Increasing the Sensitivity of BCR-ABL Monitoring By Digital PCR By Reducing the Background Noise

Background

Droplet digital PCR (dPCR) is a potentially appealing technology for quantitative detection of specific mutations with simultaneous measurement of the reference gene and without the need for standard curves. This makes dPCR particularly suitable for minimal residual disease (MRD) diagnostics in chronic myeloid leukaemia. However, a limitation of the dPCR assays compared to standard quantitative PCR (qPCR) is the background (termed lower limit of blank, LoB) of 1 or 2 positive droplets for BCR-ABL (Franke et al, ASH 2015, Cross et al, Leukemia 2016). The resulting false positive rate (FPR) limits the sensitivity and the ability of the test to detect deep molecular remissions.

Aims

The aim was to establish BCR-ABL (b2/a2 and/or b3/a2 transcript) dPCR assays with a sensitivity at least equivalent to that of qPCR, in which <5% of negative controls generate a single positive drop (FPR<5%, LoB=0).

Methods

One assay (Assay 1) was developed in Leipzig and another (Assay 2) by Bio-Rad laboratories. Sample preparation was identical for both assays up to the stage of reverse transcription (RT). Assay 1 used routine local RT procedure followed by a duplex BCR-ABL / ABL dPCR reaction modified from the M-BCR-ABL PCR published by Jennings et al. 2014. We tested ABL primer combinations from exons 4, 8, 10, and 11 generating a range of product lengths and optimised quenchers and PCR conditions to maximise the separation between positive and negative droplets. Assay 2 consisted of a reverse transcriptase reaction followed by a duplex PCR specific for ABL and both the b2/a2 and b3/a2 transcripts of BCR-ABL. Digital droplet PCR was performed in both cases using the Bio-Rad QX200 system.

Background noise and cross hybridisation were assessed in non-template controls (NTC), BCR-ABL negative cell lines and healthy wild-type donor samples. In addition, assay 2 was tested on 40 CML patient samples with MRD levels ranging from BCR-ABL negative to 1% BCR-ABL/ABL as defined by standard qPCR.

Results

Assay 1 yielded a FPR of 4% (4/94) in NTC with 1 positive droplet for BCR-ABL and a FPR of 3% for ABL (n = 3/94), resulting in a specificity of >95% and a LOB of 0. The BCR-ABL FPR using wild type target (HEK 293T cell line and healthy donors) was 5% (5/92 and 2/37, respectively).

Assay2: The specificity was >95% for both BCR-ABL and ABL in NTC and wild type samples. Extensive NTC analysis yielded no false positives for BCR-ABL PCR (n = 0/176; LoB = 0) and 1% false positives in ABL PCR (n = 2/176; 1-2 positive droplets, LoB = 0). The BCR-ABL negative cell line was confirmed negative for BCR-ABL PCR (n = 44), while the BCR-ABL assays of healthy donors were positive in 2% (n = 5/234) with 1 positive droplet per positive sample. The BCR-ABL copy numbers of the 40 CML samples were lower by dPCR than by RT-PCR (median 3/6) with similar ABL transcripts (median 55500/54949). This resulted in lower ratios by dPCR (mean 0.0313 vs. 0.0461 by qPCR). However, after conversion of the qPCR results with our lab specific conversion factor (CF) <1, the dPCR ratios were higher (mean 0.0313 vs. 0.0134). The MR classes were concordant in 60% of the samples, but 30% had more MRD while 10% had less MRD by dPCR than by qPCR, confirming our previous results using the EAC protocol (Franke, ASH 2015). More samples were MRD positive by dPCR than by qPCR (58% vs 50%).

Conclusions

An assay with a FPR of <5% and a LoB = 0 for the detection of BCR-ABL was developed. The BIO-RAD QX 200-Dx BCR-ABL Kit is fully functional and has a high specificity for BCR-ABL and ABL with a LoB of 0. The high specificity and low FPR/LoB of both assays enables sensitive and accurate monitoring of deep molecular remissions in CML.