Evolution of genome and immunogenome in esophageal squamous cell carcinomas driven by neoadjuvant chemoradiotherapy

doi:10.21203/rs.3.rs-3912458/v1

Download PDF

Research Article

Evolution of genome and immunogenome in esophageal squamous cell carcinomas driven by neoadjuvant chemoradiotherapy

https://doi.org/10.21203/rs.3.rs-3912458/v1

This work is licensed under a CC BY 4.0 License

Version 1

posted

You are reading this latest preprint version

Background

Neoadjuvant chemoradiotherapy (NCRT) followed by surgery is a standard treatment for locally advanced esophageal squamous cell carcinomas (ESCCs). However, evolution of genome and immunogenome in ESCCs driven by NCRT remain incompletely elucidated.

Methods

We performed whole-exome sequencing of 51 ESCC tumors collected before and after NCRT, 36 of which were subjected to transcriptome sequencing.

Results

Clonal analysis identified clonal extinction in 13 ESCC patients wherein all pre-NCRT clones disappeared after NCRT, and clonal persistence in 9 patients wherein clones endured following NCRT. Clone-persistent patients showed higher pre-NCRT genomic intratumoral heterogeneity and worse prognosis than clone-extinct ones. In contrast to clone-extinct patients, clone-persistent patients demonstrated a high proportion of subclonal neoantigens within pre-treatment specimens. Transcriptome analysis revealed increased immune infiltrations and up-regulated immune-related pathways after NCRT, especially in clone-extinct patients. The number of T cell receptor-neoantigen interactions were higher in clone-extinct patients than clone-extinct ones. Decrease in T cell repertoire evenness positively correlated to the decreased number of clonal neoantigens after NCRT, especifically in clone-extinct patients.

Conclusions

We identified two prognosis-related clonal dynamic modes driven by NCRT in ESCCs. This study extended our knowledge in the field of ESCC genome and immunogenome evolutions driven by NCRT.

esophageal squamous cell carcinoma

neoadjuvant chemoradiotherapy

genome

immunogenome

Esophageal cancer is the sixth most frequent cause of cancer-related deaths worldwide causing more than half a million cancer-related deaths annually¹. As the main pathological type, esophageal squamous cell carcinomas (ESCCs) account for approximately 85% of esophageal cancers². For patients with locally advanced ESCCs, neoadjuvant chemoradiotherapy (NCRT) followed by surgery is proved to be the standard treatment³. However, the outcomes of NCRT are heterogeneous, with better survivals in patients achieving major histomorphologic response after NCRT⁴. Many patients are still underserved by current treatment, partially due to our poor understandings of the antitumor effects of NCRT and tumor evolution during NCRT, which have essential implications in treatment response.

Genomic alterations account for critical drivers of tumor evolution, and employing clonal and mutational signature analyses helps reveal their influence on tumor progression as well as therapeutic resistance⁵. Clonal analysis determines the clonal architecture through clustering the cancer cell fractions (CCFs) of somatic nonsynonymous mutations and infers relationships among clones through the construction of phylogenetic trees⁶. Genomic variations originate from diverse mutational processes, which characteristically yield combinations of distinct mutation types, denoted as "signatures"⁷. Mutational signatures offer valuable insights into the particular mutational processes acting upon tumor genomes, as well as the dynamic alterations in the activity of these processes⁷. The analysis of genome evolution has been performed in a variety of solid tumors under treatment pressures. In triple-negative breast cancers, Kim et al.⁸ found that resistant genotypes were pre-existing and adaptively selected by neoadjuvant chemotherapy. However, in esophageal adenocarcinomas, post-chemotherapy samples could acquire mutations in known cancer driver genes that were absent from the paired pre-treatment sample⁹. In a previous study of whole-exome sequencing (WES) analyses of pretreatment and locally recurrent lesions from five ESCC cases received radical chemoradiotherapy (CRT), Hirata et al.¹⁰ found that most driver gene-altered clones remained under CRT selection pressure, while few driver gene alterations were acquired at recurrence.

In addition to the genetic mechanisms, there are increasing evidences that the interaction between cancer cells and the immune-microenvironment affect treatment response¹¹. Park et al.¹² found that gene sets related to immune signaling pathways, as well as the proportion of neutrophils, were enriched in ESCCs after CRT, and increased immune score after CRT was associated with favorable survival outcomes. The tumor neoantigens arising from somatic nonsynonymous mutations during anti-cancer treatment play a crucial role in initiating anti-cancer immunity, especially T cell immunity¹³. However, factors affecting the immunogenicity of neoantigens during NCRT, such as their presentation on tumor cells, and the extent of T cell expansion after neoantigen recognition, have not been systematically elucidated in ESCCs.

In our study, based on WES and transcriptome profile of paired pre- and post-NCRT specimens, we investigated the correlations between patients' prognosis and genome alterations of tumors, such as clonal evolution and mutational signatures. Additionally, we identified the interplay among neoantigens landscape, tumor microenvironment, and clonal dynamics during NCRT. Our results increased understanding of therapeutic mechanisms of NCRT and supported the rationale for precision NCRT.

Genome landscapes of ESCCs before and after NCRT

WES profiling data of 22 pairs of pre- and post-NCRT tumor samples, 7 additional pre-NCRT tumor samples, and corresponding 29 pre-NCRT white blood cell (WBC) samples were obtained from 29 patients with locally advanced ESCCs who received NCRT and surgery (Fig. 1A, Supplemental Table 1). There was no significant correlation between tumor regression grades (TRGs) and NCRT chemotherapy regimens as well as baseline clinical characteristics (all P > 0.05). The median sequencing depth was 250-fold (range: 228–363) for the 29 pre-NCRT tumors, 238.5-fold (range: 221–496) for the 22 post-NCRT tumors, and 143-fold (range: 116–214) for the corresponding 29 pre-NCRT WBC samples (Supplemental Table 2).

WES of 51 tumor specimens identified a total of 3,562 somatic nonsynonymous mutations involving 2,704 genes (Supplemental Table 3). The mutation rates of known ESCC driver genes in pre-NCRT samples were similar to those in the TCGA ESCC dataset and the largest whole genome sequencing dataset of ESCC published to date¹⁴, including TP53 (90%), NOTCH1 (21%), AJUBA (14%), and NFE2L2 (14%) (Supplemental Fig. 1A). Distinct alterations in driver mutations were observed between pre- and post-treatment samples. Although the majority of these mutations vanished following treatment, a subset either persisted during NCRT or emerged after NCRT (Fig. 1B). Notably, in post-NCRT samples with histopathological purity estimation at 0% (TRG0, n = 3), WES detected low-frequency residual mutations, which might be due to sampling bias between samples used for sequencing and pathological examination. Besides, it is possible that these post-NCRT TRG0 samples contained a few cells with genomic changes but not morphological changes that could be detected by traditional histopathology, which was reported in TRG0 patients after neoadjuvant treatment in non-small cell lung cancers¹⁵. Our observation is consistent with previous reports that TRG0 ESCC patients may nonetheless experience tumor recurrence after NCRT and surgery¹⁶.

Somatic nonsynonymous mutations were detected in 22 post-NCRT samples with a median tumor mutation burden (TMB) value of 14 (range: 3–93), significantly lower than that (median: 104.5, range: 29–224) in paired pre-NCRT samples (P < 0.001, Fig. 1C). Significant decrease in tumor purity estimated by ABSOLUTE¹⁷ was also observed after NCRT (P = 0.020, Fig. 1C). To figure out whether the decrease in TMB was due to decreased tumor purity or not, we conducted a simulation analysis¹⁸ and found that the observed TMB per post-NCRT sample was lower than expected one, indicating that magnitude of TMB decrease could not simply attributed to NCRT-induced tumor shrinkage (P = 0.006, Supplemental Fig. 1B).

As for the somatic copy number variation (SCNV) analysis, we identified an average of 1,851.0 MB amplification and 53.5 MB deletion in 22 pre-NCRT tumors, which reduced to 871.4 MB amplification and 27.9 MB deletion in paired post-NCRT tumors (P < 0.001 and P = 0.272, respectively, Supplemental Fig. 1C). The most recurrent SCNVs in 22 pre-NCRT samples were similar to the results identified previously in treatment-naive ESCC genome¹⁹, including the amplifications of 11q11.3 and 8q24.3, as well as the deletions of 9q21.3 and 21q21.1 (Fig. 1D). However, SCNV patterns significantly changed in post-NCRT samples, in which the most recurrent SCNVs were amplifications of 9q24.3, 8q21.1, 4q35.2 and 1q36.22 and deletion of 12q21.31 (Fig. 1D).

Genomic predictors of response to NCRT were intended to be identified in pre-NCRT samples. In 29 pre-treatment samples, NFE2L2 mutations tended to be enriched more in patients who achieved TRG2/3 than those TRG0/1 (P = 0.076, Supplemental Fig. 1D), suggesting a potential association between NFE2L2 mutations and resistance to NCRT.

Clonal evolution of ESCC under NCRT

The CCF of a somatic nonsynonymous mutation was estimated after copy-number and purity normalization²⁰. Using the assumption that somatic nonsynonymous mutations with similar CCFs represented a tumor clone²¹, unsupervised clustering of the CCFs of somatic nonsynonymous mutations identified the average of 2.40 clones (range: 1–8) in 22 pre-NCRT samples, which were much higher than the average of 1.73 clones (range: 1–7) (P = 0.039) in paired post-NCRT samples. Genomic intratumoral heterogeneity (ITH) for each sample was evaluated by Shannon index²² based on inferred clone architecture, which was positively correlated with single cell RNA sequencing (scRNA)-ITH estimated by the tumor cell-specific transcriptomic diversity scores among 8 samples with both WES and scRNA profiles available in our previous work²³ (rho = 0.795, P = 0.018, Supplemental Fig. 2A). With the median of genomic ITH as the cutoff point, 29 pre-NCRT samples were classified into two groups, and the patients with high pre-NCRT genomic ITH had worse survival than those with low pre-NCRT genomic ITH [P = 0.026 and P < 0.001 for overall survival (OS) and disease free survival (DFS), respectively, Fig. 2A]. However, patients evaluated as TRG0/1 or TRG2/3 showed no statistical difference in either prognosis (P = 0.470 for OS and P = 0.157 for DFS, repectively, Supplemental Fig. 2B) or pre-NCRT genomic ITH (P = 0.295, Supplemental Fig. 2C).

The frequency variation of tumor clone in pre- and post-NCRT samples provides an evolutionary perspective. We identified two clonal dynamics in 22 patients with paired tumors: 13 patients exhibited clonal extinction, whose clones in the pre-treatment samples all disappeared after NCRT (Fig. 2B left, Supplemental Fig. 2D), while 9 patients exhibited clonal persistence, who had at least one pre-NCRT clone persisted during NCRT (Fig. 2B right, Supplemental Fig. 2D). We utilized fishplot²⁴ to visualize temporal clonal evolution. In a representative case (P1) from the clone-extinct group, the pre-NCRT clone (Clone 1) extinguished after NCRT. And the post-NCRT-specific clone (Clone 2) seemed to be newly developed during NCRT (Fig. 2C left), although it is possible that Clone 2 existed in the pre-NCRT sample at a very low frequency, which could not be detected due to the limited WES sequencing depth (median 250-fold). However, in the clone-persistent P14 case, pre-NCRT-existed Clone 2 and 4 remained present in the post-NCRT sample with shifted clonal frequencies (Fig. 2C right). Patients experiencing these two evolution modes showed distinct outcomes, with those in clone-persistent group having a worse prognosis (P = 0.014 for OS and P = 0.014 for DFS, Fig. 2D). However, there was no significant correlation between clonal evolution modes and pre-NCRT clinical characteristics, chemotherapy regimens, post-NCRT TRGs, or the time spans from the end of NCRT to esophagectomy (Supplemental Table 4). To validate the prognostic implications of the clonal dynamics, we performed clonal dynamic analyses in another cohort of 21 ESCC patients with WES data of paired pre- and post-treatment samples available.²⁵ Among the 21 ESCCs, 12 patients were with clonal extinction while 9 with clonal persistence after treatment (Supplemental Fig. 3A). And the clone-persistent patients also had a worse prognosis than the clone-extinct patients (P = 0.001 for OS and P = 0.033 for local regional recurrence survival, Supplemental Fig. 3B). Consistent with the results in our data, the patients with high pre-NCRT genomic ITH had worse survival than those with low pre-NCRT genomic ITH (P = 0.001 and P = 0.016 for OS and local regional recurrence survival, respectively, Supplemental Fig. 3C).

Mutational landscape was further compared between clone-extinct and clone-persistent patients. No single mutation with statistical bias towards either clone-extinct or clone-persistent groups was found in pre-NCRT samples, however, mutations in TP53, NBPF20, and NOTCH1 were only identified in post-NCRT samples from clone-persistent patients (Supplemental Fig. 4A). Post-NCRT mutations in clone-extinct group were absent in paired pre-NCRT samples (Supplemental Fig. 4B), while most of post-NCRT mutations in clone-persistent group were shared by paired pre-NCRT sample (Supplemental Fig. 4C). To dissect the functional impact of these persistent mutations in clone-persistent patients, we conducted pathway enrichment analysis using REACTOME database²⁶, and identified the enrichment of mutations in pathways involved in NOTCH signaling pathways, apoptosis and programmed cell death, and extracellular matrix interaction, suggesting that these pathways may have an important role in the NCRT-driven clonal persistence (Fig. 2E).

Besides, the pre-NCRT genomic ITH was significantly correlated with the different clonal dynamic modes (P < 0.001, Fig. 2F). Further, the receiver operating characteristic (ROC) curve showed a good performance of pre-NCRT genomic ITH status in discriminating clonal dynamic modes with an area under the curve (AUC) of 0.906 (95% confidence interval: 0.774–1.00, Fig. 2G).

Mutational signatures related to NCRT-driven clonal evolution

Different mutational processes often generate various combinations of mutation types, termed mutational signatures, which can provide an etiological perspective on clonal evolution⁷. To characterize the evolution of mutational signatures during NCRT, somatic mutations were decoded into six classes of base substitution. After NCRT, significant decreases in C > T, C > G and C > A mutations and significant increases in T > C and T > G mutations were observed (all P < 0.05, Fig. 3A). Incorporating information on the bases immediately 5′ and 3′ to each mutated base, 96 mutation classes were further identified, which showed some shifts in the relative contributions between pre- and post-NCRT samples (Supplemental Fig. 5A).

A total of 6 single base substitution (SBS) signatures were extracted using the data of 96 mutations classes (Fig. 3B), including SBS1, caused by age-related accumulation of 5-methylcytosine deamination events; SBS3, a predictor of homologous recombination deficiency (HRD); SBS5, whose aetiology is unknown; SBS6, associated with defective DNA mismatch repair; SBS13, attributed to Apolipoprotein B mRNA Editing Catalytic Polypeptide-like mutagenesis; and SBS16, previously associated with alcohol consumption in ESCC²⁷. The comparison of SBS signatures between pre- and post-NCRT samples showed that SBS1 and SBS13 decreased (P = 0.003 for SBS1, P < 0.001 for SBS13) while SBS6 and SBS5 increased after NCRT (P = 0.046 for SBS6, P = 0.015 for SBS5, Fig. 3C). Subgroup analyses showed that the SBS1 and SBS13 decreased (P < 0.001 and P = 0.003, respectively) while SBS5 and SBS6 increased (P = 0.021 and P = 0.017, respectively) in the clone-extinct group after NCRT. In the clone-persistent group, SBS13 decreased significantly after NCRT (P = 0.036), however, SBS1, SBS5, and SBS6 did not change significantly after NCRT (all P > 0.05, Supplemental Fig. 5B). We further sought to determine whether there were any biased mutational signatures between clone-extinct and clone-persistent groups. SBS16 was enriched more in the clone-persistent group than in the clone-extinct group in pre-NCRT samples (P = 0.037, Fig. 3D), while SBS5 and SBS6 were enriched more in the clone-extinct group than in the clone-persistent group in post-NCRT samples (P = 0.044 for SBS5, P = 0.016 for SBS6, Fig. 3E).

In addition to SBS signatures, other genomic alteration signatures including doublet-base substitution (DBS), short insertion and deletion (INDEL) and CNV signatures were also analyzed (Fig. 3F). We mainly focused on pre-NCRT samples as few alterations could be detected after NCRT. In terms of DBS and INDEL signatures, DBS3 (Supplemental Fig. 5C), whose proposed aetiology is polymerase epsilon exonuclease domain mutations, and ID1, which is composed predominantly of insertions of thymine (Supplemental Fig. 5D) and is left on the replicated DNA strand by slippage error during DNA replication²⁸, were found to be highly enriched in the pre-NCRT samples from clone-extinct patients (P = 0.043 for DBS3 and P = 0.044 for ID1; Fig. 3G). In addition, analyzing the CNV signatures in pre-NCRT samples also found that HRD-related signature CN17, which consisted of loss of heterozygosity (LOH) segments with total copy numbers between 2 and 4, as well as heterozygous segments with total copy numbers between 3 and 8 (Supplemental Fig. 5E)²⁹, were highly enriched in clone-extinct patients instead of clone-persistent ones (P = 0.045, Fig. 3G). Besides, HRD scores, which were calculated based on CNV profiling data with scarHRD³⁰, were higher in pre-NCRT samples from clone-extinct patients than their counterparts (P = 0.045, Fig. 3G). In summary, mutational signatures provide us with an etiological perspective on clonal dynamics, including potential associations between exogenous factors like alcohol metabolism (SBS16) and clonal persistence, as well as intrinsic factors such as DNA repair deficiencies (DBS3, ID1 and CN17) and clonal extinction.

High subclonal neoantigens and persistent human leucocyte antigen (HLA)-LOH in clone-persistent patients

To investigate how tumor neoantigen landscape evolved during NCRT, we conducted analysis of neoantigens. A total of 2494 neoantigens were computationally identified and tumor neoantigens burden (TNB) was positively correlated with TMB (rho = 0.967, P < 0.001, Fig. 4A) in all 29 pre- and 22 post-NCRT samples. Although total TNB was not significantly different between clone-extinct and clone-persistent pre-NCRT samples (P = 0.442, Fig. 4B), the proportion of subclonal neoantigens in pre-NCRT samples was higher in clone-persistent group than that in clone-extinct group (P = 0.038, Fig. 4C). After NCRT, significant decrease in TNB was observed in both clone-extinct and clone-persistent group (P = 0.002 and P = 0.042) and the magnitude of clonal TNB decrease was larger in clone-extinct group than that in clone-persistent group (P = 0.047, Fig. 4D). To further investigate whether the decrease of neoantigens were simply due to overall tumor regression or selective depletion of neoantigen-bearing cells, we estimated the theoretical TNB merely caused by overall tumor regression and found that the observed number of neoantigens was significantly lower than the expected number in the clone-extinct group, but not in the clone-persistent group (P = 0.016 and P = 0.377, Fig. 4E), suggesting the critical roles of immune elimination of neoantigen-bearing cancer cells in clone-extinct patients.

Alteration of HLA, the important protein for antigen presentation, was a common way for tumor escaping from immune clearance, we therefore investigated whether HLA dysfunction contributed to distinct neoantigen changes between two clonal dynamics. Loss of an HLA haplotype resulted in reduced antigen presentation and thus facilitate immune evasion³¹. Fifteen out of 51 ESCCs exhibited loss of 1–3 HLA haplotypes, resulting in the failure to present 0-100% of neoantigens (Fig. 4F). There were no significant differences in the frequency of HLA-LOH or in the number of neoantigens that could not be presented due to HLA-LOH in either pre- or post-NCRT samples between clone-extinct and clone-persistent groups (all P > 0.05, Supplemental Fig. 6A-B). However, persistently existed HLA-LOH and consequent failure in neoantigen presentation were only observed in two cases (P3 and P30) of the clone-persistent group after NCRT (Fig. 4F). In P3, peptides derived from somatic nonsynonymous mutations might not be recognized by immune system with loss of HLA-B*13:01, which was used for presentation of 100% neoantigens in both pre- and post-NCRT samples in P3. In P30, loss of HLA-A*11:01 resulted in 6.3% of neoantigens being unpresented and newly occurred loss of HLA-C*04:01 after NCRT resulted in the other 6.3% neoantigens failed to be present on the surface of tumor cells. These results implied that persistence of HLA dysfunction-related immune evasion was a critical driver of clonal persistence in ESCC (Fig. 4F).

Activated immune programs and increased immune infiltrations after NCRT were pronounced in clone-extinct patients

In addition to inducing genome evolution, NCRT may sculpt tumor and tumor microenvironment transcriptomes, which might be associated with NCRT response. Gene-set enrichment analysis conducted on pre-NCRT transcriptome data revealed that pathways related to TNFα signaling via NFKB, IL6-JAK-STAT3 and inflammatory response were significantly upregulated in the clone-extinct group, indicating the presence of a pre-existing inflamed environment (Supplemental Fig. 7A). Concordantly, immune-related pathways were significantly overexpressed in post-NCRT samples from clone-extinct group (Supplemental Fig. 7B). Further analysis of differentially expressed genes (DEGs) between paired pre- and post-NCRT samples identified 6654 DEGs in clone-extinct group, and 3661 DEGs in clone-persistent group (fold change ≥ 2 & adjust P value < 0.05, Supplemental Fig. 7C). Generally, pathways of tumor proliferation, including E2F targets, MYC targets were downregulated after NCRT, while pathways of angiogenesis, epithelial mesenchymal transition, myogenesis, and immune response, such as complement, inflammatory response, and allograft rejection were enriched in post-NCRT samples from both clone-extinct and clone-persistent groups (Fig. 5A). These results suggested that immune programs were activated after NCRT, which were more pronounced in the clone-extinct group.

We next investigated the adaptive changes of microenvironment composition in response to NCRT. Immune infiltration assessed by MCPcounter³³ revealed that monocytic lineage cells were enriched in clone-extinct patients before NCRT (P = 0.027, Supplemental Fig. 7D), while CD8 + T cells were enriched in clone-extinct patients after NCRT (P = 0.015, Supplemental Fig. 7E). By comparing the differential infiltration between pre- and post-NCRT samples, we identified increased infiltrations of a series of immune cells, endothelial cells and fibroblasts after NCRT regardless of the clonal dynamic modes (all P < 0.05, Fig. 5B). Furthermore, the increase of CD8 + T cells in clone-extinct patients was significantly higher than that in clone-persistent ones after NCRT (P = 0.027, Fig. 5C). Quantification of immune infiltration by other algorithms supported the pronounced elevation of CD8 + T cell infiltration in clone-extinct patients following treatment as compared to the alterations observed in clone-persistent patients (Supplemental Fig. 7F). Immunohistochemistry staining of samples with RNA sequencing in our study also confirmed that the increase in proportion of CD8 + T cell after NCRT were higher in the clone-extinct group than in the clone-persistent group (P = 0.043, Fig. 5D). Our results suggested that NCRT might induced an adaptive immune response during NCRT and the pronounced elevation of CD8 + T cell infiltration correlated significantly to clonal extinction.

Integrated T cell clone and diversity metrics associated with clonal evolution

Tumor neoantigens need to be specifically recognized by T cell receptors (TCRs) to induce an immunogenic response. Hence, complementarity-determining region 3 (CDR3) sequences in TCRs that bind specifically to neoantigens were inferred using transcriptome data. Unique CDR3 represents T cell clonetype and the number of unique CDR3 negatively correlated to genomic ITH (rho = -0.352, P = 0.035, Fig. 6A). The numerical abundance of CDR3s in pre-NCRT samples did not differ between clone-extinct and clone-persistent group (P = 0.717, Fig. 6B). However, not all T-cell clonetypes recognize the binding neoantigens, and we questioned if there was a difference in the number of T-cell clonetypes that bind neoantigens between the two groups. Applying deep-learning model pMTnet³⁴ to predict which T cell clonetypes are able to bind neoantigens, the absolute number of TCR-neoantigen interactions were significantly higher in pre-NCRT samples from clone-extinct patients than those from clone-persistent patients (P = 0.044, Fig. 6C). We further adopted neoantigen immunogenicity effectiveness score (NIES), which is based on the product of the variant allele frequency of the neoantigen’s corresponding somatic nonsynonymous mutation and the clonal fraction of the TCRs that bind the same neoantigen, to measure this antitumor effect of interactions between neoantigens and TCRs³⁴. The higher the NIES score, the more expanded TCRs are specific against clonal neoantigens, which is generally favorable for clinical outcomes³⁴. We found that the NIES score was higher in pre-NCRT samples from clone-extinct patients than those from clone-persistent patients (P = 0.046, Fig. 6D). Analysis of TCR-neoantigens was not performed in post-NCRT samples as few neoantigens could be detected after NCRT.

As a surrogate for the level of T cell infiltration in the microenvironment, the numerical abundance of CDR3s undergoes augmentation in both clone-extinct and clone-persistent groups after NCRT (P = 0.003 and P = 0.078, respectively), although statistical significance was not achieved in clone-persistent group (Fig. 6E). The ability to bind neoantigens and the clonality of binding neoantigens will affect the expansion of T cell clones. Therefore, different T cell clones are likely to expand unevenly during treatment. We introduced D50 index, an indicator of T cell repertoire evenness of which higher values indicated the expansion of general clonal populations and lower values indicated the expansion of specific clonal populations³⁵, to evaluate the change of T cell clonal distribution evenness. After NCRT, the evenness of T cell clonetype distribution decreased significantly in clone-extinct patients (P = 0.042), but increased significantly in clone-persistent patients (P = 0.016, Fig. 6F). Furthermore, reduction of clonal neoantigens were positively associated with the decrease of D50 index during NCRT, an indicator of increased skewness of T cell clonetypes, in the clone-extinct group (rho = 0.718, P = 0.017, Fig. 6G) but not in the clone-persistent group. No other significant correlations were identified between the changes of TNB and the distribution of T cell clonetypes during NCRT. In conclusion, our investigation of T cell repertoire and clonal evolution suggested that there was a more effective T cell-mediated antitumor immunity in pre-NCRT samples from clone-extinct group, which correlated to a significant reduction of tumor clones bearing clonal neoantigens.

Clonal dynamics following cancer therapies including cytotoxic treatments have been reported in multiple cancer types^{8, 9, 36}. NCRT significantly improved outcomes of ESCC patients, but those non-responders still undergo high risk of disease relapse and treatments for these patients are limited, partially due to our insufficient understanding of the association between cancer evolution and treatment responses^{10, 37}. In this study, we collected pre- and post-treatment samples from ESCC patients receiving NCRT and performed whole-exome and transcriptome sequencings to explore the tumor genome and immunogenome evolution of ESCCs. Compared with previous studies on the clonal evolution of ESCC, we are the first to systematically describe the changes in clonal structure and immune genome during NCRT and thereby categorized ESCC patients into two clonal dynamics groups in the setting of NCRT: clonal extinction and clonal persistence. The two clonal dynamics were coherently associated with distinct patient outcomes superior to the pathological assessments in our cohort and an external validation cohort, implying the important role of the clonal dynamics in evaluating NCRT efficacy. In addition, ITH of pre-NCRT samples were highly correlated with NCRT-driven clonal dynamics and patients’ outcomes, highlighting the importance of dissection of pre-treatment ITH in predicting therapeutic outcomes.

NCRT contains treatments that cause DNA damage to kill tumors, for example, platinum induces DNA interstrand crosslink and radiation induces DNA double-strand breaks, suggesting that the functional status of DNA damage repair machinery might be critical determinants of NCRT responses and NCRT-driven clonal evolution³⁸. In our mutational signature analyses, DBS3 which occurred during DNA synthesis due to polymerase epsilon exonuclease domain mutations and ID1 that was associated with frame-slippage during DNA replication³⁹ were enriched in pre-NCRT samples of clone-extinct patients, implying that the disturbed DNA replication process sensitized ESCC cells to genotoxic therapies. The HRD status, reflected by frequent LOH (CN17)²⁹ and higher HRD scores, was also associated with clonal extinction after NCRT in ESCCs, in concordance with previous reports describing HRD breast cancer subgroups as beneficiaries of adjuvant chemotherapy⁴⁰. Like the broad applications of HRD detection in other cancer types, our results raised the promising potential of HRD assessment in guiding treatment decisions in ESCC. In addition, SBS16 was considered to be related to NCRT resistance as higher prevalence of SBS16 was found in pre-treatment samples from clone-persistent ESCCs. Previous study issued the associations between SBS16 and alcohol consumption as well as poor survival in ESCCs⁴¹. Future large-scale studies are needed to confirm the role of SBS16 and alcohol consumption in affecting NCRT responses in ESCCs.

In the study, TMB significantly decreased after NCRT as previously reported³⁰, which might be caused by the decreased sample purity due to tumor shrinkage. Interestingly, our simulation analysis indicated an extraordinary decline of TMB in clone-extinct patients which could not be simply attributed to the decrease of tumor purity, stressing other antitumor mechanisms targeting mutation bearing cells in addition to direct cytotoxic effects. In clone-extinct patients, the observed reductions of TNB far exceeded the TNB decrease owing to tumor regression, suggesting neoantigen-bearing cancer cells were selectively eliminated by NCRT. Importantly, the degree of clonal neoantigen depletion positively correlated with specific T cell clone expansion in clone-extinct patients but not clone-persistent ones, suggesting specific T cell clone expansion and attack towards tumor clones as an additive antitumor approach of NCRT in clone-extinct patients. It is speculated that the generation of neoantigens via DNA damage induced by chemotherapy and (or) radiotherapy might increase the immunogenicity of a cell during treatment, which finally results in the elimination of cancer cells by active immune cells together with chemotherapy and radiotherapy⁴². This is in accordance with our findings that immune pathways and infiltrates including cytotoxic lymphocyte CD8 + T cells significantly increased after NCRT, especially in clone-extinct patients by transcriptomic analyses. However, the anti-tumor immunity was not significantly enhanced by NCRT in clone-persistent patients compared to clone-extinct ones, probably due to the higher proportion of subclonal neoantigens before NCRT and persistent HLA-LOH after NCRT in clone-persistent patients. An active immune response often started from the regular neoantigen presentation from tumor cells. The subclonal neoantigens stimulated a futile T cell response, resulting in immune clearance of only a minor proportion of cells carrying these subclonal neoantigens but not the whole populations¹⁸. Loss of HLA haplotype leads to impaired presentation of cancer neoantigens and enables tumor cells to escape NCRT-induced immune attack³¹. Overall, our results revealed the NCRT-enhanced antitumor immunity as a critical mechanism on eliminating ESCC cells, and provided evidences of potential additive effects of combinative or subsequent immunotherapy with NCRT.

There were some limitations in this study. First, the sample size used for clonal evolution analysis is relatively small, due to the difficulty in obtaining paired pre- and post-NCRT tumor specimens. However, we tried to validate our identification of the relationship between ESCC clonal evolution modes under treatment and prognosis in an external validation cohort with public data available²⁵. Besides, no recurrent mutations associated with NCRT sensitivity or resistance was identified by current WES analyses. Future whole-genome sequencing which could disclose more genomic variants, such as non-coding mutations and structural variation, would help to reveal more genomic factors related to ESCC clonal evolution and therapeutic responses. In addition, other omics besides genomic studies can further elucidate the therapy-driven clonal evolution beyond genome evolution. For example, cancers can also evolve through epigenetic modifications, which interact closely with the changing tumor microenvironment⁴⁴.

Collectively, we identified two tumor evolution modes following NCRT in ESCCs: clonal extinction and clonal persistence, characterized by distinct pre-treatment genomic ITH, NCRT-driven clonal dynamics and patients’ prognoses. Immunogenomic analyses emphasized the crosstalk between tumor cell genomic features and the tumor immune microenvironment during NCRT to affect responses to NCRT. Our study provides insights into the tumor clone and immunogenomic evolution of ESCCs driven by NCRT, which would provide insights into identifying determinants of treatment responses and developing treatment strategies for improving ESCC outcomes.

Patients and samples

Pre- and post-NCRT tumor and corresponding WBC samples were obtained from pathologically diagnosed ESCC patients, who were clinically staged as T1-4N1-3M0 potentially resectable thoracic ESCCs, and received NCRT and complete esophagectomy after NCRT in Sun Yat-sen University Cancer Center from November 2018 to March 2020. All patients provided written informed consent for treatment and molecular analyses, with ethical approval from the Research Ethics Committee of Sun Yat-sen University Cancer Center (GZR2018-110). NCRT scheme and sample collection were described in the Supplemental materials.

Whole-exome sequencing and transcriptome sequencing analysis

A total of 22 paired pre- and post-NCRT and additional 7 pre-NCRT tumor samples and 29 WBC samples were collected for WES, among which 18 pairs of pre- and post-NCRT samples were used for transcriptome sequencing. Other methods were described in the Supplemental materials.

Statistics and software

All statistical analyses were performed using R software v4.1.2, or GraphPad Prism v7.0, and P values were two-sided. The Wilcoxon rank sum test and Wilcoxon signed rank test were used as the nonparametric tests to compare differences between two groups or between paired samples, respectively. Fisher’s exact test was used to analyze the correlation between pre-treatment characteristics and TRGs as well as clonal dynamic modes. Correlations between two continuous variables were measured using Spearman rank correlation test.

OS was calculated as the time from the date of first NCRT treatment to the date of death or the last follow-up. DFS was calculated as the time from the date of R0 resection to the date of first disease recurrence or the last follow-up. The cutoff value was determined using the maxstat R package. OS and DFS were analyzed by the Kaplan–Meier method and compared by the log-rank test.

AUTHOR CONTRIBUTIONS

J.W. and Z.L.W. conceived of and designed the study. Z.L.W., Z.H.M., and J.Y.Y. conducted the data analyses and wrote the manuscript. K.J.L., Q.W.L., H.Y., X.Y.X., X.D.L., T.L., J.Y.C and X.L.W. collected the patients’ information and specimens. F.Q.D. performed the immunohistochemistry analyses. Y.H.L. contributed to the pathological evaluation. J.W., J.H.F., and K.J.L. supervised the study. All authors read and approved the final manuscript.

ACKNOWLEDGMENTS

This study was funded by grants from the National Natural Science Foundation of China (81871975, 82072607, 82273032, and 82002476), Natural Science Foundation of Guangdong Province (2019A1515010715), and Fundamental Research Funds for the Central Universities (16ykjc39). The funders played no role in study design, data collection, analysis and interpretation of data, or the writing of this manuscript.

CONSENT FOR PUBLICATION

Not applicable.

CONFLICT OF INTEREST STATEMENT

The authors declare no conflict of interest.

ETHICS STATEMENT

All patients provided written informed consent for treatment and molecular analyses, with ethical approval from the Research Ethics Committee of Sun Yat-sen University Cancer Center (GZR2018-110).

FUNDING STATEMENT

This work was supported by grants from the National Natural Science Foundation of China (81871975, 82072607, 82273032, and 82002476), Natural Science Foundation of Guangdong Province (2019A1515010715), and Fundamental Research Funds for the Central Universities (16ykjc39).

DATA AVAILABILITY STATEMENT

The whole-exome sequencing and RNA sequencing data are stored in GSA-Human with accession number HRA002346, while single cell RNA sequencing data is under HRA001861. These raw data related to this study was requested from the corresponding author JW ([email protected]).

Sung H, Ferlay J, Siegel RL et al. Global Cancer Statistics 2020: GLOBOCAN Estimates of Incidence and Mortality Worldwide for 36 Cancers in 185 Countries. CA Cancer J Clin. 2021; 71 (3): 209-249.
Arnold M, Ferlay J, van Berge Henegouwen MI et al. Global burden of oesophageal and gastric cancer by histology and subsite in 2018. Gut. 2020; 69 (9): 1564-1571.
Ajani JA, D’Amico TA, Bentrem DJ et al. Esophageal and Esophagogastric Junction Cancers, Version 2.2023, NCCN Clinical Practice Guidelines in Oncology. J Natl Compr Canc Netw. 2023; 21 (4): 393-422.
Schneider PM, Baldus SE, Metzger R et al. Histomorphologic tumor regression and lymph node metastases determine prognosis following neoadjuvant radiochemotherapy for esophageal cancer: implications for response classification. Ann Surg. 2005; 242 (5): 684-692.
McGranahan N, Swanton C. Clonal Heterogeneity and Tumor Evolution: Past, Present, and the Future. Cell. 2017; 168 (4): 613-628.
Dentro SC, Wedge DC, Van Loo P. Principles of Reconstructing the Subclonal Architecture of Cancers. Cold Spring Harb Perspect Med. 2017; 7 (8).
Alexandrov LB, Nik-Zainal S, Wedge DC et al. Signatures of mutational processes in human cancer. Nature. 2013; 500 (7463): 415-421.
Kim C, Gao R, Sei E et al. Chemoresistance Evolution in Triple-Negative Breast Cancer Delineated by Single-Cell Sequencing. Cell. 2018; 173 (4): 879-893 e813.
Findlay JM, Castro-Giner F, Makino S et al. Differential clonal evolution in oesophageal cancers in response to neo-adjuvant chemotherapy. Nat Commun. 2016; 7: 11111.
Hirata H, Niida A, Kakiuchi N et al. The Evolving Genomic Landscape of Esophageal Squamous Cell Carcinoma Under Chemoradiotherapy. Cancer Res. 2021; 81 (19): 4926-4938.
Marine JC, Dawson SJ, Dawson MA. Non-genetic mechanisms of therapeutic resistance in cancer. Nat Rev Cancer. 2020; 20 (12): 743-756.
Park S, Joung JG, Min YW et al. Paired whole exome and transcriptome analyses for the Immunogenomic changes during concurrent chemoradiotherapy in esophageal squamous cell carcinoma. J Immunother Cancer. 2019; 7 (1): 128.
Ji D, Yi H, Zhang D et al. Somatic Mutations and Immune Alternation in Rectal Cancer Following Neoadjuvant Chemoradiotherapy. Cancer Immunol Res. 2018; 6 (11): 1401-1416.
Cui Y, Chen H, Xi R et al. Whole-genome sequencing of 508 patients identifies key molecular features associated with poor prognosis in esophageal squamous cell carcinoma. Cell Res. 2020; 30 (10): 902-913.
Hu J, Zhang L, Xia H et al. Tumor microenvironment remodeling after neoadjuvant immunotherapy in non-small cell lung cancer revealed by single-cell RNA sequencing. Genome Med. 2023; 15 (1): 14.
Oppedijk V, van der Gaast A, van Lanschot JJ et al. Patterns of recurrence after surgery alone versus preoperative chemoradiotherapy and surgery in the CROSS trials. J Clin Oncol. 2014; 32 (5): 385-391.
Carter SL, Cibulskis K, Helman E et al. Absolute quantification of somatic DNA alterations in human cancer. Nat Biotechnol. 2012; 30 (5): 413-421.
Riaz N, Havel JJ, Makarov V et al. Tumor and Microenvironment Evolution during Immunotherapy with Nivolumab. Cell. 2017; 171 (4): 934-949 e916.
Integrated genomic characterization of oesophageal carcinoma. Nature. 2017; 541 (7636): 169-175.
Roth A, Khattra J, Yap D et al. PyClone: statistical inference of clonal population structure in cancer. Nat Methods. 2014; 11 (4): 396-398.
Engle EK, Fisher DA, Miller CA et al. Clonal evolution revealed by whole genome sequencing in a case of primary myelofibrosis transformed to secondary acute myeloid leukemia. Leukemia. 2015; 29 (4): 869-876.
Yang F, Wang Y, Li Q et al. Intratumor heterogeneity predicts metastasis of triple-negative breast cancer. Carcinogenesis. 2017; 38 (9): 900-909.
Wen J, Fang S, Hu Y et al. Impacts of neoadjuvant chemoradiotherapy on the immune landscape of esophageal squamous cell carcinoma. EBioMedicine. 2022; 86: 104371.
Miller CA, McMichael J, Dang HX et al. Visualizing tumor evolution with the fishplot package for R. BMC Genomics. 2016; 17 (1): 880.
Yang L, Zhang X, MacKay M et al. Identification of Radioresponsive Genes in Esophageal Cancer from Longitudinal and Single Cell Exome Sequencing. Int J Radiat Oncol Biol Phys. 2020; 108 (4): 1103-1114.
Jassal B, Matthews L, Viteri G et al. The reactome pathway knowledgebase. Nucleic Acids Res. 2020; 48 (D1): D498-d503.
Wei R, Li P, He F et al. Comprehensive analysis reveals distinct mutational signature and its mechanistic insights of alcohol consumption in human cancers. Brief Bioinform. 2021; 22 (3).
Alexandrov LB, Kim J, Haradhvala NJ et al. The repertoire of mutational signatures in human cancer. Nature. 2020; 578 (7793): 94-101.
Steele CD, Abbasi A, Islam SMA et al. Signatures of copy number alterations in human cancer. Nature. 2022; 606 (7916): 984-991.
Sztupinszki Z, Diossy M, Krzystanek M et al. Migrating the SNP array-based homologous recombination deficiency measures to next generation sequencing data of breast cancer. NPJ breast cancer. 2018; 4: 16.
McGranahan N, Rosenthal R, Hiley CT et al. Allele-Specific HLA Loss and Immune Escape in Lung Cancer Evolution. Cell. 2017; 171 (6): 1259-1271 e1211.
Song D, Wang X. DEPTH2: an mRNA-based algorithm to evaluate intratumor heterogeneity without reference to normal controls. J Transl Med. 2022; 20 (1): 150.
Becht E, Giraldo NA, Lacroix L et al. Estimating the population abundance of tissue-infiltrating immune and stromal cell populations using gene expression. Genome Biol. 2016; 17 (1): 218.
Lu T, Zhang Z, Zhu J et al. Deep learning-based prediction of the T cell receptor-antigen binding specificity. Nat Mach Intell. 2021; 3 (10): 864-875.
Shugay M, Bagaev DV, Turchaninova MA et al. VDJtools: Unifying Post-analysis of T Cell Receptor Repertoires. PLoS Comput Biol. 2015; 11 (11): e1004503.
Kim H, Zheng S, Amini SS et al. Whole-genome and multisector exome sequencing of primary and post-treatment glioblastoma reveals patterns of tumor evolution. Genome Res. 2015; 25 (3): 316-327.
Park S, Joung J-G, Min YW et al. Paired whole exome and transcriptome analyses for the Immunogenomic changes during concurrent chemoradiotherapy in esophageal squamous cell carcinoma. J Immunother Cancer. 2019; 7 (1): 128-128.
Goldstein M, Kastan MB. The DNA damage response: implications for tumor responses to radiation and chemotherapy. Annu Rev Med. 2015; 66: 129-143.
Alexandrov LB, Kim J, Haradhvala NJ et al. The repertoire of mutational signatures in human cancer. Nature. 2020; 578 (7793): 94-101.
Telli ML, Timms KM, Reid J et al. Homologous Recombination Deficiency (HRD) Score Predicts Response to Platinum-Containing Neoadjuvant Chemotherapy in Patients with Triple-Negative Breast Cancer. Clin Cancer Res. 2016; 22 (15): 3764-3773.
Wei R, Li P, He F et al. Comprehensive analysis reveals distinct mutational signature and its mechanistic insights of alcohol consumption in human cancers. Brief Bioinform. 2020; 22 (3).
Thol K, Pawlik P, McGranahan N. Therapy sculpts the complex interplay between cancer and the immune system during tumour evolution. Genome Med. 2022; 14 (1): 137.
Kelly RJ, Ajani JA, Kuzdzal J et al. Adjuvant Nivolumab in Resected Esophageal or Gastroesophageal Junction Cancer. N Engl J Med. 2021; 384 (13): 1191-1203.
Nicholson JG, Fine HA. Diffuse Glioma Heterogeneity and Its Therapeutic Implications. Cancer Discov. 2021; 11 (3): 575-590.

No competing interests reported.

Download PDF

Version 1

posted

You are reading this latest preprint version

Evolution of genome and immunogenome in esophageal squamous cell carcinomas driven by neoadjuvant chemoradiotherapy

Status:

Version 1

Abstract

Background

Methods

Results

Conclusions

Figures

Introduction

Results

Genome landscapes of ESCCs before and after NCRT

Clonal evolution of ESCC under NCRT

Mutational signatures related to NCRT-driven clonal evolution

High subclonal neoantigens and persistent human leucocyte antigen (HLA)-LOH in clone-persistent patients

Activated immune programs and increased immune infiltrations after NCRT were pronounced in clone-extinct patients

Integrated T cell clone and diversity metrics associated with clonal evolution

Discussion

Methods and Materials

Patients and samples

Whole-exome sequencing and transcriptome sequencing analysis

Statistics and software

Declarations

References

Additional Declarations

Supplementary Files

Status:

Version 1