Diagnostic performance of the molecular BCR-ABL1 monitoring system may impact on inclusion of CML patients in stopping trials.

In chronic myeloid leukemia (CML), the duration of deep molecular response (MR) before treatment cessation (MR4 or deeper, corresponding to BCR-ABL1 ≤ 0.01% on the International Scale (IS)) is considered as a prognostic factor for treatment free remission in stopping trials. MR level determination is dependent on the sensitivity of the monitoring technique. Here, we compared a newly established TaqMan (TM) and our so far routinely used LightCycler (LC) quantitative reverse transcription (qRT)-PCR systems for their ability to achieve the best possible sensitivity in BCR-ABL1 monitoring. We have comparatively analyzed RNA samples from peripheral blood mononuclear cells of 92 randomly chosen patients with CML resembling major molecular remission (MMR) or better and of 128 CML patients after treatment cessation (EURO-SKI stopping trial). While our LC system utilized ABL1, the TM system is based on GUSB as reference gene. We observed 99% concordance with respect to achievement of MMR. However, we found that 34 of the 92 patients monitored by TM/GUSB were re-classified to the next inferior MR log level, especially when LC/ABL1-based results were borderline to thresholds. Thirteen patients BCR-ABL1 negative in LC/ABL1 became positive after TM/GUSB analysis. In the 128 patients included in the EURO-SKI trial identical molecular findings were achieved for 114 patients. However, 14 patients were re-classified to the next inferior log-level by the TM/GUSB combination. Eight of these patients relapsed after treatment cessation; two of them were re-classified from MR4 to MMR and therefore did not meet inclusion criteria anymore. In conclusion, we consider both methods as comparable and interchangeable in terms of achievement of MMR and of longitudinal evaluation of clinical courses. However, in LC/ABL1 negative samples, slightly enhanced TM/GUSB sensitivity may lead to inferior classification of clinical samples in the context of TFR