Review Reports

Round 1

Reviewer 1 Report

Comments and Suggestions for Authors

This manuscript presents a single-center retrospective comparison of alpelisib plus endocrine therapy (ET) versus everolimus plus ET in patients with PIK3CA-mutant HR+/HER2− metastatic breast cancer after CDK4/6 inhibitor progression. However, several methodological and interpretive issues must be addressed to improve the validity and clarity. For this reason, I suggest significant revisions as presented below.

-Significant disparities exist between treatment groups, especially in:
the number of metastatic sites, the percentage of bone-only disease, and previous treatment options. OS comparisons are likely complicated by these well-known prognostic variables. A multivariable Cox regression analysis that accounts for important baseline variables (such as metastatic burden, bone-only disease, previous chemotherapy, and treatment line) should be included.

-Since ctDNA was unavailable, only tissue-based NGS was utilized.
This limitation is significant given the known spatial and temporal heterogeneity. It is advised to go into more detail about how the lack of ctDNA testing could have resulted in incorrect classification or a delay in the identification of actionable mutations.

Author Response

Reviewer 1

 

First, we thank the reviewer for taking the time to read our manuscript and for the comments.

 

Comment:

‘’ -Significant disparities exist between treatment groups, especially in:
the number of metastatic sites, the percentage of bone-only disease, and previous treatment options. OS comparisons are likely complicated by these well-known prognostic variables. A multivariable Cox regression analysis that accounts for important baseline variables (such as metastatic burden, bone-only disease, previous chemotherapy, and treatment line) should be included. ‘’

Answer:

Indeed, between the two groups, there are imbalances, most striking in the number of metastatic sites and bone-only disease. A multivariate analysis is essential for determining the impact of multiple covariates on the OS. Unfortunately, we did not include a multivariate analysis due to the very small sample size, which precluded the extraction of reliable results. However, after adjusting for baseline prognostic factors, including the number of metastatic sites and visceral disease, the treatment cohort was still not independently associated with OS. Please find explanations between lines 172-180 and 383-389.

 

Comment:

‘’ -Since ctDNA was unavailable, only tissue-based NGS was utilized.
This limitation is significant given the known spatial and temporal heterogeneity. It is advised to go into more detail about how the lack of ctDNA testing could have resulted in incorrect classification or a delay in the identification of actionable mutations. ‘’

Answer:

ctDNA is a valuable tool in metastatic breast cancer. The ESMO and NCCN guidelines currently recommend it for the detection of actionable mutations (e.g., PIK3CA, ESR1, BRCA1/2), while recent studies have demonstrated its role as a guide for treatment selection before radiological progression of the disease. In our country, there are mainly financial impediments to accessing any licensed ctDNA assay. We agree with the reviewer that the lack of ctDNA-derived data is a significant limitation of the study. Utilising ctDNA could enable us to include more patients in the study, as many patients' genomic profiles remain unknown due to unavailable primary tissue or difficulties obtaining a biopsy from the metastatic site. Moreover, we believe that a ctDNA assay could spare patients from receiving chemotherapy after the disease progression on CDK4/6 inhibitors, given its non-invasive nature and the rapid turnaround time for results. Please find the paragraph discussing the above limitations owing to the lack of ctDNA, between lines 390-416. 

 

Reviewer 2 Report

Comments and Suggestions for Authors

The manuscript addresses a highly relevant clinical question: the optimal sequencing of targeted therapies after CDK4/6 inhibitor failure in PIK3CA-mutant breast cancer. Given the lack of prospective randomized head-to-head trials between these agents in the post-CDK4/6 setting, real-world evidence is valuable. However, the manuscript requires major revisions before acceptance.

Comments:

1. The Abstract mentions the comparison with Everolimus in PIK3CA-wild-type patients, but does not report these results in the Abstract Results section. Either remove the mention from the Abstract methods or include the finding in the Abstract results to ensure consistency.
2. Table 1 illustrates substantial baseline imbalances between the treatment groups, which likely bias overall survival in favor of everolimus. Patients in the everolimus arm had fewer metastatic sites (<2 sites: 59.1% vs. 32.5%) and a markedly higher rate of bone-only disease (45.5% vs. 22.5%), both of which are indicators of more indolent disease biology. Although briefly acknowledged in the Conclusion, this critical limitation is insufficiently addressed in the Abstract and Discussion. The observed OS difference (18.3 vs. 9.6 months), therefore, appears more plausibly attributable to baseline prognostic differences rather than treatment effect. A multivariate Cox regression model adjusting for metastatic burden and visceral disease is warranted; if precluded by the small sample size (N = 62), this limitation should be clearly emphasized in the Abstract and early Discussion.
3. In Table 2, the Confidence Intervals are quite wide due to the small sample size. The authors ensure the p-value for the comparison of ORR between the two groups is provided.
4. In the Discussion, the authors stated "Conversely, in terms of median OS, everolimus outperformed alpelisib by nearly doubling the OS... Despite the numerical difference, the result was not statistically significant..." It is scientifically inaccurate to state that one drug "outperformed" the other when the p-value is 0.41. In statistics, a p-value of 0.41 implies the numerical difference could easily be due to chance. The author should rephrase all instances claiming superiority and state clearly that "no statistically significant difference in OS was observed." The numerical trend can be mentioned, but it must be immediately contextualized in relation to the baseline imbalances.
5. The manuscript uses commas for decimals (e.g., "4,9 months", "p=0,53") in the text and tables, but occasionally uses points. Standard English scientific publishing requires the use of points (e.g., "4.9 months"). Please standardize this throughout the text and tables.

Author Response

Reviewer 2

 

First, we thank the reviewer for taking the time to read our manuscript and for the comments.

 

Comment:

‘’ The Abstract mentions the comparison with Everolimus in PIK3CA-wild-type patients, but does not report these results in the Abstract Results section. Either remove the mention from the Abstract methods or include the finding in the Abstract results to ensure consistency. ‘’

Answer:

We included in the Abstract (see Methods and Results) the results of the comparison between PIK3CA-mutant and PIK3CA-wild-type groups.  Lines 35-37, 42-45.

 

Comment:

‘’ Table 1 illustrates substantial baseline imbalances between the treatment groups, which likely bias overall survival in favor of everolimus. Patients in the everolimus arm had fewer metastatic sites (<2 sites: 59.1% vs. 32.5%) and a markedly higher rate of bone-only disease (45.5% vs. 22.5%), both of which are indicators of more indolent disease biology. Although briefly acknowledged in the Conclusion, this critical limitation is insufficiently addressed in the Abstract and Discussion. The observed OS difference (18.3 vs. 9.6 months), therefore, appears more plausibly attributable to baseline prognostic differences rather than treatment effect. A multivariate Cox regression model adjusting for metastatic burden and visceral disease is warranted; if precluded by the small sample size (N = 62), this limitation should be clearly emphasized in the Abstract and early Discussion. ‘’

Answer:

Indeed, between the two groups, there are imbalances, most striking in the number of metastatic sites and bone-only disease. A multivariate analysis is essential for determining the impact of multiple covariates on the OS. Unfortunately, we did not include a multivariate analysis due to the very small sample size, which precluded the extraction of reliable results. However, after adjusting for baseline prognostic factors, including the number of metastatic sites and visceral disease, the treatment cohort was still not independently associated with OS. Please find explanations between lines 172-180 and 383-389.

 

Comment:

‘’ In Table 2, the Confidence Intervals are quite wide due to the small sample size. The authors ensure the p-value for the comparison of ORR between the two groups is provided. ‘’

Answer:

Confidence intervals were wide due to the small sample size, and ORR comparisons should be interpreted cautiously. ORR was numerically higher in the Alpelisib group compared with the Everolimus group (20.7% vs 5.3%). However, this difference is not statistically significant using a Fisher’s exact test, yielding a p-value of 0.22. Lines 224, 225.

 

Comment:

‘’ In the Discussion, the authors stated "Conversely, in terms of median OS, everolimus outperformed alpelisib by nearly doubling the OS... Despite the numerical difference, the result was not statistically significant..." It is scientifically inaccurate to state that one drug "outperformed" the other when the p-value is 0.41. In statistics, a p-value of 0.41 implies the numerical difference could easily be due to chance. The author should rephrase all instances claiming superiority and state clearly that "no statistically significant difference in OS was observed." The numerical trend can be mentioned, but it must be immediately contextualized in relation to the baseline imbalances. ‘’

Answer:

We restate the above phrase and ensure that every statistical result is presented within the proper scientific background throughout the manuscript. Lines 246-248.

 

Comment:

‘’ The manuscript uses commas for decimals (e.g., "4,9 months", "p=0,53") in the text and tables, but occasionally uses points. Standard English scientific publishing requires the use of points (e.g., "4.9 months"). Please standardize this throughout the text and tables. ‘’

Answer:

All commas, ‘’,’’ that were used as decimals in the manuscript have been substituted by points, ‘’.’’.