Review Reports - Molecular Subtypes of Pancreatic Cancer: A Review of the Literature

Round 1

Reviewer 1 Report

Comments and Suggestions for Authors

The manuscript addresses a timely and clinically relevant topic and provides a useful overview of the major molecular classification systems in pancreatic ductal adenocarcinoma. The topic is important, and the review has clear potential value for readers interested in PDAC biology and translational oncology.

Introduction
The introduction gives a clear rationale for why molecular subtyping in PDAC deserves attention. I suggest the authors clarify a little earlier whether the main focus of the review is prognostic value, predictive value, or both, because these concepts are closely related but not identical, and separating them from the beginning would improve the flow of the manuscript.

Materials and Methods
The Materials and Methods section is helpful in showing the databases searched, the search concept, and the general eligibility criteria. I suggest the authors clarify the review design more carefully. Since the manuscript includes a structured search strategy and a study selection flowchart, it may appear to readers as a systematic review; however, the manuscript also states that no protocol was registered and that no risk-of-bias assessment was performed. For this reason, it may be better either to frame the article more clearly as a narrative review with a structured search or to strengthen the methodological section accordingly.

Limitations of the Review
I appreciate that the manuscript openly acknowledges key limitations such as overlapping study populations, pre-analytical variability, and the absence of protocol registration and bias assessment. I suggest the authors reflect these limitations more clearly in the abstract and final conclusion as well, so that the overall message remains fully balanced with the level of evidence presented.

Results and Study Selection
The Results section provides a broad overview of the included studies and helps the reader follow the literature selection process. I suggest the authors clarify whether study screening and full-text assessment were performed by one reviewer or more than one reviewer, as this would improve transparency and make the review process easier to interpret.

Prognostic Value of Molecular Subtyping
This is one of the more informative parts of the review, especially because it shows a recurring pattern in which basal-like, squamous, or quasi-mesenchymal phenotypes tend to be associated with poorer outcomes. I suggest the authors slightly moderate the wording when discussing the consistency of prognostic findings across all classification systems, because the evidence summarized in the manuscript also indicates that not all classifiers perform equally well across studies or in multivariate analyses. A more synthesized comparison of which systems appear most reproducible would strengthen this part further.

Predictive Value and Treatment Implications
I also appreciate that the manuscript goes beyond prognosis and tries to discuss treatment implications. This adds translational value to the review. I suggest the authors use slightly more cautious wording in the section on predictive utility, because the evidence appears promising but still preliminary. Since only a limited number of studies directly address treatment response, and one of them is preclinical, it may be more appropriate to present these findings as emerging evidence rather than as evidence ready for routine clinical decision-making.

Discussion
The Discussion is informative, especially where it explains the relationship and overlap between the Collisson, Moffitt, Bailey, TCGA, and later purity-aware classifications. I suggest the authors make this section a little more synthetic and concise in places, with a clearer summary of the main common message across these classification systems. This would improve readability and help readers identify the key take-home points more easily.

Methodological Variability and Sample-Related Issues
The discussion of methodological differences, tumor purity, stromal admixture, and sample source is very relevant and adds important context to the review. I suggest the authors connect this point even more directly to the final interpretation of the literature, because these factors likely explain at least part of the variability seen across studies and should be emphasized when discussing the current limitations of molecular subtyping.

Clinical Implementation
The section on clinical implementation is useful and practical, particularly the discussion of FFPE material, turnaround time, cost, and the possible use of surrogate markers such as GATA6 and KRT17. I suggest the authors distinguish a little more clearly between what is currently feasible in research settings and what is truly ready for routine clinical practice, as this would make the translational message more precise and realistic.

Figures, Tables, and Supplementary Material
The figures and tables are helpful and improve the readability of the manuscript. I suggest the authors carefully review consistency between the main text and the supplementary material, especially study numbering, formatting, and the presentation of outcomes, because a few inconsistencies make cross-checking slightly difficult for the reader.

Conclusion
The conclusion is relevant and appropriately highlights the growing interest in molecular profiling of PDAC. I suggest the authors revise the final paragraph slightly so that it more clearly separates prognostic association, predictive potential, and routine clinical applicability. At present, the wording feels a little stronger than the evidence summarized earlier in the review.

Overall Recommendation
Overall, this manuscript addresses an important topic and brings together a substantial body of literature in a useful way. I suggest the authors revise the manuscript to improve methodological clarity, moderate a few of the broader interpretations, strengthen the synthesis across studies, and carefully edit the language for clarity and consistency. With these revisions, the review could become a valuable contribution to the field.

Comments on the Quality of English Language

The manuscript is understandable and communicates the main scientific points well. I suggest the authors revise the English for grammar, sentence flow, punctuation, spacing, and consistency of phrasing, as careful language editing would improve readability and give the review a more polished presentation.

Author Response

RESPONSE TO REVIEWERS

Manuscript: Molecular subtypes of pancreatic cancer: a review of the literature

Authors: Jakub Wnuk, Wiktoria Skowron, Anna Długaszek, Joanna Sadurska, Łukasz Pietrzyński, Jacek Kabut, Iwona Gisterek-Grocholska

We would like to thank the reviewers or their valuable and detailed comments, suggestions, and time spent reviewing the manuscript. We believe that after completing the suggested revisions, the manuscript’s overall presentation and clarity have improved.

We look forward hearing from you soon.

Sincerely, on behalf of all the authors,

Jakub Wnuk

Introduction

The introduction gives a clear rationale for why molecular subtyping in PDAC deserves attention. I suggest the authors clarify a little earlier whether the main focus of the review is prognostic value, predictive value, or both, because these concepts are closely related but not identical, and separating them from the beginning would improve the flow of the manuscript.

Thank you for your comment. We would amended the 'Aims' paragraph in accordance with your advice:

“It is a valid and important task to check the clinical utility of the molecular subtyping of PDAC. The primary aim of this study was to review the current evidence for the prognostic utility of PDAC molecular profiling. The second aim was to evaluate whether the molecular subtyping of PDAC could be useful in predicting systemic treatment outcomes.”

Materials and Methods

The Materials and Methods section is helpful in showing the databases searched, the search concept, and the general eligibility criteria. I suggest the authors clarify the review design more carefully. Since the manuscript includes a structured search strategy and a study selection flowchart, it may appear to readers as a systematic review; however, the manuscript also states that no protocol was registered and that no risk-of-bias assessment was performed. For this reason, it may be better either to frame the article more clearly as a narrative review with a structured search or to strengthen the methodological section accordingly.

Limitations of the Review

I appreciate that the manuscript openly acknowledges key limitations such as overlapping study populations, pre-analytical variability, and the absence of protocol registration and bias assessment. I suggest the authors reflect these limitations more clearly in the abstract and final conclusion as well, so that the overall message remains fully balanced with the level of evidence presented.

We would like to make this clear to readers. We have added the phrase,“Due to its limitations, this article should be considered a narrative review with a structured search strategy” to the beginning of the Materials and Methods section. We also added statements to the Abstract (“Due to its limitations, including the lack of a registered protocol and risk of bias assessment for the included studies and those whose results were not included, this study should be considered a narrative review with a structured search strategy rather than a systematic review.”) and Conclusions(“Due to the limitations mentioned earlier in the manuscript, we would like to emphasise that this review should be considered a narrative review with a structured search strategy. Therefore, the strength of the recommended directions of development of PDAC molecular subtyping and the conclusions are diminished.”).

Results and Study Selection

The Results section provides a broad overview of the included studies and helps the reader follow the literature selection process. I suggest the authors clarify whether study screening and full-text assessment were performed by one reviewer or more than one reviewer, as this would improve transparency and make the review process easier to interpret.

We have added the following sentence to the 'Materials and Methods' section: 'Two authors (J.W. and W.S.) performed the study screening and full-text assessment.'

Prognostic Value of Molecular Subtyping

This is one of the more informative parts of the review, especially because it shows a recurring pattern in which basal-like, squamous, or quasi-mesenchymal phenotypes tend to be associated with poorer outcomes. I suggest the authors slightly moderate the wording when discussing the consistency of prognostic findings across all classification systems, because the evidence summarized in the manuscript also indicates that not all classifiers perform equally well across studies or in multivariate analyses. A more synthesized comparison of which systems appear most reproducible would strengthen this part further.

Thank you for this comment.

We have adjusted this part of manuscript, adding several paragraphs:

“The studies included in this analysis consistently suggest that quasi-mesenchymal, squamous, or basal-like subtypes of PDAC are associated with poorer prognosis based on OS, DFS, or DSS. However, while the direction of this association appears relatively stable across studies, its strength and statistical robustness are not uniform. In particular, not all classification systems retain independent prognostic value in multivariate analyses, and their performance may vary depending on study design, patient population, and methodological factors.”

“Taken together, these findings suggest that although multiple classification systems capture a similar underlying biological signal, they differ in their reproducibility and prognostic performance. Among the commonly used classifiers, those proposed by Moffitt and Bailey appear to demonstrate more consistent prognostic value across studies and are more likely to retain significance in multivariate analyses. In contrast, other systems, such as the Collisson classification, show greater variability in their prognostic performance, which may reflect differences in methodology, sample composition, or cohort characteristics.

Overall, while a recurring prognostic pattern can be observed across different molecular classification systems, the degree of its reproducibility and statistical independence varies, hig-hlighting the need for further standardization and prospective validation.”

Predictive Value and Treatment Implications

I also appreciate that the manuscript goes beyond prognosis and tries to discuss treatment implications. This adds translational value to the review. I suggest the authors use slightly more cautious wording in the section on predictive utility, because the evidence appears promising but still preliminary. Since only a limited number of studies directly address treatment response, and one of them is preclinical, it may be more appropriate to present these findings as emerging evidence rather than as evidence ready for routine clinical decision-making.

Thank you for your comment on this matter. We added the following paragraph to our manuscript:

“A growing but still limited body of evidence suggests that PDAC molecular subtypes may have predictive value for systemic therapy. Although several studies report consistent differences in treatment response between classical and basal-like tumors, the available data remain relatively sparse and are derived largely from retrospective analyses, with only a small number of studies directly addressing treatment outcomes in this context. Therefore, these findings should currently be interpreted as emerging evidence rather than as a basis for routine clinical decision-making.

Classical/pancreatic–progenitor tumors appear to be better treated by intensified regimens such as (modified) FOLFIRINOX, whereas basal‑like/squamous tumors have consistently poorer outcomes and seem relatively chemoresistant across both FOLFI-RINOX and gemcitabine plus nab‑paclitaxel. This pattern supports the potential role of mo-lecular subtyping in identifying patients who may be more likely to benefit from aggressive com-bination chemotherapy and in recognizing a subgroup in whom escalation of standard cytotoxics may offer limited additional benefit. Basal-like tumors appear to show biological features associated with treatment resistance, including a loss of pancreatic lineage specifiers, enrichment of EMT and stem‑like programs, and activation of MYC/ΔNp63‑driven signaling, which provides a plausible mechanistic basis for their inferior responses.”

“Taken together, these observations suggest a biologically plausible relationship between mo-lecular subtype and treatment sensitivity; however, the current level of evidence remains insufficient for clinical implementation. Importantly, the number of studies directly evaluating treatment response by molecular subtype is limited, and their heterogeneity—in terms of study design, patient populations, and therapeutic regimens—makes direct comparison challenging. In addition, part of the available evidence is derived from preclinical models, which further limits its immediate translational applicability.

Overall, while the available data point toward a potentially important predictive role of mo-lecular subtyping in PDAC, prospective validation in biomarker-stratified clinical trials is required before these approaches can be reliably incorporated into routine treatment decision-making.”

Discussion

The Discussion is informative, especially where it explains the relationship and overlap between the Collisson, Moffitt, Bailey, TCGA, and later purity-aware classifications. I suggest the authors make this section a little more synthetic and concise in places, with a clearer summary of the main common message across these classification systems. This would improve readability and help readers identify the key take-home points more easily.

We appreciate your feedback on the discussion. I agree that we can summarize the topic more clearly; to summarize this part, we have added the following text:

“Taken together, despite differences in methodology, nomenclature, and sample composition, the major transcriptomic classification systems of PDAC converge on a shared biological framework. In practice, most tumors can be consistently mapped onto two principal epithelial lineages: a classical/pancreatic progenitor group, associated with preserved differentiation and more favorable outcomes, and a basal-like/squamous group, characterized by a loss of epithelial identity and poorer prognosis.

In retrospect, while preparing this review, it became increasingly clear to us that many of the additional subtypes described across studies—such as ADEX, immunogenic, or exocrine-like—are less consistently reproduced and are often influenced by stromal, immune, or low-purity sample components rather than representing stable, tumor-intrinsic categories. This perspective does not invalidate earlier classifications, but rather places them within a more unified and biologically coherent framework

Framing the literature in this way may help reconcile some of the apparent inconsistencies between studies and provides a clearer conceptual basis for interpreting the clinical relevance of molecular subtyping in PDAC.”

Methodological Variability and Sample-Related Issues

The discussion of methodological differences, tumor purity, stromal admixture, and sample source is very relevant and adds important context to the review. I suggest the authors connect this point even more directly to the final interpretation of the literature, because these factors likely explain at least part of the variability seen across studies and should be emphasized when discussing the current limitations of molecular subtyping.

We appreciate this remark and have added several paragraphs addressing this subject to the discussion of sample-related issues:

“Importantly, these methodological and sample-related differences are not merely technical considerations but have direct implications for the interpretation of the available evidence. In re-trospect, while analyzing the included studies, it became increasingly apparent to us that at least part of the variability in reported prognostic and predictive effects may stem from differences in tumor purity, stromal admixture, and sample processing rather than true biological discrepancies between cohorts. For instance, classifications derived from bulk RNA sequencing of low-purity samples may partially reflect stromal or immune components, which can lead to apparent subtype distinctions that are not entirely tumor-intrinsic.

This may help explain why some studies report weaker or inconsistent associations between molecular subtypes and survival outcomes, despite an overall trend favoring the classical versus basal-like dichotomy. In other words, the heterogeneity observed across studies should not be in-terpreted solely as a lack of robustness of molecular subtyping itself, but rather as a consequence of differences in how the underlying biological signal is captured.

At the same time, this reinforces one of the central limitations of the current literature: without standardized approaches to sample selection, purity assessment, and data processing, direct comparison between studies remains challenging. Therefore, methodological variability and sample composition should be explicitly considered when interpreting subtype classifications, and future research should aim to integrate purity-aware or deconvolution-based approaches more consistently.”

Clinical Implementation

The section on clinical implementation is useful and practical, particularly the discussion of FFPE material, turnaround time, cost, and the possible use of surrogate markers such as GATA6 and KRT17. I suggest the authors distinguish a little more clearly between what is currently feasible in research settings and what is truly ready for routine clinical practice, as this would make the translational message more precise and realistic.

We have provided additional comments in each paragraph of this section to make it clearer which options require further research and which remain experimental. These parts are marked in red in the manuscript.
We have also added a newparagraph to the Conclusions.

Figures, Tables, and Supplementary Material

The figures and tables are helpful and improve the readability of the manuscript. I suggest the authors carefully review consistency between the main text and the supplementary material, especially study numbering, formatting, and the presentation of outcomes, because a few inconsistencies make cross-checking slightly difficult for the reader.

The adjustments have been made in the main text and supplementary material.

Conclusion

The conclusion is relevant and appropriately highlights the growing interest in molecular profiling of PDAC. I suggest the authors revise the final paragraph slightly so that it more clearly separates prognostic association, predictive potential, and routine clinical applicability. At present, the wording feels a little stronger than the evidence summarized earlier in the review.

We have adjusted the Conclusions section as follows:

“Molecular profiling of PDAC is a developing technique for which there is growing evidence of clinical utility as a prognostic marker. Despite their methodological variability, a consistent pattern emerges across the reviewed studies, with most classification systems converging on a fundamental dichotomy between classical and basal-like tumor lineages. Of the selected studies, the three most commonly used PDAC molecular subtype classifiers were the Moffitt, Bailey, and Collisson classifications, which had the strongest evidence of prognostic utility. However, differences in the technologies and analytical approaches used in the various classifications mean that further research on this topic is required.

The predictive value of molecular subtyping in PDAC remains to be fully described, with clinical trials in progress. Nevertheless, a worse response to chemotherapy has been observed in cases of the basal-like and squamous subtypes. As only five studies on this topic were included in this analysis, the predictive value of PDAC molecular subtyping requires further research.

Both the lack of unified nomenclature and costs hinder the clinical implementation of molecular subtyping into clinical practice. The development of methods for molecular subtyping should focus on the use of easily accessible clinical materials, like FFPE tissues, and reducing the testing time and cost.

Analysis of the material included in this review and the developmental directions of molecular subtyping of PDAC suggests that it is moving towards the use of more readily available materials (such as FFPE tissue samples, including biopsy samples) and analytical methods based on data mining techniques (such as those used in the Moffit classification and the PurIST algorithm). This could reduce the time required to obtain results and the cost of the procedure.

However, due to the limitations mentioned earlier in this article, we would like to emphasize that this review should be considered a narrative review with a structured search strategy. Therefore, the strength of the recommended directions of developing PDAC molecular subtyping and the conclusions is diminished.”

Overall Recommendation

Overall, this manuscript addresses an important topic and brings together a substantial body of literature in a useful way. I suggest the authors revise the manuscript to improve methodological clarity, moderate a few of the broader interpretations, strengthen the synthesis across studies, and carefully edit the language for clarity and consistency. With these revisions, the review could become a valuable contribution to the field.

Comments on the Quality of English Language

On behalf of the entire team, I would like to thank you once again for your advice and recommendations.

I would like to extend our kind regards.
Jakub Wnuk

Reviewer 2 Report

Comments and Suggestions for Authors

Line 29 - space needed between for467,005

Line 30 - space needed between 9thmost commonly

Line 39 - space needed between andPRSS1 (I will let editors pick up these spacing issues from here - there are many).

Line 40 - it would be good to include an indication of approximately what % of PDAC cases are ressectable at the time of diagnosis.

Line 44 - typo - mFOLFIRINOX

Line 229 - "Four of the included studies directly referenced systemic therapy..." but 5 studies are listed in Table 3. Please amend text in line 229.

Figure 3 - many spelling mistakes throughout

The authors provide a sectipon on the Challenges in implementing the molecular subtyping into clinical practice but I feel that there needs to be some clearer recommendations as to hpow to overcome these challenges in a real world setting so as to help patient groups and others advocate for positive systems change with clinical decision makers and policy makers (either in a new section after Challenges or in the Conclusion section).

Comments on the Quality of English Language

English is fine but many spelling mistakes and typos throughout (especially Fig 3 which is almost illegible).

Needs a thorough edit throughout.

Author Response

RESPONSE TO REVIEWERS

Manuscript: Molecular subtypes of pancreatic cancer: a review of the literature

Authors: Jakub Wnuk, Wiktoria Skowron, Anna Długaszek, Joanna Sadurska, Łukasz Pietrzyński, Jacek Kabut, Iwona Gisterek-Grocholska

We would like to thank the reviewers for their valuable and detailed comments, suggestions, and time spent reviewing the manuscript. We believe that after completing the suggested revisions, the overall presentation and clarity of the manuscript have improved.

We look forward hearing from you soon.

Sincerely, on behalf of all authors,

Jakub Wnuk

Comments and Suggestions for Authors

Line 29 - space needed between for 467,005

Line 30 - space needed between 9thmost commonly

Line 39 - space needed between andPRSS1 (I will let editors pick up these spacing issues from here - there are many).

Thank you for pointing this out; we have checked the spacing in the manuscript. Unfortunately, the problem likely stems from different authors editing the text with different software, as this is not the first time I have encountered this issue. I hope that after this round of corrections, this will not happen again.

Line 40 - it would be good to include an indication of approximately what % of PDAC cases are ressectable at the time of diagnosis.

Thank you for this comment. We have added a citation and percentages to the manuscript.

Line 44 - typo - mFOLFIRINOX

Thank you for this advice. We have changed the phrase from “mFOLFIRINOX” to “modified FOLFIRINOX”.

Line 229 - "Four of the included studies directly referenced systemic therapy..." but 5 studies are listed in Table 3. Please amend text in line 229.

We have correctedthismistake.

Figure 3 - many spelling mistakes throughout
The authors provide a sectipon on the Challenges in implementing the molecular subtyping into clinical practice but I feel that there needs to be some clearer recommendations as to hpow to overcome these challenges in a real world setting so as to help patient groups and others advocate for positive systems change with clinical decision makers and policy makers (either in a new section after Challenges or in the Conclusion section).

We have decided to remove Figure 3 from the manuscript. The potential solutions for overcoming the challenges in a real-world setting are now presented in a different paragraph in the Conclusions section, as follows:.

“Molecular profiling of PDAC is a developing technique for which there is growing evidence of clinical utility as a prognostic marker. Despite their methodological variability, a consistent pattern emerges across the reviewed studies, with most classification systems converging on a fundamental dichotomy between classical and basal-like tumor lineages. Of the selected studies, the three most commonly used PDAC molecular subtype classifiers were the Moffitt, Bailey, and Collisson classifications, which had the strongest evidence of prognostic utility. However, differences in the technologies and analytical approaches used in the various classifications mean that further research on this topic is required.

Comments on the Quality of English LanguageEnglish is fine but many spelling mistakes and typos throughout (especially Fig 3 which is almost illegible).Needs a thorough edit throughout

After making the suggested revisions, the manuscript was submitted for English editing by MDPI Author Services.

Once again, thank you for taking the time to write this review and for your advice.

On behalf of the entire team, I would like to extend our kind regards.
Jakub Wnuk