Actionable Genomic Alterations and Survival in Gallbladder Cancer: A Documented Stage- and Treatment-Matched Real-World Global Analysis

Round 1

Reviewer 1 Report

Comments and Suggestions for Authors

This manuscript by Solangi and colleagues leverages the TriNetX Global Collaborative Network to investigate whether the presence of documented actionable genomic alterations (AGAs) is associated with overall survival in gallbladder cancer (GBC), independent of traditional clinical confounders. The authors employ a tiered propensity score matching strategy, first controlling for demographics (Model 1) and subsequently adding key oncologic variables—metastatic status, surgical resection, and chemotherapy history (Model 2). The core finding is that the AGA-positive cohort consistently exhibited a higher absolute risk of mortality compared to the cohort without documented AGAs, even after rigorous stage- and treatment-matching. The study also highlights stark disparities in genomic documentation, with a significantly higher proportion of "unknown race" in the cohort lacking documented AGAs.

The key contributions of this work lie in its scale (a global, real-world cohort) and its methodological attempt to isolate the prognostic signal of genomics from clinical decision-making. The observation that the survival disadvantage persists after controlling for treatment intensity is a clinically relevant finding that supports the push for broader molecular profiling. Furthermore, the documented demographic disparity in testing access serves as an important, albeit secondary, call to action regarding health equity in precision oncology.

 

Detailed Evaluation

Introduction. The introduction is concise and effectively sets the stage. It correctly identifies the central problem in GBC research: the difficulty in decoupling the prognostic impact of tumor biology from the confounding effects of late-stage presentation and variable treatment. The rationale for the study is clearly stated, and the authors appropriately position their work as an attempt to provide real-world evidence to address this knowledge gap. However, the introduction could be strengthened by a brief acknowledgment of the specific limitations of the TriNetX database regarding granularity of treatment data (e.g., margin status, lymph node yield, specific chemotherapy regimens), as this would pre-emptively address a major limitation discussed later.

Methodology. This is the most critical section of the manuscript, and it contains both strengths and notable areas of concern.

• Strengths: The tiered matching strategy (Model 1 and Model 2) is methodologically sound and demonstrates the authors' understanding that demographic balance alone is insufficient. The use of standardized mean differences (SMD) to assess balance is appropriate. The decision to restrict the analysis to patients diagnosed after 2015 (the contemporary genomic era) is a wise one to reduce historical bias.

• Critical Concerns: Cohort Definition (The "Gene-unmutated" Group): This is the most significant methodological flaw. The authors define the comparator group as the "Gene-unmutated cohort" with "No documented actionable genomic alteration." However, they later acknowledge in the discussion that this group "likely included patients who were never genomically profiled." This is not just a limitation; it fundamentally misrepresents the comparator. This cohort is not a "wild-type" or "unmutated" group; it is a "genomically untested or undocumented" group. This distinction is crucial. The association being measured is not "AGA-positive vs. wild-type," but rather "documented AGA-positive vs. not documented." This conflates tumor biology with testing behavior and healthcare access. A more accurate and honest label for Cohort 1 would be "No Documented AGA (including untested patients)." This distinction must be made unequivocally throughout the manuscript, starting with the title and abstract. Matching Covariates: While including "surgical resection" and "chemotherapy history" is a step in the right direction, it is a relatively blunt instrument for a surgical oncologist. There is no granularity regarding the quality of the surgery (e.g., R0 vs. R1 resection, extent of lymphadenectomy) or the type of chemotherapy (e.g., gemcitabine-based vs. targeted therapies). For a study aiming to isolate biological impact, the failure to match for margin status (R0 resection) is a major limitation, as R0 status is one of the most powerful prognostic factors in GBC and is likely correlated with the decision to perform genomic testing. AGA Panel Heterogeneity: The selected gene panel (KRAS, TP53, ERBB2, IDH1, FGFR1, PIK3CA, ARID1A) is a reasonable starting point. However, treating all "actionable" alterations as a single, monolithic entity is problematic. The biology, therapeutic implications, and prognostic impact of a KRAS mutation are vastly different from an IDH1 mutation or an ERBB2 amplification. The authors acknowledge this in their exploratory subgroup analysis, but the primary analysis, which drives the main conclusion, lumps them all together. This risks oversimplifying a complex molecular landscape.

Results. The results are presented clearly, and the disparity in demographic documentation (51.0% unknown race in the comparator group vs. 3.7% in the AGA group) is a striking and important finding that underscores a major selection bias. Model 1 vs. Model 2: The observation that the absolute risk difference (RD) increased from 10.3% in Model 1 to 13.1% in Model 2 is counterintuitive. One would expect that controlling for stage and treatment (confounders on the causal pathway) would attenuate the effect size, not increase it. The authors do not comment on this. This paradoxical finding could suggest that the matching in Model 2, while statistically balanced, inadvertently selected for a sicker subgroup within the AGA-positive cohort, or that unmeasured confounders (like R0 status) are at play. This warrants a deeper discussion. Survival Analysis and Proportional Hazards: The authors correctly identify the violation of the proportional hazards assumption in Model 2 and appropriately pivot to absolute risk measures. This is a sophisticated and correct analytical decision. The interpretation that early interventions may stabilize survival before underlying biology drives late divergence is plausible and insightful. However, the Kaplan-Meier curves in Figure 2b for Model 2 do not visually support the narrative of a "late accelerator" effect; the curves appear to overlap significantly for the entire duration, with the p-value for the log-rank test being non-significant (0.925). The authors rely on the risk difference (56.2% vs. 43.0%) as the primary evidence, but the visual representation of survival over time should align with this interpretation. Exploratory Subgroup Analyses: The analysis of KRAS and TP53 is interesting but must be interpreted with extreme caution, as the authors do. However, comparing the KRAS-mutated group (N=?) to a matched cohort without a documented KRAS alteration suffers from the same fundamental flaw as the main analysis. Furthermore, performing a survival analysis on a subgroup derived from a previously matched cohort (Model 2) can introduce statistical artifacts. This section should be clearly labeled as hypothesis-generating only and should not be over-interpreted.

Discussion and Conclusions. The discussion is well-written and balanced. The authors acknowledge the major limitations, including the problematic definition of the comparator cohort, the retrospective design, and the lack of granular treatment data. The emphasis on disparities in testing access is a valuable and relevant addition to the literature. The conclusion states that "documented actionable genomic alterations... were associated with higher mortality." While this is true based on their analysis, the more accurate conclusion is that "in this real-world dataset, patients with a documented AGA had worse outcomes, even after accounting for stage and treatment, highlighting both the potential prognostic significance of these alterations and the profound inequities in who receives genomic testing." The current conclusion slightly overstates the evidence regarding the direct biological impact, given the comparator group issue.

Constructive Feedback for Authors

Dear Authors,

Thank you for the opportunity to review your interesting and timely manuscript. You have tackled a critically important question regarding the independent prognostic value of molecular profiling in GBC, a disease where real-world evidence is sorely needed. The scale of your analysis and your careful methodological approach are commendable. However, several key issues need to be addressed to strengthen the manuscript and ensure its findings are interpreted correctly.

1. Reframe the Comparator Cohort: This is the most critical revision. Please rename the comparator group throughout the manuscript (title, abstract, methods, results, discussion) from "gene-unmutated" to something more accurate, such as "no documented AGA (including untested)." The current terminology is misleading and implies a biological certainty that does not exist. Your study is comparing patients with a documented test result against those without. This distinction is fundamental to the validity of your conclusions.

2. Clarify the Abstract and Conclusions: The abstract and conclusion sections must be revised to reflect the accurate definition of the comparator group. Instead of stating that the "AGA cohort had higher mortality," it would be more precise to state that "patients with a documented AGA experienced higher mortality compared to those without documented alterations, even after rigorous matching." This subtle change more accurately represents your study design and findings.

3. Address the Paradoxical Risk Difference: Please provide a commentary in the discussion on why the absolute risk difference increased from Model 1 (10.3%) to Model 2 (13.1%) after controlling for stage and treatment. This is a non-intuitive finding that readers will likely question. Was there a difference in the composition of the cohorts before matching for Model 2? Could this be explained by a higher proportion of patients with certain high-risk AGAs (like KRAS) being more likely to have complete oncologic data documented? Exploring this will demonstrate a deeper engagement with your data.

4. Acknowledge the Bluntness of Treatment Matching: Please expand on the limitations of using binary variables for surgical resection and chemotherapy. Acknowledge that you were unable to match for margin status (R0 vs. R1/2), the extent of lymphadenectomy, or specific chemotherapy regimens. Given that surgical quality is a dominant prognostic factor in GBC, this is a significant unmeasured confounder that should be prominently discussed.

5. Re-evaluate the KRAS Subgroup Analysis: I recommend removing the formal survival analysis comparing a KRAS-mutated group to a "no documented KRAS" group from the main results section. Given the issues with the main comparator cohort, this subgroup analysis is highly susceptible to bias. If you wish to keep it, it must be moved to a supplementary section and prefaced with even stronger cautionary language about its exploratory nature and the inherent limitations of the comparator definition.

6. Improve Visual Alignment: Please ensure that the narrative regarding the Model 2 survival curves (a late divergence in risk) is visually supported by the graph in Figure 2b. If the curves truly overlap for the entire follow-up period with a non-significant log-rank test, it may be more accurate to state that the survival difference was not statistically significant in the time-to-event analysis and that the primary finding rests on the difference in cumulative mortality risk at the end of follow-up.

Your work has the potential to make a valuable contribution to the field, particularly by highlighting disparities in genomic testing. Addressing these points will significantly enhance the manuscript's clarity, accuracy, and impact.

Comments on the Quality of English Language

Overall, the manuscript is written in a clear, professional, and largely idiomatic scientific English. The syntax, terminology, and flow are appropriate for a journal of Cancers’ scope. However, several recurring issues and minor errors detract from the overall polish and should be corrected during revision.

Typographical and formatting errors:

• The abbreviation “AGA” is introduced in the abstract and used frequently, but in the introduction it is spelled out again; standardization is needed.

• The CONSORT-like figure description refers to “Figure 1” in the methods, but the figure is placed after the methods with a caption; the citation in text should be checked for accuracy.

Redundancy and wordiness:

• Some sentences are overly long and could be split for clarity. For example, in the introduction: “Because a large proportion of patients remain asymptomatic during early disease, diagnosis frequently occurs at advanced stages, necessitating aggressive multidisciplinary care.” This is acceptable, but elsewhere, such as in the discussion, there are sentences that combine multiple ideas (e.g., “In this large real-world analysis of patients with gallbladder cancer, documented actionable genomic alterations were associated with higher mortality, and this association persisted after matching for major demographic, stage-related, and treatment-related variables.”). Streamlining would improve readability.

Precision and consistency:

• The term “gene-unmutated” is used throughout the methods and results, but as noted in the scientific critique, this is a misnomer. While this is a conceptual issue, the language itself should be revised to “cohort without a documented AGA” to avoid misleading terminology.

• In Table 1, the column headers “Gene-unmutated” and “Gene-mutated” are used; these should be changed to align with the corrected terminology.

• Some abbreviations are used without definition (e.g., “HCOs” in the methods is defined, but “TNX, UMLS” in the genomic status section are mentioned without explanation; consider clarifying or removing if not essential).

• In the results section, “p<0.001” sometimes has a space, sometimes not; consistency is needed.

Grammar and syntax:

• Minor subject–verb agreement issues: “The proportional hazards assumption was violated (χ² = 9.979, p = 0.002), indicating a non-constant effect over time. Accordingly, for Model 2, absolute mortality risk and risk difference were considered the primary effect measures As shown in the survival curves...” – there is a missing period and a run-on sentence.

• In the discussion: “By contrast, the TP53 subgroup analysis was limited by small sample size and should not be interpreted as evidence of lack of prognostic relevance.” This is grammatically correct but slightly awkward; consider “...should not be interpreted as evidence that TP53 lacks prognostic relevance.”

• In the conclusion: “These findings support the prognostic relevance of molecular profiling in GBC and reinforce the importance of improving equitable access to genomic testing and documentation.” – the phrase “improving equitable access” is somewhat redundant; “ensuring equitable access” would be more natural.

Figure legends and in-text references:

• The legend for Figure 2 uses “A. GBC documented AGA Vs. no documented AGA. Model 1.” The use of “Vs.” is informal; “versus” or a more descriptive phrase would be better.

• In the text, the forest plot (Figure 3) is mentioned but not fully described; the legend is minimal. Ensure all figures are referenced sequentially and have complete, self-explanatory legends.

Recommendation. The English is generally competent and does not impede understanding, but a careful copyedit to correct typographical errors, standardize terminology, and improve sentence flow is recommended. I would classify the language quality as needing minor to moderate revision.

Author Response

Reviewer 1

 

Open Review

(x) I would not like to sign my review report 
( ) I would like to sign my review report 

Quality of English Language

(x) The English could be improved to more clearly express the research. 
( ) The English is fine and does not require any improvement. 

 

	Yes	Can be improved	Must be improved	Not applicable
Does the introduction provide sufficient background and include all relevant references?	(x)	( )	( )	( )
Is the research design appropriate?	( )	(x)	( )	( )
Are the methods adequately described?	( )	(x)	( )	( )
Are the results clearly presented?	( )	(x)	( )	( )
Are the conclusions supported by the results?	(x)	( )	( )	( )
Are all figures and tables clear and well-presented?	(x)	( )	( )	( )

Comments and Suggestions for Authors

This manuscript by Solangi and colleagues leverages the TriNetX Global Collaborative Network to investigate whether the presence of documented actionable genomic alterations (AGAs) is associated with overall survival in gallbladder cancer (GBC), independent of traditional clinical confounders. The authors employ a tiered propensity score matching strategy, first controlling for demographics (Model 1) and subsequently adding key oncologic variables—metastatic status, surgical resection, and chemotherapy history (Model 2). The core finding is that the AGA-positive cohort consistently exhibited a higher absolute risk of mortality compared to the cohort without documented AGAs, even after rigorous stage- and treatment-matching. The study also highlights stark disparities in genomic documentation, with a significantly higher proportion of "unknown race" in the cohort lacking documented AGAs.

The key contributions of this work lie in its scale (a global, real-world cohort) and its methodological attempt to isolate the prognostic signal of genomics from clinical decision-making. The observation that the survival disadvantage persists after controlling for treatment intensity is a clinically relevant finding that supports the push for broader molecular profiling. Furthermore, the documented demographic disparity in testing access serves as an important, albeit secondary, call to action regarding health equity in precision oncology.

 

Detailed Evaluation

Introduction. The introduction is concise and effectively sets the stage. It correctly identifies the central problem in GBC research: the difficulty in decoupling the prognostic impact of tumor biology from the confounding effects of late-stage presentation and variable treatment. The rationale for the study is clearly stated, and the authors appropriately position their work as an attempt to provide real-world evidence to address this knowledge gap. However, the introduction could be strengthened by a brief acknowledgment of the specific limitations of the TriNetX database regarding granularity of treatment data (e.g., margin status, lymph node yield, specific chemotherapy regimens), as this would pre-emptively address a major limitation discussed later.

Response:

We thank the reviewer for this insightful suggestion. We agree that acknowledging the limitations of real-world data early in the manuscript enhances transparency and manages reader expectations regarding variables like margin status and treatment granularity. We have added a statement to the third paragraph of the Introduction, specifically identifying these limitations of the TriNetX database and explaining how our tiered matching strategy was designed to mitigate their impact.

Methodology. This is the most critical section of the manuscript, and it contains both strengths and notable areas of concern.

• Strengths: The tiered matching strategy (Model 1 and Model 2) is methodologically sound and demonstrates the authors' understanding that demographic balance alone is insufficient. The use of standardized mean differences (SMD) to assess balance is appropriate. The decision to restrict the analysis to patients diagnosed after 2015 (the contemporary genomic era) is a wise one to reduce historical bias.
• Critical Concerns: Cohort Definition (The "Gene-unmutated" Group): This is the most significant methodological flaw. The authors define the comparator group as the "Gene-unmutated cohort" with "No documented actionable genomic alteration." However, they later acknowledge in the discussion that this group "likely included patients who were never genomically profiled." This is not just a limitation; it fundamentally misrepresents the comparator. This cohort is not a "wild-type" or "unmutated" group; it is a "genomically untested or undocumented" group. This distinction is crucial. The association being measured is not "AGA-positive vs. wild-type," but rather "documented AGA-positive vs. not documented." This conflates tumor biology with testing behavior and healthcare access. A more accurate and honest label for Cohort 1 would be "No Documented AGA (including untested patients)." This distinction must be made unequivocally throughout the manuscript, starting with the title and abstract. Matching Covariates: While including "surgical resection" and "chemotherapy history" is a step in the right direction, it is a relatively blunt instrument for a surgical oncologist. There is no granularity regarding the quality of the surgery (e.g., R0 vs. R1 resection, extent of lymphadenectomy) or the typeof chemotherapy (e.g., gemcitabine-based vs. targeted therapies). For a study aiming to isolate biological impact, the failure to match for margin status (R0 resection) is a major limitation, as R0 status is one of the most powerful prognostic factors in GBC and is likely correlated with the decision to perform genomic testing. AGA Panel Heterogeneity: The selected gene panel (KRAS, TP53, ERBB2, IDH1, FGFR1, PIK3CA, ARID1A) is a reasonable starting point. However, treating all "actionable" alterations as a single, monolithic entity is problematic. The biology, therapeutic implications, and prognostic impact of a KRAS mutation are vastly different from an IDH1 mutation or an ERBB2amplification. The authors acknowledge this in their exploratory subgroup analysis, but the primary analysis, which drives the main conclusion, lumps them all together. This risks oversimplifying a complex molecular landscape.

Response:

We thank the reviewer for this fundamental critique regarding the potential conflation of tumor biology with testing behavior. We agree that the comparison cohort cannot be definitively labeled "wild-type."

 

Terminology: We have revised the title, abstract, methods, discussion and conclusions to include "Documented" and renamed the comparator cohort to "No Documented AGA (including untested patients)" throughout the manuscript and figures to ensure transparency.

Surgical Granularity: We have explicitly added the lack of R0/R1 margin status and lymph node yield as a limitation the limitations section, acknowledging these as powerful prognostic factors not captured by the TriNetX platform.

Heterogeneity: We have revised the Discussion (Paragraph 5) to acknowledge that grouping these mutations into a monolithic "actionable" panel is a simplification, framing our subgroup analysis as an essential first step in parsing this molecular complexity.

Results. The results are presented clearly, and the disparity in demographic documentation (51.0% unknown race in the comparator group vs. 3.7% in the AGA group) is a striking and important finding that underscores a major selection bias. Model 1 vs. Model 2: The observation that the absolute risk difference (RD) increased from 10.3% in Model 1 to 13.1% in Model 2 is counterintuitive. One would expect that controlling for stage and treatment (confounders on the causal pathway) would attenuate the effect size, not increase it. The authors do not comment on this. This paradoxical finding could suggest that the matching in Model 2, while statistically balanced, inadvertently selected for a sicker subgroup within the AGA-positive cohort, or that unmeasured confounders (like R0 status) are at play. This warrants a deeper discussion. Survival Analysis and Proportional Hazards: The authors correctly identify the violation of the proportional hazards assumption in Model 2 and appropriately pivot to absolute risk measures. This is a sophisticated and correct analytical decision. The interpretation that early interventions may stabilize survival before underlying biology drives late divergence is plausible and insightful. However, the Kaplan-Meier curves in Figure 2b for Model 2 do not visually support the narrative of a "late accelerator" effect; the curves appear to overlap significantly for the entire duration, with the p-value for the log-rank test being non-significant (0.925). The authors rely on the risk difference (56.2% vs. 43.0%) as the primary evidence, but the visual representation of survival over time should align with this interpretation. Exploratory Subgroup Analyses: The analysis of KRAS and TP53 is interesting but must be interpreted with extreme caution, as the authors do. However, comparing the KRAS-mutated group (N=?) to a matched cohort without a documented KRAS alteration suffers from the same fundamental flaw as the main analysis. Furthermore, performing a survival analysis on a subgroup derived from a previously matched cohort (Model 2) can introduce statistical artifacts. This section should be clearly labeled as hypothesis-generating only and should not be over-interpreted.

Responses:

Comment on Paradoxical Risk Difference (10.3% to 13.2%): We thank the reviewer for this insightful observation. We agree that the increase in effect size after more restrictive matching is noteworthy. We have added a dedicated paragraph to the Discussion (Paragraph 3) exploring this. We hypothesize that when clinical confounders (stage/treatment) are balanced, the independent prognostic weight of genomic status becomes more apparent, potentially identifying a subset of patients less responsive to standard interventions. We also acknowledge this may reflect a "documentation bias" where patients with more aggressive clinical courses are more likely to undergo comprehensive testing.

Comment on Survival Curves (Figure 2b): We agree with the reviewer’s assessment that the curves in Figure 2b exhibit significant overlap (p=0.925). We have revised Section 3.4 (Results) and Paragraph 5 (Discussion) to remove the "late accelerator" narrative and instead provide a more accurate interpretation: the early overlap likely reflects the stabilizing effect of initial surgical and systemic interventions (which were balanced in Model 2), while the significant difference in final cumulative mortality (56.2% vs. 43.0%) reflects the long-term biological impact. We have also updated the Figure 2b legend to clarify that the findings rest on absolute risk due to the statistically significant violation of proportional hazards (p=0.002).

Comment on exploratory subgroup analyses (KRAS/TP53): We agree. We have revised the Results section and paragraph 6 (Discussion). We have maintained the KRAS subgroup analysis but have moved it to a strictly "hypothesis-generating" context. We have added cautionary language in the Results and Discussion (Paragraph 6) acknowledging the potential for statistical artifacts when performing subgroup analyses on a matched population and emphasizing that these comparisons suffer from the same "documentation" limitations as the primary analysis.

Discussion and Conclusions. The discussion is well-written and balanced. The authors acknowledge the major limitations, including the problematic definition of the comparator cohort, the retrospective design, and the lack of granular treatment data. The emphasis on disparities in testing access is a valuable and relevant addition to the literature. The conclusion states that "documented actionable genomic alterations... were associated with higher mortality." While this is true based on their analysis, the more accurate conclusion is that "in this real-world dataset, patients with a documented AGA had worse outcomes, even after accounting for stage and treatment, highlighting both the potential prognostic significance of these alterations and the profound inequities in who receives genomic testing." The current conclusion slightly overstates the evidence regarding the direct biological impact, given the comparator group issue.

Response:

We completely agree with the reviewer’s perspective. We have revised the Conclusion to reflect that our study compares patients with a "documented AGA" to those "without documented alterations" in a real-world setting. This revised wording, largely based on the reviewer's helpful suggestion, explicitly highlights that our findings represent both the potential prognostic significance of these alterations and the systemic inequities in who receives genomic documentation. We believe this shift significantly strengthens the scientific and clinical accuracy of our final statement.

Constructive Feedback for Authors

Dear Authors,

Thank you for the opportunity to review your interesting and timely manuscript. You have tackled a critically important question regarding the independent prognostic value of molecular profiling in GBC, a disease where real-world evidence is sorely needed. The scale of your analysis and your careful methodological approach are commendable. However, several key issues need to be addressed to strengthen the manuscript and ensure its findings are interpreted correctly.

1. Reframe the Comparator Cohort: This is the most critical revision. Please rename the comparator group throughout the manuscript (title, abstract, methods, results, discussion) from "gene-unmutated" to something more accurate, such as "no documented AGA (including untested)." The current terminology is misleading and implies a biological certainty that does not exist. Your study is comparing patients with a documented test result against those without. This distinction is fundamental to the validity of your conclusions.

Response:

We have replaced the terms "gene-unmutated" and "wild-type" with "no documented AGA" throughout the entire manuscript, including the Title, Abstract, Methods, Results, Discussion, and Figure/Table legends. This more accurately reflects the nature of EHR-derived data and avoids implying biological certainty for patients who may not have undergone testing.

1. Clarify the Abstract and Conclusions: The abstract and conclusion sections must be revised to reflect the accurate definition of the comparator group. Instead of stating that the "AGA cohort had higher mortality," it would be more precise to state that "patients with a documented AGA experienced higher mortality compared to those without documented alterations, even after rigorous matching." This subtle change more accurately represents your study design and findings.

Response:

We have revised the Abstract and Conclusion using the reviewer’s suggested phrasing. We now explicitly state that patients with a documented AGA experienced higher mortality compared to those without documented alterations. This clarifies the study design and aligns with the revised terminology mentioned above.

1. Address the Paradoxical Risk Difference: Please provide a commentary in the discussion on why the absolute risk difference increased from Model 1 (10.3%) to Model 2 (13.1%) after controlling for stage and treatment. This is a non-intuitive finding that readers will likely question. Was there a difference in the composition of the cohorts before matching for Model 2? Could this be explained by a higher proportion of patients with certain high-risk AGAs (like KRAS) being more likely to have complete oncologic data documented? Exploring this will demonstrate a deeper engagement with your data.

Response:

This is a very insightful point. We have added a dedicated paragraph to the Discussion (Paragraph 3) exploring this observation. We hypothesize that after clinical confounders (stage and treatment) are balanced, the independent prognostic weight of genomic alterations becomes more apparent. We also discuss the possibility that this reflects a "documentation bias" where patients with more aggressive disease trajectories are more likely to have complete oncologic data and molecular profiling documented in the EHR.

1. Acknowledge the Bluntness of Treatment Matching: Please expand on the limitations of using binary variables for surgical resection and chemotherapy. Acknowledge that you were unable to match for margin status (R0 vs. R1/2), the extent of lymphadenectomy, or specific chemotherapy regimens. Given that surgical quality is a dominant prognostic factor in GBC, this is a significant unmeasured confounder that should be prominently discussed.

Response:

We agree that surgical quality is a critical prognostic factor in GBC. We have significantly expanded the Limitations section to acknowledge the lack of granularity regarding R0/R1 margin status, lymph node yield, and specific chemotherapy regimens. We have explicitly stated that unmeasured surgical bias remains a potential constraint of the TriNetX platform.

1. Re-evaluate the KRAS Subgroup Analysis: I recommend removing the formal survival analysis comparing a KRAS-mutated group to a "no documented KRAS" group from the main results section. Given the issues with the main comparator cohort, this subgroup analysis is highly susceptible to bias. If you wish to keep it, it must be moved to a supplementary section and prefaced with even stronger cautionary language about its exploratory nature and the inherent limitations of the comparator definition.

Response:

We appreciate this caution. Given that our KRAS cohort is quite large (N=563), we have retained it in the main results but have completely reframed Section 3.5 as ". Hypothesis-Generating Subgroup Observations." We have added prominent cautionary language acknowledging the "no documented alteration" comparator flaw and the potential for statistical artifacts when performing subgroup analyses on a previously matched population.

1. Improve Visual Alignment: Please ensure that the narrative regarding the Model 2 survival curves (a late divergence in risk) is visually supported by the graph in Figure 2b. If the curves truly overlap for the entire follow-up period with a non-significant log-rank test, it may be more accurate to state that the survival difference was not statistically significant in the time-to-event analysis and that the primary finding rests on the difference in cumulative mortality risk at the end of follow-up.

Response:

We have revised the Results section and the Figure 2b legend to explicitly state that the log-rank test was not statistically significant in Model 2. We have shifted the primary evidence to the absolute cumulative mortality risk difference, which we justify by citing the statistically significant violation of the proportional hazards assumption (p=0.002). We have also added a clinical interpretation in the Discussion regarding why these curves may overlap during early follow-up (initial treatment stabilization) before diverging later.

Your work has the potential to make a valuable contribution to the field, particularly by highlighting disparities in genomic testing. Addressing these points will significantly enhance the manuscript's clarity, accuracy, and impact.

 

Comments on the Quality of English Language

Overall, the manuscript is written in a clear, professional, and largely idiomatic scientific English. The syntax, terminology, and flow are appropriate for a journal of Cancers’ scope. However, several recurring issues and minor errors detract from the overall polish and should be corrected during revision.

Typographical and formatting errors:

• The abbreviation “AGA” is introduced in the abstract and used frequently, but in the introduction it is spelled out again; standardization is needed.
• The CONSORT-like figure description refers to “Figure 1” in the methods, but the figure is placed after the methods with a caption; the citation in text should be checked for accuracy.

Response:

Standardized "AGA" usage and defined it at first mention in the Introduction.

The CONSORT figure 1 was moved to the methods sections as suggested.

Redundancy and wordiness:

• Some sentences are overly long and could be split for clarity. For example, in the introduction: “Because a large proportion of patients remain asymptomatic during early disease, diagnosis frequently occurs at advanced stages, necessitating aggressive multidisciplinary care.” This is acceptable, but elsewhere, such as in the discussion, there are sentences that combine multiple ideas (e.g., “In this large real-world analysis of patients with gallbladder cancer, documented actionable genomic alterations were associated with higher mortality, and this association persisted after matching for major demographic, stage-related, and treatment-related variables.”). Streamlining would improve readability.

Response:

We thank the reviewer for this constructive feedback. We have performed a comprehensive "clarity edit" of the manuscript, focusing on breaking down complex, multi-clause sentences into shorter, punchier statements. We have specifically addressed the examples provided by the reviewer in the Introduction and Discussion.

Precision and consistency:

• The term “gene-unmutated” is used throughout the methods and results, but as noted in the scientific critique, this is a misnomer. While this is a conceptual issue, the language itself should be revised to “cohort without a documented AGA” to avoid misleading terminology.
• In Table 1, the column headers “Gene-unmutated” and “Gene-mutated” are used; these should be changed to align with the corrected terminology.
• Some abbreviations are used without definition (e.g., “HCOs” in the methods is defined, but “TNX, UMLS” in the genomic status section are mentioned without explanation; consider clarifying or removing if not essential).
• In the results section, “p<0.001” sometimes has a space, sometimes not; consistency is needed.

 

Response:

We have standardized the terminology and formatting throughout the manuscript to ensure precision and consistency.

• We have removed the term "gene-unmutated". We have replaced it with "no documented AGA" or "cohort without a documented AGA" in the text, Table 1 headers, and all figure legends.
• We have defined technical acronyms at first mention, specifically TNX (TriNetX) and UMLS (Unified Medical Language System). Non-essential jargon has been removed to improve accessibility.
• We have standardized p-value formatting to ensure a consistent space between the variable and the operator (e.g., p<0.001) throughout the Results and Tables.

Grammar and syntax:

• Minor subject–verb agreement issues: “The proportional hazards assumption was violated (χ² = 9.979, p = 0.002), indicating a non-constant effect over time. Accordingly, for Model 2, absolute mortality risk and risk difference were considered the primary effect measures As shown in the survival curves...” – there is a missing period and a run-on sentence.
• In the discussion: “By contrast, the TP53 subgroup analysis was limited by small sample size and should not be interpreted as evidence of lack of prognostic relevance.” This is grammatically correct but slightly awkward; consider “...should not be interpreted as evidence that TP53 lacks prognostic relevance.”
• In the conclusion: “These findings support the prognostic relevance of molecular profiling in GBC and reinforce the importance of improving equitable access to genomic testing and documentation.” – the phrase “improving equitable access” is somewhat redundant; “ensuring equitable access” would be more natural.

Response:

Thank you! We have revised and corrected the run-on sentence regarding the proportional hazards assumption violation.

We have the revised the TP53 discussion sentence and corrected as suggested. Thank you. It now reads more clear.

We have also revised and improved the equitable access phase in the conclusion as advised. Thank you!

Figure legends and in-text references:

• The legend for Figure 2 uses “A. GBC documented AGA Vs. no documented AGA. Model 1.” The use of “Vs.” is informal; “versus” or a more descriptive phrase would be better.
• In the text, the forest plot (Figure 3) is mentioned but not fully described; the legend is minimal. Ensure all figures are referenced sequentially and have complete, self-explanatory legends.

Response:

We have formalized and expanded our figure legends to ensure they are self-explanatory and maintain a professional tone.

Figure 2: We have replaced the informal "Vs." with "versus" and added more descriptive subtitles for Panel A and Panel B.

Figure 3: The legend for the forest plot has been significantly expanded to describe the statistical measures (Hazard Ratios and 95% CIs) and the subgroups represented.

We have verified that all figures are referenced in numerical order within the text, and we have confirmed that Figure 1 is correctly cited in the Methods section.

We have updated the legend for Figure 1 to be fully self-explanatory, detailing the attrition process and the specific variables used for each propensity score-matched (PSM) model. We have also verified that the cohort numbers (N) in the text match the final Consort diagram exactly and that the figure is cited sequentially within the Methods section where the matching strategy is first described.

We have updated Table 1 to align with the revised terminology. The headers "Gene-unmutated" and "Gene-mutated" have been replaced with "No Documented AGA" and "≥1 Documented AGA", respectively. Additionally, we have ensured that all demographic and oncologic characteristics are presented with consistent decimal places and that the legend clearly defines the populations represented in each model.

We have updated Table 2 accordingly as well as its legend.

Recommendation. The English is generally competent and does not impede understanding, but a careful copyedit to correct typographical errors, standardize terminology, and improve sentence flow is recommended. I would classify the language quality as needing minor to moderate revision.

Response:

Thank you. We have conducted a copyedit review of the manuscript and revised as suggested.

 

 

Reviewer 2 Report

Comments and Suggestions for Authors

The manuscript by Solangi et al analyzed the association between documented actionable genomic alterations (AGA) status and their co-relation with the clinic-pathological parameters and overall survival (OS) in GBC. The strength of the study is the large sample size used for the analysis, however, as similar analysis has been done earlier, the significance of the outcome is limited.

1. The term ‘without a documented actionable genomic alteration’ suggests as if there is no information available regarding the genomic alterations. Ther term should be changed ‘No genomic alterations’ for better clarity.
2. The supplementary table should provide the details of the genomic alterations, age, gender, TNM, Stage, chemotherapy etc.
3. The spread of each genomic alterations should be provided as it will provide an information regarding the prevalence of genomic alterations observed in the patient cohort.
4. Discussion section: The results are not compared with the previous studies. The corelation with the previous studies may be provided in the Tabular form.
5. It would be interesting to study the corelation of the genomic alterations with the chemoresistant vs chemosensitive cases.

Author Response

Reviewer 2

 

Review Report Form 

Open Review

(x) I would not like to sign my review report 
( ) I would like to sign my review report 

Quality of English Language

( ) The English could be improved to more clearly express the research. 
(x) The English is fine and does not require any improvement. 

 

	Yes	Can be improved	Must be improved	Not applicable
Does the introduction provide sufficient background and include all relevant references?	(x)	( )	( )	( )
Is the research design appropriate?	( )	(x)	( )	( )
Are the methods adequately described?	( )	(x)	( )	( )
Are the results clearly presented?	( )	( )	(x)	( )
Are the conclusions supported by the results?	( )	( )	( )	( )
Are all figures and tables clear and well-presented?	( )	( )	(x)	( )

Comments and Suggestions for Authors

The manuscript by Solangi et al analyzed the association between documented actionable genomic alterations (AGA) status and their co-relation with the clinic-pathological parameters and overall survival (OS) in GBC. The strength of the study is the large sample size used for the analysis, however, as similar analysis has been done earlier, the significance of the outcome is limited.

1. The term ‘without a documented actionable genomic alteration’ suggests as if there is no information available regarding the genomic alterations. Ther term should be changed ‘No genomic alterations’ for better clarity.

 

Response:

 

We appreciate the reviewer's focus on clarity. However, in accordance with the suggestions of Reviewer 1 and the inherent limitations of real-world electronic health record (EHR) data, we have opted to use the term "no documented AGA." Because the TriNetX database includes patients who may not have undergone sequencing, labeling them as having "no genomic alterations" would imply a confirmed wild-type status that we cannot verify. We have added a clarifying statement in the Methods (Section 2.2) and Limitations to explicitly define this cohort as a mix of wild-type and untested patients.

 

1. The supplementary table should provide the details of the genomic alterations, age, gender, TNM, Stage, chemotherapy etc.
2. The spread of each genomic alterations should be provided as it will provide an information regarding the prevalence of genomic alterations observed in the patient cohort.

 

Response:

 

We thank the reviewer for this insightful suggestion. We agree that understanding the clinical landscape of individual genomic drivers is essential for contextualizing the pooled results.

 

In response, we have added Supplementary Table 1, which provides a granular breakdown of the demographic characteristics (age, sex, and race/ethnicity) for the three most prevalent genomic subgroups: KRAS (N=568), TP53 (N=34), and ERBB2 (N=17).

 

Our analysis revealed several notable trends:

 

The TP53 subgroup exhibited a higher proportion of Asian (29.4%) and Hispanic/Latino (29.4%) patients compared to the KRAS group.

 

The ERBB2 and TP53 subgroups showed a higher female predominance (88.2% and 76.5%, respectively) compared to the KRAS cohort (58.8%).

 

Regarding clinical staging and treatment, these characteristics were consistently balanced across the KRAS and TP53 groups. However, for the ERBB2 subgroup and other rare alterations (e.g., PIK3CA, ARID1A, IDH1, FGFR1), specific clinical frequencies were suppressed by the TriNetX platform's privacy protocols to ensure patient de-identification for cohorts with small sample sizes.

 

This inherent limitation of individual subgroup analysis, specifically the rapid decrease in statistical power and data availability for non-KRAS alterations, emphasizes our methodological decision to utilize a pooled AGA-positive cohort for the primary survival analysis. This approach ensured a robust and statistically valid comparison against the matched wild-type population while accounting for the diverse genomic landscape of gallbladder cancer.

 

 

1. Discussion section: The results are not compared with the previous studies. The corelation with the previous studies may be provided in the Tabular form.

 

Response:

 

We have expanded the Discussion to contextualize our findings within the existing literature. As requested, we have added Table 3 to the Discussion, comparing our mortality and hazard ratio findings with key landmark studies in GBC genomics.

 

1. It would be interesting to study the corelation of the genomic alterations with the chemoresistant vs chemosensitive cases.

 

Response:

While our dataset does not allow for a direct assessment of chemoresistance versus chemosensitivity, the persistence of a 13.2% mortality risk difference after matching for chemotherapy history (Model 2) suggests that patients with documented AGAs may derive less benefit from standard cytotoxic regimens. This potentially indicates a phenotype of relative chemoresistance driven by oncogenic pathways such as KRAS signaling.

 

 

Reviewer 3 Report

Comments and Suggestions for Authors

Solangi et. al. analyzed gallbladder cancer patients to see if having actionable genomic alterations affects survival. Using real world data and carefully matched models to account for demographics, disease stage, and treatments, patients with documented AGAs showed higher mortality and slightly shorter overall survival than those without AGAs. KRAS mutations were linked to particularly poor outcomes. The study highlights that genomic profiling has prognostic value in GBC and points to disparities in access to genomic testing.

I have some concerns regarding this manuscript:

1. KRAS alterations are associated with significantly worse survival, suggesting potential clinical relevance. The manuscript should clarify whether KRAS effects are independent of co-occurring mutations or other clinical factors.
2. The TP53 analysis are limited by very small sample size and wide confidence intervals, which reduces power. The authors appropriately describe these results as exploratory, but explicit caution regarding interpretation is recommended; additionally, the manuscript should note that these findings are hypothesis generating and require validation in independent cohorts.

Author Response

Reviewer 3

 

Open Review

(x) I would not like to sign my review report 
( ) I would like to sign my review report 

Quality of English Language

( ) The English could be improved to more clearly express the research. 
(x) The English is fine and does not require any improvement. 

 

	Yes	Can be improved	Must be improved	Not applicable
Does the introduction provide sufficient background and include all relevant references?	( )	( )	(x)	( )
Is the research design appropriate?	( )	( )	(x)	( )
Are the methods adequately described?	( )	( )	(x)	( )
Are the results clearly presented?	( )	( )	(x)	( )
Are the conclusions supported by the results?	( )	( )	(x)	( )
Are all figures and tables clear and well-presented?	( )	( )	(x)	( )

Comments and Suggestions for Authors

Solangi et. al. analyzed gallbladder cancer patients to see if having actionable genomic alterations affects survival. Using real world data and carefully matched models to account for demographics, disease stage, and treatments, patients with documented AGAs showed higher mortality and slightly shorter overall survival than those without AGAs. KRAS mutations were linked to particularly poor outcomes. The study highlights that genomic profiling has prognostic value in GBC and points to disparities in access to genomic testing.

I have some concerns regarding this manuscript:

1. KRAS alterations are associated with significantly worse survival, suggesting potential clinical relevance. The manuscript should clarify whether KRAS effects are independent of co-occurring mutations or other clinical factors.

 

Response:
We thank the reviewer for this insightful observation regarding the clinical significance of KRAS. We agree that the interplay between KRAS and concurrent mutations is a critical factor in understanding tumor aggression.

 

Clarification of Independence: We have updated the Discussion (Paragraph 5) to clarify that while co-occurring mutations are present, the adverse survival signal in our cohort appears largely driven by the intrinsic aggression of RAS-pathway activation.

 

We have strengthened this argument by citing https://doi.org/10.64898/2026.03.29.26349383 (2026), whose recent multi-cohort genomic modeling demonstrated that KRAS alterations maintain a distinct and independent prognostic weight even when accounting for concurrent TP53 status.

 

Furthermore, our matched survival analysis (Model 2) specifically adjusted for key clinical factors including metastatic stage and treatment history, suggesting that the KRAS effect is not just a surrogate for advanced clinical presentation at baseline.

 

1. The TP53 analysis are limited by very small sample size and wide confidence intervals, which reduces power. The authors appropriately describe these results as exploratory, but explicit caution regarding interpretation is recommended; additionally, the manuscript should note that these findings are hypothesis generating and require validation in independent cohorts.

 

Response:
We appreciate the reviewer's caution regarding the TP53 subgroup analysis. We acknowledge that the rarity of this malignancy, combined with the limitations of real-world EMR documentation, resulted in a smaller sample size (N=34) and wider confidence intervals for this specific driver. We have updated the discussion including cautionary language and added citations of recent studies that provide the necess

Round 2

Reviewer 2 Report

Comments and Suggestions for Authors

The manuscript is accepted for publication.

Reviewer 3 Report

Comments and Suggestions for Authors

The authors have satisfactorily addressed my queries.