Impact of Antidiabetic Medication on Therapy Outcomes in Metastatic Urothelial Cancer Patients Receiving Enfortumab Vedotin Monotherapy

Round 1

Reviewer 1 Report

Comments and Suggestions for Authors

In their manuscript entitled “Impact of antidiabetic medication on therapy outcomes in metastatic urothelial cancer patients receiving enfortumab vedotin monotherapy”, Schneidewind et al examine the effect of diabetes medications on response rate to enfortumab vedotin (EV). The purpose of the study was to determine the effect of glitazones on the efficacy of EV, however the study was limited by the lack of use of glitazone therapy in the modern management of diabetes. The authors pivoted their study to examine alternative diabetes medications. In this manuscript, they examine the ORR at 3 months, 6 months and the CSS between patients on various diabetes medications.

The authors used a dataset of 125 patients curated from 11 medical centers. Of these patients, 24 had a diagnosis of diabetes (19.2%), 10 were using insulin (8%) and 9 were on metformin (7.2%). 16 of these patients had UTUC. These are very small sample sizes, particularly the metformin group which drives the largest claim made in the manuscript. Overall, I have several concerns regarding the comparisons made between groups.

In the methods, the authors describe a series of statistical tests, but apart from the use of log rank testing for median CSS, there is not a clear description of which test was used for which analysis. The methods should clearly state how the various statistical testing was utilized for the claims made in the paper.

The results begin with an assessment of the demographics of the patients. This includes a statistical comparison between the demographics of the UTUC and the bladder cancer cohorts. It is unclear what the purpose of this comparison is, as there is no hypothesis about expected differences, and the age, gender, metastatic rate, performance status and therapy line are not provided for UTUC until much later in the manuscript. Should the authors wish to better demonstrate the similarities between the groups, a table of the demographics broken down by subtype would be more effective.

The authors next describe the ORR of the patients on diabetes medications. The description of the primary result of ORR is confusing and requires clarification. ORR refers to the percentage of the population that has a partial or complete response, and is not a binary variable, yet the authors treat it as such in their comparisons. For example, in the sentence at line 151 the authors state “After 3 months of therapy 77 patients (61.6%) had an ORR and 50 patients (40.0%) 151 after 6 months”. By definition ORR is a rate, and not something a patient can have. This should state that 61.6% is the ORR, and 77 patients had a response. This directly impacts table 1, in which the comparison of ‘ORR no versus yes’ is not correct. I believe the comparison is being made between responders and non-responders. This needs correction or clarification. Additionally, the table should include the number of patients given the very small sample size to assist in better interpretation.

Regarding cancer specific survival, the authors must clarify how they are assessing this variable.  At line 152, the statement “Fifty-one 152 patients (40.8%) have already died due to their malignant disease” needs context as to when in their treatment course this is being assessed. In table 1, there is no clear explanation as to when the cancer specific mortality is being assessed. If this is a comparison at a particular time point (3 months? 6 months?) then a portion of patients that are alive may be a reasonable comparison. Without the context of how this is being assessed, a reader cannot interpret whether the reported difference in CS mortality is meaningful.  

In the analysis of UTUC, the authors claim there is a significant association between metformin use and CSS. This is based on analysis of a single patient that received metformin. This is not a reasonable statistical comparison to make.

In the discussion, the authors rightfully mention studies that show potential benefit of metformin, and state their results directly conflict with these studies. The small sample size in this manuscript is not sufficient to contradict the cited studies. The authors’ explanation of the confounding effect of hyperglycemia and insulin use for EV toxicity is well stated and important to include.

Regarding the citations, EV-201 is not cited in the introduction despite being mentioned, and is only cited much later in the discussion. This citation should be moved up. The authors also fail to cite the practice changing EV-301 study of EV+pembrolizumab in untreated bladder cancer. This study should be included in portion of the introduction that references the new expanded use of EV in management of bladder cancer.

Overall, the manuscript needs substantial revision. Revisions should focus on explanation of the statistical methods and clarification of comparisons. I would recommend expanding the sample size substantially if able to better support the claims made in the manuscript.

Author Response

In their manuscript entitled “Impact of antidiabetic medication on therapy outcomes in metastatic urothelial cancer patients receiving enfortumab vedotin monotherapy”, Schneidewind et al examine the effect of diabetes medications on response rate to enfortumab vedotin (EV). The purpose of the study was to determine the effect of glitazones on the efficacy of EV, however the study was limited by the lack of use of glitazone therapy in the modern management of diabetes. The authors pivoted their study to examine alternative diabetes medications. In this manuscript, they examine the ORR at 3 months, 6 months and the CSS between patients on various diabetes medications. The authors used a dataset of 125 patients curated from 11 medical centers. Of these patients, 24 had a diagnosis of diabetes (19.2%), 10 were using insulin (8%) and 9 were on metformin (7.2%). 16 of these patients had UTUC. These are very small sample sizes, particularly the metformin group which drives the largest claim made in the manuscript. Overall, I have several concerns regarding the comparisons made between groups.

Thank you for your valuable time and efforts to improve our work. Furthermore, we must state that our work is mainly limited due to the small sample sizes, but we have discussed that issue in the limitations section of our discussion.

In the methods, the authors describe a series of statistical tests, but apart from the use of log rank testing for median CSS, there is not a clear description of which test was used for which analysis. The methods should clearly state how the various statistical testing was utilized for the claims made in the paper.

Thank you, this is certainly true. We tried to make it clear which variable required which test in the statistics section and have highlighted this again.

The results begin with an assessment of the demographics of the patients. This includes a statistical comparison between the demographics of the UTUC and the bladder cancer cohorts. It is unclear what the purpose of this comparison is, as there is no hypothesis about expected differences, and the age, gender, metastatic rate, performance status and therapy line are not provided for UTUC until much later in the manuscript. Should the authors wish to better demonstrate the similarities between the groups, a table of the demographics broken down by subtype would be more effective.

Thanks, we just mentioned this in the text because we only wanted to show that the groups are not significantly different in major demographic characteristics. Furthermore, we state that UTUC had significant higher T stages, which is important, at least in our opinion, to justify a comprehensive subgroup analysis.

The authors next describe the ORR of the patients on diabetes medications. The description of the primary result of ORR is confusing and requires clarification. ORR refers to the percentage of the population that has a partial or complete response, and is not a binary variable, yet the authors treat it as such in their comparisons. For example, in the sentence at line 151 the authors state “After 3 months of therapy 77 patients (61.6%) had an ORR and 50 patients (40.0%) 151 after 6 months”. By definition ORR is a rate, and not something a patient can have. This should state that 61.6% is the ORR, and 77 patients had a response. This directly impacts table 1, in which the comparison of ‘ORR no versus yes’ is not correct. I believe the comparison is being made between responders and non-responders. This needs correction or clarification. Additionally, the table should include the number of patients given the very small sample size to assist in better interpretation.

Thank you and we want to make this clear: Overall response rate (ORR) is defined as the proportion of patients who have a partial or complete response to therapy; it does not include stable disease and is a direct measure of drug tumoricidal activity. We use this definition correctly and it is correct that one patient can not have ORR, but one patient is accounted in having ORR or not. We gave the correct rates in table 1, but yes and no means for the diabetic parameter – we tried to make this clear in the table.

Regarding cancer specific survival, the authors must clarify how they are assessing this variable.  At line 152, the statement “Fifty-one 152 patients (40.8%) have already died due to their malignant disease” needs context as to when in their treatment course this is being assessed. In table 1, there is no clear explanation as to when the cancer specific mortality is being assessed. If this is a comparison at a particular time point (3 months? 6 months?) then a portion of patients that are alive may be a reasonable comparison. Without the context of how this is being assessed, a reader cannot interpret whether the reported difference in CS mortality is meaningful.  

Thank you, this is a very good point. We used the correct definition of CSS, also given in the methods section: CSS is defined as the time from randomization or treatment initiation to patient death caused by the index cancer, whether due to the original tumor or to a second primary of the same cancer type, but you are correct that you need to give time point of assessment or the follow up time, so we have added this information to the text.

In the analysis of UTUC, the authors claim there is a significant association between metformin use and CSS. This is based on analysis of a single patient that received metformin. This is not a reasonable statistical comparison to make.

This is a very good point and it is certainly true that the main limitation of our work is the small sample size as well as the low event rate, we have therefore expanded our limitation section.

In the discussion, the authors rightfully mention studies that show potential benefit of metformin, and state their results directly conflict with these studies. The small sample size in this manuscript is not sufficient to contradict the cited studies. The authors’ explanation of the confounding effect of hyperglycemia and insulin use for EV toxicity is well stated and important to include.

Thank you, as mentioned above, we have expanded our limitation section.

Regarding the citations, EV-201 is not cited in the introduction despite being mentioned, and is only cited much later in the discussion. This citation should be moved up. The authors also fail to cite the practice changing EV-301 study of EV + pembrolizumab in untreated bladder cancer. This study should be included in portion of the introduction that references the new expanded use of EV in management of bladder cancer.

Correct, we have made the requested changes and added the EV-301 study also to the introduction section.

Overall, the manuscript needs substantial revision. Revisions should focus on explanation of the statistical methods and clarification of comparisons. I would recommend expanding the sample size substantially if able to better support the claims made in the manuscript.

Thanks a lot again for your valuable time and efforts to improve our manuscript and even our research work. We highly appreciate this. We tried to make more clear definition and expanded our limitation section according to your true concern of small sample sizes.

 

Reviewer 2 Report

Comments and Suggestions for Authors

This is a novel angle nvestigating the effect of antidiabetic medications—particularly metformin—on clinical outcomes in patients with metastatic urothelial carcinoma. 

Given the recent shift to first-line EV + pembrolizumab regimens, evaluating how comorbidities like diabetes affect outcomes is extremely relevant.

• The retrospective design does not account for critical variables:: Glycemic control (HbA1c, glucose levels), duration and dose of metformin, other comorbidities, renal function, BMI, corticosteroid use. Even with retrospective data, some of these should be collected and adjusted for in multivariable analyses. Is it possibile? or at least acknowledged more systematically I would say. 
• “metformin use may be linked to poorer survival”… I suggest to temper these conclusions due to premature data without adjusting for confounding by indication. In addition, subgroup analysis on UTUC (n=16) and patients receiving metformin (n=9) is too underpowered to support strong conclusions. Please, clearly acknowledge these are exploratory findings. 
• While the PPAR-γ and Nectin-4 link is intriguing, it’s currently speculative and lacks direct data in this study. Move speculative pathways to a “Future Directions” section or clearly label them as hypotheses.
• Looking to UTUC subgroup, the topic must also be contextualized in light of the considerable heterogeneity in clinical practice guidelines. As shown in a recent systematic comparison, substantial differences exist in recommendations and management across guidelines. This lack of standardization may contribute to variability in treatment responses and highlights the need for further research tailored specifically to this population. Please state this ad discuss: doi: 10.3390/cancers16061115
• Dicussion: In your UTUC population, the interpretation of outcomes—particularly the observed association between metformin use and cancer-specific mortality—may be affected by upstream treatment variability. A recent systematic review confirmed that while NAST may improve pathological response, its benefit over adjuvant therapy is still unclear, and patient selection remains challenging DOI: 10.23736/S2724-6051.22.04659-6. Such inconsistency in pre-treatment strategies could have influenced both response to enfortumab vedotin and overall prognosis in this subgroup. Please discuss it also in limitation 
• Please better present table and also replace “patients having/not having the parameter” with “patients with/without [condition]”
• Check typos

Comments on the Quality of English Language

not bad, some minor edit needed

Author Response

This is a novel angle investigating the effect of antidiabetic medications—particularly metformin—on clinical outcomes in patients with metastatic urothelial carcinoma.

Given the recent shift to first-line EV + pembrolizumab regimens, evaluating how comorbidities like diabetes affect outcomes is extremely relevant.

Thank you for your valuable time and efforts to improve our work.

    The retrospective design does not account for critical variables:: Glycemic control (HbA1c, glucose levels), duration and dose of metformin, other comorbidities, renal function, BMI, corticosteroid use. Even with retrospective data, some of these should be collected and adjusted for in multivariable analyses. Is it possible? or at least acknowledged more systematically I would say.

Thank you and this is a very good as well as important point. Unfortunately, due to retrospective study design, we are not able to perform this multivariate analysis, but since you are certainly true, we have added these concerns to our limitations section.

    “metformin use may be linked to poorer survival”… I suggest to temper these conclusions due to premature data without adjusting for confounding by indication. In addition, subgroup analysis on UTUC (n=16) and patients receiving metformin (n=9) is too underpowered to support strong conclusions. Please, clearly acknowledge these are exploratory findings.

Thanks, it is very true that the major limitation of our work are small sample sizes as well as low event rates. That is why, we also we expanded our limitation section and consequently, we tempered down our conclusion.

    While the PPAR-γ and Nectin-4 link is intriguing, it’s currently speculative and lacks direct data in this study. Move speculative pathways to a “Future Directions” section or clearly label them as hypotheses.

Thank you, this is very true and we have labeled it as hypothesis.

    Looking to UTUC subgroup, the topic must also be contextualized in light of the considerable heterogeneity in clinical practice guidelines. As shown in a recent systematic comparison, substantial differences exist in recommendations and management across guidelines. This lack of standardization may contribute to variability in treatment responses and highlights the need for further research tailored specifically to this population. Please state this ad discuss: doi: 10.3390/cancers16061115 [Titel anhand dieser DOI in Citavi-Projekt übernehmen]

Thank you and this is a very important point since we did not discuss the results for UTUC in detail. Therefore, we have added a UTUC section into the discussion.

    Discussion: In your UTUC population, the interpretation of outcomes—particularly the observed association between metformin use and cancer-specific mortality—may be affected by upstream treatment variability. A recent systematic review confirmed that while NAST may improve pathological response, its benefit over adjuvant therapy is still unclear, and patient selection remains challenging DOI: 10.23736/S2724-6051.22.04659-6 [Titel anhand dieser DOI in Citavi-Projekt übernehmen] . Such inconsistency in pre-treatment strategies could have influenced both response to enfortumab vedotin and overall prognosis in this subgroup. Please discuss it also in limitation

Thanks again, as we mentioned above, we introduced a UTUC section into the discussion and expanded our limitation section.

    Please better present table and also replace “patients having/not having the parameter” with “patients with/without [condition]”

Sorry, we have used this wording, because other wording might be even more confusing.

    Check typos

Thank you, we have done that. We highly appreciate your time and valuable discussion to improve our research.

Round 2

Reviewer 1 Report

Comments and Suggestions for Authors

See attached document

Comments for author File: Comments.pdf

Author Response

Dear Editor, Dear Reviewer,

 

Thank you for your valuable time and your thoughtful comments to further improve our manuscript and even our research. We highly appreciate this.

Please, find our comments to your concerns below and we have marked all changes in the manuscript in red color.

Thanks again and Kind Regards,

 

Laila Schneidewind, on behalf of all authors

 

The authors have made no change to the text. The authors should state variables were parametric and which were not and also should state which are continuous and which categorical.Version 2: “For parametric continuous variables the Student t test was used and for parametriccategorical variables the chi-square test or the Fisher exact test was used. For non-parametric data the Mann-Whitney U test was used for categorical and continuous variables, respectively.”Version 1: “For parametric continuous variables the Student t test was used and for parametric categorical variables the chi-square test or the Fisher exact test was used. For non-parametric data the Mann-Whitney U test was used for categorical and continuous variables, respectively”

Thank you, this is certainly true, but as we mentioned in review round 1 we just have highlighted the section to make this clear. In our opinion, it should be clear to a medical doctor what continuous and categorial means for our results. It would make the text or tables more confusing, adding this information every time.

I again recommend a table to better display this comparison.

Thank you, we initially did not insert the table because it was not a study aim and does not provide any further information, but as requested we have inserted the table.

The authors misunderstand my meaning. The literal text they use is incorrect. The sentence reads “After 3 months of therapy 77 patients (61.6%) had an ORR”. If you rewrite this without the parenthesis and abbreviation, this would read “After 3 months of therapy 77 patients had an overallresponse rate”. 77 patients did not have an overall response rate. 77 patients had a response (completeresponse or partial response). The overall response rate is 61.6%. This must be rewritten. Options include “After 3 months of therapy 77 patients had a response (ORR = 61.6%). After 6 months, 50patients had a response (ORR = 40.0%).” or “The ORR was 61.6% at 3 months (77/X patients) and 40.0% at 6 months (50/X patients).” I appreciate the authors additional of the clarification to the table, however diabetic is spelled wrong. The authors also did not add the number of patients in each category to the table, which would greatly assist in interpretation. The authors could also clarify if they are using the best overall response rate or the response rate at the particular point in time. How would a patient who had a partial response at 3 months and stable disease at 6 months be included? Are the authors using best response (i.e the partial response from 3 months) or response at that time point (i.e. stable disease).

Thank you, we used the correct definition of ORR given in the methods section and like stated in the previous review. This is the international correct definition used in oncological studies. If you give the time point, it is the ORR at that time. It is a very fair point and makes our results more clear – thank you for that advise, so we rephrased the sentences with the rates according to your advice.

Thank you for the advise with the spelling error at the table. Sorry, for our error, we have corrected it. Unfortunately, we cannot agree with you to add the patient numbers into the table, because it will make the table more confusing without adding any information, because we have given the correct percentage rates.

The authors added the sentence at line 153 “The median follow-up time was 7 monts (IQR 5.0 – 10.0).”“Months” is spelled incorrectly in this new sentence. This also fails to address my question is what comparison is being used for CSM. It appears the comparison is whether a patient in either group had died of their cancer during the 6 months of analysis. I would recommend clarification of this variable.

Thank you and sorry for our spelling error, we have corrected it. As also stated in the methods section and in the previous review, we used the correct definition of CSS and also compared CSS at the end of follow-up for each patient or death to the index cancer according to the definition. That is why we added follow-up period. What do you mean by CSM?

The authors still state that there is a significant association between metformin medication and UTUC. This is not a reasonable question to attempt to use statistics to answer given the n of 1 in the group ofinterest. The statement that metformin is statistically significant should be removed.

Thank, for the medical interpretation you are totally right. That is why we extended our limitation section so much and toned down our conclusion and you are certainly right that it just might be a bias. Unfortunately, mathematically you are wrong, e. g. you can apply a statistical test with only one event and it can also be not significant. It always a test in mathematical setting for your hypothesis or study aim and then its medical interpretation.

At line 283, the new text reads “Additionally, in your UTUC population”. This is either a typo that was meant to read “Additionally, in our UTUC population” or a comment copied from a reviewer.I appreciate the authors adding EV-301, however there was an error in my comment, I intended tosuggest a reference to EV-302: https://www.nejm.org/doi/full/10.1056/NEJMoa2312117

Alright, we also added this reference to our paper.

I appreciate the authors attempts to address my concerns, however I recommend further modification

Thanks again, for your valuable time and thoughtful comments to improve our work. It helped a lot and we very much appreciated it.

 

 

 

Reviewer 2 Report

Comments and Suggestions for Authors

Overall, good improvement in your revision 

Comments on the Quality of English Language

Good 

Author Response

Dear Editor, Dear Reviewer,

 

Thank you for your valuable time and your thoughtful comments to further improve our manuscript and even our research. We highly appreciate this.

Please, find our comments to your concerns below and we have marked all changes in the manuscript in red color.

Thanks again and Kind Regards,

 

Laila Schneidewind, on behalf of all authors

 

No further comments.

We highly appreciate your time and valuable discussion to improve our research.