ADVERTISEMENT
O-020
Copy number variation in longitudinal liver metastases biopsies in colorectal cancer identifies biomarker candidates of resistance to standard chemotherapy
Introduction
We have previously defined ‘next generation specimens’ as longitudinal biopsy samples collected at different time points during the disease trajectory, in a specific treatment context, either for discovery research or to guide clinical decisions (Basik, M. et al. Nat Rev Clin Oncol. 2013). The objective of the present study was to track tumor evolution at the copy number (CN) level and capture emerging resistance mechanisms in metastatic colorectal cancer (mCRC) using a standardized collection of high-quality metastasis samples, at baseline and at the time of clinical resistance, in the context of a phase IV multicenter clinical trial (NCT00984048).
We have previously defined ‘next generation specimens’ as longitudinal biopsy samples collected at different time points during the disease trajectory, in a specific treatment context, either for discovery research or to guide clinical decisions (Basik, M. et al. Nat Rev Clin Oncol. 2013). The objective of the present study was to track tumor evolution at the copy number (CN) level and capture emerging resistance mechanisms in metastatic colorectal cancer (mCRC) using a standardized collection of high-quality metastasis samples, at baseline and at the time of clinical resistance, in the context of a phase IV multicenter clinical trial (NCT00984048).
Methods
One hundred and forty fresh frozen liver metastasis (LM) specimens collected at baseline (pre, n = 97) and at the time of clinical resistance (post, n = 43) from 119 mCRC patients undergoing standard first-line therapy were profiled using whole exome sequencing and characterized for CN variation using Nexus Copy Number software. RNA sequencing data available for the same sample set allowed filtering the CN aberration (CNA) candidates for mRNA correlation in each analysis performed. Objective response was attributed based on the RECIST v1.0 criteria. Genomic Identification of Significant Target in Cancer (GISTIC) test was used to identify focal CNA. CNA frequency reported as being significantly different between 2 groups met a minimum p-value of 0.005 on a two-tailed Fisher’s exact test. The log-rank test was used to identify CNA associated with progression-free survival (PFS). The threshold used for significance was permutated p-value < .005.
One hundred and forty fresh frozen liver metastasis (LM) specimens collected at baseline (pre, n = 97) and at the time of clinical resistance (post, n = 43) from 119 mCRC patients undergoing standard first-line therapy were profiled using whole exome sequencing and characterized for CN variation using Nexus Copy Number software. RNA sequencing data available for the same sample set allowed filtering the CN aberration (CNA) candidates for mRNA correlation in each analysis performed. Objective response was attributed based on the RECIST v1.0 criteria. Genomic Identification of Significant Target in Cancer (GISTIC) test was used to identify focal CNA. CNA frequency reported as being significantly different between 2 groups met a minimum p-value of 0.005 on a two-tailed Fisher’s exact test. The log-rank test was used to identify CNA associated with progression-free survival (PFS). The threshold used for significance was permutated p-value < .005.
Results
The CN variation landscape of 97 pre LM samples confirmed the most common chromosome (chr) arm amplifications being on chr 7p, 7q, 8q, 13q, and 20q and the most frequent deletions on chr 1p, 1q, 4p, 4q, 8p, 17p, 18p, 18q, and 22q. Twenty-nine focal aberrations were also identified by GISTIC. Comparative analyses identified 22 focal CNA with significantly different frequencies between pre- and post- samples. The comparison between partial responder (PR, n = 47) and intrinsically resistant (IRES, n = 5) lesions revealed 34 CNA with different frequencies at baseline. Acquired resistant (ARES) and IRES post-samples (n = 8) showed significant enrichment in 36 CNA compared to pre-sample (n = 97), and these CNA were not captured when PR (n = 10) post-samples are compared to pre-samples. One hundred thirty-seven CNA regions were found to be significantly associated with PFS. Interrogation of primary tumors and LM data available in public databases validated the predictive potential of 18 regions and also revealed the metastasis specificity of 15 of them. Candidate genes in these regions with concordant genomic/transcriptomic aberrations were identified.
The CN variation landscape of 97 pre LM samples confirmed the most common chromosome (chr) arm amplifications being on chr 7p, 7q, 8q, 13q, and 20q and the most frequent deletions on chr 1p, 1q, 4p, 4q, 8p, 17p, 18p, 18q, and 22q. Twenty-nine focal aberrations were also identified by GISTIC. Comparative analyses identified 22 focal CNA with significantly different frequencies between pre- and post- samples. The comparison between partial responder (PR, n = 47) and intrinsically resistant (IRES, n = 5) lesions revealed 34 CNA with different frequencies at baseline. Acquired resistant (ARES) and IRES post-samples (n = 8) showed significant enrichment in 36 CNA compared to pre-sample (n = 97), and these CNA were not captured when PR (n = 10) post-samples are compared to pre-samples. One hundred thirty-seven CNA regions were found to be significantly associated with PFS. Interrogation of primary tumors and LM data available in public databases validated the predictive potential of 18 regions and also revealed the metastasis specificity of 15 of them. Candidate genes in these regions with concordant genomic/transcriptomic aberrations were identified.
Conclusion
Our effort to collect and profile next-generation biospecimens provided global genomic and transcriptomic data of the largest LM cohort to date. By analyzing the association between molecular aberrations and patient outcomes, we identified novel biomarker candidates predictive of drug response in LM samples, supporting the utility of collecting metastatic samples in clinical settings.
Our effort to collect and profile next-generation biospecimens provided global genomic and transcriptomic data of the largest LM cohort to date. By analyzing the association between molecular aberrations and patient outcomes, we identified novel biomarker candidates predictive of drug response in LM samples, supporting the utility of collecting metastatic samples in clinical settings.
Publisher
Oxford University Press
Source Journal
Annals of Oncology
