ALBERT

Description: BACKGROUND: Cytogenetic assessment to measure CML response to Glivec or Tasigna is expressed as a ratio of Ph+ CML to total cells. Molecular response determination for CML by RQ-PCR achieves a similar goal by obtaining a ratio of transcript levels of BCR-ABL expressed exclusively in Ph+ CML cells to the BCR or ABL control gene which is expressed in both Ph+ and normal cells. However when a RQ-PCR assay result is obtained, a different RQ-PCR ratio will be obtained for the same sample using BCR vs. ABL as control gene due to the intrinsic expression level differences in the two control genes. Experiments have demonstrated empirically that international standardization of RQ-PCR data obtained by assays using these different control genes is possible, thus allowing direct comparison of data. We have developed a mathematical simulation that explains the underlying basis for international standardization and this was experimentally confirmed to show how standardization is feasible in practice as a part of the Tyrosine Kinase Optimization and Selectivity (TOPS) trial. METHODS: We have mathematically simulated a correlation between theoretical cytogenetic response (Ph+ to total cell ratio) with the theoretical molecular response (BCR-ABL to control gene transcript ratio) by using the following assumptions; 1) BCR and BCR-ABL expression occur from similar promoters and likely have similar expression levels. 2) Ph+ cells have one normal BCR resulting in 1x level of BCR transcription and normal cells have two normal copies of BCR resulting in 2x level of BCR transcription (x=BCR transcription rate from one promoter). This simulation was extended to the ABL control gene as data of the relative expression level of ABL vs. BCR was available from samples in the TOPS trial, and the ABL transcript level is similar in both Ph+ and normal cells as the ABL PCR primers amplify both ABL and BCR-ABL transcripts. To experimentally test the simulation, RQ-PCR samples from patients in TOPS were exchanged between the study’s regional RQ-PCR laboratories. Each sample was assayed at 2 labs, one using BCR and the other ABL as the control gene. RESULTS: Mathematical simulation suggests that RQ-PCR ratios with BCR but not ABL as control gene has a nonlinear relationship with the Ph+ to the total cell ratio across the entire range from 0 to 100%. However at clinically relevant RQ-PCR ratio of ≤10%, irrespective of whether BCR or ABL is used as a control gene, RQ-PCR ratios are linear but with different slopes due to higher ratios obtained with ABL as control gene. Despite the different slopes, RQ-PCR ratios obtained with either control gene can be easily inter-converted via a multiplicative factor to neutralize the bias between methods as is currently recommended in international standardization. Experimental testing in the TOPS trial showed that; 1) ABL transcript copy number is significantly lower than BCR transcript copy number leading to higher RQ-PCR ratios with ABL as control gene. 2) Mathematical correction of bias between BCR and ABL as control gene allows for direct comparison of RQ-PCR ratios with an 88.2% concordance when ratios are below 10%. CONCLUSIONS: Mathematical simulation shows that for BCR and ABL as the control gene, the basis for international standardization is a consistent linear relationship that can be easily inter-converted when PCR ratios are ≤ 10%. This can be experimentally demonstrated by the high level of concordance obtained when the same sample is analyzed using different control genes. For RQ-PCR ratios 〉 10%, standardization is not possible using the currently applied neutralization of bias method of conversion, but may be possible by using a formula that accounts for the non-linear relationship of BCR as a control gene. Simulation studies also highlight the importance of not only obtaining precise but also accurate RQ-PCR ratios thus enabling consistent standardization.