Personalized tumor-loaded monocyte-derived dendritic cell vaccination in combination with atezolizumab as maintenance treatment in extensive-stage small cell lung cancer (ES-SCLC): phase Ib-II VENEZOLUNG trial

Abstract

Background

Despite the incorporation of atezolizumab, an anti-programmed death-ligand 1 (PD-L1) antibody, into first-line chemotherapy regimen, survival in patients with extensive-stage small cell lung cancer (ES-SCLC) remains poor. Given the role of dendritic cells (DCs) in modulating responses to anti-programmed cell death protein 1 (PD-1) antibodies, a promising avenue is the development of vaccines based on DCs, together with the evaluation of how DC phenotypic modifications shape treatment efficacy.

Materials and methods

This is a single-arm, phase Ib/II, open-label clinical trial investigating the safety and activity of combining autologous tumor lysate pulsed DCs administered intradermally with intravenous atezolizumab as maintenance therapy for patients with ES-SCLC. Following leukapheresis, monocytes were differentiated ex vivo into mature DCs and loaded with irradiated autologous tumor lysate. After standard induction therapy (four cycles of carboplatin, etoposide, and atezolizumab), patients received up to six doses of mature DCs intradermally, along with intravenous atezolizumab 1,200 mg every 3 weeks, until disease progression.

Results

20 patients were enrolled, including six patients with brain metastases. 18 patients received DC vaccination, with a median of three doses (range 1–6). Most grade 3–4 treatment-related adverse events were hematological and related to chemotherapy. DC vaccination was well tolerated, with only grade 1–2 transient skin irritation at the injection site. Among 18 evaluable patients, 17 (94.4%) achieved an objective response. After a median follow-up of 31.4 months, progression-free survival at 2 years (PFS_2y) was 20% (95% CI 6.4% to 39.1%), median overall survival (OS) 11.2 months (95% CI 6.3 to 26.0) and OS_3y 22.5% (95% CI 6.8% to 43.6%). Six patients achieved long survival (median 30.5 months, range 23.4–36.2). Expansion of conventional type DC1 (cDC1s) XCR1⁺ and stem-like progenitor-exhausted (CXCR5⁺PD-1⁺) CD8⁺ T cells following chemotherapy plus atezolizumab was associated with improved PFS (HR 0.94, 95% CI 0.90 to 0.99 and HR 0.59, 95% CI 0.39 to 0.89).

Conclusion

DC immunization combined with atezolizumab maintenance following standard chemotherapy plus atezolizumab treatment is well tolerated. Its clinical activity cannot be definitively determined in this study, as it was administered as maintenance therapy; however, the observed rate of long-term survivors suggests a potential benefit. Our findings suggest a pivotal role for cDC1s in mediating the antitumor activity in long-term survivors.

Trial registration number

NCT04487756.

WHAT IS ALREADY KNOWN ON THIS TOPIC

• Patients with extensive-stage small cell lung cancer treated with programmed death-ligand 1 antibodies plus chemotherapy have poor survival (progression-free survival at 2 years (PFS_2y) 3–11% and overall survival (OS)_3y 16–18%).

WHAT THIS STUDY ADDS

• In the VENEZOLUNG trial, dendritic cell vaccination plus atezolizumab maintenance was feasible and well tolerated, with PFS_2y of 20% and OS_3y of 22.5%. Clinical benefit correlated with expansion of XCR1⁺ conventional type dendritic cell 1 (cDC1) and CD8⁺ progenitor-exhausted (CXCR5⁺PD-1⁺) CD8⁺ T (TPEX) cells.

HOW THIS STUDY MIGHT AFFECT RESEARCH, PRACTICE OR POLICY

• These findings encourage further research on strategies to enhance the cDC1-TPEX axis and to integrate approaches that improve dendritic cell function to strengthen responses to PD-L1 blockade.

Introduction

Small-cell lung cancer (SCLC) is a highly aggressive malignancy strongly associated with tobacco exposure and accounts for 15% of all lung cancers. For decades, platinum-etoposide chemotherapy was the only standard treatment, providing limited benefit with a median overall survival (OS) less than 1 year in extensive-stage disease (ES-SCLC).1 The advent of immune checkpoint inhibitors, mainly anti programmed cell death protein 1 (PD-1)/programmed death-ligand 1 (PD-L1) antibodies, has modestly improved outcomes. In the IMpower133 trial, adding the PD-L1 inhibitor atezolizumab to chemotherapy median OS increased to 12.3 months compared with 10.3 months with chemotherapy alone (HR=0.70; 95% CI 0.54 to 0.91),1 while OS_3y was 16% (95% CI 11% to 21%).2 A second phase III study, the CASPIAN trial, confirmed that a different anti-PD-L1 antibody, durvalumab, combined with platinum-etoposide also improved OS (HR=0.71; 95% CI 0.60 to 0.86), OS_3y was 17.6% (95% CI 13.3% to 22.4%)3 and progression-free survival at 2 years (PFS_2y) was 11% (95% CI 7.5% to 15.2%).4 Despite these advances, most patients derive limited benefit, underscoring the need for improved therapeutic strategies.

Dendritic cells (DCs) are central regulators of antitumor immunity, but their function is often impaired in cancer. DCs are central orchestrators of immune responses, capable of activating naïve T cells. Following antigen uptake, DCs mature, upregulate costimulatory molecules (CD80, CD86, CD40), secrete pro-inflammatory cytokines, and migrate to lymph nodes to prime T helper type 1 responses and cytotoxic T lymphocytes.5 6 Conventional DC type 1 (cDC1) in the mature state are defined by XCR1 expression. They are key mediators of tumor antigen cross-presentation and CD8⁺ T-cell priming. These cells support the maintenance of stem-like progenitor-exhausted (CXCR5⁺PD-1⁺) CD8⁺ T cells (TPEX) that drive responses to immune checkpoint blockade. Accordingly, the presence of cDC1 in the tumor microenvironment, and their interaction with CD8⁺ T cells in tumors has been consistently associated with improved clinical responses to immunotherapy.7 8

Ex vivo generation and modification of monocyte-derived DCs can restore their function, enhancing type I interferon production and costimulatory signaling. As monotherapy, DC vaccines have shown limited clinical efficacy.9 However, their combination with checkpoint inhibitors is particularly attractive as PD-L1 expressed on DCs restricts T-cell priming by sequestering B7.1. Blocking PD-L1 on DCs disrupts the PD-L1-mediated sequestration of B7.1, thereby enabling enhanced T-cell priming through the B7.1-CD28 interaction.5 Thus, combining DC vaccination with PD-L1 blockade may synergistically enhance antitumor immunity by simultaneously restoring DC function and releasing PD-1 mediated inhibition of effector T cells. Previous data support this rationale in hepatocellular carcinoma9 and other tumor types.10 11 Moreover, our previous data in a model of chronic viral infection and other studies support an important role of cDC1s (XCR1⁺) during anti-PD-1 checkpoint inhibitor therapy through the regulation of exhausted CD8⁺ T-cell states: they are dispensable for the proliferative burst of progenitor-exhausted (CXCR5⁺PD-1⁺) CD8⁺ T cells, but required to sustain the effector function of more differentiated exhausted CD8⁺ T cells, underscoring a key role for cDC1s in maintaining effective CD8⁺ immunity under checkpoint therapy.7 8 12

In this study, we evaluated the activity of a DC-based vaccine generated from autologous peripheral blood monocytes, matured and loaded with autologous tumor lysate, in combination with the PD-L1 inhibitor atezolizumab administered as maintenance therapy following standard first-line chemo-atezolizumab in patients with ES-SCLC.

Methods

This is an independent investigator-initiated trial, single arm, phase Ib/II multicenter open-label study of atezolizumab plus autologous DC product obtained from peripheral blood monocytes stimulated with autologous tumor antigens as maintenance treatment in ES-SCLC. Eligible patients were adults with histologically confirmed ES-SCLC who had not received prior treatment for advanced disease and had measurable disease according to Response Evaluation Criteria in Solid Tumors (RECIST) 1.1. Patients were required to have an ECOG (Eastern Cooperative Oncology Group) performance status of 0–1, a life expectancy of at least 16 weeks, and adequate hematologic, renal, and hepatic function. Additional key criteria included the ability to undergo leukapheresis, and controlled or asymptomatic brain metastases if present. All patients provided written informed consent prior to enrollment. The study was conducted in three hospitals in Barcelona (Spain). DC product was developed under GMP conditions at the Immunology Department laboratory of Hospital Clinic, Barcelona, Spain.

Treatment schedule

The timing of study procedures was standardized as follows: a tumor biopsy to obtain material for lysate preparation was performed within 1 week if archival tissue was not available, following confirmation of eligibility during screening assessments. Leukapheresis was scheduled either before or after initiation of the first treatment cycle, depending on apheresis unit availability, ensuring that treatment initiation was not delayed by more than 1 week. DC vaccine production required approximately 1 week, after which the final product was cryopreserved and stored until administration.

After four cycles of standard treatment induction with carboplatin and etoposide, plus atezolizumab, there was a maintenance phase with the administration of atezolizumab and DC immunotherapy. The induction phase consisted of four 21-day cycles of carboplatin area under the curve 5, administered intravenously on day 1 of each cycle and etoposide 100 mg per square meter of body-surface area, administered intravenously on days 1 through 3 of each cycle with atezolizumab at a dose of 1,200 mg, administered intravenously on day 1 of each cycle. The induction phase was followed by a maintenance phase during which patients received atezolizumab at a dose of 1,200 mg, administered intravenously on day 1 of each cycle with DC immunotherapy intradermally for a maximum of six doses, every 3 weeks for the first four doses and every 12 weeks for the last two doses. Continuation of the trial regimen after the occurrence of disease progression during either phase was allowed if there was evidence of clinical benefit (see online supplemental figure 1).

DC immunotherapy was based on the generation of the product from peripheral monocytes obtained by apheresis in an Amicus separator (Fresenius Kabi, Germany) performed before the first cycle of induction phase or between days 10 and 19 of the first induction cycle. Apheresis was performed to obtain 5×10⁸ monocytes per patient collection. Autologous monocytes were selected by adherence to culture flasks and then differentiated to DCs by culturing adherent monocytes for 7 days in X-VIVO 15 (Lonza, Walkersville, Maryland, USA) supplemented with 2% autologous serum, 800 U/mL GM-CSF (Miltenyi Biotech) and 500 U/mL IL-4 (Miltenyi Biotec) and then for an additional 24 hours in the presence of 20 ng/mL tumor necrosis factor-α (Miltenyi Biotech), 10 ng/mL IL-1β (Miltenyi Biotech), 20 ng/mL IL-6 (Miltenyi Biotech), 1 µg/mL prostaglandin E2 (Dinoprostona; Pfizer, New York, USA), 20 µg/mL poly (I:C) (Hiltonol; Oncovir, Washington DC, USA) (courtesy of Dr Andres Salazar) and autologous tumor lysate, using good manufacturing practices standard procedures.

Tumor lysates were prepared from fresh or formalin-fixed paraffin-embedded tumor tissue using physical disruption under sterile conditions. Briefly, tumor samples were mechanically dissociated and subjected to repeated freeze-thaw cycles (liquid nitrogen—water bath at 37°C) to induce cell lysis and release a broad repertoire of tumor-associated antigens. The resulting material was centrifuged to remove cellular debris, and the supernatant containing soluble tumor antigens was collected, irradiated, and stored under controlled conditions (–20°C) until use. Autologous DCs were subsequently pulsed ex vivo with the tumor lysate to facilitate antigen uptake, processing, and presentation prior to administration to the patient. The tumor lysate was generated from biopsy cores measuring 1.5 mm in diameter. Due to the small volume of the tumor sample, vaccine production was prioritized, and unfortunately, it was not feasible to assess tumor-reactive responses.

Maturation was confirmed by immunophenotyping (see online supplemental figure 2). After 1 week of culture, fully mature DCs with release criteria including >80% CD80⁺, CD83⁺, CD86⁺, HLA-DR+, absence of T lymphocytes (CD3⁺) and monocytes (CD14⁺) (less than 15% CD14, CD3 and CD19 positive cells) and negative microbial test (bacterial, fungi, Mycoplasma), which constitute the product, were aliquoted, cryopreserved and stored in nitrogen until its use. Each vial containing 10×10⁶ of matured DCs was thawed at 37°C and injected intradermally. Patients received up to six doses of DC at maintenance phase (week 1, 4, 7, 10, and week 22 and 34), thereafter until progressive disease (maximum of six doses).

Trial design

The two primary endpoints were the investigator-assessed toxicity and PFS rate at 6 months, both in the intention-to-treat population. In the phase Ib safety lead-in portion, six patients were enrolled into and evaluated for dose-limiting toxicities during the first two cycles of the maintenance therapy. If less than two out of six patients present with limiting toxicities, the investigational products administration would be deemed safe in the phase Ib lead-in; therefore, enrollment would be expanded into the phase II cohort including up to 20 patients.

Adverse events (AEs) were assessed throughout the study period and evaluated by using National Cancer Institute Common Technology Criteria version of Adverse Events V.5.0. Tumor measurements were performed by CT scan at baseline, at week 6, at week 12, and every 9 weeks thereafter, to evaluate tumor response and PFS. Response was evaluated using the RECIST 1.1.

Immune cell phenotyping

Peripheral blood samples were collected pre-induction treatment (“pre-induction sample” T0), pre-maintenance phase (“post-induction sample” T1), after 3 weeks of the first DC immunotherapy administration (T2), after 9 weeks of the first DC immunotherapy administration (T3), after completing the DC immunotherapy (“post-vaccination samples”, T4) and at progression of disease (T5).

Cryopreserved PBMCs (Peripheral Blood Mononuclear Cells) collected from patients at distinct time points (pre-induction, post-induction and post-vaccine) were thawed, washed, and maintained in cRPMI (RPMI (Gibco) supplemented with 10% FBS (Fetal Bovine Serum), 1% pen/strep, 0.05 mM β-mercaptoethanol, and 1 mM Sodium Pyruvate) (see online supplemental table 1). For flow cytometry analysis, cells were plated and stained with viability dyes (Live/Dead Fixable Violet (564996, BD) or Fixable Viability Stain 780 (ref 565388, BD) and Fc Block (ref 564220, BD) diluted 1:50 with phosphate-buffered saline (PBS) to ensure data quality and prevent non-specific binding. For more details about the specific markers and antibody references used for each panel, see online supplemental table 2.

Extracellular staining with specific antibodies was performed to identify regulatory T cell (Treg), γδ T cells, DCs, and exhausted T-cell populations using FACS (Fluorescence-Activated Cell Sorting) buffer (PBS supplemented with 5% FCS (Fetal Calf Serum), 0.5% BSA (Bovine Serum Albumin), and 0.07% NaN₃ (sodium azide). For deeper functional characterization, intracellular staining was conducted following cell fixation (2% formaldehyde) and permeabilization using Perm/Wash buffer (PBS with 1% FCS, 0.1% NaN₃, and 0.1% Saponin). Treg populations were further defined by FOXP3 expression, while T-bet and ROR-γt transcription factors were analyzed in γδ T cells to determine their potential for interferon-γ or interleukin (IL)-17 secretion, respectively. Samples were fixed in FACS Fix buffer and acquired on an Aurora 4L (Cytek) flow cytometer. Data were analyzed using FlowJo V.10 software (The Star). For details of gating strategies for all analyzed populations, see online supplemental figure 3.

Statistical analysis

Sample size calculation for this study is based on PFS at 6 months of 30.9% observed in the IMpower133 trial.1 To obtain a power of 80% to detect differences in the contrast of the null hypothesis (PFS at 6 months, H₀: p=31%) using a bilateral χ² test for one sample, taking into account that the level of significance is (alpha) 5%, and assuming that the expected proportion (PFS at 6 months) is 61%, it will be necessary to include at least 20 patients in the study.

To evaluate longitudinal changes in circulating immune cell subsets, percentages and mean fluorescence intensities were compared between time points (pre-induction, post-induction and post-vaccination) using paired Wilcoxon signed-rank tests. Furthermore, all immunological parameters were correlated with clinical outcomes, specifically OS and PFS of the patients. Statistical analyses and graphical representations were performed using GraphPad Prism V.9. Correlations between variables were assessed via Spearman rank correlation tests. Statistical significance was defined as p<0.05 and denoted in figures as follows: *p<0.05; **p<0.01; ***p<0.001; ****p<0.0001.

Changes in percentages of immune cell subsets were analyzed for timed endpoints (PFS and OS) with a multivariate Cox regression model. For binary outcomes (ORR -overall response rate-) multivariate logistic regression models were used; for continuous outcomes, continuous regression models were used. Analysis was performed with SAS V.9.4.

Data cut-off for this final analysis was September 26, 2025.

Results

Recruitment has been completed with a total of 20 patients enrolled, including 6 (30%) patients with brain metastases. Other patients’ characteristics are presented in table 1. Most patients had tumors PD-L1 negative (75%), ASCL1 dominant subtype (58.3%) and high tumor mutation burden (25, range 20–30 mut/Mb) (table 1). The most common mutated genes were TP53 (10/12 cases; 83%), RB1 (8/12; 67%) and PTEN (3/12; 25%); followed by KMT2C, FAT1, EPHA6, EP300 and ATRX (see online supplemental table 4).

Table 1.

Patient characteristics

Baseline demographics
	Total (N=20)
Sex, n (%)
Female	10 (50.00)
Race, n (%)
White non-Spanish	18 (90.00)
White Spanish	1 (5.00)
UK	1 (5.00)
ECOG, n (%)
0	5 (25.00)
1	15 (75.00)
Age years, median (range)	61 (53–75)
Number of metastasis sites, median (range)	2 (0–6)
Metastatic sites, n (%)
Bone metastasis	7 (35)
Brain metastasis	6 (30)
Liver metastasis	7 (35)
Pericardium	2 (10)
Pleura	4 (20)
Adrenal	4 (20)
Renal	1 (5)
IHC subtype, n (%)*	Total (N=12)
IHC-A/A only	7 (58)
IHC-AN/N	3 (25)
IHC-P	1 (8)
IHC-TN	1 (8)
IHC PD-L1 <1%, n (%)*	9 (75)
TMB, median (range)*	25 (20–30)

'IHC: Immunohistochemistry; ECOG: Eastern Cooperative Oncology Group; TMB: Tumor Mutational Burden

*Analyzed in 12 cases with available tumor tissue sample; IHC-A: ASCL1 dominant; IHC-A only: ASCL1 only; IHC-AN: ASCL1/NEUROD1 doble-positive; IHC-N: NEUROD1 dominant; IHC-P: ASCL1/NEUROD1 doble-negative and POU2F3-positive; IHC-TN (triple negative): ASCL1/NEUROD1 doble-negative and POU2F3-negative.

PD-L1, programmed death-ligand 1.

In 18 patients a tumor biopsy was performed before apheresis in order to obtain fresh non-paraffin-embedded tissue used to pulse DCs; in two cases, DCs were pulsed with archival paraffin-embedded tissue. The apheresis was performed in 10 patients before administration of the first cycle of treatment. In the remaining 10 cases, apheresis was performed after receiving the first cycle of chemotherapy plus atezolizumab. Six doses of the DC vaccine were produced for each patient, except in two cases in which the number of generated DCs yielded only two and five doses, respectively. DCs production fulfilled the release criteria showing purity up to 80% and an increased expression of costimulatory (CD80), maturation (CD83), antigen presenting (major histocompatibility complex (MHC) class II) and cytokine receptor (CCR7) molecules on DCs surface membrane when compared with immature DCs. We did not observe differences between when cells were produced before or after the first cycle of chemotherapy plus atezolizumab.

All patients completed the four cycles of induction with chemotherapy plus atezolizumab, except two cases due to early deaths non-drug related. The remaining 18 patients were treated during the maintenance phase with atezolizumab plus DC vaccination. The median number of vaccine doses received was 3 (range 1–6). Four patients completed the DC vaccination with all the produced doses (including two cases with only two and five doses produced instead of six planned). The reasons for non-receiving the full number of doses of DC vaccine in 14 patients, was progression of the disease.

During the chemotherapy plus atezolizumab induction phase, most patients achieved an objective response (94%) including one complete response (5%) (figure 1). Treatment-related adverse events (TRAEs) were observed in 100% of patients (see online supplemental tables 5 and 6, figure 2). Grade 3–4 TRAEs were observed in 11 (55%) of patients. Most atezolizumab TRAEs were grade 1–2, and consisted of asthenia (35%), skin toxicities/pruritus (25%) and elevated transaminases (15%) (see online supplemental table 7). Toxicities related to DC vaccination were uncommon and mainly consisted of grade 1–2 site irritation (15%) (all short duration and spontaneously resolved) (see online supplemental table 8). Toxicities related to chemotherapy, most were hematological toxicities (anemia 45%, neutropenia 45% and thrombocytopenia 25%) (see online supplemental table 9).

Figure 1.

Longitudinal changes in tumor burden over time for individual patients. The percentage change in tumor size relative to baseline is shown at each radiological assessment. Each line represents one patient and is color-coded according to best overall response: CR (green), PR (yellow), and SD (red). The horizontal dashed line indicates no change from baseline. CR, complete response; PR, partial response; SD, stable disease.

Figure 2.

Treatment-related AEs according to maximum toxicity grade experienced by each patient during the study. Bars represent the percentage of patients experiencing each adverse event, stratified by NCI toxicity grade (G1–G4). Each patient is counted once per adverse event at the highest grade observed. Percentages are calculated based on the total number of treated patients. AEs, adverse events; ALT. Alanine Aminotransferase; AST: Aspartate Aminotransferase; NCI, National Cancer Institute.

After a median follow-up of 31.4 months, median PFS was 4.7 months (95% CI 4.2 to 6.9), PFS_6m 30% (95% CI 12.25% to 50.14%) and PFS_2y was 20% (95% CI 6.4% to 39.1%) (figure 3A). Median OS was 11.2 months (95% CI 6.3 to 26.0), OS_2y 30% (95% CI 12.2% to 50.1%) and OS_3y 22.5% (95% CI 6.83% to 43.66%) (figure 3B). Six patients achieved long survival (median 30.5 months, range 23.4–36.2), four without progression after more than 23 months (23.4+ months, 26.3+ months, 27.4 months and 35.5+ months) (figure 4). Long-term survival was not significantly associated with clinical or molecular characteristics, although no patients with bone metastases were long-term survivors, 50% of long-term survivors had brain metastases and the number of metastatic sites tended to be lower in long-term survivors (see online supplemental table 10, online supplemental table 11 and online supplemental figures 4 and 5).

Figure 3.

Progression-free survival (A) and overall survival (B). Time is expressed in months from treatment initiation. Tick marks indicate censored observations. Median survival times with 95% CIs, number of events, and number of censored patients are shown within each panel. The number of patients at risk at selected time points is displayed below each plot. OS, overall survival; PFS, progression-free survival.

Figure 4.

Treatment exposure, vaccination timeline, and clinical events by patient. Swimmer plot depicting treatment duration, vaccination schedule, and clinical events for individual patients. Each horizontal bar represents one patient, with time expressed in months from treatment start date. The solid line indicates duration of study treatment. Triangles denote initiation of dendritic cell vaccination, stars indicate individual vaccine doses, diamonds represent PD, and squares indicate death (exitus). Patients are displayed according to total treatment duration. PD, disease progression.

Dynamic of circulating immune cells along treatment

Immune cell subtype dynamics correlates with progression-free survival after induction with chemotherapy plus atezolizumab

We explored the association between changes in circulating immune cell populations from pretreatment (pre-induction, T0) to post-chemotherapy plus atezolizumab (post-induction, T1) and PFS using univariate Cox proportional hazards models (see online supplemental table 12). Increased proportions of XCR1-expressing cDC1 were associated with a reduced risk of progression, both when expressed as a percentage of total cDC1 (HR 0.945, 95% CI 0.900 to 0.993; p=0.0253) and as a percentage of total cDCs (HR 0.209, 95% CI 0.044 to 0.994; p=0.0492). Within the CD8⁺ T-cell compartment, a higher proportion of TPEX was strongly associated with prolonged PFS (HR 0.597, 95% CI 0.399 to 0.893; p=0.0121) (see online supplemental figure 6). Similarly, a higher frequency of Vδ2 TCR+ γδ T cells was significantly associated with improved PFS (HR 0.935, 95% CI 0.879 to 0.993; p=0.0298). In contrast, increased expression of TIGIT on terminally exhausted CD8⁺ T cells (CXCR5⁻ PD-1⁺) (TEX) was associated with an increased risk of progression (HR 1.088, 95% CI 1.009 to 1.172; p=0.0275). No significant associations with PFS were observed for regulatory T-cell subsets or γδ T cells.

Changes in immune cell populations between pre-induction (T0) and post-induction (T1) revealed a significant positive association between dendritic cells and TPEX. Specifically, the variation in XCR1⁺ cDC1 frequency within the cDC1 compartment positively correlated with the variation in CXCR5⁺ PD-1⁺ CD8⁺ non-naïve T cells (TPEX). This correlation was statistically significant using both Pearson (r=0.6759, p=0.0158) and Spearman (r=0.6434, p=0.0278) analyses (see online supplemental figure 7). Notably, patients with longer survival tended to cluster in the subgroup showing increases in both populations, suggesting coordinated changes between antigen-presenting cells and progenitor-exhausted CD8⁺ T cells after chemotherapy plus atezolizumab (figure 5 and see online supplemental figure 8).

Figure 5.

Coordinated changes in XCR1⁺ cDC1 and TPEX prior to vaccination. Scatter plot showing the correlation between changes in the frequency of XCR1⁺ cDC1 within the cDC1 population and changes in CXCR5⁺ PD-1⁺ CD8⁺ non-naïve T cells (TPEX) between screening and the pre-vaccination time point (Δ pre-vaccine—screening). Each dot represents one patient; long survivors are shown in orange and non-long survivors in blue. A positive association between ΔXCR1⁺ cDC1 and ΔTPEX was observed using both Pearson (r=0.6759, p=0.0158) and Spearman (r=0.6434, p=0.0278) correlation analyses. The regression line represents the linear fit. These data suggest coordinated modulation of dendritic cell and progenitor-exhausted CD8⁺ T-cell compartments prior to vaccination. cDC1, conventional type dendritic cell 1; PD-1, programmed cell death protein 1; TPEX, progenitor-exhausted CD8⁺ T cells.

Immune cell subtype dynamics correlate with overall survival after induction with chemotherapy plus atezolizumab

We next evaluated associations between immune cell dynamics before (T0) and after the induction phase (T1) and OS (see online supplemental table 13).

An increased proportion of plasmacytoid dendritic cells (pDCs) among CD45⁺ cells was strongly associated with worse OS (HR 13.146, 95% CI 1.507 to 114.657; p=0.0197). Likewise, a higher percentage of cDC1 within the total CD45⁺ compartment was associated with inferior OS (HR 2.586, 95% CI 1.008 to 6.631; p=0.048) (see online supplemental table 13). In contrast, several CD8⁺ T-cell parameters were associated with improved OS. Increased levels of TPEX cells were associated with improved OS (HR 0.744, 95% CI 0.561 to 0.986; p=0.0398) (figure 6). A trend toward improved OS was also observed for CXCR5⁻ PD-1⁺ TEX cells, although this did not reach statistical significance (p=0.0504). No other immune parameters analyzed pre and post induction phase were significantly associated with OS.

Figure 6.

OS according to dynamics of CXCR5⁺PD1⁺ CD8⁺ T cells (TPEX) after chemotherapy plus atezolizumab (induction phase). Patients with an increase in TPEX had longer OS (cut point 0.42): median OS 31.1 months (95% CI 10.8 to NR) versus 6.8 months (95% CI 5 to 7.5) (p=0.0011). NR: not reached;OS, overall survival; TPEX, progenitor-exhausted CD8⁺ T cells.

Immune cell subtype dynamics correlate with overall survival after maintenance with DC vaccination plus atezolizumab

Finally, we analyzed immune cell changes from pre-induction (T0) to post-vaccination (maintenance phase, T2-4) in relation to OS (see online supplemental table 14).

Consistent with findings during induction, increased proportions of immature pDCs among CD45⁺ cells remained significantly associated with worse OS (HR 8.661, 95% CI 1.050 to 71.464; p=0.045). Similarly, a higher frequency of cDC1 within the CD45⁺ compartment was associated with inferior OS (HR 2.924, 95% CI 1.012 to 8.447; p=0.047) (see online supplemental table 14 and online supplemental figure 9). In the CD8⁺ T-cell compartment, a significant association emerged for CXCR5⁻ PD-1⁺ TEX cells, where higher proportions correlated with improved OS during the maintenance phase (HR 0.895, 95% CI 0.825 to 0.972; p=0.008) (see online supplemental table 14). No statistically significant associations were observed for TPEX cells during maintenance, although HRs remained below 1. Other immune cell subsets, including Tregs and γδ T cells, were not significantly associated with OS in this phase.

Discussion

Despite the incorporation of PD-L1 inhibitors into first-line treatment, outcomes for patients with ES-SCLC remain poor, with only a minority achieving durable benefit. In this investigator-initiated phase Ib/II trial, we evaluated the safety, clinical activity, and immune correlates of combining a personalized autologous DC vaccine with atezolizumab as maintenance therapy following standard chemo-immunotherapy. Our results indicate that this strategy is feasible and well tolerated and results suggest a possible increase in the proportion of long-term survivors that compares favorably with historical data, while providing novel translational insights into the role of DC-CD8⁺ T-cell interactions during PD-L1 blockade.

The safety profile observed in this study was favorable and consistent with prior experience using this DC platform in other malignancies. Our previous studies in other tumors, including patients with diffuse pontine glioma13 (NCT02840123) and colorectal cancer14 (NCT01413295), have also demonstrated the safety of this DC immunotherapy generated at our laboratory in similar conditions, including also the DC immunotherapy in combination with an anti-PD-L1 antibody in metastatic colorectal cancer demonstrating a good safety profile.15 TRAEs in the present study were largely driven by chemotherapy, with immune-related toxicities from atezolizumab mostly grade 1–2, and DC vaccination associated only with mild, transient injection-site reactions. Importantly, the combination of DC vaccination with PD-L1 blockade did not reveal any new safety signals, supporting the feasibility of this approach in ES-SCLC. In addition, neither chemotherapy nor the DC product appeared to adversely affect the immune response, although the design of our study does not allow definitive conclusions to be drawn on this matter.

In terms of activity, although the trial did not reach the prespecified rate of PFS at 6 months, a proportion of patients remained progression-free at 2 years (20%) and alive at 3 years (22.5%).2–4 Median PFS and OS in the current study were within the range reported in pivotal randomized trials such as IMpower133 and CASPIAN, in which median OS ranged from 12.3² to 12.9³ months and median PFS from 5.1³ to 5.2 months². However, cross-trial comparisons should be interpreted with caution due to differences in study design, patient populations, and sample size. In particular, the small, single-arm nature of this phase Ib/II study limits the robustness of long-term survival estimates. While the proportion of long-term survivors (PFS_2y 20%) observed in our cohort may appear numerically higher than that reported in historical trials (PFS_2y 11%),3 these observations are descriptive and not intended for formal comparison. Similarly, comparisons with real-world cohorts should be interpreted cautiously (PFS_2y 5%13 to 8%).14 Overall, these findings should be considered exploratory and hypothesis-generating, and do not allow definitive conclusions regarding clinical efficacy.

Previous studies with DC-based vaccines in SCLC are scarce. Two studies from the same group reported limited clinical efficacy in SCLC with a DC-based vaccine transduced with an adenovirus expressing p53, used as maintenance treatment,16 17 but no studies had been previously performed combining DC-based vaccines with anti-PD-1/PD-L1 antibodies in SCLC.

Beyond clinical outcomes, a central contribution of this study lies in its detailed longitudinal immune monitoring, which provides a translational bridge between our previously published mechanistic work12 and human immunotherapy. Growing evidence supports a key role for cDC1 in the initiation of tumor-specific CD8⁺ T-cell responses within the tumor microenvironment, particularly in patients receiving immune checkpoint blockade.18 cDC1s represent a distinct and relatively infrequent dendritic cell population that, on activation, upregulate antigen-presentation machinery and co-stimulatory molecules, including MHC class I, MHC class II, and CD40, as well as characteristic markers such as XCR1. In preclinical models, we and others demonstrated that cDC1s are essential for sustaining the effector function of CD8⁺ T cells during PD-L1 blockade.8 12 In tumors, these cells facilitate antitumor immunity by recruiting T cells through chemokine production and by cross-presenting tumor-derived antigens, ultimately supporting effective T-cell-mediated immune responses.18 19

The present study extends these observations into the clinical setting. After the induction phase with chemotherapy plus atezolizumab, improved PFS and OS were associated with the increase in frequencies of TPEX, consistent with accumulating evidence that these progenitor-like exhausted T cells represent the principal reservoir responsive to immune checkpoint blockade. In parallel, increased proportions of cDC1s were associated with improved PFS, supporting the concept that effective antigen cross-presentation and DC and T-cell crosstalk underpin early clinical benefit from PD-L1 inhibition. Notably, treatment-induced changes in cDC1s correlated with the expansion of TPEX cells, whereas no such relationship was observed at baseline, underscoring the importance of immune dynamics rather than static biomarkers. Our longitudinal immune profiling links the expansion of circulating cDC1 and TPEX cells after chemotherapy plus atezolizumab with improved clinical outcome and shows that long survivors preferentially exhibit an increase in TPEX cells following induction chemo-immunotherapy. This extends prior work demonstrating that cDC1 support stem-like/TPEX CD8⁺ states and anti-PD-1 antibody responsiveness, largely established in preclinical systems and tumor-tissue studies, by providing prospective, patient-level, peripheral blood evidence of a coordinated cDC1 and TPEX axis associated with benefit in an immune-refractory cancer such as ES-SCLC.

In contrast, higher expression of inhibitory receptors such as TIGIT on TEX cells after induction was associated with inferior PFS, highlighting that deep exhaustion states may limit responsiveness. Together, these findings suggest that during induction, clinical benefit is driven by a favorable balance between reinvigoratable progenitor-exhausted CD8⁺ T cells and supportive antigen-presenting DC subsets.

A striking and consistent observation across both induction and maintenance phases was the association between decreased circulating DC frequencies, particularly pDCs and total cDC1 within the CD45⁺ compartment, and improved OS. Although it appears counterintuitive given the established role of cDC1s in antitumor immunity, it should be noted that these measurements were performed in peripheral blood. In the context of effective immunotherapy, activated DCs are expected to migrate from the circulation into tumor tissue and tumor-draining lymph nodes, where antigen presentation and T-cell priming occur.20 Therefore, persistently elevated circulating DC levels may reflect impaired tissue recruitment or ineffective immune engagement, rather than a protective systemic immune state. Although speculative, this interpretation is consistent with the observation that reduced circulating DC frequencies after vaccination correlated with improved survival. Supporting this interpretation, DCs are known to acquire tissue-specific signatures on migration to lymphoid organs and to be actively recruited from the circulation to inflamed tissues during immune responses.21–23 Consistent with this dynamic trafficking, immune activation such as influenza vaccination has been shown to induce mobilization of peripheral blood myeloid DCs.24 In line with these observations, studies in humans have demonstrated that DCs undergo rapid redistribution following administration, with limited persistence in the peripheral circulation. In a seminal study using radiolabeled DCs, Morse et al showed that intravenously injected DCs initially localized to the lungs and subsequently redistributed to the liver, spleen, and bone marrow, with minimal detection in peripheral blood and no significant accumulation in lymph nodes or tumor sites. Notably, only a small fraction of intradermally injected DCs migrated to regional lymph nodes, suggesting that most DCs exit the circulation and relocate to tissue compartments rather than remaining detectable in blood.25 Furthermore, recent work by Iborra et al highlights that both cDC1s and pDCs must physically localize within tissues to successfully cross-prime and amplify protective CD8⁺ tissue-resident memory T-cell responses.26 Unfortunately, the absence of paired tumor or lymph-node biopsies precludes direct assessment of DC trafficking and function at the tissue level.

During the maintenance phase incorporating DC vaccination plus atezolizumab, the immune correlates of outcome appeared to shift. While TPEX cells dominated during induction, increased proportions of CXCR5⁻ PD-1⁺ TEX cells were associated with improved OS after vaccination. Although TEX CD8⁺ T cells are frequently viewed as functionally constrained, particularly in terms of proliferative capacity, TEX subsets can retain measurable effector potential in supportive inflammatory and antigen-presenting contexts. In our prior preclinical work, XCR1⁺ cDC1 were required to sustain effector function in more differentiated exhausted CD8⁺ T-cell states during PD-1/PD-L1 axis checkpoint blockade, even when progenitor-exhausted responses could still proliferate.12 Accordingly, an increase in TEX cells in patients may reflect ongoing antigen-driven engagement and maintenance of effector activity rather than irreversible dysfunction, although persistent antigen exposure can also promote terminal differentiation. These observations raise the possibility that DC vaccination could modulate the balance across exhaustion states over time, a hypothesis that warrants direct functional and longitudinal validation.

Notably, regulatory T-cell subsets, γδ T-cell populations, and most other immune cells were not consistently associated with outcomes, indicating that the observed effects are not driven by generalized immune activation or suppression but rather by specific DC and CD8⁺ T-cell axes. This specificity reinforces the biological plausibility of our findings and aligns with the mechanistic framework proposed in prior work.

This study has important limitations. The number of patients included was low and immune analyses were exploratory. Immune profiling was restricted to peripheral blood, and direct evaluation of intratumoral immune responses was not possible. Consequently, causal relationships cannot be established, and the observed associations should be validated in larger, controlled studies incorporating paired blood and tissue analyses.

In conclusion, the VENEZOLUNG trial provides prospective clinical and translational evidence supporting the feasibility and potential impact of combining autologous tumor-lysate loaded DC vaccination with PD-L1 blockade as maintenance therapy in ES-SCLC. The integrated immune monitoring reveals longitudinal immune dynamics associated with durable benefit. In particular, long-term survival was associated with the expansion of circulating cDC1 XCR1⁺ and TPEX cells following induction with chemotherapy plus atezolizumab, as well as with subsequent modulation of circulating cDC1 after DC vaccination, suggesting active remodeling of antigen-presenting cell compartments during effective antitumor immunity. These findings extend prior mechanistic and spatial studies7 8 of cDC1 and CD8⁺ T-cell interactions into a prospective clinical setting, providing patient-level evidence that DC-based immunization strategies may help reinforce CD8⁺ T-cell responses and contribute to durable outcomes in an immune-refractory disease such as ES-SCLC. Together, our results support further investigation of DC-targeted immunotherapies as a strategy to enhance long-term benefit from checkpoint blockade in SCLC.

Data availability statement

Data are available upon reasonable request.

Ethics statements

Patient consent for publication

Not applicable.

Ethics approval

The VENEZOLUNG study was conducted following the International Ethical Guidelines for Biomedical Research Involving Human Subjects, Good Clinical Practice guidelines, and the Declaration of Helsinki, 1964; local laws; SAS order 3470/2009; and Organic Law 3/2018, of December 5, on the Protection of Personal Data and guarantee of digital rights. The study was reviewed and approved by the ethics committee of Grupo Hospitalario Quirónsalud-Catalunya (OR-IISML42037). All patients provided informed consent before inclusion.