Comparation of Digital PCR and Quantitative Pcr for Molecular Monitorization of Chronic Myeloid Leukemia Patients

Introduction: Chronic myeloid leukemia (CML) is a neoplasm produced by the formation of the BCRABL fusion gene. Tyrosine kinase inhibitors (TKI) have changed dramatically the clinical course of the disease. Quantitative PCR (qPCR) is the reference technique for measuring response in CML patients. Droplet digital PCR (ddPCR) is, in brief, a technology based on partitioning of the sample in a high number of small fractions which, in a further step, are read as positive or negative by its fluorescence. ddPCR performs an absolute quantification of the number of transcripts and their results do not depend on PCR efficiency. In spite of these theoretical advantages over qPCR, ddPCR is not routinely applied for BCRABL monitoring. In this study, we sought to compare qPCR and ddPCR techniques in similar amplification conditions.

Methods: We included samples from 43 CML patients with different levels of response determined by qPCR: 23 patients with BCRABL/ABL1 levels <15%, 7 patients with MR3.0, 10 patients with MR4.0 and 3 patients who showed MR4.5 (according to European LeukemiaNet guidelines). Additionally, 3 non-CML patients were included. High-Capacity cDNA Reverse Transcription Kit was employed, following manufacturer instructions, in a unique reaction for both subsequent technologies utilized. qPCR was carried out with primers and probes for BCRABL and ABL1 from Europe Against Cancer report with 5 microL of non-measured cDNA in duplicates. The standards were BCR-ABL P210 ELITe Standard (ELITech Group) and the conversion factor was obtained with the Philadelphia P210 RNA Reference (Elitech Group). qPCR assys were run in an ABI 7500 platform in standard quantitation mode. EUTOS recommendations for BCRABL quantification were followed.

ddPCR reaction was performed with the same primers and probes. Median cDNA concentration was 4,485ng/µL (1,771-7,787ng/µL). In order to avoid saturation of the technology, 2microL of cDNA at a concentration of 600ng/microL (i.e. 1,200ng) were added and run in duplicates in a C1000 touch Thermal Cycler (Bio-RAD) with the same amplification conditions used in qPCR. The same RNA reference material was used to obtain the calibration factor for the ddPCR technique. Bio-Rad QX200 Droplet Digital PCR System was employed for analysis. We used the mode of analysis for rare event detection and thresholds were set individually in each reaction. Qualitative comparations were carried out using Kappa coefficient with quadratic weight. For quantitative comparations, we applied Pearson correlation, intraclass correlation coefficient for absolute agreement and Passing-Bablok tests. A subgroup analysis, including samples with MR3.0 or deeper, was also performed.

Results: All the samples showed ≥10000 ABL copies per well by qPCR. More than 10000 events were analyzed in every ddPCR reaction well. Good replicates were observed in qPCR and ddPCR. Adequate separation of positive and negative events could be performed and no saturation was observed in the ddPCR reations. No signal from BCRABL was detected in non-CML patients in ddPCR. The Kappa coefficient between qPCR and ddPCR measurements was 0.76 (P< 0.001, 95% confidence interval (CI): 0.63-0.88). In the 20 samples with MR 3.0 or deeper, kappa=0.13 (P >0.05 ) .

For quantitative comparison, 32 samples with detectable BCRABL were included. Pearson test yielded an R²=0.91 (P< 0.001). The intraclass correlation coefficient for absolute agreement showed a value of 0.96 (P <0.001). Passing-Bablok test provided a non significant constant difference (0.041, CI 95%: -0.04-0.05) and a significant proportional difference (0.8, CI 95%: 0.7-0.96) (i.e. ddPCR results= 0.041 + (0.8*qPCR results). There were only 9 samples with a BCRABL/ABL1 ratio and quantitative test were not performed.

Conclusion: Although a global excelent qualitative correlation is observed, in patients with MR3.0 or deeper which are the more important to classify correctly, qualitative correlation is absent. The R² and intraclass correlation coefficient for absolute agreement showed a high concordance but Passing-Bablok test yielded a significant proportional difference among qPCR and ddPCR. We sought to test both technologies in similar conditions but different ddPCR strategies could be implemented in order to improve reproducibility of both techniques: higher amount of replicates or, given the lack of saturation, a higher cDNA concentration in the reactions.

Table 1

View largeDownload slide

Table 1

View largeDownload slide

Close modal