Evaluation of progression-free survival as a surrogate endpoint for overall survival in small cell lung cancer: a systematic review

Abstract

Background

Patients with small cell lung cancer (SCLC) face a poor prognosis and have limited treatment options, which has prompted the initiation of clinical trials. Overall survival (OS) is the gold-standard endpoint in clinical trials, reflects the benefits patients derive from interventions. Nevertheless, utilizing OS as an endpoint in clinical trials requires large sample sizes and greater financial supports than using progression-free survival (PFS). This study aims to evaluate whether PFS can serve as a surrogate endpoint for OS, thereby conserving human and financial resources in SCLC clinical trials.

Methods

We conducted a systematic review of randomized controlled trials (RCTs) from online databases (PubMed, EMBASE, and the Cochrane Central Library) and international conferences up to January 1, 2025. Hazard ratios (HRs) for PFS and OS, as well as PFS and OS rates, were collected. Weighted linear regression analyses were performed to assess the correlation between PFS benefit and OS benefit, with subgroup analyses based on treatment lines and the trial phase.

Results

A total of 43 RCTs involving 15,119 patients with SCLC were analyzed, including 28 trials focusing on treatment-naïve patients and 15 trials evaluating non-first-line therapies. PFS exhibited a moderate correlation with OS benefit in treatment-naïve patients, particularly for those receiving first-line immunotherapy. Moreover, the 1-year PFS rate showed a strong association with the 1.5-year OS in first-line trials and 2-year OS in subsequent-line trials. No significant correlation was observed in patients who had received prior treatments.

Conclusions

This study supports the utilization of PFS as a surrogate endpoint for OS in clinical trials involving treatment-naïve SCLC patients, especially receiving first-line immunotherapy. This finding can enhance the efficiency of clinical trial design and facilitate the evaluation of novel therapeutic interventions in this challenging patient population.

Highlight box.

Key findings

• In treatment-naïve small cell lung cancer (SCLC) patients, particularly those receiving first-line immunotherapy, progression-free survival (PFS) showed a moderate correlation with overall survival (OS) benefit. The 1-year PFS rate exhibited a strong linear correlation with 1.5-year OS and 2-year OS.

• Using PFS as a surrogate endpoint can enhance the efficiency of clinical trial design, facilitating the evaluation of novel therapeutic interventions in treatment-naïve SCLC patients, especially those receiving first-line immunotherapy.

What is known and what is new?

• PFS is commonly used as a surrogate endpoint in clinical trials for various cancers, including lymphoma and thyroid cancer. However, its applicability in SCLC has been less explored.

• This study provides robust evidence supporting the use of PFS as a surrogate endpoint for OS in treatment-naïve SCLC patients, particularly those receiving first-line immunotherapy. The findings highlight the strong correlation between 1-year PFS and longer-term OS, suggesting that PFS can reliably predict OS benefits in this specific patient population.

What is the implication, and what should change now?

• The use of PFS as a surrogate endpoint can significantly streamline clinical trial design and evaluation in SCLC, reducing the need for large sample sizes and long follow-up periods. This can accelerate the development and approval of new therapies, ultimately improving patient outcomes.

• Clinical trial protocols for treatment-naïve SCLC patients should consider incorporating PFS as a primary endpoint, especially in studies evaluating first-line immunotherapy. This approach can lead to more efficient and timely assessments of treatment efficacy, supporting faster advancements in SCLC treatment strategies.

Introduction

Small cell lung cancer (SCLC), a rare and highly aggressive neuroendocrine tumor, accounts for approximately 13% of all lung cancer cases (1). SCLC is typically characterized by its propensity for early metastasis, rapid progression, robust resistance mechanisms, and poor prognosis (2). The Veterans Administration Lung Cancer Study Group (VALCSG) staging system classifies SCLC into two distinct stages: limited-stage (LS) and extensive-stage (ES) (3). The advent of immunotherapy has significantly influenced first-line treatment for ES-SCLC, where traditional platinum-etoposide regimens are frequently combined with immunotherapy or other drugs in clinical trials (4). Despite these advancements, substantial improvements in survival outcomes remain elusive. Consequently, the identification of novel therapeutic strategy is critical to addressing this longstanding challenge and overcoming the current clinical limitations of SCLC. Overall survival (OS) is generally regarded as the gold standard primary endpoint in clinical trials evaluating tumor treatments. However, the rigorous assessment of OS is fraught with challenges, including constraints related to sample size, extended follow-up periods, and the significant financial demands of such research. Moreover, the early metastasis and high incidence of severe complications associated with SCLC hinder trial enrollment, thereby limiting the pool of eligible participants (5).

Surrogate assessment is a methodological approach designed as an alternative to direct evaluation, often employed in research contexts where direct measurement is impractical, unethical, or prohibitively expensive. This approach relies on indirect measures, referred to as surrogate markers, which aim to approximate or predict the outcomes of interest. Surrogate assessment originated in medical and environmental sciences, where directly measuring long-term outcomes, such as disease progression or ecosystem degradation, is frequently unfeasible within the time constraints of a study. In these fields, surrogate markers—such as biomarkers in clinical trials or specific environmental indicators in ecological studies—provide a practical means of estimating probable outcomes without the extended timeframes required for direct observation. While surrogate endpoints can offer valuable insights into complex phenomena, they remain indirect representations of the primary outcomes being studied (6).

In the field of malignant cancer treatment, progression-free survival (PFS), event-free survival (EFS), objective response rate (ORR), and imaging or biochemical outcomes are commonly utilized as potential surrogate endpoints. These alternatives mitigate the constraints of sample size and follow-up duration while offering the additional advantage of promptly evaluating intervention effects. Recent studies sanctioned by the Food and Drug Administration (FDA) have validated the use of PFS as a surrogate endpoint for OS in clinical trials for lymphoma and thyroid cancer (7,8). However, evidence regarding the appropriateness of PFS as a surrogate endpoint for OS in SCLC remains limited. To address this gap, the present study conducts a comprehensive review of existing clinical trials to assess the feasibility of employing PFS as a surrogate endpoint for OS in SCLC at the trial level. This investigation aims to provide valuable insights to inform and refine treatment strategies in ongoing and future SCLC clinical trials. We present this article in accordance with the PRISMA reporting checklist (available at https://jtd.amegroups.com/article/view/10.21037/jtd-2025-469/rc).

Methods

Search strategy

A systematic search was performed in the PubMed, Cochrane Central Library and EMBASE databases up to January 1, 2025, with no language limitation. The detailed search strategies are provided in Table S1. Additionally, proceedings from the World Conference on Lung Cancer, the American Society of Clinical Oncology, and the European Cancer Conference were reviewed to ensure comprehensive data collection. The protocol was registered in the Prospective Register of Systematic Reviews (CRD42023394567).

Inclusion criteria

Studies were included based on the following criteria: (I) randomized clinical trials; (II) trials with at least two arms; (III) availability of complete data; (IV) clinical reporting of both PFS and OS; and (V) patients with histologically or clinically documented SCLC according to the eighth edition of the American Joint Committee on Cancer staging guidelines. The exclusion criteria were as follows: (I) case reports, meta-analyses, or systematic reviews; and (II) studies where endpoint data could not be extracted from the original research.

Data extraction

Two researchers (G.L., F.G.) independently screened eligible trials and extracted endpoint data into electronic spreadsheets. The data included the trial name, author, publication year, trial phase, pathological type, number of treatment lines, sample size, therapy regimen, hazard ratios (HRs) for PFS and OS, PFS rates (6-month and 1-year), and OS rates (1-, 1.5-, and 2-year). PFS was defined as the time from diagnosis to any tumor progression or death from any cause. OS was defined as the time from diagnosis to death from any cause.

Risk of bias assessment

The risk of bias for each eligible trial was assessed using Review Manager (version 5.3) and the Cochrane Risk of Bias Tool, considering the following aspects: random sequence generation, allocation concealment, blinding of participants and personnel, blinding of outcome assessment, incomplete outcome data, selective reporting, and other bias. Items were categorized as low, high, or unclear risk of bias.

Data analysis

The most recent survival endpoint data were extracted from clinical trials and used to analyze the correlations between PFS and OS. A weighted linear regression analysis of log HRs for PFS and OS was conducted, with weights based on the number of patients in each trial. The Pearson correlation coefficient (R) from the weighted linear regression model was used to assess the degree of correlation between PFS benefit and OS benefit. According to the ReSEEM guidelines, an R≥0.7 indicates a strong correlation, 0.5≤R<0.7 indicates a moderate correlation, and R<0.5 indicates a weak correlation. For both first-line and non-first-line trials, the correlation between surrogate endpoints and OS was explored. For the intervention arm, correlations between the 6-month and 1-year PFS rates and the 1-, 1.5-, and 2-year OS rates were investigated using the Pearson correlation coefficient. Leave-one-out sensitivity analyses were conducted to assess the robustness of the findings. All analyses and plots were generated using the ggpubr package in R software (version 4.2.3, http://cran.r-project.org/).

Results

Eligible trials and characteristics

A literature search identified 1,523 records from online databases and international conferences. After excluding 976 studies that lacked complete survival data, 148 non-randomized controlled trials (RCTs), and 279 duplicates, the full texts of 120 publications were further screened. Subsequently, 77 biomarker studies were excluded. Ultimately, 43 RCTs that met the inclusion criteria and were rated as having a low risk of bias were included in the analysis (Figure 1 and Figure S1).

Figure 1.

The flowchart of study selection.

The 43 eligible RCTs reported data on 15,119 patients with SCLC, with 28 trials including treatment-naïve patients (Table S2, No. 1–28), and 15 trials focusing on non-first-line studies (Table S2, No. 29–43). All trials had two arms, except for NCT02289690, which had three arms (veliparib plus etoposide plus platinum/carboplatin vs. veliparib vs. etoposide plus platinum/carboplatin), and NCT02538666, which also had three arms (nivolumab plus ipilimumab vs. nivolumab vs. placebo). All first-line trials included ES-SCLC patients, except for the LUNGSTAR trial, which did not report the SCLC subtype. All non-first-line trials were second-line, except for NCT03059797, which was a third- or further-line trial. Detailed information on the eligible trials is summarized in Tables 1,2.

Table 1. The clinical information of included first-line trials.

Trial name	Publication year	Phase	Number of treatment line	Pathology	Treatment	Sample size	PFS, %				OS, %
							HR	6-month	1-year		HR	1-year	1.5-year	2-year
NCT04256421	2024	3	First-line	ES-SCLC	Tiragolumab + atezolizumab + etoposide + platin/carboplatin	243	1.08	35.2	14.2		1.09			21
					Atezolizumab + etoposide + platin/carboplatin	247		42.4	17.3					26
NCT03043872	2021	3	First-line	ES-SCLC	Durvalumab + tremelimumab + etoposide + platin/carboplatin	268	0.84	43.2	16.9		0.81		30.7
					Etoposide + platin/carboplatin	269		45.8	5.3				24.8
NCT03066778	2020	3	First-line	ES-SCLC	Pembrolizumab + etoposide + platin/carboplatin	228	0.75		13.6		0.80	45.1		22.5
					Etoposide + platin/carboplatin	225			3.1			39.6		11.2
NCT03043872	2022	3	First-line	ES-SCLC	Durvalumab + etoposide + platin/carboplatin	268	0.8		84.2		0.71
					Etoposide + platin/carboplatin	269			85.2
NCT02763579	2021	3	First-line	ES-SCLC	Atezolizumab + etoposide + platin/carboplatin	201	0.77				0.76	51.9	34
					Etoposide + platin/carboplatin	202						39	21
NCT04063163	2022	3	First-line	ES-SCLC	Serplulimab + etoposide + platin/carboplatin	389	0.47				0.63	60.7		43.1
					Etoposide + platin/carboplatin	196						47.8		7.9
NCT03711305	2022	3	First-line	ES-SCLC	Adebrelimab + etoposide + platin/carboplatin	230	0.67	49.4	19.7		0.72	62.9		31.3
					Etoposide + platin/carboplatin	232		37.3	5.9			52		17.2
NCT01450761	2016	3	First-line	ES-SCLC	Ipilimumab + etoposide + platin/carboplatin	566	0.85				0.94
					Etoposide + platin/carboplatin	566
SALUTE trial	2011	2	First-line	ES-SCLC	Bevacizumab + etoposide + platin/carboplatin	52	0.52				1.16
					Etoposide + platin/carboplatin	50
NCT00930891	2015	2	First-line	ES-SCLC	Bevacizumab + etoposide + platin/carboplatin	37	1.05				0.8
					Etoposide + platin/carboplatin	37
GOIRC-AIFA	2017	3	First-line	ES-SCLC	Bevacizumab + etoposide + platin/carboplatin	101	0.72				0.78	37
					Etoposide + platin/carboplatin	103						25
10.1097/JTO.0000000000000343	2015	2	First-line	ES-SCLC	rh-Endostatin + etoposide + platin/carboplatin	69	0.801	59.3			1.048	50
					Etoposide + platin/carboplatin	69		46.6				54.6
NCT02499770	2019	2	First-line	ES-SCLC	Trilaciclib + etoposide + platin/carboplatin	38	0.7				0.87
					Etoposide + platin/carboplatin	37
NCT02161419	2019	2	First-line	ES-SCLC	Roniciclib + etoposide + platin/carboplatin	71	1.242	24.7			1.218
					Etoposide + platin/carboplatin	71		28.2
NCT02289690	2021	2	First-line	ES-SCLC	Veliparib + etoposide + platin/carboplatin	59	0.979				1.46
					Veliparib	61	0.665				1.432
					Etoposide + platin/carboplatin	61
NCT01642251	2019	2	First-line	ES-SCLC	Veliparib + etoposide + platin/carboplatin	64	0.75				0.83
					Etoposide + platin/carboplatin	64
NCT01439568	2017	2	First-line	ES-SCLC	LY2510924 + etoposide + platin/carboplatin	47	1.01				1.51
					Etoposide + platin/carboplatin	43
NCT00791154	2017	2	First-line	ES-SCLC	Rilotumumab + etoposide + platin/carboplatin	62	1.05				0.84
					Etoposide + platin/carboplatin	61
NCT00660504	2016	3	First-line	ES-SCLC	Amrubicin + etoposide + platin/carboplatin	149	0.88				0.81	48.6
					Etoposide + platin/carboplatin	150						41.9
NCT00168896	2011	3	First-line	ES-SCLC	Irinotecan + etoposide + platin/carboplatin	106	1.29				1.34
					Etoposide + platin/carboplatin	110
NCT00349492	2019	3	First-line	ES-SCLC	Irinotecan + etoposide + platin/carboplatin	173	0.846				0.879
					Etoposide + platin/carboplatin	189
jRCTs031180193	2023	3	First-line	ES-SCLC	Carboplatin + etoposide	129	0.85				0.85
					Carboplatin + irinotecan	129
NCT00453154	2015	2	First-line	ES-SCLC	Sunitinib + chemotherapy	44	1.62				1.28	62.6
					Chemotherapy	41						43.9
NCT03043872	2019	3	First-line	ES-SCLC	Durvalumab + chemotherapy	268	0.73	45	18		0.78	54	34
					Chemotherapy	269		46	5			40	25
NCT04063163	2022	3	First-line	ES-SCLC	Serplulimab	389	0.48				0.63	60.7		43.1
					Placebo	196
LUNGSTAR	2017	3	First-line	SCLC	Pravastatin + chemotherapy	422	0.98		25.3		1.01			14.4
					Chemotherapy	424			24.2					13.2
JCOG 0509	2014	3	First-line	ES-SCLC	Amrubicin + cisplatin	142	1.42				1.43
					Irinotecan + cisplatin	142
ETER701	2023	3	First-line	ES-SCLC	Anlotinib + TQB2450 + etoposide + platin/carboplatin	246	0.32				0.61
					Placebo + etoposide + platin/carboplatin	247

ES, extensive-stage; HR, hazard ratio; OS, overall survival; PFS, progression-free survival; SCLC, small cell lung cancer.

Table 2. The clinical information of included non-first-line trials.

Trial name	Publication year	Phase	Number of treatment line	Pathology	Treatment	Sample size	PFS, %				OS, %
							HR	6-month	1-year		HR	1-year	2-year
NCT03059797	2021	2	Third- or further-line	SCLC	Anlotinib	81	0.19				0.53	30.6
					Placebo	38						13.1
NCT02566993	2023	3	Second-line	SCLC	Lurbinectedin + doxorubicin	307	0.83	31	11		0.97	31	9
					Chemotherapy	306		24	4			28	10
UMIN000000828	2016	3	Second-line	SCLC	Topotecan	90	0.5				0.67
					Chemotherapy	90					0.95
ETOP/IFCT 4-12 STIMULI	2022	2	Second-line	LS-SCLC	Nivolumab	78	1.02		48.1				62.9
					Placebo	75			52.8		1.03		66.4
NCT03516084	2021	3	Second-line	ES-SCLC	Niraparib	125	0.88
					Placebo	60					0.92
NCT02538666	2021	3	Second-line	ES-SCLC	Nivolumab + ipilimumab	279	0.72				0.84
					Nivolumab	280	0.67
					Placebo	275
KCSG-LU12-07	2018	2	Second-line	ES-SCLC	Pazopanib	48	0.44				1.14
					Placebo	47
UMIN000001755	2017	3	Second-line	ES-SCLC	Prophylactic cranial irradiation	113	0.98				1.27	48.4	15
					Placebo	111						53.6	18.8
NCT01497873	2021	2	Second-line	SCLC	Belotecan	80	1.65				0.69	58
					Topotecan	81						27
NCT01017601	2020	2	Second-line	ES-SCLC	Ntx-010	26	1.03				1.49
					Placebo	24
NCT03059667	2019	2	Second-line	SCLC	Atezolizumab	49	2.26				0.84
					Chemotherapy	24
10.1200/JCO.2013.54.5392	2014	3	Second-line	SCLC	Amrubicin	424	0.8				0.88
					Topotecan	213
NCT03033511	2021	3	Second-line	ES-SCLC	Tesirine	372	0.48				1.07
					Placebo	376
NCT01928394	2021	3	Second-line	SCLC	Nivolumab	284	1.41	19.7	10.9		0.86	36.3
					Chemotherapy	285		26.5	10			34.1
NTR1527	2014	3	Second-line	ES-SCLC	Radiotherapy	247	0.73	24			0.84	33	13
					Control	248		20				28	3

ES, extensive-stage; HR, hazard ratio; LS, limited-stage; OS, overall survival; PFS, progression-free survival; SCLC, small cell lung cancer.

Trial-level correlations between treatment effects of PFS and OS

Correlation analyses were performed according to the treatment line. For first-line trials, weighted correlation analysis demonstrated a significant correlation between the 29 pairs of PFS HRs and OS HRs. The log HR of PFS was moderately correlated with the log HR of OS (R=0.616; 95% CI: 0.322–0.802; P<0.001; Figure 2). For non-first-line trials, no significant correlation was observed between PFS and OS (P=0.63; Figure 2).

Figure 2.

Trial level correlations between PFS and OS in first-line and non-first-line RCTs. Nodes present the trial. Line represents fitted weighted linear regression line, wathet zone represents its 95% CI. CI, confidence interval; HR, hazard ratio; OS, overall survival; PFS, progression-free survival; R, correlation coefficient; RCT, randomized controlled trial.

To increase the robustness of the analysis, we conducted exploratory analyses for phase III and phase II trials separately. As shown in Table S3, only phase III first-line trials exhibited a strong correlation between PFS and OS (R=0.96; 95% CI: 0.90–0.99; P<0.001). Phase II first-line trials and phase II non-first-line trials showed no significant correlation between PFS and OS. These results suggest that PFS benefits from interventions can reliably estimate OS benefits for treatment-naïve patients with SCLC, but not for those who have received prior therapy. Sensitivity analyses confirmed the robustness of the correlations between PFS and OS (Figure S2).

Treatment arm-level correlations between treatment effects of PFS on OS

For first-line trials, the 6-month PFS, 1-year PFS, 1-year OS, 1.5-year OS, and 2-year OS rates were reported in 28 trials. The correlation trend were observed between the 6-month PFS (R=0.030; 95% CI: −0.801 to 0.822; P=0.95), 1-year PFS (R=0.797; 95% CI: −0.041 to 0.977; P=0.058) and 1-year OS, yet without statistical significance. The 1-year PFS (R=0.971; 95% CI: 0.142 to 0.999; P=0.03) exhibited a linear correlation with 1.5-year OS. The 6-month PFS (R=0.919; 95% CI: −0.359 to 0.998; P=0.08) and 1-year PFS (R=0.184; 95% CI: −0.598 to 0.787; P=0.66) showed a linear correlation with 2-year OS. These results indicate a strong correlation between 1-year PFS and 1.5-year OS for treatment-naïve SCLC patients (Figure 3). Additionally, the associations between PFS and OS were stronger for patients receiving immunotherapy, either alone or in combination with chemotherapy, compared to those receiving only chemotherapy (Figure S3).

Figure 3.

Intervention-level correlations between PFS and OS in clinical trials. First-line trials: line 1–2 (green); non-first-line trial: line 3–4 (yellow). The color in the heat map represents the correlation coefficient (95% confidence interval) and hazard ratio, darker color represents stronger correlations. na, not applicable; OS, overall survival; PFS, progression-free survival.

For non-first-line trials, the 6-month PFS, 1-year PFS, 1-year OS, and 2-year OS rates were extracted from 15 trials. The 1-year PFS (R=0.777; 95% CI: −0.726 to 0.995; P=0.22) showed the positive correlation trend with 1-year OS. The 6-month PFS (R=0.430; 95% CI: −0.905 to 0.984; P=0.57) and 1-year PFS (R=0.991; 95% CI: 0.642 to 0.998; P=0.009) showed the positive correlation trend with 2-year OS. A strong correlation was observed between 1-year PFS and 2-year OS in SCLC patients who received antitumor therapy (Figure 3). For patients who received further-line chemotherapy, the 6-month PFS (R=0.344; P=0.78) and 1-year PFS (R=0.787; P=0.42) showed no significant correlation with 1-year OS (Figure S4).

Patient-level analyses of correlations between PFS and OS

PFS and OS data were collected based on the baseline characteristics of patients. Only 4 first-line trials reported both PFS and OS at the patient level. The patient-level analysis revealed a strong correlation between PFS and OS for patients younger than 65 years of age (R=0.99, P=0.04), while no significant linear relationship was observed for patients older than 65 years of age. Additionally, differences in sex, Eastern Cooperative Oncology Group (ECOG) performance status, and programmed cell death ligand 1 (PD-L1) expression did not result in a correlation between PFS and OS. Detailed information is provided in Table S4.

Discussion

The use of surrogate endpoints in clinical trials is increasingly common, particularly in settings where the primary outcome of interest is difficult to measure directly due to long follow-up periods or large sample size requirements. The validation of surrogate endpoints is crucial to ensure that they reliably predict the treatment effects on the target outcomes. Recent frameworks for the validation of surrogate endpoints emphasized the need for rigorous statistical evidence demonstrating that the treatment effect on the surrogate endpoint is strongly predictive of the treatment effect on the target outcome (9). In this study, the statistical significance of the observed correlations between PFS and OS was rigorously assessed using weighted linear regression analyses. For first-line trials, the log HR of PFS was moderately correlated with the log HR of OS. This indicates that the observed correlation is statistically significant, suggesting that improvements in PFS can reliably estimate improvements in OS for treatment-naïve patients with SCLC. Specifically, the 1-year PFS rate showed a strong linear correlation with 1.5-year OS and 2-year OS in first-line trials. These findings underscore the robustness of PFS as a surrogate endpoint for OS in this patient population. In contrast, no significant correlation was observed between PFS and OS in non-first-line trials. This suggests that the relationship between PFS and OS may be influenced by prior treatment exposure, and PFS may not be a reliable surrogate endpoint for OS in patients who have received prior therapies. This highlights the importance of considering the treatment history of patients when evaluating the utility of PFS as a surrogate endpoint. In brief, these findings highlight the importance of incorporating PFS as a metric to inform the design of subsequent therapeutic regimens for patients diagnosed with SCLC.

Traditionally, OS has served as the benchmark endpoint in oncology trials, providing a comprehensive measure of a treatment’s impact on the long-term well-being of patients. However, OS assessments present challenges related to prolonged follow-up periods, substantial financial costs, and potential confounding factors, particularly subsequent lines of therapy. In contrast, PFS emerges as an appealing alternative endpoint, capable of capturing early treatment effects by delineating the duration until disease progression or death. This surrogate endpoint assumes particular significance in SCLC, a disease characterized by its aggressive pathophysiology and rapid progression (10). By offering a more immediate indication of treatment efficacy compared to OS, PFS holds promise for accelerating the evaluation of interventions in clinical trials. The versatility of PFS is highlighted by its adaptability to various clinical trial scenarios. In the specific context of SCLC, where patient enrollment is complicated by early metastasis and severe complications (11), PFS facilitates a more pragmatic recruitment process. Its focus on disease progression, rather than long-term survival, enhances the practicality and efficiency of clinical trials in this challenging patient population. Furthermore, the efficacy of PFS is underscored by its demonstrated ability to navigate diverse clinical trial landscapes. This adaptability is especially crucial in establishing its utility as a surrogate endpoint in SCLC, where the exigencies of the disease necessitate a careful approach to endpoint selection.

PFS offers a more immediate measure of therapeutic benefit, enabling quicker assessments of treatment impact. By focusing on disease progression rather than mortality, PFS can capture early signs of treatment efficacy, making it particularly useful in diseases with prolonged survival times. The adoption of PFS as an endpoint has also transformed the landscape of clinical research, facilitating the acceleration of drug development and regulatory approvals. This is particularly significant in contexts where patient survival is extended due to improved care, and where immediate indicators of drug effectiveness are essential. Moreover, because PFS provides data on the therapeutic effect while excluding deaths unrelated to the disease, it is a more sensitive measure in heterogeneous cohorts with varying comorbidities. However, the use of PFS also introduces challenges, as it requires rigorous and consistent definitions of disease progression, and its interpretation may be influenced by measurement biases. Nevertheless, PFS remains a valuable endpoint in clinical trials, contributing to more efficient evaluations of treatment benefits and supporting the timely development of effective therapies in contemporary clinical research.

The empirical foundation supporting PFS as a surrogate endpoint is further strengthened by robust correlations observed in various malignancies, as demonstrated by relevant studies in lymphoma (7) and thyroid cancer (8). This analysis rigorously evaluates the existing evidence to assess the applicability of PFS as a surrogate endpoint for OS, specifically in the complex context of SCLC. However, the degree of association between PFS and OS is not consistent across different cancers (12). For example, in highly aggressive malignancies with limited effective treatment options, PFS often correlates more closely with OS, as disease progression is tightly linked to mortality (13). In contrast, in slower-progressing cancers or those with high survival rates, improvements in PFS may not translate into meaningful gains in OS, due to the availability of subsequent lines of therapy that can extend survival independent of the initial treatment’s effect on PFS (14). In cancers where progression does not significantly impact quality of life or symptom burden, the clinical relevance of PFS as an endpoint is less clear (15,16). Additionally, treatment mechanisms can influence this relationship; for instance, targeted therapies or immunotherapies may stabilize the disease for extended periods without immediately affecting OS, whereas traditional chemotherapies may exhibit parallel trends in PFS and OS due to their systemic effects (13). Given these variable outcomes, understanding the correlation between PFS and OS across different malignancies is critical for determining whether PFS can serve as a reliable surrogate endpoint (6,17). Such evaluation not only informs the design of future clinical trials but also assists regulatory agencies and stakeholders in assessing the therapeutic value of new interventions, ensuring that endpoints used in clinical trials accurately reflect patient-centered outcomes and treatment efficacy across diverse cancer types (18).

Over the past three decades, addressing the significant challenge of improving the prognosis for patients with SCLC has remained a persistent endeavor. A previous study conducted a comprehensive assessment of the robust correlation between PFS and OS in SCLC patients receiving first-line treatment, revealing a notable coefficient of 0.75 (19). Significant developments in the clinical trial landscape have introduced a diverse range of therapeutic agents, including immune checkpoint inhibitors (e.g., atezolizumab, nivolumab, pembrolizumab), cell cycle regulation inhibitors (e.g., alisertib, trilaciclib), DNA damage repair inhibitors (e.g., olaparib, berzosertib), transcription inhibitors (e.g., lurbinectedin), DLL3-targeted therapies (e.g., rovalpituzumab), and various modalities of radiation therapy (including low-dose radiotherapy, high-dose radiotherapy, and stereotactic radiotherapy). This analysis encompassed all published drugs, revealing a moderate correlation, specifically in treatment-naïve SCLC patients. The increased diversity and heterogeneity inherent in these therapeutic agents contributed to a discernible attenuation in the correlation between PFS benefit and OS benefit. Given the characteristic rapid progression of SCLC, most trials reported PFS data at 6-month and 1-year intervals. Our evaluation revealed a robust correlation between 1-year PFS benefit and 1.5-year OS benefit in treatment-naïve patients, as well as a similar correlation between 1-year PFS benefit and 2-year OS benefit in second- or subsequent-line patients with SCLC. These findings underscore the temporal limitations associated with the use of PFS as a surrogate endpoint, reflecting the inherently aggressive nature of SCLC.

However, 6-month PFS has shown a relatively weak correlation with OS, which also implying that shorter-term PFS may not be as reliable in predicting long-term outcomes. This discrepancy can be attributed to several factors: (I) shorter-term PFS (e.g., 6 months) primarily reflects the immediate impact of treatment on disease progression. While this is valuable for assessing early therapeutic effects, it may not fully capture the long-term dynamics of disease control and patient survival, especially in aggressive cancers like SCLC. (II) The biological behavior of SCLC can vary significantly among patients. Some may experience rapid progression despite initial treatment response, while others may have a more indolent course. Shorter-term PFS may not account for this variability, leading to less accurate predictions of long-term survival. (III) Immunotherapies can induce complex immune responses that may not be evident in the short term (20). For instance, “pseudo-progression” can occur where the tumor initially appears to grow due to immune cell infiltration but eventually shrinks, leading to improved long-term outcomes (21). This phenomenon may not be captured by 6-month PFS assessments. Therefore, understanding these temporal limitations is crucial for designing clinical trials that effectively use PFS as a surrogate endpoint. Firstly, given the stronger correlation between 1-year PFS and longer-term OS, clinical trials should consider using 1-year PFS as a primary endpoint, especially in studies evaluating first-line immunotherapy. This aligns with the observed temporal dynamics and provides a more reliable indicator of long-term patient outcomes. Then, clinical trials should consider stratifying patients based on risk factors and comorbidities that may influence treatment response and survival. This can help in identifying subgroups where shorter-term PFS may still be predictive and in understanding the variability in outcomes. Lastly, combining PFS with other endpoints, such as ORR or time to next treatment, and incorporating real-world data alongside RCT findings can provide additional insights into the predictive value of PFS in diverse patient populations. Real-world data can help bridge the gap between trial results and clinical practice, ensuring that surrogate endpoints are applicable in broader contexts.

Moreover, the correlation between PFS and OS is stronger for immunotherapy compared to chemotherapy in treatment-naïve patients. Chemotherapy typically works by directly targeting and destroying rapidly dividing cancer cells, often resulting in an immediate but transient reduction in tumor burden. However, the cytotoxic effects of chemotherapy tend to diminish quickly, and the likelihood of relapse or disease progression is generally higher once treatment ceases. As a result, improvements in PFS within chemotherapy cohorts may not consistently translate into OS gains, as disease progression may resume relatively swiftly due to the absence of a lasting anti-tumor immune response. In contrast, immune checkpoint inhibitors stimulate the host immune system to recognize and target tumor cells, inducing immune memory effects. Cytotoxic T cells, primed during immunotherapy, retain their ability to recognize and destroy residual cancer cells, contributing to long-term disease control. Additionally, immunotherapy can induce a phenomenon known as “pseudo-progression”, where the tumor temporarily appears to grow as immune cells infiltrate it, before ultimately shrinking. This reflects an ongoing immune response rather than true disease progression and may delay the appearance of progression. However, it is often followed by extended survival, further reinforcing the correlation between PFS and OS. The analysis of treatment-naïve patients in this study is characterized by robustness, highlighting the significant translational value of PFS in identifying novel surrogate endpoints within clinical trials. Traditionally, patient prognosis has been predicted using response evaluation criteria in solid tumors (RECIST) and tumor mutation burden (22); however, their clinical utility has been limited by the inherent heterogeneity within patient populations. While the occurrence of death from causes unrelated to the condition may introduce confounding factors in the assessment of OS, it is important to note that such instances constitute a minority within the dataset. This comprehensive investigation demonstrates the clinical practicality of PFS as a surrogate endpoint for early efficacy assessment.

In clinical practice, this correlation aids in making more informed decisions. For instance, if a treatment demonstrates a significant improvement in PFS, clinicians can reasonably infer that this may translate into a survival benefit, thereby supporting the continuation of the treatment regimen (23). Given the aggressive nature of SCLC, early indications of treatment efficacy are crucial for timely intervention and management. Moreover, these findings can enhance patient counseling. Patients often seek early reassurance about the effectiveness of their treatment. By leveraging PFS as a surrogate endpoint, clinicians can provide more immediate feedback on treatment response, potentially reducing patient anxiety and fostering a more proactive approach to therapy. Additionally, the stronger correlation between PFS and OS observed in immunotherapy compared to chemotherapy highlights the unique mechanisms of action of immunotherapeutic agents. Immunotherapy’s ability to induce long-term immune responses and potentially extend survival without immediate cytotoxic effects underscores its potential for sustained disease control (24). This insight can guide treatment selection, particularly when immunotherapy is a viable option, and can inform discussions about the potential long-term benefits of such treatments.

In contrast to previous investigations, the present study distinguishes itself through its reliance on a large sample size, encompassing rigorous randomized clinical trials of both phase II and phase III, as well as the inclusion of updated survival data. Notwithstanding these strengths, there are also has several limitations in this study. First, all data were derived from published clinical trials, which limited our ability to control for potential confounding factors related to baseline clinical characteristics, including but not limited to age, race, and smoking status. The inherent heterogeneity of the population represents a notable constraint. Second, only a limited subset of trials provided survival rates at specific time points, and not all trials reported survival outcomes for patient subgroups, thereby restricting the depth of analyses regarding treatment arm-level correlations. Furthermore, the inconsistency in the first-line treatment approach among the second-line treatment options introduced confounding factors that could not be controlled in the analysis of the second-line patient cohort. These factors included patient baseline information, the treatment modality or medication received in the first line of treatment, and other such inconsistencies. This might be one of the reasons for the negative results observed in this group. And the absence of statistical differences in the subgroup analysis of phase II trials may be attributable to the following factors: (I) the smaller sample size in phase II trials limits the power to detect treatment benefits and hampers the ability to perform meaningful subgroup analyses; (II) the more heterogeneity patient population in phase II trials restricts the generalizability of the findings, as a treatment that performs well in a carefully selected cohort may not yield similar results in the broader, more diverse population encountered in clinical practice; (III) the reduced rigor in phase II trials increases the risk of bias, which may skew the results and reduce their reliability compared to the more robust design of phase III trials; and (IV) the lack of a strong control group in phase II trials makes it challenging to discern whether the observed effects are truly attributable to the treatment or are simply part of the natural disease progression. Finally, the exclusive inclusion of RCTs ensured a high level of evidence quality but may limit the generalizability of our findings to real-world populations. RCTs typically employ strict eligibility criteria to ensure patient homogeneity and minimize confounding factors, which can result in the exclusion of older patients and those with severe comorbidities. These patient groups are often underrepresented in clinical trials but constitute a significant proportion of the actual patient population. For instance, older patients with SCLC often face unique challenges, including frailty, polypharmacy, and a higher prevalence of comorbid conditions such as cardiovascular disease and diabetes. These factors can influence treatment tolerance, response, and overall prognosis. Similarly, patients with comorbidities may experience altered pharmacokinetics and pharmacodynamics, affecting the efficacy and safety of interventions. The underrepresentation of such patients in RCTs means that the observed correlations between PFS and OS may not fully capture the complexities and variations seen in broader clinical practice. Moreover, the findings from RCTs may not be directly applicable to patients receiving treatments outside of clinical trial protocols. Real-world data often reveal more diverse treatment patterns, adherence issues, and patient preferences, all of which can significantly impact outcomes. Therefore, while RCTs provide robust evidence for evaluating the efficacy of interventions in controlled environments, their applicability to heterogeneous patient populations in routine clinical care settings may be limited. Future research should consider incorporating real-world data and pragmatic trials to complement the findings from RCTs. This approach can help bridge the gap between trial results and real-world outcomes, ensuring that the insights gained are more reflective of the diverse patient populations encountered in clinical practice.

Conclusions

In conclusion, this analysis represents a comprehensive investigation into the potential role of PFS as a surrogate endpoint for OS in treatment-naïve patients with SCLC. This thorough review of the existing literature and clinical trial data provides valuable insights that contribute to the ongoing discourse on appropriate endpoints in SCLC research. Ultimately, the findings of this study aim to enhance the precision and efficiency of clinical trial design, facilitating the evaluation of novel therapeutic interventions in this challenging patient population.

Supplementary

The article’s supplementary files as

Ethical Statement: The authors are accountable for all aspects of the work in ensuring that questions related to the accuracy or integrity of any part of the work are appropriately investigated and resolved.