Combination Strategies for Immune-Checkpoint Blockade and Response Prediction by Artificial Intelligence

Round 1

Reviewer 1 Report

In this manuscript, Florian Huemer et al. collected the latest data on complex approaches to anti-cancer therapy which included the combination of immune checkpoint inhibitors and other treatment types. The topic of the review is interesting and important. Hopes for using the combination treatment are based on possibility to enhance antitumor immune cell activity at multiple points. The manuscript includes, not counting the introduction, several major sections considering various combinations of PD-1inhibition strategies with applications of kinase inhibitors, anti-cancer vaccines, and CAR-T-cell therapy. Item of special interest is the section covering ways of predicting response to immune-checkpoint inhibition with the help of artificial intelligence approaches.  The manuscript is very well written and, despite of the complexity of the issue, reads well. It provides actual information, which is highly relevant to the designated topic. However, it needs to be improved in some parts.

Several concerns should be addressed to increase the overall quality of the manuscript.

1. Figure 2 does not look clear. Authors need to improve the figure by reflecting more causal relationships between the represented items.  
2. Some parts of the text in sections 2 and 5 (for example, lines 130-135 and lines 283-285, 288-291) are mostly repetitive, it should be improved.
3. Importantly, the excessive activation of the immune cells may lead to the development of autoimmune aggression and tissue damage. Combination anti-cancer therapy should be discussed more in the aspect of undesirable consequences.
4. The title of section 2 looks not good. The enumeration of an object and a phenomenon does not reflect their relationship.
5. The language of the text is mostly good; still, the use of T-cells with a hyphen in many cases is confusing.  

Author Response

We thank the reviewer for the valuable input, which has definitely helped to improve our manuscript. The raised questions are answered in a point-by-point fashion and changes in the revised manuscript are highlighted using the “Track Changes” function.

Reviewer 1:

 “In this manuscript, Florian Huemer et al. collected the latest data on complex approaches to anti-cancer therapy which included the combination of immune checkpoint inhibitors and other treatment types. The topic of the review is interesting and important. Hopes for using the combination treatment are based on possibility to enhance antitumor immune cell activity at multiple points. The manuscript includes, not counting the introduction, several major sections considering various combinations of PD-1inhibition strategies with applications of kinase inhibitors, anti-cancer vaccines, and CAR-T-cell therapy. Item of special interest is the section covering ways of predicting response to immune-checkpoint inhibition with the help of artificial intelligence approaches.  The manuscript is very well written and, despite of the complexity of the issue, reads well. It provides actual information, which is highly relevant to the designated topic. However, it needs to be improved in some parts.

Several concerns should be addressed to increase the overall quality of the manuscript.”

 

#1: “Figure 2 does not look clear. Authors need to improve the figure by reflecting more causal relationships between the represented items.”

Authors’ response: Figure 2 has been adapted and has been substituted (page 5).

#2: “Some parts of the text in sections 2 and 5 (for example, lines 130-135 and lines 283-285, 288-291) are mostly repetitive, it should be improved”

Authors’ response:

The following repetitive sentences have been deleted: “CTLA-4 is upregulated in activated T-cells as well as in Tregs and is important for T-cell activation processes in secondary lymphoid organs [6]. PD-1 is expressed on activated T-cells, B-cells, macrophages, natural killer (NK) cells and several antigen-presenting cell types [6].“ (page 7)

#3: “Importantly, the excessive activation of the immune cells may lead to the development of autoimmune aggression and tissue damage. Combination anti-cancer therapy should be discussed more in the aspect of undesirable consequences”

Authors’ response:

A new chapter has been added to the review “6. Toxicity associated with immune-checkpoint blockade combination strategies” with the following text: “The encouraging and exciting activity of ICI therapy comes at the cost of immune related adverse events (IRAE). IRAE are thought to arise from an “over-activation” of the immune system leading to autoimmune inflammatory events affecting virtually any organ, most commonly the skin, gastrointestinal tract, liver, endocrine system and lung [104, 105]. Excellent guidelines on management of these conditions have been recently published and can be found elsewhere [106]. In this review, we have highlighted potential combination strategies with ICI. We have seen in the past that the combination of different ICI (i.e. nivolumab with ipilimumab) results in an increased rate of severe IRAE as reviewed recently in a meta-analysis [107]. Consequently, there is a relevant concern that the combination of ICI with abovementioned potent therapies may lead to excessive toxicity. Overall, clinical experience with the combination strategies discussed in this review is limited. However, a recent phase 3 trial comparing the combination of the tyrosine kinase inhibitor axitinib and pembrolizumab with sunitinib for the treatment of renal cell carcinoma reported an increased rate of grade 3-5 liver toxicity in the combination arm [15]. Similar evidence for the potential of additive toxicity comes from a phase II trial evaluating nivolumab in combination with ibrutinib for patients with advanced CLL [108]. In the latter study diarrhea was the most commonly reported adverse event, probably indicating additive toxicity caused by the two drug classes. However, none of the eleven fatal adverse events that have been reported in this trial where deemed to be drug-related [108].

Regarding CAR-T cell treatment, the most relevant drug-specific adverse events are CRS and neurotoxicity[109]. Since CRS results from an over-activation of immune effector cells, combination with ICI causes significant concerns regarding excessive toxicity. Up to now, clinical experience with combination strategies of ICI and CAR-T cells are very limited. In the preliminary studies discussed above, the rate and severity of reported CRS was similar to CAR-T cell monotherapy and no life threatening CRS was reported. Overall, larger randomized studies will be required to evaluate the actual risk for severe adverse events with ICI combinations.” (page 15-16)

 

#4: “The title of section 2 looks not good. The enumeration of an object and a phenomenon does not reflect their relationship”

 

Authors’ response:

The title of section 2 has been renamed to “Programmed death 1 (PD-1) and its key role in T cell exhaustion (page 2).

#5: “The language of the text is mostly good; still, the use of T-cells with a hyphen in many cases is confusing”

Authors’ response:

The hyphen between T cells has been removed in the entire manuscript.

Reviewer 2 Report

The authors did a good job on summarizing strategies for immune-checkpoint blockade.

However, they paid little attention to the prediction of ICI therapy responses by "artificial intelligence". First, the authors only mentioned (a) alterations in the antigen presenting pathway and (b) radiomics. Other methods should be discuss in more detail.

Second, several publications regarding neoantigen prediction were not included or discussed.

Third, a table to summarize all work about ICI therapy responses by "artificial intelligence" should be provided.

Forth, too many long sentences make the manuscript hard to read and understand. 

 

Author Response

We thank the reviewer for the valuable input, which has definitely helped to improve our manuscript. The raised questions are answered in a point-by-point fashion and changes in the revised manuscript are highlighted using the “Track Changes” function.

Reviewer 2:

 “The authors did a good job on summarizing strategies for immune-checkpoint blockade. However, they paid little attention to the prediction of ICI therapy responses by "artificial intelligence".”

 

#1: “First, the authors only mentioned (a) alterations in the antigen presenting pathway and (b) radiomics. Other methods should be discuss in more detail.”

Authors’ response:

We totally agree with the reviewer that the manuscript does not cover each topic of ICI response prediction by artificial intelligence. We have focused on the antigen presenting pathway as well as radiomics. Covering all aspects of artificial intelligence would go beyond the scope of this review. However, we have mentioned and described the publication by Xie et al. (Clinical Cancer Research, 2020) as an example of a comprehensive analysis incorporating genomic analyses and RNA sequencing data aiming at predicting ICI responses (page 20).

 

#2: “Second, several publications regarding neoantigen prediction were not included or discussed”

Authors’ response:

An updated overview of (tumor) neoantigen prediction models including references has been added as Table 4.

#3: “Third, a table to summarize all work about ICI therapy responses by "artificial intelligence" should be provided”

Authors’ response:

Two additional tables have been added to the manuscript: Table 5 summarizes studies investigating the impact of the antigen presenting pathway as well as of the TCR repertoire on clinical outcome with immune-checkpoint blockade. Table 6 summarizes response prediction studies by radiomics in patients undergoing ICI therapy.

 

#4: “Forth, too many long sentences make the manuscript hard to read and understand”

 

Authors’ response:

We have made an effort to shorten several sentences.

Reviewer 3 Report

In this review the authors discuss the use of immune checkpoint inhibitors (ICI) in cancer immunotherapy. Unlike most reviews of the field, which refer to clinical aspects of the use of ICI as monotherapy, this review focuses on promising strategies for combinational therapies employing ICI, as well as novel genetic engineering approaches and emerging prediction tools, all aiming at maximizing the clinical benefit of ICI treatment.

This is an outstanding review, which systemically covers a wide spectrum of approaches, going in each from basic immunological principles and how these had been elucidated in pioneering works to current technologies and achievements, including ongoing clinical trials. The idea of exploiting artificial intelligence to create a powerful prediction platform incorporating analytical tools with experimental and clinical data is intriguing and by itself justifies publication of this review.

The only major reservation I have is that the review lacks a summarizing section/paragraph.

Minor comments are technical:

• Fig. 1. Font is small and it is difficult to decipher the color code.
• Line 159 should be ‘similarly’.
• Line 218 please rephrase sentence.
• Line 378 ‘seams’.

 

Author Response

We thank the reviewer for the valuable input, which has definitely helped to improve our manuscript. The raised questions are answered in a point-by-point fashion and changes in the revised manuscript are highlighted using the “Track Changes” function.

Reviewer 3:

“In this review the authors discuss the use of immune checkpoint inhibitors (ICI) in cancer immunotherapy. Unlike most reviews of the field, which refer to clinical aspects of the use of ICI as monotherapy, this review focuses on promising strategies for combinational therapies employing ICI, as well as novel genetic engineering approaches and emerging prediction tools, all aiming at maximizing the clinical benefit of ICI treatment.

This is an outstanding review, which systemically covers a wide spectrum of approaches, going in each from basic immunological principles and how these had been elucidated in pioneering works to current technologies and achievements, including ongoing clinical trials. The idea of exploiting artificial intelligence to create a powerful prediction platform incorporating analytical tools with experimental and clinical data is intriguing and by itself justifies publication of this review.”

#1: “The only major reservation I have is that the review lacks a summarizing section/paragraph.”

Authors’ response:

A conclusion section has been added to the end of the review (page 24-25).

 

#2: “Fig. 1. Font is small and it is difficult to decipher the color code.”

Authors’ response:

The font and color code have been improved.

#3: “Line 159 should be ‘similarly’”

Authors’ response:

Similar has been changed to similarly.

 

#4: “Line 218 please rephrase sentence.”

Authors’ response:

The sentence has been rephrased to “Despite the success of vaccinations against microbes or viral diseases therapeutic vaccinations against cancer cells have not yielded similar success so far.”

 

#5: “Line 378 ‘seams’.”

 

Authors’ response:

Seams has been changed to seems.

Round 2

Reviewer 3 Report

All comments have been referred to satisfactorily.