Brief Report: Multi-Institution Real-World Analysis Evaluating Safety, Efficacy, and ctDNA Dynamics After Tarlatamab in Patients With Previously Treated SCLC and LCNEC.

Introduction
SCLC is an aggressive malignancy with poor prognosis accounting for approximately 15% of all lung cancer cases.1 For extensive-stage disease (ES-SCLC), despite frontline therapy incorporating combination platinum-doublet chemotherapy with immune checkpoint inhibitor (ICI), approximately 70% of patients will experience progression within 6 months,2 with mixed results noted after available second-line chemotherapy.3, 4, 5
Tarlatamab, a bispecific T-cell engager (BiTE) targeting delta-like ligand 3 (DLL3) and CD3 on T cells, was recently approved by the Food and Drug Administration (FDA) in May 2024 for treatment of relapsed ES-SCLC, following positive results from the phase II DeLLphi-3016 and the subsequent confirmatory phase III DeLLphi-304 trial demonstrating significant improvements in both median progression-free survival (PFS) and overall survival (OS) compared with chemotherapy.7 Although these results are encouraging, tarlatamab is associated with unique toxicities, including cytokine release syndrome (CRS) and immune effector cell-associated neurotoxicity syndrome (ICANS), which require close monitoring.
With the recent FDA approval of tarlatamab, there remains an active question on how outcomes of this novel immunotherapeutic agent will translate into real-world practice. To address this, we report a multi-institutional, retrospective analysis detailing the safety and efficacy of tarlatamab at two National Cancer Institute–designated comprehensive cancer centers.

Materials and Methods

Study Design and Patients
We retrospectively reviewed all patients who received tarlatamab for relapsed ES-SCLC or large cell neuroendocrine carcinoma (LCNEC) at the University of Maryland Medical Center and Case Western University Hospital between July 2024 and May 2025. Data were collected under institutional review board–approved protocols at each respective center. Owing to the retrospective nature of the study, the requirement for written informed consent was waived. Patients who received at least 1 dose of tarlatamab were included in our safety analysis. Patients who underwent restaging imaging after tarlatamab were included in our efficacy analysis.
Patients received tarlatamab per FDA label in 28-day cycles, with inpatient observation for at least 22 hours after tarlatamab infusion on day 1 and day 8 of cycle 1. Pending tolerability during inpatient monitoring, subsequent tarlatamab infusions were administered outpatient using recommended observation periods as per drug label.

Study End Points
Treatment response was determined using clinical interpretation of radiological disease assessment from patients with available post-treatment imaging performed before C3D1. CRS and ICANS grading were reported using American Society for Transplantation and Cellular Therapy consensus guidelines. Other side effects of special interest (SESI), including dysgeusia and cytopenias, were reported using Common Terminology Criteria for Adverse Events (CTCAE) version 5.0. A subcohort of patients had circulating tumor DNA (ctDNA) collection using Guardant360 assay at pre- and post-treatment time points (drawn before C1D1 and on or after C3D1) and were included for molecular response assessment by measuring trends in tumor fraction. Molecular response was defined as more than or equal to 50% reduction in tumor fraction as validated in a recent cohort of patients with SCLC receiving immunotherapy.8

Statistical Considerations
OS, PFS, and duration of response were calculated from the date of treatment initiation, with a data cutoff of May 31, 2025. The follow-up period was defined as the duration of time from treatment initiation to date of last contact or death. Median PFS and OS were estimated using Kaplan-Meier methods. Survival analyses were performed with GraphPad PRISM, version 10.5.

Results

Patient Characteristics
A total of 48 patients received at least 1 dose of tarlatamab. Most had SCLC histology (94%). Median age at C1D1 was 68 years, with 58% females and 92% being current/former smokers. Furthermore, 33% of patients had Eastern Cooperative Oncology Group more than or equal to 2, with median prior lines of therapy of 2 (range 1–5), and 67% having intracranial involvement. In addition, 44% of patients had less than 90-day response after platinum-doublet chemotherapy (Table 1).

Side Effects of Special Interest
Of 48 patients, 42 (87.5%) experienced a SESI after treatment initiation. CRS and ICANS were observed in 54.2% and 33.3% of patients, respectively. Most CRS (53.8%) and ICANS (62.5%) events occurred within 24 hours of C1D1. Both CRS and ICANS had a median onset time from preceding start of tarlatamab infusion of 11 hours. All CRS events were grade 1 or 2 in severity. Grade 3 ICANS was observed in two patients, both of whom had intracranial involvement. Furthermore, 13 of 32 patients (41%) with intracranial involvement at the start of treatment developed ICANS compared with three of 16 (19%) without intracranial involvement. Of the 13 patients with intracranial involvement, 12 had previously treated lesions. All three patients without intracranial involvement who developed ICANS had simultaneous CRS compared with eight of 13 patients with brain metastases. Rates of ICANS after different modalities of intracranial radiation were comparable: whole brain radiation therapy (WBRT, 4/8, 50%), stereotactic radiosurgery (SRS, 5/11, 45%), gamma knife radiosurgery (GKRS, 2/3, 67%), or a combination of WBRT/SRS/GKRS (2/4, 50%). Furthermore, one of six patients (17%) with untreated brain metastases developed ICANS. Median duration of both CRS and ICANS to symptomatic resolution was less than 24 hours. Other frequently observed SESI included dysgeusia (47.9%), thrombocytopenia (33.3%), and neutropenia (14.6%) (Supplementary Table 1).

Survival Outcomes
A subset of 30 patients were included in our survival analysis. Of the patients, 20 (66.7%) experienced objective response, with two patients (6.7%) achieving complete response (CR) (Fig. 1A). Disease control rate (DCR) was 73%. After a median follow-up time of 5.7 months, median PFS was 4.9 months (95% confidence interval [CI]: 4–not reached) and median OS had not been reached (95% CI: 4.8– not reached) (Fig. 1B). Median duration of response among the 20 patients who experienced objective response was 3.8 months (95% CI: 1.8–5.8). In addition, 14 patients had pretreatment ctDNA collected, with five patients not undergoing post-treatment collection (two patients passed before C3D1, the remaining three patients achieved partial response but never had follow-up ctDNA testing). Of the nine patients who underwent ctDNA analysis, five demonstrated molecular response (Fig. 2). The average reduction in tumor fraction among molecular responders was 79.3%. All five patients determined to have molecular response were also found to have radiographic response and remain without disease progression at time of data cutoff. All nine patients in the ctDNA cohort also had comparable baseline characteristics to that of the study population as a whole: median age was 67 years, median Eastern Cooperative Oncology Group was 1, 100% were current/former smokers, and median line of prior treatment was 1.

Discussion
Tarlatamab has demonstrated promising efficacy and a manageable safety profile based on initial clinical trial experience. Given its recent FDA approval, there remains a paucity of published data reflecting the antitumor response and tolerability of tarlatamab in real-world clinical practice. This retrospective study—the largest study of real-world efficacy of tarlatamab to date—further confirms the valuable role of tarlatamab in the treatment of ES-SCLC while highlighting variations in toxicity profile compared with trial data.
Within our cohort of patients with heavily pretreated SCLC and LCNEC, we observed a best overall response rate (BOR) of 67% with two patients experiencing CRs. Median PFS (4.9 mo) compared favorably to results from both DeLLphi-301 (4.9 mo), DeLLphi-304 (4.2 mo), and existing real-world retrospective analysis (2.7 mo).9 Furthermore, 14 of 20 patients with initial response had sustained benefit without progression at time of data cutoff highlighting the durability of this agent. The 6-month OS rate of 60% was also comparable to DeLLphi-304 results (70%), despite our cohort having a lower baseline performance status and higher incidence of intracranial involvement. Collectively, these findings affirm the practice-changing efficacy of tarlatamab when delivered in a standard-of-care setting.
ctDNA dynamics have demonstrated prognostic value in patients with SCLC receiving immunotherapy, with ctDNA clearance between 6 and 15 weeks post-treatment correlated to improved survival outcomes.8,10 Five patients met the criteria for molecular response of more than or equal to 50% reduction in tumor fraction with all five patients having objective treatment response. These findings underscore the potential utility of this diagnostic tool as a dynamic marker of clinical response in patients treated with tarlatamab. Further validation through prospective studies is still needed to confirm these initial findings.
Both the rate (54% versus 56%) and onset (<24 h from C1D1) of CRS within our cohort mirrored the DeLLphi-304 findings, albeit with a shorter duration of symptoms noted in our real-world cohort. This difference is likely representative of the higher utilization of tocilizumab within our cohort (23%) compared with those in DeLLphi-304 (4%). There are ongoing research efforts evaluating prophylactic administration of tocilizumab to prevent CRS, with important financial implications that need consideration (S Puri, Department of Thoracic Oncology, Moffitt Cancer Center, personal communication, 2025).9
Notably, ICANS was observed in 33% of patients, a significant increase from the 6% in Dellphi-304 and consistent with recent real-world data.9 Grade 3 ICANS was reported in two patients within our cohort, both with previously treated brain metastases. Furthermore, 81.2% of patients who developed ICANS had brain metastases at time of treatment, emphasizing the need for close neurologic monitoring, particularly in patients with intracranial involvement. Of 16 patients, 15 required dexamethasone for management of ICANS, with rapid resolution noted within 1 day of onset. We suspect that differences in ICANS rate from our population stem from the high proportion of patients in our cohort with intracranial disease (58.3% versus 44% in Dellphi-304), including those with untreated brain lesions. The comparatively early onset of neurotoxicity observed in our cohort may also reflect a spectrum of neurocognitive symptoms classically categorized as ICANS, but it may in fact represent unique pathophysiologic mechanisms of neurologic dysfunction warranting further characterization and elucidation.11,12
Important limitations of our analysis include the abbreviated median follow-up of 5.7 months, which affects our estimation of survival impact. In addition, although our cohort comes from two research centers, it is still a relatively small, ethnically homogenous patient sample, with larger multicenter efforts to summarize real-world outcomes still needed.
In summary, initial experience with tarlatamab among our real-world patient cohort compared favorably in terms of clinical outcomes with previously reported clinical trial results. Important differences were noted in terms of rates and timing of neurologic toxicity, specifically ICANS and dysgeusia, but these were often low grade and manageable. To this end, tarlatamab represents an exciting therapeutic breakthrough for patients with ES-SCLC, with future efforts needed to help optimize delivery and accessibility within standard-of-care practice.

CRediT Authorship Contribution Statement
Graeme Fenton: Conceptualization, Data curation, Formal analysis, Investigation, Validation, Methodology, Visualization, Writing - original draft, Writing - review & editing.
Annie L. Zhang: Data curation, Investigation, Writing - review & editing.
Sharon Zhong: Data curation, Investigation, Writing - review & editing.
Curts Tatsuoka: Formal analysis, Validation, Writing - review & editing.
Michelle Sitg: Investigation, Data curation, Project administration.
Alexandra Simms: Project administration, Resources.
Ranee Mehra: Investigation, Supervision, Data curation, Project administration, Writing - review & editing.
Afshin Dowlati: Investigation, Supervision, Data curation, Project administration, Writing - review & editing.
Taofeek K. Owonikoko: Investigation, Supervision, Project administration, Writing - review & editing.
Katherine Scilla: Investigation, Supervision, Data curation, Project administration, Writing - review & editing.
Melinda Hsu: Conceptualization, Data curation, Formal analysis, Investigation, Methodology, Project administration, Supervision, Validation, Visualization, Writing - original draft, Writing - review and editing.
Samuel Rosner: Conceptualization, Data curation, Formal analysis, Investigation, Methodology, Project administration, Supervision, Validation, Visualization, Writing - original draft, Writing - review & editing.

Disclosure
The authors declare no conflict of interest.