Integrating Functional Genomic Screens and Multi-Omics Data to Construct a Prognostic Model for Lung Adenocarcinoma and Validating SPC25

Round 1

Reviewer 1 Report

Comments and Suggestions for Authors

Overall Evaluation

This manuscript integrates DepMap CRISPR screening data with TCGA and GEO datasets to construct a prognostic model for lung adenocarcinoma. While the concept is interesting, there are several major concerns regarding the limited novelty, methodological consistency, and depth of experimental validation.

In particular, the model-building strategy overlaps substantially with previously published work, and the experimental validation of SPC25 remains at a preliminary level. To substantiate the main conclusions, additional analyses, external validation, and mechanistic experiments are indispensable.

 

Major Comments

1. Lack of Clinical Significance

The correlation between the risk score and conventional clinical variables such as age and tumor stage is expected, and therefore, the model’s value as a novel clinical stratification tool is currently limited.

To strengthen its clinical relevance, the authors should:

Provide a multivariate comparison (HR, C-index) with existing prognostic models (e.g., TNM stage, LIPI score, IASLC immune score) in a dedicated table.

Analyze the association between the risk score and PFS/ORR in a clinical cohort treated with immune checkpoint inhibitors.

Perform stage- and treatment-specific subgroup analyses (e.g., Stage I–II vs. Stage III–IV) to verify the generalizability of the prognostic performance.

2. Superficial Experimental Validation of SPC25

Although the SPC25 knockdown results showing reduced proliferation and migration are interesting, the evidence for its direct mechanistic involvement remains insufficient.

The following additional experiments are strongly recommended:

Rescue experiments (re-expression of SPC25) to confirm phenotypic reversal.

Phosphorylation signal analyses (Western blot, RT-qPCR) to explore involvement in the mTORC1, E2F, and MYC pathways.

In vivo xenograft studies to validate the tumorigenic potential of SPC25.

RNA-seq or GSEA following SPC25 knockdown to present a comprehensive view of transcriptomic alterations and provide a unified mechanistic interpretation.

3. Insufficient Depth of Tumor Immune Microenvironment Analysis

The current immune analysis, based on ssGSEA and CIBERSORT, remains predictive and lacks spatial or functional validation.

The authors should further:

Examine correlations between the risk score and the expression of immune checkpoint molecules (PD-1, PD-L1, LAG3, TIGIT, etc.).

Assess the relationship between the risk score and tumor mutation burden (TMB) as well as immune phenotypes (inflamed vs. cold tumors).

Provide a detailed visualization (e.g., UMAP plots) in the Supplementary Materials, showing the immune cell composition and cluster information of SPC25-high cells identified in single-cell RNA-seq analysis.

4. Quantitative Evaluation of HPA Immunostaining

In the Human Protein Atlas (HPA) immunohistochemistry images, representative examples alone are insufficient. The authors should include quantitative data (e.g., H-score distribution) to show the expression trend across all available cases.

Author Response

Q1: Lack of Clinical Significance
The correlation between the risk score and conventional clinical variables such as age and tumor stage is expected, and therefore, the model’s value as a novel clinical stratification tool is currently limited.To strengthen its clinical relevance, the authors should:
Q(1.1) Provide a multivariate comparison (HR, C-index) with existing prognostic models (e.g., TNM stage, LIPI score, IASLC immune score) in a dedicated  table.                                                    

A(1.1): Thank you for your comments. We have already demonstrated the independent prognostic value of our model in the original manuscript through: Multivariate Cox regression (forest plot) incorporating TNM stage, age, and sex, confirming the risk score as an independent prognostic factor（Figure 3C）.Time-dependent ROC curves, calibration curves, and DCA, demonstrating the model's predictive accuracy and clinical utility（Figure 3D-Figure 3G).A new comparison in the supplements with recently published LUAD gene signatures, further underscoring the superior performance of our model(Supplementary Material 2).Regarding comparison with LIPI and IASLC scores, we were unable to compute these directly due to the lack of necessary raw data (blood parameters and pathological cell counts) in public databases. The uniqueness of our model lies in its foundation of intrinsic dependency genes from CRISPR screening. This provides a complementary dimension to assessments based on systemic inflammation (LIPI) or immune morphology (IASLC), directly revealing core vulnerabilities driving tumor growth.

Q (1.2）: Analyze the association between the risk score and PFS/ORR in a clinical cohort treated with immune checkpoint inhibitors.               

A (1.2): Thank you for your comments. We fully acknowledge the value of validating our risk score in an independent immunotherapy cohort with genomic data. However, as most clinical trial datasets with treatment response (PFS/ORR) lack publicly available gene expression data required to compute our risk score, such direct validation remains technically unfeasible. To address this limitation, we employed established computational proxies: TIDE analysis demonstrated significantly higher scores in our high-risk group, predicting innate immunotherapy resistance（Figs. 6E, F）, whileanalysis of the real-world Patil2022-OAK cohort confirmed that high expression of every signature gene predicts significantly worse survival following immunotherapy（Fig 9）. These convergent results provide robust indirect evidence supporting our model's relevance to immunotherapy outcomes, despite the current data accessibility constraints.

Q(1.3): Perform stage- and treatment-specific subgroup analyses (e.g., Stage I–II vs. Stage III–IV) to verify the generalizability of the prognostic performance.

A(1.3): We performed the recommended subgroup analysis(Supplementary Material 1). The risk score demonstrated powerful and statistically significant prognostic stratification in early-stage (Stage I-II) patients (Log-rank P < 0.001), but not in advanced-stage (Stage III-IV) patients.                               

We interpret this not as a limitation, but as a key finding that refines the clinical utility of our model. It strongly suggests that our signature is particularly effective for risk stratification in early-stage LUAD, where identifying high-risk patients for adjuvant therapy is a major clinical challenge. In contrast, the prognosis of advanced-stage disease is likely dominated by treatment history and acquired resistance, which may overshadow the signal from our core dependency genes. This result precisely positions our model as a valuable tool for personalized management in early-stage disease

Q2: Superficial Experimental Validation of SPC25
Although the SPC25 knockdown results showing reduced proliferation and migration are interesting, the evidence for its direct mechanistic involvement remains insufficient.
The following additional experiments are strongly recommended:       

Rescue experiments (re-expression of SPC25) to confirm phenotypic reversal. Phosphorylation signal analyses (Western blot, RT-qPCR) to explore involvement in the mTORC1, E2F, and MYC pathways.                                                             

In vivo xenograft studies to validate the tumorigenic potential of SPC25.  

RNA-seq or GSEA following SPC25 knockdown to present a comprehensive view of transcriptomic alterations and provide a unified mechanistic interpretation.

A2: We thank the reviewer for the constructive feedback. To directly address the concern regarding in vivo tumorigenic potential, we have now performed xenograft assays, which provide definitive evidence that SPC25 knockdown suppresses tumor growth in vivo (Fig. 10E). We believe this critical experiment, combined with our  in vitro and in vivo data, solidly validates SPC25's oncogenic role within the scope of this prognostic model study. Deeper mechanistic explorations (rescue, signaling, transcriptomics) are part of an ongoing, separate investigation.

Q3: Insufficient Depth of Tumor Immune Microenvironment Analysis
The current immune analysis, based on ssGSEA and CIBERSORT, remains predictive and lacks spatial or functional validation.
The authors should further:
Q(3.1): Examine correlations between the risk score and the expression of immune checkpoint molecules (PD-1, PD-L1, LAG3, TIGIT,etc.).        

A(3.1): We thank the reviewer for this suggestion. We examined the correlation between our risk score and key immune checkpoint molecules. The results revealed a nuanced and biologically important pattern:

The expression of CTLA-4 and TIGIT was significantly higher in the low-risk group. The expression of PD-1 and PD-L1 showed no statistically significant difference between the risk groups. We interpret these findings as strong evidence that the low-risk group possesses a more active, T-cell-inflamed tumor microenvironment. CTLA-4 and TIGIT are often upregulated on activated T cells as part of a natural feedback mechanism to prevent over-activation. Therefore, their higher expression in the low-risk group is likely a consequence of robust T-cell activation and ongoing anti-tumor immune responses, consistent with their favorable prognosis. In contrast, the lack of difference in PD-1/PD-L1 suggests that this particular axis may be similarly regulated or saturated in both groups and is not the primary driver of the prognostic differences captured by our model. This analysis further clarifies the distinct immunological landscape associated with our risk signature.

Q(3.2): Assess the relationship between the risk score and tumor mutation burden (TMB) as well as immune phenotypes (inflamed vs. cold tumors).   

A(3.2): We thank the reviewer for raising this important point. In direct response, we have performed the requested analyses(Figs 6J-K).

Q(3.3): Provide a detailed visualization (e.g., UMAP plots) in the Supplementary Materials, showing the immune cell composition and cluster information of SPC25-high cells identified in single-cell RNA-seq analysis. 

A(3.3)As requested, the UMAP plot in Figure 10F (generated from TISCH) already provides the detailed visualization, showing both immune cell composition and specific localization of SPC25-high cells within tumor and Treg clusters. This directly addresses your point.

Q4: Quantitative Evaluation of HPA Immunostaining
In the Human Protein Atlas (HPA) immunohistochemistry images, representative examples alone are insufficient. The authors should include quantitative data (e.g., H-score distribution) to show the expression trend across all available cases.

A4: We thank the reviewer for this valuable suggestion. We have now performed a quantitative analysis of available LUAD and normal lung tissue images from the HPA for our seven genes. The results, presented as Average Optical Density (AOD) in the new Figure 8, confirm that all seven proteins are significantly upregulated in LUAD. This quantitative data strengthens the validation of our signature and has been added to the Results section.

Author Response File: Author Response.pdf

Reviewer 2 Report

Comments and Suggestions for Authors

1. The abstract is crisp and clearly written.
2. Introduction: It is advised to add a few previous studies. Mention what past studies have accomplished and what remains to be explored in this area.
3. The background on lung cancer is good. When mentioning tumour heterogeneity, please add one sentence to explain the main type of heterogeneity seen in LUAD.
4. Explain why CRISPR screens are better than RNAi for finding dependency genes, and give one example from other cancer research.
5. Add a short discussion on current treatments for LUAD and the challenges in improving outcomes.
6. Please highlight the novelty of this study at the end of the introduction. Mention clearly if SPC25 has or has not been reported before in LUAD.
7. Please state how many genes were entered into the univariate Cox analysis.
8. For the ROC curves, mention the time points used (e.g., 1-, 3-, and 5-year survival).
9. Correct the term to “Gene Set Enrichment Analysis (GSEA).”
10. In the drug response section, please explain how drug sensitivity was compared between high-risk and low-risk groups.
11. Mention whether the predicted drug responses agree with previously published studies.
12. Consider adding a simple workflow figure showing all major analysis steps.
13. Clarify which medium was used for each cell line and why.
14. State whether more than one shRNA was tested, and confirm that a control shRNA was used.
15. Add 1–2 lines explaining the BCA protein assay.
16. For the wound healing assay, mention the medium used, incubation conditions, and magnification of images. A brief explanation for all experimental studies is required. Authors are advised to go through and cite the papers for references: 10.3390/pharmaceutics17020270
17. For the Western blot, mention protein loading volume and transfer conditions.
18. Check units for consistency (e.g., use “h”, ensure spacing before °C).
19. Add the incubation time for the colony formation assay.
20. Authors need to add the number of biological replicates (e.g., n = 3) for all experiments.
21. Mention the microscope model and magnification used for taking images for all experiments.
22. Include the risk score formula also in the Supplementary Material.
23. Comment briefly on how your AUC values compare with other LUAD signatures.
24. The wound healing images are blurry. Please replace them with higher-resolution images.
25. Write the conclusion as a separate section with its own heading, and include the limitations and drawbacks of the current study in the conclusion part.

Author Response

Q1: The abstract is crisp and clearly written.

A1: We thank the reviewer for their positive comment regarding the clarity and quality of our abstract. We are pleased that the core message of our work was effectively communicated.

Q2：Introduction: It is advised to add a few previous studies. Mention what past studies have accomplished and what remains to be explored in this area.

A2：We fully agree with this suggestion. As requested, we have now added a dedicated sentence in the second paragraph to acknowledge the significant contributions of previous studies and to clearly define the research gap our work aims to fill. The added text states: “Indeed, previous studies have established numerous prognostic signatures based on gene expression, mutations, or epigenetic alterations, demonstrating their utility in risk stratification and subtype identification. However, many of these models are derived primarily from transcriptomic correlations with patient survival, lacking a direct foundation in functional genetic essentiality data, which may better reflect core tumor vulnerabilities”.

Q3: The background on lung cancer is good. When mentioning tumour heterogeneity, please add one sentence to explain the main type of heterogeneity seen in LUAD.

A3: Thank you for this positive feedback and specific suggestion. We have now added a sentence to clarify the primary forms of heterogeneity in LUAD. The new sentence reads: “In LUAD, this heterogeneity manifests predominantly at the genomic, transcriptomic, and proteomic levels, leading to diverse cellular phenotypes and functional variation within the tumor. ”

Q4:Explain why CRISPR screens are better than RNAi for finding dependency genes, and give one example from other cancer research.

A4: This is an excellent point that strengthens the rationale for our methodology. We have incorporated the requested explanation and example into the third paragraph: “CRISPR-based screens are superior to earlier RNA interference (RNAi) techniques for identifying dependency genes due to their higher efficiency, greater specificity, and reduced off-target effects, enabling more complete gene knockout and thus more reliable identification of essential genes. For instance, CRISPR screens in other malignancies like triple-negative breast cancerhave successfully uncovered novel dependencies such as the Cop1, which would have been missed by less robust methods.”

Q5: Add a short discussion on current treatments for LUAD and the challenges in improving outcomes.

A5:We have expanded the background on LUAD by adding a new sentence that succinctly summarizes current treatments and the persistent challenges, as suggested: “ Current standard-of-care for LUAD includes surgical resection for early-stage disease, platinum-based chemotherapy, and increasingly, targeted tThank you for this suggestion. We have revised the end of the introduction to clearly highlight the novelty of our study and the status of SPC25 in LUAD. The final paragraph now reads: "To experimentally corroborate the clinical relevance of our model, we validated SPC25—a key gene emerging from our signature—and confirmed its oncogenic role. The primary novelty of our work lies in this integrative framework that prioritizes prognostic genes based on functional essentiality. By demonstrating that a model-derived gene like SPC25 possesses significant biological and clinical relevance, we strengthen the credibility of both the target and the modeling approach itself.." This revision:Explicitly states the methodological novelty of our functionally-informed integrative framework. Accurately positions SPC25 not as a completely novel discovery, but as a key model-derived candidate whose biological and clinical relevance we have experimentally validated, thereby strengthening the credibility of our overall approach. or acquired resistance, tumor heterogeneity, and a lack of predictive biomarkers, which ultimately results in a stubbornly low five-year survival rate for LUAD of approximately 20.5% .”

Q6:Please highlight the novelty of this study at the end of the introduction. Mention clearly if SPC25 has or has not been reported before in LUAD.

A6: Thank you for this suggestion. We have revised the end of the introduction to clearly highlight the novelty of our study and the status of SPC25 in LUAD. The final paragraph now reads: "To experimentally corroborate the clinical relevance of our model, we validated SPC25 — a key gene emerging from our signature — and confirmed its oncogenic role. The primary novelty of our work lies in this integrative framework that prioritizes prognostic genes based on functional essentiality. By demonstrating that a model-derived gene like SPC25 possesses significant biological and clinical relevance, we strengthen the credibility of both the target and the modeling approach itself." This revision:Explicitly states the methodological novelty of our functionally-informed integrative framework. Accurately positions SPC25 not as a completely novel discovery, but as a keymodel-derived candidate whose biological and clinical relevance we have experimentally validated, thereby strengthening the credibility of our overall approach.

Q7: Please state how many genes were entered into the univariate Cox analysis.

A7: We thank the reviewer for this suggestion. We have now specified the number of genes in the ‘Construction of a prognostic model’ section of the Materials and Methods. The revised text reads: "A univariate regression analysis was performed on the 38 candidate genes, and only those with a p-value less than 0.05 were chosen for subsequent analysis. "

Q8: For the ROC curves, mention the time points used (e.g., 1-, 3-, and 5-year survival).

A8: We have added the specific time points to the ‘Verification and assessment of a predictive model’ section. The text now states: "The 'timeROC' package in R was utilized to generate the receiver operating characteristic (ROC) curve and assess the model's discriminative power through the computation of area under the curve (AUC) metrics at 1, 3, and 5 years."

Q9: Correct the term to “Gene Set Enrichment Analysis (GSEA).

A9: We apologize for this oversight. The term has been corrected to "Gene Set Enrichment Analysis (GSEA)" in the ‘Functional enrichment analysis associated with risk score’ section.

Q10: In the drug response section, please explain how drug sensitivity was compared between high-risk and low-risk groups.

A10:We have added the following explanation in the ‘Chemotherapy response prediction’ section: “The differences in drug sensitivity (as reflected by the predicted IC50 values) between the high-risk and low-risk groups were statistically compared using the Wilcoxon rank-sum test. A lower IC50 value in a group indicates higher sensitivity to the drug.”

Q11: Mention whether the predicted drug responses agree with previously published studies.

A11: We have added a discussion on this point in the Results section. For example: ”The predicted sensitivity of the low-risk group to p53-targeting agents (Nutlin-3a and PRIMA-1MET) is mechanistically supported. As Nutlin-3a activates wild-type p53 and PRIMA-1MET reactivates mutant p53, the efficacy in our low-risk group implies these tumors rely on functional p53 pathways — a finding consistent with prior studies on PRIMA-1METcontext-specific efficacy.”

Q12: Consider adding a simple workflow figure showing all major analysis steps.

A12:An illustrative diagram has been added, depicting the key analytical steps from data acquisition to model construction and validation, as Figure 1.

Q13: Clarify which medium was used for each cell line and why.

A13: We have specified the media in the ‘Cell culture’ section: “Calu3 and PC9 lung adenocarcinoma cells (Key Laboratory of Respiratory Inflammatory Injury and Precision Diagnosis and Treatment, Chongqing Municipal Health Commission) were cultured in DMEM or RPMI 1640 medium (supplemented with 10% FBS and 1% penicillin/streptomycin), as these are the standard media optimized for the growth of these specific cell lines.respectively.”

Q14: State whether more than one shRNA was tested, and confirm that a control shRNA was used.

A14: We have added this detail: “Three distinct SPC25-targeting shRNA sequences and a non-targeting control shRNA were used (Obio Technology). Stable SPC25-knockdown lines were established by lentiviral transduction followed by puromycin selection.”

Q15: Add 1–2 lines explaining the BCA protein assay.

A15: We have added a brief explanation: “Total proteins were extracted using RIPA lysis buffer (Beyotime), quantified via a bicinchoninic acid (BCA) assay kit (Beyotime), which determines protein concentration through a colorimetric reaction of protein with Cu ² ⁺ in an alkaline medium. ”

Q16: For the wound healing assay, mention the medium used, incubation conditions, and magnification of images. A brief explanation for all experimental studies is required. Authors are advised to go through and cite the papers for references: 10.3390/pharmaceutics17020270.

A16: We thank the reviewer for this suggestion. We have now revised the ‘Wound healing assay’ section in the Materials and Methods to include the specific medium used, incubation conditions, and image magnification as requested. A brief explanation of the assay's purpose has also been added. The recommended reference has been duly cited to provide methodological context. “Cell migration ability was assessed using a wound healing assay. In brief, cells were plated in 6-well plates and cultured to 80 – 100% confluence in complete growth medium. A uniform scratch wound was created using a sterile 200 μ L pipette tip. The dislodged cells wereremoved by washing twice with PBS, and then fresh serum-free medium was added to minimize the influence of cell proliferation. The plates were incubated under standard culture conditions (37 ° C, 5% CO ₂ ). Microscopic images of identical fields were captured at 0 and 48 h using an inverted microscope (Olympus IX73) at 100 × magnification. The wound closure percentage was quantified using ImageJ software by measuring the residual cell-free area at 48 h relative to the initial wound area at 0 h, as previously described.”

Q17: For the Western blot, mention protein loading volume and transfer conditions.

A17: We have added these details: “ Equal amounts of protein (25ug per lane) were separated by 12.5% SDS-PAGE and electrophoretically transferred to PVDF membranes using a wet transfer system at 200mA for 50 minutes. ”

Q18: Check units for consistency (e.g., use “h”, ensure spacing before °C).

A18: We have thoroughly checked and corrected the units throughout the manuscript, ensuring a space before "°C" (e.g., "37 °C").

Q19: Add the incubation time for the colony formation assay.

A19: We have added the incubation time: “Cells (5 × 10 3 /well)were seeded in each well of a 6-well plate for the colony formation experiment. When the colonies were visible to the naked eye (240 h). ”

Q20: Authors need to add the number of biological replicates (e.g., n = 3) for all experiments.

A20: We have added specific descriptions in the Statistics and Methods section. “All experiments were conducted with at least three biological replicates.”

Q21: Mention the microscope model and magnification used for taking images for all experiments.

A21: As partially addressed in Comment 16, we have now specified the microscope model (Olympus IX73) and magnifications used for all imaging experiments (e.g., wound healing, EdU assay) in the respective method sections.

Q22: Include the risk score formula also in the Supplementary Material.

A22:We thank the reviewer for this suggestion. The risk score formula is already presented in the main text of the manuscript (in the Methods section, Page 13). To avoid redundancy, we have not duplicated it in the Supplementary Material.

Q23: Comment briefly on how your AUC values compare with other LUAD signatures.

A23: We added comparisons with similar studies to the results: “These robust AUC values validate the competitive performance of our model against existing prognostic signatures for LUAD (Supplementary Material 2)”.

Q24: The wound healing images are blurry. Please replace them with higher-resolution images.

A24: We apologize for the image quality. The wound healing images have been replaced with higher-resolution versions.

Q25: Write the conclusion as a separate section with its own heading, and include the limitations and drawbacks of the current study in the conclusion part.

A25:We have restructured the manuscript according to this suggestion. The final part of the Discussion has been moved to a new, standalone section titled "Conclusion." This section now succinctly summarizes the study's findings and explicitly includes the limitations: In conclusion, this study successfully developed and validated a novel 7-gene prognostic signature based on lung adenocarcinoma dependence genes (LADGs). The signature demonstrated robust performance in predicting overall survival and was established as an independent prognostic factor for LUAD patients. Through comprehensive bioinformatics analyses and experimental validation, we revealed the critical roles of these LADGs in tumor proliferation, metastasis, metabolic reprogramming, and immune modulation. The signature's significant association with the tumor microenvironment and drug sensitivity profiles provides valuable insights for developing personalized treatment strategies, particularly in immunotherapy. Notwithstanding these findings, several limitations of our study should be acknowledged. Firstly, the primary analysis was conducted using publicly available databases, which may contain inherent biases and lack certain clinical details. Future prospective, multicentre studies with substantial real-world sample sizes are necessary for clinical validation. Secondly, while we validated the oncogenic role of SPC25 through in vitro and in vivo experiments, the functional characterization of the other six genes in our signature remains to be fully elucidated. Thirdly, the practical application of our risk score model requires certain technical expertise and resource inputs, which might limit its immediate widespread clinical adoption. Despite these limitations, our study provides a valuable framework for understanding genotype-specific vulnerabilities in LUAD and offers expanded.

Author Response File: Author Response.pdf

Reviewer 3 Report

Comments and Suggestions for Authors

This study aimed to construct a prognostic model for lung adenocarcinoma (LUAD) and verify the function of the hub gene SPC25. By integrating CRISPR-Cas9 screening data from the DepMap database with transcriptome data from The Cancer Genome Atlas (TCGA) cohort, we identified 317 LUAD-associated dependency genes (LADGs), based on which a prognostic model consisting of seven genes including CCT6A and MCM7 was constructed. This model showed good predictive performance in both the TCGA and Gene Expression Omnibus (GEO) datasets, where high-risk patients had significantly shorter overall survival times, advanced tumor stages, and immunosuppressive microenvironments. Functional experiments demonstrated that knockdown of SPC25 significantly inhibited the proliferation, migration, and colony formation abilities of LUAD cells. Single-cell sequencing revealed increased expression of SPC25 in tumor cells and specific immunosuppressive T cell subsets. In addition, overexpression of the model genes was associated with poor response to immunotherapy and could successfully predict different chemosensitivity. Our research provides a reliable evaluation system for the prognosis of LUAD and suggests that SPC25 is a potential therapeutic target. It is recommended that the author revise the manuscript in accordance with the following comments.

1. It is recommended that the author allocate additional space in the introduction section to include comprehensive background information on lung cancer.
2. The following references are highly relevant to the author's topic and are recommended for citation.

[1] Wu YL, Lu S, Zhou Q, Zhang L, Cheng Y, Wang J, et al. Expert consensus on treatment for stage III non-small cell lung cancer. Med Adv. 2023; 1(1): 3–13. https://doi.org/10.1002/med4.7

[2] A. Gu, J. Li, M.-Y. Li, Y. Liu, Patient-derived xenograft model in cancer: establishment and applications. MedComm, 2025, 6, e70059. DOI: 10.1002/mco2.70059

[3] Tokhanbigli S, Haghi M, Dua K, Oliver BGG. Cancer-associated fibroblast cell surface markers as potential biomarkers or therapeutic targets in lung cancer. Cancer Drug Resist. 2024;7:32. http://dx.doi.org/10.20517/cdr.2024.55

3. Incorporating more patient data from different regions, ethnicities and medical institutions in future studies is recommended to enhance the statistical power and generalizability of the model, thereby strengthening the robustness of results.
4. Conduct animal model experiments to observe the effects of SPC25 knockout or inhibition on tumor growth and metastasis, comprehensively evaluating its potential as a therapeutic target to support clinical translation.
5. Consider integrating additional clinical indicators (such as patients’ lifestyle and comorbidities) and molecular biomarkers during model construction, utilizing advanced algorithms such as machine learning to improve the predictive performance and practicality of the model.
6. Validate the model in more independent cohorts and various types of lung cancer patients while comparing it with other published prognostic models to highlight the advantages and uniqueness of the proposed model.

Author Response

Q1: It is recommended that the author allocate additional space in the introduction section to include comprehensive background information on lung cancer.

A1: We thank the reviewer for this valuable suggestion. We have significantly expanded the background information on lung cancer in our introduction. Specifically, we have added:

“(1)Details on the therapeutic challenges of stage III NSCLC and reference to recent expert consensus .

(2)A discussion on the importance of advanced research models like patient-derived xenografts in studying tumor heterogeneity.

(3)Information about the role of cancer-associated fibroblasts in therapy response and resistance mechanisms.

(4)An overview of current LUAD treatments and the persistent challenges in improving patient outcomes.”

Q2: The following references are highly relevant to the author's topic and are recommended for citation.

A2: We appreciate the reviewer's suggestion and have incorporated all three recommended references into our introduction。

Q3: Incorporating more patient data from different regions, ethnicities and medical institutions in future studies is recommended to enhance the statistical power and generalizability of the model, thereby strengthening the robustness of results.
A3: We completely agree with this important recommendation. In the final paragraph of our introduction, we have explicitly acknowledged this limitation and outlined it as a direction for future research:

"We acknowledge that future validation in more diverse cohorts incorporating additional clinical variables and advanced computational approaches will be essential to enhance the model's generalizability and clinical utility."

Q4: Conduct animal model experiments to observe the effects of SPC25 knockout or inhibition on tumor growth and metastasis, comprehensively evaluating its potential as a therapeutic target to support clinical translation.

A4: We thank the reviewer for the constructive feedback. To directly address the concern regarding in vivo tumorigenic potential, we have now performed xenograft assays, which provide definitive evidence that SPC25 knockdown suppresses tumor growth in vivo (Fig. 10E).

Q5: Consider integrating additional clinical indicators (such as patients’ lifestyle and comorbidities) and molecular biomarkers during model construction, utilizing advanced algorithms such as machine learning to improve the predictive performance and practicality of the model.

A5:Thank you for this valuable suggestion. We fully agree that integrating additional clinical indicators (such as lifestyle and comorbidities) and utilizing machine learning algorithms are crucial for enhancing the predictive performance and clinical practicality of the model.

We would like to clarify that our current model was constructed using public genomic databases like TCGA. While these databases provide high-quality genomic and basic clinical data (e.g., stage, overall survival), they lack the detailed clinical information you mentioned, such as specific patient lifestyles and comorbidities. This data limitation prevented us from directly incorporating these factors into the current study.

Notwithstanding this, the primary value of our present model lies in its novel perspective—it is the first prognostic signature built upon the concept of "lung adenocarcinoma dependency genes," offering complementary insights into the biological underpinnings of the tumor. Our plan for future work, using prospective cohorts with richer clinical data, is to explicitly follow the reviewer's advice by employing advanced machine learning techniques to build a more comprehensive prognostic platform.

Q6:Validate the model in more independent cohorts and various types of lung cancer patients while comparing it with other published prognostic models to highlight the advantages and uniqueness of the proposed model.

A6: We sincerely thank the reviewer for this insightful suggestion. In response, we have now added a systematic comparison with several recently published LUAD prognostic signatures in the Results section, demonstrating that our model shows comparable or superior predictive performance (Supplementary Material 2).

Furthermore, while the current study validates the model in two independent LUAD cohorts, we fully acknowledge the value of broader validation and have explicitly stated in the Discussion that extending this work to other lung cancer subtypes constitutes an important direction for our future research.

Author Response File: Author Response.pdf

Round 2

Reviewer 2 Report

Comments and Suggestions for Authors

May be accepted for publication. 

Author Response

We sincerely thank Reviewer 2 for their positive feedback and for accepting our manuscript for publication.