Search

Simple Summary: At a genomic level metastatic castrate resistant prostate cancer state is highly heterogeneous and no clear genome-based prognostic or predictive biomarkers exist in practice. We evaluated multiple copy number somatic alterations in two castrate resistant patient cohorts to determine if a genome-based risk score at the copy number level can predict clinical outcomes. The first cohort included patients in a prospective clinical-trial in which abiraterone acetate was given and the other comprised of a real-world hospital-registry. We extracted plasma cell free DNA in both cohorts and performed low pass whole genome sequencing. Copy number alterations were identified for 24 candidate genes and a final composite score developed from 11 genes. This risk score was able to predict survival in castrate resistant patients after adjusting for known clinical biomarkers. Additionally, the multi-gene copy number alteration based risk score algorithm also predicted if abiraterone acetate would be effective in castrate resistant patients. A plasma cell-free DNA (cfDNA) multi-gene copy number alteration (CNA)-based risk score was evaluated to predict clinical outcomes in metastatic castrate resistant prostate cancer (mCRPC) patients. Methods: Plasma specimens from two independent mCRPC patient cohorts (N = 88 and N = 92 patients) were used. A treatment-naïve mCRPC cohort (prospective clinical-trial cohort) included plasma samples before treatment with abiraterone acetate/prednisone and serially at 3-months. A separate real-world hospital-registry (RWHR) mCRPC cohort included a single blood sample collected prior to mCRPC treatments in 92 mCRPC patients following ADT failure. Low pass whole genome sequencing was performed on plasma cell-free DNA (cfDNA) and copy number alterations (CNAs) were identified for 24 candidate genes of interest. Associations of individual gene CNAs with 3 month primary resistance to therapy, progression-free survival (PFS) in the prospective trial cohort and overall survival (OS) in both cohorts was evaluated by Cox regression. A multi-gene risk score was determined for significantly associated candidate CNAs for predicting clinical outcomes. Clinical factors were included in the risk model for survival. Statistical significance for all tests was set at 0.05. Results: In the prospective trial cohort, patients responding to treatment were observed to have a significant copy number decrease in AR (p = 0.001) and COL22A1 (p = 0.037) at 3 months, while the non-responder group showed a significant CNA decrease in NKX3.1 (p = 0.027), ZBTB16 (p = 0.025) and CNA increases in PIK3CB (p = 0.006). Based on the significance level of each gene, CNAs in 11 of the 24 genes (AR, COL22A1, MYC, NCOR1, NKX3.1, NOTCH1, PIK3CA, PIK3CB, TMPRSS2, TP53, ZBTB16) were selected to develop a Cox-regression coefficient-based weighted multi-gene risk score for predicting mCRPC outcomes in both cohorts. A higher multi-gene risk score was observed to have poor OS in mCRPC patients in the prospective trial cohort (p = 0.00019) and for the RWHR cohort, (p < 0.0001). A higher risk score was also associated with poor PFS in the prospective cohort (p = 0.0043). Conclusions: A multi-gene CNAs-based risk score derived from plasma cfDNA may predict treatment response and prognosticate survival in mCRPC and warrants prospective validation of risk-based algorithms. [ABSTRACT FROM AUTHOR]

Molecular analysis of cell-free DNA (cfDNA) that circulates in plasma and other body fluids represents a "liquid biopsy" approach for non-invasive cancer screening or monitoring. The rapid development of sequencing technologies has made cfDNA a promising source to study cancer development and progression. Specific genetic and epigenetic alterations have been found in plasma, serum, and urine cfDNA and could potentially be used as diagnostic or prognostic biomarkers in various cancer types. In this review, we will discuss the molecular characteristics of cancer cfDNA and major bioinformatics approaches involved in the analysis of cfDNA sequencing data for detecting genetic mutation, copy number alteration, methylation change, and nucleosome positioning variation. We highlight specific challenges in sensitivity to detect genetic aberrations and robustness of statistical analysis. Finally, we provide perspectives regarding the standard and continuing development of bioinformatics analysis to move this promising screening tool into clinical practice. [ABSTRACT FROM AUTHOR]