Data show 40% of chronic myeloid leukemia (CML) patients maintain complete molecular remission (CMR), i.e. failure to detect BCR-ABL1, consenting to termination of Imatinib mesylate (IM) therapy, following undetectable disease for ≥2 years by quantitative PCR (Q-PCR). These findings suggest majority of the patients experience molecular relapse. Furthermore, majority relapse in the first 6 months, implying Q-PCR assay sensitivity is suboptimum, to confidently identify patients for discontinuation of IM. Droplet digital PCR (ddPCR) is suggested to have sensitivity that is one log greater than the Taqman (Q-PCR) assay. If verified, ddPCR would enhance safe withdrawal of IM therapy from CML patients. Here we present data comparing ddPCR with Q-PCR.

In total we assayed 161 samples, of these 6 were serial dilutions of the International reference (IR) BCR-ABL1 plasmid, the remainder were cDNA samples. The 161 samples comprised of 4 sample groups; I: CML chronic phase samples (n=118); II: CML samples post stem cell transplant (SCT) (n=22); III: normal control (NC) samples (n=16); IV: Serially diluted BCR-ABL1 1.04x10e2 copies/µl, IR ERM-AD623e (n=6). Group I comprised of 121 samples from 21 CML patients in chronic phase treated with IM. Group II comprised of 21 samples from 19 CML patients (2 samples each for 2 of these 19 patients) who had undergone SCT. Group II comprised of samples from normal adult blood donor volunteers. Finally, Group IV included 6 serially diluted samples, ranging 100 to 0.001 copies of the IR (1.04x10e2 copies/µl) plasmid (Sigma, Munich, Germany). All the Group I, III and IV samples were subjected to Q-PCR Taqman assay and ddPCR (Biorad, California, USA). The Group III samples were in addition subjected to nested PCR. Only those samples with cycle threshold (Ct) <37 in 2 or more replicates by Q-PCR were recorded positive. For ddPCR those samples with sum of ≥3 positive droplets with a minimum 5500 droplets per well were reported positive. All the PCR reactions were performed in triplicate in final volume of 20µl, which included 5µl of cDNA or reference plasmid.

Among Group I, Q-PCR detected BCR-ABL1 in 57 of the 118 patient samples assayed, with a median of 16.72 transcripts (range 1.38-47450). In only 2 (2.25 and 4.96 transcripts) of these 57 samples, ddPCR failed to detect BCR-ABL1 transcripts. Q-PCR and ddPCR failed to detect BCR-ABL1 in 45 (38.1%) samples. For 14 (11.8%) of the samples ddPCR was positive, median 3.4 copies (range 3-15) but negative by Q-PCR. Among Group II, one of the 21 samples was excluded from the analysis because <5500 droplets were generated. Of the 21 samples 7 were negative by ddPCR. Nested PCR was negative for all 7 samples. Three samples were positive by all 3 technologies, nested PCR, Q-PCR and ddPCR. Remarkably, 5 of the 21 samples were negative by nested, but positive by ddPCR; median 18.0 copies (6.2-22.0). These 5 samples were not subjected to Q-PCR. In Group III all 16 NC samples were negative by Q-PCR and ddPCR. In Group IV, ddPCR did detect BCR-ABL1 in the serially diluted IR sample calculated to have 1 copy (26 positive droplets of the 106095 total droplets), but Q-PCR failed. However, the lower dilutions, calculated to contain 0.1, 0.01 and 0.001 copies were negative by ddPCR and Q-PCR.

We assayed 161 samples by ddPCR and Q-PCR, of these 139 were from CML patients. In addition 21 of the samples were also subjected to nested PCR. Our data support the notion ddPCR is at least one log more sensitive than Q-PCR. Of the 140 patient samples assayed, 19 (13.5%) were positive by ddPCR but negative by Q-PCR. Only 2 of the 118 samples in Group I were negative by ddPCR but positive by Q-PCR. There was in insufficient sample to repeat these 2 assays. The increased sensitivity of ddPCR as implied by the clinical samples was supported by assays performed using the IR. The serial dilution equivalent to 1.0 BCR-ABL1 copy was reliably detected by ddPCR, but was negative by Q-PCR. In summary, these data suggest ddPCR is more sensitive. However, the clinical significance of this must be assessed in context of long-term clinical outcome of patients with detectable BCR-ABL1 by ddPCR and negative by Q-PCR. But, clearly increased sensitivity is likely to enhance safe withdrawal of IM therapy for CML patients in CMR. Furthermore, regular monitoring of these patients by ddPCR would enable early detection of molecular relapse and thereby minimize the risk of disease progression.


Le Coutre:Novartis: Honoraria.

Author notes


Asterisk with author names denotes non-ASH members.