Inflammation induced by incomplete radiofrequency ablation accelerates tumor progression and hinders PD-1 immunotherapy

Radiofrequency ablation (RFA) promotes tumor antigen-specific T cell responses and enhances the effect of immunotherapy in preclinical settings. Here we report that the existence of remnant tumor masses due to incomplete RFA (iRFA) is associated with earlier new metastases and poor survival in patients with colorectal cancer liver metastases (CRCLM). Using mouse models, we demonstrate that iRFA promotes tumor progression and hinders the efficacy of anti-PD-1 therapy. Immune analysis reveals that iRFA induces sustained local inflammation with predominant myeloid suppressor cells, which inhibit T cell function in tumors. Mechanistically, tumor cell-derived CCL2 is critical for the accumulation of monocytes and tumor-associated macrophages (TAMs). The crosstalk between TAMs and tumor cells enhances the CCL2 production by tumor cells. Furthermore, we find that administration of a CCR2 antagonist or the loss of CCL2 expression in tumor cells enhances the antitumor activity of PD-1 blockade, providing a salvage alternative for residual tumors after iRFA.


Introduction
-Reference 12 is not properly cited and is an important part of the rationale for this study -It would be helpful to more specifically describe findings from previous studies in terms of iRFA leading to progression within or outside of the treated lesions given the current study findings.

Results
-Despite the authors use of a MVA, it is hard to account for bias where the iRFA lesions are inherently more aggressive in contrast to the incomplete ablation leading to worse outcomes -this should be mentioned as a limitation in the discussion. -As mentioned above, more comparisons with cRFA need to be performed to demonstrate the inhibitory effects are not a result of RFA in general -for example, does cRFA also cause increase in MDSC in the peripheral blood? -Is there any additional benefit to PD-1 inhibition in the setting of iRFA and CCL2 knock out with CRISPR?
Method -It should be noted whether local response in the iRFA lesions was determined blinded to the development of distant progression, which would have been understandably difficult. However, if this was not blinded there is the potential for bias. -There are several typos in the methods -e.g. "presented local tumor", "baseline imaging evaluations liver metastases" -How were the treatment parameters chosen for partial ablation? Were other parameters tried? Was there a minimum threshold below which the suppressive effects were not observed?
RFA was an independent risk factor for earlier metastasis and poor survival. Using mouse models they show incomplete RFA of cancers was associated with immune supressive myeloid cells that impaired T cell function and is thought to be driven by CCL2. The strength of the mechanistic findings are considerable given a a CCR2 antagonist could restore some T cell function.
How do the authors resolve these findings with the statements they make in the introduction about "the synergistic 54 antitumor effects of RFA in combination with anti-PD-1 therapy in a preclinical 55 study 11. Recently, a phase II clinical trial showed that local ablation combined 56 with anti-PD-1 therapy significantly improves the survival of patients with 57 oligometastatic non-small cell lung cancer (NSCLC)>" This is a well done study, clearly presented and tells an interesting story. The reliance on a single model of CRC is a weakness. There are otherwise no major deficiencies.

Point-to-point answers to reviewer comments
All changes in the text are in red color.

Reviewer #1
This is an intriguing study that suggests that incomplete local treatment in the form of RFA may give rise to distant metastases and systemic tumor progression due to upregulated antitumor immunity. This is suggested clinically using a retrospective controlled study involving more than 500 treated tumors where incomplete ablation was found to be an independent risk factor for earlier development of metastases and inferior survival. In corresponding animal models, iRFA was associated with local resistance to future treatment with PD-1 inhibitors, and interestingly also associated with systemic resistance to PD-1 therapy and the development of distant metastases as well. There were complex genetic changes in the incompletely treated primary tumor reflecting inflammatory changes led to an influx of inhibitory cells. This recruitment of TAM was particularly dependent on CCL2 induced by TNFalpha, which are identified as potential therapeutic targets in this setting.
Overall this is a provocative and well done translational study with potential clinical and therapeutic implications, and is therefore worthy of publication. However, major issues include the need to (1) more conclusively demonstrate that inhibitory mechanisms demonstrated are specifically related to iRFA and not also present in the setting of cRFA. (2) All experiments except for PD-1/CCR2 inhibition also seem to have been performed on CT26 and therefore it is hard to know whether these results are generalizable to other model systems. (3) The manuscript would also benefit from more careful grammatical editing throughout. Other more specific comments are below.

Introduction
(4)-Reference 12 is not properly cited and is an important part of the rationale for this study. It would be helpful to more specifically describe findings from previous studies in terms of iRFA leading to progression within or outside of the treated lesions given the current study findings.

Results
(5) -Despite the authors use of a MVA, it is hard to account for bias where the iRFA lesions are inherently more aggressive in contrast to the incomplete ablation leading to worse outcomes -this should be mentioned as a limitation in the discussion.
(1) As mentioned above, more comparisons with cRFA need to be performed to demonstrate the inhibitory effects are not a result of RFA in general -for example, does cRFA also cause increase in MDSC in the peripheral blood?
(6)-Is there any additional benefit to PD-1 inhibition in the setting of iRFA and CCL2 knock out with CRISPR? Method (7)-It should be noted whether local response in the iRFA lesions was determined blinded to the development of distant progression, which would have been understandably difficult. However, if this was not blinded there is the potential for bias. Comment 2: All experiments except for PD-1/CCR2 inhibition also seem to have been performed on CT26 and therefore it is hard to know whether these results are generalizable to other model systems.

Response:
To extend the previous data, we performed additional experiments using another colon cancer model MC38. In general, the new data confirmed the previous findings in CT26 and hepa1-6 models. The new data now are presented in parallel with the previous data obtained from CT26 model in the revised Fig 2, 4-8, Fig S1, 2,

4-6.
Comment 3: The manuscript would also benefit from more careful grammatical editing throughout.
Response: According to the Reviewer's suggestion, we have asked a native English speaker to check grammatical editing.

Comment 4:
Reference 12 is not properly cited and is an important part of the rationale for this study. It would be helpful to more specifically describe findings from previous studies in terms of iRFA leading to progression within or outside of the treated lesions given the current study findings Response: According to the Reviewer's suggestion, we removed the reference and added two reference in terms of rapid tumor progression after iRFA.
Comment 5: Despite the authors use of a MVA, it is hard to account for bias where the iRFA lesions are inherently more aggressive in contrast to the incomplete ablation leading to worse outcomes -this should be mentioned as a limitation in the discussion.

Response:
We agree that the target lesions might be inherently more aggressive and lead to worse outcomes. We have discussed about this point in the revised manuscript and we have updated the references accordingly (page 12). Response: Due to nature of RFA treatment, it is difficult to use blind method to evaluate local response after treatment. When designing this study, we realized the potential bias. So, we also compared the weight of local tumors examined on day 14 after iRFA by dissection the mice. This may be helpful to reduce bias.

Comment 8:
There are several typos in the methods -e.g. "presented local tumor", "baseline imaging evaluations liver metastases" Response: We thank this reviewer for kindly pointing out these errors. We have carefully checked typos in the revised manuscript. Cancer Letters, 2018). We think this may be a more precise method to control the extent of ablation, and it is worth trying in future studies. Response: This is a critical issue worthy of in-depth discussion. Further studies are warranted, as it not only is helpful to find new strategies to prevent the recurrence of ablated tumors, but also may provide new insights into the relationship between inflammation and tumor progression. Therefore, we wish to explore an experimental model for precise control of residual tumors, and ultimately understand the threshold for initiating immunosuppression.

Reviewer #2 :
Major findings: Retrospective review of patient CT's showed that patients with residual metastatic to the liver after RFA was an independent risk factor for earlier metastasis and poor survival. Using mouse models they show incomplete RFA of cancers was associated with immune suppressive myeloid cells that impaired T cell function and is thought to be driven by CCL2. The strength of the mechanistic findings are considerable given a a CCR2 antagonist could restore some T cell function.
(1) How do the authors resolve these findings with the statements they make in the introduction about "the synergistic antitumor effects of RFA in combination with anti-PD-1 therapy in a preclinical study? Recently, a phase II clinical trial showed that local ablation combined 56 with anti-PD-1 therapy significantly improves the survival of patients with oligometastatic non-small cell lung cancer (NSCLC)" . This is a well done study, clearly presented and tells an interesting story. (2) The reliance on a single model of CRC is a weakness. There are otherwise no major deficiencies.
Comment 1：How do the authors resolve these findings with the statements they make in the introduction about "the synergistic antitumor effects of RFA in combination with anti-PD-1 therapy in a preclinical study?

Response:
The present study was aimed to evaluate the immune response in residual tumors after incomplete RFA (iRFA). We have deleted the inappropriate statement in the section of introduction in the revised manuscript, and more specifically describe findings from previous studies in terms of iRFA leading to progression.
Comment 2：The reliance on a single model of CRC is a weakness.

Response:
We performed additional experiment using another colon cancer model MC38. In general, the new data confirmed the previous finding in CT26 and hepa1-6 models. The new data now are presented in parallel with the previous data obtained from CT26 model in the revised Fig 2, 4-8, Fig S1, 2, 4-6.