Impact of Antibiotic Use in the Primary Treatment of Nasopharyngeal Carcinoma

Round 1

Reviewer 1 Report

Comments and Suggestions for Authors

The manuscript entitled “Impact of antibiotic use in the primary treatment of nasopharyngeal carcinoma” investigates an interesting and clinically relevant topic regarding the potential association between peri-treatment antibiotic exposure and survival outcomes in nasopharyngeal carcinoma. The study benefits from a relatively large cohort and long follow-up duration. I would like to offer the following major points for consideration by the authors towards the improvement of the manuscript:

1- The major limitation of this study is confounding by indication. Patients who received antibiotics were more likely to have advanced disease, undergo concurrent chemoradiotherapy, receive neoadjuvant chemotherapy, develop mucositis, neutropenic fever, or infections, all of which independently reflect poorer prognosis and treatment-related toxicity. Therefore, antibiotic use may simply represent a surrogate marker of frailty, treatment intensity, or complications rather than an independent prognostic factor

2- Importantly, antibiotic use around primary treatment lost statistical significance in the multivariate Cox regression model. Despite this, the manuscript continues to emphasize a negative prognostic effect of antibiotics throughout the Discussion and Conclusions

3- The subgroup analysis between “short-term survivors (<5 years)” and “long-term survivors (≥5 years)” appears to have been done after the fact and lacks a clear methodological justification. This approach may lead to survival bias and inflate apparent relationships.

4- The study does not adequately account for the clinical severity of infections, mucositis grades, hospitalization status, treatment interruptions, chemotherapy dose reductions, or sepsis-related complications.

5- Some subgroup analyses show incredibly large hazard ratios with extremely wide confidence intervals (e.g., HR >35 or >40 for antibiotics within 1-2 weeks before and after treatment).

6- The article groups together highly heterogeneous antibiotic regimens despite significant differences in spectrum, route of administration, duration, and indication. More detailed analyses classified by antibiotic class, cumulative exposure time, intravenous and oral administration, and indication would significantly strengthen the article.

7- Important prognostic biomarkers in NPC, particularly plasma EBV DNA levels, were not available and were not included in the analysis.

Author Response

Comments 1: The major limitation of this study is confounded by indication. Patients who received antibiotics were more likely to have advanced disease, undergo concurrent chemoradiotherapy, receive neoadjuvant chemotherapy, develop mucositis, neutropenic fever, or infections, all of which independently reflect poorer prognosis and treatment-related toxicity. Therefore, antibiotic use may simply represent a surrogate marker of frailty, treatment intensity, or complications rather than an independent prognostic factor.

Response 1: We highly appreciate the reviewer’s critical insight and completely agree that confounding by indication is a substantial limitation of our study. In Figure2, it demonstrates the effects of many potential factors on the prognosis of patients with RT only who were mainly at an early stage of NPC without any preference on Abx administration. In this patient subgroup, the disturbance of tumor stage and treatment strategies was little, therefore the influence of Abx was obvious. This strongly supports our hypothesis that Abx use around the primary treatment of NPC would impair or somehow affect the therapy. However, as you point out, patients requiring antibiotics inherently possessed higher risks of complications, while our retrospective data do not allow us to establish a definitive causal link between Abx itself and survival. Since the study started from 2010, the flaws in electrical medical record system limited our collection for more clinical information, such as the definite indication of Abx, the exact Abx period, etc. And to eliminate the possible biases brought by these factors, we conducted the multivariate analysis and even though the result of multivariate analysis on the effects of Abx were statistically negative, we kept showing it in the manuscript to make an honest statement. Therefore, we consider and highlight the Abx use during primary treatment of NPC as a strong and easily trackable clinical indicator (or "red flag") to identify patients at higher risk of treatment failure and poor prognosis to arouse more attention from clinical doctors for careful Abx prescription and following-up their relapse and other adverse events. These opinions have been stated in the first paragraph of DISCUSSION part from LINE 243 to LINE 248 to avoid misleading the readers about our conclusions and purposes.

 

Comments 2: Importantly, antibiotic use around primary treatment lost statistical significance in the multivariate Cox regression model. Despite this, the manuscript continues to emphasize a negative prognostic effect of antibiotics throughout the Discussion and Conclusions

Response 2: We deeply appreciate the reviewer’s questioning about the exact effects of Abx, especially with the absence of statistical difference when conducting a multivariate Cox regression model. According to the result of our univariate and multi-variable analysis, only tumor stage remained a significant independent prognostic factor model. We thank to this chance of revision that we could supplement a new citation of Weng et al.’s study as REF [16], which reported that within an NPC cohort sustaining severe mucositis, patients receiving antibiotic therapy sustained significantly worse OS and DFS. These peer-published data validate that the detrimental impact of antibiotic exposure remains robust even after rigorously adjusting for tumor burden. However, we also agree with your opinion that our study design and conclusions are still limited, we have carefully revised the entire manuscript, particularly the Results, Discussion, and Conclusion sections, improved our description and statement with a specific focus on not misleading the readers.

 

Comments 3: The subgroup analysis between “short-term survivors (<5 years)” and “long-term survivors (≥5 years)” appears to have been done after the fact and lacks a clear methodological justification. This approach may lead to survival bias and inflate apparent relationships.

Response 3: We appreciate the reviewer’s comment, but we would like to respectfully clarify that there appears to be a misunderstanding regarding our study design and selection of observing landmark. First, we did NOT stratify patients into ‘5-year’ and ‘10-year’ groups to do experiments separately to compare Abx exposure and survival status to pursue a positive result. The setting of ‘5-year’ and ‘10-year’ survival timepoints was decided based on our clinical practice, rather than an after-the-fact picking strategy. In the published paper talking about Hong Kong’s experience and authoritative population-based data from the Hong Kong Cancer Registry (Hospital Authority), we usually choose ‘5-year’ as the commonly used observation time point [1,2]. And according to these common sense and our data, patients with NPC tended to have a long survival timeline, so we looked at both 5 and 10 year.

Ref:

• Lee AW, et al. Treatment results for nasopharyngeal carcinoma in the modern era: the Hong Kong experience. Int J Radiat Oncol Biol Phys. 2005 Mar 15;61(4):1107-16. doi: 10.1016/j.ijrobp.2004.07.702. PMID: 15752890.
• https://www3.ha.org.hk/cancereg/pdf/survival/Stage-specific%20Survival%20of%20NPC%20in%20HK.pdf

 

Comments 4: The study does not adequately account for the clinical severity of infections, mucositis grades, hospitalization status, treatment interruptions, chemotherapy dose reductions, or sepsis-related complications.

Response 4: We sincerely appreciate the reviewer’s comment and completely agree that our study is subject to this limitation, which primarily attribute to the constrained information collection as well as the limited access within our electronic medical record system. We have transparently addressed this deficiency in the revised DISCUSSION section from LINE 362 to LINE 369.

 

Comments 5: Some subgroup analyses show incredibly large hazard ratios with extremely wide confidence intervals (e.g., HR >35 or >40 for antibiotics within 1-2 weeks before and after treatment).

Response 5: We sincerely thank the reviewer for pointing out this questionable statistical finding.  We completely understand the reviewer’s concern regarding the data stability implied by these extremely high HR value and wide 95%CIs. As the reviewer mentioned, these results reflected the possible effects of the antibiotic use within 2 weeks (HR = 35.996 [3.999, 323.979]) or 1 week (HR = 42.932 [4.759, 387.309]) on NPC primary treatment, which was in the population of patients who received RT only. Within this specific cohort, there were only 73 patients in total and only 17 patients received antibiotic intervention. Statistically, a small sample size with unbalanced event distribution introduced this sparse data bias and caused surprisingly high HR value and 95%CI. Despite the statistical instability, the reason why we still decided to retain this subgroup analysis in the manuscript was because it revealed that even in early-stage patients with favorable baseline prognoses, antibiotic use within primary treatment window displayed their potentials as a ‘red-flag’ for high-risk clinical vulnerability. We have already added more explanation about this part, stated this sparse data limitation and its clinical implications in the DISCUSSION part of revised manuscript from LINE355 to LINE 359.

 

Comments 6: The article groups together highly heterogeneous antibiotic regimens despite significant differences in spectrum, route of administration, duration, and indication. More detailed analyses classified by antibiotic class, cumulative exposure time, intravenous and oral administration, and indication would significantly strengthen the article.

Response 6: Agree. We tried to trace back the complications and indications, however due to the limitations of our electrical medical record system and our limited access to patient’s full medical records, including their attendance to other departments and the results of corresponding laboratory tests. We completely understand the importance of clarifying this aspect in the text, and we deeply apologize that we are unable to do so.

 

Comments 7: Important prognostic biomarkers in NPC, particularly plasma EBV DNA levels, were not available and were not included in the analysis.

Response 7: We are grateful and agree with the reviewer for this professional concern. First, since this study was retrospective cohort design conducted from 2010, EBV DNA level was not routinely used to evaluate or dynamically monitor tumor progression at that time, and it could not be analyzed commonly. Besides, retrospectively, we found that EBV test results were generally positive among these NPC patients. As you mentioned, the consideration about EBV infection status is indeed a flaw of our study, and it has been pointed out in the DISCUSSION, especially the paragraph stating the limitation of this study, from LINE 375 to LINE 379.

Reviewer 2 Report

Comments and Suggestions for Authors

The authors investigated the impact of antibiotic administration before and after treatment on the prognosis of nasopharyngeal cancer patients and found that the use of antibiotics around treatment with radiotherapy (RT), chemoradiotherapy (CCRT), and neoadjuvant chemotherapy (NC) in combination with CCRT was associated with a decrease in disease-specific survival.

Previous studies on head and neck cancer, small cell lung cancer, colorectal cancer, and ovarian cancer have suggested that antibiotics may adversely affect patient prognosis. Based on these prior studies, the aim of this study was to demonstrate the harmful effects of antibiotic administration peri-treatment on the outcomes of nasopharyngeal patients. However, the results do not necessarily support this hypothesis and depend on the patient’s disease stage and the type of radiation therapy or chemotherapy administered. To clarify the distinctive features of this paper, this point should be explicitly stated in the conclusion. There are several points that warrant further consideration.

 

If the authors wish to elucidate how antibiotics affect patient outcomes through their impact on the oral and gut microbiomes, it would be more appropriate to investigate the effects of antibiotics prior to RT and CCRT. The authors need to explain why they included the post-RT and post-CCRT periods in this study.

 

Adverse events of chemoradiotherapy such as severe mucositis and febrile neutropenia appear to be associated with the use of antibiotics during PTx. Since severe adverse events can lead to malnutrition and inflammation, they may be a major cause of poor prognosis, and antibiotics may be administered in response to these adverse events. If possible, it is preferable to report on the occurrence of serious adverse events.

 

It is necessary to explain why the effects of antibiotics administered peri-treatment were observed only in 5-year and 10-year disease-specific survival (DSS) rates, and why they had no impact on recurrence-free survival (RFS) or overall survival (OS).

 

Abstract, page 1, line 33: The authors state that “Multivariate analysis and subgroup analysis revealed that it also predicted a worse 5-year DSS among patients with shorter survival (p=0.019)”. The results for short-term survivors (less than 5 years) and long-term survivors (5 years or more) should also be presented in Table 4.

 

Table 2: What does “decreased” mean? An explanation is needed.

 

Figure 2. NC (B) and CCRT (C) need to be expanded as in (A). This suggests that the negative role of antibiotics becomes apparent when analyzing only RT patients, which is consistent with the findings in References 9, 38, and 39. Conversely, in the case of the NC and CCRT groups, while antibiotics may reduce the ability of the oral and intestinal microbiota to maintain antitumor immunity, they prevent bacterial infections associated with severe mucositis and neutropenia, thereby playing a positive role in patient prognosis. Combining the RT group and the NC/CCRT group may obscure the negative impact of antibiotics on prognosis. This issue warrants further consideration.

Author Response

Comments 1: If the authors wish to elucidate how antibiotics affect patient outcomes through their impact on the oral and gut microbiomes, it would be more appropriate to investigate the effects of antibiotics prior to RT and CCRT. The authors need to explain why they included the post-RT and post-CCRT periods in this study.

Response 1: We appreciate the comments from Reviewer 2. First, we agree that restricting our cohort strictly to pre-treatment Abx users would optimize the temporal ‘cause-and-effect’ logic. Nevertheless, as for our study design, we considered that NPC treatment was a prolonged status, and our dataset recorded the definitive start & end dates of chemoradiotherapy. We hypothesized that any Abx use within this window could disrupt microbial and immunological homeostasis, thus potentially affecting outcomes, which is also supported by two papers we added this time, Ref 16 and Ref 18. Therefore, we included all the cases ‘around’ this window but not just before the start date. We still agree with the review that the setting of ‘Abx around PTx’ somehow lacks a direct causal logic, and we have supplemented this point in the LIMITATIONS paragraph in DISCUSSION part from LINE 278 to LINE 300.  

 

Comments 2: Adverse events of chemoradiotherapy such as severe mucositis and febrile neutropenia appear to be associated with the use of antibiotics during PTx. Since severe adverse events can lead to malnutrition and inflammation, they may be a major cause of poor prognosis, and antibiotics may be administered in response to these adverse events. If possible, it is preferable to report on the occurrence of serious adverse events.

Response 2: Agree. We tried our best to track the complications, however due to the limitations of our electrical medical record system and our limited access to patient’s full medical records, including their attendance to other departments and the results of corresponding laboratory tests, we cannot supplement more details about their adverse events. And this critical limitation has been stated in the DISCUSSION part from LINE362 to LINE 369.

 

 

Comments 3: It is necessary to explain why the effects of antibiotics administered peri-treatment were observed only in 5-year and 10-year disease-specific survival (DSS) rates, and why they had no impact on recurrence-free survival (RFS) or overall survival (OS).

Response 3: We deeply appreciate the reviewer’s question, which is quite valuable. By definition, Disease-Specific Survival (DSS), Recurrence-Free Survival (RFS), and Overall Survival (OS) reflect different aspects of tumor burden and various underlying etiologies of adverse prognostic events. First, we suppose that DSS based on precise records could unmask a direct relationship between the variable and cancer-related mortality. RFS relatively focus on their impacts on tumor recurrence, whereas OS takes more other diseases and general body conditions into consideration, including systemic host immunity, and general nutritional status.

To explain the statistical discrepancy observed in our study that Abx exposure correlates with 5-year and 10-year DSS but demonstrates no significant association with RFS or OS, we admit that the independent biological impact of peri-treatment Abx use on NPC outcomes is restricted. The lack of correlation with RFS indicates that Abx would not accelerate tumor recurrence or promote aggressive malignant progression of tumor itself.

As for OS, non-cancer-related moralities carry a great statistical weight in long-term all-cause mortality, which was also consistent the finding of a large-scaled Hong Kong NPC Survivor Study (HKNPCSS) published in 2024 [1]. They tracked 7,893 NPC patients and monitor their 5- and 10-year survival status, and indicated that up to 59.9% of late deaths were attributed to  non-cancer complications rather than NPC itself [1]. The leading causes include respiratory diseases (predominantly recurrent aspiration pneumonia secondary to radiation-induced dysphagia and neck fibrosis) and cardiovascular events (such as ischemic heart disease and stroke).

Critically, it justifies our strict methodological decision to restrict the exposure definition to ‘Abx around Primary Treatment (PTx)’. By isolating this ‘window’, the selective correlation between peri-PTx Abx use and DSS successfully filters out competing noises, which strongly reinforces our primary hypothesis that acute antibiotic exposure around primary treatment may perturb the mucosal and systemic microbiome, thereby indirectly affecting the therapeutic effects of primary treatment of NPC or increasing the risks of complications. Related explanation about the concept of ‘Abx around PTx’ is supplemented in the DISCUSSION part from LINE 278 to LINE 281.

Ref : [1] Chi Leung Chiang et al. Late mortality and chronic health conditions in long-term survivors of nasopharyngeal cancers: The 20-year follow-up of territory-wide Hong Kong Nasopharyngeal Cancer Survivor Study (HKNPCSS). Journal of Clinical Oncology, 42(16_suppl), 12045-12045 (2024). DOI: 10.1200/JCO.2024.42.16_suppl.12045

 

Comments 4: Abstract, page 1, line 33: The authors state that “Multivariate analysis and subgroup analysis revealed that it also predicted a worse 5-year DSS among patients with shorter survival (p=0.019)”. The results for short-term survivors (less than 5 years) and long-term survivors (5 years or more) should also be presented in Table 4.

Response 4: We sincerely thank the reviewer for this suggestion. Since our original phrasing may have inadvertently caused some misunderstanding, we appreciate this chance to clarify more about this point. In Table 4, the analysis was based on the whole patient cohort; whereas the specific finding mentioned in the reviewer's comment was achieved by subgroup analysis after stratifying the whole patient population into short-term and long-term survivors. Merging these distinct subgroup results into the main population analysis could introduce substantial methodological confusion. If the reviewer feels that retaining this exploratory subgroup analysis might bring confusion for readers, we could remove the corresponding text from LINE 224 to LINE231.  

 

Comments 5: Table 2: What does “decreased” mean? An explanation is needed.

Response 5: Thank you for this question. In Table 2, we use the word ‘deceased’ to indicate patient death.

 

Comments 6: Figure 2. NC (B) and CCRT (C) need to be expanded as in (A). This suggests that the negative role of antibiotics becomes apparent when analyzing only RT patients, which is consistent with the findings in References 9, 38, and 39. Conversely, in the case of the NC and CCRT groups, while antibiotics may reduce the ability of the oral and intestinal microbiota to maintain antitumor immunity, they prevent bacterial infections associated with severe mucositis and neutropenia, thereby playing a positive role in patient prognosis. Combining the RT group and the NC/CCRT group may obscure the negative impact of antibiotics on prognosis. This issue warrants further consideration.

Response 6: We sincerely thank the reviewer for this comment. The explanation regarding the balancing effects of antibiotics in the NC and CCRT groups helps a lot to explain our findings. As recommended, we have expanded panels B and C in Figure 2 to match the scale of panel A to fully present the survival curves of these cohorts. We completely agree that in the NC and CCRT groups, the effects of antibiotics via microbiota disruption is likely counterbalanced by their life-saving benefit. Following your insightful guidance, we have enriched our DISCUSSION section from LINE 249 to LINE255, LINE 261 to LINE 264, LINE 266 to LINE 277 to detail it.

Reviewer 3 Report

Comments and Suggestions for Authors

This is an interesting and well conducted work on NPC and antibiotic use, providing - as reported by the authors - a retrospective basis for future prospective and translational studies on microbiome.

- I would underline that standard of care has changed for NPC since 2010-2014
- I would differentiate staging in locally advanced IV and metastatic IV stage
- i would focus the analysis on negative prognostic factors in the results chapter only on antibiotic therapy or significantly related to antibiotic therapy, since it is the aim of the paper
- as reported in the discussion chapter "The underlying cause may not only be attributed to the direct impact of antibiotics themselves on cancer treatment, but rather primarily to clinical indications for antibiotic use that potentially lead to rapid tumor deterioration and disease progression." If feasible (I undestand that it can be difficult to recollect all the informations), I'd report reasons for antibiotic therapy in a more extensive manner, even if it can be available only on a part of the population. Similarly, I would stress that this is a significant limitation of the study in line 340 and following.
- there are some typo "et.al" in line 282 and 284

Author Response

Comments 1: I would underline that standard of care has changed for NPC since 2010-2014.

Response 1: We thank the reviewer a lot for this constructive comment and emphasize the importance of the inductive chemotherapy/ ‘neoadjuvant chemotherapy (NC) in our study and we supplement some discussion about this point in the manuscript from LINE 369 to LINE 375, LINE 385 to LINE 387.

 

Comments 2: I would differentiate staging in locally advanced IV and metastatic IV stage.

Response 2: This is a great point which we also considered when doing data analysis. However, in our patient cohort, there was only 1 patient with distant metastasis, which made the grouping impossible to conduct. We have already supplemented some description in RESULTS part on LINE 133.

 

Comments 3: I would focus the analysis on negative prognostic factors in the results chapter only on antibiotic therapy or significantly related to antibiotic therapy, since it is the aim of the paper.

Response 3: We agree with the reviewer's insightful perspective. But we still decide not to exclude the analysis of other factors, the factors here are most basic and essential. Introducing these demographic characteristics of NPC patients with or without Abx use can also disclosed that no underlying relationship between these factors and Abx use was found but only patients with advanced tumor stages might have more Abx use. This finding caught our attention and explained why we focused on Abx use and patients’ prognosis. In Table 3, we displayed the analysis of Abx use and different endpoints, and different Abx characteristics, like the timepoint, types, type of administration on the 5-year and 10-year DSS.

 

Comments 4: as reported in the discussion chapter "The underlying cause may not only be attributed to the direct impact of antibiotics themselves on cancer treatment, but rather primarily to clinical indications for antibiotic use that potentially lead to rapid tumor deterioration and disease progression." If feasible (I undestand that it can be difficult to recollect all the informations), I'd report reasons for antibiotic therapy in a more extensive manner, even if it can be available only on a part of the population. Similarly, I would stress that this is a significant limitation of the study in line 340 and following.

Response 4: We tried our best to track the complications, however due to the limitations of our electrical medical record system and our limited access to patient’s full medical records, including their attendance to other departments and the results of corresponding laboratory tests, we feel so sorry but have no way to supplement more information. And this critical limitation has been stated in the DISCUSSION part from LINE 362 to LINE 369. 

 

Comments 5: there are some typo "et.al" in line 282 and 284

Response 5: We appreciate the reviewer’s careful reading and we have corrected all possible typo of ‘et.al’ in the manuscript.

Reviewer 4 Report

Comments and Suggestions for Authors

This manuscript addresses an important and clinically relevant question: whether antibiotic therapy administered during treatment for nasopharyngeal carcinoma influences patient survival. The authors analyze a retrospective cohort of 455 patients with more than 10 years of follow‑up, which is a notable strength. The study is generally well executed, and the results are clearly presented. However, some revisions are required for its improvement.

1. The cohort includes patients treated between 2010 and 2014. The authors should specify the date of last follow‑up or last data collection for surviving patients to ensure transparency regarding follow‑up duration.
2. Over the past decade, targeted therapies and immunotherapies have become integral to the management of locally advanced, recurrent or metastatic nasopharyngeal carcinoma. The authors should discuss how the introduction of these agents might influence or modify the relevance of their findings, particularly regarding antibiotic exposure and long‑term outcomes.
3. The study uses staging definitions contemporary to the treatment period. The authors should comment on how restaging the cohort using the current AJCC staging system (8th edition) might affect interpretation. If restaging is not feasible, acknowledging this as a limitation would strengthen the discussion.
4. The manuscript does not sufficiently reference prior work on this topic. Important studies such as Weng et al. (doi: 10.1177/0300060519874899) and the recent analysis by Fang et al. (doi: 10.1097/CJI.0000000000000556) should be included, as both support and contextualize the present findings.
5. The discussion should be expanded to address how clinicians should interpret and apply these findings. Specifically, the authors should comment on how to balance the need for antibiotic therapy during nasopharyngeal carcinoma treatment with the potential prognostic implications identified in the study.
6. On line 62, reference 9 pertains only to head and neck tumors. It should not be used to support statements regarding leukemia or lymphoma, and this citation should be corrected.

Author Response

Comments 1: The cohort includes patients treated between 2010 and 2014. The authors should specify the date of last follow‑up or last data collection for surviving patients to ensure transparency regarding follow‑up duration.

Response 1: We are grateful to the reviewer for raising this point to improve the manuscript, and we supplemented corresponding information as ‘The first decease of patient was noted on November 21st 2010, about 5 months after primary treatment, and the last follow-up of surviving patient was recorded as August 4th 2021.’ from LINE 147 to LINE 149.

 

 

Comments 2: Over the past decade, targeted therapies and immunotherapies have become integral to the management of locally advanced, recurrent or metastatic nasopharyngeal carcinoma. The authors should discuss how the introduction of these agents might influence or modify the relevance of their findings, particularly regarding antibiotic exposure and long-term outcomes.

Response 2: We highly appreciate this chance to have a revision on the manuscript and improve our paper, thank to the reviewer's professional suggestion and recommendation for one representative paper in Comment 4. As for the potential impacts brought by the Abx administration on the immunotherapy cancer patients, there is no published study addressing the underlying mechanism between anti-PD-1 monoclonal antibody and the Abx use. We discussed several possible pathways in the DISCUSSION part from LINE 295 to LINE 319. Besides, referring to the paper you mentioned, we agreed that the timepoint of introducing the antibiotics really mattered in many cases. Therefore, we made a major improvement about discussion this part from LINE 278 to LINE 295.      

 

 

Comments 3: The study uses staging definitions contemporary to the treatment period. The authors should comment on how restaging the cohort using the current AJCC staging system (8th edition) might affect interpretation. If restaging is not feasible, acknowledging this as a limitation would strengthen the discussion.

Response 3: We totally agree with the reviewer's comment. There was a large transition from the AJCC 7th to the 8th edition, not only on changing clinical decision-making and treatment strategies, but also on the specific staging. It subsequently affected the grouping of patients in our study. However, because this is a retrospective cohort, our clinical records from the treatment period just documented tumor parameters based on the AJCC 7th standards. Unfortunately, some key information, such as the exact site (specifically the involvement of the medial or lateral pterygoid muscles) of the tumor and the precise evaluation of lymph node stations, were not available in my current data records. Consequently, we cannot re-stage our cohort according to the AJCC 8th edition. We sincerely thank the reviewer for this constructive guidance, and we accordingly, have explicitly acknowledged this limitation in the revised DISCUSSION section from LINE 369 to LINE 375, LINE 385 to LINE 387. We hope the improvement would help to remind readers that our findings should be interpreted within the context of the AJCC 7th edition evaluation system.

 

Comments 4: The manuscript does not sufficiently reference prior work on this topic. Important studies such as Weng et al. (doi: 10.1177/0300060519874899) and the recent analysis by Fang et al. (doi: 10.1097/CJI.0000000000000556) should be included, as both support and contextualize the present findings.

Response 4: We would like to express our sincere gratitude to the reviewer for these two highly valuable literature recommendations, which significantly help us to improve our discussion and interpretation about the results. The controlled cohort by Weng et al. perfectly complements our work by restricting their target population to NPC patients with severe radiation-induced mucositis. They elegantly isolated and proved that antibiotic exposure acts as an independent adverse prognostic factor. Similarly, Fang et al. defined a "periprocedural treatment window" to effectively filters out the overlapping noise of antibiotic prescriptions driven by long-term chronic complications, such as recurrent late-onset aspiration pneumonia. It may facilitate the investigation into the direct biological impact of antibiotics on oncological efficacy.

We once again thank the reviewer for recommending these two publications, and we have seamlessly integrated them into our revised Discussion section from LINE 271 to LINE 277, LINE 286 to 295.

 

 

Comments 5: The discussion should be expanded to address how clinicians should interpret and apply these findings. Specifically, the authors should comment on how to balance the need for antibiotic therapy during nasopharyngeal carcinoma treatment with the potential prognostic implications identified in the study.

Response 5: This is a vital question, and it aligns perfectly with the objective of our study. Following the current guideline, neo-adjuvant chemotherapy with chemoradiotherapy is the most recommended strategy for locally advanced NPC, whereas this recipe is along with the risk of severe complications, such as mucositis and aspiration pneumonia, which inevitably require antibiotic intervention. And based on our findings and the conclusion of other studies, we call for more caution by clinical doctors about strictly avoiding prophylactic antibiotic use and implementing a culture-guided administration. Moreover, the antibiotic exposure should be carefully decided and avoid the therapeutic window of primary treatment as much as possible, since we noticed that antibiotics serve as a more pronounced indicator of poor prognosis when administered within 1 to 2 weeks around the PTx. The optimization of Abx intervention timing is likely to minimize their possible worsening effects. In the revised version, we have put more emphasis on this opinion in the DISCUSSION part from LINE 286 to 295.

 

 

Comments 6 On line 62, reference 9 pertains only to head and neck tumors. It should not be used to support statements regarding leukemia or lymphoma, and this citation should be corrected.

Response 6: We sincerely thank the reviewer for this careful reviewing our manuscript and this suggestion to improve. We have now incorporated the study by Pflug N, et al. (doi: 10.1080/2162402X.2016.1150399) as REF [9] to further support our statement regarding the negative impact of antibiotics on the OSPFS in chronic lymphocytic leukemia and relapsed lymphoma.

Round 2

Reviewer 1 Report

Comments and Suggestions for Authors

I am satisfied that the authors have addressed all of my previous concerns about the article. It is now much improved and I feel that it is now suitable for publication.

 

Author Response

Comment: I am satisfied that the authors have addressed all of my previous concerns about the article. It is now much improved and I feel that it is now suitable for publication.

Response: We sincerely appreciate the reviewer for his/her insightful suggestions. These comments are very helpful, providing valuable guidance, not only for improving this manuscript but also for our future research designs.

Reviewer 2 Report

Comments and Suggestions for Authors

The authors respond appropriately to most points raised by the reviewer.

Regarding Comment 4, as this is a key finding of this study, it is recommended to reconsider the description in the abstract (page 1, lines 33–34).

For example,

In the subgroup of patients who experienced disease-specific events within 5 years, antibiotic use remained an independent predictor of DSS (p=0.019) in multivariate analysis.

Author Response

Comment: The authors respond appropriately to most points raised by the reviewer. Regarding Comment 4, as this is a key finding of this study, it is recommended to reconsider the description in the abstract (page 1, lines 33–34). For example, in the subgroup of patients who experienced disease-specific events within 5 years, antibiotic use remained an independent predictor of DSS (p=0.019) in multivariate analysis.

Reply: We greatly appreciate the reviewers' comment and have already made corresponding revision.

Reviewer 3 Report

Comments and Suggestions for Authors

All previous requests, when feasible, have been satisfied and those who were not feasible have been explained correctly in the paper.

Author Response

Comment: All previous requests, when feasible, have been satisfied and those who were not feasible have been explained correctly in the paper.

Response: As the reviewer correctly pointed out, while our study demonstrates novelty and clinical significance, it also carries certain limitations. We will keep these insightful suggestions in mind to improve both our clinical practice and future research. Thank you once again for these feedbacks.