Molecular classification of small cell lung cancer subtypes: Characteristics, prognostic factors, and clinical translation

Abstract

Small cell lung cancer (SCLC) is a highly malignant tumor with a very poor prognosis; therefore, more effective treatments are urgently needed for patients afflicted with the disease. In recent years, emerging molecular classifications based on key transcription factors of SCLC have provided more information on the tumor pathophysiology, metastasis, immune microenvironment, and acquired therapeutic resistance and reflected the intertumoral heterogeneity of the various SCLC phenotypes. Additionally, advances in genomics and single-cell sequencing analysis have further revealed the high intratumoral heterogeneity and plasticity of the disease. Herein, we review and summarize these recent lines of evidence and discuss the possible pathogenesis of SCLC.

Introduction

Lung cancer is one of the most common malignant tumors that affect humans worldwide, with small cell lung cancer (SCLC) accounting for approximately 15% of all lung cancers.^[1] Occurring frequently in smokers,^[2] SCLC is characterized by its high malignancy, invasiveness, easy recurrence, and metastasis, and has a 5-year survival rate of less than 7%.^[3] Platinum-based chemotherapy and radiation therapy remain the first-line treatments for patients with SCLC,^[4] and there have been no significant improvements in survival and therapeutic approaches for more than 30 years.^[5] Although SCLC is initially highly sensitive to chemotherapy, with a first-response rate of up to 75%–80%, the development of resistance is essentially a universal occurrence.^[6]

In recent years, a breakthrough in the treatment of SCLC has come in the form of immunotherapy, especially with immune checkpoint inhibitors. Nivolumab and pembrolizumab, which are anti-programed cell death protein 1 (PD1) monoclonal antibodies, were granted accelerated approval by the US Food and Drug Administration for the third-line treatment of patients with metastatic SCLC.^[7–9] Both atezolizumab and durvalumab are recommended for use in combination with chemotherapy for the first-line treatment of patients with extensive-stage SCLC in National Comprehensive Cancer Network (NCCN) Guidelines for Small Cell Lung Cancer Version 1.2021.^[10] However, because patients with SCLC generally have a poor overall response to immune checkpoint inhibitors and the drugs therefore have limited survival benefit, clinical research on new treatments for SCLC is still urgently needed.

As a highly heterogeneous disease, the subtypes of SCLC are defined by their differential expression of transcriptional regulators, such as achaete-scute homolog 1 (ASCL1), neurogenic differentiation factor 1 (NEUROD1), yes-associated protein 1 (YAP1), and POU class 2 homeobox 3 (POU2F3). At the same time, exploration of the intratumoral heterogeneity of SCLC is also expected to reveal its mechanisms of drug resistance and evolution, which may help in the search for new therapeutic targets. In this paper, we review recent advances in the study of the classification and translational research of SCLC as well as the elucidation of its intra- and intertumoral heterogeneity. A better understanding of these biological features may yield new therapeutic targets and combination drug strategies for the treatment of SCLC in the future.

Molecular Classification of SCLC and the Intertumoral Heterogeneity of Its Phenotypes

The first classification model of SCLC, which was proposed in 1986, divides the disease into the following two subtypes based on cytological characteristics: "classical SCLC," which grows in the form of floating spherical aggregates in cell culture, with or without central necrosis; and "variant SCLC", which grows as loosely adherent aggregates or tightly adherent monolayers.^[11] In 2015, the World Health Organization attributed SCLC to one of four tumors with a neuroendocrine (NE) morphology, the key features of which were the following: a dense cell arrangement (palisade, chrysanthemum conglomerate, or trabecular), a unique cell morphology (scant cytoplasm, dense core granules, and lack of significant nucleoli), high grade (high proliferation rate and very high Ki-67 proliferation index [50%–100%]), and positive NE markers (including synaptophysin, chromogranin A [CgA], neural cell adhesion molecule 1 [NCAM1], and insulinoma-associated protein 1 [INSM1]).^[12] However, a few SCLC specimens were negative for all standard NE markers and therefore could not be correctly classified.

Gene expression studies have revealed that some SCLC cells express high levels of CgA, gastrin-releasing peptide (GRP), ASCL1, and protein delta homolog 1 (DLK1) (thought to represent tumors with low NOTCH signaling pathway activity) and have thus been named "NE SCLCs". By contrast, other SCLC cells lack the expression of these NE markers and are therefore called "non-NE SCLCs".^[13]

The roles of transcription factors in shaping tumor behavior and regulating the various different NE patterns of SCLCs have recently been elucidated. ASCL1 and NEUROD1 are considered to be key transcription factors affecting the development and maturation of pulmonary NE cells.^[14,15] The ASCL1 gene, which is expressed in approximately 75% of SCLCs, controls several crucial cellular mechanisms, including cell growth and survival.^[16] SCLC cell lines that express high levels of ASCL1 grow as loosely adherent clusters in a "variant" morphology.^[17] NEUROD1 is expressed in approximately 24% of human tumor specimens of SCLC.^[18] The morphology of NEUROD1-expressing SCLC cells ranges from "classical" to "variant" forms, growing loosely in adherent monolayers.^[15] Borromeo et al^[15] found that ASCL1 and NEUROD1 distinguished heterogeneity in SCLC with distinct genomic landscapes and gene expression profiles, and the tumors could be divided into ASCL1^high, NEUROD1^high, and "double-negative SCLC" subtypes (with low levels of both ASCL1 and NEUROD1 expression). The use of clustered regularly interspaced short palindromic repeats (CRISPR) gene editing technology for the screening of important proteins that affect tumor cell growth revealed that the transcription factor POU2F3 was overexpressed in only a few tumor specimens of SCLC that expressed NE markers at a low level. POU2F3 was also demonstrated to be a major regulator of tuft cells, a variant of non-NE SCLC cells.^[19]

The MD Anderson Cancer Research Center in the United States recognizes four biologically distinct clusters among 81 resected SCLC tumor samples and 62 SCLC cell lines. Three of the clusters are defined almost solely by their differential expression of ASCL1 (SCLC-A, 36%), NEUROD1 (SCLC-N, 31%), and POU2F3 (SCLC-P, 16%), and the remaining cluster is defined by the absence of all three (SCLC-inflamed/mesenchymal [SCLC-I/IM], 17%).^[20] By studying the expression of ASCL1, NEUROD1, YAP1, and POU2F3 in 81 SCLC tumor samples and 54 SCLC cell lines, the following four subtypes of non-NE SCLC: SCLC-A type (70%), SCLC-N type (11%), SCLC-P type (10%), and SCLC-Y type (2%) were defined.^[21] However, in another study of samples from 174 patients with SCLC, the comprehensive immunohistochemical and histopathological analyses revealed that YAP1 was expressed at low levels and primarily in combined SCLC variants, and its expression was not exclusive of other subtypes.^[22]

It was shown in a recent study that the level of hes family bHLH transcription factor (HES1) expression could be used to identify a population of long-term tumor-propagating cells (TPCs) of SCLC,^[23] which also showed higher levels of NE differentiation than non-TPCs.^[24] Wooten et al^[25] classified SCLCs into four subtypes based on differential HES1 expression levels and NE, non-NE, and NEv1 marker expression. Three of them corresponded to the known subtypes SCLC-A, SCLC-Y, and SCLC-N, respectively, whereas the fourth was a previously undescribed ASCL1 + NE variant (designated NEv2 or SCLC-A2).^[25]

Simpson et al^[26] described a biobank of 38 circulating-tumor-cell-derived explant (CDX) models, the transcriptomic analysis of which confirmed three of four previously described subtypes based on ASCL1, NEUROD1, and POU2F3 expression and identified a previously unreported subtype based on another NE transcription factor, atonal BHLH transcription factor 1 (ATOH1). A high prevalence of ATOH1 in these CDX models reflects their potential as a circulating tumor cell source for in vivo clonal selection and expansion. However, according to other studies, ATOH1 is present in only approximately 1% (1/81) of surgically resected SCLC samples.^[13,18] At present, there are different opinions on the classification of SCLC, and thus more data are needed to confirm these findings. Figure 1 summarizes the different classifications of SCLC mentioned in the studies to date.

Figure 1.

Molecular subtypes of small cell lung cancer. ASCL: Achaete-scute homolog; ATOH1: Atonal BHLH transcription factor 1; NE: Neuroendocrine; NEUROD1: Neurogenic differentiation factor 1; POU2F3: POU class 2 homeobox 3; SCLC: Small cell lung cancer.

Key Genetic and Molecular Pathways of SCLC

Whole-genome analysis has revealed two key tumor suppressor genes (TP53 and retinoblastoma 1 [RB1]) that are commonly mutated and inactivated in human SCLC cells (>90% and ~65%, respectively).^[13,27–29] In a study of the molecular characterization of NE carcinoma in a Chinese population, mutations of both TP53 and RB1 occurred in 83% (29/35) of patients with SCLC.^[30]

The Hedgehog (HH) signaling pathway regulates the fate and self-renewal of cancer cells.^[31] The activation of this signaling pathway has also been detected in SCLC cells.^[32] Constitutive activation of the HH signaling molecule Smoothened (Smo) promoted the clonogenicity of human SCLC cells in vitro and the initiation and progression of the disease in mice in vivo.^[33] These results suggest that the HH pathway may be a therapeutic target in SCLC.

The transcription factor INSM1 is also a sensitive and specific marker of NE neoplasms and functions as an NE differentiation factor.^[34] Because 93% of SCLCs have been found to be positive for INSM1,^[35] this transcription factor has emerged as a sensitive and specific biomarker for the disease.^[36]

Through the use of in vitro and mouse models, it has been validated that loss of function of the tumor suppressor phosphatase and the tensin homolog deleted on chromosome 10 (PTEN),^[37] NOTCH receptors,^[38] and the chromatin regulator CREB-binding protein (CREBBP)^[39] can also lead to SCLC. The frequent amplification of myelocytomatosis oncogene cellular homolog (MYC) family proto-oncogenes (MYC, MYCL, and MYCN) has also been observed in SCLC.^{[13,27,40,41]} Members of the MYC gene family are associated with high-frequency mutations in epigenetic regulators (e.g., histone lysine acetyltransferases eP300 and CREBBP, and histone methyltransferase KMt2D) and inactivating mutations of NOTCH family genes.^[42] The ASCL1^high subtype of SCLC is highly associated with the expression of MYCL,^[17] whereas the NEUROD1^high and POU2F3^high subtypes are related to the upregulation of MYC.^[19,43,44] Aurora kinase (AURK) is a main regulatory serine/threonine kinase of mitosis and provides a growth advantage in SCLC when overexpression is accompanied by MYC family amplification.^[41,45,46]

Epithelial–Mesenchymal Transition and Drug Resistance in SCLC

SCLC cells, which often have a round or wattle-shaped, immature, and small appearance, have weak adhesive properties and reduced E-cadherin expression levels. Moreover, they exhibit morphological characteristics typical of epithelial–mesenchymal transition (EMT), which has been implicated in the tumor cell resistance to treatments.^[47,48]

The NOTCH signaling and ASCL1 expression status of SCLC cells may determine their resistance to treatments by modifying the EMT phenotype. The ASCL1 gene drives the expression of many oncogenes, such as SRY-box 2 (SOX2), MYCL, and B-cell lymphoma 2 (BCL-2), and is associated with EMT in SCLC.^[15,16] The NOTCH signaling pathway plays an inhibitory role in SCLC tumorigenesis,^[49] with the majority of cancerous cells exhibiting low NOTCH pathway activity with a concomitant high expression level of Delta-like 1 homolog (DLK1), a non-canonical NOTCH ligand.^[50] NOTCH signaling can enhance cell adhesion and inhibit the EMT process in SCLC cells by inducing the production of E-cadherin. It can also repress ASCL1 expression at both the transcriptional and post-transcriptional levels.^[51] Approximately 25% of SCLC tumors have an inactivation mutation of the NOTCH signaling pathway,^[13] which has also been associated with EMT. Chemotherapy and radiation therapy can inactivate NOTCH signaling, thereby inhibiting NE differentiation of the SCLC cells and promoting the EMT process.^[42]

Neuropeptides Associated with SCLC

SCLC cells have the capacity to produce neuropeptides for promoting their survival and proliferation as well as for communicating with the microenvironment in autocrine, paracrine, and endocrine manners. Both ASCL1-dominant and NEUROD1-dominant SCLC subtypes are associated with an NE marker^high profile, whereas POU2F3 and other ASCL1/NEUROD1 double-negative subtypes are NE marker^low.^[22]

The NE function of SCLC is mainly related to tumor growth and metastasis. It is possible that the migration potential of SCLC cells is inherently related to the striking migratory phenotype of NE cells during lung development.^[52] c-Kit and its natural ligand, stem cell growth factor (SCF), are co-expressed in approximately 40–70% of tumor specimens and cell lines of SCLC,^[53] where the SCF/c-Kit pathway is functional in an autocrine or a paracrine fashion.^[54] SCLC cells also express a high level of thyroid transcription factor-1 (TTF-1), which regulates the expression of the BCL-2 gene family and can partly coordinate with ASCL1 to promote growth of the tumor cells and contribute to their NE and antiapoptotic gene expression.^[55] The chemoattractant stromal cell-derived factor-1 (SDF-1) and its receptor C-X-C motif chemokine receptor 4 (CXCR4) are key modulators of cancer development and crosstalk between tumors and their microenvironment.^[56] CXCR4 is highly expressed in SCLC cells,^[57] where it cooperates with c-Kit to enhance cell motility and metastasis as well as induce the phosphorylation of viability-signaling molecules, such as Akt and p70 S6 kinase.^[58] It has been shown that the transcription factor polyomavirus enhancer activator 3 (Pea3) plays a critical role in facilitating the metastasis of SCLC.^[59] In an intravenous transplantation model, paracrine signaling between NE and non-NE subclones of SCLC via fibroblast growth factor 2 (FGF2) and mitogen-activated protein kinase (MAPK) caused enhanced expression of Pea3, which then resulted in metastatic dissemination of the NE tumor subclones.^[59]

Bombesin (BOM), neuromedin B (NMB), and GRP, which are all neuropeptides produced by SCLC cells, can inhibit the maturation of dendritic cells in a dose-dependent manner. BOM and GRP also inhibit the production of interleukin-12 (IL-12) by dendritic cells and thus reduce their ability to activate T cells.^[60] This suggests that the paracrine function may be a mechanism of tumor immune escape and new target for the treatment of SCLC could be discovered upon paracrine function related pathways.

Neural autoantibodies, such as antineuronal nuclear antibody-type 1 (ANNA-1) and N-type voltage-gated calcium channel (VGCC-N) autoantibodies, are commonly found in the serum of patients with SCLC with neurological manifestations,^[61,62] reflecting the presence of mutated cancer autoantigens.^[63] These patients tend to have a better prognosis than those without neurological complications.^[62,64–66] The comorbidity of endocrine paraneoplastic syndromes in patients with SCLC suggests the existence of communication between SCLC cells and other cells in the body, which is a phenomenon that has yet to be fully studied.^[67]

Tumor Evolution and Intratumoral Heterogeneity of SCLC

The fact that SCLC tumors can be assigned to different molecular subtypes based on their differential expression of transcription factors suggests that they have significant intertumoral heterogeneity. However, some studies have indicated that different molecular subtypes of SCLC may exist in a single tumor, which implies the intratumoral heterogeneity of the disease.^[26,68] For example, frequent dual-high expression (19.8%) of ASCL1 and NEUROD1 was found in a small set of formalin-fixed paraffin-embedded samples (N = 81) of human SCLC tissue.^[18] In another study with a larger sample set (N = 174), significant co-expression of ASCL1 and NEUROD1 was found in 37% of the specimens, with 22% exhibiting dual-high expression (H-score >50 for both markers).^[22] The apparent difference in transcription factor expression levels among different studies may be due to the different types of tissues being analyzed (cell cultures, resection specimens, biopsies, genetically engineered mouse models [GEMMs]), which in turn reflects the difference and variability in tumor compositions. Samples derived from animal and pathological models of SCLC are generally more homogeneous than their native counterparts. For example, CDX models derived from SCLC patient samples and RNA sequencing data were used to explore the co-expression of subtype-defining transcription factors, whereupon fewer than 1% of cells in any of the models were found to express POU2F3; and of these rare POU2F3-positive cells, all exhibited ASCL1 co-expression.^[20] Furthermore, this study also showed an ASCL1/NEUROD1 co-expression rate of approximately 10% in one of the models.^[20] Overall, although most SCLC tumors express only one of these transcription factors, the expression is not mutually exclusive. It is worth emphasizing that SCLC is not just a heterogeneous tumor, but this heterogeneity should underlie different tumors of different origins and/or pathogenesis.

Some reports have also confirmed the impact that intratumoral heterogeneity may have on plasticity transformation, or subtype switching, as mechanisms of the therapeutic resistance of SCLC.^[23] NE SCLCs can transform into non-NE tumors (e.g., by MYC amplification), and these two components are likely to coexist within the same tumor, thus supporting the different susceptibility of these cell compartments to therapy.^[68] NE cells are fast growing and prone to metastasis, whereas the non-NE component(s) account for chemoresistance, with the traits often connecting over time, confirming the profoundly heterogeneous nature of these tumors.^[41,69]

The development of GEMMs through targeted alteration of the mouse tumor suppressors Rb and transformation-related protein 53 (Trp53), which are also present in almost all human SCLC cells, has provided an invaluable preclinical platform for identifying and characterizing the mechanisms of SCLC tumorigenesis and progression.^[13,27–30] These GEMMs have a high tumor incidence and are highly similar to human SCLC tumors.^[70] One such model has demonstrated the predominance of ASCL1 in early lesions with Trp53/Rb1/Rbl2 deficiency, suggesting a possible hierarchy where SCLC-A is a precursor of SCLC-N.^[15,41] In vitro studies with SCLC cell lines have shown that surface marker profiles of the disease can shift upon drug treatment, indicating a phenomenon of tumor evolution under subtype-specific therapeutic selection.^[71]

However, the difficulty in obtaining sufficient biopsies that are suitable for pathological diagnosis and subsequent study is a common problem in the field of SCLC research. Tools for the preclinical study of malignant diseases, such as human cancer cell lines and derived xenografts, GEMMs, patient-derived xenografts, or cell-derived xenografts, are expected to provide us with more evidence on the evolutionary history of SCLCs in the future.

Treatment Strategies and Prognostic Biomarkers for Different SCLC Subtypes

Despite the significant intratumoral heterogeneity of SCLC, all patients with this disease are currently treated with first-line platinum-based chemotherapy, radiation therapy, and immunotherapy.^[72] Advances in molecular typing and mechanistic research at the genetic level may provide new therapeutic strategies for patients with SCLC.

SCLC-A

SCLC-A, an NE epithelial tumor with an ASCL1^high/c-MYC^low profile, is targetable by BCL-2 inhibitors (e.g., venetoclax, nativoclax),^[55,73] which can block the growth of the tumor and induce its regression in mice bearing the high BCL-2-expressing subtype. Preclinical studies have shown that BCL-2 expression levels may be an effective predictive biomarker of a tumor's sensitivity to treatment.^[74,75] However, BCL-2 inhibitor monotherapy has not shown a significant clinical impact.^[74] Receptor tyrosine kinase-like orphan receptor 1 (ROR1) is an oncofetal protein that is co-expressed with BCL-2 in multiple tumor types, including SCLC. Preclinical studies have shown that ROR1 inhibition (KAN0441571C) occurs synergistically with BCL-2 inhibition in SCLC models, thus presenting a novel therapeutic target for the disease.^[76] The bromodomain and extra-terminal (BET) proteins have been shown to regulate the expression of key genes in oncogenesis, such as MYC, cyclin D2 (CCND2), and BCL2-like protein 1 (BCL2L1). Co-treatment of SCLC tumors with a BET inhibitor and the BCL-2 inhibitor venetoclax resulted in strong synergetic antitumor effects both in vitro and in vivo.^[77] Additionally, anthracyclines (e.g.,doxorubicin) and cyclin-dependent kinase 9 (CDK9) inhibitors (e.g., dinaciclib) were also shown to act synergistically with a BCL-2 inhibitor (e.g., venetoclax) in downregulating myeloid leukemia-1 (MCL-1); therefore, the combinations of venetoclax with doxorubicin or dinaciclib provide yet another effective therapeutic strategy against SCLC growth and proliferation.^[78]

Inactivating mutations in members of the NOTCH family and the abnormally high expression of delta-like 3 protein (DLL3), an atypical NOTCH ligand, are also often observed in SCLC-A tumors.^[79] Expression of the NOTCH1 receptor in SCLC serves as a clinically important prognostic factor.^[80] Rovalpituzumab tesirine (Rova-T), an antibody–drug conjugate that targets DLL3, had shown encouraging single-agent antitumor activity in phase I/II studies on patients with recurrent SCLC and high DLL3 expression, despite the occurrence of serious adverse events.^[81,82] However, a subsequent phase III clinical study showed that compared to that with topotecan, Rova-T resulted in a poorer overall survival rate and higher rates of serosal effusions, photosensitivity reactions, and peripheral edema, which led to the discontinuation of patient enrollment.^[83] Lysine-specific histone demethylase 1A (LSD1, also known as KDM1A) binds to the NOTCH1 locus, resulting in the suppression of NOTCH1 expression and its downstream signaling events. In SCLC models, LSD1 inhibitors were shown to activate the NOTCH pathway, thereby suppressing ASCL1 expression and the repression of tumorigenesis.^[38] However, in a phase I study (NCT02034123), GSK2879552 (a potent selective LSD1 inhibitor) provided poor disease control and a high adverse event rate in patients with SCLC, and the study was subsequently terminated as the risk–benefit profile did not favor its continuation.^[84]

In another study, the quantitative secretome analysis of 13 cell lines representing different NE lung cancer subtypes showed that ASCL1 regulates insulin-like growth factor binding protein 5 (IGFBP5) transcription directly by binding to its E-box element.^[85] The study also identified IGFBP5 as a secreted marker for ASCL1^high SCLC in vitro and in mouse models, rendering it a potential drug target for SCLC.^[85] SCLC-A also expresses a high level of Schlafen 11 (SLFN11) in a bimodal manner which has prognostic predictive value for a variety of cancer treatments (etoposide, topotecan, carboplatin, temozolomide/veliparib, and poly-(ADP)-ribose polymerase [PARP] inhibitors).^[48,86,87]

SCLC-N

SCLC-N is an NE MYC^high tumor with susceptibility to AURK inhibitors (AURKi).^[73,88] The A-type AURK stabilizes the MYC family of oncoproteins, which have long been considered an undruggable target. It has been confirmed that AURKi can potently inhibit the proliferation of high-MYC-expressing SCLC cell lines.^[88] In a phase I/II study, single-agent treatment with alisertib, an investigational AURKi, showed an objective response rate of 21% (10/48) in patients with SCLC.^[89] In another phase I study, the combination of alisertib with nab-paclitaxel proved to be synergistically active against rapidly proliferative high-grade NE tumors, especially SCLC, and displayed a manageable side-effect profile.^[90] In a phase II study, promising preliminary efficacy was seen with combination alisertib/paclitaxel therapy in relapsed or refractory SCLC, with the drug combination resulting in a median progression-free survival period of 3.32 months vs. 2.17 months with paclitaxel alone (hazard ratio [HR] = 0.71, P = 0.038).^[91] Moreover, the c-MYC expression level and mutations in cell cycle regulators may be potential predictive biomarkers of alisertib efficacy.^[91] These data support the important need for further clinical assessments of alisertib in patients with SCLC. Another potential strategy against SCLC-N involves Seneca Valley Virus (SVV-001), an NE cancer-selective oncolytic picornavirus that has shown selective efficacy in primary heterotransplant mouse models of NEUROD1^high SCLC.^[92]

SCLC-P

SCLC-P, a non-NE epithelial tumor, accounts for approximately 7% of all SCLC cases.^[22] This tumor subtype is vulnerable to poly (ADP-ribose) polymerase (PARP) inhibitors and anti-metabolites, including anti-folates and nucleoside analogs.^[20] The use of small-molecule inhibitors may also be a future direction in the research and development of drugs that specifically target unique dependencies in POU2F3-expressing SCLC cell lines, such as the transcription factors SRY-box transcription factor (SOX) 9 and ASCL2 and the receptor tyrosine kinase insulin-like growth factor 1 receptor (IGF1R).^[19] Additionally, the SCLC-P subtype seemed to be a poor prognostic marker for immunotherapy as it does especially poorly relative to the other three subtypes in the subtype-by-subtype analysis comparing survival between atezolizumab vs. placebo in the cohort of IMpower133.^[20]

SCLC-Y

SCLC-Y, a non-NE inflamed phenotype that has a better-differentiated tumor histology and an inflamed tumor microenvironment, accounts for 5%–10% of SCLC cases.^[93] SCLC-Y cell lines are distinct among the SCLC subtypes, bear transcriptome similarity to that of NSCLC, and are associated with shorter patient survival and increased chemoresistance.^[46,94] This subtype is also associated with high expression levels of YAP1, interferon-gamma response genes, T-cell receptor genes, human leukocyte antigen (HLA) genes, and antigen-presenting machinery genes; high T-cell-inflamed gene expression profile scores; and low levels of cancer testis antigens, RB1 mutations, and ASCL1 expression.^[93,95] Other characteristics of SCLC-Y cells include their high expression of the ephrin type-A receptor (EPHA2), mesenchymal marker vimentin (VIM), cytoskeleton component and regulator calponin 2 (CNN2), and NOTCH pathway genes.^[95] In addition, it has been shown that YAP1 can upregulate programmed death-ligand 1 (PD-L1) expression levels and induce an immunosuppressive tumor microenvironment,^[96,97] which suggest that the SCLC-Y subtype may be more likely to benefit from immunotherapy, especially immune checkpoint inhibitors. According to the results of gene expression and in silico studies, SCLC-Y tumors are also likely vulnerable to inhibitors targeting mammalian target of rapamycin (mTOR), polo-like kinase 1 (PLK1), YAP1, and NOTCH1.^{[51,93,97,98]}

SCLC-I

SCLC-I, a non-NE mesenchymal inflamed phenotype of SCLC, is characterized by its high expression of inflammatory markers, immune checkpoint proteins, and stimulator of interferon genes (STING)-related genes, with greater benefit for its suppression being derived from immune checkpoint inhibitors.^[30] In the IMpower133 study, patients with SCLC-I tumors, who constituted 18.5% of all participants, showed a significantly higher overall survival rate than those with other SCLC subtypes in the atezolizumab vs. placebo in addition to chemotherapy trial (HR = 0.57; 95% CI: 0.32–0.99).^[20,99] However, there is insufficient evidence for the SCLC-I to be a prognostic maker of immunotherapy as the addition of atezolizumab produces more modest gains in median OS in SCLC-A and -N than in SCLC-I.^[20] As the most of the clinical trials of immunotherapy were not designed for different subtypes of SCLC, future clinical studies are still needed to confirm whether the molecular classification of SCLC can be used as a prognostic marker for immunotherapy. SCLC-I cells also show a high expression of Bruton's tyrosine kinase (BTK), rendering them potentially sensitive to BTK inhibitors, such as ibrutinb.^[20] Table 1 summarizes the gene expression profile and corresponding treatment strategies of different subtypes of SCLC.

Table 1.

Summary of gene expression profile and corresponding treatment strategies of subtypes of SCLC.

Major transcriptional factors	Associated subtype and tumor behavioral characteristics of SCLC	Related genes	Potential therapeutic approaches
ASCL1	SCLC-A; NE tumor, "classic" form, EMT	Upregulate SOX2, MYCL,BCL-2, DLK1, INSM1, E-cadherin, DLL3, TTF1 Downregulate NOTCH signaling, MYC	BCL-2 inhibitors (venetoclax, nativoclax); anthracyclines (e.g., doxorubicin); CDK9 inhibitors (e.g., dinaciclib); ROR1 inhibitor (KAN0441571C); BET inhibitor; DLL3 inhibitors (rovalpituzumab tesirine [Rova-T]); LSD1 inhibitors (NCT02034123)
NEUROD1	SCLC-N; NE tumor, "variant" form, growing loosely in adherent monolayers	Upregulate MYC, AURK, HES6,INSM1; Downregulate E-cadherin	AURK inhibitors, Seneca Valley Virus (SVV-001)
POU2F3	SCLC-P; Non-NE tumor	Upregulate MYC, AURK, E-cadherin, SOX9, ASCL2, IGF1R	PARP inhibitors, anti-metabolites (including anti-folates and nucleoside analogs), IGF1R inhibitors
YAP1	SCLC-Y; non-NE tumor, associated with increased chemoresistance, high expression of immune suppressive microenvironment-related markers	Upregulate YAP1, MYC, interferon-gamma response genes, T-cell receptor genes, HLA genes, AURK, EPHA2, VIM, CNN2, NOTCH signaling, PD-L1, mTOR, PLK1, RB1 Downregulate MYC	Immune checkpoint inhibitors, mTOR inhibitors, PLK1 inhibitors, YAP1 inhibitors
ASCL1–/NEUROD1–/POU2F3–	SCLC-I/IM; Inflamed/mesenchymal, EMT	Upregulate BTK	Immune checkpoint inhibitors, BTKi inhibitors (ibrutinib)

ADP: Adenosine diphosphate; ASCL: Achaete-scute homolog; AURK: Aurora kinase; BCL-2: B cell lymphoma protein-2; BET: Bromodomain and extra-terminal; BTK: Bruton's tyrosine kinase; CDK9: Cyclin-dependent kinase 9; CNN2: Calponin 2; DLK1: Delta-like homolog 1; DLL3: Delta-like 3 protein; EMT: Epithelial–mesenchymal transition; EPHA2: Ephrin type-A receptor 2; HES6: Hes family bHLH transcription factor 6; HLA: Human leukocyte antigen; IGF1R: Insulin-like growth factor 1 receptor; INSM1: Insulinoma-associated protein 1; LSD1: Lysine-specific histone demethylase 1A; mTOR: Mammalian target of rapamycin; MYC: Myelocytomatosis oncogene cellular homolog; NE: Neuroendocrine; NEUROD1: Neurogenic differentiation factor 1; PARP: Poly-(ADP)-ribose polymerase; PD-L1: Programmed death-ligand 1; PLK1: Polo-like kinase 1; POU2F3: POU class 2 homeobox 3; RB1: Retinoblastoma 1; ROR1: Receptor tyrosine kinase-like orphan receptor 1; SCLC: Small cell lung cancer; SOX: SRY-box transcription factor; TTF1: Thyroid transcription factor-1; VIM: Vimentin; YAP1: Yes-associated protein 1.

Summary and Future Prospects

Throughout the progresses made in the field of SCLC research, immunotherapy has not only gradually changed the treatment pattern for this disease but also opened up the possibilities of finding the appropriate patient population, ideal treatment timing, and optimal treatment strategies. With the ongoing study of the disease pathogenesis and the application of new treatment methods, such as immunotherapy, the classification of SCLC is also being continuously updated and integrated with clinical practice. However, the study of molecular subtypes of SCLC at the genetic level belongs to different dimensions with different treatment types (such as surgery, chemotherapy, immunotherapy, etc.), and most clinical trials of immunotherapy have not considered the molecular subtypes of SCLC at the time of design. Whether molecular subtypes of SCLC can be used as a basis for immunotherapy remains to be revealed by more clinical study data. Such advances in knowledge acquisition will lead to better analyses of the biological behavior and clinical characteristics of the various SCLC subtypes, which will more accurately guide their treatment and the evaluation of their prognosis, thereby improving the survival of patients with SCLC in the future.

SCLC is a highly heterogeneous disease. The goal of classifying the various SCLC subtypes is to elucidate the nature of the malignancy. However, any classification model is staged and cannot picture all the characteristics of SCLC. Despite the fact that the molecular-typing classification of SCLC based on ASCL1, NEUROD1, YAP1, and POU2F3 expression levels has reached uniform consensus, some subtypes still cannot be classified. Such classification also poses new problems, and the associations of these subtypes with the tumor stage, metastatic potential, and immune microenvironment remain unknown. The elucidation of these relationships may have prognostic and therapeutic implications and therefore warrant further clinical investigation.

Funding

This work was supported by grants from the Jilin Scientific and Technological Development Program (CN) (No. 20190303146SF) and General Program of National Natural Science Foundation of China (No. 81874052).

Conflicts of interest

None.