Radiomic Analysis of Treatment Effect for Patients with Radiation Necrosis Treated with Pentoxifylline and Vitamin E
Round 1
Reviewer 1 Report
Comments and Suggestions for Authors
Comments on Manuscript ID tomography-3041306
Title Radiomic Analysis of Treatment Effect for Patients with Radiation Necrosis treated with Pentoxifylline and Vitamin E
Authors Jimmy S Patel et al
Comments:
This is interesting study to analyze the impact of Pentoxifylline and Vitamin E treatment on radiotherapy/SRS RE. The authors need to address following questions:
1. The data showed the impact of Pentoxifylline and Vitamin E in Post MRI induced radiation necrosis the Pentoxifylline and Vitamin E.
2. In the result section, the authors stated that 48 Patients identified who developed RN after radiation therapy, The authors need to provide the total number of patients included in this radiation therapy study group and what was percentile of radiation necrosis developed patients (48 count).
3. Is it possible for authors to provide any evidence study that the patient treated with Ptx + VitE prior to start radiotherapy/SRS and potential inhibitory role for mimic RN impact.
4. The results table 1 and 2 should be in the format of Bar Diagram having mean ± SD value instead of Tabulations.
5. The author needs to provide the data of any surgical intervention in the selected group patients for this study.
6. Since the Pentoxifylline is vasoactive agent having impact of cardio-vascular and RBC and blood circulatory system, the authors should provide the risk factor analysis outcomes of this study.
7. How the author would suggest including/ excluding the diabetic patients for such study regime for potential CNS and neurotoxicity interventions such as nerve fibers damaging
Author Response
Within attached
Author Response File: Author Response.pdf
Reviewer 2 Report
Comments and Suggestions for Authors
This manuscript tries to cover two claims: the use of pentoxifylline (Ptx) and vitamin E (VitE) for RN and a machine learning method to determine who will respond to treatment. A major challenge here is that the latter is only valuable if the former is proved to be an adequate treatment. We are presented here with what is basically a case report rather than a true evaluation of the treatment which limits the value of the manuscript and the value of the radiomics. Ideally the radiomic model would be applicable to any treatment of RN.
1. This study is not set up to properly evaluate the effectiveness of the Ptx + VitE treatment as it lacks proper controls. Namely, there should be a group of patients receiving dexamethasone only as that is the closest to a standard of care that we have for RN.
2. Line 98: “400 mg BID of Ptx + 1000 IU qd of VitE.” Not all the readers will have the clinical background to understand the BID and qd acronyms. Please use “twice a day” and “once a day”
3. Line 141 to 144 state that only the post-therapy scans are used for the radiomics model. Was this scan performed at the same time as the assessment of response to therapy? The response assessment subsection on lines 104 to 109 does not mention any timing of when the assessment is performed. I finally found this on the results section in Line 201 and table 4 which suggests that in at least one case the evaluation was made within the timeline of the scan used for radiomics. Also, this timeline for assessment is horribly inconsistent. The likelihood of progression is way different at 0.66 months versus 12.68 months.
4. You identified 48 patients per line 156 but only 43 were used for radiomics per line 120. Why were the 5 patients excluded?
5. If the focus is on SRS then why did you include the patients treated with conventional radiotherapy? When looking at Table 1, inclusion of these three patients makes the parametric statistics (mean and standard deviation) meaningless because the distribution is likely bimodal rather than normal. In general, the inclusion of these three patients makes it impossible to determine the distribution of doses for the SRS cases from the data presented in Table 1.
6. Since you have multiple fractionation schemes and since radiation dose does not add up linearly, you should be including either BED or 2-Gy-Equivalent dose in Table 1. I would also want to see lesion size as that may also be linked to risk of RN since a larger lesion leads to more normal tissue receiving a larger dose.
7. You state that of 25 patients that started on Ptx + VitE alone, 11 required dexamethasone to be added (Line 179-180). There is a group of 7 that starts Ptx + VitE after dexamethasone (Line 184). Was this also because dexamethasone was not sufficient and if so do we know in what percentage of cases dexamethasone is not sufficient? Effectively, is the percentage of patients that respond to Ptx + VitE alone higher or lower than dexamethasone alone?
8. For the groups that combined Ptx + VitE and dexamethasone but did not start them simultaneously, was the assessment of response done in comparison to the original lesion or to the lesion at the time the second therapy was implemented?
9. Line 202 to 207 focuses exclusively on the monotherapy group which is flawed and should be remove. Going back to the point above, the monotherapy group excludes the 11 patients where it was deemed that the monotherapy failed and dexamethasone needed to be added. If anything, I am surprised that the percent of patients that were stable or showing improvement was not 100% as the failures should have been part of the group that had dexamethasone added. Also, because this section comes right before discussing Table 5, it can create confusion that Table 5 only focuses on that group when it focuses on all 48 patients.
10. Lines 211-214 and Table 6. P=0.064 is not significant. These lines and table can be removed instead of trying to interpret this as being meaningful. You don’t have the statistical power to make any claims here.
11. Given the small sample size and the fact that the ROC curves without the wavelet filters are no better than chance, it looks like this radiomic model is not useful.
12. Lines 245 to 250 speak of dosimetric factors that relate to RN risk, namely lesion size and max hot spot dose. If you know these are important why where they not included in your analysis?
13. Lines 238 to 272 are a great introduction not a discussion.
Author Response
within attached
Author Response File: Author Response.pdf
Reviewer 3 Report
Comments and Suggestions for Authors
It's an interesting and well written manuscript. I have several comments/questions.
- In the introduction: a clearer statement of the study's objectives should be found at the end of the introduction.
- In the MM: More information on the criteria for defining treatment response would enhance understanding.
- In the results: The section on radiomics is detailed, but the inclusion of supplementary material could be better organized to avoid overwhelming the reader with data. Table 5 in particular could be simplified. Table 4 can be presented as a diagram or as figures.
- The rationale for choosing specific radiomic features and preprocessing steps should be more explicitly justified, with several references in the litterature.
- Radiomics Model: The development and performance of the SVM model are well-documented, but the clinical relevance of the AUC value (0.69) needs further discussion. And more clarity on how radiomic features translate into clinical decision-making would be helpful.
- The conclusion could briefly touch on future research directions.
Author Response
Within attached
Author Response File: Author Response.pdf
Reviewer 4 Report
Comments and Suggestions for Authors
Lisa J. Sudmeier et al. reported an interesting work about utilizing Ptx + VitE to treat radiation necrosis. The topic was of a certain significance nowadays, and would arouse a certain impact in its field. Overall, the submission fell within the scope of Tomography, and could be considered for publication after a Minor Revision. Detailed comments:
1) Please reconsider the abbreviation for pentoxifylline. Ptx might be commonly short for paclitaxel.
2) The significance of this study should be stated at the end of Introduction.
3) The demographic data of the included patients must be provided.
4) The synergistic effects of Ptx plus VitE should be discussed from a pharmacological perspective, in Section 4. In other words, why consider to combine Ptx VitE and are there any evidence from basic studies?
5) The format of References should be unified.
Author Response
within attached
Author Response File: Author Response.pdf
Reviewer 5 Report
Comments and Suggestions for Authors
The primary objectives of the research were to evaluate the efficacy of pentoxifylline (Ptx) combined with vitamin E (VitE) in treating radiation necrosis (RN) in patients who have undergone stereotactic radiosurgery (SRS), to analyze radiomic features from MRI images before and after treatment to identify predictive features for treatment response, and to determine the safety and tolerability of Ptx + VitE in these patients.
To achieve these objectives, the researchers conducted a retrospective review of patient records from the Emory Department of Radiation Oncology database. They identified patients treated with Ptx + VitE for RN and analyzed their MRI images before and after treatment. Radiomic features were extracted from these images using software tools such as Velocity and 3D Slicer. The features included first-order statistics, shape-based features, and various matrix-based features. These features were then analyzed to develop a predictive model using Support Vector Machine (SVM), and the model’s performance was assessed with the receiver operating characteristic (ROC) curve. Additionally, statistical analyses were performed using descriptive statistics, chi-squared tests, Fisher’s exact tests, and ANOVA to compare treatment responses across patient characteristics. A multivariable logistic regression model was used to evaluate the interaction of different variables with treatment response.
The research identified several limitations. The small sample size of 48 patients may affect the generalizability and statistical power of the findings. The retrospective nature of the study introduces biases inherent in patient record reviews and lacks the rigor of a prospective randomized trial. Furthermore, the study highlights the need for randomized controlled trials to definitively assess the efficacy of Ptx + VitE and to validate the predictive radiomic models developed.
To enhance the quality and impact of the research, the researchers recommend conducting prospective randomized controlled trials, increasing the sample size and including a more diverse patient population, and standardizing imaging protocols and treatment regimens across different institutions to reduce variability and enhance the reliability of the radiomic analysis.
Several language corrections were noted in the original document. For example, "RN is usually diagnosed radiographically although the appearance of RN versus tumor recurrence has significant overlap on standard MRIs" should be corrected to "Radiation necrosis (RN) is typically diagnosed radiographically; however, distinguishing RN from tumor recurrence on standard MRIs presents significant challenges due to overlapping imaging features." Another correction is changing "MRI has been used for visual assessment of tumor response to therapy" to "MRI is commonly used for the visual assessment of tumor response to therapy." Lastly, "Our results align with other studies that demonstrate the effectiveness of preprocessing filters in improving machine learning model performance" should be corrected to "Our results are consistent with other studies demonstrating that preprocessing filters enhance the performance of machine learning models."
In conclusion, your dedication to exploring innovative treatment options for radiation necrosis is commendable. The integration of radiomic analysis with traditional treatment approaches holds great promise for improving patient outcomes. Your efforts in addressing this challenging condition contribute significantly to the field of radiation oncology. Keep pushing forward with your research, and know that your work is making a meaningful difference in the lives of patients.
Comments on the Quality of English Language
The primary objectives of the research were to evaluate the efficacy of pentoxifylline (Ptx) combined with vitamin E (VitE) in treating radiation necrosis (RN) in patients who have undergone stereotactic radiosurgery (SRS), to analyze radiomic features from MRI images before and after treatment to identify predictive features for treatment response, and to determine the safety and tolerability of Ptx + VitE in these patients.
To achieve these objectives, the researchers conducted a retrospective review of patient records from the Emory Department of Radiation Oncology database. They identified patients treated with Ptx + VitE for RN and analyzed their MRI images before and after treatment. Radiomic features were extracted from these images using software tools such as Velocity and 3D Slicer. The features included first-order statistics, shape-based features, and various matrix-based features. These features were then analyzed to develop a predictive model using Support Vector Machine (SVM), and the model’s performance was assessed with the receiver operating characteristic (ROC) curve. Additionally, statistical analyses were performed using descriptive statistics, chi-squared tests, Fisher’s exact tests, and ANOVA to compare treatment responses across patient characteristics. A multivariable logistic regression model was used to evaluate the interaction of different variables with treatment response.
The research identified several limitations. The small sample size of 48 patients may affect the generalizability and statistical power of the findings. The retrospective nature of the study introduces biases inherent in patient record reviews and lacks the rigor of a prospective randomized trial. Furthermore, the study highlights the need for randomized controlled trials to definitively assess the efficacy of Ptx + VitE and to validate the predictive radiomic models developed.
To enhance the quality and impact of the research, the researchers recommend conducting prospective randomized controlled trials, increasing the sample size and including a more diverse patient population, and standardizing imaging protocols and treatment regimens across different institutions to reduce variability and enhance the reliability of the radiomic analysis.
Several language corrections were noted in the original document. For example, "RN is usually diagnosed radiographically although the appearance of RN versus tumor recurrence has significant overlap on standard MRIs" should be corrected to "Radiation necrosis (RN) is typically diagnosed radiographically; however, distinguishing RN from tumor recurrence on standard MRIs presents significant challenges due to overlapping imaging features." Another correction is changing "MRI has been used for visual assessment of tumor response to therapy" to "MRI is commonly used for the visual assessment of tumor response to therapy." Lastly, "Our results align with other studies that demonstrate the effectiveness of preprocessing filters in improving machine learning model performance" should be corrected to "Our results are consistent with other studies demonstrating that preprocessing filters enhance the performance of machine learning models."
In conclusion, your dedication to exploring innovative treatment options for radiation necrosis is commendable. The integration of radiomic analysis with traditional treatment approaches holds great promise for improving patient outcomes. Your efforts in addressing this challenging condition contribute significantly to the field of radiation oncology. Keep pushing forward with your research, and know that your work is making a meaningful difference in the lives of patients.
Author Response
within attached
Author Response File: Author Response.pdf
Round 2
Reviewer 2 Report
Comments and Suggestions for Authors
My comments remain the same
Author Response
Please see attached.
Author Response File: Author Response.pdf
Reviewer 5 Report
Comments and Suggestions for Authors
I've carefully reviewed this paper as a seasoned years veteran in the medical field. The study's main objective was to evaluate the effects of combining pentoxifylline and vitamin E in treating radiation necrosis patients, and to explore whether radiomic features from MRI scans could help predict treatment response.
The researchers used a retrospective cohort design, relying on radiation oncologists' subjective assessments of imaging changes to analyze treatment response. They also developed a machine learning-based radiomic model to try and forecast how patients would fare on this therapy.
While this is a valuable effort, the study does have some limitations, like the relatively small patient sample size and its single-institution nature. Still, I have a few suggestions that could strengthen the work:
First, validating the pentoxifylline-vitamin E combination in a larger, multi-center prospective study would help confirm its efficacy. Second, the radiomic response prediction model should be further validated to ensure its reliability. And third, delving deeper into the underlying mechanisms of radiation necrosis and how this treatment might work could provide important insights.
In terms of the writing, there are a few areas that could be tightened up. For instance, the phrasing "was used to treat" could be updated to "has been used to treat" for better flow. And phrases like "was capable of predicting" might read more naturally as "was able to predict."
Overall though, I think this is a worthwhile study that adds to our understanding of radiation necrosis management. The authors are to be commended for exploring this alternative therapeutic approach and leveraging advanced imaging analysis. I'm hopeful that with continued research, they'll be able to further refine the treatment and develop more reliable predictive tools to benefit patients. Please keep up the great work!
Comments on the Quality of English Language
I've carefully reviewed this paper as a seasoned years veteran in the medical field. The study's main objective was to evaluate the effects of combining pentoxifylline and vitamin E in treating radiation necrosis patients, and to explore whether radiomic features from MRI scans could help predict treatment response.
The researchers used a retrospective cohort design, relying on radiation oncologists' subjective assessments of imaging changes to analyze treatment response. They also developed a machine learning-based radiomic model to try and forecast how patients would fare on this therapy.
While this is a valuable effort, the study does have some limitations, like the relatively small patient sample size and its single-institution nature. Still, I have a few suggestions that could strengthen the work:
First, validating the pentoxifylline-vitamin E combination in a larger, multi-center prospective study would help confirm its efficacy. Second, the radiomic response prediction model should be further validated to ensure its reliability. And third, delving deeper into the underlying mechanisms of radiation necrosis and how this treatment might work could provide important insights.
In terms of the writing, there are a few areas that could be tightened up. For instance, the phrasing "was used to treat" could be updated to "has been used to treat" for better flow. And phrases like "was capable of predicting" might read more naturally as "was able to predict."
Overall though, I think this is a worthwhile study that adds to our understanding of radiation necrosis management. The authors are to be commended for exploring this alternative therapeutic approach and leveraging advanced imaging analysis. I'm hopeful that with continued research, they'll be able to further refine the treatment and develop more reliable predictive tools to benefit patients. Please keep up the great work!
Author Response
Please see attached
Author Response File: Author Response.pdf