Determination of Octanol–Water Partition Coefficients for Corticosteroids and Its Application in a Screening-Level In Silico Environmental Risk Prioritization for Aquaculture Systems

Round 1

Reviewer 1 Report (New Reviewer)

Comments and Suggestions for Authors

Interesting work on computational workflow to bridge the gap between the molecular structure of corticosteroids and its potential risk in aquaculture systems. The discussion was done very well. Only some minor comment on the tables and figures which can be just provided in supplementary.  

1. Table 2 data of Adj R2 and SSE can be combined to table 1. remove other info on table 2 to supplementary info.
2. Similarly remove Fig 1 to supplementary
3. Page 15, para 2 and 3 needs literature support on the justification made.
4. Table 3 move to supplementary.

Author Response

# Reviewer 1:

Interesting work on computational workflow to bridge the gap between the molecular structure of corticosteroids and its potential risk in aquaculture systems. The discussion was done very well. Only some minor comment on the tables and figures which can be just provided in supplementary.  

1. Table 2 data of Adj R2 and SSE can be combined to table 1. remove other info on table 2 to supplementary info.

Response: Thank you — we agree. We have moved the full regression parameter table (intercept, slope, SE, etc.) to the Supplementary Information and condensed the main-text table to show only the key model-performance metrics required for comparison.

2. Similarly remove Fig 1 to supplementary

Response: Thank you for this suggestion. We respectfully chose to retain Fig. 1 in the main text because it provides the key visual evidence supporting the comparative evaluation of the nine computational programs and the selection of XLOGP3 as the preferred predictor. As this figure is central to the manuscript’s core methodological objective, moving it to the Supplementary Information would weaken the logical flow and make it harder for readers to follow the model-selection rationale. To improve readability, we have revised the figure layout and caption for greater clarity.

3. Page 15, para 2 and 3 needs literature support on the justification made.

Response: Thank you for this helpful suggestion. We added citations 46 and 47. Reference 46 supports the statement that esterification is a common medicinal chemistry strategy to increase corticosteroid lipophilicity by reducing hydrogen-bonding capacity and enhancing hydrophobic interactions. It also helps justify the observed trend that longer ester chains lead to higher logKow. Reference 47 supports the discussion of acetonide (cyclic ketal) formation, explaining that this modification masks two polar hydroxyl groups, reduces polarity, and therefore substantially increases lipophilicity and logKow.

4. Table 3 move to supplementary.

Response: Thank you for this suggestion. We respectfully chose to retain Table 3 in the main manuscript because it is not merely supporting information, but one of the core results of the study. Table 3 integrates the predicted logKow values with the derived logBCF and logKoc values for all 63 corticosteroids and steroid hormones, and it provides the basis for the environmental risk ranking discussed in Section 3.3. In this sense, the table is essential for readers to follow the logic of the screening-level assessment and to identify the priority compounds directly from the main text. Moving Table 3 to the Supplementary Information would weaken the readability and interpretability of the manuscript, since the table contains the central data set used to support the conclusions and is closely linked to the discussion of risk classes, Figures 3–5, and the prioritization of compounds for monitoring and follow-up studies. To improve presentation without removing the table, we have ensured that it is clearly formatted and properly referenced in the Results and Discussion.

Reviewer 2 Report (New Reviewer)

Comments and Suggestions for Authors

1、This study systematically evaluated the predictive performance of 9 computational programs for the logKow of corticosteroids and applied them to environmental risk assessment. The research approach is clear and the data is detailed. However, similar validation studies of computational programs have been reported more frequently for other compound categories (such as polychlorinated biphenyls, perfluorinated compounds). It is suggested that the authors should more prominently highlight the unique contributions of this study in the introduction section.

2、The author used the experimental logKow values of 50 steroid hormones to validate 9 calculation programs, and selected the best one, XLOGP3. Then, this program was used to predict the logKow values of 32 synthetic corticosteroids, and the BCF and Koc were further estimated. However, the predicted values of these 32 compounds lacked any experimental verification, which made the subsequent risk assessment based on the unverified predicted values. Suggestion:

It is clearly acknowledged that this is a screening-level risk assessment based on computational chemistry, rather than a definitive conclusion. If the logKow determination of a few representative compounds could be added to the experiments, it would significantly enhance the credibility of the conclusions.

3、The author used the BCF model and the Koc model for parameter estimation. Both of these models are based on single-parameter regression of logKow and have a simple form, which is suitable for high-throughput screening. However, their limitations have not been fully discussed:

The BCF model does not consider metabolic transformation, bioavailability, and the special nature of ionized compounds;

Corticosteroids are mostly weakly polar compounds, and some of their structures may partially ionize under environmental pH. Directly using the neutral logKow to predict Koc may lead to overestimation of the adsorption capacity.

It is suggested to include an analysis of uncertainties in the discussion section and to introduce machine learning or multi-parameter QSAR models in subsequent studies to improve the prediction accuracy.

4、Table 3 presents 63 compounds' names, abbreviations, logKow values from multiple sources, mean values, median values, logBCF and logKoc in a single table, with extremely high information density but poor readability.

5、The conclusion section does not provide sufficient information regarding the limitations of the study.

It states in the conclusion that "these in silico predictions generally represent the worst-case scenario", but does not specify which factors may lead to overestimation and which factors may result in underestimation.

6、The discussion on the relationship between structure and lipophilicity is comprehensive, but Figure 2 lacks quantitative characterization.
Figure 2 visually presents the effects of hydroxylation, esterification, and ketone formation on logKow using a structure + arrow format. The example compounds are selected reasonably. However, all the changes are only qualitatively described as "significantly increased/decreased", without providing specific numerical values for ΔlogKow. It is recommended: In the figure or the figure caption, supplement the logKow change amount corresponding to each structural modification (such as Δ = +2.12) to enhance the quantitative persuasiveness and to correspond with the text in Section 3.2.

Author Response

# Reviewer 2:

1、This study systematically evaluated the predictive performance of 9 computational programs for the logKow of corticosteroids and applied them to environmental risk assessment. The research approach is clear and the data is detailed. However, similar validation studies of computational programs have been reported more frequently for other compound categories (such as polychlorinated biphenyls, perfluorinated compounds). It is suggested that the authors should more prominently highlight the unique contributions of this study in the introduction section.

Response: We sincerely thank the reviewer for this constructive suggestion. We agree that the uniqueness of this study should be stated more explicitly in the Introduction. In response, we revised the Introduction to better emphasize the specific contributions of this work and to distinguish it from previous validation studies on other chemical classes such as PCBs and PFAS.

In particular, the unique contributions of this study are:

The first systematic benchmark of nine logKow prediction programs specifically for corticosteroids, a structurally complex and pharmacologically important class for which experimental lipophilicity data remain limited. A curated validation dataset of 50 steroid hormones with experimental logKow values, enabling a class-specific comparison rather than a generic benchmark across chemically unrelated compounds. A structure–lipophilicity relationship analysis tailored to corticosteroids, showing how hydroxylation, esterification, and acetonide formation systematically alter lipophilicity within this class. The translation of validated logKow predictions into screening-level environmental risk indicators by estimating BCF and Koc for 63 corticosteroids and prioritizing compounds of concern in aquaculture systems. The application context is especially relevant to aquaculture wastewater and microalgae-based remediation/feed recycling systems, which has not been addressed in previous logKow benchmark studies. To strengthen this point, we revised the Introduction to clearly state that this study is not simply another general validation of calculation software, but a class-specific benchmarking and screening framework for corticosteroids, with direct implications for environmental monitoring and risk prioritization in aquaculture

2、The author used the experimental logKow values of 50 steroid hormones to validate 9 calculation programs, and selected the best one, XLOGP3. Then, this program was used to predict the logKow values of 32 synthetic corticosteroids, and the BCF and Koc were further estimated. However, the predicted values of these 32 compounds lacked any experimental verification, which made the subsequent risk assessment based on the unverified predicted values. Suggestion:

It is clearly acknowledged that this is a screening-level risk assessment based on computational chemistry, rather than a definitive conclusion. If the logKow determination of a few representative compounds could be added to the experiments, it would significantly enhance the credibility of the conclusions.

Response: We sincerely thank the reviewer for this thoughtful and constructive suggestion. We fully agree that experimental determination of representative synthetic corticosteroids would further strengthen the study. However, in the present work we intentionally framed the environmental assessment as a screening-level, in silico prioritization based on a validated logKow prediction workflow rather than as a definitive fate assessment for each compound. Our primary objective was to establish and benchmark the most reliable computational tool for corticosteroids using a validation set of compounds with high-quality experimental data, and then apply that validated model to a broader set of synthetic corticosteroids for which experimental logKow values are largely unavailable in the literature. We did not add new experimental measurements in this revision because such a validation would require acquisition of additional reference standards, method development and calibration for chromatographic or shake-flask measurements, and substantial additional time and resources beyond the scope of the current study. Importantly, the 32 synthetic corticosteroids were selected precisely because experimental logKow data are scarce, which is one of the key reasons a computational screening approach is needed in the first place. To address the reviewer’s concern, we have revised the manuscript to state more explicitly that the predicted logKow, BCF, and Koc values for the 32 synthetic corticosteroids should be interpreted as prioritization indicators rather than definitive values. We also strengthened the Discussion and Conclusions to emphasize that these results provide a conservative first-step assessment for identifying compounds that warrant future experimental verification, monitoring, and toxicological follow-up.

3、The author used the BCF model and the Koc model for parameter estimation. Both of these models are based on single-parameter regression of logKow and have a simple form, which is suitable for high-throughput screening. However, their limitations have not been fully discussed:

The BCF model does not consider metabolic transformation, bioavailability, and the special nature of ionized compounds;

Corticosteroids are mostly weakly polar compounds, and some of their structures may partially ionize under environmental pH. Directly using the neutral logKow to predict Koc may lead to overestimation of the adsorption capacity.

It is suggested to include an analysis of uncertainties in the discussion section and to introduce machine learning or multi-parameter QSAR models in subsequent studies to improve the prediction accuracy.

Response: We sincerely thank the reviewer for this insightful and valuable comment. We agree that the limitations of the single-parameter BCF and Koc models should be discussed more explicitly. In the revised manuscript, we have strengthened the Discussion to clarify that both models are used here strictly for screening-level prioritization rather than definitive fate prediction. We now explicitly note that the BCF model does not account for metabolic transformation, bioavailability, or the special behavior of ionizable compounds, and that the Koc model may overestimate sorption for corticosteroids that can partially ionize under environmentally relevant pH conditions when neutral logKow values are used as the sole descriptor. In addition, we added a statement emphasizing that the reported BCF and Koc estimates should be interpreted as first-step indicators of potential environmental concern, especially in data-poor settings. We also noted that future studies could improve prediction accuracy by integrating multi-parameter QSAR/QSPR models, machine learning approaches, and pH-dependent descriptors, which may better capture the complexity of corticosteroid behavior in real aquatic environments.

4、Table 3 presents 63 compounds' names, abbreviations, logKow values from multiple sources, mean values, median values, logBCF and logKoc in a single table, with extremely high information density but poor readability.

Response: Thank you for this helpful comment. We agree that the original Table 3 contained too much information in a single table, which reduced readability. To improve clarity, we revised Table 3 by streamlining the main manuscript table to include only the key parameters used for environmental prioritization, namely compound name, abbreviation, XLOGP3-predicted logKow, logBCF, and logKoc.

5、The conclusion section does not provide sufficient information regarding the limitations of the study. It states in the conclusion that "these in silico predictions generally represent the worst-case scenario", but does not specify which factors may lead to overestimation and which factors may result in underestimation.

Response: We sincerely thank the reviewer for this important comment. We agree that the original Conclusion did not describe the study limitations with sufficient specificity. In the revised manuscript, we have expanded the Conclusion to clearly distinguish the main factors that may lead to overestimation or underestimation in the predicted values. Specifically, we now note that the predicted BCF values may be overestimated because the single-parameter model does not account for metabolic biotransformation, species-specific bioavailability, or reduced uptake due to ionization and physiological barriers. Likewise, the predicted Koc values may be overestimated for corticosteroids that can partially ionize under environmentally relevant pH conditions, since the model uses neutral logKow as the sole descriptor. At the same time, we also clarified that underestimation may occur for certain compounds because the simple logKow-based models cannot capture all structure-specific interactions, such as strong specific sorption, unconventional molecular conformations, or compound-specific behavior not represented in the model training domain. In addition, for some heavily modified corticosteroids, the predicted lipophilicity may be affected by structural complexity and intramolecular interactions that are not fully captured by linear single-descriptor models. Accordingly, we revised the Conclusion to emphasize that the results should be interpreted as screening-level prioritization outputs rather than definitive environmental fate values, and that future work should combine experimental verification with more advanced multi-parameter QSAR/QSPR or machine-learning approaches to reduce uncertainty.

6、The discussion on the relationship between structure and lipophilicity is comprehensive, but Figure 2 lacks quantitative characterization.

Figure 2 visually presents the effects of hydroxylation, esterification, and ketone formation on logKow using a structure + arrow format. The example compounds are selected reasonably. However, all the changes are only qualitatively described as "significantly increased/decreased", without providing specific numerical values for ΔlogKow. It is recommended: In the figure or the figure caption, supplement the logKow change amount corresponding to each structural modification (such as Δ = +2.12) to enhance the quantitative persuasiveness and to correspond with the text in Section 3.2.

Response: We thank the reviewer for this excellent suggestion. We agree that adding quantitative values for the change in lipophilicity significantly strengthens the figure’s impact and provides better alignment with the discussion in Section 3.2.

As suggested, we have revised Figure 2 to include the specific numerical changes associated with each structural modification. We also updated the Figure 2 caption to explicitly state these numerical shifts, thereby transitioning the analysis from a qualitative description to a quantitative characterization of the structure–lipophilicity relationships (SLRs) in corticosteroids.

Reviewer 3 Report (New Reviewer)

Comments and Suggestions for Authors

Dear Authors,

The manuscript addresses an important and timely topic, namely, the environmental relevance of corticosteroids in aquaculture systems. Overall, the study is well structured, and the computational analyses are conducted with care and methodological rigor. Nevertheless, despite the technical correctness of the analyses, several essential scientific and conceptual issues need to be addressed before the manuscript can be considered for publication in Water.

First, a formatting issue affects the review process. Line numbering is provided only up to Table 1, while it is no longer visible in the subsequent sections. As a result, precise attribution of reviewer comments to specific parts of the manuscript is difficult. For clarity and to facilitate a fair and efficient peer-review process, the authors are requested to ensure consistent line numbering throughout the entire manuscript.

Major Comments

1. The computational steps (calculation of logKow, estimation of BCF, and Koc based on QSPR models) are described across multiple sections of the manuscript; however, explicitly highlighting the novel aspects of this workflow will help readers appreciate its unique contribution and clarify the distinction between new methods and established tools.

- Explicitly define the computational workflow, including a schematic representation and step-by-step description (inputs, outputs, and potential decision points);

- Adjust the statements referring to "workflow development" and reframe the study as a structured application and comparative evaluation of existing computational tools for screening-level risk prioritization.

2. The manuscript states that it "links" molecular structure to environmental risk. While this is an appealing objective, the analysis presented in its current form does not fully support it.

The environmental risk assessment relies exclusively on estimated intrinsic physicochemical properties (logKow, BCF, Koc), without including exposure scenarios relevant to aquaculture systems or explicit risk indicators. This should be acknowledged as a limitation to maintain transparency and trust within the study's scope.

The resulting risk ranking, therefore, represents a screening-level prioritization rather than a comprehensive environmental risk assessment. Given that the predictions correspond to a worst-case scenario, this important limitation is not sufficiently reflected in the main conclusions.

It is recommended either to define the risk assessment as preliminary and screening-based explicitly, or to provide additional contextual information (e.g., hypothetical concentration ranges, comparisons with available monitoring data, or the use of simple risk indicators) to support claims of environmental risk relevance more convincingly.

3. Although the manuscript emphasizes the applicability of the results to aquaculture systems, a clearer discussion of how the estimated BCF and Koc values relate to realistic exposure pathways, such as bioaccumulation in aquatic organisms or sediment interactions, will help readers feel more confident in the environmental relevance of the findings.
4. To make a clearer distinction between results that are directly demonstrated (comparative evaluation of logKow, structure–lipophilicity relationships, screening-level prioritization), and prospective implications, which require further experimental validation.

Minor comments

I recommend clarifying from the outset that the environmental risk assessment is conducted in silico and intended for screening purposes.

Although the discussion of the XLOGP3 model's performance is detailed, including a summary table that highlights its strengths and limitations.

The limitation related to the lack of data on biotransformation and metabolism is relevant and important, but explicitly emphasizing these data limitations in the discussion section would make the audience feel respected.

Author Response

# Reviewer 3:

The manuscript addresses an important and timely topic, namely, the environmental relevance of corticosteroids in aquaculture systems. Overall, the study is well structured, and the computational analyses are conducted with care and methodological rigor. Nevertheless, despite the technical correctness of the analyses, several essential scientific and conceptual issues need to be addressed before the manuscript can be considered for publication in Water.

First, a formatting issue affects the review process. Line numbering is provided only up to Table 1, while it is no longer visible in the subsequent sections. As a result, precise attribution of reviewer comments to specific parts of the manuscript is difficult. For clarity and to facilitate a fair and efficient peer-review process, the authors are requested to ensure consistent line numbering throughout the entire manuscript.

Response: We sincerely apologize for the inconvenience caused by the inconsistent line numbering in the original submission. We noticed that the line numbers were automatically generated by the journal’s editorial system during the PDF conversion process, and unfortunately, they were truncated after Table 1. In this revised version, we have manually ensured that continuous and consistent line numbering is provided throughout the entire manuscript, from the title page to the references and supplementary information. We trust that this will facilitate a more efficient and precise review process.

Major Comments

1. The computational steps (calculation of logKow, estimation of BCF, and Koc based on QSPR models) are described across multiple sections of the manuscript; however, explicitly highlighting the novel aspects of this workflow will help readers appreciate its unique contribution and clarify the distinction between new methods and established tools.

- Explicitly define the computational workflow, including a schematic representation and step-by-step description (inputs, outputs, and potential decision points);

- Adjust the statements referring to "workflow development" and reframe the study as a structured application and comparative evaluation of existing computational tools for screening-level risk prioritization.

Response: We sincerely thank the reviewer for this helpful comment. We agree that the core value of this study lies in its systematic and class-specific implementation of existing computational tools. In response, we have made the following revisions to clarify the study’s scope: (1) We have adjusted the language throughout the manuscript, replacing “workflow development” with “structured application” and “comparative evaluation framework.” This clarifies that the study focuses on optimizing the use of existing tools for corticosteroids rather than creating new algorithms. (2) We added a dedicated paragraph in Section 2.5 (previously under 2.4) that explicitly outlines the sequential logic of the study: from input structure preparation, through benchmarking and selection of the optimal logKow tool, to the final risk-prioritization output for 63 compounds. We believe these textual enhancements provide a clear roadmap for the reader without the need for additional figures, ensuring the focus remains on the screening-level risk prioritization logic.

2. The manuscript states that it "links" molecular structure to environmental risk. While this is an appealing objective, the analysis presented in its current form does not fully support it.

The environmental risk assessment relies exclusively on estimated intrinsic physicochemical properties (logKow, BCF, Koc), without including exposure scenarios relevant to aquaculture systems or explicit risk indicators. This should be acknowledged as a limitation to maintain transparency and trust within the study's scope.

The resulting risk ranking, therefore, represents a screening-level prioritization rather than a comprehensive environmental risk assessment. Given that the predictions correspond to a worst-case scenario, this important limitation is not sufficiently reflected in the main conclusions.

It is recommended either to define the risk assessment as preliminary and screening-based explicitly, or to provide additional contextual information (e.g., hypothetical concentration ranges, comparisons with available monitoring data, or the use of simple risk indicators) to support claims of environmental risk relevance more convincingly.

Response: We agree with the reviewer’s perspective that a comprehensive environmental risk assessment requires the integration of both hazard (intrinsic properties) and exposure (environmental concentrations). Since experimental monitoring data and exposure scenarios for many synthetic corticosteroids in aquaculture are currently extremely sparse, a complete PEC/PNEC-based risk characterization (Risk Quotient) was beyond the current scope of this study. In response to your valuable suggestion, we have made the following major revisions: Reframing the Title and Objectives: We have changed the terminology from “comprehensive environmental risk assessment” to “screening-level environmental risk prioritization.” This more accurately reflects the study’s focus on ranking compounds based on their chemical “hazard potential” derived from molecular structure. Explicit Acknowledgment of Limitations: We added a detailed paragraph in Section 3.3 explaining that the ranking is based on intrinsic partitioning potential (Kow, BCF, Koc) and represents a “worst-case scenario” because it assumes persistence and does not account for metabolic biotransformation or specific exposure levels in different aquaculture systems. Strengthening Conclusions: The Conclusion section has been updated to emphasize that this prioritization serves as a preliminary screening tool to identify which compounds warrant urgent experimental monitoring and full-scale risk assessment in the future. We believe these changes maintain the scientific integrity of the work while providing a more transparent and realistic interpretation of our results.

3. Although the manuscript emphasizes the applicability of the results to aquaculture systems, a clearer discussion of how the estimated BCF and Koc values relate to realistic exposure pathways, such as bioaccumulation in aquatic organisms or sediment interactions, will help readers feel more confident in the environmental relevance of the findings.

Response: We sincerely thank the reviewer for this constructive suggestion. We agree that bridging the gap between intrinsic physicochemical properties and actual aquaculture exposure pathways is essential for demonstrating the environmental relevance of our findings. In the revised manuscript, we have expanded Section 3.3 to include a detailed discussion on how predicted BCF and Koc values translate into specific risks within the aquaculture ecosystem. Specifically, we now discuss: The “Sediment Reservoir” Effect: How high Koc values indicate that certain corticosteroids will persistently accumulate in pond sediments, creating a long-term exposure source for benthic organisms even after water exchange. Microalgae-Mediated Transfer: How the partitioning of lipophilic compounds into microalgae (used for bioremediation) creates a direct dietary exposure pathway for cultured fish, leading to bioaccumulation (high BCF). Food Safety Implications: The link between high BCF/lipophilicity and the potential for these contaminants to remain in the edible tissues of fish, poses a risk to human consumers. We believe these additions provide a much clearer ecological context for our screening-level prioritization.

4. To make a clearer distinction between results that are directly demonstrated (comparative evaluation of logKow, structure–lipophilicity relationships, screening-level prioritization), and prospective implications, which require further experimental validation.

Response: We thank the reviewer for this insightful suggestion. We agree that it is crucial to distinguish between the empirically validated findings of our study and the forward-looking environmental implications that stem from them. In the revised manuscript, we have carefully re-structured our Discussion and Conclusion sections to achieve this clarity: Directly Demonstrated Results: We clearly present the benchmarking of calculation programs, the identification of XLOGP3 as the optimal tool, the quantitative structure–lipophilicity relationships (SLRs), and the resulting screening-level prioritization list as the firm outputs of this work. Prospective Implications: We have adjusted our language when discussing downstream environmental fate (BCF and Koc) and ecological risks, using more cautious terms such as “potentially,” “suggested,” and “prospective.” Future Validation: We explicitly state that while the prioritization provides a necessary first step for managing data-poor corticosteroids, these prospective risks must be validated through future field monitoring and in vivo toxicological research. We believe this clearer distinction enhances the transparency and scientific rigor of the paper.

Minor comments

I recommend clarifying from the outset that the environmental risk assessment is conducted in silico and intended for screening purposes.

Response: We agree with the reviewer’s recommendation. To ensure the scope and methodology of our study are transparent from the very beginning, we have made the following revisions: Revised the Title: We have incorporated the terms “screening-level” and “in silico” into the title to clearly define the study’s nature. Updated the Abstract: The first few sentences of the Abstract now explicitly state that the risk assessment is an in silico screening-level prioritization. Reframed the Introduction: We have clarified the study’s objectives in the final paragraph of the Introduction, emphasizing that the output is a preliminary screening tool rather than a definitive environmental assessment. We believe these changes ensure that the “in silico screening” context is clear to the reader from the outset.

Although the discussion of the XLOGP3 model's performance is detailed, including a summary table that highlights its strengths and limitations.

Response: We thank the reviewer for this constructive suggestion to synthesize the performance characteristics of the XLOGP3 model. While we agree that a summary of its strengths and limitations is highly valuable, the manuscript already contains several comprehensive tables (including the revised Table 2 and supplementary tables). To maintain a concise presentation and avoid over-extending the manuscript’s length, we have opted to incorporate a dedicated discussion paragraph in Section 3.1 that explicitly summarizes these points instead of adding a new table.

In this new paragraph, we highlight XLOGP3’s strengths, such as its robust atom-additive logic and extensive training set which ensure high consistency for steroid scaffolds, as well as its limitations, such as potential inaccuracies in capturing complex long-range electronic interactions or specific stereochemical nuances not fully covered by its fragment-based parameters. We believe this addition provides the requested clarity while keeping the manuscript’s structure streamlined.

The limitation related to the lack of data on biotransformation and metabolism is relevant and important, but explicitly emphasizing these data limitations in the discussion section would make the audience feel respected.

Response: We sincerely appreciate the reviewer’s thoughtful guidance on this point. We fully agree that providing a transparent and explicit discussion of data limitations—particularly regarding biotransformation and metabolism—is essential for maintaining academic rigor and respecting the audience’s expertise. In the revised manuscript, we clarify that our BCF and risk predictions are based on “neutral partitioned” models, which do not account for the metabolic clearance rates of corticosteroids in aquatic organisms. We have highlighted that the lack of compound-specific metabolic data for many synthetic corticosteroids represents a significant knowledge gap, and that our current “worst-case” predictions should be interpreted with this context in mind. We believe this addition provides a more balanced and honest perspective on the study’s scope.

Round 2

Reviewer 2 Report (New Reviewer)

Comments and Suggestions for Authors

Agree to publish this version

Reviewer 3 Report (New Reviewer)

Comments and Suggestions for Authors

Dear Authors,

Thank you for your efforts to enhance the manuscript; the revisions are appreciated. 

This manuscript is a resubmission of an earlier submission. The following is a list of the peer review reports and author responses from that submission.

Round 1

Reviewer 1 Report

Comments and Suggestions for Authors

It is in the file attached.

Comments for author File: Comments.pdf

Reviewer 2 Report

Comments and Suggestions for Authors

Please check the attachment.

Comments for author File: Comments.pdf

Reviewer 3 Report

Comments and Suggestions for Authors

This paper presents a computational framework to evaluate the environmental risks of corticosteroids (CSs) in aquaculture. The authors assessed nine predictive programs for estimating the octanol–water partition coefficient (logKow), a key indicator of chemical lipophilicity and environmental behavior, using 50 steroid hormones as benchmarks.The authors then used the best performing orgram XLOGP3 to calculated additional 32 CSs, and used existing models to calculated BCF and Koc from the Kow values. 

While the topic is of importance, the manuscript lacks novelty to justify its publication. All the models used are existing models - The study introduces no fundamentally new computational or QSPR methodology. The models used to calculate Kow are existing models, the validation set comes from existing EPA database, and the equation used to calculate Koc and BCF comes from existing regression models. The authors introduced no new model, nor did they perform any new experimental measurement. The contribution is incremental, and the results are perhaps more suitable to publish as a dataset, rather than a new study.

The finding that XLOGP3 outperforms other logKow prediction models is not unexpected, as similar trends have been observed for other classes of hydrophobic organic compounds. The work still has its practical value by confirming this behavior for CSs. However, the practical contribution still does compensate for the lack in originality

Reviewer 4 Report

Comments and Suggestions for Authors

Dear authors,

I thank the authors for their efforts.
The manuscript titled "ADetermination of octanol-water partition coefficients for corti-costeroids and its application in a comprehensive in silico envi-ronmental risk assessment for aquaculture systems" presents interesting and valuable findings; however, some could be improved to strengthen the overall impact of the study.
The abstract should be shortened and highlight the main research results and recommendations if possible.
Figure 4:  remover line
Generally, the results have not been discussed. Please present the discussion.
Best regards,