Performance evaluation of an amplicon‐based next‐generation sequencing panel for BRCA1 and BRCA2 variant detection

Abstract Background As next‐generation sequencing (NGS) technology matures, various amplicon‐based NGS tests for BRCA1/2 genotyping have been introduced. This study was designed to evaluate an NGS test using a newly released amplicon‐based panel, AmpliSeq for Illumina BRCA Panel (AmpliSeq panel), for detection of clinically significant BRCA variants, and to compare it to another amplicon‐based NGS test confirmed by Sanger sequencing. Methods We reviewed BRCA test results done by NGS using the TruSeq Custom Amplicon kit from patients suspected of hereditary breast/ovarian cancer syndrome (HBOC) in 2018. Of those, 96 residual samples with 100 clinically significant variants were included in this study using predefined criteria: 100 variants were distributed throughout the BRCA1 and BRCA2 genes. All target variants were confirmed by Sanger sequencing. Duplicate NGS testing of these samples was performed using the AmpliSeq panel, and the concordance of results from the two amplicon‐based NGS tests was assessed. Results Ninety‐nine of 100 variants were detected in duplicate BRCA1/2 genotyping using the AmpliSeq panel (sensitivity, 99%; specificity, 100%). In the discordant case, one variant (BRCA1 c.3627dupA) was found only in repeat 1, but not in repeat 2. Automated nomenclature of all variants, except for two indel variants, was in consensus with Human Genome Variation Society nomenclature. Conclusion Our findings confirm that the analytic performance of the AmpliSeq panel is satisfactory, with high sensitivity and specificity.


| INTRODUC TI ON
In 1994, linkage analysis in large numbers of families identified BRCA1 and BRCA2 as genes associated with predisposition for hereditary breast/ovarian cancer syndrome (HBOC). 1,2 Approximately 5%-10% and 20% of breast and ovarian cancer cases are considered hereditary tumors, 3,4 but only 25% of HBOC are associated with BRCA1/2 pathogenic variants, which affect DNA repair mechanisms. 5 Carriers with BRCA1/2 pathogenic variants have a higher risk of developing breast cancer (60%-85%) and ovarian cancer (15%-40%) over their lifetime. 6,7 In BRCA-mutated patients, both intensive screening (including MRI) and prophylactic surgery or chemical treatment decrease cancer risk and overall mortality. [8][9][10] Among triple-negative breast cancer patients, platinum-based chemotherapeutic agents are favorable for BRCA1/2 variant carriers. 11 Recently, poly ADPribose polymerase (PARP) inhibitors were reported to improve prognosis in patients with BRCA-mutated metastatic ovarian cancer. 12 Collectively, these reports indicate that testing for BRCA1/2 mutation plays a significant role in the choice of therapy, as well detection of the genetic cause.
Next-generation sequencing (NGS) was introduced to clinical laboratories for multi-gene and high-throughput analysis. 13 Subsequently, NGS has been developed as a powerful tool for detecting BRCA1/2 variants. 14-17 Although the high performance and cost-effectiveness of the NGS technique are well known, the diversity of NGS platforms, enrichment methods, and analytic pipelines represents a potential obstacle to implementation. Because amplicon-based methods for enrichment have several strengths, including lower cost, shorter preparation time, and smaller quantities of input DNA in comparison with capture methods, 18

| MATERIAL S AND ME THODS
The Institutional Review Board/Ethics Committee of Asan Medical Center waived the requirement for informed consent for this study (2019-0044).

| Sample selection and DNA extraction
This study was performed at a single center. In 2018, 883 patients diagnosed with breast or ovarian cancer suspected to be HBOC A schematic workflow of this study is shown in Figure 1.

| AmpliSeq panel-based NGS
A single NGS platform, MiSeqDx, was adopted for the detection of small indel and single-nucleotide variants. The AmpliSeq for Illumina

| Bioinformatic analysis
Human genome build 19 (hg19) was used for alignment. Analysis

| Statistical analysis
To evaluate performance, the results from NGS using AmpliSeq panel were compared with the NGS results obtained using TruSeq kit and confirmed by Sanger sequencing. Sensitivity, specificity, negative predictive value, and positive predictive value were determined, and 95% confidence intervals were calculated using the efficientscore method. The NGS test using AmpliSeq panel was performed in duplicate to examine reproducibility.

| Target variants
We analyzed a total of 100 variants, comprising 66 single-nucleotide variants and 34 indel variants. Characteristics of these variants are shown in Figure 2, and all variants are listed in Table S1. Only two

| Technical performance
A total of eight runs were performed: four batches, each containing 24 samples, were repeated. The quality control (QC) parameters of sequencing using the AmpliSeq panel were acceptable for all runs. The specific values of QC parameters are listed in Table 1.
In the discordant case, one variant (BRCA1 c.3627dupA) was called only in repeat 1, but not in repeat 2. However, the variant was visible with low variant allele frequency (VAF) (19.4%) (Figure 3).
After detailed review, we determined that all sixteen variants except one (from another sample in the same batch) were called: The exceptional case was detectable only on IGV due to low VAF. These observations suggested a possible error in sample preparation.

| Variant annotation
After reclassification, eight variants from 39 target variants previously classified as VUS were designated as benign or likely benign. All were missense variants: five in BRCA1 and four in BRCA2. Reclassification was mainly due to observations with a pathogenic variant. Thus, 31 variants remained as VUS. These are also listed in Table S1.
The nomenclature of all but two of the detected target variants was consistent with HGVS recommendations. In the two exceptional cases, indel variants (BRCA1 c.922_924delAGCinsT and BRCA1 c.5496_5506delinsA) were detected as two individual variants (one insertion variant and one deletion variant). These variants were observed in cis after sequence confirmation on IGV and were therefore reclassified as single indel variants.  19,20,28 Collectively, these findings confirm the high performance of NGSbased BRCA1/2 testing.

| D ISCUSS I ON
In regard to the error in sample preparation, we note that the QC results from this run were acceptable. To avoid such errors, detected variants should be compared with variants found in other samples from the same run, and abnormal samples should be re-examined.
Other validation studies reported limitations due to technical factors such as low average coverage depth. 19,28 In addition to those sources of errors, procedural errors, such as in our case, do occur (albeit rarely) in the clinical laboratory. Therefore, this report emphasizes the need for clinical laboratories to make their best efforts to decrease errors in procedures.
In this study, indel variants were separated into insertion and de-  In conclusion, this study shows that the analytic performance of AmpliSeq panel is satisfactory, with high sensitivity and specificity.
Therefore, the AmpliSeq panel performs sufficiently well to be implemented in the clinical laboratory for detection of BRCA1/2 variants. Further improvement in testing and bioinformatic platforms will be required to overcome the remaining limitations with regard to detection of CNVs detection and calling and annotation of indel variants.

ACK N OWLED G M ENTS
The authors thank Illumina, Inc for providing some library preparation kits and sequencing reagents.

CO N FLI C T O F I NTE R E S T
The authors thank Illumina, Inc for providing some library preparation kits and sequencing reagents.