Early Changes in Alpha-Fetoprotein and Des-γ-Carboxy Prothrombin Are Useful Predictors of Antitumor Response to Durvalumab Plus Tremelimumab Therapy for Advanced Hepatocellular Carcinoma

Round 1

Reviewer 1 Report

Comments and Suggestions for Authors

In this study, Kuzuya et al. seek the potentials of AFP and DCP as factors to predict response to durvalumab plus tremelimumab in HCC patients. Although the topic is tempting, basic research design is questionable. Therefore, I cannot review this manuscript unless the authors address the points listed below.

·       The authors compare two groups, 8W-OR vs. Non-8W-OR and 8W-DC vs. Non-8W-DC. The SD group is sometimes with CR+PR but sometimes with PD+NE. I do not think this is appropriate because in Figure 1b, the authors show that the SD group is clearly different from CR+PR or PD+NE groups. The authors do not provide evidence to show that they can merge SD with CR+PR or PD+NE. These three groups should be separated because they are clearly different in Figure 1b.

·       In Table 1, the authors show parameters for each group (SD, CR+PR, and PD+NE) and need to show that there are no significant differences between groups for parameters, such as treatment line or grades or stages. Otherwise it is impossible to conclude if the differences in AFP or DCP or PFS are due to differences in parameters or sensitivity against treatments.

Author Response

Thank you very much for reviewing our study. Your valuable comments and insights have greatly improved the manuscript.

Initially, we had results for 32 out of 40 patients. With the extended follow-up, we now have data for all 40 patients and have performed a new analysis. The overall results remain similar and we believe the increased number of cases improves the quality.

We have made revisions based on your feedback. Changes due to the updated data for 40 patients are highlighted in blue, changes made in response to reviewer comments are highlighted in red, and all figures have been updated accordingly.

1.

The authors compare two groups, 8W-OR vs. Non-8W-OR and 8W-DC vs. Non-8W-DC. The SD group is sometimes with CR+PR but sometimes with PD+NE. I do not think this is appropriate because in Figure 1b, the authors show that the SD group is clearly different from CR+PR or PD+NE groups. The authors do not provide evidence to show that they can merge SD with CR+PR or PD+NE. These three groups should be separated because they are clearly different in Figure 1b.

 

Thank you for your very important comments. As you mentioned, the SD group is included in both the non-OR and DC groups.

In our institution, as in other institutions, anti-tumor efficacy is determined by RECIST and evaluated in 4 groups: CR, PR, SD and PD. In addition, OR (CR+PR) and DC (CR+PR+SD), which are commonly used in clinical trials, are also evaluated at the same time and are used as criteria for deciding whether to continue treatment.

The presence or absence of DC is an important indicator for treatment continuation. In the era when only molecularly targeted therapies were available, evaluation of DC was important because the proportion of patients who could achieve OR was low. On the other hand, since the availability of ICIs such as Atz/Bev and Dur/Tre, the percentage of OR cases has increased, and once OR is achieved, treatment strategies aimed at CR with other local therapies are also gaining attention. Therefore, early prediction of patients with good antitumor response (OR) and poor antitumor response (PD, non-DC) may be useful in Dur/Tre therapy. Therefore, we focused on OR vs. non-OR and Dc vs. non-DC to investigate the treatment outcome.    

 

2.  

In Table 1, the authors show parameters for each group (SD, CR+PR, and PD+NE) and need to show that there are no significant differences between groups for parameters, such as treatment line or grades or stages. Otherwise it is impossible to conclude if the differences in AFP or DCP or PFS are due to differences in parameters or sensitivity against treatments.

 

Thank you very much for pointing out this very important point. Results for background factors according to RECIST1.1 are shown in Supplementary Table 1. There were no significant differences in the relationship between anti-tumor efficacy and background factors between the OR group and the non-PR group, or the DC group and the non-DC group.

The following texts have been added to “Result 3.2”.

“Baseline characteristics at Dur/Tre initiation, stratified by 8W-RECIST 1.1, are shown in Supplementary Table 1. There were no significant differences in baseline factors between the 8W-OR and non-8W-OR groups or between the 8W-DC and non-8W-DC groups.”

Reviewer 2 Report

Comments and Suggestions for Authors

In this manuscript, authors identified that the dynamic change of AFP and DCP can predict the response of durvalumab plus tremelimumab therapy for advanced hepatocellular carcinoma. A high AFP or DCP ratio at 4 week was associated with an increased risk of no response at 8 week. However, the analyses presented in this manuscript are flawed (see below), and some crucial information was omitted.

Below some specific comments:

1.      The changes in tumor markers are unknown before receiving treatment, thus it is not possible to guide the formulation of the treatment plan beforehand. Many studies have shown that baseline levels of tumor markers can predict the response to immunotherapy, so it is recommended to evaluate the predictive value of baseline levels of tumor markers for treatment response.

2.      It is recommended to discuss novel HCC prognosis models related to AFP, AFP-L3, and DCP in the Discussion (PMID: 37683735, 33950167, 34618932, 37988413).

3.      For the Cox regression analysis, how is the cutoff value for continuous variables determined? For example, why was the cutoff value for AFP chosen to be 163 ng/mL? Additionally, the biomarker ratios should also be incorporated into the Cox analysis.

4.      If possible, please add the results of biopsy pathology and genetic testing for the HCC patients in this study. For Dur/Tre therapy, tumor MSI and PD-L1 expression levels also could predict the therapeutic responses.

5.      The definition of PFS and 1^st-line treatment need to be mentioned in the Methods.

6.      Detailed inclusion and exclusion criteria for the HCC patients should be added in the Methods.

7.      The predictive value of early changes has not been evaluated in patients with low baseline levels of biomarkers, thus the conclusions are limited. It is necessary to emphasize the application population of the conclusions in the text.

8.      The survival curve for PFS and OS stratified by 8W-mRECIST need to be added.

9.      It is suggested to replace "8W-RECIST" with "8W-RECIST 1.1"

10.  The absolute values of AFP, DCP and AFP-L3 should also be provided in Table 3-5.

Comments on the Quality of English Language

 The language needs to be polished.

Author Response

Thank you very much for reviewing our study. Your valuable comments and insights have greatly improved the manuscript.

Initially, we had results for 32 out of 40 patients. With the extended follow-up, we now have data for all 40 patients and have performed a new analysis. The overall results remain similar and we believe the increased number of cases improves the quality.

We have made revisions based on your feedback. Changes due to the updated data for 40 patients are highlighted in blue, changes made in response to reviewer comments are highlighted in red, and all figures have been updated accordingly.

 

1.

The changes in tumor markers are unknown before receiving treatment, thus it is not possible to guide the formulation of the treatment plan beforehand. Many studies have shown that baseline levels of tumor markers can predict the response to immunotherapy, so it is recommended to evaluate the predictive value of baseline levels of tumor markers for treatment response.

Thank you very much for your valuable comments. It is very important to check whether pre-treatment tumor marker levels are predictive biomarkers of anti-tumor efficacy, so we analyzed their relationship. Supplementary Table 1 shows the relationship with anti-tumor efficacy in all 40 patients, and Supplementary Tables 2-4 show the results of the analysis of the relationship with anti-tumor efficacy in the patients whose tumor marker trends were examined. There was no association between pre-treatment tumor marker levels and anti-tumor effect.

The following texts have been added to “Result 3.2”, “Discussion, Paragraph 7.

“Baseline characteristics at Dur/Tre initiation, stratified by 8W-RECIST 1.1, are shown in Supplementary Table 1. There were no significant differences in baseline factors between the 8W-OR and non-8W-OR groups or between the 8W-DC and non-8W-DC groups.”

” In the present study, no correlation was found between baseline actual tumor marker levels and PFS, 8W-OR, and 8W-DCR (Table 6, Supplementary Tables 1-4).”

 

2.

It is recommended to discuss novel HCC prognosis models related to AFP, AFP-L3, and DCP in the Discussion (PMID: 37683735, 33950167, 34618932, 37988413).

Thank you for the very useful information. I have read these four English papers. They are all useful and I have added them as references (Refs. 24-27).

The following text was added to the “Discussion, 6th paragraph.

“Combining multiple tumor markers or integrating markers with age and tumor factors has been reported as a useful indicator for predicting anti-tumor efficacy and prognosis [24-27]. Therefore, in Dur/Tre treatment, the use of a combination of these factors may be a more effective biomarker for predicting treatment outcome.”

 

 

3.

For the Cox regression analysis, how is the cutoff value for continuous variables　determined? For example, why was the cutoff value for AFP chosen to be 163 ng/mL? Additionally, the biomarker ratios should also be incorporated into the Cox analysis.

Thank you for your valuable advice regarding cutoff values. We have redefined the cutoffs for baseline tumor marker and NLR levels under the following conditions. We have adopted the cutoff values used in similar analyses in general, which are close to the median values of the patients in this study. Specifically, the following cutoff values. AFP level (100 ng/mL), DCP level (400 mAU/mL), AFP-L3 level (10.0%), and NLR (3.00).

Thank you for your comments regarding the Cox analysis of factors contributing to PFS. The purpose of this study was to determine if there are useful factors that contribute to PFS prior to treatment initiation. Therefore, we did not include the change in tumor markers after treatment initiation because we only used factors at treatment initiation in our analysis. However, you raise an important point, and we did perform an ROC analysis using tumor marker ratios at 2 and 4 weeks to calculate the cutoff for tumor marker ratios that are useful for predicting 8W-OR and 8W-DC.

The following texts were added to “Materials and methods 2.5.”. “Results 3.4.”, “Results 3.5.”, “Results 3.6.”, “Discussion, 3th paragraph”, “Discussion, 4th paragraph”.

“Cutoff values were determined using the receiver operating characteristic (ROC) curve and area under the curve (AUC).”

“ROC curves were plotted using 8W-OR and 8W-DC as state variables and AFP ratios at weeks 2 and 4 as test variables. For 8W-OR, the AUCs at weeks 2 and 4 were 0.64 and 0.84, with an optimal week 4 cutoff of 0.53 (75% sensitivity, 89.5% specificity). For 8W-DC, the AUCs at weeks 2 and 4 were 0.78 and 0.89, with an optimal week 4 cutoff of 1.01 (86.7% sensitivity, 91.7% specificity).”

“ROC curves were plotted using 8W-OR and 8W-DC as state variables and DCP ratios at weeks 2 and 4 as test variables. For 8W-OR, the AUCs at weeks 2 and 4 were 0.84 and 0.98, with an optimal week 4 cutoff of 0.48 (100% sensitivity, 92.6% specificity). For 8W-DC, the AUCs at weeks 2 and 4 were 0.63 and 0.72, with an optimal week 4 cutoff of 0.73 (61.9% sensitivity, 81.2% specificity).”

“ROC curves were plotted using 8W-OR and 8W-DC as state variables and the AFP-L3 ratio at week 4 as the test variable. The AUC was 0.60 for 8W-OR and 0.51 for 8W-DC, both indicating low correlation.”

“The optimal cut-off values for the AFP ratio at week 4 predicting 8W-OR and 8W-DC were 0.53 and 1.01, respectively.”

“The optimal cut-off values for the DCP ratio at week 4 predicting 8W-OR and 8W-DC were 0.48 and 0.73, respectively.”

 

 

4.

If possible, please add the results of biopsy pathology and genetic testing for the HCC patients in this study. For Dur/Tre therapy, tumor MSI and PD-L1 expression levels also could predict the therapeutic responses.

   Thank you for your valuable input. However, in the 40 cases that were the subject of this review, not a single case was tested for MSI or PD-L1 expression.

 

 

5.

The definition of PFS and 1^st-line treatment need to be mentioned in the Methods.

   Thank you for your suggestion. We have added the following statement to “Methods 2.5”., Methods 2.2”. In addition to PFS, the definition of OS is also provided.

   “PFS was measured as the time from the start of Dur/Tre treatment to date of progression by RECIST 1.1 or date of death or last hospital visit. Overall survival (OS) was measured as the time from the start of Dur/Tre treatment to death or last hospital visit.”

“The treatment line refers to the sequence of systemic therapy with MTA or immune checkpoint inhibitors (ICIs).”

 

 

6.

Detailed inclusion and exclusion criteria for the HCC patients should be added in the Methods.

   Thank you very much for pointing this out. We have added the following statement to “Methods 2.1”.,

“Between March 2023 and May 2024, 40 consecutive patients with advanced unresectable HCC started Dur/Tre therapy at our institution. Eligibility criteria included: 1) HCC diagnosis by imaging or biopsy, 2) Barcelona Clinic Liver Cancer (BCLC) stage C or B ineligible for surgery or locoregional therapy, 3) Eastern Cooperative Oncology Group performance status (ECOG-PS) of 0 or 1, and 4) Child-Pugh score of 7 or less. In addition, three patients were included: one with multinodular fused BCLC A and a maximum tumor diameter of 100 mm, and two with Child-Pugh scores of 8 and 9 due to large tumor volume despite chronic hepatitis. All 40 patients were included in this study for retrospective evaluation of treatment outcomes.”

 

 

7.

The predictive value of early changes has not been evaluated in patients with low baseline levels of biomarkers, thus the conclusions are limited. It is necessary to emphasize the application population of the conclusions in the text.

   Thank you very much for your very important remarks. The following text has been added to the “Discussion 9 paragraphs”.

   “This study included only patients with baseline tumor marker levels above the upper limit of normal. Thus, it did not evaluate the relationship between tumor marker changes and antitumor efficacy in patients with normal levels, limiting the conclusions.”

 

 

8.

The survival curve for PFS and OS stratified by 8W-mRECIST need to be added.

Thank you very much for your helpful advice. I have analyzed the data as you suggested and added the PFS by mRECIST as Figure 1c. We have added the following statement to “Results 3.3”.,

“Median PFS stratified by 8W-mRECIST was NR (95%CI: 7.8–NR) in the CR group (n=3), NR (95%CI: 6.4–NR) in the PR group (n=9), 6.5 months (95%CI: 3.7–7.6 months) in the SD group (n=11) and 1.8 months (95%CI: 1.5–1.8 months) in the PD+NE group (n=17) (Figure 1c).”

Regarding OS, the PD+NE group in RECIST 1.1 is the same as the PD+NE group in mRECIST. Also, no deaths have occurred in the CR, PR or SD groups at this time. Therefore, only the results of the analysis of the CR+PR+SD group vs. the PD+NE group in RECIST 1.1 are included, as they are the same for both RECIST 1.1 and mRECIST (Figure 2b).

 

 

9.

It is suggested to replace "8W-RECIST" with "8W-RECIST 1.1"

Thank you very much for pointing this out. We have changed everything from "RECIST" to "RECIST 1.1" as you suggested.

 

 

10.

The absolute values of AFP, DCP and AFP-L3 should also be provided in Table 3-5.

Thank you very much for your useful advice.

 

We have analyzed the actual tumor marker values by RECIST 1.1. Supplementary Table 2-4.

The following texts were added to “Results 3.4.”, “Results 3.5.”, “Results 3.6.”, “Discussion, Paragraph 7”.

“Actual AFP levels at 0, 2, 4, and 8 weeks after Dur/Tre initiation, stratified according to the 8W-RECIST 1.1 are shown in Supplementary Table 2. Median actual AFP levels in the 8W-DC group at weeks 4 and 8 were significantly lower than those in the Non-8W-DC group (p=0.0128 and p=0.0063, respectively).”

 

“Actual DCP levels at 0, 2, 4, and 8 weeks after Dur/Tre initiation, stratified according to the 8W-RECIST 1.1 are shown in Supplementary Table 3. Median actual DCP levels in the 8W-OR group at weeks 4 and 8 were significantly lower than those in the Non-8W-OR group (p=0.0167 and p=0.0001, respectively). Median actual DCP level in the 8W-DC group at week 8 was significantly lower than that in the Non-8W-DC group (p=0.0048).”

 

“Actual AFP-L3 levels at 0, 2, 4, and 8 weeks after Dur/Tre initiation, stratified according to the 8W-RECIST 1.1 are shown in Supplementary Table 4. Median actual AFP-L3 level in the 8W-DC group at week 8 was significantly lower than that in the Non-8W-DC group (p=0.0390).”

 

“In the present study, no correlation was found between baseline actual tumor marker levels and PFS, 8W-OR, and 8W-DCR (Table 6, Supplementary Tables 1-4).”

Round 2

Reviewer 1 Report

Comments and Suggestions for Authors

No further comments.

Reviewer 2 Report

Comments and Suggestions for Authors

The author has made detailed revisions to the manuscript and found no other errors.