Harmonization of BCR-ABL mRNA Quantification Using an Uniform Control Plasmid in 36 International Laboratories.

Quantitative determination of residual BCR-ABL transcript levels has been accepted as integral part of the management of CML patients. However, heterogeneity of molecular approaches results in a lack of comparability between different studies. Thus, there is an unmet need for harmonization of both procedures and expression of results. In a series of consensus meetings within the European LeukemiaNet a list of prerequisites to achieve an optimal sensitivity and standardization has been elaborated:

• use of at least 10ml peripheral blood processed within 36 hrs;

• bedside RNA stabilization for multicenter trials;

• standardized PCR protocols optimized for each platform;

• use of a single plasmid containing target and housekeeping genes to avoid dilution errors;

• use of total ABL and/or beta glucuronidase (GUS) as internal controls.

To substantiate these theses an international multicenter trial within 36 labs in 14 countries was initiated. The aim of the study was to assess the variability of results obtained from different labs using the PAXgene Blood RNA System^® (PreAnalytiX, Hombrechtikon, Switzerland) for RNA extraction, individual protocols for cDNA synthesis, 3 different PCR platforms (TaqMan^®, TM, n=24, LightCycler^®, LC, n=14, Rotorgene^® n=1), and optimized quantitative RT-PCR conditions. In order to standardize results, b3a2 BCR-ABL and GUS sequences were cloned into a pCR 2.1-TOPO vector^® (Invitrogen, Carlsbad, CA), which was distributed to all participants in serial dilutions as external control for quantification of BCR-ABL, total ABL, and GUS mRNA transcripts. Ten samples containing dilutions (10, 2, 1, 0.1%) of b3a2 or b2a2 BCR-ABL positive cells in normal leukocytes and negative controls were prepared, blinded, and shipped to the participants. Transcript numbers were determined in triplicates, ratios BCR-ABL/ABL and BCR-ABL/GUS were calculated and expressed in %. Median ratios BCR-ABL/ABL for b3a2 samples were 8.9, 1.7, 0.85, and 0.11%; for b2a2 samples 9.1, 1.6, 0.82, and 0.10%. Median ratios BCR-ABL/GUS for b3a2 samples were 3.4, 0.77, 0.37, and 0.042%; for b2a2 samples 2.8, 0.48, 0.29, and 0.031%. Four of 36 participants (11%) detected low BCR-ABL copy numbers in negative control samples. The coefficients of variation (CV) for all participants, TM, and LC users were 0.62, 0.57, and 0.59 for ratios BCR-ABL/ABL; 1.03, 0.85, and 1.22 for ratios BCR-ABL/GUS, respectively. Standard errors to the regression line were significantly lower evaluating ratios BCR-ABL/GUS (median 0.075, range 0.0046–0.90) compared to ratios BCR-ABL/ABL (median 0.18, range 0.022–2.2, p<0.001). Overall, mean TM ratios were 1.7times higher than LC ratios indicating a difference of the amplification efficiencies between target and control genes. We conclude, that harmonization of BCR-ABL mRNA quantification is feasible employing a common plasmid for BCR-ABL, total ABL, and GUS. However, the remaining variability of results indicates minor differences of the PCR efficiencies using individual protocols. We therefore suggest (i) the use of a common standard plasmid, (ii) the introduction of a calibrator, and (iii) regular control rounds to achieve and maintain comparability of results between individual laboratories.