Real-World Efficacy and Safety of First-Line Chemo-immunotherapy in Extensive-Stage Small Cell Lung Cancer and Its Association with Molecular Subtype

Article information

J Korean Cancer Assoc. 2025;.crt.2025.971

Publication date (electronic) : 2025 November 5

doi : https://doi.org/10.4143/crt.2025.971

Miran Han, Sehhoon Park, Se-Hoon Lee, Junkyu Kim, Jin-yong Kim, Jong-Mu Sun, Jin Seok Ahn, Myung-Ju Ahn

Division of Hematology-Oncology, Department of Medicine, Samsung Medical Center, Sungkyunkwan University School of Medicine, Seoul, Korea

Correspondence: Myung-Ju Ahn, Division of Hematology and Oncology, Department of Medicine, Samsung Medical Center, Sungkyunkwan University School of Medicine, 81 Irwon-ro, Gangnam-gu, Seoul 06351, Korea Tel: 82-2-3410-3438 E-mail: silkahn@skku.edu

Received 2025 September 3; Accepted 2025 November 4.

Abstract

Introduction

According to the 2024 annual report from the Korea Central Cancer Registry using 2022 national data, lung cancer accounts for 11.5% of all malignancies, ranking third in cancer incidence in Korea [1]. However, it remains the most common cause of cancer death. Among these, small cell lung cancer (SCLC) accounts for approximately 9.9% of these. SCLC is a poorly differentiated, high-grade neuroendocrine carcinoma characterized by rapid proliferation, early metastatic dissemination, and poor prognosis [2,3]. Approximately two-thirds of patients present with metastatic disease at diagnosis. SCLC is traditionally classified into two stages. According to the 2025 National Comprehensive Cancer Network guidelines, limited-stage disease is defined as stage I-III (T any, N any, M0), in which a curative radiation dose can be safely delivered. Extensive-stage disease includes T3-4 tumors unsuitable for definitive radiation or any disease with distant metastasis (M1).

For patients with limited-stage SCLC (LS-SCLC), concurrent chemoradiation has been the standard of care, and subsequent durvalumab consolidation has recently improved both progression-free survival (PFS) and overall survival (OS) in LS-SCLC [4]. For extensive-stage disease, systemic chemotherapy has been the mainstay of treatment [5]. Historically, platinum-based chemotherapy—typically etoposide combined with either cisplatin or carboplatin—has served as the standard first-line regimen. Despite high initial response rates, most patients relapse rapidly due to the high proliferative index and genomic instability of SCLC, leading to treatment resistance and poor long-term outcomes [6].

The introduction of immune checkpoint inhibitors (ICIs) has recently transformed the treatment landscape for extensive-stage SCLC (ES-SCLC). The pivotal phase III IMpower133 (2018) and CASPIAN (2019) trials demonstrated improved survival with the addition of anti–programmed cell death-ligand 1 (PD-L1) agents (atezolizumab and durvalumab, respectively) to platinum-etoposide chemotherapy, establishing chemo-immunotherapy as the new first-line standard [7,8].

However, the generalizability of these findings to real-world practice, particularly in Korea, remains uncertain. Patients with poor performance status or comorbidities are often excluded from clinical trials, raising concerns about external validity. Therefore, this study aimed to evaluate the real-world effectiveness and safety of first-line chemo-immunotherapy in Korean patients with ES-SCLC. We included all patients regardless of baseline performance status or metastatic sites to reflect real-world clinical heterogeneity and assess whether outcomes align with those observed in pivotal trials. Additionally, SCLC subtypes were defined, and treatment responses were evaluated by molecular subgroup. Given that brain metastases occur in approximately 40%-60% of SCLC cases and are associated with neurologic symptoms as well as reduced quality of life, we conducted a subgroup analysis to evaluate the intracranial efficacy of chemo-immunotherapy in this population [9,10].

Materials and Methods

Clinical data were retrospectively collected from electronic medical records of patients diagnosed with ES-SCLC at Samsung Medical Center between 2018 and 2021. Eligible patients included those with an initial diagnosis of ES-SCLC as well as those with disease progression from limited-stage to extensive-stage who subsequently received first-line chemo-immunotherapy.

Patients were categorized by clinical variables including age, sex, Eastern Cooperative Oncology Group (ECOG) performance status, smoking history, stage at initial diagnosis, and the presence of brain or liver metastases, and line of systemic therapy (chemo-line; first-line vs. second-line or later). PD-L1 expression was assessed using tumor proportion score (TPS), based on either the 22C3 or SP263 immunohistochemistry assay, as documented in pathology reports. However, due to the absence of PD-L1 data in over 60% of cases, it was excluded from efficacy analyses.

All patients received the standard-of-care regimen approved in Korea, consisting of four 21-day cycles of chemoimmunotherapy with carboplatin (area under the curve 5 mg/mL/min, administered intravenously on day 1), etoposide (100 mg/m² intravenously on days 1-3), and atezolizumab (1,200 mg intravenously on day 1 of each cycle). From cycle 5 onward, maintenance atezolizumab (1,200 mg intravenously every 21 days) was continued until disease progression, unacceptable toxicity, withdrawal of consent, or physician decision.

OS was defined as the time from treatment initiation to death from any cause, with censoring at the date of last follow-up for surviving patients. PFS was defined as the time from treatment initiation to radiographic progression or death. PFS2 was defined as the interval from the first objective progression after first-line chemo-immunotherapy to the second objective progression after initiation of second-line therapy or death. Time to next treatment (TTNT) was defined as the time from treatment initiation to the start of the next systemic therapy, with censoring at last follow-up for patients without additional treatment. OS, PFS, and TTNT were estimated using the Kaplan-Meier method. The overall response rate (ORR) was calculated as the proportion of patients who achieved either a complete response or partial response as their best overall response, according to Response Evaluation Criteria in Solid Tumors (RECIST) ver. 1.1 criteria.

Subgroup analyses were conducted based on clinical variables including age (< 65 vs. ≥ 65 years), sex, ECOG performance status (0-1 vs. ≥ 2), smoking history (never vs. ever), number of treatment cycles, and presence of brain or liver metastases. Multivariate Cox proportional hazards models identified independent prognostic factors for OS and PFS. In patients with baseline brain metastases, OS and PFS were separately estimated using the Kaplan-Meier method. Intracranial responses were assessed by magnetic resonance imaging and classified according to RECIST ver. 1.1.

Molecular subtyping was performed using immunohistochemical (IHC) staining for ASCL1, NEUROD1, POU2F3, and YAP1 to classify tumors into A, N, P, and I subtypes. Tumor samples obtained from outside institutions were not subjected to molecular subtyping due to the unavailability of tissue blocks, and patients who did not consent to molecular analysis were also excluded.

Treatment-related adverse events (TRAEs) were graded according to the Common Terminology Criteria for Adverse Events (CTCAE), ver. 5.0. All statistical analyses were performed using R software ver. 4.5.1 (R Foundation for Statistical Computing).

Results

1. Patient baseline characteristics

A total of 177 patients with ES-SCLC who received first-line chemo-immunotherapy were included. The median age at diagnosis was 66 years (range, 42 to 91 years), and most were male (92.7%). At initial presentation, 95.5% had de novo extensive-stage disease, while 4.5% progressed from limited-stage. ECOG performance status was 0-1 in 91.5% of patients (ECOG 0, 46.9%; ECOG 1, 44.6%), and ≥ 2 in 5.6%. Smoking history was present in 80.8%, with 7.9% never-smokers—slightly higher than in the IMpower133 trial. Brain metastases were detected in 27.7% (n=49); among them, 23 had received prior local brain-directed therapy, including Gamma Knife surgery (GKS) and whole-brain radiotherapy (WBRT), while the remaining 26 untreated patients were asymptomatic at the initiation of chemo-immunotherapy. Liver metastases were present in 26.0% of patients (Table 1). PD-L1 expression was assessed in 63.3% and 68.4% of patients using the 22C3 and SP263 assays, respectively. Among those tested with 22C3 (n=65), TPS < 1% was observed in 60 patients (92.3%), while TPS ≥ 50% was seen in 2 patients (3.1%). Among those tested with SP263 (n=56), TPS < 1% was found in 53 patients (94.6%) and TPS ≥ 50% in two patients (3.6%), indicating predominantly low PD-L1 expression with few high expressers. The median number of chemotherapy cycles was 6 (range, 1 to 48), reflecting a right-skewed distribution (S1 Fig.). Of these, 127 patients (71.8%) proceeded to maintenance therapy, with a median of 3 atezolizumab maintenance cycles (interquartile range, 2 to 6; range, 1 to 44). Among 177 evaluable patients, 59 (33.3%) received only first-line chemo-immunotherapy, whereas 118 (66.7%) proceeded to second-line or later treatment. The most common second-line regimens were irinotecan/carboplatin (n=40, 33.9%), irinotecan/cisplatin (n=24, 20.3%), and irinotecan monotherapy (n=23, 19.5%). Other regimens, including topotecan, amrubicin, or etoposide re-challenge, were used in 31 patients (26.3%). Median follow-up was 27.2 months (range, 3.9 to 43.2 months), with 14.7% followed for more than 3 years.

Table 1.

Baseline clinical characteristics of patients

2. Treatment response and OS

The ORR was 74.5% (95% confidence interval [CI], 67.1 to 81.1), with a median time to response of 3.42 months. The median OS in the entire cohort was 12.4 months (95% CI, 11.6 to 14.9). As shown in Fig. 1, the Kaplan-Meier curve for OS demonstrated an early decline in survival during the first few months, which likely coincided with the completion of cytotoxic chemotherapy, followed by a more gradual slope during the atezolizumab maintenance phase among responders. At the time of analysis, the estimated OS rates at 12, 24, and 36 months were 53.2% (95% CI, 46.3 to 61.2), 23.0% (95% CI, 17.0 to 31.1), and 16.4% (95% CI, 10.2 to 26.2), respectively. A total of three patients (1.7%) survived ≥ 3 years after treatment initiation. In these long-term responders, the median age was 57 years (range, 52 to 72 years), indicating that this subgroup was relatively younger than the overall cohort. All were male, had ECOG performance status 0-1, no liver metastasis at baseline, and received a longer duration of chemo-immunotherapy (median 13 vs. 6 cycles).

Fig. 1.

Overall survival (OS) curve for the total study cohort (n=177). The median OS was 12.4 months. The blue line represents the survival probability over time, and the shaded region indicates the 95% confidence interval. Censored observations are marked with vertical ticks.

In multivariable Cox regression (S2 Fig.), ECOG performance status ≥ 2 was associated with worse OS (hazard ratio [HR], 2.97; 95% CI, 1.44 to 6.15; p=0.003), while receiving ≥ 6 chemotherapy cycles—reflecting the cohort-median total first-line exposure (four induction etoposide–carboplatin plus atezolizumab with maintenance from cycle 5)—was independently associated with improved OS (HR, 0.33; 95% CI, 0.22 to 0.49; p < 0.001). Univariate analysis showed a strong linear relationship between cycle count and survival, with each additional cycle reducing death risk by 16% (HR, 0.84; 95% CI, 0.80 to 0.89; p < 0.001). Age, sex, smoking history, brain metastasis, and liver metastasis were not significantly associated with OS (all p > 0.3).

3. PFS and multivariate analysis

Median PFS was 5.3 months (95% CI, 5.1 to 5.87), and median TTNT was 5.7 months (95% CI, 5.27 to 6.23). PFS rates at 6 and 12 months were 38.1% (95% CI, 31.5 to 46.0) and 16.3% (95% CI, 11.6 to 22.8), respectively. The Kaplan-Meier curves (Fig. 2A and B) showed a steep early decline within the first few months, reflecting early disease progression events during the induction phase. After completion of cytotoxic chemotherapy, the curve became more gradual, consistent with disease stabilization during atezolizumab maintenance. Among patients who received second-line therapy with evaluable data (n=118), the median PFS2 was 4.6 months (95% CI, 4.17 to 5.23). PFS2 rates at 6 and 12 months were 27.1% (95% CI, 18.4 to 36.5) and 10.7% (95% CI, 4.5 to 19.9), respectively.

Fig. 2.

(A) Progression-free survival (PFS) in the overall cohort (n=177). This figure shows the PFS curve of the overall cohort. The median PFS was 5.3 months. The survival probability was estimated using the Kaplan-Meier method, with the shaded region indicating the 95% confidence interval. (B) Time to next treatment (TTNT) in the overall cohort (n=177). This figure shows TTNT curve of the overall cohort. The median TTNT was 5.7 months (95% confidence interval, 5.27 to 6.23). The curve illustrates the probability of patients remaining without initiation of subsequent treatment over time.

In multivariable Cox regression for PFS (S3 Fig.), receiving ≥ 6 chemotherapy cycles was significantly associated with prolonged PFS (HR, 0.51; 95% CI, 0.32 to 0.81; p=0.004). Liver metastasis was also independently associated with worse PFS (HR, 1.89; 95% CI, 1.15 to 3.11; p=0.012). Other clinical variables, including age (HR, 0.99; 95% CI, 0.97 to 1.02; p=0.514), sex (HR, 1.57; 95% CI, 0.51 to 4.83; p=0.434), ECOG ≥ 2 (HR, 2.10; 95% CI, 0.83 to 5.29; p=0.116), smoking history (HR, 0.74; 95% CI, 0.30 to 1.78; p=0.496), and brain metastasis (HR, 1.14; 95% CI, 0.74 to 1.74; p=0.551) were not significantly associated with PFS.

4. Subgroup analysis: patients with brain metastases

Among patients with baseline brain metastases (n=49), the median OS was 12.2 months (95% CI, 10.6 to 20.6), comparable to that of the overall cohort. The median PFS in this subgroup was 5.7 months (95% CI, 4.93 to 6.37), and the median TTNT was 5.53 months (95% CI, 5.20 to 6.20). The 6- and 12-month PFS rates were 40.8% (95% CI, 29.1 to 57.2) and 12.2% (95% CI, 5.8 to 25.9), respectively. In a direct comparison between patients with and without brain metastases, no significant differences were observed in OS (HR, 0.99; 95% CI, 0.67 to 1.45; p=0.944) or PFS (HR, 0.96; 95% CI, 0.65 to 1.40; p=0.816) (Fig. 3A, S4 Fig.).

Fig. 3.

(A) Comparing overall survival (OS) between patients with and without baseline brain metastases (n=49 vs. n=128). The overlapping curves indicate no significant difference in survival between the two groups (hazard ratio, 0.99; 95% confidence interval, 0.67 to 1.45; p=0.944), demonstrating comparable outcomes regardless of brain metastasis status. (B) Comparing OS between patients with brain metastases who received local treatment (Gamma Knife surgery or radiotherapy) and those untreated and asymptomatic (n=23 vs. n=26). Patients who underwent local treatment demonstrated significantly improved overall survival compared to untreated patients (p=0.009), indicating a survival benefit associated with local brain-directed therapies.

Of the patients with brain metastases, 23 had received prior local treatment (e.g., GKS or WBRT), while 26 untreated patients were asymptomatic and did not undergo local brain-directed therapy before starting chemo-immunotherapy. Those who received prior local treatment showed significantly prolonged OS compared to untreated patients (HR, 0.42; 95% CI, 0.21 to 0.83; p=0.012), confirmed by Kaplan-Meier analysis (p=0.009) (Fig. 3B). In contrast, PFS did not differ significantly between two groups (HR, 0.70; 95% CI, 0.37 to 1.33; p=0.277). Intracranial efficacy assessed by RECIST ver. 1.1 showed an ORR of 63.2%, including complete and partial responses; 36.8% had stable or progressive diseases.

6. Treatment-related adverse events

TRAEs of any grade were common. The most frequent hematologic toxicities included anemia (136 patients, 76.8%), neutropenia (87 patients, 49.1%), and thrombocytopenia (79 patients, 44.6%). Grade ≥ 3 cytopenia occurred in 16.9% (anemia), 32.2% (neutropenia: grade 3, 24.3%; grade 4, 7.9%), and 12.4% (thrombocytopenia). These events primarily occurred during cytotoxic chemotherapy phase and were rare during maintenance ICI monotherapy, suggesting chemotherapy as the main cause of myelosuppression.

Non-hematologic adverse events were less frequent and mostly low grade, including nausea (6.2%), vomiting (1.7%), diarrhea (1.1%), and anorexia (4.5%). Elevated liver transaminases were noted in 13.6%, with grade 3 severity in 2.3%. Immune-related adverse events included skin rash in 10.7% of patients (n=19), with grade 1 in 9.0% (n=16) and grade 2 in 1.7% (n=3). Hypothyroidism occurred in 1.1% (n=2) of patients, with one patient each experiencing grade 2 (0.6%) and grade 3 (0.6%) (Table 2).

Table 2.

Incidence and severity of treatment-related adverse events in the overall cohort

Six patients developed severe toxicity-related complications; three discontinued treatments due to events such as acute kidney injury and cytopenia, while three died from treatment-related causes, including septic shock, severe neutropenia complications, and immune-mediated hepatitis. Data on progression sites causing death were limited due to retrospective data collection and follow-up loss.

Discussion

Real-world data from this cohort of 177 patients demonstrated median OS and PFS of 12.4 months and 5.3 months, respectively, which are consistent with outcomes reported in previous pivotal trials. In the IMpower133 trial, the addition of atezolizumab to carboplatin and etoposide improved median OS from 10.3 to 12.3 months (HR, 0.70; 95% CI, 0.54 to 0.91; p=0.007) and PFS from 4.3 to 5.2 months (HR, 0.77; 95% CI, 0.62 to 0.96; p=0.02). Similarly, the CASPIAN trial reported a median OS of 13.0 months with durvalumab plus chemotherapy, compared with 10.3 months in the control arm (HR, 0.73; 95% CI, 0.59 to 0.91; p=0.0047). Given that our cohort included a broader and unselected patient population often excluded from clinical trials, these results are encouraging [7,8].

The median TTNT was 5.7 months, compared with a similar median PFS. Notably, our cohort included a slightly higher proportion of never-smokers than global studies, possibly reflecting regional epidemiologic differences in East Asian SCLC patients [11].

With a long-term follow-up of 27.2 months, the 3-year OS rate in our cohort was 16.4%, which is consistent with CASPIAN trial, where the 3-year OS rate was 17.6% [12]. These results reinforce that chemo-immunotherapy has significantly improved long-term survival compared to chemotherapy alone in both clinical trials and real-world practice.

Multivariate analyses identified several key prognostic factors in ES-SCLC patients undergoing chemo-immunotherapy. Receiving six or more cycles of chemotherapy was independently associated with significantly improved OS and PFS, underscoring the importance of adequate chemotherapy exposure for optimal disease control. Conversely, ECOG performance status of ≥ 2 was associated with poorer OS, consistent with its established role reflecting patient functional reserve. Liver metastasis emerged as a strong predictor of shorter PFS but was not significantly associated with OS, suggesting its impact on early disease progression despite potential modulation of survival by subsequent therapies. Other factors—including age, sex, smoking status, and baseline brain metastases—were not significantly associated with survival in multivariate models. Future prospective studies should validate these findings and explore tailored approaches addressing high-risk features such as poor performance status and hepatic involvement.

While IMpower133 included asymptomatic, treated brain metastases, specific outcomes for this subgroup were not separately reported [7]. Our real-world data support the feasibility and benefit of chemo-immunotherapy in ES-SCLC patients with brain metastases. Subgroup analyses revealed no significant differences in OS (p=0.944) or PFS (p=0.816) between patients with and without brain metastases. These results suggest that brain metastases do not diminish chemoimmunotherapy efficacy. Among patients with brain metastases, those receiving local treatment exhibited significantly improved OS compared to untreated patients, likely reflecting the benefit of concurrent local therapies such as GKS or WBRT.

The safety profile observed was consistent with pivotal trials. Hematologic toxicities (anemia, neutropenia, and thrombocytopenia) predominated during chemotherapy, with minimal cytopenia during ICI maintenance, indicating chemotherapy as the main driver of myelosuppression. Non-hematologic and immune-related adverse events were infrequent and manageable, confirming the real-world tolerability of chemo-immunotherapy.

Despite clinical benefit, the lack of reliable predictive biomarkers remains a major challenge in SCLC management. Unlike non–small cell lung cancer, where PD-L1 is a validated biomarker, its role in SCLC is unclear due to low tumor cellularity and inconsistent correlation with response [13,14]. In line with prior studies, PD-L1 status was unavailable for most patients and could not be used for response stratification. As an alternative biomarker, tumor mutational burden has been investigated as a predictive marker in SCLC, but results are inconsistent and no definitive link to treatment response has been established [15].

Recent transcriptomic studies have identified four molecular subtypes of SCLC—SCLC-A, SCLC-N, SCLC-P, and SCLC-I—based on expression profiles of ASCL1, NEUROD1, POU2F3, or the absence of all three accompanied by an inflamed gene signature [16]. In our cohort, transcriptional subtyping was available for only 68 patients, limiting the ability to assess treatment efficacy by subtype. Although differences in OS and PFS between subtypes were not statistically significant (p=0.574 and p=0.680, respectively), the SCLC-I subtype (90.9%) showed a trend toward a higher response rate compared with both the A subtype (65.6%) and the unknown group (62.4%).

Supporting evidence from Gay et al. [17] identified SCLC-I as the subtype most likely to benefit from immune checkpoint blockade (ICB), with median OS over 18 months on etoposide plus cisplatin (EP) plus atezolizumab versus just over 10 months on EP plus placebo, while other subtypes derived only modest benefit. Together with preclinical data indicating an immunogenic, relative platinum resistance phenotype, these findings support SCLC-I as a biologically distinct, ICB-sensitive subset [17]. Interestingly, the dual (A/N) subtype in our cohort demonstrated an unexpectedly high objective response rate (90.9%), comparable to that of the SCLC-I subtype. Recent single-cell and epigenetic studies have revealed that SCLC subtypes exhibit considerable intratumoral heterogeneity and lineage plasticity, with tumors co-expressing ASCL1 and NEUROD1 representing a transitional or hybrid state between neuroendocrine and non-neuroendocrine lineages [17-19]. Such dual tumors may retain partial inflamed or immune-related features, including residual interferon-γ and PD-L1 signaling, thereby maintaining sensitivity to chemoimmunotherapy [19]. This biological plasticity may underlie the favorable outcomes observed in the dual subtype, warranting further mechanistic validation.

Nevertheless, transcriptomic subtyping is not yet routinely performed in clinical practice, limiting its immediate applicability to treatment selection.

To our knowledge, this study represents the first and one of the largest Korean real-world retrospective analyses evaluating first-line chemo-immunotherapy in patients with ES-SCLC. Importantly, we incorporated both brain metastasis status and transcriptomic subtypes into our analysis, indicating potential variation in treatment outcomes and response patterns. However, several limitations should be acknowledged. As a single-center retrospective study, it is susceptible to selection and information bias and unmeasured confounding, and the modest sample size limits power, particularly for subgroup analyses. These factors warrant cautious interpretation of the results and highlight the necessity for prospective, biomarker-driven validation in larger, multicenter cohorts.

In conclusion, first-line chemo-immunotherapy shows real-world effectiveness and tolerability in ES-SCLC, and integrating molecular with clinical features may guide more precise treatment selection.

Electronic Supplementary Material

Article information Continued