Single-cell RNA sequencing to map tumor heterogeneity in gastric carcinogenesis paving roads to individualized therapy

Abstract

Gastric cancer (GC) is a highly heterogeneous disease with a complex tumor microenvironment (TME) that encompasses multiple cell types including cancer cells, immune cells, stromal cells, and so on. Cancer-associated cells could remodel the TME and influence the progression of GC and therapeutic response. Single-cell RNA sequencing (scRNA-seq), as an emerging technology, has provided unprecedented insights into the complicated biological composition and characteristics of TME at the molecular, cellular, and immunological resolutions, offering a new idea for GC studies. In this review, we discuss the novel findings from scRNA-seq datasets revealing the origin and evolution of GC, and scRNA-seq is a powerful tool for investigating transcriptional dynamics and intratumor heterogeneity (ITH) in GC. Meanwhile, we demonstrate that the vital immune cells within TME, including T cells, B cells, macrophages, and stromal cells, play an important role in the disease progression. Additionally, we also overview that how scRNA-seq facilitates our understanding about the effects on individualized therapy of GC patients. Spatial transcriptomes (ST) have been designed to determine spatial distribution and capture local intercellular communication networks, enabling a further understanding of the relationship between the spatial background of a particular cell and its functions. In summary, scRNA-seq and other single-cell technologies provide a valuable perspective for molecular and pathological disease characteristics and hold promise for advancing basic research and clinical practice in GC.

Supplementary Information

The online version contains supplementary material available at 10.1007/s00262-024-03820-4.

Background

GC ranks fifth in malignancy incidence worldwide and is also the fifth leading cause of cancer-related death [1]. It is a highly heterogeneous disease with multiple histological and molecular subtypes, which are distinguished with the Lauren classification system [2], The Cancer Genome Atlas subtypes (TCGA) [3], and Asian Cancer Research Group subtypes (ACRG) [4]. Due to the lack of effective clinical markers, delayed diagnosis is common in patients with an advanced stage, and the five-year survival rate is still less than 5% for GC patients with metastasis [5]. Although the classifications are beneficial for distinguishing subtypes of GC and confirming treatment decisions, they may not be comprehensive enough to guide personalized therapies leading to improved outcome. The complexity of identifying standard therapy methods is increasing via the significant variation in histopathological or molecular subtypes among different patients [6]. Additionally, the mechanisms driving GC development have not been fully elucidated. TME is a complex ecosystem that includes cancer cells, immune cells, and stromal cells, and it plays a vital role in tumorigenesis, progression, invasion, and drug resistance [7, 8]. The interactions among these cells contribute to the constantly evolving TME during tumor progression, and in turn this dynamic process significantly impacts tumor development. Thus, there is an urgent need for understanding the TME to identify new therapeutic targets and advance the clinical practice of GC.

Recent advancements in single-cell technologies, such as scRNA-seq, ST, etc., have facilitated a more in-depth analysis of tumorigenesis, tumor development, TME heterogeneity and individualized therapies in multiple dimensions at the single-cell resolution [9–12]. Unlike traditional bulk RNA sequencing, which averages gene expression levels across all cells, scRNA-seq captures the transcripts of individual cell, allowing for an insightful understanding of the gene expression profiles of each cell. Single-cell technique not only facilitates the discovery of differences in cellular composition and features but also identifies certain rare cell populations that are often obscured in bulk RNA-seq analyses [13]. However, single-cell technologies produce a mass of high-dimensional and complicated data, making it unfeasible to use conventional computational methods. Deep learning, a potential alternative to the traditional machine learning algorithms, could solve these problems in single-cell studies [14]. Especially for some intricate pathological phenotypes such as carcinoma, drug resistance, and neurobiology, deep learning could make a profound significance for the single-cell studies. In the field of GC, several reviews have shown the overview procedures of single-cell data to facilitate a better understanding and application of the technologies [15–17]; there is also one paper about breast cancer that summarized the advantages of several single-cell multi-omics approaches, including scRNA-seq and ST [18].

This review will highlight recent advances in the TME landscape obtained through single-cell technologies. In particular, we focused on characterizing features of the cancer cells, immune cells, and stromal cells, including their heterogeneity, dynamics, and potential roles in the TME. Additionally, this review delves into the interactions among various cell subpopulations and discusses advancements in therapeutic approaches that were inspired via predictive biomarkers and novel targets identified by advanced single-cell technologies in GC.

Single-cell technologies reveal the origin and evolution of GC

The human stomach, a muscular organ for food storage and digestion, contains cardia, fundus, corpus (body), and antrum (pylorus). Its single-layered epithelium consists of glandular units known as oxyntic, antral, and pyloric glands depending on their anatomic location in the cardia, corpus, or the pylorus region, respectively [19]. Human gastric single-cell atlas has identified all known and rare cell types, including the pit/foveolar cells (GKN1 and GKN2), parietal cells (ATP4A and ATP4B), proliferating isthmus cells (MKI67 and STMN1), mucous neck cells (MUC6), chief cells (LIPF and PGA5), endocrine cells (GAST, GHRL, SST, and HDC), and tuft cells (TRPM5 and SH2D6) [20]. Wang et al. [21] revealed that chief cells, parietal cells, and enteroendocrine cells were seldom observed in the malignant epithelial subpopulation; in contrast, gland and pit mucous cells, along with AQP5⁺ stem cells, were predominant during the progression of malignancy. Tsubosaka et al. [12] generated a “stomach encyclopedia” by integrating scRNA-seq and ST to uncover cell diversity and homeostatic regulation of human stomach; they also identified LEFTY1 as a novel stem cell marker in the gastric mucosa. Yet, molecular knowledge of human gastric corpus epithelium is still limited, through single-cell multi-omics technologies, including scRNA-seq, ST, and single-cell assay for transposase-accessible chromatin sequencing (scATAC-seq); Dong et al. [22] have provided a new perspective for a systematic comprehension of cellular diversity and homeostasis of human gastric corpus epithelium. In their paper, a stem/progenitor cell population was discovered in the isthmus of human gastric corpus, where EGF and WNT signaling pathways were activated. In addition, FABP5 and NME1 were identified and confirmed to be important for both normal gastric stem/progenitor cells and GC cells.

According to Correa hypothesis, GC originated from chronic atrophic gastritis (CAG), intestinal metaplasia (IM), and eventually developed GC [23]. IL-13, secreted by various immune cell subsets including mast cells, B cells, macrophages, innate lymphoid cells type 2 (ILC2s), and T cells, can act directly on the gastric epithelium to promote metaplasia development during chronic gastritis [24]. IM is a pre-malignant condition of the gastric mucosa, and IM cells are considered to have a precancerous property with an increased opportunity of transformation into cancer cells [25]. scRNA-seq and ST profiles highlight IM lineage heterogeneity with SOX9 being a new IM driver gene, and IM molecular subtypes are correlated with inflammation and microbial communities [26]. Another study validated that SOX9 marked gastric stem cells and modulated biased symmetric cell division, which seemed to be required for the malignant transformation of gastric stem cells [27]. In addition, CAG and IM involving the gastric mucosa are considered crucial steps in intestinal-type GC (IGC) pathogenesis [28]. Change in cell types plays an essential role in the cascade from premalignant lesions to the malignant lesions. Zhang et al. [29] established a single-cell atlas using scRNA-seq from gastric antral mucosa biopsies of patients, including non-atrophic gastritis, CAG, IM, and early GC. Their results demonstrated that gland mucous cells tended to acquire an intestinal-like stem cell phenotype during metaplasia, and they also identified OR51E1 as a distinctive marker for endocrine cells in the early-malignant lesion. Additionally, HES6 could potentially serve as a marker for pre-goblet cells to help identifying the metaplasia at the early stage. Meanwhile, Wang et al. [9] using scRNA-seq depicted a comprehensive single-cell profiling of normal tissues, precancerous lesions (including CAG and IM), localized GC, and metastatic GC, revealing alterations of cell states and compositions within TME as GC progression. Specially, they found that IgA⁺ plasma cells are abundant in premalignant lesions, whereas immunosuppressive myeloid and stromal cells dominated late-stage GC.

There are two main forms of metaplasia commonly discovered in the stomach: IM and spasmolytic polypeptide-expressing metaplasia (SPEM). SPEM may serve as a precursor to IM and is considered a significant biomarker in the GC development [30–32]. Several scRNA-seq datasets have depicted the role of SPEM in GC progression. Bockerstett et al. [33] demonstrated that gastrokine 3 (Gkn3) mRNA served as a specific marker for SPEM in mice; mucous neck cells as well as chief cells exhibited plasticity and converged into a pre-metaplastic cell type that progressed to metaplasia under chronic inflammatory conditions. At the same time, they expanded the definition of SPEM to include MUC6⁺ TFF2⁺ cells that did not express mature chief cell transcripts such as gastric intrinsic factor [34]. Zhang et al. [35] also provided molecular evidences for potential transition from gastric chief cells into MUC6⁺ TFF2⁺ SPEM by analyzing non-malignant epithelium. Additionally, ILC2s were increasing in stomach tissues of patients with SPEM compared to those with normal corpus mucosa, indicating their involvement in coordinating the metaplastic response to severe gastric injury [36]. scRNA-seq of gastric leukocytes in mice showed that ILC2s expressed high levels of glucocorticoid receptor (GR) and androgen receptor (AR), and 5α-dihydrotestosterone treatment in mice significantly suppressed the expression of the proinflammatory cytokines I113 and Csf2 by ILC2s; moreover, ILC2s depletion could protect the stomach from SPEM development [37]. Thus, it indicated that sex hormones played a critical role in regulating gastric inflammation and metaplasia. Meanwhile, scRNA-seq data about mice with autoimmune gastritis revealed that IL27 could suppress gastritis and SPEM by suppressing CD4⁺ T cells-mediated inflammation in the mucosa [38].

scRNA-seq technologies uncover GC heterogeneity

Tumor heterogeneity derives from the presence of cell groups with distinct genotypes among growth process about tumor cells, and these various cell groups could contribute to different phenotypes. Tumors with high intratumoral heterogeneity may result in inferior clinical outcomes for GC patients. Increasing evidence suggests that tumor heterogeneity allows cancer cells to survive conventional chemotherapy, radiotherapy, and targeted therapies [39, 40]. The emergence of drug resistance may be attributed to selective pressure from treatment, which either promotes the expansion of the pre-existing subclonal or facilitates the evolution of drug-tolerant cells. Meanwhile, tumor heterogeneity will also have a significant impact on the efficacy of immunotherapy, especially for immune checkpoint inhibitors (ICIs) [41]. Deciphering tumor heterogeneity is crucial for comprehending the biology of primary GC and its metastasis, as well as pinpointing potential therapeutic targets. scRNA-seq allows researchers to investigate the heterogeneity within GC tumors at a single-cell solution of cellular and molecular landscapes, revealing the complex biology features of these tumors [35, 42–45]. Lu et al. [42] generated a series of GC models using genome-edited gastric organoids to elucidate the progression of tumorigenesis from dysplasia to well-differentiated, poorly differentiated, and metastatic GC in mice, and their results highlighted the interaction between stomach-specific macrophages and GC cells via the fibronectin-integrin α6β4, which promoted the progression and metastases of GC. Kim et al. [43] identified 10 gastric cell subtypes and found that the IGC and diffuse-type GC (DGC) have distinct cell populations, IGC cells differentiated along the IM lineage, whereas DGC cells resembled the de novo pathway. In addition, molecular signatures of intratumoral heterogeneities and differentiation states of GC have been uncovered using integrative single-cell multi-omics analyses [44]. Ge et al. [45] performed scRNA-seq and single-cell T cell receptor sequencing (scTCR-seq) on treatment-naïve patients of GC, and they identified distinct ITH groups, indicating that ITH could potentially be served as a prognostic marker for predicting GC recurrence.

Metastasis is the main cause of cancer-related death in patients with GC, and scRNA-seq technology plays a significant role in the mechanism studies of GC metastasis. Studies have shown that epithelial-to-mesenchymal transition (EMT) is one of the key points in the acquisition of invasive and metastatic ability of tumor cells [46–48]. Circulating tumor cells (CTCs) migrate from primary tumor into blood vessels and contribute to the metastasis of GC to distant tissues. Analysis of the transcriptome profile of single-CTCs indicated that a large proportion of gastric CTCs experienced EMT, with platelet adhesion identified as a potential key factor in driving EMT progression and acquisition of chemoresistance [49]. In another study, DAZ interacting zinc finger protein 1 (DZIP1) was identified to be upregulated in cancer-associated fibroblasts (CAFs) and malignant epithelial cells, and it is involved in remodeling the immunosuppressive microenvironment and inducing EMT through participating in tumor-stromal signaling crosstalk [50].

Features of GC metastases in different organs have been characterized at single-cell resolution (Fig. 1). Peritoneum, liver, lymph nodes, spleen, pancreas, ovary are all friendly sites for GC cells to grow and form metastatic lesions [51]. Peritoneal metastasis (PM) of GC is the most common form of distant metastasis and is associated with significantly unfavorable prognosis and high mortality rates [52]. The rapid advancement of scRNA-seq technology has significantly enhanced our understanding of the molecular and cellular features of PM in GC, and numerous studies have been conducted to explore the specific characteristics of PM at the single-cell resolution [53–56]. Wang et al. [53] constructed a single-cell transcriptome map of peritoneal carcinomatosis from GC patients and profiled the transcriptome states of tumor cell populations, incisively depicted the ITH of malignant PC cells, and identified its significant correlation with patient survival. During PM in GC, there was an obvious increase of monocyte-like dendritic cells (DCs) that exhibited pro-angiogenic properties and reduced antigen-presenting capacity and were associated with a poor prognosis in GC [54]. Additionally, immunosuppressive macrophages can transition from cathepsin^high (CTS^high) tumor-associated macrophages (TAMs) to complement 1q^high (C1Q^high) TAMs in ascites from GC patients with PM [55]. Fibroblasts also played a crucial role, with CLDN11 secreted by CXCR7⁺ fibroblasts promoting GC proliferation and PM; moreover, CXCR7⁺ fibroblasts were closely associated with M2-macrophage infiltration in tissues [56].

Fig. 1.

Overviews of GC organ metastasis and corresponding cells changes discovered by scRNA-seq. The picture shows the tissues or organs (liver, lymph node, peritoneum) where the GC cell subpopulations metastasize and corresponding alternations of immune cells and stromal cells. CSCs marker gene CXCR4 might be a key gene affecting the LM of GC. Meanwhile, the interaction between SPP1⁺ TAMs and CD8⁺ Tex cells is also essential for the immunosuppressive microenvironment of LM in GC through GDF15-TGFBR2 axis. A 20-gene signature of LNM-derived CD8⁺ Tex cells might forecast LNM and disordered neutrophils might contribute to LNM in GC. Monocyte-like DCs are increasing and immunosuppressive macrophages transit from CTS^high to C1Q^high TAMs in ascites from patients with PM in GC; in addition, CXCR7-positive fibroblasts also play an important role in PM of GC via secreting CLDN11. GC, gastric cancer; scRNA-seq, single-cell RNA sequencing; CSC, cancer stem cell; CXCR, C-X-C motif chemokine receptor; LM, liver metastasis; TAM, tumor-associated macrophage; Tex, exhausted T cell; LNM, lymph node metastasis; DC, dendritic cell; PM, peritoneal metastasis; CTS^high, cathepsin^high; C1Q^high, complement 1q^high

Lymph node metastasis (LNM) is a common route of metastasis in GC. Based on scRNA-seq, Qian et al. [57] demonstrated that disordering polarization and maturation of neutrophils, along with the activation of immune checkpoint SPP1, could contribute to LNM in GC. Hence, this study offered a new perspective for the mechanisms and potential therapeutic targets for LNM in GC. Furthermore, a 20-gene signature of CD8⁺ exhausted T (Tex) cells derived from lymph nodes could be used to predict LNM, and targeting the HLA-E-KLRC1/KLRC2 signaling pathway may present new clinical treatment opportunities for GC [58]. Cancer stem cells (CSCs) are crucial for cancer treatment and prognosis, as they are involved in tumor recurrence and metastasis [59–61]. In a study, it was found that through scRNA-seq combined with bulk RNA-seq, CSCs marker gene C-X-C motif chemokine receptor 4 (CXCR4) may be a pivotal gene facilitating phenotypes of CSCs, contributing to tumor growth and liver metastasis (LM) in GC [62]. In the meanwhile, the interaction between SPP1⁺ TAMs and CD8⁺ Tex cells is crucial in the immunosuppressive microenvironment of LM in GC; among these interactions, GDF15-TGFBR2 may play an indispensable immunosuppressive role [63]. Overall, GC frequently exhibits intratumoral heterogeneity, which can significantly impact therapeutic responses and clinical outcomes. The advent of scRNA-seq has enabled the evaluation of intratumoral heterogeneity in GC at a single-cell resolution which facilitates researchers to elucidate the substantial biological complexity in tumors.

scRNA-seq enables the discovery of GC microenvironment

TME is a complex ecosystem comprising heterogeneous tumor cells, immune cells, and stromal cells, including CAFs andendothelial cells, proangiogenic mediators, soluble growth factors, and extracellular matrix (ECM), contributing to tumor initiation, progression, invasion, and drug resistance [64, 65]. The defining markers of these cells are listed in Table 1. Comprehending the intricate interactions between tumor cell-intrinsic, cell-extrinsic, and systemic factors is crucial for the development of effective anticancer treatment [7]. In the TME, scRNA-seq can differentiate various cell types of GC at the single-cell level, and this technology provides valuable insights into the interactions and regulatory mechanisms among cells (Supplementary Fig. S1). Highly aggressive gastric cancer (HAGC) is a subtype of GC known for bone marrow metastasis and disseminated intravascular coagulation (DIC). Researchers utilized scRNA-seq to elucidate the immune response of patients with HAGC by analyzing peripheral blood mononuclear cells (PBMCs) [66], and their study uncovered that the induction of activated yet immature neutrophils in HAGC played a key role in the immunosuppression of lymphoid cells and the activation of signaling pathways correlated to DIC progression. ARID1A is the second most commonly mutated driver gene in GC and analysis of scRNA-seq profiles demonstrated that prevalent ARID1A inactivation across GC molecular subtypes was associated with a NFκB-driven proinflammatory TME [67]. Epigenomic alterations in cancer interacting with the immune microenvironment also influence tumor progression and therapeutic response, Sundar et al. [68] utilized scRNA-seq to show that alternate promoter burden (APB)_high GC exhibited reduced levels of T cell cytolytic activity and signatures of immune depletion; furthermore, immunotherapy-treated gastrointestinal cancer (GI) patients displayed resistance of APB_high tumor to ICIs. Altogether, these results highlighted an intimate connection between APB_high and tumor immune microenvironment, leading to immune evasion and resistance to immunotherapy.

Table 1.

GC immune microenvironment containing various cell subtypes

Cell types	Subtypes	Gene signatures	References
T cell		CD3D, CD3E	[54]
Activated T cell		CD69	[69]
Proliferative T cell		CCNB2, DLGAP5, MKI67	[69]
CD4+ T cell	Naïve CD4+ T cell	CCR7, LEF1	[58]
	CD4+ Tex cell	PDCD1, CXCL13, TIGIT	[58, 69]
	CD4+ TEM	CCL5, ANXA1, GZMA	[58]
	CD4+ Treg	FOXP3, CTLA4	[58]
	GADD45B+ Th1-like CD4+ T cell	GADD45B, TNF	[58]
	CD4+ Tconv cell	LAYN, TNFRSF9, IL2RA, CCR8, CTLA4	[70]
CD8+ T cell	Naïve CD8+ T cell	LEF1, SELL	[58]
	CTL	NKG7, GZMA, GNLY	[58, 69]
	CD8+ Tex cell	CTLA4, LAG3, TIGIT	[58]
	CD8+ TEM	GZMK, CXCR4, EOMES	[58]
	MAIT cell	SLC4A10, KLRB1	[58]
	Tc17	IL17, CXCL12	[71]
Treg cell		FOXP3, ICOS, CTLA4	[58, 69]
B cell		CD79B, MS4A1, CD79A	[35, 54]
	CXCR4+ B cell	CXCR4	[72]
NK cell		KLRC1, KLRF1	[54]
	CD8− NK cell	NKG7, GZMA	[69]
Macrophage		CD14	[35]
	SPP1+ TAM	GDF15	[63]
	CTShigh TAM	CTSA, CTSD, CXCL2, LAMC1, NRP1	[55]
	C1Qhigh TAM	C1QA, C1QB, C1QC	[55]
DC		CD1C	[69]
	cDC1	CLEC9, XCR1	[54]
	cDC2	CLEC10A, CD1C	[54]
	Monocyte-like DC	CCL2, CCL3, CCL4, IL1B	[54]
Neutrophil		FCGR3B, CSF3R	[54]
Mast cell		CPA3	[35]
Fibroblast		COL3A1, COL1A2, ACTA2, NNMT,	[35, 54]
	Myofibroblast	ACTA2, MYH11	[73]
	Pericyte	RGS5, ACTA2	[73]
	eCAF	MMP14, LOXL2, POSTN	[73]
	iCAF	CXCL12, IL-6, CXCL14	[73]
	CXCR7+ fibroblast	CXCR7	[56]
	CAFEndMT	RGS5, ACTA2, PLVAP, VWF	[74]

Tex exhausted T cell, T_EM effector memory T cell, Treg regulatory T cell, Th1 T helper type 1 cell, Tconv conventional T cell, CTL cytotoxic CD8⁺ T cell, MAIT mucosal-associated invariant T cell, Tc17 IL-17 CD8⁺ T cell, NK natural killer cell, CTS^high cathepsin^high, TAM tumor-associated macrophage, C1Q^high complement 1q^high, DC dendritic cell, cDC conventional DC, CAF cancer-associated fibroblast, eCAF extracellular matrix CAF, iCAF inflammatory CAF, CAF_EndMT endothelial-to-mesenchymal CAF

T cells

As a major component of the TME, immune cells take part in tumor immunity response processes. T cells have a crucial role in orchestrating the immune response and eliminating damaged cells [75]. scRNA-seq analysis demonstrated that T cells are the most significantly enriched immune cells in the TME of GC based on the expression of canonical marker genes [35, 58, 76, 77]. In general, these immune functions are mediated by CD4⁺ and CD8⁺ T cells, respectively. CD8⁺ T cells are commonly viewed as a population of cells that produce high levels of interferon gamma (IFN-γ) and granzyme B, working synergistically to kill infected or cancerous cells. CD8⁺ T cells can be classified into the conventional IFN-γ-producing Tc1s, interleukin (IL)-4-producing Tc2s, IL-9-producing Tc9s, IL-17-producing Tc17s, and IL-22-producing Tc22 subsets [78]. As typical cytotoxic CD8⁺ T cells, Tc1s show potent cytotoxic activity against tumor cells and cells harboring intracellular pathogens. However, Tc1s could be rendered ineffective primarily by the suppression of regulatory T cells (Tregs) [76]. Tc17s, derived from tissue-resident memory T cell populations, have the ability to differentiate into Tex cells, and Tc17s may promote tumor progression through IL-17, IL22 and IL26 signaling [71]. In addition, Tc17s isolated from gastric tumor can stimulate tumor cells to produce CXCL12, which recruited myeloid-derived suppressor cells (MDSCs) to suppress CD8⁺ T cells [79]. Ke et al. [70] analyzed scRNA-seq data of GC patients and employed flow to investigate the glucocorticoid-induced tumor necrosis factor receptor (GITR) expression across T cell subsets, and their results indicated that the abundance of intratumoral GITR⁺ CD4⁺ Treg cells and CD4⁺ conventional T cells was associated with immunosuppressive TME and worse overall survival of GC patients. Sathe et al. [80] utilized ex vivo tumor slice cultures obtained from fresh surgical resections of gastric and colon cancer to investigate the effects of GITR agonist or T cell Ig and ITIM domain (TIGIT) antagonist. Through the application of paired scRNA-seq and TCR sequencing, they uncovered novel cellular mechanisms underlying the action of GITR and TIGIT immunotherapy within the TME. While the GITR agonist had a limited transcriptional response restricted to cytotoxic CD8⁺ T cells, the TIGIT antagonist orchestrated a multicellular transcriptional reprogramming of the TME involving activation of cytotoxic and dysfunctional CD8⁺ T cells, T follicular helper-like cells, and DCs together with a reduction of Treg cells phenotype. Hence, understanding the cellular and transcriptional mechanisms of response or resistance will help prioritization of targets and clinical translation. Gastric signet ring cell carcinoma (GSRCC), characterized by prominent mucin in cytoplasm and eccentric nucleus, is associated with a worse prognosis in advanced stages [81, 82]. Single-cell profiling of TME revealed that C-X-C motif chemokine ligand 13 (CXCL13)-producing CD8⁺ Tex cells played an important role in promoting antitumor response [83]. Compared to non-GSRCC, impaired CD8⁺ Tex cells-derived modulation was responsible for the inadequate immune response in GSRCC. Thus, enhancing the CXCL13-producing ability of CD8⁺ Tex cells could be crucial for reversing the refractory condition in GSRCC patients.

B cells

Tumor-infiltrating B cells (TIBs) play a crucial role in cellular and humoral immunity, but their contribution to anti-tumor immunity is still controversial. TIBs can be found within tissues and are a significant component of tertiary lymphatic structures (TLSs). Through scRNA-seq analysis, Jia et al. [69] discovered a substantial presence of mucosal-associated lymphoid tissue (MALT)-B cells in gastric adenocarcinoma tissues with mature TLSs, and MALT-B cells may promote antitumor immunity by activating the complement pathway. B cells from the gastric TME and normal sites have a similarity in transcriptional features in GC, however, plasma cell clusters exhibited obvious differences in the expression of genes encoding immunoglobulin isotypes with increased IgA encoding genes in normal tissues and IgG in gastric TME [84], this isotype switching may impact B cell antitumor immunity by limiting IgA-mediated complement activation. As we have mentioned before, CXCR4 as a marker gene of CSCs contributes to tumor growth and LM in GC [62], and by combining scRNA-seq and bulk RNA-seq data, Su et al. [72] confirmed that CXCR4 is also a hub gene for TIBs, and thus CXCR4 may be a novel target for GC therapy.

Macrophages

Macrophages distribute throughout the body and play a crucial role in the innate immune system that is essential in immune reactions and the maintenance of tissue homeostasis [85]. A recent study has shown that histamine signaling plays an important role in tissue macrophage differentiation and maintenance of gastric homeostasis via the suppression of bacterial overgrowth in stomach [86]. Diversity and plasticity are important properties of TAMs; these cells can polarize into opposite functional phenotypes in specific TME conditions, i.e., anti-tumor M1 macrophage and promoting-tumor M2 macrophage [87–90]. However, Sathe et al. [84] performed scRNA-seq on patients with GC and discovered that macrophages were transcriptionally heterogeneous and did not conform to a binary M1/M2 paradigm. Nevertheless, TAMs seem to exhibit more characteristics of the M2 subtype in other studies. Eum et al. [91] utilized scRNA-seq to investigate heterogeneity and functional interactions within immune cells and tumor cells in ascites of GC patients, their finding revealed that a majority of macrophages exhibited an alternatively activated ‘M2’ phenotype, which appeared to be affected by the metastatic sites and tumor types; and M2 TAMs-derived exosomes could promote GC progression by metastasis-associated lung adenocarcinoma transcript 1 (MALAT1)-mediated regulation of glycolysis [92]. Therapeutic effectiveness of immune checkpoint blockade (ICB) is limited in GC, but targeting myeloid checkpoints might be a promising approach to current ICB treatments. scRNA-seq data demonstrated that C5aR1⁺ and siglec-10⁺ TAMs showed immunosuppressive properties that could led to CD8⁺ T cell dysfunction, ultimately promoted immune evasion [93, 94], thus, targeting myeloid checkpoint may be a potential strategy of immunotherapy for GC patients. Overall, the rapid development of scRNA-seq holds promise to further delineate the different subsets of macrophages and their role in immunotherapy.

Stromal cells

In the TME, stromal cells can produce cytokines and chemokines that influence tumor growth and invasion via promoting ECM formation and angiogenesis [95]. Studies have identified distinct roles of CAFs and endothelial cells in regulating the antitumor immune response and facilitating cancer progression and metastasis. The gastric TME underwent a significant remodeling about its stromal component with EN10-SERPINE1 endothelial cells and F13-CTHRC1-activated fibroblasts representing tumor-specific cell populations, and cell communication network showed crosstalk between EN10-SERPINE1 and F13-CTHRC1 and proposed interaction axes that may impact angiogenesis, migration, and the EMT of GC cells [96]. Based on scRNA-seq, ACKR1 is specifically expressed in GC endothelial cells, which correlated with poor prognosis [76]. In addition, understanding how the stroma manipulates the TME could potentially enhance the effectiveness of GC treatment. Platelet-derived growth factor receptor beta (PDGFRβ) was predominantly expressed in DGC stroma, and scRNA-seq and ST uncovered that PDGFRα/β blockade reversed the immunosuppressive TME through stromal modification, highlighting the impact of stromal reprogramming on immune reactivation [97].

GC is a type of cancer where CAFs have been identified as key components in promoting tumor progression through the remodeling of the TME [98–100]. Hence, it is essential to investigate the dynamic and complex interactions of CAFs with all the TME components to develop effective therapy methods against CAFs and ultimately improve survival rates in GC. Li et al. [73] identified four CAF subclusters with distinct characteristics in GC using scRNA-seq. Within these CAF subsets, inflammatory CAFs (iCAFs) interacted with CD8⁺ T cells by secreting IL-6 and CXCL12, and periostin (POSTN)⁺ extracellular matrix CAFs (eCAFs) showed an increased chemotaxis ability of attracting M2 macrophages and were associated with a decreased overall survival time of GC patients. By analyzing extensive scRNA-seq profiling, a variety of CAF-secreted proteins were discovered, among which the SERPINE2 was significantly abundant in stromal fibroblasts within GC tissues and contributed to a pro-tumor and immunosuppressive microenvironment [101]. Meanwhile, Luo et al. [74] conducted a pan-cancer analysis on 10 solid cancer types, including GC, to characterize the TME at single-cell resolution. Their finding highlighted the shared characteristics or plasticity of heterogeneous CAFs, particularly, SPP1⁺ TAMs may be involved in tumor angiogenesis via interacting with adjacent endothelial-to-mesenchymal transition CAF (CAF_EndMT), which was regarded as the initial step of angiogenesis.

Single-cell sequencing facilitates GC treatment

GC is a highly heterogeneous disease, so different patients respond diversely to therapeutic regimens [102]. scRNA-seq is a valuable method for identifying optimal therapeutic strategies to target heterogeneous cell populations [103–105]. Furthermore, it may identify alterations associated with therapeutic resistance in distinct cell clusters, facilitating the development of personalized medicine for patients with GC [106]. Use of single-cell technologies to detect or predict drug response as well as alterations within TME in GC patients is summarized in Table 2.

Table 2.

Overview of drug treatment response via scRNA-seq in GC

Treatment regimens	Sample types	Key findings	References
Chemotherapy	GC patient tissues	After NACT, TME became an immunosuppressive environment with increased percentages of the endothelial cells and fibroblasts	[107]
	GC patient tissues	Responders showed chemotherapy-induced NK cell infiltration, macrophage repolarization, and increased antigen presentation. Increased LAG3 and decreased DC abundance were observed in non-responders	[108]
	GEO database	IGF1+ CAFs may induce drug-resistant phenotype through IGF1-α6β4 integrin ligand-receptor binding and activation of EMT biological process	[109]
	GEO database	Depleted ECM components and increased immune processes are two vital TME features associated with 5-FU beneficial responses in GC patients	[110]
	Gastric organoids from mice and GC patients	Pyrvinium specifically targeted CD133+/CD166+ stem cell populations and proliferating cells in dysplastic organoids	[111]
Immunotherapy	Mice	Combining PDGFRα/β blockade and anti–PD-1 treatment synergistically suppressed the growth of fibrotic tumors	[97]
	GC patient tissues	MSI-H GC patients had a genomic, immunologic, and response heterogeneity treated with pembrolizumab	[112]
	GEO database (GSE183904)	C5aR1 is a myeloid checkpoint, and C5aR1 blockade combining with PD-1 inhibitor displayed a synergistic effect	[93]
	GEO database (Kumar et al., and Jeong et al.)	Siglec-10 is a myeloid checkpoint, blocking it reinvigorates the antitumor immune response and synergistically enhances anti-PD-1 immunotherapy response in GC	[94]
	Ex vivo tumor slice cultures from fresh surgical resections of GC	GITR agonist generated a limited transcriptional response, while TIGIT antagonist orchestrated a multicellular response involving CD8+ T cells, Tfh-like cells, DCs, and Tregs	[80]
	GC patient tissues	APBhigh tumor exhibited immunotherapy resistance to immune checkpoint inhibitor	[68]
Immunochemotherapy	GC patient tissues	A high baseline IFN-γ signature in CD8+ T cells can better predict the response to the neoadjuvant immunotherapy plus chemotherapy	[113]
	GC patient tissues	ISG15+ CD8+ T cells, enriched in the EBV+ GC patients, indicated benefit from immunochemotherapy for GC patients	[114]
Targeted therapy plus immunotherapy	Humanized PDX models	CXCL5/CXCR2 blockade via apatinib can enhance anti-PD-1 immunotherapy for GC	[115]

scRNA-seq single-cell RNA sequencing, GC gastric cancer, NACT neoadjuvant chemotherapy, TME tumor microenvironment, NK natural killer cell, LAG3 lymphocyte activation gene 3, DC dendritic cell, GEO gene expression omnibus, CAF cancer-associated fibroblast, IGF1 insulin-like growth factor 1, EMT epithelial–mesenchymal transition, ECM extracellular matrix, 5-FU fluorouracil, PDGFR platelet-derived growth factor receptor, PD-1 programmed cell death protein 1, MSI-H microsatellite instability-high, GITR glucocorticoid-induced tumor necrosis factor receptor, TIGIT T cell Ig and ITIM domain, Tfh follicular helper T cell, Treg regulatory T cell, APB, alternate promoter burden, IFN-γ interferon gamma, ISG15 interferon-stimulated gene 15, EBV Epstein-Barr virus, PDX patient-derived xenograft, CXCL5 C-X-C motif chemokine ligand 5, CXCR2 C-X-C motif chemokine receptor 2

Pre-cancerous metaplasia progression to dysplasia can enhance risk of GC, yet effective strategies to specifically target these pre-cancerous lesions are currently lacking. Chemotherapy is prevalent in first-line treatment of advanced GC, yet responses are heterogeneous. Based on the scRNA-seq, Kim et al. [111] revealed that pyrvinium, a putative chemotherapeutic agent, can specifically target CD133⁺/CD166⁺ stem cell populations and proliferating cells in dysplastic gastric organoid models established from GC patient, suggesting the effect of pyrvinium to reprogram the pre-cancerous milieu to prevent of GC progression. 5-FU is one of the most universally used chemotherapy drugs, in the treatment of advanced GC, whereas the innate or acquired drug resistance significantly hinders its survival benefit for patients with GC. Combining bulk sequencing with scRNA-seq data, Dong et al. [110] firstly deciphered that depleted ECM components and increased immune processes were two vital TME features associated with 5-FU beneficial responses in GC patients. Coupling whole-exome sequencing (WES), bulk RNA-seq, and scRNA-seq, Kim et al. [108] identified chemotherapy-induced natural killer (NK) cells infiltration, macrophages repolarization, and increased antigen presentation among responders using paired pretreatment and on-treatment specimens during standard first-line chemotherapy; on the other hand, non-responders exhibited increased lymphocyte activation gene 3 (LAG3) expression and decreased DC abundance, highlighting the TME remodeling during chemotherapy response and resistance.

Currently, curative surgical resection with perioperative chemotherapy is the principal treatment method for GC [116]. Neoadjuvant chemotherapy (NACT), also known as preoperative chemotherapy, is considered one of the viable options for patients with advanced GC [117–119]. TME can influence GC progression and metastasis, and meanwhile TME can be changed after NACT. Hence, a comprehensive analysis of the alterations in cell composition post-NACT could provide valuable insights for the subsequent treatments targeting therapeutic vulnerabilities within the TME to effectively control the progression of GC. Through scRNA-seq, Chen et al. [107] uncovered that compared to pre-treatment, TME became an immunosuppressive environment with increased percentages of the endothelial cell and fibroblast after NACT, and their results firstly demonstrated that myofibroblasts played an important role in chemotherapy resistance in GC.

A deeper comprehension of the TME can significantly promote immunotherapy for GC. In contrast to traditional radiotherapy and chemotherapy, immunotherapy has shown promising effectiveness and tolerable toxicity. ICIs, such as programmed death-ligand 1 (PD-L1) and cytotoxic T-lymphocyte-associated antigen 4 (CTLA-4) inhibitors, have been developed and extensively studied in both preclinical and clinical settings [120]. PD-L1 is expressed at high levels in various malignant tumors and is notably present in GC, particularly among patients with Epstein-Barr virus (EBV) infection or microsatellite instability (MSI) [121]. While the emergence of ICIs represents a significant breakthrough in the treatment of GC, their efficacy and toxicity continue to limit their widespread clinical application [122]. GC with MSI-high (MSI-H) is observed in about 20% cases [123], and MSI-H tumors are correlated with elevated tumor-infiltrating lymphocytes and enrichment of PD-L1 expression. Nevertheless, 50% GC with MSI-H are intrinsically resistant to programmed cell death protein 1 (PD-1) treatments [124]. scRNA-seq data revealed that only MSI-H GC patients with more elevated tumor mutational burden (TMB), abundant T cells infiltration, greater TCR clonal diversity, and less stem-like Tex cells at baseline may not require anything beyond pembrolizumab to achieve optimal outcomes [112]. Therefore, this study emphasized the response heterogeneity among patients with MSI-H GC treated with PD-1 antibody and highlighted the potential for extended baseline and early on-treatment biomarker analyses to identify responsers. In another study, using scRNA-seq and humanized patient-derived xenograft (PDX) models, Luo et al. [115] demonstrated that anti-PD-1 immunotherapy may facilitate the recruitment of protumor tumor-associated neutrophils (TANs) through CXCL5/CXCR2 axis to remodel an immunosuppressive TME; surprisingly, apatinib, a vascular endothelial growth factor receptor (VEGFR) inhibitor, could inhibit upregulation of CXCL5 induced by anti-PD-1 therapy in GC. In accordance, their study provided insights into the dynamic changes of GC TME during anti-PD-1 immunotherapy and underlined potential targets to address immune checkpoint immunotherapy resistance.

Li et al. [113] conducted scRNA-seq on GC samples treated with neoadjuvant immunotherapy plus chemotherapy to elucidate the interaction between TME characteristics and combination therapy. They demonstrated that CD8⁺ T cells with IFN-γ signature could better predict response to the combination therapy; meanwhile, they observed a significant decrease in the IFN-γ signature across various cell types, as well as a notable suppression of the CD8⁺ T cells with exhausted signature during the neoadjuvant therapy. Therefore, their results underscored the importance of the signature of CD8⁺ T cells in predicting response to the combined therapy for GC. EBV⁺ GC elicits fascinating immunotherapy response and exhibits an inflamed-immune phenotype with increased T cell and B cell infiltration. Qiu et al. [114] analyzed the features with the dynamic tumor immune landscape of EBV⁺ GC treated with immunochemotherapy using longitudinal scRNA-seq and paired scTCR/single-cell B cell receptor sequencing (scBCR-seq). Notably, interferon-stimulated gene 15 (ISG15)⁺ CD8⁺ T cells, significantly enriched in EBV⁺ GC patients, were identified as potential biomarker for predicting immunotherapy responsiveness in GC patients; reemerged clonotypes of pre-existing ISG15⁺ CD8⁺ T cells could be observed after immunochemotherapy, which contributed to a CXCL13-expressing effector cell populations in responsive EBV⁺ GC. Nevertheless, retention of LAG3 may lead to ISG15⁺ CD8⁺ T cells into a state of terminal exhaustion in non-responsive EBV⁺ tumors. In sum, anti-LAG3 therapy could effectively reduce tumor burden in refractory EBV⁺ GC patients.

Recent advances in spatial transcriptomes utilization in GC

While scRNA-seq can identify cell subpopulations within tissues, it lacks the ability to identify spatial distribution or capture local intercellular communication networks. Emerging technologies for ST can provide spatial information on different cell subpopulations and gene expression profiles in specific tissue regions, enabling a further understanding of the relationship between the spatial background of a particular cell and its functions [125, 126]. The mesenchymal phenotype in cancer is known to be associated with treatment resistance and a poor prognosis. Jang et al. [127] conducted ST analysis, indicating that intratumoral heterogeneity in GC may arise from an intermediate EMT status represented in the GC modules identified through intertumoral analysis. To our knowledge, Sun et al. were the first to utilize spatial multi-omics to investigate the immense TME in GC [10]. Detailed, they creatively proposed an integrated approach of spatial metabolomics (SM), spatial lipidomics (SL), and ST to hierarchically visualize the intratumor metabolic heterogeneity and cell metabolic interactions, these spatial multi-omics methods could effectively identify cell types and distributions within the intricate TME, and a region known as the “tumor-normal interface,” dominated by immune cells, exhibited unique transcriptional signatures and obvious immunometabolic changes when tumor cells came into contact with surrounding tissues. In a word, their approach not only facilitated the depiction of ITH for mapping molecular architecture of tissues, but also enhanced the comprehension of metabolism for GC on a systematic level. Further advancements in these spatial technologies will be crucial for mapping high-resolution cellular atlases from heterogeneous tumor ecosystems and for elucidating the molecular pathways involved in GC.

Summary and outlook

ITH poses a significant obstacle to successful cancer treatment and personalized medicine. Nevertheless, advancements in technologies like single-cell analysis and spatial pathologies are enabling researchers to better understand the intricate composition and changes within tumor ecosystem and therapeutic drug responses as the disease progression (Fig. 2). This knowledge is expected to uncover new predictive biomarkers and suitable therapy strategies, although the full integration of these finding into clinical practice is still pending. Here, this study delves into the applications and advancements of single-cell omics in uncovering the origin and progression of GC, pathogenesis, TME, and corresponding therapy strategies. In addition, the obvious heterogeneity between different GC patients appears to correlate with patient survival and response to immunotherapy, underscoring the critical need for a comprehensive understanding of TME alternations. We firmly believe that single-cell technologies could significantly facilitate the research in the field of GC. Firstly, scRNA-seq facilitates the identification of precancerous lesions that possess an elevated risk of progression to cancer from a transcriptional perspective. Secondly, single-cell technologies can capture the dynamic changes in the immune microenvironment and characterize the inflammatory infiltration at different stages of GC. Lastly, scRNA-seq can discover novel immune components serving as immunotherapy targets, especially for GC patients who do not respond to ICIs. Altogether, application of scRNA-seq in GC biology research has enhanced our comprehension of tumor heterogeneity, shed light on molecular mechanisms involved in tumor evolution and metastasis, and facilitated the identification of new therapeutic regimens for personalized cancer treatment. Despite its benefits, current scRNA-seq technology still faces limitations including strict sample quality standards, limited representation of primary tumors, and high costs. However, it is anticipated that these limitations will be overcome with time and technological progress, leading to increased utilization of scRNA-seq in the future.

Fig. 2.

Recent advances of GC studies using single-cell technologies. Various single-cell technologies, including scRNA-seq and ST, etc. , revealed recent progress in the field of GC in five aspects: GC origin and evolution, tumor heterogeneity, oncogenesis and metastasis, including LM, LNM, and PM, TME and drug treatment response such as chemotherapy, targeted therapy, immunotherapy, and a combination of these treatments. GC, gastric cancer; scRNA-seq, single-cell RNA sequencing; ST, spatial transcriptome; LM, liver metastasis; LNM, lymph node metastasis; PM, peritoneal metastasis; CAG, chronic atrophic gastritis; IM, intestinal metaplasia; SPEM, spasmolytic polypeptide-expressing metaplasia; TME, tumor microenvironment; GITR, glucocorticoid-induced tumor necrosis factor receptor; Tc17, IL-17⁺ CD8⁺ T cell; CXCL13, C-X-C motif chemokine ligand 13; Tex cell, exhausted T cell; MALT, mucosal-associated lymphoid tissue; CXCR4, C-X-C motif chemokine receptor 4; TIB, tumor-infiltrating B cell; TAM, tumor-associated macrophage; ACKR1, atypical chemokine receptor 1; CAF, cancer-associated fibroblast; iCAF, inflammatory CAF; eCAF, extracellular matrix CAF; CAF_EndMT, endothelial-to-mesenchymal CAF; PD-1, programmed cell death protein 1; XELOX, oxaliplatin + capecitabine; FOLFOX, leucovorin + 5-FU + oxaliplatin; NACT, neoadjuvant chemotherapy

The emergence of single-cell multi-omics technologies characterizes cell states and activities by integrating various omics methods that profile the genome, epigenome, transcriptome, proteome, and metabolome [128]. The advantages of scRNA-seq mainly capture the transcripts of individual cell and identify certain rare cell populations [129]. Whereas scDNA-seq has been widely used to detect single-cell copy-number aberration [130], single-cell assay for transposase-accessible chromatin sequencing (scATAC-seq) and single-cell ChIP sequencing (scChIP-seq) can identify heterogeneity of chromatin and explore epigenetically chromatin accessibility [131, 132]. Collectively, these integrated multi-omics methods not only comprehensively delineate genetic landscape of single cell, but also contribute to our comprehension of cellular heterogeneities and intricate interaction, leading to unravel numerous unsolved mysteries in GC.

Supplementary Information

Below is the link to the electronic supplementary material.

Abbreviations

ACRG

Asian Cancer Research Group

APB

Alternate promoter burden

AR

Androgen receptor

CAF

Cancer-associated fibroblast

CAF_EndMT

Endothelial-to-mesenchymal transition CAF

CAG

Chronic atrophic gastritis

CSC

Cancer stem cell

CTC

Circulating tumor cell

CTLA-4

Cytotoxic T-lymphocyte-associated antigen 4

CTS^high

Cathepsin^high

CXCL

C-X-C motif chemokine ligand

CXCR

C-X-C motif chemokine receptor

C1Q^high

Complement 1q^high

DC

Dendritic cell

DGC

Diffuse-type GC

DIC

Disseminated intravascular coagulation

DZIP1

DAZ interacting zinc finger protein 1

EBV

Epstein-Barr virus

eCAF

Extracellular matrix CAF

ECM

Extracellular matrix

EMT

Epithelial-to-mesenchymal transition

GC

Gastric cancer

GI

Gastrointestinal cancer

GITR

Glucocorticoid-induced tumor necrosis factor receptor

Gkn3

Gastrokine 3

GR

Glucocorticoid receptor

GSRCC

Gastric signet ring cell carcinoma

HAGC

Highly aggressive gastric cancer

ICI

Immune checkpoint inhibitor

ICB

Immune checkpoint blockade

iCAF

Inflammatory CAF

IFN-γ

Interferon gamma

IL

Interleukin

ILC2

Innate lymphoid cell type 2

IM

Intestinal metaplasia

ITH

Intratumor heterogeneity

IGC

Intestinal-type GC

ISG15

Interferon-stimulated gene 15

LM

Liver metastasis

LAG3

Lymphocyte activation gene 3

LNM

Lymph node metastasis

MDSC

Myeloid-derived suppressor cell

MALT

Mucosal-associated lymphoid tissue

MALAT1

Metastasis-associated lung adenocarcinoma transcript 1

MSI

Microsatellite instability

MSI-high

Microsatellite instability-high

NACT

Neoadjuvant chemotherapy

NK cell

Natural killer cell

PM

Peritoneal metastasis

PBMC

Peripheral blood mononuclear cell

PDGFRβ

Platelet-derived growth factor receptor beta

POSTN

Periostin

PD-L1

Programmed cell death protein ligand 1

PD-1

Programmed cell death protein 1

PDX

Patient-derived xenograft

scATAC-seq

Single-cell assay for transposase-accessible chromatin sequencing

scBCR-seq

Single-cell B cell receptor sequencing

scRNA-seq

Single-cell RNA sequencing

scTCR-seq

Single-cell T cell receptor sequencing

scCHIP-seq

Single-cell ChIP sequencing

SPEM

Spasmolytic polypeptide-expressing metaplasia

SL

Spatial lipidomics

SM

Spatial metabolomics

ST

Spatial transcriptome

TAN

Tumor-associated neutrophil

TAM

Tumor-associated macrophage

Tex

Exhausted T cell

TCGA

The Cancer Genome Atlas

TIB

Tumor-infiltrating B cell

TIGIT

T cell Ig and ITIM domain

TLS

Tertiary lymphatic structure

TMB

Tumor mutational burden

TME

Tumor microenvironment

Treg

Regulatory T cell

VEGFR

Vascular endothelial growth factor receptor

WES

Whole-exome sequencing

Author contributions

Jiao Xu conceptualized the study, collected the literature search, and drafted the manuscript. Jiao Xu and Bixin Yu made the tables and figures. Fan Wang and Jin Yang revised the manuscript. Fan Wang and Jin Yang were responsible for project administration and funding acquisition. All authors have read and agreed to the published version of the manuscript.

Funding

This study was supported by Gastric Cancer-National Major Disease Multidisciplinary Collaborative Diagnosis and Treatment Capacity Building Program (QT252), Cancer Precision Medical Science System and Service Platform Building-National Major Disease Multidisciplinary Collaborative Diagnosis and Treatment Capacity Building Program (QT264), National Natural Science Foundation of China (81902680), and Xi'an Science and Technology Association Young Talent Support Program (095920221304).

Data availability

No datasets were generated or analyzed during the current study.

Declarations

Conflict of interest

The authors declare no competing interests.

Ethical approval

Not applicable.

Consent for publication

Not applicable.