION-1	≥2	II			Nivolumab (Nivo)	Mono	65	SCC	100	17	1.5	10.8	-	-		Kudo et al. 142
ATTRACTION-3	2	III				Nivo versus Chemo †	419	SCC	100	19 versus 22	1.7 versus 3.4	10.9 versus 8.4	1.08 (0.87–1.34)	0.77 (0.62–0.96)		Kato et al. 19
CheckMate 648	1	III				CF=Nivo	970	SCC	100	47 versus 27 for ITT (53 versus 20 for TPS ≥ 1%)	5.8 versus 5.6 for ITT (6.9 versus 4.4 for TPS > 1%)	13.2 versus 10.7 for ITT (15.4 versus 9.1 for TPS > 1%)	0.81 (0.64–1.04) for ITT and 0.65 (0.46–0.92) for TPS ≥ 1%	0.74 (0.58–0.96) for ITT and 0.54 (0.37–0.80) for TPS ≥ 1%	Primary endpoints: OS and PFS in TPS PD-L1 ≥ 1%	Doki et al. 143
			CTLA-4		Ipilimumab (Ipi)	Ipi + Nivo versus CF		SCC	100	28 versus 27 for ITT (35 versus 20 for TPS ≥ 1%)	2.9 versus 5.6 for ITT (4.0 versus 4.4 for TPS > 1%)	12.8 versus 10.7 for ITT (13.7 versus 9.1 for TPS > 1%)	1.26 (1.04–1.52) for ITT and 1.02 (0.73–1.43) for TPS ≥ 1%	0.78 (0.62–0.98) for ITT and 0.64 (0.46–0.90) for TPS ≥ 1%
	1	II	PD-1		Camrelizumab (Camre)	Camre + Apa + Chemo¶	30	SCC	100	80	6.85	19.43	–	–	–	Zhang et al. 144
ESCORT	2	III				Camre versus Chemo ‡	457	SCC	100	20.2 versus 6.4	1.9 versus 1.9	8.3 versus 6.2	0.69 (0.56–0.86)	0.71 (0.57–0.87)	Primary endpoint: OS for ITT	Huang et al. 20
ESCORT-1st	1	III				TP ± Camre	596	SCC	100	72.1 versus 62.1	6.9 versus 5.6	15.3 versus 12.0	0.56 (0.46–0.68)	0.70 (0.56–0.88)	Primary endpoints: OS and PFS for ITT	Luo et al. 145
NCT03469557	1	II			Tislelizumab (Tisle)	CF + Tisle	15	SCC	100	46.7	10.4	N/R	–	–		Xu et al. 146
RATIONALE 306	1	III				Chemo†† ± Tisle	649	SCC	100	63.5 versus 42.4	7.3 versus 5.6	17.2 versus 10.6	0.62 (0.52–0.75)	0.66 (0.54–0.80)	Primary endpoint: OS for ITT	Yoon et al. 147
RATIONALE 302	2	III				Tisle versus Chemo §	512	SCC	100	20.3 versus 9.8	1.6 versus 2.1	8.6 versus 6.3 for ITT (10.3 versus 6.8 for CPS ≥ 10%)	0.83 (0.67–1.01)	0.70 (0.57–0.85) for ITT and 0.54 (0.36–0.79) for CPS ≥ 10%	Primary endpoint: OS for ITT	Shen et al. 21
JUPITER-06	1	III			Toripalimab (Tori)	TP ± Tori	514	SCC	100	69.3 versus 52.1	5.7 versus 5.5	17.0 versus 11.0	0.58 (0.46–0.74)	0.58 (0.46–0.74)	Primary endpoints: OS and PFS for ITT	Wang et al. 54
ORIENT-2	2	II			Sintilimab (Sin)	Sin versus chemo║	190	SCC	100	12.6 versus 6.3	1.6 versus 2.9	7.2 versus 6.2	1.00	0.70	Primary endpoint: OS for ITT	Xu et al. 148
ORIENT-15	1	III				Chemo* ± Sin	659	SCC	100	75.5 versus 56.9 for ITT (78.7 versus 57.5 for CPS ≥ 10%)	7.2 versus 5.7 for ITT (8.3 versus 6.4 for CPS ≥ 10%)	16.7 versus 12.5 for ITT (17.2 versus 13.6 for CPS ≥ 10%)	0.56 (0.46–0.68) for ITT and 0.58 (0.45–0.75) for CPS ≥ 10%	0.63 (0.51–0.78) for ITT and 0.64 (0.48–0.85) for CPS ≥ 10%	Primary endpoints: OS in the pts with CPS ≥ 10 and in all pts	Shen et al. 149
KN046-204	1	II	PD-L1/CTLA-4	BiAb	KN046	KN046 ± TP	15	SCC	100	58.3	–	–	–	–		Xu et al. 150

A search for clinical trials included in Table 4 was conducted using the PubMed databases (https://pubmed.ncbi.nlm.nih.gov/). We used the search terms ‘clinical trial’ and ‘medication’ in combination with the term ‘esophageal cancer’. The literature screening process used the following inclusion criteria: (i) prospective phase I–III trials and (ii) studies that evaluated the therapeutic efficacy of medication. The exclusion criteria were as follows: (i) studies for esophageal adenocarcinoma, (ii) studies that evaluated the therapeutic efficacy of perioperative medication for patients with early-staged ESCC, and (iii) studies with results published in languages other than English. Several results of trials, which have been presented at a conference or meeting of cancer specialists but have not yet been published in specialist medical journals, were examined using the European Society for Medical Oncology (ESMO, https://www.esmo.org/) and the American Society of Clinical Oncology (ASCO, https://conferences.asco.org/) website.

5-FU, 5-fluorouracil; 95% CI, 95% confidence interval; Ab, antibody; AC, adenocarcinoma; BiAb, bispecific antibody; CDDP, cisplatin; CF, CDDP plus 5-FU; Chemo§, investigator’s choice (PTX, DTX, or CPT-11); Chemo†, investigator’s choice (PTX or DTX); Chemo††, investigator’s choice (platinum plus fluoropyrimidine or platinum plus PTX); Chemo¶, liposomal PTX and nedaplatin; Chemo‡, investigator’s choice (DTX or CPT-11); Chemo║, investigator’s choice (PTX or CPT-11); Chemo*, TP or CF; CPS, combined positive score; DTX, docetaxel; ESCC, esophageal squamous cell carcinoma; HR, hazard ratio; ITT, intent-to-treat population; mono, monotherapy; ORR, overall response rate; OS, overall survival; PFS, progression-free survival; Pts, patients; PTX, paclitaxel; TPS, tumor proportion score; TP, PTX plus CDDP.

In the second-line setting, monotherapy with anti-PD-1 monoclonal antibody (mAb) demonstrated a survival benefit compared with chemotherapy in phase III ATTRACTION-03, ¹⁹ KEYNOTE-181, ¹⁸ ESCORT, ²⁰ and RATIONALE-302 ²¹ trials. Of note, the survival curves crossed and subsequently sustained separation of the curves in favor of anti-PD-1 mAb monotherapy with duration of response, ranging from 6.9 to 8.5 months.^{18 –20} The ORR of anti-PD-1 mAb monotherapy was approximately only 20%.^{18 –21} Collectively, although a subset of patients achieved a durable response, there is still a large proportion of patients who do not benefit from anti-PD-1 mAb treatment. Chemotherapy can promote immune responses by increasing the immunogenicity of cancer cells or inhibiting immunosuppressive circuitries.^151,152 Recently, the clinical benefits of anti-PD-1 mAb in combination with first-line chemotherapy for ESCC patients were demonstrated in phase III KEYNOTE-590, ⁵³ CheckMate 648, ¹⁴³ ESCORT-lst, ¹⁴⁵ JUPITER-06, ⁵⁴ ORIENT-15, ¹⁴⁹ and RATIONALE-306 ¹⁴⁷ trials. The survival curves were separated from the beginning of treatment in favor of anti-PD-1 mAb plus chemotherapy compared with chemotherapy alone, and the combined therapy had a manageable safety profile. Thus, the combination of anti-PD-1 mAb and chemotherapy represents a new standard first-line treatment for ESCC patients. Although the percentage of patients who had an objective response was higher with anti-PD-1 mAb plus chemotherapy than with chemotherapy alone, the absolute difference in ORR was 10.0–21.1% in an intent-to-treat population. Thus, there is an urgent need for potential predictive biomarkers to select the most suitable patients. The status of microsatellite instability-high (MSI-H) is a robust predictive biomarker for treatment with ICIs, but MSI-H is rare in ESCC.^55,153 PD-L1 expression status is among the candidate biomarkers of the response to treatment with PD-1 mAb for ESCC patients.^{53,143,145,147,154} However, the expression of PD-L1 on the cell surface was highly heterogeneous, ²³ and much more complicated factors arising from cancer cells and the tumor immune microenvironment are likely to contribute to the clinical benefit of ICI treatment.^23,155,156 Therefore, further development of predictive biomarkers and more effective immunotherapy are required to improve prognostic outcomes.

Cytotoxic T-lymphocyte-associated antigen-4 (CTLA-4) acts as a negative regulator of the initial priming of T cells in the early stage of the immune response process, whereas PD-1/PD-L1 acts in later stages by turning off antitumor T-cell responses. ¹⁵⁷ Therefore, dual inhibition of PD-1/PD-L1 and CTLA-4 synergistically promotes an antitumor immune response by blocking complementary mechanisms. In a phase III CheckMate 648 trial, the combination of nivolumab plus an anti-CTLA-4 mAb ipilimumab versus chemotherapy alone was also evaluated. ¹⁴³ Treatment with nivolumab plus ipilimumab resulted in a longer OS (median, 12.7 versus 10.7 months) and duration of response (median, 11.1 versus 7.1 months) than chemotherapy alone in ESCC patients, suggesting that the dual inhibitor is also a new standard first-line regimen. However, another primary endpoint PFS was not met, with a crossover of the survival curves. The percentage of patients assessed as having progressive disease in treatment with nivolumab plus ipilimumab was twice as high as the percentage in chemotherapy alone (30 versus 15%), regardless of the high ORR (35 versus 20%). In addition, treatment-related serious adverse events of any grade were more common with nivolumab plus ipilimumab (32%) than with chemotherapy alone (16%), despite chemotherapy-free treatment. Special attention should be paid to treatment with nivolumab plus ipilimumab because of the double-edged sword that has benefits with durable response and risk of rapid progression and serious adverse events.

EGFR-targeted therapy

EGFR overexpression and amplification are reported in 18–89% and 6–26% of ESCCs, respectively.^{8,10,12,13,22 –26,48,52,55,160,175 –177} A meta-analysis including 802 patients with ESCC showed that EGFR overexpression was associated with lymph node metastatic status and poor OS. ¹⁷⁸ In addition, there was a trend toward a worse prognosis for ESCC patients with a high EGFR gene copy number than for those with a low copy number. ¹⁷⁹

Based on the frequent EGFR overexpression of gene amplification in ESCC, several phase II/III trials have been conducted to evaluate the treatment efficacy of EGFR inhibitors, including mAb and tyrosine kinase inhibitors (TKIs). Unfortunately, these trials have shown limited efficacy (Table 5). A phase III COG trial of EGFR TKI gefitinib versus placebo in previously treated patients with EC, including 106 patients with ESCC, was conducted. ¹⁶⁴ Although gefitinib showed the limited benefits of PFS (1.6 versus 1.2 months), the primary endpoint, OS, was not met. Another phase III POWER trial of first-line chemotherapy with or without EGFR mAb panitumumab in ESCC patients was terminated early because of potential safety concerns on interim analysis. ¹⁶⁷ There was no improvement in OS in addition to panitumumab, with more frequent treatment-related deaths. Similar findings were found in a randomized phase II AIO trial of the addition of EGFR mAb cetuximab to first-line chemotherapy in ESCC patients, which resulted in no significant improvement in terms of OS, PFS, and ORR. ¹⁶⁶ Of note, only a subset of patients had marked clinical responses.^{160,161,163,164,168,180} The identification of predictive biomarkers is vital for improving the clinical outcomes of patients treated with EGFR inhibitors. In a phase Ib trial of EGFR TKI larotinib in pretreated advanced ESCC with EGFR overexpression or amplification, larotinib demonstrated promising antitumor activity, with an ORR of 20.0% and median PFS of 3.4 months at a dose of 350 mg 149. A phase III trial of larotinib versus chemotherapy is ongoing for ESCC with EGFR overexpression (NCT04415853; Table 6).

Although miRNA-338-5p inhibits the signaling pathways of MET and EGFR by directly targeting the 3′ untranslated regions of these genes, ¹⁸¹ its expression is downregulated in the majority of ESCC cases. ¹⁸² Amivantamab is a bispecific antibody (BiAb) targeting EGFR and MET, with immune cell-directing activity. ¹⁸³ The U.S. Food and Drug Administration granted accelerated approval to amivantamab for patients with non-small-cell lung cancer harboring the EGFR exon 20 insertion mutation, based on the results of the CHRYSALIS trial. ¹⁸⁴ The activity of amivantamab is assessed in ESCC with overexpression or amplification of EGFR and MET (NCT04945733 and NCT05117931) (Table 6). The EGFR signaling pathway is activated via hetero-dimerization with other HER families as well as EGFR homo-dimerization. ¹⁸⁵ A phase II trial of an irreversible pan-HER TKI dacomitinib was conducted to assess whether combined inhibition of all HER family kinases has more potent antitumor activity than EGFR inhibition alone in ESCC patients. ¹⁶⁸ The ORR was 12.5% (6 of 48 patients), with a median PFS of 3.3 months (Table 5). There are ongoing early trials of EGFR/HER3, a bispecific antibody (BiAb) SI-B001 (NCT05022654), and a combined blockade of EGFR using TKI afatinib and mAb JMT101 (NCT05164848) (Table 6). In addition, aberrations of cell cycle-related genes coexist in the majority of EGFR-amplified ESCC, and the blockade of CDK4/6 may enhance the efficacy of the EGFR inhibitor. ¹⁸⁶

FGFR-targeted therapy

The FGFR family consists of four highly conserved receptors (FGFR1, FGFR2, FGFR3, and FGFR4). The FGFR pathway is cancer specifically dysregulated by its overexpression and genetic alterations, driving cancer development and progression.^{187 –189} FGFR genomic alterations were detected in 7% of the 4853 solid tumors, of which gene amplifications were the most common alteration, with 66% of the cases having FGFR aberrations. ¹⁸⁸ In ESCC, FGFR1 has the most frequent genetic alteration among FGFR family members, and FGFR1 gene amplification was detected in 14.3%. ¹⁸⁸ A meta-analysis including 2326 ESCC patients showed that FGFR1 amplification was significantly correlated with lymph node metastasis, poorer differentiation, and unfavorable OS with an HR of 1.50 (95% CI: 1.25–1.81). ¹⁹⁰ In addition, FGFR ligands FGF3, FGF4, and FGF19 genes are more frequently amplified in ESCC. ⁵⁵ Futibatinib (TAS-120) is an orally bioavailable, highly selective, and irreversible FGFR1–4 inhibitor. ¹⁹¹ A phase I/II trial of futibatinib is ongoing in patients with advanced solid tumors harboring FGF/FGFR aberrations (NCT02052778; Table 6).

PI3K-targeted therapy

PI3K/Akt signaling is one of the most frequently dysregulated biochemical pathways in ESCC.^{8,10 –13,22 –24,48} Class I PI3Ks consist of a p110 catalytic subunit and a p85 regulatory subunit, and they are divided into four catalytic isoforms, p110α, p110β, p110γ, and p110δ, which are encoded by PIK3CA, PIK3CB, PIK3CG, and PIK3CD, respectively. ¹⁹² BKM120 is an oral pan-class I PI3K inhibitor targeting all four catalytic isoforms of class I PI3K. In a phase II EPOC1303 trial of BKM120 in 41 ESCC patients, the primary endpoint was met, with a disease control rate (DCR) of 51.2% ¹⁷⁰ (Table 5). Although pan-class I PI3K inhibitors have the advantage of on-target inhibition of each of the class I PI3K catalytic isoforms, they have the risk of broader off-target effects, resulting in an unfavorable toxicity profile and limited clinical utility. ¹⁹³ PI3K isoform-specific inhibitors may offer a superior therapeutic window and less toxicity than pan-class I PI3K inhibitors. ¹⁹² Double PIK3CA mutations in cis may be a predictive biomarker for selecting patients who would benefit from PI3Kα inhibitors. ¹⁹⁴

There are alternative strategies for the blockade of the PI3K/Akt signaling pathway. The insulin-like growth factor-II (IGF-II)/IGF-I Receptor (IGF-IR) signaling pathway plays an important role in ESCC progression by activating Akt, and treatment with anti-IGF-IR mAb cixutumumab resulted in the enhanced antitumor effect and chemosensitivity in a preclinical study. ¹⁹⁵ However, in a phase II trial of second-line paclitaxel plus cixutumumab in patients with metastatic EC (NCT01142388), the addition of cixutumumab did not improve clinical outcomes ¹⁶⁹ (Table 5). Focal adhesion kinase (FAK) is a cytoplasmic tyrosine kinase that activates the PI3K/Akt pathway as a mediator of cell signaling downstream of growth factor and cytokine receptors. ¹⁹⁶ In a preclinical study, a specific FAK inhibitor defactinib dose and time dependently induced the dissociation of PI3K, which led to the blockade of Akt signaling and the expression of several oncogenes, such as SOX2, MYC, EGFR, or MET. ¹⁹⁷ Furthermore, treatment with defactinib suppressed tumor growth and metastatic ability in ESCC xenograft models. AXL is a RTK belonging to the Tyro3/Axl/Mer (TAM) family, ¹⁹⁸ and it is frequently overexpressed in ESCC. ¹⁹⁹ AXL dimerizes with EGFR to activate the protein kinase C (PKC), leading to the activation of mTOR in an PI3K-independent manner. ²⁰⁰ Consequently, AXL overexpression mediates resistance to PI3K inhibitors by activating the EGFR/PKC/mTOR axis in ESCC, and the blockade of AXL may revert resistance to PI3Ka inhibitors. Because several other signaling pathways connect with the PI3K/Akt signaling network, ¹⁹³ the major targets downstream of the PI3K pathway may be compensated for via bypass or crosstalk signaling when PI3K is blocked.

Angiogenesis-targeted therapy

Angiogenesis contributes to the progression of tumorigenesis and metastasis by providing nutrition, growth factors, and an oxygen supply. ²⁰¹ In addition, tumor vasculature suppresses the delivery of chemotherapy agents to the tumor site due to its leaky properties. ²⁰² VEGF and its receptor VEGFR are key mediators responsible for angiogenesis, and therapies directed against VEGF/VEGFR are the focus of major molecular-targeted research in solid tumors. In ESCC, high levels of VEGF expression were associated with poor survival in a meta-analysis that included 2043 patients with ESCC, ²⁰³ and several trials of anti-VEGFR inhibitors were conducted.

Apatinib is a novel VEGFR2 tyrosine kinase inhibitor, and treatment with apatinib resulted in inhibited proliferation and induced apoptosis in ESCC cell line models. ²⁰⁴ Two phase II trials showed similar clinical activity for apatinib, with ORR of 7.5–7.7%, DCR of 61.5–65.0%, and PFS of 3.8 months–4.6 months, as a second- or later-line therapy in ESCC patients^172,173 (Table 5). Anlotinib is an oral small-molecule TKI targeting VEGFR 1–3, FGFR 1–4, platelet-derived growth factor receptor (PDGFR) α and β, RET, and c-KIT. ²⁰⁵ In a randomized, double-blind, placebo-controlled phase II trial to evaluate the antitumor activity and safety of anlotinib as a second- or later-line therapy in ESCC patients, the PFS as a primary endpoint was significantly improved in anlotinib compared with placebo, although there was no significant difference in terms of ORR (7.3 versus 3.6%) ¹⁷¹ (Table 5). Currently, several early phase trials of agents targeting angiogenesis in combination with chemotherapy are ongoing: (i) TKIs, including apatinib (NCT02645864) and anlotinib (NCT04063683 and NCT03387904), and (ii) mAb directed against VEGFR2 (ramucirumab, NCT03762564) (Table 6).

CDK4/6-targeted therapy

The cyclin D1-CDK4/6 pathway is triggered by amplifications of CCND1 and CDK4/6, and losses of CDKN2A and RB1 in ESCC,^{10,12,13,24,48,58} suggesting a rational target for ESCC. CDK4/6 has proven to be an effective target in hormone receptor-positive advanced breast cancer. ²⁰⁶ A phase II trial of palbociclib, a SMI of CDK4/6, was conducted with patients with EC and gastric cancer, among which one of five ESCC patients had stable disease lasting for more than 4 months ¹⁷⁴ (Table 5). Dalpiciclib (SHR6390) is an orally bioavailable, small-molecule CDK4/6 inhibitor, and treatment with dalpiciclib suppressed cell proliferation and tumor growth by inducing cell cycle arrest at the G1 phase in ESCC cell lines and xenograft models. ²⁰⁷ An exploratory clinical trial of dalpiciclib is ongoing in ESCC patients (NCT04866381; Table 6).

Because CDK4/6 inhibitor alone may limit clinical efficacy, the combination with other therapeutic strategies can be of potential benefit. Therefore, the identification of a therapeutic vulnerability in ESCC harboring cell cycle dysregulation is required. Aberration of the cyclin D1-CDK4 pathway also leads to reprogramming of cellular metabolism, which contributes to glutamine addiction and mitochondrial dysfunction. ²⁰⁸ Dual blockade of glutaminase and mitochondrial respiration may effectively reduce tumor burden by interfering with cell proliferation and survival.

NRF2-targeted therapy

Transcription factor NRF2, encoded by the NFE2L2 gene, regulates cellular antioxidant, detoxification, inflammation, and metabolic pathways by modulating the constitutive and inducible expression of over 200 genes that contain antioxidant response elements in their regulatory regions.^{209 –211} NRF2 also induces upregulation of multidrug resistance-associated proteins, which are a major drug efflux pump. ²¹² Consequently, the NRF2 signaling pathway confers cancer cells with a proliferative advantage, as well as resistance to antitumor drugs and radiotherapies. In ESCC, the NRF2 pathway is aberrantly activated by genetic and epigenetic alterations, including gain-of-function mutations of the NFE2L2 gene and loss-of-function mutations of negative regulators KEAP1 and CUL3 genes.^{9,12,13,24,88} ESCC cells with mutations of the NFE2L2 gene showed increased cell proliferation and resistance to 5-fluorouracil and γ-irradiation, ⁸⁷ and blockage of NRF2 repressed the migration and invasion of ESCC cells in the hypoxic microenvironment. ²¹³ NRF2 overexpression was associated with lymph node metastases, unfavorable response to neoadjuvant chemoradiation, and poor postoperative outcomes in ESCC patients.^87,214 Thus, the NRF2 signaling pathway may contribute to the hallmarks of cancer through cellular metabolic processes leading to ‘NRF2 addiction’ in ESCC, and have potential as a relevant therapeutic target for NRF2-addicted ESCC patients.

Several strategies have been proposed to target the NRF2 signaling pathway. ²¹⁵ Potent inhibitors of NRF2, including brusatol, ²¹⁶ halofuginone, ²¹⁷ AEM1, ²¹⁸ ML385, ²¹⁹ and trigonelline, ²²⁰ have been identified by high-throughput screening. ²²¹ Many flavonoids, such as luteolin, chrysin, and apigenin, have been reported as regulators of NRF2 mRNA expression in cancer cells. ²²² NRF2 downstream pathways can also be targeted. Nuclear receptor subfamily 0 group B member 1 (NR0B1) was identified as a druggable vulnerability by chemical proteomics using NRF2-addicted lung cancer cells, and NR0B1 protein complexes were disrupted by novel covalent SMIs, such as BPK-26 and BPK-29. ²²³ As an important safety concern, NRF2 inhibitors may cause ‘off-target’ side effects on cancer-bearing hosts because of the essential role of NRF2 in cytoprotection.

Metabolic rewiring evokes cellular mechanisms that reduce therapeutic mightiness. Cancer cells avidly consume glutamine to fuel biosynthesis and proliferation, and glutaminases regulate glutathione and oxidative stress, which are activated through NRF2-driven metabolic adaptations.^211,224 NRF2-addicted cancer cells depend on glutamine metabolism and show vulnerability to glutaminase inhibition via depletion of glutathione and increased replication stress.^{225 –228} Telaglenastat (CB-839) is a first-in-class, oral, highly selective glutaminase inhibitor that has demonstrated preclinical activity in NRF2-addicted models^{225 –228} and antitumor activity in patients with previously treated renal cell carcinoma in the phase II ENTRATA trial. ²²⁹ In addition, a combined therapy of glutaminase inhibitor and the poly (ADP-ribose) polymerase (PARP) inhibitor may be an effective synthetic lethal approach. ²²⁶ Competition between cancer cells and immune cells for glutamine as a critical nutrient creates a metabolic checkpoint that induces immunosuppressive status in TME. Telaglenastat may enhance the antitumor effect of ICIs through the activation of immune cell function via blocking tumor glutamine consumption. ²³⁰

YAP1-targeted therapy

The Hippo pathway controls organ development and tissue homeostasis. ²³¹ YAP1 is a downstream transcription coactivator of the Hippo pathway and forms transcription complexes with TFs, including the transcriptional enhancer activator domain (TEAD), activator protein 1 (AP-1), bromodomain-containing protein 4 (BRD4), and E2 factor (E2F).^{231 –234} YAP1 signaling is abnormally dysregulated in various tumor types, which is implicated in tumorigenesis, EMT, and CSC traits, a central bypass mechanism that results in resistance to EGFR and RAS/MAPK inhibitors, and plays a protective role against chemotherapy-induced apoptosis.^{231 –239}

In ESCC, YAP1 is constitutively activated due to YAP1 gene amplification, mutations in FAT families as upstream activators of neurofibromin 2 (NF2)/merlin in the Hippo pathway, and deletion of vestigial-like family member 4 (VGLL4)/autophagy related 7 (ATG7) as a direct competitor with YAP for binding to TEADs.^6,12,240 In addition, YAP1 is frequently overexpressed in 56–81% of ESCC cases, which are associated with aggressive phenotype.^241,242 YAP1 controls the squamous differentiation of esophageal basal cells through crosstalk with Notch signaling, and ectopic nuclear YAP1 activation causes a hyperplastic phenotype of the esophageal epithelium in gain-of-function murine models. ²⁴³ Knockdown of YAP1 results in increased apoptosis and decreased properties of proliferation, invasion, chemoresistance, and CSCs in ESCC.^{240,244 –246} Thus, YAP is a promising target for ESCC.

Currently, there are potentially viable agents targeting YAP1 signaling, including TEAD inhibitors (CA3, MYF-01–37, K-975, and IK-930),^{238,247 –249} BRD4 inhibitor (JQ1) as blockade of both YAP1 transcription and direct interaction of BRD4 with YAP1,^239,250 tankyrase inhibitor (XAV939) as an indirect inhibitor of angiomotin-mediated YAP1 activity, ²⁵¹ and potent next-generation constrained ethyl antisense oligonucleotides to selectively target YAP1 (ON-537). ²⁵² As the YAP1-TEAD transcription complex controls the expression of CDK6, which leads to cellular senescence, ²⁵³ dual blockade of YAP1 and CDK6 may also be effective in ESCC. ²⁵⁴ IK-930 is a selective TEAD inhibitor that disrupts the TEAD-dependent transcription of key genes involved in cancer progression, metastases, and therapeutic resistance. A preclinical study showed the antitumor activity of IK-930 both as a monotherapy and in combination with targeted agents, such as MEK and EGFR inhibitors, in multiple tumor models. ²⁴⁹ A phase I first-in human trial of IK-930 is ongoing to evaluate the safety and preliminary antitumor activity in advanced solid tumors (NCT05228015). ION-537 is a next-generation antisense oligonucleotide inhibitor of YAP1, which has resulted not only in a markedly decreased expression of YAP1 protein and preclinical treatment efficacy, but also in enhanced immune activation, such as an increased number of infiltrating T cells and myeloid lineage cells in the tumors. ²⁵² A phase I trial of ION537 is ongoing in molecularly selected advanced solid tumors (NCT04659096). Although YAP1 downstream pathway-targeted therapy represents a rational approach, it is unlikely that targeting one downstream effector will fully recapitulate the effects of the direct blockade of YAP1 because of the diverse functions and escape mechanisms of YAP1-targeted molecules.

Combined immunotherapy

Although ESCC represents a potential anti-tumorigenic immune environment through activated T cells and increased dendritic cells (DCs), the infiltration of cytotoxic T cells into tumors is hindered by exclusionary pressures. ²⁵⁵ In a single-cell transcriptome analysis of the TME, ESCCs had profound heterogeneity in the cell composition of individual tumors and their cellular status, leading to distinct immune status and tumor biology.^155,156 In addition, different subsets of T cells express distinct inhibitory immune checkpoint molecules, such as CTLA-4, T-cell immunoreceptor with Ig and ITIM domains (TIGIT), CD96, and lymphocyte activation gene 3 protein (LAG3), in the TME of ESCC.^155,156 Collectively, most ESCCs may be ‘cold’ tumors, with low TILs and high expression of co-inhibitory molecules, and the immunologically ignorant phenotype may have a poor response to ICIs. Therefore, several treatment strategies have been examined to turn immunologically ‘cold’ tumors with poor immune activation into ‘hot’ tumors with strong immune infiltration in clinical trials combining the anti-PD-1/PD-L1 mAb with other immune-modulating treatments, including other ICIs, anti-angiogenetic inhibitors, and other agents (Table 6).

KN046 is a novel recombinant humanized PD-L1/CTLA-4 bispecific single-domain antibody-Fc fusion protein that can bind effectively to both PD-L1 and CTLA-4. ²⁵⁶ A phase II KN046-204 trial (NCT03925870) of KN046 monotherapy or combined with chemotherapy in ESCC is ongoing, in which the ORR and DCR are 58.3% and 91.6%, respectively, in a cohort with KN046 plus first-line chemotherapy. ¹⁵⁰ The treatment efficacy of a fully human immunoglobin G4 anti-PD-L1 mAb sugemalimab (CS1001) in combination with first-line chemotherapy is also assessing in a phase III trial (NCT04187352) (Table 6).

The PD-1/PD-L1 interaction is not the only immune checkpoint pathway regulating T-cell activation in the TME. TIGIT, LAG3, and T-cell immunoglobulin mucin receptor 3 (TIM3) are overexpressed on effector CD4+ and CD8+ T cells, regulatory T cells (Tregs), and natural killer (NK) cells, which act as inhibitory immune checkpoint modulators. ²⁵⁷ TIGIT binds to CD155 with high affinity and competes with its activating counter-receptor CD226, which suppresses antitumor immunity through inactivating immune cell effector functions, either directly, or indirectly through Treg-mediated suppression. ²⁵⁸ The anti-TIGIT mAb tiragolumab plus the anti-PD-L1 mAb atezolizumab showed a clinically meaningful benefit of ORR and PFS compared with atezolizumab alone as a first-line treatment for patients with PD-L1-positive lung cancer. ²⁵⁹ In ESCC, a phase III SKYSCRAPER-08 trial (NCT04540211) is ongoing to evaluate the efficacy of tiragolumab plus atezolizumab in combination with first-line chemotherapy versus chemotherapy alone (Table 6). In a randomized phase II AdvanTIG-203 (NCT04732494), the clinical benefit of anti-TIGIT mAb ociperlimab (BGB-A1217) plus anti-PD-1 mAb tislelizumab versus tislelizumab alone was assessed in a second-line setting. LAG-3 has high-affinity binding to the major histocompatibility complex class II, which acts as an inhibitory immune checkpoint via the activation of Tregs and the suppression of DCs and CD8 + T cells. ²⁶⁰ LAG-3 and PD-1 synergistically regulate T-cell function in promoting immune escape. ²⁶¹ A phase I trial of anti-PD-1 mAb BI 754,091 alone and in combination with anti-LAG-3 mAb BI 754,111 was conducted in Asian patients with advanced solid tumors, including ESCCs. ²⁶² In a cohort of BI 754,091 plus BI 754,111 in EC patients who had received ≥1 line of prior systemic therapy and no prior anti-PD-1/PD-L1 therapy, the ORR was 21.6% (8/37). A randomized phase II trial (NCT04785820) of a PD-1/LAG-3 BiAb RO7247669 versus anti-PD-1 mAb nivolumab in ESCC is ongoing. T-cell immunoglobulin mucin-3 (TIM-3) expressed on innate immune cells is a crucial checkpoint that can induce T-cell apoptosis and exhaust type 1 CD4+ T cells and type 1 CD8+ T cells by binding to its ligand glycan-binding protein-9 expressed on cancer cells.^257,263 The blockade of TIM-3 preclinically restored the functions of TILs, ²⁶⁴ overcame resistance to an anti-PD-1 mAb 234, and promoted a synergistic effect in combination with anti-PD-1 mAb. ²⁶⁵ The randomized phase II trial (NCT04785820) is assessing the efficacy of a PD-1/TIM-3 BiAb RO7121661 versus nivolumab in ESCC (Table 6).

As VEGF has been found to exert immunosuppressive effects via several mechanisms, such as decreasing the number of TILs, activating immune checkpoint molecules, inhibiting DC differentiation, and downregulating MHC,^136,266 antiangiogenic agents could have immunomodulatory effects. In fact, the most compatible partners of ICIs were an anti-angiogenetic inhibitor and platinum chemotherapy in a cross-sectional study from 98 clinical trials including 24,915 patients. ²⁶⁷ Thus, the combination of antiangiogenetic agents with ICI has emerged as an alternative strategy. In a phase II multiple cohort trial of regorafenib, a novel oral multi-kinase inhibitor including angiogenic kinases, in combination with nivolumab in advanced solid tumors, the ESCC cohort showed an ORR of 43% and median PFS of 6.9 months in 30 patients. ²⁶⁸ Currently, there are several ongoing trials combining ICI with apatinib (NCT05049681), anlotinib (NCT05038813, NCT05013697, NCT04797507, and NCT04503967), famitinib (NCT04866381), and cabozantinib (NCT05007613; Table 6). Lenvatinib targets VEGFR 1, 2, and 3, PDGFR α, FGFR, c-KIT, and RET tyrosine kinases. A systematic review showed more promising antitumor activity in solid cancers with lenvatinib plus pembrolizumab than with either lenvatinib or pembrolizumab alone. ²⁶⁹ A phase III LEAP-014 trial (NCT04949256) of lenvatinib plus pembrolizumab versus pembrolizumab in combination with first-line chemotherapy for ESCC patients is ongoing.

The EGFR signaling pathway establishes the non-inflamed and immunosuppressive TME, with recruitment and activated function of Tregs, downregulated MHC expression, and decreased numbers of effector CD8+ T cells, DCs, and M1-like tumor-associated macrophages. ²⁷⁰ Therefore, blockade of EGFR signaling may synergistically enhance the efficacy of PD-1/PD-L1 inhibitors in ESCC (NCT04880811, NCT04839471, NCT03766178, and NCT04229537) (Table 6).

Activation of FGFR suppresses cancer immunity through inhibition of interferon-gamma-stimulated JAK/STAT signaling pathway ²⁷¹ and downregulation of MHC class I and MHC class II. ²⁷² In preclinical models, the combination of FGFR-TKIs with anti-PD-1/PD-L1 mAb augmented antitumor immunity in renal cell carcinoma ²⁷¹ and head and neck squamous cell carcinoma. ²⁷² As FGFR is frequently activated and associated with poor prognosis in ESCC,^188,190 the treatment strategy targeting both FGFR and PD-1/PD-L1 may be a novel immunologic approach for treating patients with FGFR-aberrant ESCC. In a phase Ib trial of futibatinib plus pembrolizumab in advanced or metastatic EC harboring FGFR mRNA overexpression, the interim analysis showed acceptable safety, with an ORR of 44% in ICIs naïve patients and 20% in ICI refractory patients. ²⁷³

Src homology 2 domain-containing phosphatase (SHP2) is a non-receptor ubiquitous protein tyrosine phosphatase, ²⁷⁴ which serves as an important hub to connect several intracellular oncogenic signaling pathways. ²⁷⁵ It not only suppresses T-cell activation, but also causes T-cell anergy as a downstream molecule in the PD-1 signaling pathway.^276,277 In preclinical studies, SHP2 inhibition reversed immunosuppression in the TME via restored Th1 immunity and activated T-cell function. ²⁷⁷ Of note, the combined inhibition of SHP2 and anti-PD-1 had a higher antitumor activity than either inhibition alone. ²⁷⁸ Thus, the dual blockade of SHP2 and PD-1/PD-L1 may enhance immune-mediated responses. The rationale for dual blockade has been assessed in phase I trials of anti-PD-1 mAb plus allosteric SHP2 inhibitor JAB-0368 (NCT04721223) or TNO155 (NCT04000529; Table 6).

Interleukin-2 (IL-2) is an immunoregulatory cytokine that drives effector T and NK cell expansion through the IL-2 receptor beta/gamma complex-JAK/STAT signaling axis. ²⁷⁹ In contrast, IL-2 is essential for generation of Tregs, which are more sensitive to IL-2 stimulation than the other immune cell populations, such as conventional T cells and NK cells, due to constitutive expression of IL-2 receptor alpha. ²⁸⁰ These conflicting functions suggest that a more selective approach is required for therapeutic applications. THOR-707 is a recombinant human IL-2 molecule irreversibly bound to a pegylated chain to block alpha-binding while retaining affinity for beta/gamma IL-2 receptor subunits. ²⁸¹ Treatment with THOR-707 resulted in enhanced tumor infiltration of CD8+ T cells and NK cells, without Treg expansion, and a reduction of tumor growth in vitro. In an ongoing multicohort phase II trial, the clinical benefit of THOR-707 plus anti-PD-1 mAb was assessed in a subcohort of ESCC (NCT05104567; Table 6).

Histone deacetylase (HDAC) inhibitors enhance immunogenicity and antitumor immune responses through the reduced number and function of myeloid-derived suppressor cells and Tregs,^{282 –284} enhanced tumor antigens, ²⁸⁵ and increased MHC presentation. ²⁸⁵ On the other hand, treatment with HDAC inhibitors resulted in upregulation of PD-L1, ²⁸⁶ and nuclear acetylated PD-L1 stimulated antigen presentation. ²⁸⁷ These preclinical data provide a rationale for combining HDAC inhibitors with anti-PD-1/PD-L1 therapies. Several early clinical trials have shown encouraging effects of combination treatment with the HDAC inhibitor and the PD-1/PD-L1 inhibitor in melanoma, lung cancer, or head and neck cancer.^{288 –290} Tucidinostat is an orally bioavailable benzamide-type inhibitor of HDAC isoenzymes 1, 2, 3, and 10. In ESCC, two phase II trials of tucidinostat plus PD-1 inhibitors are ongoing (NCT04984018 and NCT05163483; Table 6).