Gastric Cancer TMEscore Ultimate: Combined with ICB Cohort Analysis

Last week, the public account also shared the second TMEscore of gastric cancer: What is the difference between the same gastric cancer TMEscore having 8+ after two years? Now the third chapter is here, this time it is a combined immunotherapy cohort analysis
.

More innovative TMEscore ideas scan code consultation Tumor microenvironment assessment to promote the precise treatment of advanced gastric cancer immune checkpoints.
This article was published in the Journal for Immunotherapy of Cancer.
The impact factor of the journal in the last year: 13.
751, compared with the previous year Increased 3.
838
.

Chinese Academy of Sciences: Medical Division 2
.

Chinese Academy of Sciences subcategory: Division 2 Immunology
.

Background knowledge: Immune checkpoint blockade (ICB) continues to be effective in a small number of patients with metastatic gastric cancer (MGC)
.

The author developed an open source TMEcore R package to quantify the tumor microenvironment (TME) to help solve this dilemma
.

Two advanced gastric cancer cohorts (RNAseq, N=45 and NanoString, N=48) and other advanced gastric cancer (N=534) received ICB treatment to study the predictive value of TMEScore
.

At the same time, multi-omics data from the Gastric Adenocarcinoma Cancer Genome Atlas (TCGA-STAD) and the Asian Cancer Research Group (ACRG) were asked to find the underlying mechanism
.

Problem found: The decisive biomarker of immune checkpoint block (ICB) response is unclear
.

Interpretation of results: TME value predicts the efficacy of ICB in gastric cancer.
In order to optimize the evaluation of TME for more effective clinical applications, the authors conducted feature screening in 6 ICB data sets and reduced the TMEscore signature genes from 244 to 44
.

As previous studies have shown, genes negatively related to ICB treatment response are enriched in the immune rejection phenotype (EMT/transforming growth factor-β pathway), while immune-related genes are positively related to efficacy (figure 1A)
.

In several GC cohorts, the authors found consistent and close correlations between the TME values ​​of 44 genes and the previously measured TME values ​​of 244 genes
.

It is worth noting that the TME score can be used as a prognostic biomarker for immunotherapy quasi-cohorts (figure 1B)
.

In the advanced GC cohort that received anti-PD-1 immunotherapy, the TME score produced the highest AUC (AUC=0.
891), surpassing other mainstream biomarkers, including MSI status, TME, CPS, and Epstein-Barr virus infection (AUC= 0.
708, 0.
672, 0.
817, and 0.
708) (figure 1C)
.

In addition, TME score corresponds to several transcription-based prediction counterparts, comprehensive gene expression profile score (GEP), immune score, CD8+T effect score, and pan-fibroblast transforming growth factor-β response characteristics (PAN-F-TBR) In comparison, the TME score is also the highest (figure 1D)
.

The authors further used the NanoString nCounter platform to isolate RNA from baseline tumor tissues of 48 patients with multi-advanced gastric cancer before ICB to further determine the expression of TMEcore gene in the tumor microenvironment
.

Obviously, the TME score has reached the overall accuracy of AUC=0.
877, which is higher than the predicted value of other popular gene signatures (figure 1E, F)
.

Consistent with the author’s previous studies, the TMEscoreA score of degenerative tumors (complete remission (CR)/partial remission (PR)) is significantly higher than that of stable and advanced tumors (progressive disease (PD)/stable disease (SD)), TMEscoreB is negatively correlated with the treatment effect of advanced GC (figure 1F), suggesting that interstitial activation is an important mechanism of ICB resistance
.

The TMEscoreB (matrix-related) gene is a more accurate biomarker and is significantly related to treatment resistance, while the TMEscoreA (immune-related) gene is highly expressed in a small number of non-responders (figure 1G, H)
.

Figure 1 TMEscore holds promise in predicting immunotherapeutic response.
TMEscore predicts the efficacy of single checkpoint immunotherapy or combined chemotherapy or angiogenesis inhibitors.
In order to provide an accurate picture to understand the performance of TMEscore in single and combined immunotherapies, the author further studied NanoString results of 48 gastric cancer patients
.

The expression of PD-L1 is mainly concentrated in a subset (CR/PR) relative to patients in the advanced stage, rather than in the middle stage (figure 2A-C)
.

PD-L2 and TIM3 are significantly elevated in unresponsive tumors, indicating that the upregulation of other corresponding or bypass checkpoint pathways may resist PD-1 blockers (figure 2B–D), and the mechanism may be induced interstitial activation And T cell rejection
.

In addition, it is reported that SYNPO is up-regulated during CAF activation, which is a key mechanism of ICB resistance
.

ICB monotherapy has limited clinical benefits in the treatment of advanced gastric cancer.
Recent clinical trials have shown that ICBs combined with chemotherapy, anti-vascular targeted therapy or other molecular targeted therapies such as CHECKMate-649 can significantly improve the treatment results
.

Most ICB-related genes and immune-related genes are positively correlated with a single immunotherapy response (figure 2E, F)
.

In the combination therapy subgroup, especially markers related to immune activation, its predictive effect has declined significantly (figure 2G, H)
.

However, TME scores still have strong predictive power in these two groups
.

It may be attributed to the most important influence exerted by matrix activation in the course of synergistic treatment
.

In the co-treatment, the expression of PD-L2 and TIM3 also showed a similar trend
.

Their up-regulation in advanced patients suggests a potential key molecular feature (figure 2G, I) in the formation of tumor immune evasion
.

This also implies the existence of simultaneous up-regulation of immune checkpoint-related genes, suggesting that this subgroup of patients may benefit from PD-L2 or TIM3 treatment
.

Figure 2 TMEscore predicts efficacy of checkpoint immunotherapy alone or combination with chemotherapy.
In MGC, TMEscore can accurately identify more patients than MSI, EBV and TMB.
In order to systematically evaluate the predictive value and potential mechanism of TMEscore in advanced gastric cancer, the author accepts Multiple omics data of patients with advanced gastric cancer who were rescued by pebrizumab were analyzed comprehensively
.

Compared with TME score (AUC=0.
921) alone, TME score combined with TMB or CP (AUC value 0.
964, 0.
973, respectively), a slight increase in AUC value was observed
.

Although no statistically significant differences were observed in the pairwise comparisons
.

In the Kim cohort, TME values ​​were not related to tumor somatic mutations and histological subtypes (figure 2J, K)
.

However, when it comes to the level of some biomarkers related to ICB response, such as tumor PD-L1 expression, CPS, MSI status and EBV infection are used to evaluate (figure 2L-N)
.

The author's analysis shows that compared with the inherent genomic characteristics of tumors, TME score can be used as a reliable biomarker for predicting ICB response in advanced gastric cancer
.

The author described the TME feature scores, clinicopathological features, and molecular features of patients with metastatic gastric cancer treated with anti-PD-1 immunotherapy to investigate factors that may be related to the effect of ICB treatment
.

The authors observed that patients with better responses are more likely to have EBV and MSI-H molecular subtypes, but are rarely rich in chromosomal instability (CIN), genome stable (GS), and EMT molecular subtypes (figure 3A)
.

Consistent with the authors' recent studies in the TCGA-STAD and ACRG cohorts, the TME values ​​of patients with MSI-H and EBV subtypes were significantly higher than CIN and GS (figure 3B)
.

It is suggested that the predictability of TME value is mainly attributed to molecular phenotypic stratification
.

Next, the authors examined the predictive power of genetic signatures and epidemic biomarkers in patients stratified by EBV and MSI-H molecular subtypes, and reportedly showed a better response to ICBS
.

ROC analysis showed that TME value (AUC=0.
895) predicts EBV and MSI-H molecular subtypes better than MSI status, TMB, CPS, EBV status, GEPs, ImmuneScore, Pan-F-TBR and immune checkpoints
.

To verify the above findings, the authors performed the same statistical analysis in two large multi-omics GC cohorts
.

Next, the author will focus on the TCGA-STAD cohort and analyze its clinical characteristics (figure 3C)
.

In the low TME group, most of the MSI and EBV subtypes were missing, while in the high TME group, the MSI and EBV subtypes accounted for a large proportion (figure 3D)
.

The author's analysis showed that in the ACRG cohort, the TME values ​​of EBV-infected tumors and MSI-H tumors were comparable (figure 3E)
.

In the TCGA cohort, its TME value was even higher than that of MSI-H tumor (figure 3F).
However, in the TCGA-STAD and ACRG cohort, the tumor mutation count was still significantly lower than that of MSI-H tumor (figure 3G, H).
The author also noted In the TCGA cohort, compared with the GC of MSI-H, the neoantigen load in EBV-infected tumors was significantly reduced (figure 3I)
.

Correlation analysis showed that in the two sets of data, the TME value was positively correlated with tumor mutation burden (figure 3J)
.

Overall, in the Pan-Caner cohort, EBV subtypes remained at low levels of TMB and neoantigens, which are associated with higher TME values ​​and immune-related signatures
.

Previous studies have shown that when gastric cancer patients receive ICBS treatment, patients with EBV infection and MSI-H phenotype are more likely to benefit from ICB treatment
.

These observations further confirm that TMB, as a widely used predictive biomarker, cannot identify GC patients with EBV subtypes and tumor patients with viral infections
.

As expected, the TME score can identify the EBV and MSI subtypes of all patients in the TCGA-STAD and ACRG cohorts, and the accuracy rate is significantly higher than that of TMB, GEP, 1Pan-F-TBRs, and immune check scores (figure 3K)
.

Figure 3 TMEscore is closely correlated with microsatellite instability-high (MSI- H) and Epstein- Barr virus (EBV) infective status in gastric cancer.
ARID1A and PIK3CA defects enhance the therapeutic anti-tumor immunity of gastric cancer.
Mutations in somatic cells can alter cancer cells Susceptibility to T cells and T cell immunotherapy
.

In two large patient cohorts, the authors tried to reveal the immune genomic determinants of tumor treatment response and the activation of GC's tumor immune microenvironment
.

Wilcoxon test and Fisher's exact test were used to identify mutations associated with TMEScore (figure 4A)
.

The author's analysis highlights mutations in ARID1A and PIK3CA (figure 4A)
.

The authors found that ARID1A and PIK3CA were significantly correlated with the level of TMEScore in the TCGA-STAD cohort, whether it was continuous evaluation or binary evaluation (figure 4B, C), which was verified in the ACRG cohort
.

At the same time, TMB was divided into high TMB group and low TMB group (cutoff value=400), and the relationship between TME score and ARID1A or PIK3CA mutation was analyzed
.

In the low TMB group, patients with ARID1A or PIK3CA mutations had significantly higher TME values
.

However, no obvious trend was observed in the high TMB group
.

The above results indicate that under low TMB conditions, both ARID1A and PIK3CA mutations are related to TME activation, while under high TMB conditions, the effects of ARID1A and PIK3CA mutations may be overwhelmed by the increase of new antigens caused by a large number of mutations to further activate TME
.

PIK3CA is the most common mutated oncogene in all solid tumors
.

ARID1A deficiency is also a common mutation in a variety of malignant tumors.
It has been reported to cause genomic characteristics of impaired mismatch repair (MMR), increased mutations and microsatellite instability, and may have a synergistic effect with anti-PD-L1 therapy
.

It is worth noting that the author further investigated the specific mutation location to determine the repetitive mutation with the highest mutation frequency in the binary TMEScore setting, so that the results can be visualized by trackViewer
.

Interestingly, in tumors with high TME values ​​(figure 4D), ARID1A mutations p.
D18550Tfs*33 and p.
F2141Sfs*59 are highlighted (figure 4D)
.

And it is closely related to the level of TMEScore (figure 4E)
.

Gastric cancers with p.
E545K and p.
H1047R mutations of PIK3CA were significantly enriched in the high-TMEscore group
.

However, despite the significant difference between the mutant and wild type, the statistical differences observed in consecutive TME values ​​are limited
.

Two recent studies have shown that mutations in signaling pathways can be used as biomarkers for immunotherapy and suggest the opportunity for combination therapy
.

Current research shows that pathway mutations are mainly derived from MSI molecular subtypes (figure 4F)
.

And in the high TME score group, almost all pathways have significant accumulation of mutations (figure 4F)
.

However, based on previous results, compared with GS and CIN subtypes, a higher frequency of PI3K pathway mutations was also observed in EBV subtypes, indicating that there is a potential interaction between EBV infection and PI3K signaling pathways, which may be partly Explains the reason for the significant increase in TME in patients with EBV infection (figure 3F)
.

In conclusion, the analysis of a large number of data of gastric cancer TME clarified that the assessment of ARID1A and PIK3CA mutation status can be used as potential biomarkers for gastric cancer immunotherapy strategies
.

Figure 4 ARID1A and PIK3CA mutation potentiate antitumor immunity.
TME-related metabolic characteristics.
Considering the interesting metabolic laws observed under different ARID1A mutation states, we further explored the transcription profile and analyzed the potential inherent influencing the key predictive ability of TME score Mechanism
.

Metabolic characteristics were estimated by the PCA method and comprehensively studied in the TCGA-STAD cohort
.

Correlation analysis showed that kynurenine metabolism, purine metabolism and cysteine ​​metabolism were activated in the high TME group, while glycogen metabolism, transsulfuration and glycine serine metabolism were significantly up-regulated in the low TME group
.

(figure 5A, B) Statistical analysis shows that kynurenine metabolism is closely related to high TME and immunotherapy-favorable molecular subtypes (including EBV and MSI-H) (figure 5C)
.

In addition, down-regulation of kynurenine metabolism was observed, suggesting rejection of T cells, which may indicate insensitivity to ICB treatment (figure 5E)
.

The metabolic process of kynurenine may restore the immunogenicity of tumor suppressor T cells, thereby improving the efficacy of ICB (such as DO1 inhibitors) in the treatment of gastric cancer
.

The authors observed that in the TCGA-STAD cohort and ACRG cohort, the glycogen metabolism of tumors and immune rejection molecular subtypes with low TME values ​​was significantly activated (figure 5D)
.

This suggests that glycogen metabolism may be closely related to the immune rejection phenotype
.

In summary, the authors identified a collection of metabolic characteristics and biological processes associated with TME, which reflected the complexity of TME and pointed out potential combined treatment opportunities
.

A previous study showed that the methylation region is associated with immune activity.
A high m6A value means immune rejection of the TME phenotype, interstitial activation, reduced survival rate, and reduced neoantigen load.
Thereafter, the author tried to determine its involvement in anti-tumor immunity and The apparent immunomodulation of tumor immunoediting may be the basis for understanding the inflammatory response that occurs in the disease
.

A comprehensive study of the location pattern of DNA methylation showed that the demethylation of VAMP8 was enriched in clusters with low TME values, while the demethylation of ATG7 was enriched in clusters with high TME values ​​(figure 5F–I)
.

The enrichment of differentially methylated genes highlights the importance of VAMP8 methylation in the TME regulatory network by up-regulating immune pathways, compressing leukocyte activation regulatory pathways, protein membrane localization, antigen processing and presentation, coating vesicles and recycling endosomes Role
.

This indicates that VAMP8 plays a key role in complex gene interactions and crosstalk in a wide range of signaling pathways
.

In addition, ATG7 demethylation, as a gene marker of autophagy, is significantly related to TMEScore
.

Further analysis of the relationship between ATG7-related signals (positive regulation of autophagy) that has been found indicates that ATG7 demethylation is one of the causes of TME immune rejection, TMEcore B increases, and fibrosis in the TCGA-STAD and ACRG cohorts Cell infiltration
.

All in all, DNA methylation, such as the different methylation regions of VAMP8 and ATG7, may provide an insight into the complexity and diversity of TME and immunological activity assays, which may help optimize immunotherapy strategies thereafter
.

Figure 5 Tumor microenvironment (TME) associated metabolism and methylation characteristics.
Summary of this article: The author optimized a TME assessment tool, which can be used as a powerful biomarker and integrated into an open source R package for further clinical application Implement
.

The predictive ability of TMEScore was verified in two cohorts of advanced gastric cancer, highlighting the predictive effect of tumor microenvironment assessment
.

The intrinsic characteristics involving ARID1A and PIK3CA mutations, kynurenine metabolism, glycogen metabolism, ATG7 and VAMP8 methylation provide new insights into the underlying mechanisms of precision immunotherapy guided by TME scores
.

Reprinting is not allowed without permission.
For friendly copy, please contact the good talking editor butler at 15510012760 (same number on WeChat).
Innovative TMEscore idea.
Scan code to consult Youshengxin related questions.
Contact: 18501230653 (same number for WeChat).
Sequencing | Single cell | m6A | Multiomics cytoscape | limma | WGCNA | Water bear legend | Linux electrophoresis | PCR | A brief history of sequencing | Karyotype | NIPT | Basic experimental genes | 2019-nCoV | Enrichment analysis| Joint analysis|Micro-environment plague chasing the killer| Summary of ideas| Scholars| Scientific research| Retraction| PhD reading|Work