Review Reports

Round 1

Reviewer 1 Report (New Reviewer)

Comments and Suggestions for Authors

Title: Real-World Efficacy and Safety of Tarlatamab in Small Cell Lung Cancer: A Systematic Review

This manuscript is a systematic review design and well-formulated; however, some major issues are raised and should be revised.

1. Introduction:

The authors want to summarize real-world evidence on tarlatamab in small cell lung cancer. This is a worthwhile aim, because the drug is now in routine use and trial data do not cover every patient group.

The introduction is clear and well argued. One citation is off: a clinical trial (ref 3) is cited to support claims about real-world populations. The title uses "efficacy," but “effectiveness” is the right term for real-world data.

           2. Materials and methods

Only three databases were used (PubMed, Cochrane Library, and ScienceDirect). Embase, Scopus, Web of Science, and conference proceedings are missing, even though early real-world data usually appear first as abstracts.

The search strings are simple phrases, not structured Boolean searches, and they differ between databases. No search date is given, and the search is now about seven months old. The PROSPERO number appears to have one digit too many, and the registration date relative to the search is not stated.

Figure 1 gives two exclusion reasons, while the text lists many others. Nearly all full texts were excluded (65 of 70), which suggests the first screening was too loose. A list of excluded studies and a measure of reviewer agreement are missing.

Eligibility rules were not applied consistently. Mixed cohorts were allowed only if SCLC-specific results could be extracted. Yet three included studies mix SCLC with other tumors, and their safety data are not separated. Patient overlap between the academic centers has not been checked.

The risk-of-bias method is also inconsistent. The authors describe cohort and case-series checklists but applied only the case-series one. 

       3. Results

The numbers are mostly consistent, but several small errors need fixing. Cooper’s stable disease is given as 8% when it is 7.1%. Cooper’s CRS rate (57%) fits a denominator of 14, not 16. Bolte’s liver metastasis rate matches neither denominator. The Joshi row in Table 3 is garbled, and the prophylaxis group’s CRS and ICANS rates are missing. Table 5 still contains an unresolved note: “requires confirmation of the exact number of contributing studies.”

Effectiveness is reported only in evaluable subsets. Fenton’s 66.7% response rate covers 30 of 48 patients. Overall, treated patients are at most 41.7%.

The ICANS rates of 23–48% are several times higher than in trials. That may be the most important finding, but it is under-discussed.

The abstract has no numerical results at all. The PRISMA checklists were not updated and marked items “Yes” that are missing from the text.

      4. Discussion

The discussion is careful about prophylactic tocilizumab, outpatient dosing, and extrapolation to transformed SCLC. I welcome that caution. However, it uses data that never appear in the results: fewer events after the first dose, fewer events with prophylaxis, and clinical predictors of CRS and ICANS. Everything cited in the discussion should be shown in the results first.

The long section on molecular subtypes (ASCL1, NEUROD1, POU2F3, and YAP1) is speculative, and none of the five studies measured these. It should be shortened. It also contrasts classical SCLC with transformed SCLC in a way that implies transformed tumors lack TP53 and RB1 loss. They often carry it too.

Several references are not cited in the text, are corrupted, or are incomplete. Two methods papers (GRADE and ROBINS-I) are miscited as support for publication bias.

Author Response

Comment 1

The authors want to summarize real-world evidence on tarlatamab in small cell lung cancer. This is a worthwhile aim, because the drug is now in routine use and trial data do not cover every patient group. The introduction is clear and well argued. One citation is off: a clinical trial (ref 3) is cited to support claims about real-world populations. The title uses "efficacy," but “effectiveness” is the right term for real-world data.

Response:

We thank the reviewer for their positive assessment of the Introduction and for these helpful suggestions. We agree that "effectiveness" is the more appropriate terminology when describing treatment outcomes in routine clinical practice. Accordingly, we have revised the manuscript title to "Real-World Effectiveness and Safety of Tarlatamab in Small Cell Lung Cancer: A Systematic Review."

Additionally, we have corrected the citation in the Introduction concerning real-world patient populations. Reference [3], which describes the DeLLphi-301 clinical trial, has been replaced with references [5,6,8], which report real-world experiences with tarlatamab in patients with characteristics frequently underrepresented in clinical trials, including reduced performance status, comorbidities, and central nervous system involvement.

These revisions improve the accuracy of the supporting evidence and ensure consistency between the manuscript's terminology and its focus on real-world clinical outcomes.

 

Comment 2

Response:

We thank the reviewer for their detailed methodological assessment and constructive suggestions. We acknowledge the limitations identified and have revised the manuscript to improve transparency and clarify the methodological decisions underlying this systematic review.

Regarding the literature search, our review focused on identifying published real-world clinical outcomes of tarlatamab using PubMed, the Cochrane Library, and ScienceDirect. We employed both focused and broader keyword combinations to capture relevant studies, including publications that did not explicitly identify themselves as real-world evidence. We recognize that the inclusion of additional databases and conference proceedings could have increased the comprehensiveness of the search. We have therefore explicitly acknowledged this limitation in the revised manuscript. We have also clarified the original search date to define the period covered by our findings.

Regarding the PROSPERO registration, we have clarified the registration information and the timing of registration relative to the literature search.

We also appreciate the observations concerning the PRISMA flow diagram and study-selection process. The relatively large number of full-text exclusions reflects the broad initial screening approach, which was intended to minimize premature exclusion of potentially relevant publications. We have clarified this approach and the application of eligibility criteria during full-text assessment. Study selection was conducted independently by two reviewers, with disagreements resolved through discussion and consensus.

Regarding mixed histologic populations, we agree that distinguishing SCLC-specific outcomes from outcomes reported for combined cohorts is important. We have clarified how outcome denominators were considered and emphasized that mixed-cohort estimates should not be interpreted as exclusively SCLC-specific findings. We have also acknowledged the potential for unrecognized patient overlap between institutional reports as a limitation.

Finally, we have revised the description of the methodological quality assessment to accurately reflect the use of the JBI Critical Appraisal Checklist for Case Series. We acknowledge that application of this tool across retrospective observational studies may not fully address all bias domains relevant to cohort comparisons, particularly confounding, and have incorporated this limitation into the Discussion.

We believe these clarifications improve the transparency and interpretability of the review while maintaining its focus on the real-world evidence identified during the original literature search.

 

Comment 4

Response:

We thank the reviewer for their thoughtful assessment of the Discussion and for recognizing the cautious interpretation of our findings regarding prophylactic tocilizumab, outpatient administration, and EGFR-mutant transformed SCLC. We agree that the Discussion should be firmly grounded in the findings presented in the Results and that speculative molecular considerations should be appropriately limited.

Consistency Between Results and Discussion: We have expanded the Results section to include additional safety and toxicity-management findings reported in the included studies. Specifically, we have incorporated the observations by Mullangi et al. regarding the more favorable safety profile following subsequent doses of tarlatamab. We have also added the subgroup-specific CRS and ICANS rates reported by Joshi et al. among patients receiving prophylactic tocilizumab compared with those who did not receive prophylaxis. Furthermore, we have included the exploratory clinical predictors identified by Joshi et al., including elevated LDH and liver metastases for CRS, and diabetes mellitus and cardiovascular disease for ICANS. These additions ensure that the corresponding interpretations in the Discussion are supported by findings presented in the Results. We have maintained cautious language emphasizing the preliminary and hypothesis-generating nature of these observations.

Molecular Subtypes and Transformed SCLC: We agree that the previous discussion of molecular subtypes was disproportionately detailed given that none of the five included real-world studies systematically evaluated transcriptional subtype classification. We have therefore substantially shortened this section, retaining only a brief discussion of molecular heterogeneity and its potential relevance to DLL3 expression while emphasizing that its predictive significance for tarlatamab remains unestablished.

We also appreciate the clarification regarding TP53 and RB1 alterations. We have revised the discussion to explicitly acknowledge that these alterations occur in both conventional and EGFR-mutant transformed SCLC. The revised text now emphasizes the distinct evolutionary origin and lineage plasticity of transformed SCLC rather than implying that the absence of TP53 or RB1 inactivation distinguishes it from conventional disease.

Reference Accuracy and Consistency: We have reviewed the reference list to identify uncited, incomplete, and incorrectly attributed references. In particular, we have removed references to the GRADE and ROBINS-I methodological papers from the statement concerning potential publication and reporting bias, as these publications do not directly support that assertion. The GRADE methodology reference has instead been cited in the appropriate Methods section. We have also corrected incomplete bibliographic information and identified unused references for removal.

We believe these revisions improve the consistency between the Results and Discussion, reduce unnecessary speculation, and strengthen the scientific accuracy and clarity of the manuscript.

Reviewer 2 Report (Previous Reviewer 1)

Comments and Suggestions for Authors

Comments to the Authors

The manuscript presents a systematic review of real-world evidence regarding the effectiveness and safety of tarlatamab in patients with small cell lung cancer. The topic is clinically relevant, especially because patients treated in routine practice may differ from those included in clinical trials. The manuscript is generally understandable and provides useful preliminary information. However, several issues should be addressed before publication.

Major Comments

1. The literature search should be more comprehensive.

The authors searched PubMed, the Cochrane Library, and ScienceDirect. However, other important databases, such as Embase, Scopus, or Web of Science, were not included. Since only five studies were finally selected, the possibility of missing relevant studies should be considered. The authors should either expand the search or explain the reason for selecting only these databases.

1. The final search date should be clearly reported.

The exact date of the last literature search should be stated in the Methods section. This is important because evidence concerning tarlatamab is developing rapidly.

1. The complete search strategy should be provided.

The search terms presented in the manuscript are quite limited. The authors should provide the complete search strategy for each database, including Boolean operators and alternative terms such as AMG 757, IMDELLTRA, observational study, retrospective study, and real-world evidence.

1. The approach to conference abstracts should be clarified.

The eligibility criteria indicate that conference abstracts could be included, but the manuscript does not explain whether conference websites or abstract databases were searched systematically. The authors should clarify how unpublished studies and conference abstracts were identified.

1. The study-selection process needs more information.

A large number of full-text articles were excluded, but only general reasons are provided. It would be helpful to include a supplementary table listing the main excluded full-text articles and the reason for excluding each one.

1. The included patient populations are heterogeneous.

Some included studies contained patients with conventional small cell lung cancer, while others included transformed small cell lung cancer, large-cell neuroendocrine carcinoma, or extrapulmonary disease. These conditions may have different clinical behavior and treatment responses.

The authors should clearly separate the results for conventional small cell lung cancer from those for other neuroendocrine tumors whenever possible.

1. Possible overlap between study populations should be considered.

Some studies were conducted during similar periods at academic institutions. The authors should check whether the same patients may have been reported in more than one study. If this cannot be excluded, it should be mentioned as a limitation.

1. The risk-of-bias assessment requires clarification.

The manuscript states that different JBI tools were selected according to study design, but the table appears to use the same checklist for all studies. The authors should clearly identify which assessment tool was used for each study.

The method used to classify studies as having low, moderate, or high risk of bias should also be explained and supported by a reference.

1. The response-analysis denominators should be clearly presented.

Some studies calculated response rates only among patients who had follow-up imaging, while other studies may have used all treated patients. This can produce different response estimates.

For every study, the authors should report:

• the total number of treated patients;
• the number evaluated for response;
• the number excluded from response assessment; and
• the reason for exclusion.

1. A separate outcome table would improve the manuscript.

The manuscript would be easier to understand if the authors added a clear table summarizing the main clinical outcomes for each study, including:

• objective response rate;
• disease-control rate;
• progression-free survival;
• overall survival;
• follow-up duration;
• cytokine-release syndrome;
• immune effector cell-associated neurotoxicity syndrome;
• severe adverse events; and
• treatment discontinuation.

1. The safety results should include toxicity grades.

The manuscript reports rates of cytokine-release syndrome and neurological toxicity, but the severity of these events is not always clear. The authors should distinguish between any-grade events and grade 3 or higher events.

Information about hospitalization, intensive-care admission, corticosteroid use, tocilizumab use, and treatment discontinuation should also be presented where available.

1. Comparisons with clinical trials should be cautious.

The comparison between real-world studies and the DeLLphi clinical trials is useful. However, these are indirect comparisons involving different populations, follow-up periods, and methods of outcome assessment. The authors should avoid concluding that real-world effectiveness is equal to clinical-trial efficacy or that toxicity is higher without a direct comparative study.

1. The discussion of prophylactic tocilizumab should be moderated.

The number of patients who received prophylactic tocilizumab was small, and the comparison was not randomized. Therefore, the findings may be affected by differences between patients or treatment centers. These results should be described as exploratory and should not be used to make firm clinical recommendations.

1. The GRADE assessment should be explained more clearly.

The authors classified the evidence as having very low certainty, which appears reasonable. However, the judgments for risk of bias, inconsistency, indirectness, imprecision, and publication bias should be shown more clearly.

The editorial note stating that the exact number of contributing studies requires confirmation should be removed and replaced with the correct number.

1. The conclusions should be more cautious.

The current evidence is based on a small number of retrospective studies with limited sample sizes and short follow-up periods. Therefore, the conclusion should describe the findings as preliminary.

A more balanced conclusion would be that available real-world studies suggest possible clinical activity of tarlatamab, but larger prospective studies are required to confirm its effectiveness and safety.

1. The limitations of the review should be expanded.

The authors should discuss the small number of included studies, limited sample sizes, retrospective designs, mixed patient populations, inconsistent response assessment, possible selection bias, short follow-up, and possible publication bias.

1. The manuscript contains some repetition.

The numbers of identified, screened, excluded, and included studies are repeated in several sections. Similar statements about the limitations of the evidence also appear more than once. The manuscript could be shortened and made clearer by removing repeated information.

1. The protocol-registration information should be verified.

The authors should provide the date of PROSPERO registration and clarify whether the review was registered before study screening and data extraction. Any differences between the registered protocol and the completed review should be reported.

Minor Comments

1. The title is generally appropriate, but “effectiveness” may be more suitable than “efficacy” for real-world observational studies.
2. The Abstract should report the total number of patients included across the five studies, while avoiding double counting if study populations overlap.
3. The full form of PRISMA should be written correctly at first mention.
4. Sections describing PRISMA guidelines and study selection contain some repeated information and could be combined.
5. The abbreviations LCNEC and EPSCC should be defined at first use.
6. The authors should consistently distinguish between de novo small cell lung cancer, transformed small cell lung cancer, and other neuroendocrine cancers.
7. The study period should be reported for every included study. If it is unavailable, it should be stated as “not reported.”
8. Tables should clearly distinguish between the total treated population, the safety population, and the response-evaluable population.
9. The percentages should be reported using a consistent number of decimal places.
10. Table headings and footnotes should define all abbreviations.
11. The PRISMA flow diagram should include the number of articles excluded for each main reason.
12. The reference list should be carefully checked for incomplete entries and inconsistent formatting.
13. ClinicalTrials.gov references should include the study title, registration number, and access date.

The manuscript should be proofread for grammar, spacing, broken words, and formatting errors.

Author Response

Comment 1. Complete Search Strategy

Response: We thank the reviewer for this suggestion. Our original search strategy included both focused searches incorporating "real world evidence" and broader searches combining "tarlatamab" and "small cell lung cancer." These broader searches were intended to identify relevant observational studies regardless of whether they explicitly used terms such as "retrospective" or "real-world" in their titles or abstracts. The database-specific search expressions are presented in Table 2. We acknowledge that including additional alternative terms could have improved search sensitivity, and this limitation has been recognized in the revised manuscript.

Comment 2. Conference Abstracts

Response: We appreciate the reviewer highlighting this point. Although conference abstracts were eligible for inclusion when sufficient clinical data were available for extraction, conference proceedings and abstract repositories were not systematically searched as independent sources. We have clarified this distinction in Section 2.4 and acknowledged that some early or unpublished findings may not have been captured.

Comment 3. Excluded Full-Text Articles

Response: We thank the reviewer for this recommendation. The original screening process identified 141 records, of which 70 underwent full-text assessment and 65 were excluded according to the predefined eligibility criteria. The principal exclusion categories are summarized in the Results and PRISMA flow diagram. We recognize that an article-level supplementary table would provide greater transparency; however, the current review presents the aggregate reasons for exclusion rather than an individual listing of excluded articles.

Comment 4. Heterogeneous Patient Populations

Response: We agree that conventional SCLC, EGFR-mutant transformed SCLC, and other high-grade neuroendocrine malignancies represent clinically distinct populations. We have clarified the differences between these groups and emphasized SCLC-specific outcomes wherever available. In particular, the outcomes reported by Cooper et al. involving EGFR-mutant transformed SCLC are discussed separately from those involving predominantly conventional SCLC. Results derived from mixed histologic cohorts are identified and interpreted cautiously.

Comment 5. Potential Overlap Between Patient Populations

Response: We appreciate this observation. Although the included studies reported separate institutional or multicenter cohorts, potential patient overlap cannot be definitively excluded from the available aggregate data. We have acknowledged this limitation in Section 4.3 and clarified that the combined study sample sizes should not necessarily be interpreted as representing entirely unique patients. No quantitative pooling was performed.

Comment 6. Risk-of-Bias Assessment

Response: We thank the reviewer for identifying this inconsistency. We have clarified that the JBI Critical Appraisal Checklist for Case Series was applied across the five included studies. Furthermore, we have revised Table 4 to retain the individual checklist judgments while removing the overall categorical classifications based on numerical thresholds, which were not derived from a standardized JBI scoring system. This approach improves methodological transparency while preserving the original critical appraisal findings.

Comment 7. Response-Analysis Denominators

Response: We agree that distinguishing treated populations from response-evaluable populations is essential for interpreting real-world effectiveness. We have clarified these denominators in the Results and added a summary table presenting the treated and response-evaluable populations for each study. Reasons for missing response assessments are provided where available, while unavailable information is designated as not reported. We have also emphasized the potential for selection and survivorship bias when response estimates are restricted to evaluable subsets.

Comment 8. Summary of Clinical Outcomes

Response: We thank the reviewer for this helpful suggestion. We have added a consolidated table summarizing the principal effectiveness and safety outcomes of the five included studies, including response rates, disease control, progression-free survival, overall survival, CRS, and ICANS. The table also clarifies the relevant denominators and distinguishes outcomes from mixed histologic populations where applicable.

Comment 9. Toxicity Severity and Management

Response: We appreciate the reviewer's recommendation. We have clarified the distinction between any-grade and severe CRS/ICANS and emphasized that toxicity outcomes should be interpreted using the definitions and denominators reported by the original studies. Available safety-management findings are described where reported, while outcomes without sufficient information are identified as not reported rather than assumed to be absent.

Comment 10. Comparisons With Clinical Trials

Response: We agree that comparisons between real-world observational studies and the DeLLphi clinical trials are indirect and subject to differences in patient populations, follow-up, and outcome ascertainment. We have maintained a cautious interpretation and clarified that numerical similarities or differences do not establish equivalent effectiveness or higher toxicity risk in routine clinical practice.

Comment 11. Prophylactic Tocilizumab

Response: We thank the reviewer for this observation. We agree that the findings regarding prophylactic tocilizumab should be interpreted cautiously given the small sample size and non-randomized study design. As emphasized in Sections 4.2 and 4.4, these observations are exploratory and hypothesis-generating and do not establish a causal benefit or justify routine prophylactic administration.

Comment 12. GRADE Assessment

Response: We appreciate the request for additional clarity. The manuscript explains that the very low certainty ratings were primarily influenced by retrospective study designs, potential selection bias, heterogeneous populations, differences in outcome assessment, small sample sizes, and limited follow-up. We have clarified these considerations in the GRADE assessment and removed the unresolved editorial note from the table. We acknowledge that severe-toxicity outcomes were not consistently reported across studies, limiting certainty in these estimates.

Comment 13. Conclusions

Response: We agree that the conclusions should appropriately reflect the limited evidence base. We have revised the Conclusion to emphasize that the available real-world studies suggest preliminary clinical activity of tarlatamab, while recognizing the limitations associated with small retrospective cohorts, heterogeneous populations, and short follow-up. The need for larger prospective studies has been further emphasized.

Comment 14. Limitations

Response: We thank the reviewer for this suggestion. We have expanded the limitations section to acknowledge the restricted database search, small number of studies, retrospective designs, mixed histologic populations, inconsistent response assessment, potential selection and survivorship bias, limited follow-up, possible patient overlap, and publication bias. These limitations reinforce the need for cautious interpretation of our findings.

Comment 15. Repetition

Response: We appreciate this observation. We have streamlined the Methods by removing repeated descriptions of the PRISMA reporting framework and study-selection numbers. We have also reduced overlapping statements regarding the limitations of the available evidence to improve readability and conciseness.

Comment 16. PROSPERO Registration

Response: We thank the reviewer for emphasizing the importance of protocol-registration transparency. We agree that the registration date and its relationship to study screening and data extraction should be reported explicitly. The registration information will be clarified in Section 2.1 based on the original PROSPERO record, including any applicable protocol deviations.

Minor Comments

Response: We thank the reviewer for these additional recommendations. We have addressed the relevant terminology, numerical, and formatting issues throughout the manuscript, including the use of "effectiveness" in the title, correction of the full PRISMA terminology, definition of LCNEC and EPSCC, consistent distinction between conventional and transformed SCLC, and clarification of study-specific outcome denominators.

We have also revised the Abstract to better summarize the included populations, standardized the presentation of percentages and table abbreviations, and reviewed the reference list for incomplete or unused entries. Unused ClinicalTrials.gov references have been identified for removal.

We appreciate the reviewer's constructive feedback, which has helped improve the clarity, transparency, and clinical interpretation of our systematic review.

 

Reviewer 3 Report (Previous Reviewer 2)

Comments and Suggestions for Authors

Minor revision. All four substantive Round 1 points are closed. Remaining
work is confined to: reconciling Table 4 with the stated dual-checklist
methodology, removing the unresolved internal note in Table 5, correcting
the supplementary file naming and PRISMA checklist page references to match
the resubmitted pagination, and confirming the PROSPERO registration number
through direct registry lookup rather than restating the unverified ID.

2. NEW ISSUES INTRODUCED IN THIS REVISION
2.1 Table 4 / Section 3.3 inconsistency
Section 3.3 now claims a differentiated risk-of-bias approach: cohort studies
assessed with the JBI Cohort Studies checklist, case series assessed with the
JBI Case Series checklist. Table 4 is still titled "JBI Case Series checklist"
and shows content identical to the single-checklist version from the prior
submission, applied to all five studies. Either the text was revised without
updating the table, or the stated dual-checklist methodology was not actually
implemented. This must be reconciled before acceptance.

2.2 Unresolved internal note in Table 5
The GRADE summary table (Table 5) ends with the line: "The severe-toxicity
row requires confirmation of the exact number of contributing studies." This
reads as the authors' own working note rather than content intended for a
reader, and should be struck or resolved before this proceeds further.

2.3 Stale supplementary file metadata
The PRISMA 2020 checklist supplied in the supplementary ZIP cites page
numbers up to 25, consistent with the original 26-page submission. The
resubmitted manuscript is 19 pages. The checklist's page references were not
updated and do not currently correspond to the resubmitted manuscript's
pagination. Note also that the supplementary ZIP folder and both checklist
filenames still carry the prior manuscript number (cancers-4622263), not the
current resubmission number (cancers-4669438).

Author Response

Response to Reviewer – Minor Revision

We sincerely thank the reviewer for their careful reassessment of our manuscript and for acknowledging that all four substantive concerns raised during the first round of review have been satisfactorily addressed.

1. Table 4 / Section 3.3 Inconsistency

We thank the reviewer for identifying this inconsistency. We have reconciled the methodological description with the risk-of-bias assessment presented in Table 4, ensuring that the manuscript accurately reflects the JBI appraisal approach used for the included studies.

2. Unresolved Editorial Note in Table 5

We agree with the reviewer and have removed the unresolved internal editorial note from the GRADE summary table.

3. Supplementary File Naming and PRISMA Checklist Pagination

We appreciate the reviewer highlighting these administrative discrepancies. We have updated the supplementary folder and checklist filenames to reflect the current manuscript number (cancers-4669438). Additionally, the PRISMA checklist page references have been revised to match the pagination of the resubmitted manuscript.

4. PROSPERO Registration Verification

We thank the reviewer for this recommendation. We have verified the PROSPERO registration information against the official registry and ensured that the registration number reported in the manuscript corresponds to the registered protocol.

We appreciate the reviewer's constructive feedback throughout the revision process, which has helped improve the methodological transparency, accuracy, and presentation of our systematic review.

 

Round 2

Reviewer 1 Report (New Reviewer)

Comments and Suggestions for Authors

The authors have performed all the modifications; thus, I recommend accepting this manuscript in the current form.

Reviewer 2 Report (Previous Reviewer 1)

Comments and Suggestions for Authors

Final Review Report

The authors have responded satisfactorily to the reviewer’s comments and have made appropriate revisions throughout the manuscript. The revised version is clearer, better organized, and more transparent in its presentation of the methodology, results, limitations, and interpretation of the available evidence.

The authors have adequately clarified the study populations and response-evaluable denominators, improved the distinction between de novo small cell lung cancer and transformed small cell lung cancer, strengthened the reporting of safety and effectiveness outcomes, and moderated conclusions that were previously stated too strongly. The limitations of the included real-world studies and the indirect nature of comparisons with clinical-trial findings are now more appropriately acknowledged.

Overall, the manuscript has been substantially improved, and the authors’ responses are satisfactory. The remaining issues, if any, are minor editorial matters that can be handled during production and do not affect the scientific value or conclusions of the study.

Recommendation

Accept in the present form.

This manuscript is a resubmission of an earlier submission. The following is a list of the peer review reports and author responses from that submission.

Round 1

Reviewer 1 Report

Comments and Suggestions for Authors

Major Comments to the Authors

I recommend that the authors specifically address the following points:

1. Strengthen and fully document the systematic search. The current reliance on PubMed, Cochrane Library, and ScienceDirect may not provide sufficiently comprehensive coverage. Major biomedical bibliographic databases should be considered or their omission explicitly justified. Complete reproducible search strings and exact search dates should be provided.
2. Recheck the PRISMA record counts and deduplication procedure. The manuscript reports 141 identified records and only seven duplicates. The authors should clearly state which reference-management or systematic-review software was used and how duplicate records were identified.
3. Clarify the risk-of-bias methodology. The Methods indicate design-specific JBI cohort/case-series tools, whereas the Results report use of the JBI Case Series Checklist for all five studies. The tool used for each study should reflect the actual study design, and this methodological discrepancy should be corrected.
4. Provide a formal GRADE summary. Since GRADE assessment is explicitly described in the Methods, outcome-level certainty ratings should be shown transparently in the Results, ideally through a summary-of-findings table including the reasons for downgrading.
5. Avoid overinterpretation of cross-study differences in toxicity. Claims that real-world CRS and ICANS rates are higher than trial rates should be framed cautiously because differences in case mix, baseline CNS involvement, toxicity ascertainment, grading practice, follow-up, and institutional monitoring may explain part of the observed variation.
6. Clarify denominators for efficacy estimates. Several studies report outcomes only among response-evaluable patients. For example, Fenton et al. included 48 treated patients but efficacy analysis was conducted in a subset of 30. Such analyses may introduce selection or survivorship bias, so treated-population and evaluable-population denominators should consistently accompany every efficacy estimate.
7. Maintain a clear distinction between de novo and transformed SCLC. Given the notably poorer outcomes reported for EGFR-mutant transformed SCLC, these patients should not be incorporated into generalized conclusions about conventional ES-SCLC without explicit qualification.
8. Reduce the length and repetition of the Discussion. Sections 4.1 to 4.6 repeatedly discuss toxicity, CNS involvement, trial generalizability, retrospective design, and the need for larger studies. A shorter, more focused Discussion would significantly improve the manuscript.
9. Moderate clinical recommendations. Statements regarding treatment positioning, outpatient administration, prophylactic tocilizumab, and risk-adapted management should clearly distinguish established evidence from preliminary hypotheses generated by small retrospective cohorts.
10. Carefully proofread the manuscript and tables. Errors such as “reponse” and “ICAS” should be corrected, and the placeholder wording“Please add: This research received no external funding” must be removed or finalized.

Author Response

please refer to the attachment

Author Response File: Author Response.docx

Reviewer 2 Report

Comments and Suggestions for Authors

This manuscript presents a systematic review of the real-world efficacy and safety of tarlatamab in previously treated small cell lung cancer. The review is registered in PROSPERO and follows PRISMA reporting guidance, and it synthesizes the five retrospective real-world studies currently available. The paper addresses a timely question: how the DeLLphi trial program's results generalize to broader, more clinically vulnerable populations, and it captures the operational realities of managing cytokine release syndrome (CRS) and immune effector cell-associated neurotoxicity syndrome (ICANS) outside the trial setting.

Methodological Strengths
The search strategy is transparent and reproducible across three databases, and the eligibility criteria in Table 1 are clearly pre-specified. Use of the Joanna Briggs Institute checklist for case series is appropriate given the single-arm, retrospective nature of the included studies. The decision to present a narrative qualitative synthesis rather than force a meta-analysis is correct: with five studies of heterogeneous populations, no comparator arms, and inconsistent response-assessment methods, pooling would have manufactured false precision. Isolating EGFR-mutant transformed SCLC as a biologically distinct subgroup with materially worse outcomes (2.7-month median OS versus the near-trial-level activity seen in Bolte et al. and Fenton et al.) is the single most clinically important distinction the review makes, and it is handled correctly.

Major Critiques
The review's methodological execution is sound, but its interpretation of comparative outcomes would benefit from closer engagement with the quantitative data already in the manuscript. Section 4.4.1 acknowledges that response and survival estimates in several cohorts are drawn from evaluable subsets rather than full treated populations (notably Fenton et al., 48 treated versus 30 analyzed), which the authors themselves identify as a survivorship-bias risk. The Discussion should state explicitly, at first mention of each cohort's response rate, whether the denominator is the full treated population or an evaluable subset, rather than leaving this distinction to the Limitations section alone. A reader skimming Section 4.1.1 could otherwise take the 66.7% objective response rate in Fenton et al. at face value.
Second, the manuscript's biological framing of why outcomes diverge between de novo and EGFR-transformed SCLC is present but thin. Section 4.1.3 correctly cites lineage plasticity and altered immune context as explanations for the poorer Cooper et al. outcomes, but does not ground this in the specific genetic and transcriptional architecture that determines tarlatamab sensitivity. Classical de novo SCLC is defined by near-universal biallelic inactivation of TP53 and RB1, which drives the neuroendocrine lineage program that DLL3 expression depends on. Transformed SCLC arising under EGFR-inhibitor selective pressure retains the original EGFR driver and acquires the SCLC phenotype through a distinct route, which plausibly explains heterogeneous and lower DLL3 expression in that population. The manuscript should add this genetic grounding, with a citation to the relevant transformation literature it already references (Wang et al., ref. 14; Aredo et al., ref. 16), rather than relying on "lineage plasticity" as an unexplained term.
Third, the manuscript should incorporate the four-subtype neuroendocrine transcriptional framework (ASCL1, NEUROD1, POU2F3, YAP1) in Section 4.1 or 4.6. ASCL1-driven tumors characteristically show higher DLL3 expression and are expected to yield deeper, more durable responses to DLL3-directed therapy. None of the five included real-world studies report transcriptional subtype, which the manuscript should state as an explicit evidence gap in Section 4.6 (Future Directions) and recommend that future real-world registries capture ASCL1/NEUROD1/POU2F3/YAP1 status alongside DLL3 immunohistochemistry.

Minor 
The toxicity discussion in Section 4.2.1 would benefit from a brief mechanistic grounding of why CRS occurs with bispecific T-cell engagers specifically (IL-6-dominant cytokine release following T-cell activation), distinct from tumor-intrinsic inflammatory biology. The manuscript already notes that patients with higher baseline oxygen requirements and CNS metastases experienced more frequent immune-mediated toxicity; a one-sentence mechanistic link between baseline inflammatory burden and CRS severity would strengthen this without requiring new analysis.
Two editorial items should also be corrected before publication. The funding statement currently reads "Please add: This research received no external funding," which appears to be an unaddressed template placeholder and should be finalized. Author 4 is listed as "Konstantinos Arnaoutakis and MD," which should read "Konstantinos Arnaoutakis, MD." The authors should also confirm the PROSPERO registration number (CRD420261345046) is correctly transcribed, as its length differs from the standard CRD42-plus-seven-digit format.

Recommendation
Major revision. The genetic and transcriptional-subtyping additions requested above are not cosmetic; they materially strengthen the manuscript's explanatory account of why tarlatamab activity diverges between de novo and transformed SCLC, which is the review's most clinically important finding. The remaining items (evaluable-subset labeling, CRS mechanism, editorial corrections) are minor and can be addressed concurrently.

Author Response

please refer to the attachment

Author Response File: Author Response.docx