“Detection of BCR-ABL1-like Subtype in Adult Acute Lymphoblastic Leukemia Using Digital Ncounter Nanostring Technology”

Introduction

A new provisional entity of “B-lymphoblastic leukaemia/lymphoma, BCR-ABL1-like” has been introduced in the 2017 revised edition of WHO classification of tumours of haematopoietic and lymphoid tissues. BCR-ABL1-like cases are negative for Ph chromosome or t(9:22)(q34;q11.2) translocation, do not express the fusion BCR-ABL1 RNA transcripts and proteins resulting from the Ph chromosome,and are characterized by similar gene expression profiles as that of BCR-ABL1-positive acute lymphoblastic leukemia (BCR-ABL1-positive ALL).There is no ‘short-cut approach’ for making an accurate diagnosis of BCR-ABL1-like ALL. Two approaches namely Gene expression profiling (GEP) or Next-generation sequencing (NGS) and TLDA (TaqMan low-density array) are used for the detection of BCR-ABL1-like ALL cases. NGS is very costly and data interpretation requires a lot of bioinformatics skills and TLDA is not commercially available in India.

Aims

We planned to study the whole transcriptome of BCR-ABL1-positive ALL cases using microarray GEP, followed by customizing targeted gene panel using nCounter NanoString technology, for the detection of BCR-ABL1-like cases.

METHODS

Flow cytometric immunophenotying (FCM-IP) and multiplex RT-PCR were performed on 200 B-ALL cases to detect BCR-ABL1 chimeric fusion transcripts. Further, 12 BCR-ABL1-positive cases were subjected to transcriptome profiling using Affymetrix microarray (Gene Chip Human Genome U133 Plus 2.0 Array). The results were analyzed using TAC 4.0 software. Finally, a targeted panel of 50 differentially expressed genes [including 5 Housekeeping genes (HKGs)] was constructed according to our microarray findings and previously published data (Harvey RC et al.ASH 2013). A total of 96 B-ALL cases (16 BCR-ABL1-positive cases & 80 BCR-ABL1-negative cases) were subjected to GEP using nCounter Platform. The results were analyzed using nSolver4.0 software.

RESULTS

In the study cohort of 200 adult B-ALL cases, BCR-ABL1 chimeric fusion transcripts were detected in 34% (b2a2 and b3a2=18.05% & e1a2=15.5%), as revealed by multiplex assay. Global transcriptome profiling of 12 BCR-ABL1 RNA transcripts revealed a total of 1574 as DE genes (460 genes in e1a2, 515 genes in b2a2 and 599 genes in b3a2). DE genes were further filtered through hierarchical clustering analysis and a total of 45 DE genes with 10- to -86-fold change were identified. These genes were further analyzed using nCounter NanoString. To further identify the best classifier genes, log2 normalized expression values were analyzed using penalized logistic regression. Based on previous literature and regression coefficient values, 15 genes were finally selected whose performance was individually analyzed using receiver operating characteristic curve (ROC) and area under the curve (AUC). Optimal thresholds for these genes were estimated as the values with maximum sensitivity and specificity. Out of 78 examined BCR-ABL1-negative cases, 33(42.30%) BCR-ABL1-negative cases were clustered together with 15 BCR-ABL1-positive cases and were attributed as BCR-ABL1-like ALL cases in principal component analysis. Further, we categorized CRLF2 in two categories; high CRLF2 cases 25/33 (75.75%) & low 8/33 (24.24%) in BCR-ABL1-like ALL cases. JAK2p.R683G mutation was screened in CRLF2 high cases and showed positivity in 19/24 (79.16%) by the Amplification Refractory Mutation (ARMS) PCR. In 25 cases, the average log fold change of -0.80 &-5.83 was seen in P2YR8 & CSF2RA respectively by qPCR. In CRLF2 low expressing cases, the average log fold change of 11 kinase genes showed -0.75 in CENPC, -0.66 FOXP1, -0.16 NUP153, 1.04 RCSD1, 1.50 PAX5, 1.12 FLT3, -5.65 EPOR, -4.03 ILR2B, -3.46 PDGRFB, -7.49 NTRK3 &-2.83 ZNF274 respectively. The average log fold change of IKZF1 in 80 BCR-ABL1-negative cases was found to be 1.07.

DISCUSSION & CONCLUSION

We have devised a method that includes 15 genes according to AUC/ROC for the detection of BCR-ABL1-like ALL cases, using nCounter NanoString technology for the first time in Indian patients. Furthermore, we are planning to validate this model in future, on 50 BCR-ABL1-positive and 150 BCR-ABL1-negative cases and devise a simple, efficient, cost-effective qPCR method. It is very important to detect BCR-ABL1-like ALL cases to start the desirable TKI therapy & aid in treatment stratification, prognostication, and improve the overall survival of these patients.