SCLC: Epidemiology, Risk Factors, Genetic Susceptibility, Molecular Pathology, Screening, and Early Detection

Abstract

We review research regarding the epidemiology, risk factors, genetic susceptibility, molecular pathology, and early detection of SCLC, a deadly tumor that accounts for 14% of lung cancers. We first summarize the changing incidences of SCLC globally and in the United States among males and females. We then review the established risk factor (i.e., tobacco smoking) and suspected nonsmoking-related risk factors for SCLC, and emphasize the importance of continued effort in tobacco control worldwide. Review of genetic susceptibility and molecular pathology suggests different molecular pathways in SCLC development compared with other types of lung cancer. Last, we comment on the limited utility of low-dose computed tomography screening in SCLC and on several promising blood-based molecular biomarkers as potential tools in SCLC early detection.

Introduction

Lung cancer is the leading cause of cancer mortality globally, contributing an estimated 130,180 deaths, or up to 21% of all cancer-related deaths in the United States in 2022.^1,2 There are approximately 2.2 million incidental lung cancer cases worldwide in 2020 and 227,875 in the United States, with SCLC constituting approximately 14% of them.^1,3 Smoking accounts for more than 95% of SCLC cases.⁴ The risk of SCLC remains poorly characterized at the genetic level, and certain germline variants may be actionable in SCLC treatment. In addition, despite the universal loss of TP53 and RB1 genes, SCLC is a heterogeneous disease, and its molecular complexity is responsible for the resistance to current therapeutic strategies.⁵ Lung cancer screening with low-dose computed tomography (LDCT) reduce lung cancer-specific mortality by up to 20%; however, the benefit is almost exclusively driven by NSCLC.⁶ Emerging blood-based biomarkers have been developed and evaluated for early lung cancer detection with several having promising results. Nevertheless, whether these biomarkers will have an impact on cancer control at a population level, especially for a cancer with aggressive biological behavior such as SCLC, remains unknown. Smoking cessation is the most effective method in reducing lung cancer burden, particularly for SCLC. Continued global and regional efforts in tobacco control are needed. Therefore, we aim to provide an up-to-date review of epidemiology, prevention (including screening), genetics, and molecular pathology of this important cancer, with the ultimate goal of contributing to reducing its disease burden.

Epidemiology

Lung cancer is the most common cancer in males with an estimated 1.4 million incident cases globally and the third most common cancer in females with 0.77 million cases in 2020.¹ It continues to be the leading cancer-related death for males and second for females.¹ The SCLC accounts for approximately 14% of all lung cancer cases.³ Approximately 250,000 patients are diagnosed with having SCLC each year globally, of which approximately 200,000 succumb to the disease.⁷

Changes in smoking habits have historically influenced lung cancer incidence and mortality, which usually lags 20 to 30 years behind the trend in smoking prevalence.⁸ Since the report of the Surgeon General’s Advisory Committee in 1964 and the deployment of tobacco control programs in many countries, smoking prevalence in males has decreased substantially in many countries (except Eastern Europe, Southeast Asia, People’s Republic of China, Spain, and several countries in Latin America) from 1970 to 2000. Consequently, there is a marked decrease in male lung cancer incidence in these regions.^9,10 In comparison, the smoking prevalence in females was higher in North America, Oceania, Latin America, and Western Europe than other parts of the world. By 2000, smoking prevalence decreased in several countries (including the United States) but remained high in parts of Europe, such as Spain, France, Switzerland, Poland, Hungary, and Chile, where lung cancer incidence continues to rise.^9,10 Worldwide, for SCLC-specific incidence, the city of Izmir in Turkey reported the highest incidence at 12.4 per 100,000 in males whereas New Zealand, Māori reported the highest incidence in females at 14.5 per 100,000.¹¹ In the United States, according to the National Cancer Institute Surveillance, Epidemiology, and End Results program (using Joinpoint Trend Analysis), the age-adjusted incidence of SCLC in males increased rapidly from 1975 to 1978 with an annual rate of 10.5% (Fig. 1) and started to decrease steadily since 1998 (annual change of 3.4%).^12–14 In females, the incidence climbed dramatically from 1975 to 1982 with an annual rate of 8.4% and continued to rise at a lower rate until 1989. It started to decrease afterward but at a slower rate than males.¹² As a consequence, the initial strong male predominance of SCLC has changed from 68.3% of males in 1975 to slightly lower than females (47.4%) in 2019.

Figure 1.

Long-term trends in SEER age-adjusted incidence rates, SEER 8 (1975–2019) by sex.¹² Figure source: National Cancer Institute SEER Cancer Statistics Explorer Network.

The rate of lung cancer overall is notably higher in males of African ancestry than in those of European ancestry (67.1 versus 60.9 per 100,000 males), with a higher rate of mortality (48.9 versus 42.4 per 100,000 males) in 2019.¹² Nevertheless, in contrast to NSCLC, individuals of African ancestry may be at a lower risk of developing SCLC relative to those of European ancestry (5.2 versus 6.4 per 100,000 in 2019).¹² Furthermore, African-ancestry patients with SCLC, despite facing disadvantages in accessing cancer care, have better 5-year survival compared with European-ancestry patients (7.8% versus 5.8%, respectively), according to the Surveillance, Epidemiology, and End Results data from 2012 to 2018.^12,15 Genetic studies would contribute to understand such ancestry-based disparities in risk for SCLC.^16,17 In addition, the incidence of SCLC is higher among people with low socioeconomic status (SES), which parallels to the patterns of tobacco consumption.¹⁸ A recent large pooled analysis of case-control studies revealed that lower SES is associated with statistically significant increased risk of lung cancer including SCLC when comparing higher SES (OR = 2.13, 95% confidence interval [CI]: 1.63–2.77 in males; OR = 2.85, 95% CI: 1.57–5.18 in females; p for trend < 0.001 for both sexes), after controlling for smoking behaviors.¹⁸

Though the 2-year relative survival for limited-stage SCLC has improved from 36% in 2001 to 2002 to 46% in 2015 to 2016, prognosis remains poor for patients with extensive disease, with 2-year relative survival at 7% to 8% and a median survival of 7 months.¹⁹ Unlike in NSCLC, there has been no major treatment advancement in the past few decades in addition to the standard combination of platinum-based chemotherapy and radiation until 2018. Several phase 3 clinical trials have evaluated the efficacy of additional immunotherapy including atezolizumab²⁰ and durvalumab²¹ (recommended by the European Society for Medical Oncology and National Comprehensive Cancer Network)^22,23 and serplulimab²⁴ and adebrelimab²⁵ in patients with extensive-stage SCLC, with the median overall survival improvement ranging from 2 to 4.5 months compared with chemotherapy alone. Nevertheless, the degree of benefit from immunotherapy was less compared with NSCLC.²⁶ In addition, clinical trial patients are generally highly selected and the long-term survival benefit in patients from the general population is largely unknown.

Risk Factors

Cigarette Smoking

Cigarette smoking is the major risk factor for lung cancer, and it accounts for more than 95% of the SCLC cases.⁴ Several metrics measuring smoking exposure including duration, intensity, pack-years, time since quitting, and age at starting smoking have been studied in relation to lung cancer risk.^27,28 Similar to NSCLC, the risk of SCLC is much higher in current smokers (OR = 42.0, 95% CI: 21.7–81.2) than former smokers (OR = 17.1, 95% CI: 9.5–31.0).²⁹ The risk decreased soon after quitting however remained higher than the baseline risk among never smokers, even 35 years after quitting.³⁰ Though pack-years of smoking has been widely used in clinical practice and research with the assumption that duration and intensity contribute equally, it has been found that for an equal total exposure, smoking at a lower intensity over a longer time period leads to a higher risk of lung cancer compared with smoking at a higher intensity over a shorter time period.^28,31 A pooled meta-analysis has revealed that the OR of developing SCLC is higher for duration (OR ranging from 17.1 to 38.6 for 30 y of smoking) than the number of cigarettes per day (OR ranging from 8.9 to 18.3 for 20 cigarettes/d).²⁹ In addition, earlier age at initiation of cigarette smoking significantly heightens the risk of lung cancer, in particular SCLC compared with other histologies.^29,30 Smoking-related epithelial cell changes in teens may induce an increased lung cancer risk independent of duration and intensity of smoking.^32,33 In a large pooled case-control study including a total of 1760 SCLC cases, Pesch et al.³⁰ suggested that compared with never smokers, males who began to smoke before 15 years old had the highest risk of developing SCLC (OR = 23.3, 95% CI: 14.3–37.9), which started to slowly decrease after 15 years old (OR = 21.2, 95% CI: 13.5–33.5 for age at initiating between 15 and 20 y). Among females, the risk seemed to be lower (OR = 6.8, 95% CI: 3.2–14.7 for age <15 y old) than males but continued to increase in the age 15 to 20 years group (OR = 8.0; 95% CI: 4.8–13.4).³⁰ In 2019, approximately 155 million young individuals (aged 15–24 y) worldwide were tobacco smokers, and 77% of current smokers (aged 20–54 y) started to smoke before the age of 21 years.³⁴ In addition, there has been a considerable increase in the use of electronic cigarettes and vaping, where the number of users has more than tripled between 2013 and 2020, from 21 million to 68 million.³⁵ Young individuals constitute approximately 12% to 45% of electronic cigarette users,³⁶ who are also more likely to smoke cigarettes, and thus have heightened risk of lung cancer later in life.³⁷

In addition, types of cigarettes, filter use, and depth of inhalation could also affect risk of lung cancer.^38,39 For instance, use of filter and lower-tar cigarettes is associated with a higher risk of peripherally located lung cancer (e.g., adenocarcinoma and large cell cancer) whereas nonfilter cigarette consumption is linked to a higher incidence of centrally located lung cancer (e.g., squamous cell lung cancer and SCLC),³⁸ which may also partially explain the decreasing incidence of SCLC.⁴⁰

Substantial international and national efforts have been made in tobacco control including smoking bans, health warning, advertising bans, and tobacco taxes. Many effective progresses have been made, but there is still considerable room for improvement, as the rate in lowering smoking prevalence has been heterogeneous by country, sex, and age group.⁴¹ Continued efforts in more targeted tobacco control approaches on the basis of specific population’s stage of the smoking epidemic and tobacco control are urgently needed.

Nonsmoking-Related Risk Factors

It is estimated that 2% to 3% of SCLCs occur among never smokers.^42,43 Several environmental and occupational exposures and hormonal factors have been suggested to play a role in SCLC. In many populations, residential exposure to radon is the second most import risk for lung cancer after cigarette smoking.⁴⁴ Prior research suggested that residential radon is associated with a marked increased risk of SCLC: the excess of relative risk per each 100 Bq/m³ was 31.2% (12.8%–60.6%), higher than for other lung cancer histology subtypes at 2.6% (0%–10%).⁴⁴ Radon exposure is associated with tumor suppressor TP53 gene somatic mutation, which is found in up to 90% of patients with SCLC compared with 23% to 65% in NSCLC.^5,45,46

The attributable fraction for lung cancer from occupational exposures has been reported to be as high as 15% in males and 5% in females, with asbestos, diesel engine emissions, and other mixtures of polycyclic aromatic hydrocarbons, crystalline silica, arsenic, and some heavy metals as some of the major contributors.^47,48 Indoor and outdoor sources of air pollution (especially in low- and middle-income countries) and secondhand tobacco smoke are risk factors for lung cancer, but most studies do not provide separate results for SCLC or are based on small sample sizes.^49,50 Hormonal, reproductive, and dietary factors may also play a role in SCLC though results from prior epidemiologic studies have been largely inconsistent, and findings were limited by the smaller sample size of SCLC than other histology types.^51–54

Genetic Susceptibility

The SCLC risk remains poorly characterized at the genetic level. Genome-wide association studies (GWAS) have identified multiple susceptibility loci as determinants of increased overall lung cancer risk with the main variations reported at 15q25, 5p15, and 6p21 loci.^55–58 Nevertheless, so far, GWAS explain a small proportion of the overall genetic variance specifically for SCLC, as most of these identified genetic variants are predominantly associated with NSCLC.^57,59,60 This suggests a different molecular landscape underlying SCLC and necessitates a larger sample size to explore genetic variations in SCLC. The only exception is the smoking-related 15q25 locus, which had a significant association with SCLC risk, reflecting the strong association with tobacco smoking.^56,61,62 In addition, a recent meta-analysis has identified five variants that had significant associations with SCLC, though all had moderate or weak cumulative evidence (CHRNA5 rs16969968 [weak], CYP1A1 rs4646903 [moderate], GSTM1 present/null [moderate], NQO1 rs1800566 [weak], XPC rs2228001 [weak]).⁵⁹

Germline variants may be actionable in SCLC treatment.⁶³ Although array-based GWAS studies have focused on overall lung cancer risk, studies on germline-somatic whole-exome sequencing, the largest of which is The Cancer Genome Atlas, have not included any SCLC samples among the 33 cancer types they have studied.⁶⁴ A relatively small-sized study has been recently published by Tlemsani et al.⁶⁵ to evaluate pathogenic germline variants in individuals with SCLC by performing germline whole-exome sequencing on 87 patients, which they validated on 79 patients. This study only focused on the coding variants of 607 cancer-related genes and reported that when compared with population-level controls, unselected patients with SCLC were more likely to carry germline pathogenic variants in RAD51D, CHEK1, BRCA2, and MUTYH, as compared with the healthy controls.⁶⁵ Nevertheless, the study did observe that patients with SCLC with germline pathogenic variant genotype were statistically significantly associated with the likelihood of a first-degree relative with lung cancer or other cancers and longer recurrence-free survival after platinum-based chemotherapy, revealing the critical importance of understanding the role of germline variants in SCLC treatment.⁶⁵ Furthermore, a more recent study has investigated pathogenic or likely pathogenic (P/LP) variants in three genes (BRCA1/2 and partner and localizer of BRCA2, PALB2) in multiple malignancies including SCLC and observed such variants in 5.1% (two of 39) of SCLC patient tumors.⁶⁶ Given the small cohort size, the percentage of patients with SCLC with germline P/LP variants in BRCA1/2 and PALB2 (over all patients with SCLC with P/LP variants in tumor) was unclear. Yet, 74.4% of such patients with other cancer types were found to exhibit germline variants classified as P/LP by ClinVar.⁶⁶ These results have important potential implications for patient selection in future SCLC treatment studies with poly (ADP-ribose) polymerase (PARP) inhibitors. Although PARP inhibitors alone have so far failed to reveal additional benefit to unselected patients with SCLC,^67,68 a recent phase 2 trial combining talazoparib and temozolomide in the second-line setting for extensive-stage SCLC has found an overall response of 39.3%.⁶⁹ Nevertheless, it remains to be investigated whether germline P/LP variants in BRCA1/2 and PALB2 can be used as biomarkers for SCLC patient selection for treatment with PARP inhibitors, analogous to their current use for this purpose in PARP inhibitor therapy in breast, ovarian, prostate, and pancreatic cancers.^70–73 Given these potential treatment implications, combined with potential familial risk for such individuals, clinical germline testing can be beneficial for carriers of P/LP variants in BRCA1/2 and PALB2 in their SCLC tumors. Nevertheless, there is still much work needed to better understand the association of the full spectrum of both coding and noncoding germline genetic variants with SCLC risk, progression, and treatment response.

Gene-environment interactions may be critical in accounting for the missing heritability of SCLC.⁵⁸ As mentioned previously, more than 95% of SCLC cases are partly attributable to tobacco smoking, the highest of all cancers.^30,74,75 Nevertheless, only a small subset of smokers develop SCLC. This is likely explained by genetic variance in addition to competing causes of death at an earlier age. Although gene-smoking interactions for SCLC risk have not been explored in detail, there is evidence to support that smoking-related SCLC risk may be modulated by genetic variations.^55,57,76,77 First, 15q24–25.1, which is the only region that has been statistically significantly associated with SCLC risk in lung cancer GWAS subgroup analyses, was found to have a smoking-related effect. This region contains several genes that encode for nicotinic acetylcholine receptor subunits (CHRNA5, CHRNA3, and CHRNB4), which bind to tobacco-related carcinogens and have increased risks for former (p = 4 × 10⁻⁷) and current (p = 3 × 10⁻¹⁰) smokers.^55,77 Two variants (rs16969968 and rs578776) in this region are also associated with cotinine levels in current smokers.⁵⁷ Second, most of the SCLC tumors have a TP53 mutation that has been found to correlate with carcinogens in tobacco smoke.⁷⁶ Third, patients with SCLC have also been observed to be at a higher risk for second smoking-related cancers.⁷⁶

Molecular Pathology

Role of Tumor Suppressor Genes

The SCLC is characterized by a biallelic inactivation of both tumor suppressor genes RB1 and TP53.⁵ Alterations of TP53 and RB1 genes were firstly identified in SCLC cell lines (Table 1).⁷⁸ Since then, high-throughput genome-sequencing studies have revealed that loss of TP53 and RB1 occurs in more than 90% of the tumors.^5,79

Table 1.

The Frequency of Selected Gene Changes Involved in SCLC, in Cell Lines, and Clinical Samples and Pathogeneses in GEMM

Gene	Cell Lines	Clinical Samples	GEMM
TP53/RB1 5,82,83,162	TP53 mutations: 75%–90%	Inactivating mutations/deletions: p53: 100% RB1: 93%	Conditional deletion of p53 and RB leads to development of multiple tumors resembling human SCLC; loss of p53, RB, and p130 accelerates tumor development maintaining the histologic and molecular features of SCLC.
NOTCH gene family5	NA	Specific mutations: 25% Low activity: 83%	Activation of Notch 1 and Notch2 (intracellular domain) to Trp53/Rb1/Rbl2 mice significantly reduces the number of tumors.
MYC gene family91,163	Amplification: 30%–60%	Amplification: 20%	Loss of RB1 and TP53 and MYC overexpression (RPM) promote SCLC with NEUROD1 expression.
CREBBP EP300 79,94,95	Mutations: 18%	Mutations: 18%	Rb1/Trp53/Crebbp-deficient mice develop a “classic” SCLC.

GEMM, genetically engineered mouse model; NA, not available; NEUROD1, neuronal differentiation 1; RPM, Rb1^fl/fl Trp53^fl/fl Myc^LSL/LSL.

Most of p53 inactivating mutations are missense mutations in the DNA binding domain and in a smaller percentage, homozygous deletions. Abnormal protein expression is observed in 40% to 70% of SCLC. Moreover, p53 mutations also correlate with cigarette smoking.^76,80 The Rb gene is affected by various types of mutations, such as deletions, nonsense mutations, and splicing abnormalities.⁸¹ In addition to the loss of p53 and RB, a significant number of SCLCs are characterized by mutations of the functional homologs RBL1, RBL2, and TP53.⁸¹

The role of TP53 and RB in SCLC development has been investigated in various mouse models. In 2003, Meuwissen et al.⁸² induced the somatic inactivation of Trp53 and RB1 in the epithelial cells of a conditional mouse model. The mutant mice developed aggressive small cell tumors, sharing with the human counterpart the histologic and immunophenotypic features, metastasis pattern, and response to chemotherapy.⁸² Further studies revealed that the ablation of p130 in a murine model associated to Rb1 and p53 loss accelerates tumor development while maintaining the overall histopathologic and molecular features of SCLC.⁸³ Nevertheless, the specific ablation of all three members of the RB family, RB1, P130, and P107 in conditional knockout mouse induced only tumorlets, benign neuroendocrine tumors, enforcing the concept of the synergic role of p53 and RB1 and the need of other mutations for carcinogenesis.⁸⁴

Molecular Complexity of SCLC

Despite the universal loss of TP53 and RB1 genes, SCLC is a heterogeneous disease and its molecular complexity is responsible for the resistance to current therapeutic strategies.⁷ High-throughput genome-sequencing projects based on whole-genome, transcriptome, exome, and copy number analyses have shed light on several relevant genetic and epigenetic mutations that account for SCLC tumor diversity.⁸⁵ In particular, the transcriptomic analysis of human samples revealed that most tumors (approximately 83%) had low Notch activity.⁵ In normal lung development, Notch signaling is critical in regulation of the neuroendocrine compartment.⁸⁶ In SCLC, NOTHC acts as a tumor suppressor that negatively regulates neuroendocrine differentiation.⁸⁷ Specific mutations of the NOTCH family genes occur in 25% of human SCLC.⁵ One of the mechanisms involved in suppression of NOTCH activity in wild-type tumor has been investigated by Oser et al.⁸⁷ The authors revealed that KDM5A histone demethylase represses NOTCH signaling to sustain neuroendocrine differentiation and promote SCLC tumorigenesis. In normal conditions, pRB binds and inhibits the activity of the H3K4 histone demethylase KDM5A. All tumors with NOTCH function impairment express traditional neuroendocrine markers and have high expression of ASCL1, a lineage oncogene of neuroendocrine cells. The ASCL1 is normally present in the lung neuroendocrine cells during development and in the adult and is essential for this lineage to develop.⁸⁷ Moreover, ASCL1 regulates the expression of Delta-like ligand 3 (DLL-3). The DLL-3 is expressed on the surface of the SCLC cells, and it is reported to inhibit Notch signaling in SCLC.⁸⁸ Meder et al.⁸⁹ revealed that one inactivating NOTCH mutation was sufficient to induce neuroendocrine differentiation from non-neuroendocrine tumor cells or precursors. Conversely, endogenous activation of NOTCH activity in a mouse model causes non-neuroendocrine differentiation in SCLC.⁵

Along with the NOTCH family, MYC gene family determines different SCLC-specific phenotypes.⁷⁹ Amplification of MYC family genes including MYC, MYCL, and MYCN occurs in approximately 20% of tumors and is mutually exclusive.^79,90 Although MYCL1 is uniquely expressed by ASCL1-positive SCLC, MYC drives a neuroendocrine-low “variant” subset of SCLC with high neurogenic differentiation factor 1 (NeuroD1) expression.⁹⁰ In Rb1^fl/fl Trp53^fl/fl Myc^LSL/LSL (RPM) murine models, carrying loss of p53 and RB1 and activation of MYC, the developed carcinomas were characterized by significantly reduced expression of the ASCL1 but high expression of NeuroD1.⁹¹ The NeuroD1 has critical roles in promoting neurogenic differentiation of cells during development and malignant behavior in SCLC cell lines.⁹² The RPM tumors have high metastatic potential, in agreement with the observations of Osborne et al.,⁹³ who associated the overexpression of NeuroD1 to the development of metastases and aggressiveness in SCLC cell lines.

Mutations of Epigenetic-Related Gene

Mutations have been identified in epigenetic-related genes in SCLC tumors. Deletions and truncating mutations in CREBBP and EP300 genes together with missense mutations in the histone acetyltransferase (HAT) domain have been observed in 18% of the cases, in a mutually exclusive manner.⁹⁴ CREBBP and EP300 are ubiquitously expressed and facilitate transcription through acetylation of histones and transcription factors.⁹⁴ The role of EP300/CBP mutations in SCLC pathogenesis has not been fully elucidated. According to Jia et al.,⁹⁵ the Rb1/Trp53/Crebbp-mutant mice developed SCLC exhibiting neuroendocrine markers and TTF1, together with pituitary and medullary thyroid carcinomas. The authors observed reduced expression of E-cadherin (CDH1) in SCLC, speculating that the reduced histone acetylation after CREBBP deletion might contribute to the reduced expression of CDH1. Moreover, the loss of CREBBP was associated with increased expression of proteins associated with epithelial-mesenchymal transition (EMT), including Zeb1, N-cadherin, vimentin, and Slug, suggesting CREBBP inactivation can induce a partial epithelial-mesenchymal transition program, with reduced expression of CDH1. The CREBBP/EP300 is also involved in lung neuroendocrine differentiation, cooperating with the Notch pathway to activate the expression of the target genes.⁹⁵

Multiple other types of mutations, including missense mutations and truncations, were discovered in the genes encoding the mixed lineage leukemia (MLL, also known as KMT2). The MLL proteins, as part of multiprotein complexes, regulate methylation of lysine 4 and 27 residues on histone H3 tails (H3K4 and H3K27) in regulatory elements of genes.⁹⁶ MLL1 (KMT2A) and MLL2 (KMT2D) were the most frequently mutated in both SCLC cell lines and patient tumors. Augert et al.⁹⁷ reported a high frequency of truncating mutations in the lysine methyltransferase 2D gene (KMT2D/MLL2) in 18% of cell lines and 8% of primary tumors.

Molecular Classification of SCLC

On the basis of the expression of ASCL1 and NeuroD1, two major SCLC subtypes have been identified (ASCL1-high and NeuroD1-high), defined as SCLC-A and SCLC-N.⁹⁸ The two subtypes share insulinoma-associated protein 1 expression, which is a driver of neuroendocrine differentiation in many organs and tissues, but they are regulated by different pathways and could represent a potential target for specific and precise therapeutic options. Moreover, SCLC-A and SCLC-N significantly express neuroendocrine markers, chromogranin A and synaptophysin, but have different expression of TTF1, which is more expressed in subtype A rather than N. Recent data are beginning to clarify transcriptional drivers relevant in the subtypes of tumors with low-level expression of both ASCL1 and NeuroD1 and with a non-neuroendocrine phenotype.⁹⁸ The profiling of a large panel of human SCLC identified a third subtype, SCLC-P, characterized by significant Notch activity and lacking neuroendocrine markers. The tumors SCLC-P are driven by POU2F3 (SCLC-P) gene.^98,99 The POU2F3 is usually expressed in tuft cells, a rare chemosensory cell type in the pulmonary and intestinal epithelium. POU2F3-expressing SCLC cell lines have an expression profile similar to that of tuft cells, suggesting the possibility of a distinct cell of origin. A fourth subtype was initially identified as SLCL-Y, expressing the YAP1—a regulator of transcription activated by the HIPPO signaling pathway. The YAP subtype is controversial. Some studies revealed that YAP1 did not define a distinct subgroup of SCLC, although its expression was mildly elevated in the ASCL1/NeuroD1 double-negative tumors and was associated with low expression of NE markers.⁹⁸ In contrast, Owonikoko et al.¹⁰⁰ recently performed an extensive surfaceome profiling of SCLC samples and stratified the tumors in five groups—A, N, Y, P, and a mixed type. The YAP1 subtype had a significative higher expression of TACSTD2 and its interacting and regulatory genes compared with A, N, and mixed subtypes. The TACSTD2 gene codes for TROP2, overexpressed in many cancers and an interesting target for new therapies. A novel subtype was recently identified by Gay et al.,¹⁰¹ who observed a group of tumors without a prevailing transcriptional signature but expressing numerous immune checkpoints and human leukocyte antigens. This group was defined as SCLC inflamed or SCLC-I. Furthermore, the authors confirmed a significative expression of the three transcription factors in each subtype using the monoclonal antibodies for ASCL1, NEUROD1, and POUF3. In contrast, the tumors with low expression of the three transcription factors belong to the inflamed group. The SCLC-I has an increased number of immune cells, T cells, natural killer cells, and macrophages. In addition, SCLC-I has the highest mesenchymal differentiation that could be responsible for resistance to therapy.¹⁰¹ A more accurate pathologic classification based on different molecular mechanism and a consequent clinical stratification of these variants could be helpful, in future, to drive new therapeutic options.

Screening and Early Detection

The U.S. Preventive Services Task Force sets forth recommendations for annual lung cancer screening with LDCT among individuals aged 55 to 80 years who have a 30 pack-year smoking history and currently smoke or have quit within the past 15 years, with recent updates to include those aged 50 to 80 years who have a 20 pack-year smoking history.¹⁰² Nevertheless, LDCT did not reveal survival benefit for SCLC.^103–105 Among lung cancer cases that were detected by LDCT in prior lung cancer screening trials, the proportion of SCLC cases of all staged combined only ranged from 0.7% to 15% (Table 2) with absolute incidence ranging from 22 to 97 in 100,000 person-years.¹⁰⁴ It has been estimated that the sensitivity of LDCT for early stage SCLC was significantly lower than that for other histology types: 8.8%, 56.6%, and 31.0% for stage IA SCLC, adenocarcinoma, and squamous cell lung cancer, respectively.⁶ Though limited by small number of SCLC cases, as high as 67.5% of SCLC in prior screening trials were not detected by LDCT, indicating the aggressive biology and rapid growth of SCLC.^{103,104,106,107} In a study analyzing the features of SCLC detected by LDCT in the NLST, 86% of patients with SCLC were diagnosed at stage III/IV compared with 36% in NSCLC by screening.¹⁰³ Among the 34.2% SCLC cases that were detected by LDCT, no survival benefit was found in the NLST compared with those who were never screened (3-y cancer-specific survival: 15.3% versus 12.8%; 3-y overall survival 14.9% versus 13.8%, respectively).¹⁰⁴ The dismal benefit is further hampered by low uptake of LDCT and other clinical and socioeconomic barriers.^108–110 Given the ineffectiveness of LDCT in improving survival for SCLC, novel approaches for early detection of SCLC are urgently needed.

Table 2.

Characteristics and Outcomes of SCLC Detected by Selected LDCT Lung Cancer Screening Trials

Study	Country	Study Design	N Screening / Control Arms	SCLC/LC Screening Arm N (%)	SCLC/LC Control Arm N (%)	SCLC Detection (Screening Arm)	Overall and LC-Specific Mortality RR and 95% CI	Median Follow-Up (y)
NLST103,164	US	3 Annual LDCT vs. 3 annual CXR	26,722/26,732	245/1701 (14)	291/1681 (17)	Screening detected: 34.2% Interval detected: 10.5% Non-screening detected: 55.2%	Overall: 0.92; 0.85–1.00; LC: 0.89; 0.80–0.997a	>10
NELSON106,b	Belgium/Netherlands	4 Annual LDCT vs. no screening	7900/7982	40/344 (12)	46/304 (15)	Screening detected: 32.5% Non-screening detected: 67.5%	Overall: 1.01; 0.92–1.11 LC: 0.76; 0.62–0.94	>10
DLCST165	Denmark	5 Annual LDCT vs. no screening	2052/2052	11/100 (11)	14/53 (26)	NA	Overall: 1.01; 0.82–1.25; LC: 1.03; 0.66–1.60	9.8
DANTE166	Italy	4 Annual LDCT vs. no screening	1264/1186	9/104 (0.7)	6/104 (0.5)	NA	Overall: 0.95; 0.77–1.17; LC: 0.99; 0.69–1.43	8.4
ITALUNG107	Italy	4 Annual LDCT vs. no screening	1613/1593	10/67 (15)	11/71 (15)	Screening detected: 30% Interval detected: 50% Non-screening detected: 20%	Overall: 0.83; 0.67–1.03; LC: 0.70; 0.47–1.03	9.3
MILD104,167	Italy	3 or 5 annual or biennial LDCT vs. no screening	2376/1723	6/98 (6)	4/60 (7)	Screening detected: 75% Non-screening detected: 25%c	Overall: 0.80; 0.62–1.03; LC: 0.61; 0.39–0.95	>10
LUSI168	Germany	5 Annual LDCT vs. no screening	2029/2023	5/63 (8)d	9/36 (25)d	NA	Overall: 0.99; 0.79–1.25; LC: 0.74; 0.46–1.19	8.8
LSS169,170	US	2 Annual LDCT vs. 2 annual CXR	1660/1658	4/40 (10)	2/20 (10)	NA	Overall: 1.20; 0.94–1.54; LC: 1.24; 0.74–2.08	5.2

^aScreening- vs. nonscreening-detected SCLC: 3-year LC-specific survival: 15.3% vs. 13.8%; 3-year OS: 14.9% vs. 13.8%.

^bLimited to male only as the number of lung cancer detected in females at 10-year follow-up is unavailable.

^cOn the basis of Silva et al.,¹⁰⁴ the analysis included MILD and another pilot lung cancer screening trial in Milan. SCLC = eight in the combined analysis.

^dCase number was based on data within 5-year postrandomization. There were additional 16 and eight LC cases that were detected in the screen and control arms beyond 5-year postrandomization, respectively; however, histology subtype was not provided.

CXR, chest radiograph; DANTE, Detection And screening of early lung cancer with Novel imaging TEchnology; DLCST, Danish Lung Cancer Screening Trial; ITALUNG, Italian Lung Study; LC, lung cancer; LDCT, low-dose computed tomography; LUS, German Lung cancer Screening Intervention; LSS, Lung Screening Study; MILD, Multicentric Italian Lung Detection; NA, not available; NELSON, Nederlands–Leuvens Longkanker Screenings Onderzoek; NLST, National Lung Screening Trial; RR, rate ratio; US, United States.

Growing numbers of molecular biomarkers are under development as potential tools for early lung cancer detection, including cytology, chromosomal abnormalities, gene expression, microRNA (miRNA) from nasal or airway epithelial cells by bronchoscopy sampling, sputum, exhaled breath condensate, and urine.¹¹¹ Nevertheless, these tests were mostly studied in NSCLC (reviewed by Seijo et al.¹¹¹). Hence, this review will focus on blood-based liquid biopsy in the early detection of SCLC including circulating tumor cells (CTCs), circulating tumor DNA (ctDNA) (including DNA methylation), miRNA, and proteins (Table 3 and Supplementary Table 1).

Table 3.

Selected Studies Evaluating Blood-Based Biomarkers in Lung Cancer Detection That Included SCLC^a

Biomarkers	Analyte	Cancer Type	Cases/Controls N	SCLC N (%)	Overall SEN/SPEb (%)	SCLC SEN/SPE (%)b	Reference
CTC
CTC count ≥ 1	Blood	Primary LC	125/25	9 (7)	30/88	67/88	Tanaka et al.118
CTC count ≥ 2	Blood	Primary LC	174/90	14 (8)	68/100	57/100	Li et al.116
ctDNA
DELFI	Plasma	Primary LC	129/236	11 (9)	81/80	90/80	Mathios et al.131
TP53 mutation	Plasma	SCLC	51/123	51(100)	49/89	49/89	Fernandez-Cuesta et al.128
DNA methylation
SHOX2	Serum	Primary LC	188/155	15 (8)	60/90	80/90	Kneip et al.134
4-Methylation panelc	Plasma	Primary LC	281/28	44 (16)	38/93 (APC/RASSF1A)	64/84 (HOXA9) 52/96 (RASSF1A)	Nunes et al.135
8-Methylation paneld	Plasma	3 Cancerse	102 (LC)/136	16 (16)	66/70 (LC)f	75/88g	Constancio et al.136
miRNA
miRNA-92a-2	Plasma	SCLC	50/30	50 (100)	56/100	56/100	Yu et al.145
14-miRNA signature	Blood	Primary LC	606/2440	157 (26)	81/96	90/71h	Fehlmann et al.146
2-miRNA signature	Serum	Primary LC	1566/2178	23 (1)	95/99	91/99	Asakura et al.147
Proteins
6-Ab paneli	Serum	SCLC	243/247	243 (100)	55/90	55/90	Chapman et al.150
7-Ab panelj	Plasma	Primary LC	973/1335	108 (11)	61/90	59/90 (LS)	Ren et al.151
7-Ab panelj	Serum	Primary LC	352/119	47 (13)	57/92	57/92	Du et al.152
6-Ab panelk	Serum	Primary LC	641/234	165 (26)	37/90	46/90	Boyle et al.155
Abs to NY-ESO-1 and NSE	Serum	SCLC	57/47	57 (100)	69/92	69/92	Yang et al.154
5-Protein panell	Serum	Primary LC	423/581	13 (3)	49/96	69/80	Mazzone et al.157
6-Protein panelm	Serum	Primary LC	1834/263	99 (5)	87/65	63/99.5	Liu et al.158
Pleiotrophin	Serum	SCLC	128/180	128 (100)	87/96	87/96	Xu et al.159
NSE	Serum	SCLC	9546 samples		69/96	69/92	Huang et al.160

^aThis table presents studies that included ≥100 LC cases, designed for lung cancer detection and reported separate results for SCLC. Other studies (n < 100) that did not report results for SCLC or designed for diagnosing indeterminate pulmonary nodules were presented in Supplementary Table 1.

^bSensitivity = true positives/(true positives + false negatives); specificity = true negatives/(true negatives + false positives).

^cAPC/HOXA9/RARβ2/RASSF1A.

^dAPC/FOXA1/GSTP1/HOXD3/RARβ2/RASSF1A/SEPT9/SOX17. Study was conducted among males only.

^eLung, prostate, and colorectal cancers.

^fResults for FOXA1, RARβ2, RASSF1A, and SOX17.

^gResults for HOXD3 and RASSF1A.

^hSensitivity and specificity of 9-miRNA signature in differentiating NSCLC vs. SCLC.

ⁱAbs to p53, NY-ESO-1, CAGE, GUB4–5, SOX2, and Hu-D.

^jAbs to p53, CAGE, GBU4–5, SOX2, GAGE7, PGP9.5, and MAGEA1.

^kAbs to p53, NY-ESO-1, CAGE, GUB4–5, SOX2, and Annexin 1.

^lCEA, CA125, CYFRA and NY-ESO-1 Ab and HGF.

^mCEA, CA125, CYFRA21–1, SCC, NSE, and ProGRP. Sensitivity and specificity were for differentiating lung cancer histology.

Abs, antibodies; adeno, adenocarcinoma; APC, adenomatous polyposis coli; CA125, cancer antigen 125; CEA, carcinoembryonic antigen; CpG site, 5′-C-phosphate-G-3′ sequence of nucleotides; CT, computed tomography; CTC, circulating tumor cell; ctDNA, circulating tumor DNA; CYFRA, cytokeratin fragments; DELFI, DNA evaluation of fragments for early interception; HGF, hepatocyte growth factor; HOX, homeobox; LC, lung cancer; LG3BP, galectin-3 recombinant protein; LS, limited stage; MAGE, melanoma-associated antigen; NA, not applicable; NSE, neuron-specific enolase; NY-ESO-1, New York esophageal squamous cell carcinoma 1; PAX, paired box 9; PET, positron emission tomography; ProGRP, progesterone-releasing peptide; PTGER4, prostaglandin E receptor 4; PTPRN2, protein tyrosine phosphatase receptor type N2; RARb2, retinoic acid receptor-β; RASSF1A, Ras association domain family 1 isoform A; SEN, sensitivity; SHOX2, Short Stature Homeobox 2; SOX, SRY-box; SPE, specificity; SqL, squamous cell lung cancer; STAG3, stromal antigen 3; TP53, tumor protein 53.

Circulating Tumor Cells

The CTCs derived from patients with SCLC were found to be tumorigenic in mice, and the CTC-derived explants mirror donor patients’ genomic profile and treatment response.¹¹² Approximately 70% to 95% of patients with SCLC have detectable CTCs, which makes CTCs a promising diagnostic biomarker.¹¹³ Nevertheless, most clinical assay and validation studies (adopting different CTC concentration cutoffs, measuring methods, or cell surface markers) evaluating CTCs in lung cancer detection only included NSCLC^114,115 or a small proportion of SCLC (1%–10%).^116–118 Ilie et al.¹¹⁹ analyzed the presence of CTCs to complement LDCT scans in 168 high-risk patients and found that among 3% of patients with elevated CTCs, all developed NSCLC. Tanaka et al.¹¹⁸ found that elevated CTCs could predict patients who are suspected to have lung cancer, but the sensitivity was only 30%. The multicenter French AIR study tested CTC as a standalone lung cancer screening tool in a prospective cohort.¹²⁰ A total of 614 high-risk participants with chronic obstructive pulmonary disease were assessed annually with LDCT, clinical examination, and CTC test using ISET (isolation by size of epithelial tumor cell technique) for three rounds with a total of 38 lung cancer cases (SCLC = 6) detected after 3 years. Nevertheless, the sensitivity of ISET was only 26% of detecting LC at baseline. The ISET did not detect four interval cancers that were missed by LDCT. Separate results for SCLC were not reported likely due to small number.¹²⁰

Circulating Tumor DNA

The ctDNA is released from cancer cells by either apoptosis, necrosis, or active secretions.¹²¹ In NSCLC, it has been widely used as a companion diagnostic test of detecting driver mutations, and emerging studies have revealed its promising role as a predictive biomarker in monitoring response and disease relapse.^122–125 In SCLC, however, the diagnostic and prognostic potentials still need to be replicated in larger studies with standardized testing method.^126,127 In the field of cancer detection, the early hematogenous spread nature of SCLC makes ctDNA a promising tool. Nevertheless, distinguishing cancer of interest from background noises, such as noncancer signals (e.g., hematopoietic mutations of indeterminate potential), continues to be a general challenge for clinical utility, with high false-positive rate as a major obstacle.^128,129 For example, a prior studied revealed that 11.4% of the noncancer controls had TP53 mutations from ctDNA. Extra hurdles may arise owing to the heterogeneity of cancer genome to adequately characterize the potential cancer-specific variants, especially from a large noncancer population.¹²¹ In a prospective study evaluating the significance of plasma DNA mutations for subsequent cancer development in healthy subjects (n = 314), 10 subjects (3.2%) had detectable TP53 mutations. Among those who had detected TP53 mutations, only one patient developed pancreatic cancer after 6.8 years.¹³⁰ Another more recent study by Mathios et al.¹³¹ revealed that by using machine learning and combining ctDNA fragmentation with clinical risk factors and carcinoembryonic antigen (CEA) level, followed by CT imaging, the test detected 94% of patients with lung cancer across stages and subtypes, including SCLC (area under the curve [AUC] SCLC = 0.96). Genome-wide fragmentation profile for the ASCL1 binding region was able to distinguish SCLC from NSCLC with a promising accuracy (AUC= 0.98). Nevertheless, given the highly selected population, these findings need to be replicated in future large-scale prospective studies before clinical applications.

DNA methylation also plays an important role in SCLC biology.¹³² A number of DNA methylation biomarkers either targeting a single region or various different regions (i.e., panel) have been tested in detecting lung cancer.¹³³ The reported sensitivity and specificity in detecting SCLC ranged from 52% to 80% and 84% to 96%, respectively.^134–136 A recent work by Rothwell et al.¹³⁷ tested the performance of SCLC-specific DNA methylation tumor/normal classifier and revealed that it could correctly assigned 93% and 100% circulating-free DNA samples from patients with limited-stage and extensive-stage SCLC, respectively.

Circulating-free DNA and DNA methylation-based biomarkers have also been studied in multicancer early detection (Supplementary Table 2), and several tests have become commercially available.^138–144 Overall, they revealed a high specificity but the sensitivity varies significantly depending on cancer stage, cancer type, histology subtype, and detecting method. The performance in SCLC detection remains unclear.^138–143

MicroRNAs

Circulating miRNAs are small noncoding RNAs that are expressed in most cancer types and are promising diagnostic biomarkers owing to their stability.^145–147 Prior studies using either single miRNA signature or RNA panels have suggested a sensitivity ranging from 56% to 91% and a specificity up to 100% in diagnosing SCLC from controls.^145–147 In a retrospective analysis including 606 symptomatic lung cancer cases (157 SCLC cases), a signature of 14-miRNA signature classifier (MSC) revealed a sensitivity of 81% and a specificity of 96% in differentiating lung cancer from controls.¹⁴⁶ It also revealed a sensitivity of 90% and a specificity of 71% in distinguishing NSCLC versus SCLC.¹⁴⁶ Similarly, Asakura et al.¹⁴⁷ reported a sensitivity of 95% and a specificity of 99% using the levels of two miRNAs combined in detecting lung cancer and a high diagnostic index in differentiating histology subtypes (91% for SCLC). In the context of population screening, Pastorino et al.¹⁴⁸ tested whether MSC at time of baseline LDCT could improve predictive ability in detecting LC in the BioMILD (Bio-Multicenter Italian Lung Cancer Detection cohort) (n = 4119) and found that there was a higher frequency of lung cancer in MSC-positive versus MSC-negative individuals (33.6% versus 18.3%, p = 0.003) in the CT arm, suggesting a potential role in more personalized screening approach. Nevertheless, no such difference was found in the CT-negative group and MSC did not alter the incidence of interval cancer. Of note, 80% of the cases were adenocarcinoma and SCLC was not reported.¹⁴⁸ Despite its great potential, prospective validation using standardized miRNA measurement assays is needed before its clinical use.

Proteins

The role of tumor-associated antibodies (TAAbs) to neoantigens expressed in SCLC has been studied.¹⁴⁹ The TAAb panel in general outperforms individual TAAb.¹⁴⁹ Among limited studies that included SCLC or primarily focused on SCLC, the sensitivity seems to be suboptimal at 46% to 69%.^150–154 In a study including 243 SCLC cases and 247 controls, a panel of six antibodies (p53, NY-ESO-1, CAGE, GUB4–5, SOX2, and Hu-D) had a sensitivity of 55% in detecting SCLC.¹⁵⁰ Another retrospective study (n = 142; 12% SCLC) using a 7-TAAb panel (EarlyCTD-Lung) revealed that the panel could detect lung cancer 4 years before clinical diagnosis.¹⁵⁵ The same panel was then tested in the prospective population screening setting among 12,208 high-risk participants. Participants were assigned to either EarlyCDT-Lung test, and if positive, followed by 6-monthly LDCT for 2 years (intervention arm) or standard clinical care (control arm). The study reported a sensitivity of 32% and a specificity of 90% in detecting lung cancer and found a statistically significant shift toward detecting earlier-stage tumors, but no difference in lung cancer-specific or overall mortality was observed at 2-year follow-up.¹⁵⁶ Separate results for SCLC were not available.¹⁵⁶ Other protein-based biomarkers, such as tumor marker plus TTAbs, complement fragment, pleiotrophin, and neuro-specific enolase, have also been studied in different patient populations with varying sensitivity ranging from 63% to 87%.^157–160

Challenges

Though several biomarkers were found to have encouraging results with high sensitivity and specificity, most of them were tested in the clinical setting where patients were high risk and symptomatic. Hence, in the real-world population screening setting where the prevalence of lung cancer is low, the positive predictive value would be lower than in reported studies, resulting in a higher false-positive rate.¹⁴⁶ In addition, the optimal timing, frequency, and selection of biomarker testing, with or without LDCT, are still uncertain, especially for cancers that are rarer and behave aggressively such as SCLC. Last, none of the above-mentioned blood-based molecular biomarkers have so far been incorporated in routine clinical use in cancer detection as the clinical impact (e.g., decrease in mortality, harm, cost, and unnecessary invasive procedures) compared with standard-of-care LDCT has not been formally tested in large population studies.¹⁶¹ Achieving the goal needs a holistic approach by incorporating tobacco control, machine-learning approaches analyzing multiomics data, and more strategic and collaborative clinical utility study design.

Conclusion

In summary, SCLC remains an aggressive and a deadly cancer despite recent treatment advancement. The incidence of SCLC has been decreasing in industrialized countries in the past three decades. Cigarette smoking is the predominant risk factor of SCLC, and the degree of increased risk varies by different smoking metrics. The SCLC possesses a distinct molecular pathogenesis compared with other lung cancer histology types, characterized by almost universal loss of both tumor suppressor genes RB1 and TP53. A better understanding of the molecular pathogenesis of SCLC is necessary to develop novel therapeutic strategies and to improve prognosis. Last, in addition to tobacco control program and emerging new therapeutics, current and future efforts are underway to explore the novel biomarker in early SCLC detection with the ultimate goal of reducing disease burden.