Serum Levels of CA125 and HE4 as a Tool for Predicting Regional Lymph Node Metastatic Involvement in Endometrial Carcinoma

Round 1

Reviewer 1 Report

Comments and Suggestions for Authors

Introduction: Can you better explain your rationale for use of HE4 and CA-125 to predict regional lymph node involvement in endometrial cancer? This was not explained in depth in the introduction. Is this in place of getting a pre-operative CT to evaluate the nodes? Or in conjunction? There are some papers published in the last 5 years looking at CA-125 and/or HE4 to predict lymph node metastases, but these are not referenced in the introduction. Consider adding some of these references. It appears some of these were included in the discussion. Would consider including information from studies summarized in table 4 in the introduction.

Methods: Is it standard of care in your institution to collect CA-125 and HE4 on the day of D&C or hysteroscopy for all patients regardless of their pathology? If it was collected on the same day of the diagnostic procedure, that implies that the tumor markers were drawn prior to knowledge of an endometrial cancer diagnosis. Do you have any issues with insurance coverage obtaining these tumor markers without a cancer diagnosis? This manuscript describes a retrospective approach, which implies that CA-125 and HE4 collection are standard of care for your practice at the time of all D&Cs or hysteroscopies.
What is involved in your pre-op work-up? Do patients get a CT prior to surgery?

Results: what is the rational for the sentence in lines 175-178? Since ITCs are not considered metastatic involvement, what is the rationale for this statement?

Tables and figures are clear and show the data and key findings well.

LNM prediction model: It is a bit confusing to introduce ultrasound evaluation of depth of invasion this far into the paper. Depth of myometrial invasion was not mentioned earlier. Why is this added at the end of the results section. Was this the primary goal of the study to create a prediction model? If so, it would be helpful to mention that earlier in the paper. Further, depth of myometrial invasion is known to correlate with risk of lymph node metastasis as are high risk histology, LVSI, and cervical involvement.

Discussion: would condense the discussion. It is very dense. The paragraphs are quite long and make it difficult to read. Would make this more concise

Discussion Lines 347-350: Are you implying that the CA-125 and HE4 could be used INSTEAD of MRI and/or PET/CT for evaluating lymph node involvement prior to surgery? This would be a big leap based on retrospective data and a relatively small sample size.

Rationale for the study is a bit confusing. Are you proposing to omit lymph node staging in patients who are "low risk" according to your proposed LNM prediction model? In the era of SLN mapping, does it truly reduce surgical risk (operative time, morbidity, etc) to omit SLN mapping and sampling?

Author Response

Thank you very much for taking the time to review this manuscript. Please find the detailed responses in the the attachment and the corresponding revisions/corrections highlighted in the re-submitted files. 

Author Response File: Author Response.pdf

Reviewer 2 Report

Comments and Suggestions for Authors

The Crha et al., manuscript provides valuable data on CA125 and HE4 as predictive markers for lymph node metastasis in endometrial cancer, enhanced by ultrasound staging. The findings are encouraging but would benefit from a more critical discussion of existing evidence, clearer acknowledgment of study limitations, and practical guidance for clinical adoption and future validation. With these improvements, your work could offer significant contributions to the field.

Introduction

1. The introduction should more explicitly state the specific clinical challenge, why current imaging techniques are insufficient (for example: limited sensitivity in micrometastasis detection, variable availability of advanced imaging, etc.), highlight the real-world consequences (e.g., overtreatment, surgical morbidity) and the potential benefits (clinical and cost) of introducing serum biomarkers.
2. The authors are recommended to add 3–4 sentences summarizing prior studies (with references), emphasizing where prior evidence has been inconsistent or limited, and highlighting the uniqueness of your study/model (e.g., To date, only a limited number of studies have combined CA125, HE4, and imaging findings for preoperative EC nodal risk stratification, and few have assessed their utility in a real-world cohort of this size).
3. The rationale for CA125 and HE4 selection was not clear. While the authors mention their established value in ovarian cancer, their specific justification for use in EC LNM prediction is underdeveloped. Please add a short explanation of the biological plausibility for these markers to reflect nodal involvement in EC (e.g., HE4 expression correlates with tumor aggressiveness or invasiveness; CA125 is associated with serous and more aggressive histotypes), considering citing evidence of their performance in endometrial, not only ovarian, tumors.
4. The introduction could be strengthened by stating how a reliable blood-based test for LNM risk would potentially change standard of care (e.g., reduce patient exposure to unnecessary nodal dissection, especially in comorbid populations or where imaging is equivocal).
5. The stated aim is broad; please strengthen it by making explicit what you will do, for example “This study aims to assess the individual and combined predictive ability of CA125 and HE4, alongside ultrasound-based uterine invasion assessment, for identifying regional lymph node metastasis in patients with endometrial carcinoma.”
6. The authors are recommended to end the introduction with a clear sentence describing how the results could lead to improved patient care, or serve as the basis for future prospective validation and biomarker implementation studies.

Methods

• Line 76: As the diagnostic methods (biopsy, D&C or hysteroscopy) could influence marker levels, the authors should state if marker levels varied by diagnostic method or reference any studies that show none/limited effect.
• Line 83: provide the citation of FIGO in this context.
• Lines 78/79: The authors should explicitly state and specify that the blood sampling was immediately prior to any intervention to reinforce rigor.
• Lines 87-89: Types of lymph node staging procedures are listed but not justified. Plaese briefly rationalize the choice/criteria for each type of lymph node staging. Were they at surgeon’s discretion?
• Lines 90-92: The inclusion criteria are a bit vague (esp. “mixed histotype…with endometrioid”). The authors should specify whether molecular classifiers (e.g., ProMisE, p53 status) or only standard histotype used; clarify if ambiguous histology was excluded.
• Lines 93/94: Please define specific cut-offs (eGFR threshold for “renal insufficiency”) and how exclusion diagnoses were ascertained (self-report/history/labs?).
• Lines 101-104: The authors should specify fasting state, time-of-day, and storage duration/temperature if possible. Were measurements batch or real-time?
• Also, the minimum detection level of each assay, the intra-and inter coefficient of variations of each assay, and the reference ranges for two markers for this analyzer should be stated as part of the quality control measures (to meet the standards of publication).
• Lines 110-112: the authors should clarify SNB protocols; specify thresholds for resection or conversion to full LND upon SNB failure.
• Lines 112-116: The authors should state whether pathologists were blinded to biomarker values; clarify if ultrastaging was for all SNBs.
• Line 118: Definitions of micrometastasis/ITC referenced (size), but threshold literature should be cited.
• - Lines 122/123: Missing data management were not described. Please explicitly state approach for missing data (imputation, exclusion, etc.).
• Lines 124-127: Did the authors apply any type of multiple comparisons correction?
• Lines 126/127: Only “highest sum sensitivity and specificity” for cut-offs; other metrics? Please consider reporting AUC, PPV/NPV, confidence intervals for all model performance parameters.
• Lines 130/131: Machine learning/decision tree method not sufficiently detailed (no cross-validation info). The authors state how decision trees were tuned, whether cross-validation or training/test split was used.
• Line 133: No mention of sample size/power calculation. Please briefly state if a power analysis was performed before study initiation.
• No statement of ethical approval or informed consent was identified in this section.

Results

• Lines 160-165: Median and IQR/means given for CA125 and HE4 by group, but no mention of multivariable analysis or adjustment for confounders.
• Lines 170-174: The treatment and statistical handling of “ITCs” (as ‘no LNM’ for most analyses) should be justified more explicitly, as it is a clinical gray area.
• Lines 193/194: Results highlight only univariate model performance. The authors should explicitly state the rationale for the chosen cut-offs (Youden’s index, clinical convention, etc.) and move swiftly to multivariable/combined analyses.
• Lines 196-224: No evidence of overfitting or internal validation (e.g., cross-validation, bootstrapping) is described. Please state how the model was validated internally (or acknowledge as a limitation if not done).

 

Table footers and Figure legends

• The type of data presentation, the statistical significance levels and the meaning of bold values should be added to be self-reported (no need to return to the related text).

Discussion

• Generally, it is recommended that the authors divide this one section into paragraphs, each one covers one theme to facilitate following the authors in their elaboration.
• Please add a brief note on why CA125/HE4 rise with LNM, are they active in tumoral stroma/vasculature/etc.?
• Lines 237/238: “Both CA125 and HE4 typically exhibit a broad range of values in the case of EC. This lacks interpretive context. Please, briefly discuss why the range complicates clinical use, and how your model addresses this limitation.
• Lines 297-304: International comparison is helpful but lacks discussion of limitations of direct comparison. Acknowledge regional, design, and histopathology differences between published cohorts (impact on generalizability).
• Lines 319-326: Practical clinical use of your proposed model is not discussed in detail. Specify what is needed for clinical deployment (e.g., training, validation, rapid assays, standardization of imaging).
• Lines 349-351: Suggest molecular classification integration, nomogram development, or machine learning approaches for future research.
• Lines 352-355: “limitations are scattered. Please summarize all main limitations in a single dedicated paragraph: retrospective design, single center, no external validation, etc.
• To be clinically oriented, add a brief “clinical recommendations” sentence or two, laying out how (or if) this algorithm can be immediately incorporated, or what is required for such adoption.
• Also, it is recommended to highlight the practical cost/benefit (or resource) implications for systems with varying access to imaging modalities.
• For future perspectives. The authors can suggest integration with genomic classifiers (e.g., ProMisE), validation in low-resource/low-imaging environments, or AI-based approaches for future work.

Author Response

Thank you very much for taking the time to review this manuscript. Please find the detailed responses in the in the attachment file and corresponding revisions/corrections highlighted in the re-submitted manuscript.

 

 

Author Response File: Author Response.pdf

Round 2

Reviewer 1 Report

Comments and Suggestions for Authors

Thank you for your revisions which have significantly improved your manuscript. I have additional suggestions for improvement.

Lines 66-69: Run-on sentence that is grammatically incorrect and has punctuation errors. Please revise.

Lines 87-91: The introduction states that the model could be used for planning extent of surgical treatment in cases where imaging is equivocal and that low-risk LNM group would not require surgical lymph node staging. Please reword this to address the role of sentinel lymph node mapping and sampling. Are you insinuating that your model would entirely replace sentinel lymph node mapping and sampling? Or that your model would be used to determine if complete lymphadenectomy is indicated in cases where sentinel lymph nodes fail to map?

Line 240 and Line 244: would relabel x-axis of Figures 1 and 2 as "Lymph Node Involvement" rather than "Metastatic Disability". Metastatic disability refers to functional limitations and health impairments related to cancer.

Lines 402-405: Upon reviewing reference 5, the ESGO/ESTRO/ESP guidelines state as follows: "Of note, preoperative ultrasound assessment of deep myometrial and cervical stromal invasion in endometrial carcinoma is best performed by an expert sonographer as, compared with gynecologists, they show a greater degree of agreement with histopathology and greater interobserver reproducibility." Would recommend including reference to the fact that ultrasound should be performed by an expert sonographer as ultrasound is not comparable to MRI in the absence of an expert sonographer. Please also include criteria for expert sonographer.

Lines 448-451: I continue to question the author's conclusions that the model could be used to stratify patients into low and high risk LNM groups to reduce extent of surgery. Would recommend rewording this to make it clear that sentinel lymph node mapping and sampling should be performed and that this model might be validated to be used to determine whether complete lymphadenectomy is warranted or could be omitted in patients who fail to map.

Author Response

Please see the attachment for the response. Thank you very much for you recommendations.

Best regards,

Tomas Crha

Author Response File: Author Response.pdf

Reviewer 2 Report

Comments and Suggestions for Authors

Thanks to the authors for addressing all the raised concerns.

Best

Author Response

Thank you very much for your recommendations.

Best regards,

Tomas Crha