Herein, we conducted a longitudinal ctDNA analysis integrating patient data from two clinical trials evaluating treatment regimens for LS-SCLC: one assessing CCRT alone and the other evaluating CCRT followed by serplulimab. The aim of this study was to evaluate dynamic changes in ctDNA and its integration with radiological tumor shrinkage to predict clinical outcomes, identify patients who may benefit from consolidation ICI therapy after CCRT, and monitor treatment efficacy in LS-SCLC patients.
A total of 100 patients who received CCRT alone and 44 patients who received consolidation ICI were included. The clinical characteristics of the patients are summarized in Table 1. Among the 144 patients, the majority were male (77.4%), were current/former smokers (66.0%), and had stage III disease at diagnosis (89.6%). As of April 20, 2025, disease progression had occurred in 75 patients (52.1%, 75/144), with 57 patients in the CCRT-only group and 18 patients in the consolidation ICI group. Similarly, 48 patients (33.3%, 48/144) died, with 41 and 7 patients in the CCRT-only and consolidation ICI groups, respectively. The median follow-up duration was 37.8 months (interquartile range [IQR], 33.3-44.5) for the CCRT-only group and 30.1 months (IQR, 25.5-32.1) for the ICI group.
A total of 490 plasma ctDNA samples were sequenced, comprising 251 samples from the CCRT-only group and 239 samples from the ICI group. These samples were collected at the following time points: 17 at baseline (t0), 144 after induction chemotherapy (ICT) but before thoracic radiotherapy (TRT) (post-ICT & pre-TRT, t1), 194 after completion of radiotherapy (including 114 post-TRT and 80 post-PCI, t2), 41 at progressive disease (PD, t3), 39 at the initiation of cycle 3 (C3), 29 at month 6 (M6), and 26 at year 1 (Y1) of ICI treatment (Fig. 1a). All t0 samples were ctDNA positive, with TP53 and RB1 being the most frequently altered genes, detected in 88 and 76% of patients, respectively (Fig. 1b). High concordance was observed between post-TRT and post-PCI samples (agreement: 88%; kappa coefficient: 0.65), supporting their interchangeability (Supplementary Fig. 1). Consequently, t2 ctDNA positivity was defined as positive detection in either post-TRT or post-PCI samples.
Compared with 100 patients who received CCRT alone, 44 patients who received consolidation ICI had longer median PFS (16.5 months vs. not reached [NR], Fig. 1c) and OS (49.4 months vs. NR, Fig. 1d). Given the potential for immortal time bias, time-dependent Cox regression models, which treat consolidation ICI as a time-dependent covariate, were applied, revealing that consolidation ICI was associated with a trend toward improved PFS (hazard ratio [HR], 0.76; 95% confidence interval [CI], 0.45-1.28; p = 0.294; Fig. 1c) and significantly superior OS (HR, 0.41; 95% CI, 0.19-0.91; p = 0.028; Fig. 1d) in LS-SCLC patients.
The swimmer plot summarizes the ctDNA detection and clinical outcomes of 100 CCRT-only patients (Fig. 2a). In the CCRT-only cohort, ctDNA concentrations significantly decreased from t1 to t2 (p = 0.005) but peaked at t3 (p < 0.001), with a similar pattern observed in paired t2 and t3 samples (p = 0.002) (Fig. 2b), suggesting an association between radiographic progression and ctDNA dynamics as measured by next-generation sequencing. Consistently, mean variant allele frequencies (VAFs) also declined from t1 to t2 (p = 0.004) but increased at t3 (p < 0.001), with a comparable difference observed in paired t2 and t3 samples (p < 0.001) (Supplementary Fig. 2a). The ctDNA detection dynamics tended toward decreased ctDNA detection rates from t1 to t2 (42.0% vs. 27.8%, p = 0.055) (Fig. 2c). A positive t2 ctDNA status, indicative of residual tumor presence after completing CCRT, predicted inferior PFS (HR, 2.20; 95% CI, 1.17-4.14; p = 0.015) (Fig. 2d) and OS (HR, 2.21; 95% CI, 1.03-4.77; p = 0.044) (Fig. 2e). Similarly, t1 ctDNA detection was also associated with worse PFS (HR, 2.46; 95% CI, 1.47-4.12; p < 0.001) (Fig. 2f) and OS (HR, 2.50; 95% CI, 1.34-4.67; p = 0.004) (Fig. 2g).
Given that all available t0 samples were ctDNA positive, we tentatively defined patients with negative t1 ctDNA as having achieved early ctDNA clearance at t1. Among CCRT-only patients, a trend toward inferior PFS was observed in 20 patients who achieved delayed ctDNA clearance at t2 compared with patients with early ctDNA clearance at t1 (HR, 1.81; 95% CI, 0.94-3.47; p = 0.075) (Fig. 2h). There were 15 patients with persistent ctDNA positivity from t1 to t2, who exhibited significantly poorer PFS than those with t1 clearance (HR, 3.00; 95% CI, 1.53-5.89; p = 0.001). Similar results were observed for the OS data (t2 clearance vs. t1 clearance: HR, 1.99; 95% CI, 0.91-4.36; p = 0.085; maintain positive vs. t1 clearance: HR, 2.69; 95% CI, 1.20-6.06; p = 0.017) (Fig. 2i).
Swimmer plots depict the length of follow-up, events, and ctDNA positivity (Fig. 3a). Given that t2 represents the latest time point before consolidation ICI initiation, we evaluated its potential as a predictive biomarker. Among 30 t2 ctDNA-positive patients, a trend toward superior PFS was observed in patients receiving consolidation ICI (HR, 0.51; 95% CI, 0.19-1.39; p = 0.187), whereas t2 ctDNA-negative patients receiving consolidation ICI or CCRT alone showed comparable PFS (HR, 0.99; 95% CI, 0.51-1.94; p = 0.980), suggesting the potential of t2 ctDNA as a predictive biomarker for consolidation ICI (Fig. 3b, c). Similar trends were observed for OS analysis (Supplementary Fig. 3a, b).
Since consolidation ICI significantly reduces the risk of extrathoracic metastasis and early ctDNA dynamic changes can predict chemotherapy outcomes in patients with SCLC, we further investigated whether t1 ctDNA detection could predict the benefit of ICI consolidation at an earlier time point. Among the 144 t1 samples, 56 (39%) were ctDNA positive. Consistent with t0, TP53 remained the most commonly mutated gene (61%), followed by RB1 (32%), although no copy number variation (CNV) events were detected in ≥2 patients in this subset (Fig. 3d). Instead, a single start-lost mutation in SMARCA4 was observed, highlighting differences in mutational profiles between time points. Among ctDNA-positive patients at t1, those who received consolidation ICI presented a significantly lower progression rate than those treated with CCRT alone (chi-square test, p < 0.001) (Fig. 3e). Time-to-event analysis consistently demonstrated similar findings for both PFS and OS (Fig. 3f, g). Specifically, t1 ctDNA-positive patients had longer PFS with consolidation ICI than with CCRT-only (median PFS, NR vs. 11.4 months). Time-dependent Cox regression, accounting for immortal time bias, revealed a significant association between consolidation ICI therapy and improved PFS in t1 ctDNA-positive patients (HR, 0.29; 95% CI, 0.11-0.77; p = 0.013). In addition, among t1 ctDNA-positive patients, those receiving consolidation ICI demonstrated significantly prolonged OS compared with those treated with CCRT alone (median OS, NR vs. 28.1 months; HR, 0.05; 95% CI, 0.00-0.36; p < 0.001). Univariate time-dependent Cox regression analysis identified several potential prognosis-related factors (Supplementary Fig. 4). The multivariate analysis further demonstrated that consolidation ICI were independently associated with improved PFS (HR, 0.33; 95% CI, 0.11-0.97; p = 0.045) and OS (HR, 0.07; 95% CI, 0.00-0.61; p = 0.010) in t1 ctDNA-positive patients (Fig. 3h; Supplementary Fig. 5). In contrast, among t1 ctDNA-negative patients, consolidation ICI therapy did not significantly improve either PFS (HR, 1.30; 95% CI, 0.68-2.48; p = 0.428) or OS (HR, 1.14; 95% CI, 0.46-2.84; p = 0.776) compared with those in patients receiving CCRT alone (Fig. 3i, j). Furthermore, t1 ctDNA demonstrated a higher C-index for PFS and OS prediction than that of t2 ctDNA in patients receiving consolidation ICI (Fig. 3k, l), supporting its key role in treatment decision-making and efficacy prediction.
We next explored recurrence patterns stratified by ctDNA dynamics and treatment modality. Among patients who achieved t2 clearance (ctDNA positive at t1 but ctDNA negative at t2), distant metastasis remained the predominant failure pattern in the CCRT-only group, whereas the majority of patients in the ICI group remained relapse free (p = 0.022) (Supplementary Fig. 6). These findings suggest that patients who achieve t2 clearance derive significant benefit from consolidation ICI.
No patients presented with PD at t1. As shown in Fig. 4a, t1 ctDNA positivity was significantly associated with stable disease (SD) rather than partial or complete response (PR/CR) (Fisher's exact test, p = 0.012). In addition, ctDNA concentrations were significantly higher in patients with SD than in those with PR (p = 0.001). Consistently, ctDNA-positive patients presented a decreased percentage of tumor shrinkage at t1 (p < 0.001, Fig. 4b). These findings suggest that t1 ctDNA can reflect residual tumor burden, supporting a prognostic stratification strategy combining ctDNA detection with radiological response (Supplementary Fig. 7a, b). Thus, a prognostic stratification model was developed for CCRT-only patients (Supplementary Fig. 7c). Fifty-two double responders (both radiological and molecular) to ICT had better prognosis than 34 partial responders (either radiological or molecular), whereas the 14 nonresponders (neither radiological nor molecular) had the poorest prognosis (Supplementary Fig. 7d, e).
To identify patients with a prolonged prognosis under CCRT, a tumor shrinkage threshold of 60% (75 percentile) was further applied (Fig. 4c). The low-risk cohort comprised 19 patients with tumor shrinkage ≥60% and a t1 ctDNA-negative status, with high 4-year PFS and OS rates of 68.4% and 78.9%, respectively (Fig. 4d). The mid-risk group included 38 patients with tumor shrinkage <60% and t1 ctDNA-negative status and 6 patients with tumor shrinkage ≥60% and t1 ctDNA-positive status. Patients with tumor shrinkage <60% and a t1 ctDNA-positive status were classified as high risk. The mid-risk group exhibited lower 2- and 4-year PFS rates than the low-risk group (2-year, 54.5% vs. 68.4%; 4-year, 43.6% vs. 68.4%). High-risk patients had significantly worse PFS (HR, 4.61; 95% CI, 1.90-11.14; p < 0.001) and OS (HR, 4.26; 95% CI, 1.46-12.43; p = 0.008) than low-risk patients (Fig. 4d).
Furthermore, in high-risk patients, significantly improved PFS (HR, 0.24; 95% CI, 0.08-0.75; p = 0.014) and OS (HR, 0.06; 95% CI, 0.00-0.42; p = 0.001) were observed in those treated with consolidation ICI than in those receiving CCRT alone (Fig. 4e). However, limited short-term benefits were observed in mid- and low-risk patients (Fig. 4f). To determine independent prognostic significance, variables with p values < 0.1 in the univariate time-dependent Cox analysis were included in a multivariate model adjusted for potential confounders (Fig. 4g; Supplementary Fig. 8). In the high-risk group, which exhibited a poor prognosis under CCRT, consolidation ICI therapy provided significant benefits. In contrast, the low-risk group, with a favorable prognosis under CCRT, showed limited benefits from consolidation ICI. For the mid-risk group, which had an acceptable short-term prognosis but a poor long-term prognosis under CCRT, limited short-term benefits were detected. However, the potential long-term benefits remain unclear due to the moderate follow-up duration. The optimal treatment and the efficacy of consolidation ICI require further investigation.
In addition to serial samples collected at t1, t2, and t3, the consolidation ICI cohort underwent additional ctDNA surveillance prior to progression during ICI treatment, including samples collected at C3, M6, and Y1 (Fig. 3a). Similar to the CCRT-only cohort, patients in the consolidation ICI cohort presented a decrease in both ctDNA concentrations and mean VAFs after RT, followed by a peak at the time of progression, with consistent trends observed in the matched t2 and t3 samples (Fig. 5a; Supplementary Fig. 2b). Among the six patients with ctDNA-positive status at t2 and paired C3 samples, one (16.7%) achieved ctDNA clearance at C3 and remained alive and progression free after 33.0 months of follow-up. The ctDNA detection rates and concentrations did not change significantly during consolidation ICI treatment. Notably, ctDNA detection at later surveillance time points appeared to predict worse PFS, especially at Y1 (HR, 7.95; 95% CI, 1.10-57.36; p = 0.040) (Fig. 5b). During surveillance, ctDNA was newly detected in three patients, one at C3 and two at Y1, all of whom developed disease progression within two years. Compared with patients with acquired ctDNA detection, those without acquired ctDNA detection exhibited better PFS (HR, 0.30; 95% CI, 0.08-1.05; p = 0.059) (Fig. 5c). In addition, among the 28 patients with serial samples between C3 and Y1, those with persistently negative ctDNA (n = 20) exhibited prolonged PFS compared with those with at least one detectable ctDNA sample (HR, 0.08; 95% CI, 0.01-0.42; p = 0.003) (Fig. 5d). However, due to the limited sample sizes within each subgroup, these findings should be interpreted with caution and warrant further validation in larger prospective cohorts.