Robust Assessment of Homologous Recombination Deficiency Genomic Instability by OncoScan Microarrays

Abstract

Genomic instability scars are markers for detecting homologous recombination deficiency (HRD) status in patients with ovarian cancer and predicting the response to poly (ADP-ribose) polymerase inhibitor treatment. Currently, only a few reliable and validated assays are available, with the Myriad myChoice CDx being the most commonly used commercial assay for genomic instability scar score determination. Given the need for a more straightforward, accessible, and reliable method for detecting genomic instability scars methods, in this work, we describe the feasibility of using the microarray OncoScan copy number variant assay and open-source software packages to quantify genomic instability scores, and the development of an open-access online platform for genomic instability score calculation. The laboratory-developed test accurately classified homologous recombination-proficient and recombination-deficient samples based on genomic instability scores derived from the OncoScan copy number variant assay. Internally evaluated genomic instability scores demonstrated a 92% overall agreement and a higher sample success rate compared with externally analyzed genomic instability scar scores. The availability of HRD determination has doubled the number of patients eligible for poly (ADP-ribose) polymerase therapy. The assay can be conveniently performed on individual samples, and the open-access online platform facilitates HRD determination without the need for specialized bioinformatics support.

Figure 1

Diagram of laboratory-developed test (LDT) process applied to the samples. CNV, copy number variant; FFPE, formalin fixed, paraffin embedded; GIS, genomic instability scar; H&E, hematoxylin and eosin; HRD, homologous recombination deficiency; Min, minimum; OvCa, ovarian cancer; PARPi, poly (ADP-ribose) polymerase inhibitor.

Figure 2

A–C: Comparison of Myriad myChoice CDx and laboratory-developed test (LDT) genomic instability score analysis results. Pathogenic BRCA1/2 variants in violet, and nonpathogenic or non-BRCA1/2 variants in green. A and B: Genomic instability scar (GIS) scores from 22 samples corresponding to the training group without (A) and with (B) GIS-corrected scores based on linear regression analysis correlation value of 0.83. Linear regression model provided the correction formula, where m represents the slope and b represents the value at which the y axis is intercepted. C: Validation cohort with applied correction, with correlation value of 0.85. In both training and validation groups, all cases with pathogenic BRCA1/2 variants have positive GIS score (≥42). Neg, negative; Pos, positive.

Figure 3

A–D: Allele-specific copy number and B-allelic frequency (BAF) plots from samples with low genomic instability scar (GIS) score (<42; A and B) and high GIS score (≥42; C and D). A and B: Allele-specific copy number plots where the major copy number is represented in red and minor copy number in green. A: Copy number alterations cover large sections of each chromosome (Chr) across the genome. C: There is large variation of copy number across the genome. B and D: BAF plot shows higher allelic imbalance (D) than in corresponding allelic-specific copy number profile plots (B).

Figure 4

A and B: Genomic instability scar (GIS) scores and BRCA1/2 mutation status of 654 samples analyzed externally by Myriad myChoice CDx (A) and 280 internally analyzed samples (B). Pathogenic BRCA1/2 variants represented as in violet, nonpathogenic or no variants in green, and unknown BRCA1/2 status in blue. Dashed line represents the GIS threshold value of 42 at which a sample is considered GIS positive (POS; ≥42) or GIS negative (NEG; <42). A: Myriad GIS-positive samples correspond to 41.4% (271/654) from which 41.3% (112/271) harbor BRCA1/2 pathogenic variants. B: In a similar manner, laboratory-developed test GIS-positive samples correspond to 48.4% (136/280) from which 33.1% (45/136) harbor pathogenic BRCA1/2 variants.