The Influence of Additional Treatments on the Survival of Patients Undergoing Transarterial Radioembolization (TARE)

Round 1

Reviewer 1 Report

Comments and Suggestions for Authors

In this article, Quartuccio and colleagues present a single-center retrospective database analysis of patients with hepatocellular carcinoma (HCC) and colorectal cancer with liver metastases (CLM) receiving transarterial radioembolization (TARE).

Overall, the article would require significant revisions to its methodology and presentation, before it can be considered acceptable for publication. Please find my comments for each section below.

Introduction:
1) Informative and very well-structured, albeit lengthy. Some of the information could be included in the Discussion instead, which is rather short.

2) Citations need to be revised - for example:

i) "There is expanding evidence that TARE provides a longer time to progression compared to transarterial chemoembolization (TACE) in patients with HCC" uses the EPOCH trial as a reference, while the EPOCH trial only provides data on CLM patients. 

ii) "the Resin trial demonstrated in a multicentre cohort of 498 patients that TARE is an effective second-line treatment for patients with CRC liver metastases" - uses an editorial commenting on the article, rather than original article of the RESIN registry analysis (PMID 35762890)

Methods:

1) Detailed, comprehensive presentation of all selection criteria and procedures involved

2) The statistics section is severely lacking:

i) There is no description of the statistical tests used for OS comparison or the units of measurement.

ii) The use of p-value needs to be clarified i.e. state that a p-value of <0.05 was used to denote statistical significance rather than mentioning "(p<0.05)" and if this was one-tailed or two-tailed.

iii) The use of simple linear regression when analyzing survival data is inappropriate, as it assumes complete information for all data, which is not the case as censoring is used. The correct approach would be a Cox proportional hazard regression analysis. Due to the potential confounding effect of the variables analyzed, these should be adjusted in the main analysis comparing survival between HCC and CLM patients.

Results:

1) The amount of information is inadequate; there needs to be a table with all the baseline characteristics of each group separately, and all potential confounders: age, sex, ECOG PS, Child-Pugh score, dose received, other second/third-line treatments, follow-up, mean radiation dose received, injected activity, type of sphere used. All these data are necessary to perform any subsequent survival comparison of the two groups.

2) Survival analyses need to be accompanied by a Kaplan-Meier survival curve with the number of patients at risk at each time point.

Discussion:

1) Rather short, could use more information, some of it presented in the Introduction section

2) There needs to be a clear limitations section

Author Response

REVIEWER #1

In this article, Quartuccio and colleagues present a single-center retrospective database analysis of patients with hepatocellular carcinoma (HCC) and colorectal cancer with liver metastases (CLM) receiving transarterial radioembolization (TARE).

Overall, the article would require significant revisions to its methodology and presentation, before it can be considered acceptable for publication. Please find my comments for each section below.

Introduction:

1) Informative and very well-structured, albeit lengthy. Some of the information could be included in the Discussion instead, which is rather short.

The following periods have been moved to the discussion “Surgical removal of CRC liver metastases has shown promising 5-year survival rates of 20-70%, making it the preferred treatment option for suitable patients. However, due to technical constraints and the severity of the disease, a significant portion (70-80%) of individuals with extensive liver metastases are unable to undergo surgery [9].  There is expanding evidence that TARE provides a longer time to progression com-pared to transarterial chemoembolization (TACE) in patients with HCC [8]. Similarly, in patients with CRC liver metastases, according to the results of the EPOCH trial including 428 patients, TARE reduces the risk of disease progression or death compared to chemo-therapy alone [8].” Furthermore, the text body of the article has been extended.

2) Citations need to be revised - for example:

1. i) "There is expanding evidence that TARE provides a longer time to progression compared to transarterial chemoembolization (TACE) in patients with HCC" uses the EPOCH trial as a reference, while the EPOCH trial only provides data on CLM patients.

Thanks for spotting this uncorrect citation. We replaced this one with the correct one. “Brown, A.M.; Kassab, I.; Massani, M.; Townsend, W.; Singal, A.G.; Soydal, C.; Moreno-Luna, L.; Roberts, L.R.; Chen, V.L.; Parikh, N.D. Tace versus tare for patients with hepatocellular carcinoma: Overall and individual patient level meta analysis. Cancer medicine 2023, 12, 2590-259.”

1. ii) "the Resin trial demonstrated in a multicentre cohort of 498 patients that TARE is an effective second-line treatment for patients with CRC liver metastases" - uses an editorial commenting on the article, rather than original article of the RESIN registry analysis (PMID 35762890)

We replaced the reference with the one indicated.

Methods:

1) Detailed, comprehensive presentation of all selection criteria and procedures involved.

2) The statistics section is severely lacking:

1. i) There is no description of the statistical tests used for OS comparison or the units of measurement.

We extensively edited the statistical session of the methods including the method used to compare survival of different groups.

1. ii) The use of p-value needs to be clarified i.e. state that a p-value of <0.05 was used to denote statistical significance rather than mentioning "(p<0.05)" and if this was one-tailed or two-tailed.

We clarified better this information and added additional details, as recommended.

iii) The use of simple linear regression when analyzing survival data is inappropriate, as it assumes complete information for all data, which is not the case as censoring is used. The correct approach would be a Cox proportional hazard regression analysis. Due to the potential confounding effect of the variables analyzed, these should be adjusted in the main analysis comparing survival between HCC and CLM patients.

Dear reviewer we are concerned about the use of Cox proportional hazard regression analysis due to the small sample of our cohort (Statistics in Biopharmaceutical Research, 10:2, 139-149, DOI: 10.1080/19466315.2017.1369899). We would prefer to test the existence of statistically significant difference in the variables (please see table 1) between the subgroups and estimate the survival with Kaplan Meier Curves and Log rank test analysis.

Results:

1) The amount of information is inadequate; there needs to be a table with all the baseline characteristics of each group separately, and all potential confounders: age, sex, ECOG PS, Child-Pugh score, dose received, other second/third-line treatments, follow-up, mean radiation dose received, injected activity, type of sphere used. All these data are necessary to perform any subsequent survival comparison of the two groups.

We added a table with almost all the requested information.

2) Survival analyses need to be accompanied by a Kaplan-Meier survival curve with the number of patients at risk at each time point.

We added the Kaplan-Meier curves in the new version of the manuscript.

Discussion:

1) Rather short, could use more information, some of it presented in the Introduction section

We modified and extended the discussion.

2) There needs to be a clear limitations section.

We added a limitation section at the end of the discussion.

Reviewer 2 Report

Comments and Suggestions for Authors

As the authors state "The aim of this study was [to] investigate the influence of additional treatments, injected activity, [and?] mean dose delivered to the tumor on OS [overall survival]  of patients treated with 90-Y microspheres."

1. I believe the authors report these results, but the results are not presented in this order, and it is somewhat difficult to determine whether these aims were achieved. I note that this manuscript is rather brief and believe that the addition of a table of results would aid the evaluation of whether the aims were achieved. 

2. The authors conclude (page 5) that HCC patients receiving TARE survive longer than those with colorectal liver mets, yet they state that additional treatments, increasing injected activity, and mean dose my be beneficial for patients with liver mets from colorectal cancer (CRC). This seems contradictory.  Please clarify or restate this conclusions. Also, I believe the authors may want to include as one of their aims a comparison of the effect of TARE on HCC vs. CRC cases since these results are reported in some depth. 

3. Although the authors use the abbreviation "OS" for overall survival, the text indicates that only cancer-specific death was used as an outcome. Is this correct? If so, then overall survival is not the outcome, but more accurately it is disease- or cancer-specific survival. This should be corrected. 

4. The authors use correlation between the predictors and outcome for survival time (OS). This is somewhat unusual in that 99% of the studies I read use the Kaplan-Meier method with the log-rank test to evaluate between group differences in survival time. Also, the results of group comparisons report differences in median survival time. I think the authors may want to  justify their use of correlation as the basis for their group comparisons, or revise the analyses to  use the Kaplan-Meier (KM) method. Survival curves from a KM analysis may offer the authors a visual representation of the group differences observed or reported. 

5. The authors note that patient receiving additional treatment with Sorafenib or Regorafenib resulted in longer survival among patients with CRC liver mets. Where there any differences between patients who received additional therapy compared with those who received only TARE? I am curious to know if patients receiving additional therapy had a larger number of metastatic sites within the liver or any factors that may have resulted in differing treatment decisions between these patient groups (TARE vs. TARE+). 

6. On page 5, 5th paragraph under "Discussion", the authors write "...OS were not proved..." I would note that even the highest quality empirical studies rarely "prove" anything, and suggest that they use a less potent term such as "demonstrated" or "observed." 

Comments on the Quality of English Language

The English writing is very good but an additional careful proof-reading is recommended.

Author Response

REVIEWER #2

As the authors state "The aim of this study was [to] investigate the influence of additional treatments, injected activity, [and?] mean dose delivered to the tumor on OS [overall survival]  of patients treated with 90-Y microspheres."

1. I believe the authors report these results, but the results are not presented in this order, and it is somewhat difficult to determine whether these aims were achieved. I note that this manuscript is rather brief and believe that the addition of a table of results would aid the evaluation of whether the aims were achieved.

We added the requested table.

2. The authors conclude (page 5) that HCC patients receiving TARE survive longer than those with colorectal liver mets, yet they state that additional treatments, increasing injected activity, and mean dose my be beneficial for patients with liver mets from colorectal cancer (CRC). This seems contradictory. Please clarify or restate this conclusions. Also, I believe the authors may want to include as one of their aims a comparison of the effect of TARE on HCC vs. CRC cases since these results are reported in some depth.

We modified the conclusions section and also the abstract to make it cleared.

3. Although the authors use the abbreviation "OS" for overall survival, the text indicates that only cancer-specific death was used as an outcome. Is this correct? If so, then overall survival is not the outcome, but more accurately it is disease- or cancer-specific survival. This should be corrected.

We corrected the term in the body text and used the acronym CSS; we also modified the title accordingly (using “survival” rather than overall survival).

4. The authors use correlation between the predictors and outcome for survival time (OS). This is somewhat unusual in that 99% of the studies I read use the Kaplan-Meier method with the log-rank test to evaluate between group differences in survival time. Also, the results of group comparisons report differences in median survival time. I think the authors may want to justify their use of correlation as the basis for their group comparisons, or revise the analyses to use the Kaplan-Meier (KM) method. Survival curves from a KM analysis may offer the authors a visual representation of the group differences observed or reported.

We modified the analysis as requested.

5. The authors note that patients receiving additional treatment with Sorafenib or Regorafenib resulted in longer survival among patients with CRC liver mets. Where there any differences between patients who received additional therapy compared with those who received only TARE? I am curious to know if patients receiving additional therapy had a larger number of metastatic sites within the liver or any factors that may have resulted in differing treatment decisions between these patient groups (TARE vs. TARE+).

We added all of this information in the subgroup results sections.

6. On page 5, 5th paragraph under "Discussion", the authors write "...OS were not proved..." I would note that even the highest quality empirical studies rarely "prove" anything, and suggest that they use a less potent term such as "demonstrated" or "observed."

We modified the term in “observed”.

Comments on the Quality of English Language

The English writing is very good but an additional careful proof-reading is recommended.

We carried out additional proof-reading.

Reviewer 3 Report

Comments and Suggestions for Authors

The article “The influence of additional treatments, injected activity and mean dose to the tumor on the overall survival of patients undergoing transarterial radioembolization (TARE)” describes a retrospective, single-center study, focused on 39 HCC and CRC patients.

The sample size is limited. Moreover, different groups may be identified CRC (16 patients) – HCC (23 HCC). For each of these groups resin or glass microsphere subgroups may be considered. So they are numerous subgroups that leads to small samples at the analytical level. This may have an impact on the final statistical analysis.

One of the main points of SIRT is based on the dosimetric approach of treatment. It is unfortunate that the main recommendations on this subject are not cited in this article (i.e. Weber et al, EJNMMI (2022) 49 : 1682-1699 ; Salem et al, EJNMMI (2023) 50:328-343).

The method of dose definition is not sufficiently developed in the article since the materials and methods only refer to “a minimal dose of 120 Gy”. The software used for dosimetric estimation must be identified as well as the method used for dose calculation. It is necessary to explain how the dose is calculated : partition model, personal dosimetry at the voxel scale?

Moreover, the analysis pools here the 2 types of microspheres (resin and glass microspheres). It turns out that the dosimetric objectives are completely different according to the type of microspheres. Was the objective of 120 Gy applied for each type of microspheres? Glass microsphere indeed may consider the dose to the perfused volume.

Then the intent has to be developed too, because it has a real impact on the dosimetric parameters and on the overall survival

For example in the context of a segmentectomy in “single compartment analysis” the dose to the perfused volume must be > 400 Gy for glass microspheres while the objective is > 150 Gy for resin microspheres.

Personalized dosimetry is now at the heart of the SIRT recommendations for several years. It would be more interesting to use personalized dosimetry tools instead of the partition model for dosimetry. Personalized dosimetry allows the definition of many different volumes (tumor, total liver, healthy/normal total liver, perfused volume, perfused healthy liver). This could have led to a more detailed analysis.

Treatment selectivity has also to be taken into account.

Indeed, the dosimetric impact and therefore the consequences at the level of the liver will be completely different depending on whether the SIRT is used for a segmentectomy, lobectomy, bridging therapy (tumoral control+ contralateral hypertrophy), palliative purpose, etc.

In short, it seems difficult to me not to take into account more precisely the objective of the treatment in the analytical methodology of this article because in the end we start from different situations to which different therapeutic strategies are applied leading to different volumes perfused by the microspheres, and therefore different dosimetric and morphologic consequences for the liver.

As this work is focused on additional treatments, it would have been interesting to develop what kind of treatment are precisely used for the two studied pathologies.

For example, for HCC Atezoluzimab + bevacizumab is not quoted and specific treatments for CRC are not developed neither.

It is nevertheless interesting to better know every synergistic effect of the different treatment.

Author Response

REVIEWER #3

The article “The influence of additional treatments, injected activity and mean dose to the tumor on the overall survival of patients undergoing transarterial radioembolization (TARE)” describes a retrospective, single-center study, focused on 39 HCC and CRC patients. The sample size is limited. Moreover, different groups may be identified CRC (16 patients) – HCC (23 HCC). For each of these groups resin or glass microsphere subgroups may be considered. So they are numerous subgroups that leads to small samples at the analytical level. This may have an impact on the final statistical analysis.

Dear reviewer, we expanded the limitations section of our manuscript.

One of the main points of SIRT is based on the dosimetric approach of treatment. It is unfortunate that the main recommendations on this subject are not cited in this article (i.e. Weber et al, EJNMMI (2022) 49 : 1682-1699 ; Salem et al, EJNMMI (2023) 50:328-343).

We cited the article of Salem et al.

The method of dose definition is not sufficiently developed in the article since the materials and methods only refer to “a minimal dose of 120 Gy”. The software used for dosimetric estimation must be identified as well as the method used for dose calculation. It is necessary to explain how the dose is calculated: partition model, personal dosimetry at the voxel scale?

Dosimetric estimations at a voxel level were obtained using the commercially-available software Simplicit⁹⁰YTM (Mirada Medical, Oxford, UK). The prescribed ⁹⁰Y activity for SIR-microspheres was determined on the basis of a multicompartment dosimetric estimation, whereas a monocompartimental method was used for TheraSpheres.

Moreover, the analysis pools here the 2 types of microspheres (resin and glass microspheres). It turns out that the dosimetric objectives are completely different according to the type of microspheres. Was the objective of 120 Gy applied for each type of microspheres? Glass microsphere indeed may consider the dose to the perfused volume.

We added information in the text. Notably, in case of glass-microspheres, in patients with HCC, the absorbed doses were calculated in order to achieve a desirable minimal dose of 120 Gy to the perfused liver for lobar treatment and a minimum tumor-absorbed dose of 205 Gy, as recommended by Salem et al.

Then the intent has to be developed too, because it has a real impact on the dosimetric parameters and on the overall survival. For example in the context of a segmentectomy in “single compartment analysis” the dose to the perfused volume must be > 400 Gy for glass microspheres while the objective is > 150 Gy for resin microspheres.

We described better the intents in the methods, in the results and in table 1.

Personalized dosimetry is now at the heart of the SIRT recommendations for several years. It would be more interesting to use personalized dosimetry tools instead of the partition model for dosimetry. Personalized dosimetry allows the definition of many different volumes (tumor, total liver, healthy/normal total liver, perfused volume, perfused healthy liver). This could have led to a more detailed analysis.

We added details regarding the dosimetry and also additional information regarding the volumes in a table. 

Treatment selectivity has also to be taken into account. Indeed, the dosimetric impact and therefore the consequences at the level of the liver will be completely different depending on whether the SIRT is used for a segmentectomy, lobectomy, bridging therapy (tumoral control+ contralateral hypertrophy), palliative purpose, etc.

The intents have been added to the results section. We are aware of this limitation and added also this as a comment at the end of the discussion.

In short, it seems difficult to me not to take into account more precisely the objective of the treatment in the analytical methodology of this article because in the end we start from different situations to which different therapeutic strategies are applied leading to different volumes perfused by the microspheres, and therefore different dosimetric and morphologic consequences for the liver. As this work is focused on additional treatments, it would have been interesting to develop what kind of treatment are precisely used for the two studied pathologies. For example, for HCC Atezoluzimab + bevacizumab is not quoted and specific treatments for CRC are not developed neither.

The text has been edited adding all the possible retrieved information.

It is nevertheless interesting to better know every synergistic effect of the different treatment.

Dear reviewer we believe that this analysis would not be correct due to the limited sample sizes.

Round 2

Reviewer 1 Report

Comments and Suggestions for Authors

The authors have provided an updated version of their article on the outcomes of TARE in patients with hepatocellular carcinoma and colorectal liver metastases.

They have made significant improvements to their manuscript - they now provide a much more detailed methodology including a statistics section, the presentation of the results has improved by adding a table with baseline characteristics and Kaplan-Meier survival curves and their discussion is now more comprehensive and covers important aspects of the topic such as future directions for research and limitations of the present study.

Only a few minor typos or grammatical errors need to be corrected during proofreading (see below).

Overall, the authors have answered my questions and followed my recommendations adequately and therefore I can now recommend the article for publication.

Comments on the Quality of English Language

Page 3 second paragraph typo - "stibl debated" instead of still

Page 5 typos right above the Table - "coninous" instead of continuous; "different" instead of difference

Table 1 - I would use the dot rather than comma as a decimal separator, as it is the standard in English literature

Page 9 second paragraph - "there is not enough data" should be changed to "there are not enough data"

Reviewer 3 Report

Comments and Suggestions for Authors

The recommended evolutions have been taken into account

Comments on the Quality of English Language

Some spelling errors to correct