Genomic Scar Score: A robust model predicting homologous recombination deficiency based on genomic instability

Abstract

Objective To develop a novel machine learning‐based algorithm called the Genomic Scar Score (GSS) for predicting homologous recombination deficiency (HRD) events. Design Method development study. Setting AmoyDx Medical Laboratory and Jiangsu Cancer Hospital. Population or sample A cohort of individuals with ovarian or breast cancer (n = 377) were collected from the AmoyDx Medical Laboratory. Another cohort of patients with ovarian cancer treated with PARP inhibitors (n = 58) was enrolled in the Jiangsu Cancer Hospital. Methods We used linear support vector machines to build a Genomic Scar (GS) model to predict HRD events, and Kaplan–Meier analyses were performed by comparing the progression‐free survival (PFS) of patients in different groups using a two‐sided log‐rank test. Main outcome measures The performance of the GS model and the result of clinical validation. Results The GS model displayed more than 97.0% sensitivity to detect BRCA‐deficient events, and the GS model identified patients that could benefit from poly(ADP‐ribose) polymerase inhibitors (PARPi), as the GS score (GSS)‐positive group had a longer progression‐free survival (PFS) (9.4 versus 4.4 months; hazard ratio [HR] = 0.54, P < 0.001) than the GSS‐negative group after PARPi treatment. Meanwhile, the GSS showed high concordance among different NGS panels, which implied the robustness of the GS model. Conclusions The GS was a robust model to predict HRD and had broad clinical applications in predicting which patients will respond favourably to PARPi treatment.

Relationship between homologous recombination deficiency and clinical features of breast cancer based on genomic scar score

“…GSS was developed by a machine learning method called the genomic scar (GS) model, which use different types of chromosomal copy number to evaluate genomic instability to predict HRD event [ 33 ]. GSS applied in this HRD test shares two prominent advantages.…”

“…Noteworthy, the development of the model incorporated data from breast and ovarian cancers. However, the predictive power of GS models in the clinic has only been evaluated in patients with ovarian cancer [ 33 ]. Our study on breast cancer patients filled this gap.…”

“…The tumor cell content for the HRD detection was not less than 30%. HRD Complete panel (Amoydx Ltd., Xiamen, China), covering 1.5 MB of human genomic regions, and including all exons of BRCA1/2 and 24,000 (24 k) target SNPs distributed across the complete human genome [ 33 ]. Sequencing was performed by NovaSeq 6000 platform (Illumina, Inc).…”

“…Finally, through support vector machine (SVM) training, the weights of entity description items were summed to obtain the Genomic Scar Score (GSS). GSS ≥50 was considered positive for HRD [ 33 ].…”

Somatic and germline aberrations in homologous recombination repair genes in Chinese prostate cancer patients

“…In terms of the quality control for calling, a sample would not pass the quality control if: the total depth of mutation and wild-type alleles was lower than 100× for somatic sequencing or 20× for germline sequencing; the depth of a mutation allele was lower than 5×; the allelic frequency was lower than 3% for somatic sequencing or 20% for germline sequencing; the base quality of a mutation was lower than 30; the base quality of mutation allele minus the average base quality of both mutation and wild-type alleles was smaller than -2; the read quality of mutation allele minus the average read quality of both mutation and wild-type alleles was smaller than -2; the mapping quality of mutation allele minus the average mapping quality of both mutation and wild-type alleles was smaller than -0.3. These parameters had been validated in previous studies (20,21).…”