Sample Exchange to Standardize Molecular Monitoring for CML: The EUTOS Experience

International management recommendations for chronic myeloid leukemia (CML) patients undergoing tyrosine kinase inhibitor therapy emphasise the importance of frequent molecular monitoring to determine the degree of response. Achievement or non-achievement of time-dependent molecular milestones helps to identify individuals who may or may not benefit from a change in treatment. Sequential monitoring enables the timely detection of relapse and may help to highlight other issues influencing response such as compliance.

Molecular monitoring is performed by reverse transcriptase quantitative polymerase chain reaction (RT-qPCR) for BCR-ABL1 mRNA, and the use of a reference gene helps to control for variation in quality from sample to sample. Guidelines specify that results should be expressed on the International Scale (IS) for BCR-ABL1 measurement, which is essentially the same scale used in the International Randomized Study of Interferon and STI571 (IRIS) study. The 'gold standard' method for testing laboratories to generate results on the IS is to apply a laboratory-specific conversion factor (CF) to their local results, with the CF derived by patient sample exchange with an established reference laboratory (Branford et al., Blood 2008). We report here the experience of the European Treatment and Outcome Study (EUTOS) to standardise molecular monitoring in Europe by an expanded programme of sample exchange.

The laboratory in Mannheim initially established itself as a European reference laboratory following derivation and validation of their own CF by patient sample exchange with the reference laboratory in Adelaide. Samples exchanges between Mannheim and Adelaide were repeated on a further 5 occasions and confirmed stability of the Mannheim CF. The Mannheim laboratory then undertook further patient sample exchanges with testing laboratories in Europe. Patient samples, typically up to 30 per round, were lysed in Trizol or guanidinium isothiocyanate solution and were provided by both the reference laboratory and the test laboratory. Samples were selected to span the range of detectable disease from approximately 10% BCR-ABL^IS to MR⁴ and were transported on dry ice. Samples were extracted and analyzed in both centres, and results compared. Any consistent bias in the results between the reference laboratory IS values and the test laboratory values was identified by Bland-Altman analysis and a CF defined as the antilog of the mean bias. Multiple rounds of sample exchange were performed to confirm the validity of individual CFs and to help identify changes in test laboratory assay performance over time.

The study was performed over a 10 year period (2006-2016) with the participation of 58 laboratories (defined as EUTOS reference centres) from 22 countries. A total of 291 sample exchanges were performed (median = 5; range 1-8), comprising 8396 samples of which 6956 were evaluable (mean 24 successful samples/exchange). Of the 291 sample exchanges, 7 were considered unsatisfactory due to poor concordance (<50% of samples within 2 fold), resulting in the issuance of 284 CF certificates. Of these, 244 (86%) were for laboratories using ABL1 as a reference gene and 44 for those using GUSB. For ABL1 laboratories the median CF was 0.56 (range 0.097-2.87) and for those using GUSB the median was 0.96 (0.32-3.82). Despite the wide range of CF values, 81% and 73% were within 2-fold of the median for ABL1 and GUSB, respectively, and the interquartile ranges were 0.37-0.81 for ABL1 and 0.68-1.34 for GUSB. Using previously described criteria for concordance (Branford et al., Blood 2008; Cross et al. Leukemia 2016), we defined a stable CF as CF_n/CF_n+1 = 0.63-1.58, with an optimal ratio of 0.83-1.2. Using these criteria, 124 CFs were stable on sequential analysis and 72 were unstable.

Overall our study demonstrates the feasibility of large scale sample exchanges to promulgate laboratory-specific CFs to the IS, but the process is onerous, challenging and expensive. We anticipate that sample exchange will remain an option for laboratories to calibrate their assays but is likely to be largely superseded by approaches that use IS-calibrated secondary reference reagents.