TTN/OBSCN ‘Double‐Hit’ predicts favourable prognosis, ‘immune‐hot’ subtype and potentially better immunotherapeutic efficacy in colorectal cancer

Abstract Colorectal cancer (CRC) remains a leading cause of cancer‐related deaths worldwide. Although treatment strategies for solid tumours have been revolutionized by immunotherapy, only a small subset of CRC patients benefit. Using two‐independent cohorts, we found the common frequently mutated genes TTN and OBSCN had the significant correlation with higher tumour mutation burden (TMB) and favourable overall survival. TTN and OBSCN also displayed significant commutation phenomenon. Therefore, based on the status of TTN and OBSCN, we stratified patients into ‘Double‐WT’ phenotype, ‘Single‐Hit’ phenotype and ‘Double‐Hit’ phenotype. Importantly, the ‘Double‐Hit’ phenotype had favourable prognosis, low malignant events propensity, and highest TMB, immune cells infiltration abundance, POLE mutation rate, microsatellite instability ratio, as well as immune checkpoints expression compared with the other two phenotypes. These results indicated that the ‘Double‐Hit’ phenotype suggested ‘immune‐hot’ tumours and potentially better immunotherapeutic efficacy. Bioinformatic algorithm assessment of immunotherapy responses also confirmed this conclusion, and the ‘Double‐Hit’ phenotype was found to be a better predictor of immunotherapy than PD‐L1, PD‐1, CTLA‐4, TMB and microsatellite status. This study revealed CRC patients with TTN/OBSCN ‘Double‐Hit’ was significantly associated favourable prognosis, ‘immune‐hot’ subtype and potentially better immunotherapeutic efficacy.

Over the past decade, immunotherapy has illustrated tremendous sensation owing to its remarkable efficacy in the treatment of solid tumours. 3 Immune checkpoint inhibitors (ICIs) facilitate the immune system to recognize and inhibit the essential targets on tumour cells such as PD-L1, PD-1 and CTLA-4. 4 In CRC, immunotherapy was approved in 2017 for the therapy of patients with DNA mismatch repair deficient (dMMR) or advanced microsatellite instability (MSI). Apart from MSI, there are other classification systems to stratify patients, such as consensus molecular subtypes, tumour mutation burden (TMB), PD-1/PD-L1, CTLA-4 and tumour immune microenvironment. 3 Nevertheless, these classification methods are not perfectly predicting the responses to immunotherapy, and only a small subset of CRC patients benefit. 5 In view of the serious adverse reactions and high cost of immunotherapy, it is necessary to explore new biomarkers for effective immunotherapy management in CRC patients.
Genetic mutations in CRC were also assessment indicators for immunotherapy. 6 For example, patients with POLE mutation generally benefit more from ICIs therapy. TMB has also been re- In the present research, we firstly demonstrated that mutations in TTN and OBSCN were significantly related to TMB and overall survival (OS) in patients with CRC TTN and OBSCN also displayed significant commutation phenomenon. Therefore, according to the status of TTN and OBSCN, we stratified patients into 'Double-WT' phenotype, 'Single-Hit' phenotype and 'Double-Hit' phenotype.
The three phenotypes displayed substantial differences in prognosis, malignant events propensity, molecular characteristics and immune landscape, which indicated that each phenotype should require a specific therapeutic strategy. Notably, the 'Double-Hit' phenotype displayed the highest TMB, MSI, immune cells infiltration, ICPs expression, POLE mutation rate and immunotherapy response rate, suggesting 'immune-hot' tumours. These results indicated the 'Double-Hit' phenotype might be a reliable biomarker for predicting immunotherapy responses. Moreover, we identified potential chemotherapeutic agents with specific sensitivity among three phenotypes, which provided a resource for precision chemotherapy in the TTN/OBSCN mutant phenotypes.

| Data sources and processing
Somatic gene mutations data of American CRC patients (n = 535) and Chinese CRC patients (n = 318) were enrolled from TCGA (http://portal.gdc.cancer.gov/) and ICGC (http://dcc.icgc.org/) databases, respectively. The RNA-seq data (FPKM normalized) of TCGA-CRC cohort were derived from the UCSC-Xena database and were further transformed to log 2 (TPM + 1). CRC patients were removed if they lacked somatic mutations, RNA-seq, survival data and received neo-adjuvant therapy.

| Estimate TMB in terms of per megabase
Tumour mutation burden was measured as the number of coding, somatic, base substitution and indels mutations per megabase (MB) of the targeted territory. We counted all base substitutions and indels in the coding region of targeted genes. Synonymous mutations that do not result in amino acid variations are not counted. Using the 'tmb' function in 'maftools' R package, we calculated the TMB for each patient. 11

| Functional analysis and immune cells infiltration assessment
To further analyse the latent functions underlying the TTN/ONSCN 'Double-Hit' phenotype, the gene-set enrichment analysis (GSEA) algorithm was carried out to identify enriched dramatically terms correlated with Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway and biological process of gene ontology (GO). We set the number of random permutations as 1000 to generate a normalized enrichment score (NES) and false discovery rate (FDR) q value.
We deemed the terms with NES >2 and FDR q value <0.001 were strikingly enriched. Subsequently, the single-sample gene-set enrichment analysis (ssGSEA) approach was utilized to evaluate the infiltration abundance of 28 immune cells in the tumour microenvironment (TME). We retrieved the gene sets for annotating each TME cell from the study of Charoentong, which contained miscellaneous innate and adaptive immune cells such as CD4/CD8 + T cell, B cell, natural killer T cell, macrophage, dendritic cell and monocyte. 12

| Bioinformatic assessment of immunotherapy
The Subclass Mapping (SubMap) algorithm is an unsupervised clustering approach, which reveals common subtypes between independent cohorts. Using SubMap method, we measured the similarity pattern of mRNA expression between the three subtypes and the patients with different immunotherapy responses. 13 An FDR for the two subclasses less than 0.05 indicated a dramatical similarity. Subsequently, we used the Tumor Immune Dysfunction and Exclusion (TIDE) web tool (http://tide.dfci.harva rd.edu/) to predict ICIs clinical response based on the pre-treatment expression data of tumours. 14 The TIDE module assesses immune evasion by integrating the expression signatures of T cell dysfunction and T cell exclusion. Using the TIDE framework, we obtained the outcome of the bioinformatic assessment of immunotherapy for each patient.

| Chemotherapeutic response prediction
A previous work has proposed a ridge regression model to evaluate the imputed response to 138 chemotherapeutic agents based on pharmacogenomics and gene expression data. 15 The 'pRRophetic' R package was applied to perform the prediction process. The halfmaximal inhibitory concentration (IC50) was utilized to quantified drug sensitivity, and the lower the IC50, the higher the sensitivity.
In order to identify potential chemotherapeutic agents with specific sensitivity among three phenotypes through the following procedure: (a) The Kolmogorov-Smirnov normality test indicated that the imputed drug response (IC50) belonged to be a skewed distribution (P < .05), thus, Kruskal-Wallis and Wilcoxon rank-sum tests were applied to compare the differences; (b) The P-values were adjusted by Benjamini-Hochberg (BH) multiple test correction, and an FDR <0.05 was considered significant; (c) For each potential chemotherapeutic agent, if the Kruskal-Wallis test FDR < 0.05 and the IC50 of one subtype was predominantly lower than that of the other two subtypes (Wilcoxon rank-sum test FDR <0.05), the subtype was thought to be more sensitive to the drug; (d) In terms of the median IC50 value of an antitumour drug with specific sensitivity in each phenotype, the sensitivity of three phenotypes was labelled 'High sensitivity', 'Intermediate sensitivity' and 'Low sensitivity' according to the magnitude of the median IC50 value.

| Statistical analysis
The waterfall plot of FMGs was summarized and visualized with 'maftools' R package, and the Fisher exact test was applied to assess the co-occurrence or mutually exclusive significance of FMGs. The Kolmogorov-Smirnov normality test P-value of TMB, immune cells infiltration abundance, immune checkpoints (ICPs) expression and IC50 were all less than 0.05. Therefore, the Wilcoxon rank-sum and Kruskal-Wallis tests were performed to compare the differences of two and multiple groups, respectively. Comparisons between categorical variables using Fisher's exact test or chi-squared test. The 'clusterProfiler' R package was utilized to perform GSEA process. 16 The Kaplan-Meier analysis was performed via 'survival' R package, and the log-rank test was applied to compare the survival differences among three phenotypes. FDR was generated via BH multiple test correction. All P-values were two-sided, and P < .05 was considered significant. All data processing, statistical analysis and plotting were performed in R 4.0.2 software.

| TTN and OBSCN mutation associated with TMB and OS
The TMB in the TCGA cohort ranged from 0.  Figure 2A). Previous study has reported that the high-TMB is dramatically associated with a favourable prognosis in CRC. 17 Hence, Kaplan-Meier survival analysis was further carried out to determine whether mutations in these FMGs related to higher TMB were also associated with OS in CRC patients. As illustrated in Figure
After controlling for age, gender and TNM stage, the TTN/OBSCN mutant phenotypes remained statistically significance (P < .05), which suggested that the TTN/OBSCN mutant phenotypes was also an independent protective factor of OS in CRC ( Figure S1). There were no statistically significant differences in the age, sex or T stage of CRC patients among three phenotypes ( Figure 3D Figure S2). Collectively, these findings indicated that TTN/OBSCN 'Double-Hit' suggested 'immune-hot' tumours, which has potential implications for optimizing immunotherapy.

| Molecular characteristics and therapy implications of the TTN/OBSCN mutant phenotypes
Previous study has demonstrated that the mutations of some genes were predominantly associated with the initiation, progression and PIK3CA and POLE. 18 The inactive mutation of DNA mismatch repair (MMR) systems was reported to be dramatically correlated with MSI. 19 We calculated the mutation rate of these genes in the three TTN/OBSCN mutant phenotypes ( Figure 5A). Of note, the 'Double-Hit' phenotype of American patients had the lowest APC mutation rate, whereas the 'Double-Hit' phenotype of Chinese patients had the highest APC mutation rate ( Figure 5A). Mutations in TP53 were most significant in the 'Double-WT' phenotype, which was associated with higher recurrence rate, advance stage and higher mortality in CRC ( Figure 5A,C). 20 There were no significant differences in the mutation frequency of RAS genes (including KRAS, NRAS and HARS) among the three phenotypes ( Figure 5A). Mutations in SMAD4 have been proven to increase resistance to oxaliplatin in CRC patients 21 and were most common in 'Double-Hit' phenotype ( Figure 5A,C).
Previous study has reported that patients with BRAF or PIK3CA mutations had superior resistance to EGFR inhibitors. 22

| Assessment of immunotherapy and chemotherapy
We applied SubMap analysis to reveals common subtypes between independent cohorts and found significant expression similarity between the 'Double-Hit' phenotype and patients with anti-PD-L1 and anti-CTLA-4 therapy (FDR < 0.05; Figure 6C). This confirms the above conclusion that immunotherapy was more effective in patients with 'Double-Hit' phenotype than the other two phenotypes. The TIDE framework was further employed to evaluate the immunotherapy outcome of each patient, and we found the proportion of responders to immunotherapy with 'Double-Hit' phenotype was the highest relative to the other two phenotypes (Double-WT vs Single-Hit vs Double-Hit: 15% vs 19% vs 87%; P < .001; Figure 6B). In addition, CRC patients were divided into high-PD-L1 and low-PD-L1 or high-PD-1 and low-PD-1 or high-CTLA-4 and low-CTLA-4 or high-TMB and low-TMB or MSS and MSI groups by the median values of PD-L1, PD-1, CTLA-4 and TMB or the status of microsatellite. In line with previous study, patients with high-PD-L1, PD-1, CTLA-4, TMB and MSI had more immunotherapy response rate compared with the other groups ( Figure 6C). 3 Notably, the proportion of responders with the 'Double-Hit' phenotype was significantly higher than that of high-PD-L1, high-PD-1, high-CTLA-4, high-TMB and MSI groups. Furthermore, using 'pRRophetic' R package, we evaluated the imputed responses to 138 chemotherapeutic agents in CRC patients to identify potential drugs with specific sensitivity to each phenotype. 15 As displayed in Figure 6D, A total of 59 drugs were determined. For instance, MG-132 targeting proteasome was more sensitive to patients with 'Double-WT' phenotype; OSI-906 targeting IGF1R signalling was more sensitive to patients with 'Single-Hit' phenotype. Importantly, fifty-six of these 59 drugs were specific sensitivity to patients with 'Double-Hit' phenotype ( Figure 6D). Of these 56 drugs, the majority of their targeted pathways were associated with tumour proliferation, such as cell cycle (seven drugs), DNA replication (seven drugs), mitosis (seven drugs), P53 pathway (two drugs) and PI3K/MTOR signalling (four drugs). Other pathways such as MAPK signalling (two drugs), kinases (five drugs), protein stability and degradation (five drugs) and RTK signalling (three drugs) were latent targets for patients with 'Double-Hit' phenotype ( Figure 6D).
These results offered many opportunities for potential targeted therapies in the 'Double-Hit' phenotype. Our works provided a resource for precision chemotherapy in the TTN/OBSCN mutant phenotypes.

| D ISCUSS I ON
In the present study, we identified the mutations of TTN and OBSCN were significantly associated with higher TMB and favourable OS.
Given the significant commutation phenomenon of TTN and OBSCN, CRC patients were stratified into 'Double-WT', 'Single-Hit' and 'Double-Hit' phenotypes. The three phenotypes presented distinct clinical and molecular characteristics as well as immune landscape, suggesting specific therapeutic and management strategies required for each phenotype. Importantly, patients with TTN/OBSCN 'Double-Hit' belonged to 'immune-hot' tumours and potentially better immunotherapeutic efficacy. We also identified plenty of latent chemotherapeutic agents with specific sensitivity among three phenotypes, providing a resource for precision chemotherapy in the TTN/OBSCN mutant phenotypes. Collectively, this study underlined F I G U R E 5 Molecular and immune landscape of three TTN/OBSCN mutant phenotypes. A, Mutation rate of driver genes among three TTN/OBSCN mutant phenotypes in TCGA and ICGC cohorts. B, Expression distribution of immune checkpoints in three phenotypes. *P < .05; **P < .01; ***P < .001. C, Summary of molecular and immune landscape of three phenotypes F I G U R E 6 Assessment of immunotherapy and chemotherapy among three phenotypes. A, SubMap algorithm evaluated the expression similarity between the three phenotypes and the patients with different immunotherapy responses. B, Distribution of immunotherapy responders predicted by TIDE algorithm in three phenotypes. C, Distribution of immunotherapy responders in different groups. D, A total of 59 potential antitumour drugs with specific sensitivity to each phenotype were identified. The left panel is the drug names and the level of sensitivity in each phenotype, the middle panel is the drug-targeted molecules, whereas the right panel represents the drug-targeted pathways the importance of FMGs in CRC and proposed a novel mutational classification for clinical translation, especially for predicting the responses to immunotherapy.
Both TTN and OBSCN belong to the family of giant sarcomere signalling proteins and have a role in the organization of myofibrils during assembly, which may mediate interactions between the sarcoplasmic reticulum and myofibrils. 25,26 Previous study has demonstrated that TTN and OBSCN were driver genes of CRC, but their clinical significance has not been elaborated. 27  which is associated with tumour progression, high recurrence and high mortality. 20 Tumour suppressor SMAD4 is an essential modulator of the transforming growth factor (TGF)-beta pathway controlling proliferation. Mutations in SMAD4 indicate resistance to oxaliplatinbased chemotherapy, 21 suggesting that the 'Double-Hit' phenotype with the most SMAD4 mutations might be the least sensitive to oxaliplatin than the other two phenotypes. BRAF and PIK3CA are key driven genes in CRC, and their mutations are significantly associated with resistance to anti-EGFR therapy such as cetuximab and panitumumab. 22,23 Obviously, the mutation rate of BRAF and PIK3CA in the 'Double-Hit' phenotype was 3-9 times higher than that in the other two phenotypes. These results suggested that patients with the 'Double-Hit' phenotype may benefit less from first-line chemotherapy for CRC such as oxaliplatin, cetuximab and panitumumab.
Subsequently, as a supplement, we identified plenty of potential antitumour drugs with specific sensitivity to each phenotype, and most drugs were more sensitive to the 'Double-Hit' phenotype. These results offered many opportunities for potential targeted therapies in the 'Double-Hit' phenotype. Therefore, after first-line chemotherapy fails, patients with the 'Double-Hit' phenotype have many other potential chemotherapeutic options other than immunotherapy alone, because combination therapy may achieve better efficacy.

CO N FLI C T O F I NTE R E S T
The authors declare that they have no competing interests.

DATA AVA I L A B I L I T Y S TAT E M E N T
Public data used in this work can be acquired from TCGA (http:// portal.gdc.cancer.gov/) and ICGC (http://dcc.icgc.org/) data sets.