INTERNATIONAL CONTROL ROUND FOR DEEP SEQUENCING ANALYSIS OF BCR-ABL KINASE DOMAIN MUTATIONS IN 11 LABORATORIES FROM 7 EUROPEAN COUNTRIES

Background
Point mutations in the BCR-ABL kinase domain are the most frequently identified mechanisms of acquired resistance towards tyrosine kinase inhibitors in chronic myeloid leukemia (CML). Although Sanger sequencing is still regarded as the gold standard technique for routine BCR-ABL mutation screening, next-generation sequencing (NGS) has evolved rapidly and is accessible to an increasing number of diagnostic laboratories. Thus far, data is limited on the technical performance of NGS for BCR-ABL mutation screening in a clinical diagnostic setting.

Aims
As an international consortium of 11 laboratories in 7 countries, we sought to investigate the robustness, precision, and reproducibility of NGS for BCR-ABL mutation analysis. The study was conducted as a subproject within the Interlaboratory Robustness of Next-generation sequencing (IRON)-II study.

Methods
Optimized PCR protocols and preconfigured 96-well plates containing lyophilized primer pairs targeting the entire BCR-ABL kinase domain were generated and distributed to each participating laboratory. To evaluate performance, 22 blinded control samples were sent to each laboratory (a total of 242 samples). Seventeen control samples contained cDNA of Ba/F3^BCR-ABL cell lines harboring 12 different BCR-ABL kinase domain mutations that were mixed with non-mutated Ba/F3^BCR-ABL to produce dilutions ranging from 1% to 100% of mutant alleles. Five control samples contained cDNA of non-mutated Ba/F3^BCR-ABL. All control samples were diluted into HL60 cells to simulate a BCR-ABL level of 10% on the International Scale. NGS was performed on 454 GS Junior sequencing instruments using 454 GS Junior Titanium chemistry for amplicon sequencing.

Results
Overall, a median of 104,732 high quality sequencing reads were generated across each of the 11 laboratories, where the number of sequencing reads ranged from 11,307 to 185,775. The mean read length obtained in all runs was 356 bases (range, 346 to 363), indicating a homogenous read length pattern across all centers. A combined number of % mixed and % dots filtered reads of less than 10% on average indicated a robust performance for each emulsion PCR (emPCR) reaction in this study. Concerning mutation analysis, 203 of 242 samples (84%) were evaluated correctly. Hereby, 9 laboratories showed an excellent performance with correct identification of 21 (2 laboratories), 20 (2 laboratories) and 19 (5 laboratories) of the 22 control samples including novel artificial variants (e.g. E282K) and five low-level mutations with less than 20% mutant alleles. The concordance of mutation quantification was high in all laboratories. All non-mutated control samples were reported correctly. The 1% T315I mutation dilution was not identified in any laboratory, a 5% M244V mutation was not detected in 7, and a 1% F311L mutation was not detected in 5 laboratories. Two laboratories failed to test 9 of the 22 control samples correctly, thereby reporting both false positive and false negative results.

Summary
This multicenter analysis demonstrated that amplicon-based deep sequencing is technically feasible, achieves a high concordance across multiple laboratories and allows a broad and in-depth characterization of BCR-ABL mutations in CML. However, data also illustrates that expertise in NGS performance and characterization of BCR-ABL mutations is advantageous and necessarily required for the report of accurate results.

Keyword(s): BCR-ABL, Chronic myeloid leukemia, Mutation analysis, Standardization

www.eha.org