Genomic evolution during locoregional recurrence in colorectal cancer determined by whole-exome sequencing: a retrospective observational study : Journal of Bio-X Research

Secondary Logo

Journal Logo

Research Articles

Genomic evolution during locoregional recurrence in colorectal cancer determined by whole-exome sequencing: a retrospective observational study

Lan, Xiaolianga,b; Wu, Xiaoxiaoc,d; Zhang, Chaoe; Wei, Genxiaa,f; Li, Bingbinga,g; Qiu, Weihaoa,g; Li, Danyia,g; Wu, Huanwenh; Ding, Yanqinga,g; Yuan, Jiee; Tai, Zaixiane; Yang, Zuoquane; Liang, Zhiyongh,*; Su, Danc,d,*; Liang, Lia,g,*

Author Information
Journal of Bio-XResearch 5(4):p 171-180, December 2022. | DOI: 10.1097/JBR.0000000000000116
  • Open

Abstract

Introduction

Colorectal cancer (CRC) is the third leading cause of cancer-related deaths worldwide.[1] Although distant metastasis remains the most common cause of death, locoregional recurrence (LR) of CRC after surgery is also an issue of increasing concern. The reported incidence of locally recurrent colon cancer varied from 10% to 29%.[2–5] The incidence of LR rectal cancer before the adoption of total mesorectal excision was 40% to 50% in patients with advanced-stage disease,[6] but the broad adoption of this approach has reduced the incidence to <10%.[7] In addition, preoperative neoadjuvant chemoradiotherapy with subsequent total mesorectal excision has further reduced the incidence of LR of rectal cancer in several clinical trials.[8,9]

LR in patients with CRC is always associated with a poor prognosis. A previous report revealed 5-year overall survival rates of 9% and 13% for patients with LR of colon cancer and rectal cancer, respectively.[10] The patterns and predictors of LR after curative resection have been investigated to optimize the therapeutic regimen and postoperative surveillance. Liska et al[11] proved that tumor stage, bowel obstruction, margin involvement, lymphovascular invasion, and local tumor invasion were significantly associated with colon cancer LR, and Rokke et al[12] further demonstrated that reoperation for postoperative complications, other than anastomotic leakage, was a significant predictor of LR of CRC. Although recent advances in molecular characterization of CRC have identified prognostic and predictive biomarkers for disease recurrence,[13–15] the pattern of genomic evolution between paired primary and LR CRC tissues and molecular evidence for the origin of recurrence are still unclear.

In this study, we used whole-exome sequencing to scrutinize the molecular events occurring during genomic evolution in a unique cohort of 23 patients with CRC, for whom paired primary and LR tumor samples were available. With the cutting-edge analysis tools, we aim to compare the LR patterns between high microsatellite instability (MSI-H) and microsatellite stability (MSS) patients, identify the disseminating mechanism of the primary tumor, and explore the potential biomarker for LR early detection and precise clinical management.

Materials and methods

Patient specimen acquisition

In this retrospective observational study, 23 patients with CRC who completed standard initial treatment were recruited from January 2011 to December 2018. Clinicopathological characteristics of patients were obtained from the hospital’s case management system. The inclusion criteria include that patients were diagnosed with colon or rectal cancer, patients had not received any treatment before surgery, both primary tumor and LR tumor can be obtained, patients were metastasis-free. The included 13 patients with high sequencing read coverage enrolled from Nanfang Hospital of Southern Medical University, Guangzhou City and 10 patients with low sequencing read coverage recruited from Zhejiang Cancer Hospital, Hangzhou City, whose sequencing data were used to validate chromosome arm copy number variation (CNV). All patients underwent radical surgery. Primary tumor, matched LR tumor, and matched normal tissue samples were collected, and formalin-fixed, paraffin-embedded samples were prepared. Genomic DNA was extracted using a TIANamp Genomic DNA kit (Tiangen Biotech, Beijing, China) following the manufacturer’s instructions. Written informed consent was obtained from all patients during sample collection. The study was approved by the Institutional Review Board of Nanfang Hospital of Southern Medical University (approval No. 2020010) on September 11, 2020 and conducted in accordance with the 1964 Declaration of Helsinki, as revised in 2013.

Library construction and high-throughput sequencing

Genomic DNA prepared as above was fragmented using an ultra-sonicator UCD-200 (Diagenode, Seraing, Belgium) and then purified and size-selected with magnetic beads (Beckman, MA, USA). The quality of the DNA was determined using a Qubit 2.0 Fluorometer with Quanti-IT dsDNA HS Assay Kit (Thermo Fisher Scientific, Waltham, MA, USA). Library construction was then performed using a custom 53 M whole-exon capturing probe (IDT, Coralville, IA, USA). The captured libraries were then pair-end sequenced in 100-bp lengths using the Geneplus-2000 sequencing platform (Geneplus, Beijing, China), following the manufacturer’s instructions. Raw data from next-generation sequencing was then filtered to remove low-quality reads and adaptor sequences. Reads were further aligned to the reference human genome (hg19) utilizing BWA aligner software (version 0.7.10)[16] for mutation calling.

Genomic data analysis

Single nucleotide variants (SNVs) were called by MuTect (version 1.1.4).[17] For quality control, somatic mutations were identified if they met the following criteria: (1) present in <1% of the population in the 1000 Genomes Project (https://www.internationalgenome.org/), the Exome Aggregation Consortium (ExAC), and the Genome Aggregation Database (gnomAD) (https://gnomad.broadinstitute.org); (2) not present in paired germline DNA from normal tissues; and (3) detected in at least three high-quality reads containing the particular base, where high-quality reads were defined as a Phred score >30, mapping quality >30, and without paired-end reads bias. Germline mutations were called using GATK software (version 4.0)[18] and an in-house script. The germline variations were validated in the ClinVar database (https://www.ncbi.nlm.nih.gov/clinvar/), to confirm its pathogenicity. Mutational signature analysis was performed with unfiltered somatic mutations using R package YAPSA[19] and matched to the COSMIC signature database (https://cancer.sanger.ac.uk/cosmic/signatures). Somatic CNVs (SCNVs) were identified using GATK (version 4.0) and the frequency of larger fragmental CNVs was detected using R package Copy number.[20] The focal level of SCNV was detected using an in-house script with BAM files. The clonal architectures of the somatic mutations were inferred by ABSOLUTE software[21] considering tumor purity and copy number alterations. Events with an estimated upper 95% confidence interval for the cancer cell fraction (CCF) of 1 were defined as clonal, whereas the rest were defined as sub-clonal. A list of driver genes was referenced from previous reports and databases (https://cancer.sanger.ac.uk/census).[22,23] The pathway enrichment was performed with an in-house script. Clusters of seeding clones were calculated using PyClone-VI software[24] and visualized by ClonEvol software.[25] The public dataset of CRC which consists of 276 cases, was downloaded from the cbioportal database (https://www.cbioportal.org/).

Statistical analysis

Data were analyzed by two-sided Mann-Whitney U test, Fisher exact test, Wilcoxon test and Pearson’s correlation using GraphPad Prism (version 7.01; GraphPad, San Diego, CA, USA) or R (version 3.6.1).[26] For all tests, a P value < 0.05 was considered statistically significant. The enrolled patients were selected according to inclusion and exclusion criteria, because it is extremely difficult to collect three samples from the same patient, which might take many years, thus 23 patients in this cohort are the maximum number we can recruit.

Results

Distribution of somatic mutations between primary and LR lesions in patients with CRC

Thirteen patients were recruited and 39 formalin-fixed, paraffin-embedded samples were subjected to whole-exome sequencing (13 sets of primary tumors, LR tumors, and adjacent normal tissues). The clinical characteristics of this cohort are summarized in Table 1. We detected averages of 291 (range 22-1642) and 383 (range 40-2451) non-synonymous somatic mutations and indels in the primary and LR tumor samples, respectively. We calculated the mutation distributions across each patient (Fig. 1A) and showed that 10 patients had 12% to 76% shared SNVs and indels across both primary and LR lesions, while the other three patients had very few shared SNVs or indels. Notably, both the primary and LR tumor samples in the 10 patients were MSS, while both samples in the other three patients were MSI-H. We examined the germline variations of the three MSI-H patients to clarify the etiology of the MSI-H status. MLH1 nonsense mutation c.5C > A, splice site mutation c.208-1G > A, and PMS2 missense mutation c.2570G > C were identified in patients P1 to P3, respectively (Additional Fig. 1A, https://links.lww.com/JR9/A36). Based on this molecular evidence, these three patients were identified with non-sporadic familial CRC with MSI-H status. In contrast, the MSS patients shared an average of 41.61% of all mutations between the primary and LR lesions, indicating the existence of common ancestral clones between the two lesions. In addition, 22.99% (5.15%-60.9%) and 35.4% (8.43%-82.39%) of mutations in the MSS patients were exclusive to the primary and LR lesions, respectively. LR lesions included significantly more mutations than the primary lesions (P< 0.05; Fig. 1B), suggesting that the mutation rate increased in LR tumors compared with their matched primary tumors after divergence. These data suggest a difference in the origin of LRs between patients with MSI-H and MSS CRC.

Table 1 - Patients’ demographic information
Characteristics No. of cases Proportion (%)
Total number 23
Age [yr, mean (range)] 58 (37-86)
Sex
 Male 16 70
 Female 7 30
Smoking history
 Smoker 9 39
 Non-smoker 14 61
Drinking 7 30
Diagnosed site
 Colon cancer 15 65
 Rectal cancer 8 35
Stage at diagnosis
 I-II 7 30
 III-IV 16 30
Chemotherapy 10 43

F1
Figure 1.:
Distribution of somatic mutations between primary and LR lesions in CRC. (A) Percentages of somatic mutations in matched primary and LR lesions in 13 patients. (B) Percentages of exclusive and shared mutations between primary and LR lesions in 10 microsatellite stability (MSS) patients. Data are expressed as the mea±SD. *P=0.0227 (Mann-Whitney U test). (C) Mutational landscapes in 13 patients with matched primary and LR lesions, showing number of somatic mutations in each patient (top), mutation frequency of each gene (right), and other clinical information including tumor site, microsatellite status, variation types (bottom). CRC=Colorectal cancer, LR = locoregional recurrence, P1–P3 = Patient 1–Patient 13.

Driver mutations in primary and LR lesions

Coding driver mutations were identified to construct a mutational landscape. The most frequently mutated drivers such as KRAS, APC, and TP53, have been well-characterized in the progression of CRC (Fig. 1C). In addition, KRAS mutation status was concordant between primary and matched LR lesions in MSS patients, but KRAS mutation only occurred in the primary tumor in one MSI-H patient. TP53 mutations were concordant between the primary and LR samples in most (8/10) MSS patients but differed between the two samples in two of the three MSI-H patients. These results suggest that KRAS and TP53 mutations might be common clonal mutations promoting recurrence of the primary tumor in MSS patients but may not be associated with recurrence in MSI-H patients. Moreover, several driver mutations with high frequencies, such as BRCA2, INTS1, KMT2B, KMT2D, and RNF43, were found exclusively in LR lesions in MSS patients and might thus play significant roles in the progression of LR. We further investigated the distributions of KRAS and TP53 mutations in MSI-H patients in The Cancer Genome Atlas (TCGA) dataset.[27] Notably, KRAS (Additional Fig. 1B, https://links.lww.com/JR9/A36) and TP53 (Additional Fig. 1C, https://links.lww.com/JR9/A36) mutations were significantly less frequent in MSI-H compared with MSS patients. These results emphasized the different impacts of KRAS and TP53 mutations on LR between MSI-H and MSS patients.

Mutational signatures in disease progression of CRC

We examined the mutational signature to identify the processes contributing to the early initiation and late evolution of CRC. The higher mutation rates in LR lesions suggested that the patterns could potentially be indicative of disease progression. The mutation signatures were stratified as mutations exclusive to primary and LR lesions, respectively, and shared mutations. In light of their discrepant mutational mechanisms, the signatures of MSI-H and MSS patients were considered separately. A total of 23 single-base substitution (SBS) signatures were matched with the COSMIC database (https://cancer.sanger.ac.uk/cosmic/signatures) in all 13 patients (Fig. 2A and B). SBS1, SBS6, and SBS15 occurred in most of these patients. SBS1 resulted from an endogenous mutational process initiated by spontaneous deami-nation of 5-methylcytosine and correlated with age at cancer diagnosis, while SBS6 and SBS15 were associated with defective DNA mismatch repair. The three MSI-H patients had no shared SBS signatures between the individual primary lesion and LR lesion, while each MSS patient had a high contribution of shared SBS signatures. Strikingly, the signature heterogeneity across each patient was greater than that across each evolutionary stage. According to their treatment records, six patients (P1, P2, P5, P7, P8, and P9) underwent postoperative chemoradiotherapy. However, the LR lesions in these patients showed a lack of new signatures, although few shifts were observed in the relative contributions of SBS signatures over different evolutionary stages. These results suggest that chemoradiotherapy is unlikely to be a major cause of driver accumulation.

F2
Figure 2.:
Mutation signatures of disease progression in patients with CRC. (A) Signature analysis of exclusive mutations in primary and LR lesions in three microsatellite instability (MSI-H) patients, the various color block represents different SBS signature. (B) Signature analysis of exclusive and shared mutations in primary and LR lesions for 10 microsatellite stability (MSS) patients, the various color block represents different SBS signature. The number represents different signatures derived from COSMIC database. CRC=Colorectal cancer, LR = locoregional recurrence, P1–P13 = Patient 1–Patient 13, SBS = single base substitution signature.

Frequent focal CNVs were enriched in LR lesions

Regarding chromosomal segment distribution, large CNVs were infrequent in MSI-H patients (Additional Fig. 2, https://links.lww.com/JR9/A36), consistent with a previous report.[28] However, there was obvious segmental variation between primary and LR lesions in MSS patients. We compared the CNV frequencies calculated by 2M segment as a unit between primary and LR lesions for all patients (Fig. 3A). LR lesions included losses in segments located in 1q, 5p, and 6q and gains in segments in 2q, 5q, and 6q. We validated large segmental variations in the additional 10 pairs of primary and LR samples with lower sequencing depth (12–123 X). Compared with the primary lesions, the major fragment variations with lower sequencing depth were consistent with the previous 13 pairs of samples (Additional Fig. 3, https://links.lww.com/JR9/A36). Although limited by the small sample size, we sought to identify the CNVs between primary and LR lesions at the driver gene level to emphasize the functional impact on disease progression. Significant copy number gain of PDCD1 (2q37.3) and loss of LMNA(1q22) were enriched in LR lesions (Fig. 3B and C).

F3
Figure 3.:
Frequent focal copy numbervariations (CNVs) were enriched in LR lesions. (A) Comparison of fragment variations between primary and LR lesions. X-axis indicates chromosome number and Y-axis indicates the percentage of copy gain (red) or loss (blue), the star indicates the significant gain or loss in LR lesions compare with primary lesions. (B) Distribution of PDCD1 gain in 13 patients, *P=0.039 indicates the PDCD1 gain significantly enriched in LR lesions. (C) Distribution of LMNA loss in 13 patients, *P=0.03 indicates the LMNA loss significantly enriched in LR lesions. LR = locoregional recurrence, P1–P13 = Patient 1–Patient 13.

Dynamics of mutation clonality between primary and LR lesions

We further investigated the evolution of LR by calculating the CCF value and defined clonal and subclonal mutations. We created a CCF scatter plot to exhibit dynamic clonal-subclonal transitions between primary and LR lesions. Most of the clonal and subclonal mutations were exclusive to either primary or LR lesions in MSI-H patients, indicating the independent attributes of the evolutionary stages (Fig. 4A and Additional Fig. 4A, https://links.lww.com/JR9/A36). In contrast, dynamic clonal transition was observed in MSS patients, suggesting the existence of a common ancestor between the primary and LR lesions (Fig. 4B and Additional Fig. 4B, https://links.lww.com/JR9/A36). In addition, we inferred the divergence points between the relapsed seeding clone and primary tumor, and showed a “molecular time,” representing the relapse divergence from the primary tumor, of 33% to 89% (Fig. 4C). These results suggest that the matched samples in MSS patients were clonally related.

F4
Figure 4.:
Dynamics of mutation clonality between primary and LR lesions. The CCF dynamics in a representative patient. (A) MSI-H and (B) MSS patients, the X-axis indicates the CCF of primary lesion, Y-axis indicates the CCF of LR lesion, the red dots indicate the clonal mutations, the green dots indicate the subclonal mutation, and the purple dots indicate the mutations transition from subclonal to clonal. (C) Divergence time in each MSS patient, the different shape on the top of the column represents a different patient. CCF=Cancer cell fraction, LR = locoregional recurrence, MSI-H = microsatellite instability-high, MSS = microsatellite stable, P3–P13 = Patient 3–Patient 13.

Evolutionary pattern of primary CRC to LR

We created a phylogenetic tree for each patient based on the mutation sites. The three MSI-H patients with no truncal mutations (Shared clonal mutations between the primary and LR lesion) were considered to follow independent evolutionary trajectories (Fig. 5A). In contrast, the 10 MSS patients showed unambiguous truncal mutations and divergence points (Fig. 5B). Notably, the LR tumors had an average of a 63% greater mutation burden than the primary tumor in MSS patients (Additional Fig. 5A, https://links.lww.com/JR9/A36), indicating that the rate of mutation accumulation increased during late LR evolution. The recurrence-specific increase in mutations was also loosely correlated with the time interval between first surgery and the diagnosis of LR (Pearson’s correlation r=0.313; Additional Fig. 5B, https://links.lww.com/JR9/A36). In addition, mutations exclusive to LR lesions were significantly enriched in broad pathways, such as histone methylation, DNA replication, regulation of the MAPK pathway, and T cell activation (Additional Fig. 5C, https://links.lww.com/JR9/A36). The mutational landscape of significantly enriched pathways is shown in (Additional Figure 5D, https://links.lww.com/JR9/A36). These results suggest that relapsing clones continually acquired driver mutations with diverse functions after dissemination from the primary tumor. To understand the seeding pattern between the primary tumor and LR lesion, we identified somatic mutations shared by matched samples to detect the disseminating cell clones, using PyClone-VI, based on a Bayesian clustering method. All the disseminating clones in the 10 MSS patients consisted of more than one cluster and showed variant allele frequency (VAF) dynamics between the primary and LR lesions (Additional Fig. 6, https://links.lww.com/JR9/A36). We also visualized the evolutionary models for representative patients using ClonEvol (Additional Fig. 7, https://links.lww.com/JR9/A36) and showed that LR required interactions among multiple distinct clones (polyclonal seeding) in MSS CRC patients.

F5
Figure 5.:
Phylogenetic trees of 13 patients with CRC and LR. Phylogenetic trees of (A) three MSI-H patients and (B) 10 MSS patients. The blue branch indicates exclusive mutations of primary lesions; The red branch indicates exclusive mutations of LR lesions; The black branch indicates the shared clonal mutations between primary and LR lesions; The purple branch indicates the subclonal mutations in primary lesion and present in LR lesion; Representative driver genes for LR lesions indicated with orange color. CRC = Colorectal cancer, LR = locoregional recurrence, MSI-H = microsatellite instability-high, MSS = microsatellite stable, P1–P13 = Patient 1–Patient 13.
F6
Figure 6.:
Biomarkers for early recurrence of CRC. (A) Mutation distribution of disseminating clone between patients with early (three patients) and late (seven patients) recurrence, the number indicates the mutation counts of each group (B) Comparison of SBS signatures between patients with early (three patients) and late recurrence (seven patients), the Y-axis indicates the relative SBS signature contribution. (C) Comparison of indel signatures between patients with early (three patients) and late (seven patients) recurrence, the Y-axis indicates the relative Indel signature contribution. (D) Comparison of DFS between ID4-positive (12 patients) and -negative (15 patients) patients from TCGA. P=0.04 (Wilcoxon test). CRC = Colorectal cancer, DFS = disease-free survival, SBS = single base substitution signature.

Clinical identification of biomarkers for early recurrence of CRC

Analysis of the clinical information for the MSS patients revealed recurrence intervals of <12 months in three patients and 21 to 48 months in the other seven patients. We investigated biomarkers of early recurrence in MSS patients using shared mutations between the primary and LR lesions as a surrogate for the disseminating clone. We assumed that the specific genomic variations in the disseminating cells could be used to discriminate between early and late recurrences. Comparison of the disseminating mutations in patients with early and late relapses revealed only one shared mutation (KRAS c.38G> A), 151 mutations exclusive to early-relapsed patients, and 446 exclusive to late-relapsed patients (Fig. 6A). However, the host genes at the mutation site showed high overlap in these two groups. Enrichment analysis further revealed that the host genes were distributed in similar pathways, including pathways in cancer, predominantly including APC, KRAS, TP53, TCF7L2, and other cancer-related genes (Additional Fig. 8A and B, https://links.lww.com/JR9/A36). We also wondered if the mutation signature of the disseminating cell could be used to differentiate between early and late recurrences. Although the SBS signatures were almost identical (Fig. 6B), there were distinct indel signatures between the early- and late-recurrence groups (Fig. 6C). Interestingly, ID4 and ID8 were specifically detected in early-relapse patients. According to the COSMIC database, the etiology of ID4 is unknown and ID8 is involved in the repair of DNA doublestrand breaks by non-homologous DNA end-joining mechanisms.

To confirm if CRC patients with a positive ID4 or ID8 signature relapsed earlier than those negative for ID4 and ID8, we downloaded the mutational and clinical data for 594 CRC patients in TCGA pan-cancer cohort from cbioportal. This cohort included 27 MSS patients who relapsed, for whom specific indel signature information was available. Disease-free survival (DFS) was significantly shorter in relapsed patients with an ID4 signature compared with those without ID4 (Wilcoxon P = 0.04) (Fig. 6D). Moreover, patients with an ID8 signature also had a shorter DFS, but the difference was not significant because of the small number of positive patients (Additional Fig. 8C, https://links.lww.com/JR9/A36). In addition, DFS was shorter in patients with either an ID4 or ID8 signature compared with patients without either signature (Additional Fig. 8D, https://links.lww.com/JR9/A36).

Discussion

To the best of our knowledge, this was the first study to systematically elucidate the genomic differences between primary and LR CRC tissues. Importantly, we revealed distinct patterns of disease progression between MSS and MSI-H patients, with few shared SNVs or indels in MSI-H patients, compared with 41.61% of somatic SNVs and indels shared between the primary and matched LR lesions in MSS patients. An average of 23% of mutations was specific to the primary lesions and 35.4% to the LR lesions, indicating that the mutation rate was higher in the LR compared with the primary clone. We also demonstrated more mutations in MSI-H than in MSS patients, consistent with a previous report.[29]

The most frequently mutated genes in the current cohort were genes with well-established roles in CRC,[30,31] such as KARS, TP53,and APC. However, the distributions of mutated KRAS and TP53 differed between MSS and MSI-H patients; mutated KRAS and TP53 were significantly less common in MSI-H compared with MSS patients, as confirmed in TCGA dataset. These results suggested that KRAS and TP53 mutations might follow a definite pattern in MSI-H patients. BRAF and KRAS mutation has been reported to be mutually exclusive, due to negative selection driven by oncogene-induced senescence.[32] Kumar et al[33] demonstrated that BRAF mutations were mainly distributed in MSI-H tumors, suggesting that KRAS mutations might also be predominantly distributed in MSS tumors. In addition, Lin et al. confirmed that the incidence of TP53 mutations was lower in MSI-H than in MSS tumors.[34] However, the mechanism remains unclear and requires further intensive investigations.

The relatively low incidence of LR means that the evolutionary pattern from primary to LR CRC remains unclear. We, therefore, explored the progression to LR. Analyses including mutation distribution, clonal dynamics, and phylogenetic trees demonstrated that LR occurred as a result of germline variations in mismatch repair genes in MSI-H patients, and evolved independently; that is, matched primary and LR lesions in MSI-H patients could be regarded as two metachronous primary tumors. This strongly supports the use of high-throughput sequencing to clarify the nature of the “recurrence,” especially in individuals with a genetic predisposition and in relation to decisions regarding systematic therapy in the adjuvant setting.

In contrast, the matched primary and LR lesions were clonally related in MSS patients, as indicated by the persistence of the mutational signature in the primary lesion through to the LR lesion, indicating that the mutation signature could be extended from the primary tumor to the disseminating cells. In addition, clonal transition analysis showed the presence of shared subclonal mutations between primary and LR lesions in each patient. Finally, the phylogenetic trees also showed shared truncal mutations with distinct numbers. Taken together, these different lines of evidence indicate distinct origins of LRs between MSS and MSI-H patients.

This study also clarified the seeding pattern of LR. The primary tumors in all 10 MSS demonstrated polyclonal seeding of the disseminating clones. This shows some conflict with the reported dissemination patterns in metastatic CRC,[35,36] with a monoclonal seeding pattern between primary CRC and metastatic lesions being relatively common. It is speculated that metastasis of the disseminating clone to distant organs is associated with greater selective pressure due to the longer transmission distance, meaning that only the best-adapted clones survive. In contrast, the relatively shorter transmission distance required for LR allowed multiple clones to evade selection pressure and establish new lesions near the primary site.

We further explored the progression of relapsed lesions after divergence from the primary tumor. Late-acquired mutations encompassed a wide range of driver genes enriched in functions including histone methylation, DNA replication, regulation of the MAPK pathway, and T cell activation. These findings expanded the range of potential therapeutic targets for relapsed CRC, such as PRMT, KDM, ATR, and KRAS inhibitors and cytokines.[37–39] Notably, CNV analysis identified copy number alterations in two driver genes that were enriched in LR lesions. PDCD1 (PD-1) encodes a cell surface membrane protein of the immunoglobulin superfamily and associates with CD3-TCR in the immunological synapse and directly inhibits T-cell activa-tion.[40] It may thus be possible to develop immunotherapy for relapsed CRC patients whose genome harbors increased copies of PDCD1 . Moreover, loss of LMNA, which encodes a two-dimensional matrix of proteins located next to the inner nuclear membrane and which contributes to nuclear stability, was also enriched in LR lesions. Loss of LMNA expression was previously reported to be associated with disease recurrence in stage II and III colon cancer.[41] These lines of evidence suggest that loss of LMNA may be a risk biomarker for LR in CRC.

In line with oncological practice, we aimed to stratify patients with LR. The mutational signatures ID4 and ID8 were identified as biomarkers for early recurrence. ID4-positivity was significantly associated with shorter DFS in patients who experienced recurrence. However, the low incidence of ID8-positivity in relapsed patients meant that it only showed a loose correlation with shorter DFS. It is important for doctors to be able to distinguish patients susceptible to LR and to thus implement the appropriate clinical management.

Limitations

Several limitations of this study should be indicated. First, the sample size of this cohort is small, and the relatively small sample size may reduce the persuasiveness of the study. The second aspect to be emphasized is certainly the difficulty to implement the biomarker detection in clinical practice due to the cost of time and money, therefore not very feasible for prognostic purposes in clinical practice. Despite these limitations, we still elaborately described the dynamics and disseminating types of LR, which provided a new perspective for further research.

Conclusion

LR of CRC is a complex process that may require cooperation among multiple subclones or may occur via multiple rounds of seeding involving distinct clones. These findings pose a challenge to the development of new therapeutic approaches targeting these interactions to inhibit LR.

Acknowledgments

None.

Author contributions

XL participated in data collection and manuscript drafting. XxW participated in investigation and manuscript drafting. CZ participated in data analysis and manuscript drafting. XhW was responsible for investigation. GW was responsible for data validation. BL, WQ, and DL were responsible for project administration. HW implemented the study. YD was responsible for funding acquisition. JY, ZT, and ZY participated in data analysis. ZL and DS were responsible for conceptualization. LL was responsible for supervision, manuscript review, and editing. All authors approved the final version of the manuscript.

Financial support

This work was supported by the National Key R&D Program of China (No. 2017YFC1309002), National Natural Science Foundation of China (Nos. 81672821, 81872041, 81472313, 81773101, 81903002, and 82003059), China Postdoctoral Science Foundation (Nos. 2019M652963 and 2020M682624), Key projects of Guangdong Natural Science Foundation (No. 2018B0303110017), and Guangdong Provincial Key Laboratory of Precision Medicine for Gastrointestinal Cancer (No. 2020B121201004). The funding sources had no role in the design of this study and did not have any role during its execution, analyses, data interpretation, or decision to submit results.

Institutional review board statement

The study was approved by the Institutional Review Board of Nanfang Hospital of Southern Medical University on September 11, 2020 (approval No. 2020010) and conducted in accordance with the 1964 Declaration of Helsinki, as revised in 2013.

Conflicts of interest

CZ, JY, ZT, and ZY are current employees of Geneplus-Shenzhen. No other actual or potential conflict of interest is declared.

References

[1]. Siegel RL, Miller KD, Jemal A. Cancer statistics, 2019. CA Cancer J Clin 2019;69:7–34.
[2]. Larkin JO, O’Connell PR. Multivisceral resection for T4 or recurrent colorectal cancer. Dig Dis 2012;30(Suppl 2):96–101.
[3]. Lehnert T, Methner M, Pollok A, et al. Multivisceral resection for locally advanced primary colon and rectal cancer: an analysis of prognostic factors in 201 patients. Ann Surg 2002;235:217–225.
[4]. Bowne WB, Lee B, Wong WD, et al. Operative salvage for locoregional recurrent colon cancer after curative resection: an analysis of 100 cases. Dis Colon Rectum 2005;48:897–909.
[5]. Kraemer M, Wiratkapun S, Seow-Choen F, et al. Stratifying risk factors for follow-up: a comparison of recurrent and nonrecurrent colorectal cancer. Dis Colon Rectum 2001;44:815–821.
[6]. Heald R. The ‘Holy Plane’ of rectal surgery. J R Soc Med 1988;81:503–508.
[7]. Wiig JN, Carlsen E, Soreide O. Mesorectal excision for rectal cancer: a view from Europe. Semin Surg Oncol 1998;15:78–86.
[8]. Sauer R, Becker H, Hohenberger W, et al. Preoperative versus postoperative chemoradiotherapy for rectal cancer. N Engl J Med 2004;351:1731–1740.
[9]. Rodel C, Liersch T, Becker H, et al. Preoperative chemoradiotherapy and postoperative chemotherapy with fluorouracil and oxaliplatin versus fluorouracil alone in locally advanced rectal cancer: initial results of the German CAO/ARO/AIO-04 randomised phase 3 trial. Lancet Oncol 2012;13:679–687.
[10]. Kogler P, Kafka-Ritsch R, Sieb M, et al. Therapeutic management and outcome of locoregional recurrence after curative colorectal cancer therapy-a single-center analysis. J Gastrointest Surg 2014;18:2026–2033.
[11]. Liska D, Stocchi L, Karagkounis G, et al. Incidence, patterns, and predictors of locoregional recurrence in colon cancer. Ann Surg Oncol 2017;24:1093–1099.
[12]. Røkke O, Heggelund T, Benth J, et al. Clinical predictors for recurrence after curative resection for colorectal cancer. J Cancer Ther 2017;8: 1107–1124.
[13]. Tie J, Wang Y, Tomasetti C, et al. Circulating tumor DNA analysis detects minimal residual disease and predicts recurrence in patients with stage II colon cancer. Sci Transl Med 2016;8:346ra92.
[14]. Matikas A, Voutsina A, Trypaki M, et al. Role of circulating free DNA in colorectal cancer. World J Gastrointest Oncol 2016;8:810–818.
[15]. Erstad DJ, Tumusiime G, Cusack JC Jr. Prognostic and predictive biomarkers in colorectal cancer: implications for the clinical surgeon. Ann Surg Oncol 2015;22:3433–3450.
[16]. Li H, Durbin R. Fast and accurate short read alignment with BurrowsWheeler transform. Bioinformatics 2009;25:1754–1760.
[17]. Cibulskis K, Lawrence MS, Carter SL, et al. Sensitive detection of somatic point mutations in impure and heterogeneous cancer samples. Nat Biotechnol 2013;31:213–219.
[18]. McKenna A, Hanna M, Banks E, et al. The Genome Analysis Toolkit: a MapReduce framework for analyzing next-generation DNA sequencing data. Genome Res 2010;20:1297–1303.
[19]. Hübschmann D, Jopp-Saile L, Andresen C, et al. Analysis of mutational signatures with yet another package for signature analysis. Genes Chromosomes Cancer 2021;60:314–331.
[20]. Nilsen G, Liestol K, Van Loo P, et al. Copynumber: Efficient algorithms for single- and multi-track copy number segmentation. BMC Genomics 2012;13:591.
[21]. Carter SL, Cibulskis K, Helman E, et al. Absolute quantification of somatic DNA alterations in human cancer. Nat Biotechnol 2012;30: 413–421.
[22]. Bailey MH, Tokheim C, Porta-Pardo E, et al. Comprehensive characterization of cancer driver genes and mutations. Cell 2018;174: 1034–1035.
[23]. Vogelstein B, Papadopoulos N, Velculescu VE, et al. Cancer genome landscapes. Science 2013;339:1546–1558.
[24]. Gillis S, Roth A. PyClone-VI: scalable inference of clonal population structures using whole genome data. BMC Bioinformatics 2020;21:571.
[25]. Dang HX, White BS, Foltz SM, et al. ClonEvol: clonal ordering and visualization in cancer sequencing. Ann Oncol 2017;28:3076–3082.
[26]. R.C.R. Team. A Language and Environment for Statistical Computing, 2018. https://www.R-project.org/. Accessed date July 06, 2020.
[27]. Cancer Genome Atlas N. Comprehensive molecular characterization of human colon and rectal cancer. Nature 2012;487:330–337.
[28]. Li E, Hu Y, Han W, et al. The mutational landscape of MSI-H and MSS colorectal cancer. J Clin Oncol 2019;37:e15122.
[29]. Timmermann B, Kerick M, Roehr C, et al. Somatic mutation profiles of MSI and MSS colorectal cancer identified by whole exome next generation sequencing and bioinformatics analysis. PLoS One 2010;5: e15661.
[30]. Yaeger R, Chatila WK, Lipsyc MD, et al. Clinical sequencing defines the genomic landscape of metastatic colorectal cancer. Cancer Cell 2018;33: 125–136.e123.
[31]. Ishaque N, Abba ML, Hauser C, et al. Whole genome sequencing puts forward hypotheses on metastasis evolution and therapy in colorectal cancer. Nat Commun 2018;9:4782.
[32]. Cisowski J, Sayin VI, Liu M, et al. Oncogene-induced senescence underlies the mutual exclusive nature of oncogenic KRAS and BRAF. Oncogene 2016;35:1328–1333.
[33]. Kumar K, Brim H, Giardiello F, et al. Distinct BRAF (V600E) and KRAS mutations in high microsatellite instability sporadic colorectal cancer in African Americans. Clin Cancer Res 2009;15:1155–1161.
[34]. Lin EI, Tseng LH, Gocke CD, et al. Mutational profiling of colorectal cancers with microsatellite instability. Oncotarget 2015;6:42334–42344.
[35]. Dang HX, Krasnick BA, White BS, et al. The clonal evolution of metastatic colorectal cancer. Sci Adv 2020;6:eaay9691.
[36]. Hu Z, Ding J, Ma Z, et al. Quantitative evidence for early metastatic seeding in colorectal cancer. Nat Genet 2019;51:1113–1122.
[37]. Song Y, Wu F, Wu J. Targeting histone methylation for cancer therapy: enzymes, inhibitors, biological activity and perspectives. J Hematol Oncol 2016;9:49.
[38]. Yazinski SA, Comaills V, Buisson R, et al. ATR inhibition disrupts rewired homologous recombination and fork protection pathways in PARP inhibitorresistant BRCA-deficient cancer cells. Genes Dev 2017;31:318–332.
[39]. Braicu C, Buse M, Busuioc C, et al. A comprehensive review on MAPK: a promising therapeutic target in cancer. Cancers (Basel) 2019;11:1618.
[40]. Syn NL, Teng MWL, Mok TSK, et al. De-novo and acquired resistance to immune checkpoint targeting. Lancet Oncol 2017;18:e731–e741.
[41]. Belt EJT, Fijneman RJA, van den Berg EG, et al. Loss of lamin A/C expression in stage II and III colon cancer is associated with disease recurrence. Eur J Cancer 2011;47:1837–1845.
Keywords:

biomarker; colorectal cancer; locoregional recurrence; polyclonal seeding; tumor evolution

Supplemental Digital Content

Copyright © 2022 The Chinese Medical Association, Published by Wolters Kluwer Health, Inc.