DEFINING LOW AND UNDETECTABLE LEVELS OF DISEASE IN CHRONIC MYELOID LEUKAEMIA: PROGRESS TOWARDS STANDARDISATION IN EUROPE

Background:

The international, collaborative effort to standardise molecular testing for chronic myeloid leukaemia (CML) to date has focused largely on detectable residual disease, with a particular emphasis on determining whether (or not) a patient has achieved major molecular response (MMR). However, with
new definitions of deep molecular responses (MR) a new problem arises in how to define assay sensitivity in a standardized manner when BCR-ABL1 mRNA is undetectable.

Aims:

We have previously established draft criteria to define deep molecular responses (MR4, MR4.5 etc) of residual disease in CML. To assess how well these definitions might work in practice and enable laboratories to benchmark their performance with regard to assay sensitivity in relation to others we sent
out a series of cell line and armoured RNA (aRNA) samples along with a common (IRMM) plasmid dilution to testing laboratories. We evaluated assay sensitivity in two ways: (a) by directly evaluating the ability of labs to detect low level positive samples and (b) measuring absolute control gene (CG) transcript numbers as a surrogate for sample quality and thus theoretical sensitivity with which BCR-ABL1 could be detected.

Methods:

Cell line and aRNA samples were sent to EUTOS (European Treatment and Outcome Study) reference laboratories who agreed to participate (n=39). Laboratories tested two different levels of e14a2 aRNA diluted in a background of ABL1 aRNA only (MMR, MR4) and three different cell line lysates (5%, 0.05%, 0.005% BCR-ABL1IS). Absolute copy numbers of BCR-ABL1 and CGand the BCR-ABL1/CG ratios before and after conversion to the international scale (IS) using a common plasmid standard curve and laboratory plasmid standard curve were recorded.

Results:

Cell line and aRNA analysis indicated good comparability of%BCRABL1 /CG results that was improved by use of the common plasmid. Analysis of the cell line lysates and aRNA samples indicated that all labs were able to detect BCR-ABL1 at the lowest dilution (which corresponded to MR4), however the
median number of CG transcripts in the cell line samples (ABL1=74,509 (n=33); GUSB=84,762 (n=6)) was higher than would be expected from patient samples. Analysis of absolute numbers of BCR-ABL1 and ABL1 transcripts using the aRNA samples revealed that most labs are underestimating the absolute number of
transcript copy numbers, even when using a common plasmid calibrator. All aRNA samples were predicted to contain 305,500 ABL1 copies per μl aRNA but when laboratory data were corrected for the amount of aRNA added to the PCR, the median number of ABL1 copies per μl aRNA reported was only 54,578 (n=31)
using the usual laboratory standard curve and 28,458 (n=32) using the IRMM standard curve. These data suggest that there is substantial scope for most labs to improve the quality of their cDNA synthesis procedures and thereby increase assay sensitivity.

Summary and Conclusions:

Good comparability was seen between the 39 participating labs for samples where BCR-ABL1 was detectable, but assay sensitivity based on CG numbers was highly variable. Although the use of a common plasmid calibrator improved the comparability, it did not correct for differences in estimates of absolute CGnumbers. The reasons for these differences are unclear but may relate to differences in the efficiency of reverse transcription. Further investigation will be required to determine if this is the case, and how to improve CG numbers/sensitivity where necessary.

Abstrakta v časopise Haematologica 2013, Suppl1

Online program