A Fast and Accurate Diagnostic Method for Ph-like ALL Using the Ncounter System

Ph-like acute lymphoblastic leukemia (also known as BCR-ABL1-like ALL) is a new disease entity of B-cell ALL (B-ALL) that exhibits a mRNA expression profile similar to that of Philadelphia chromosome-positive ALL (Ph⁺ ALL). Ph-like signature is presumably driven by kinase-activating gene alterations. Thus, both gene expression pattern and DNA mutational status should be assessed to make a definitive diagnosis for Ph-like ALL. A variety of approaches combining multiple methods, including RNA sequencing (RNA-seq), Taqman low-density array (LDA), fluorescence in situ hybridization (FISH) and targeted DNA sequencing, are being tested; however, such multi-omics approaches are available only in limited institutions. Since Ph-like ALL patients generally exhibit poor response to standard chemotherapy, and tyrosine kinase inhibitors (TKIs) may benefit them when used in a timely manner, a fast, accurate and generalizable diagnostic method is critically needed. In the present study, we have developed a nCounter-based diagnostic method for Ph-like ALL and validated it using a cohort of Japanese adult B-ALL cases.

To identify genes that are uniquely expressed (or not expressed) in Ph⁺ B-ALL, we first obtained publicly-available gene expression datasets comprising 1146 B-ALL cases and identified 82 differentially-expressed genes in Ph⁺ ALL cases. We then assessed expression levels of those genes in an independent cohort using the nCounter, which enables fast, sensitive and accurate RNA detection. We also tested whether nCounter-based methods can detect fusion transcripts relevant for Ph-like ALL pathogenesis using probes targeting ABL1, ABL2, CSF1R, PDGFRB, and JAK2.

We analyzed 123 samples (Ph⁺ = 42, Ph^- = 81, age 16 to 67) obtained from newly-diagnosed adult B-ALL patients enrolled in two clinical trials conducted by the Fukuoka Blood and Marrow Transplantation Group (FBMTG) (Nagafuji et al. Eur J Haematol 2019). Unsupervised hierarchical clustering successfully stratified 123 cases into two disease clusters: Ph⁺ and Ph^- subgroups. As expected, Ph⁺ subgroup included almost all Ph⁺ ALL cases (40 out of 42 cases), while 18 out of 81 Ph^- ALL cases (22%) were categorized into the Ph⁺ subgroup. We defined these cases as Ph-like ALL.

To validate the nCounter-based Ph-like ALL classification, we performed RNA-seq and target-capture DNA sequencing of all Ph^- ALL cases. As expected, we detected kinase-activating fusions/rearrangements, including CRLF2 rearrangements (7 cases), PDGFRB fusions (3 cases), JAK2 fusions (2 cases), EPOR rearrangements (2 cases), ABL1 fusion (1 case), and FLT3 internal tandem duplication (1 case) in 16 Ph-like ALL cases, while no genetic alterations were detected in 2 cases. Fusion genes involving PDGFRB were consistently detected by nCounter (3/3); however, detection of those involving JAK2 (1/2) and ABL1 (0/1) were inconsistent. JAK2 and/or RAS mutations were detected in 5 of 7 Ph-like ALL cases harboring CRLF2 rearrangements. Of note, CRLF2 protein expression was detected by FACS in all CRLF2-rearranged cases.

We next assessed significance of the Ph-like signature on clinical outcomes using a cohort of 40 Ph^- ALL cases, in which minimal/measurable residual disease (MRD) status, assessed by IgH and/or TCR rearrangements, as well as clinical data were available (Nagafuji et al. Eur J Haematol 2019). Ph-like ALL cases exclusively exhibited MRD positivity after induction therapy as compared to non-Ph-like cases (p=0.04), indicative of the chemo-resistant nature of Ph-like ALL as previously reported (Roberts et al. N Engl J Med, 2014 and Roberts et al. J Clin Oncol, 2017). As expected, Ph-like ALL cases exhibited significantly poor disease-free survival compared with non-Ph-like ALL cases (p=0.04); however, no significant difference was evident in overall survival (p=0.62) presumably due to the fact that all MRD-positive cases were subjected to allo-HSCT after induction therapy. These data indicate that MRD-based therapy stratification could overcome chemo-resistant nature of Ph-like ALL.

Our data suggest that nCounter-based diagnostic method is fast and accurate to identify Ph-like ALL. Since Ph-like signature generally dictates poor clinical outcomes, and upfront TKI therapy may improve them, our method could facilitate precision medicine in the treatment of Ph^- B-ALL.