[18F]FDG PET/MRI in Endometrial Cancer: Prospective Evaluation of Preoperative Staging, Molecular Characterization and Prognostic Assessment
, Gabriele Ironi 2, Tommaso Russo 2,3, Giorgio Candotti 4, Federico Fallanca 1, Carlotta Sabini 4, Ana Maria Samanes Gajate 1, Samuele Ghezzo 1, Alice Bergamini 3,4, Miriam Sant’Angelo 5, Luca Bocciolone 4, Giorgio Brembilla 2,3, Paola Scifo 1, GianLuca Taccagni 5, Onofrio Antonio Catalano 6, Giorgia Mangili 4, Massimo Candiani 3,4, Francesco De Cobelli 2,3, Arturo Chiti 1,3, Paola Mapelli 1,3,†and Maria Picchio 1,3,*,†add
Show full author list
Round 1
Reviewer 1 Report
Comments and Suggestions for AuthorsIn the manuscript entitled “[18F]FDG PET/MRI in endometrial cancer: prospective evaluation of preoperative staging, molecular characterization and prognostic assessment” the authors present data from prospective study of 80 patients with endometrial cancer (EC). They found that [18F]FDG PET/MR showed exceptional accuracy in EC primary tumor and LN detection.
This study examined the relationship between clinical factors and PET parameters in EC. This study also examined the relationship between molecular alterations and PET parameters, which is quite interesting. This paper can be accepted only after revision of this paper considering the comments below.
- Is [18F]FDG SUVmean measured within the ROI or the entire tumor?
- Please describe in detail what findings in immunostaining would lead to a diagnosis of p53abn.
- Since Figure 5 presents two cases, how about presenting the p53abn case?
- Is adjuvant therapy being administered according to any guidelines?
- Regarding LN detection, PET was superior to PET/MR. How were positive lymph node metastases determined using PET/MR?
- When diagnosing dMMR and p53abn, it is advisable to state that POLE was not excluded as a limitation.
Author Response
In the manuscript entitled “[18F]FDG PET/MRI in endometrial cancer: prospective evaluation of preoperative staging, molecular characterization and prognostic assessment” the authors present data from prospective study of 80 patients with endometrial cancer (EC). They found that [18F]FDG PET/MR showed exceptional accuracy in EC primary tumor and LN detection.
This study examined the relationship between clinical factors and PET parameters in EC. This study also examined the relationship between molecular alterations and PET parameters, which is quite interesting. This paper can be accepted only after revision of this paper considering the comments below.
Comment 1: Is [18F]FDG SUVmean measured within the ROI or the entire tumor?
Response 1: We thank the reviewer for pointing out this relevant aspect. The SUVmean was measured within the metabolically active tumor volume (MTV), which was defined using a thresholding-based approach at 41% of SUVmax, following the EANM guidelines for tumor imaging. This threshold-based ROI captures the metabolically active portion of the tumor rather than the entire anatomical tumor extent. We have clarified this in the revised manuscript in the Methods section (page 4, lines: 155-156).
Comment 2: Please describe in detail what findings in immunostaining would lead to a diagnosis of p53abn.
Response 2: We thank the reviewer for requesting this clarification. p53abn (p53 abnormal) was defined by immunohistochemistry as either overexpression of p53 in >80% of tumor cells or complete absence of nuclear staining in tumor cells. We have added this information to the Methods section (page 5, lines: 192-193) for clarity.
Comment 3: Since Figure 5 presents two cases, how about presenting the p53abn case?
Response 3: We thank the reviewer for this suggestion. In our cohort, we could not identify two patients with exactly opposite molecular profiles for all the markers investigated. However, we agree with the reviewer that presenting a p53abn case would be more informative. We have therefore revised Figure 5 and its legend to show representative positive and negative cases for each immunohistochemical markers (DNA mismatch repair proteins and p53), rather than focusing on specific patients cases.
Comment 4: Is adjuvant therapy being administered according to any guidelines?
Response 4: Both surgical procedures and adjuvant therapy were indeed performed according to standard clinical guidelines [ESMO-ESGO-ESTRO 2016, Colombo et al.] and we thank the reviewer for the opportunity to specify this aspect in the manuscript (page 5, lines: 175-176).
Comment 5: Regarding LN detection, PET was superior to PET/MR. How were positive lymph node metastases determined using PET/MR?
Response 5: We thank the reviewer for requesting this clarification. For PET/MRI combined assessment, lymph nodes were classified as positive if deemed suspicious on either PET or MRI evaluation. Specifically, a consensus reading was performed after independent evaluation of PET and MRI by Nuclear Medicine physicians and Radiologists, respectively. A lymph node was classified as positive in the combined PET/MRI assessment if it showed either pathological [18F]FDG uptake on PET or morphological criteria of metastasis on MRI. This approach explains why PET/MRI had slightly lower accuracy than PET alone in our cohort: one false positive lesion was identified by MRI (showing suspicious morphological features but negative on PET and confirmed benign at histopathology), which was then included in the combined PET/MRI positive assessment. We have added this clarification to the Methods section “2.3. [18F]FDG PET/MRI analysis” (page 4, lines 147-149).
Comment 6: When diagnosing dMMR and p53abn, it is advisable to state that POLE was not excluded as a limitation.
Response 6: We thank the reviewer for this important observation. We acknowledge that POLE mutational status was not assessed in our cohort (study started in 2018), which represents a limitation in achieving complete molecular classification according to FIGO 2023 guidelines. However, we estimate that the impact is likely limited, potentially affecting around 8% of all patients (ref: Alaa Salah Jumaah et al., 2020). We have added this limitation to the Discussion section, explicitly acknowledging that future studies incorporating comprehensive molecular profiling would be valuable to fully characterize the relationship between imaging features and all molecular subtypes as defined by FIGO 2023 (page 14, lines: 593-599).
Reviewer 2 Report
Comments and Suggestions for AuthorsThe manuscript addresses a clinically relevant topic, integrates advanced multiparametric imaging, and aligns with the updated FIGO 2023 molecular framework. Several methodological and interpretative issues should be addressed to strengthen the robustness of the conclusions, particularly regarding prognostic and molecular predictions.
Major comments
1. Risk of overfitting and exploratory nature of prognostic analyses
The study evaluates a very large number of imaging-derived parameters against multiple clinical, histological, molecular, and survival outcomes, while the number of events is limited (e.g., only 9 recurrences and 13 LN-positive cases). Although Benjamini–Hochberg correction is applied, the risk of false-positive findings and overfitting remains substantial, particularly for ROC-derived cut-offs and survival analyses.
The authors should explicitly state that prognostic and molecular association analyses are exploratory and hypothesis-generating and emphasize the lack of external validation and the need for independent cohorts.
2. ROC-derived cut-offs and Kaplan–Meier analysis on the same cohort
Cut-off values are derived using ROC analysis and then applied to Kaplan–Meier survival analysis within the same cohort. While this approach is common, it introduces optimism bias and may overestimate prognostic performance. The authors should avoid implying clinical applicability of these thresholds without validation and consider rephrasing conclusions to indicate association rather than prediction.
3. Cox regression analysis lacks clarity
The manuscript states that Cox proportional hazards models were applied, but it is unclear which variables were included. The authors should clarify whether Cox analyses were univariate or multivariate and also justify variable selection or limit analyses to univariate models only.
4. Comparison between PET, MRI, and PET/MRI for lymph nodes
Abdominal lymph nodes were excluded from MRI analysis due to limited field of view, yet diagnostic performance of PET, MRI, and PET/MRI is compared. This creates unequal conditions between modalities.
The authors should: clearly distinguish between patient-level and region-based analyses, explicitly state that MRI performance applies only to pelvic lymph nodes, avoid overinterpreting direct modality comparisons when anatomical coverage differs.
5. Molecular classification is incomplete (absence of POLE)
The study focuses on MMRd and p53abn status in line with FIGO 2023, but POLE-mutated tumors are not included.
The authors should: explicitly acknowledge the absence of POLE sequencing, discuss how this limits full molecular classification according to FIGO 2023, clarify whether any cases could potentially belong to the POLE subgroup.
Minor comments
1. Please carefully review all units in the text, figures, and supplementary table.
2. Minor grammatical and stylistic issues should be revised throughout the manuscript.
Author Response
The manuscript addresses a clinically relevant topic, integrates advanced multiparametric imaging, and aligns with the updated FIGO 2023 molecular framework. Several methodological and interpretative issues should be addressed to strengthen the robustness of the conclusions, particularly regarding prognostic and molecular predictions.
Major comments
Comment 1. Risk of overfitting and exploratory nature of prognostic analyses
The study evaluates a very large number of imaging-derived parameters against multiple clinical, histological, molecular, and survival outcomes, while the number of events is limited (e.g., only 9 recurrences and 13 LN-positive cases). Although Benjamini–Hochberg correction is applied, the risk of false-positive findings and overfitting remains substantial, particularly for ROC-derived cut-offs and survival analyses.
The authors should explicitly state that prognostic and molecular association analyses are exploratory and hypothesis-generating and emphasize the lack of external validation and the need for independent cohorts.
Response 1: We appreciate the reviewer's critical observation regarding the exploratory nature of our prognostic analyses. We fully agree that given the limited number of events relative to the large number of imaging parameters evaluated, our findings should be interpreted with appropriate caution. We have revised the manuscript by: 1) rephrasing our discussion of survival findings to emphasize "associations" rather than "predictive utility" (page 13, lines: 581-582), 2) explicitly stating in the limitations statement that our sample showed limited number of events other than class imbalances (page 13, line 590), and 3) explicitly stating that the prognostic and molecular association analyses are exploratory and hypothesis-generating (pages 13, lines: 596-599). We have also softened our Conclusions accordingly.
Comment 2. ROC-derived cut-offs and Kaplan–Meier analysis on the same cohort
Cut-off values are derived using ROC analysis and then applied to Kaplan–Meier survival analysis within the same cohort. While this approach is common, it introduces optimism bias and may overestimate prognostic performance. The authors should avoid implying clinical applicability of these thresholds without validation and consider rephrasing conclusions to indicate association rather than prediction.
Response 2: We thank the reviewer for this important methodological consideration. We acknowledge that deriving cut-off values using ROC analysis and applying them to survival analysis within the same cohort may introduce optimism bias and overestimate prognostic performance. We have revised the manuscript to address this concern by: 1) rephrasing our discussion of survival findings to emphasize "associations" rather than "predictive utility" (page 13, lines: 581-582), 2) explicitly stating in the Discussion that these cut-off values require external validation before clinical application, adding to the Limitations section a statement acknowledging the potential for optimism bias inherent in this analytical approach (pages 13, lines: 593-599). We have also emphasized in the Conclusions that validation in independent cohorts is needed.
Comment 3. Cox regression analysis lacks clarity
The manuscript states that Cox proportional hazards models were applied, but it is unclear which variables were included. The authors should clarify whether Cox analyses were univariate or multivariate and also justify variable selection or limit analyses to univariate models only.
Response 3: We thank the reviewer for this observation. We have clarified in the Statistical Analysis section that univariate Cox proportional hazards models were used to quantify relapse risk (page 5, line 209; page 9, line 470).
Comment 4. Comparison between PET, MRI, and PET/MRI for lymph nodes
Abdominal lymph nodes were excluded from MRI analysis due to limited field of view, yet diagnostic performance of PET, MRI, and PET/MRI is compared. This creates unequal conditions between modalities.
The authors should: clearly distinguish between patient-level and region-based analyses, explicitly state that MRI performance applies only to pelvic lymph nodes, avoid overinterpreting direct modality comparisons when anatomical coverage differs.
Response 4: We appreciate the reviewer's important observation regarding the unequal anatomical coverage between modalities. We acknowledge that this limitation should be stated more clearly in our manuscript. We have revised the Results section to explicitly indicate that MRI performance metrics at the patient level reflect assessment of pelvic lymph nodes only, as abdominal lymph nodes were outside the MRI field of view. Specifically, two patients with abdominal lymph node metastases were correctly detected by PET but could not be assessed by MRI due to the limited field of view, contributing to the lower sensitivity of MRI compared to PET (page 7, lines: 374-379). We have also added a statement in the Discussion section acknowledging that direct comparison between PET and MRI should be interpreted with caution due to the difference in anatomical coverage (page 13, lines: 586-590).
Comment 5. Molecular classification is incomplete (absence of POLE)
The study focuses on MMRd and p53abn status in line with FIGO 2023, but POLE-mutated tumors are not included.
The authors should: explicitly acknowledge the absence of POLE sequencing, discuss how this limits full molecular classification according to FIGO 2023, clarify whether any cases could potentially belong to the POLE subgroup.
Response 5: We thank the reviewer for this important observation. We acknowledge that POLE mutational status was not assessed in our cohort (study started in 2018), which represents a limitation in achieving complete molecular classification according to FIGO 2023 guidelines. However, we estimate that the impact is likely limited, potentially affecting around 8% of all patients (ref: Alaa Salah Jumaah et al., 2020). We have added this limitation to the Discussion section, explicitly acknowledging that future studies incorporating comprehensive molecular profiling would be valuable to fully characterize the relationship between imaging features and all molecular subtypes as defined by FIGO 2023 (page 14, lines: 593-599).
Minor comments
Comment 1. Please carefully review all units in the text, figures, and supplementary table.
Response 1: We thank the reviewer for this suggestion. We have carefully reviewed all units throughout the manuscript, including the main text, figures, and supplementary materials.
Comment 2. Minor grammatical and stylistic issues should be revised throughout the manuscript.
Response 2: We thank the reviewer for this suggestion. We have thoroughly revised the manuscript to address grammatical and stylistic issues throughout the text.
Round 2
Reviewer 1 Report
Comments and Suggestions for AuthorsThe manuscript has been revised well and is in a nice condition now.
Reviewer 2 Report
Comments and Suggestions for AuthorsI believe that the Authors have sufficiently improved the manuscript in response to the comments.
