Challenging Conventional Diagnostic Methods By Comprehensive Molecular Diagnostics: Results of a Nationwide Prospective Comparison in Children with Acute Lymphoblastic Leukemia

Background and aims Treatment stratification in pediatric acute lymphoblastic leukemia (ALL) is, besides clinical parameters, guided by (cyto)genetic aberrations. These aberrations include a variety of gene fusions, aneuploidy groups, and copy number alterations (CNA), requiring multiple diagnostic assays to yield conclusive data. Over the past decade, the number and complexity of aberrations to be addressed for stratification increased whereas the turn-around-time decreased. We aimed to optimize the diagnostic workflow with a limited number of assays while allowing detection of all relevant genetic aberrations in a clinically relevant timeframe. Methods In a consecutive cohort of 467 newly diagnosed patients (0 to <19 years) immunophenotyped as ALL, we compared the diagnostic yield of RNA sequencing, 7 fluorescence in situ hybridization (FISH) probes, 7 RT-PCR targets, karyotyping, SNP array (CytoSNP-850K), and multiplex ligation-dependent probe amplification (MLPA; P335-ALL-IKZF1) for conclusiveness, concordance and turn-around-time. Results To detect stratifying fusions (ETV6::RUNX1, BCR::ABL1, ABL-class, KMT2Ar, TCF3::HLF, IGH::MYC), RNA sequencing and FISH were conclusive for 97% and 96% of patients, respectively, with 99% (433/437) concordance of test results. RT-PCR for 6 fusion transcripts was conclusive for >99%, but false-negative for 6 patients who had alternative exons in their fusion genes. RNA sequencing detected fusion genes even in samples with a leukemic cell percentage as low as 10% or RNA integrity score as low as 2. A TCF3::HLF fusion was not detected by RNA sequencing due to the insertion of intronic sequences in the chimeric transcript. A KMT2A::USP2 and a BCR::ABL1 fusion were not detected by FISH because of an inversion and probably insertion, respectively, not causing a break-apart FISH pattern. RNA sequencing detected subtype-defining fusions not (yet) used for stratification in 14% of BCP-ALL and 33% of T-ALL as well as rare fusions in 2% of BCP-ALL and 8% of T-ALL, including a targetable ETV6::NTRK3 fusion which were by definition not detected by targeted FISH or RT-PCR. Moreover, RNA sequencing showed added value for classification based on expression profile and detection of expressed mutations. The turn-around-time for RNA sequencing (7-15 days) and FISH (6-13 days) were both compatible with the demand for diagnostic reporting by day 15 of high-risk genetics and ABL-class fusions for the ALLTogether01 protocol and of BCR::ABL1 positivity for referral to the EsPhALL protocol. Because KMT2A rearrangement should be detected within 7 days to enter the Interfant-21 protocol, the shorter turn-around-time of FISH and RT-PCR were needed in infants (<1 year). For the detection of aneuploidy groups (high hyperdiploidy, low hypodiploidy and near haploidy) and intrachromosomal amplification of chromosome 21 (iAMP21), SNP array gave conclusive results for 99% of the patients, thereby outperforming karyotyping, which was conclusive for 64% and mistook 2 cases of masked hypodiploidy for high hyperdiploidy. Based on the amplification pattern on chromosome 21, 13 iAMP21 cases were identified by SNP array. Using a combination of karyotyping and RUNX1 metaphase FISH, 9/13 iAMP21 cases were identified with the remaining 4 cases lacking informative karyotypes. To identify deletions in eight genes/regions relevant for CNA risk stratification (IKZF1, CDKN2A/B, PAX5, EBF1, ETV6, RB1, BTG1 and PAR1), SNP array was conclusive in 99% and MLPA in 95% of patients with 98% concordant CNA risk calls. SNP array was more sensitive than MLPA in aneuploid samples and samples with low leukemic cell percentage. Beyond the currently required detection of MLPA-based deletions, SNP array detected deletions in (single) exons and genes not covered by the MLPA assay as well as aberrations in low mosaicism. Conclusions Our assay conclusiveness for ≥97% of patients and concordance of results with classic methods of 99% in 467 consecutive patients has resulted in the implementation of RNA sequencing and SNP array as the primary choice in the molecular diagnostics of newly diagnosed ALL in the Netherlands with addition of FISH and RT-PCR to detect KMT2A rearrangement in the infant population only. Performing RNA sequencing and SNP array for all patients has the advantage of detecting new lesions and expressed mutations to retrospectively study their role in prognosis and pathobiology.