Multiplexed automated digital quantification of fusion transcripts: comparative study with fluorescent in-situ hybridization (FISH) technique in acute leukemia patients

In acute myeloid leukemia (AML) and acute lymphoblastic leukemia (ALL), chromosomal abnormalities play a critical role in pathogenesis and prognosis [1]. Molecular studies of these genomic anomalies have identified specific genes implicated in the process of leukemogenesis [2]. Specific recurrent chromosomal translocations and their fusion transcripts define the current WHO classification system for acute leukemia (AL) [3]. These genetic aberrations in combination with morphology, immunophenotype and clinical features are used to diagnose distinct types of AL and thereby dictate therapeutic approaches [4]. Clinical laboratories, worldwide, utilize standard FISH technology with or without conventional karyotype to investigate these chromosomal translocations [5]. Unfortunately, FISH is a labour intensive and expensive procedure with the limited potential for “multiplexing”. These limitations impact diagnosis and clinical management. In a small number of AL sub-types, fusion transcripts can be amplified through reverse transcription, polymerase chain reaction (RT-PCR) methodology [6]. However, clinical application of RT-PCR is narrow due to variation in breakpoints resulting in several splice regions [7]. Given these limitations and the expanding new discoveries of additional recurrent chromosomal translocations, there is a need to explore newer automated technologies to “accommodate” new fusion transcript discoveries and make this significant clinical diagnostic testing cost-effective and time efficient.

Digital RNA quantification by NanoString Technologies is an automated, multiplexed platform, which can accurately measure up to 800 specific targets. “Leukemia Fusion Gene Expression Assay” by NanoString has a unique design of junction probes, which target the unique sequence spanning the fusion junction of the two exons. It has been claimed that a high degree of sensitivity and specificity has been achieved through sequence modification. However, no clinical validation data are available to determine the utility of this automated technology in routine clinical practice. The objective of this study was to evaluate the potential application and clinical utility of “Leukemia Fusion Gene Expression Assay” in the diagnostic sub-classification of acute leukemia patients. To this effect, we investigated fusion transcripts in a large cohort of AML and ALL patients utilizing genomic RNA from diagnostic bone marrow biopsies. We then correlated these results with data obtained by routine FISH and conventional karyotype. Our results provide unique preliminary data comparing a traditional platform with a novel automated technology, which provides potential promise for future adoption in the diagnosis and management of acute leukemia patients.

University of Calgary ethics committee approved this study (REB-13-0380-MODI; dated August 21, 2013). The diagnosis was confirmed according to the WHO 2008 classification [3]. We employed adequate representation of samples with recurrent chromosomal translocation (true positive) in AML (n = 23) and ALL (n = 26) while AML with normal cytogenetics (n = 19), or complex cytogenetics (n = 21) and normal bone marrow (n = 5) samples were used as negative controls (true negative).

A total of 94 (63 AML, 26 ALL and five control bone marrow) samples were analysed. AML patients included 42 men and 21 women (M: F 2:1) with a median age of 63 years (range 5–87 years). Twenty-six ALL patients comprised of 17 men and 09 women (M: F 1.8:1) with a median age of 12 years (range 2–63 years) in this cohort. Table 1 outlines the sensitivity, specificities, and positive and negative predictive values for each transcript. Figure 1 outlines, principal component analysis (PCA) plot based on digital quantification of various fusion transcripts and their correlation with chromosomal/FISH results. In AML category, digital quantification for PML/RARA and RUNX1-RUNX1T1 correlated 100 % with t(15;17) and t(8;21) respectively; while non-specific positivity was not seen in any other samples across entire cohort. CBFB-MYH11 fusion probes showed “false positivity” in AML samples with complex and normal karyotype. In 4/8 (50 %) of AML samples with inv(16) or t(16;16); false negative reaction was noted. In ALL group, BCR-ABL1 fusion probes performed seamlessly with 100 % sensitivity and specificity. TCF3-PBX1 and MLL-AF4 fusion probes showed 100 % specificity but sensitivity ranged up to 86 %. Probes for ETV6/RUNX1 fusion transcript performed poorly and showed only 50 % sensitivity but 98 % specificity.

In WHO classification for AL, chromosomal translocations are central to diagnosis, sub classification and prognosis. Conventional cytogenetics, followed by FISH in specific cases, is the gold standard to detect these chromosomal aberrations, but poses the challenge regarding subjectivity, manual labour, cost and turnaround time. There is a need to explore automated, multiplexed, throughput and cost-efficient technologies to serve rapidly expanding discoveries of novel fusion transcripts in haematological malignancies. A large cohort of AL samples were analysed by multiplexed, automated digital quantification in parallel with conventional karyotype/FISH technology. We observed a high degree of correlation (>90 % sensitivity and specificity) for different translocations that are frequent in clinical practice. The complete correlation was noted for t(8;21) between conventional karyotype/FISH and digital fusion RNA quantification results. This high degree of correlation can be related to the generation of only two fusion transcripts by this translocation [8]. t(8;21) can also involve several chromosomes, and these variant rearrangements can be cryptic and easily overlooked by conventional G-banding technique [9]. However, it was interesting to note that none of these cryptic fusion transcripts were detected in a large cohort of AL patients with normal and complex cytogenetics. Similarly, almost complete correlation was also noted for t(15;17); t(9;22) and t(4;11) translocations. Hence, we established the potential for clinical utility of this automated throughput platform of digital quantification as a possible alternate to current and widely used FISH technique.

Multiplexed molecular techniques present opportunities in procedural efficiencies, but also pose challenges regarding cross-reactivity and false positivity [10, 11]. Our extensive series of AL sample sets with normal and complex cytogenetic provided an adequate safety net to detect any cross-reactivity of fusion probes. We noted minimal to no false positive reactions in the majority of fusion probes. These results are encouraging in our small patient sample; however, a more robust validation is required using a larger cohort. In AL patients with t(12;21) (p13;q22), we noted discordance between FISH results and ETV6-RUNX1 fusion probe. The breakpoints in t(12;21) (p13;q22) are variable within the several thousand base pairs of ETV6 as well as RUNX1 introns. In spite of micro-clustering of breakpoints in ETV6 and RUNX1 introns, it is still difficult to identify or to sequence the breakpoints by even streamlined long‐distance PCR in t(12;21) [12]. Hence, it appears that fusion transcript ETV6-RUNX1 is beyond the scope of this assay at this stage. We also noted that inv(16) showed the most variable pattern and weak correlation with FISH results. This translocation results in more than ten differently sized CBFB-MYH11 fusion transcripts, which may explain variable sensitivity [13]. The probes against CBFB-MYH11 showed cross-reactivity with few samples harbouring normal or complex karyotype. Since there is marked heterogeneity of breakpoints in MYH11 gene, false negative rates could be high [14]. False positivity in inv 16 cases has also been reported by RT-PCR [15]. These observations mandate re-designing of some probes for this assay for enhanced sensitivity.

Our study demonstrates initial evaluation for fusion transcript detection by a throughput automated multiplexed platform for AL patients as an alternate solution to current complex, manual and labour intensive karyotyping /FISH techniques. Our preliminary study requires validations by future studies to evaluate these emerging technologies for their potential adoption in clinical laboratories. These expression assays offer significant capacity for expansion to accommodate additional future fusion transcripts discoveries of diagnostic or prognostic value in AL patients. It also has an immense ability to include gene expression signature as a diagnostic or prognostic tool for AL patients.