Abstract

During the last decade, several studies have shown that quantification of residual tumor cells significantly correlates with clinical outcome in chronic myelogenous leukemia (CML). Detection of minimal residual disease (MRD) is now becoming routinely implemented in protocols for guiding therapy and for evaluation of new treatment modalities. The lack of standardization of the methodology represents a major barrier in the comparison of data generated in different studies. Therapeutic response can be expressed in 3 ways: (i) Calculation of the ratio of mRNA transcripts of target to reference gene, e.g. ratio BCR-ABL/ABL, (ii) Individual calculation of the relative molecular response: i.e. comparison of the MRD level after therapy vs pretherapeutic level, and (iii) use of a lab-specific reference point, e.g. a pool of diagnostic samples for calculation of the log reduction (Δlog). In the IRIS trial, a Δlog=3 after 12 mo of imatinib therapy was accompanied by a 100% relapse free survival after 30 mo and defined as "major molecular response" (MMR, Hughes et al. NEJM, 2003). The reference sample, however, is not available for widespread distribution. We sought to establish a relationship between the Δlog approach and the ratio of BCR-ABL to total ABL transcripts. To compare scales, 134 samples (68 RNA, 66 cDNA) of CML pts on imatinib therapy were exchanged between labs in Mannheim and Adelaide. 8 samples represented neg controls and 8 were degraded during shipment. Thus, 118 samples were eligible for analysis. 46 samples were pos for b3a2, 41 for b2a2, and 31 for b3a2&b2a2 BCR-ABL transcripts. In Adelaide, TaqMan PCR was performed for BCR-ABL and BCR transcripts. BCR-ABL and BCR plasmid dilutions were employed as standards, the ratio BCR-ABL/BCR was compared to the ratio of a pooled sample of 30 CML pts at diagnosis (IRIS approach). In Mannheim, quantitative LightCycler PCR was performed for BCR-ABL and total ABL transcripts with a single BCR-ABL plasmid dilution as standard curve for either transcript. The ratio BCR-ABL/ABL was calculated for the individual sample. In 4 samples, BCR-ABL was not detectable in one of the labs, 114 were pos in both labs and are eligible for comparison. None of the neg controls was tested pos in either lab. Δlog and ratio BCR-ABL/ABL correlated with r=−0.87 (p<0.0001). Regression analysis revealed a relationship between Δlog and ratio BCR-ABL/ABL according to the formula: Δlog = −0.91*[log (BCR-ABL/ABL in %)]+2.15. To demonstrate the variability of the reference point 17 paired samples were investigated from CML patients at diagnosis and prior to the start of imatinib after a brief period (10–274 days, median 53) of hydroxyurea therapy. During this interval, median leukocyte counts changed from 138/nl (range 21–605) to 7.8/nl (3.2–41, p=0.086), blasts in peripheral blood from 1% (0–12) to 0% (0–1, p=0.0085), and ratio BCR-ABL/G6PD from 5.5% (2.0–24.6) to 4.3% (1.3–17.1, p=0.036) with a Δlog of 0.2 (−0.3 to 0.8). We conclude that the calculation of the ratio BCR-ABL/ABL is a rational approach to express MRD levels in CML patients after therapy. The ability to calculate the Δlog depends on the access to a reference sample which is not currently available worldwide. We demonstrate here the option to convert results calculated as ratios of BCR-ABL/ABL into Δlog from a pre-therapy median and vice-versa. A 3-log reduction, which has been established as MMR, represents a ratio BCR-ABL/ABL of 0.12%.

Author notes

Corresponding author