A Comparison of Droplet Digital PCR and Quantitative RT-PCR for Low Level BCR-ABL in CML Patients with Molecular Responses

Droplet Digital PCR (ddPCR) is based on the separation of a standard PCR reaction into many thousand single droplets, each of which is tested independently for the presence of the target. This has the potential to generate an absolute read out that is largely tolerant of variations in PCR efficiency, reducing the requirement for internal standardisation. The ddPCR method should therefore be suitable for monitoring even very low BCR-ABL levels in CML patients with a high degree of reliability and sensitivity. We compared standardized quantitative (Q) RT-PCR with ddPCR results in patients with BCR-ABL e13/e14/a2 positive chronic myeloid leukemia and deep molecular remission (MR).

439 cDNA samples from CML patients with e13 or e14/a2 BCR-ABL fusion genes were received for routine monitoring in our laboratory, which is accredited for the detection of molecular response (MR) down to 4.5 by the European Treatment Outcome Study (EUTOS) collaboration. QRT-PCR was performed using the WHO certified ERM-AD623 BCR-ABL standard and showed molecular response (MR) 3, 4, or at least 4.5 in 78, 129 and 232 samples, respectively. For comparison, the same cDNA samples were analysed by ddPCR using a QX200 Droplet Digital PCR System (BIO-RAD). In line with the manufacturer's recommendations, samples yielding a minimum of 3 positive droplets from the 10-15.000 routinely analysed were scored as positive, although the effects of reducing this cut-off value were also investigated.

The median total ABL copies in 5µl cDNA were in close agreement between the 2 methods with 37769 (range 5032 to 470057) by QRT-PCR and 37810 (range 4200 to 110100) by ddPCR. The congruence at the level of single samples was relatively low, as may be expected at limiting target levels, with 104 samples scoring positive by both methods, 65 by QRT-PCR only and 84 by ddPCR only (table 1). Where both methods scored positive, the calculated BCR-ABL/ABL ratios were comparable with a median of 0.011 for QRT-PCR (range 0.0001-0.1) and 0.013 for ddPCR (range 0.0037-0.12).

Further analysis of the data following pre-categorization of the samples into MR3, MR4 or MR4.5 based on the standardised QRT-PCR procedure revealed a higher number of samples scoring positive by ddPCR but not QRT-PCR at the lower BCR-ABL/ABL ratios, with 75 samples registered positive by ddPCR only, compared to 8 by QRT-PCR only in the MR4.5 category. This suggests a possible advantage of ddPCR at very low target levels. Further reducing the cut-off value to 2 positive droplets results in out hands in a small (1%) incidence of positivity in the negative controls, but leads to a marked increase in the frequency of CML samples detected positive by ddPCR only.

Conclusions: 1) Although QRT-PCR and ddPCR yield similar overall results throughout all categories of molecular response, there is relatively low congruence at the level of single samples. It is likely that stochastic variation in pipetting the low numbers of target molecules involved contributes significantly to the variability between individual assays. 2) Using the recommended cut-off of 3 positive droplets, ddPCR scored more positive samples than did QRT-PCR at low BCR-ABL/ABL ratios. 3) Given the low numbers of target molecules present, the recommended use of a 3 droplet cut-off may limit the sensitivity of the ddPCR method. For these applications, it will therefore be important to define and reduce the source of background in the ddPCR procedure to a reproducible minimum.

Overall, our results suggest that ddPCR may have the potential for more sensitive quantitation at low transcript levels then the current standard QRT-PCR.

*MR groups defined on the basis of standard QRT-PCR results.