Supplementary Methods: Quality control study design.
QC I: Test materials representing five consecutive peripheral blood (PB) samples from a CML patient (BCR-ABL+), five consecutive PB samples from an AML patient (CBFβ-MYH11+ and over-expressing the WT1 gene), as well as a K562 cell line sample were included in the study. To eliminate noise from local differences in sample handling, RNA purification, and reverse transcription procedures, the samples were collected and processed to cDNA by the Aarhus laboratory (Laboratory F). Next, the cDNA samples were shipped to Ipsogen (Marseille, France) for subsequent distribution to all eight laboratories, together with kits containing primers, probes, and reference plasmids for the three target genes and the ABL control gene (Ipsogen kit numbers: FQPP-10-CE , FQPP-02 and PQPP-02-CE).
Each laboratory performed qPCR using their usual mastermix reagents and qPCR hardware. The latter included ABI 7500 and ABI 7900 (Applied Biosystems Inc, Foster City, CA), Mx3000P (Stratagene, La Jolla, CA), and LightCycler 480 (Roche, Basel, Switzerland) (Supplementary Table 1). qPCR conditions were set according to the instructions by the kit manufacturer (Ipsogen). All participants used a fluorescence threshold of 0.1 before exporting qPCR values as text files. These files, encompassing information on the sample ID, target gene ID, Ct value for each individual PCR well, and the copy number for plasmid standards, were subsequently imported into the MRD reporting software.
To test the software the participants were asked to generate 16 specified reports covering central software functions, including different choices of MRD calculation method and graphic presentation modes, generation of patient reports, as well as tabular lists of MRD data. The options for MRD calculation method were tested in both AML (relative quantification by normalization to diagnostic sample in reports 1-5, absolute quantification based on plasmid standard in reports 6-10, respectively) and CML (relative quantification by normalization to diagnostic sample in reports 11-12, relative quantification by normalization to cell line calibrator sample in reports 13-14, and absolute quantification based on plasmid standard in reports 15-16, respectively) (Supplementary Table 1). A graphic display option with color-coded evaluation of samples (PCR positive, positive but not quantifiable or negative, respectively) was employed for AML samples, whereas graphs without color-coded evaluation were chosen for CML samples. Display of two target genes (CBFβ-MYH11 and WT1) in one graph was demonstrated in AML reports 5 and 10. All MRD results were displayed as reports with an MRD curve and corresponding lists showing the MRD results of each sample, respectively.
QC II: The purpose of the IS CF in CML MRD monitoring is to allow for comparison of MRD data irrespective of local differences in sample handling, qPCR method, and MRD expression. Thus, the CF of each laboratory will be influenced by the local choice of methodology.(9) In the present study laboratories C, D, F, G, and H had acquired a CF. Six CML samples were collected in Laboratory F, lysed in red cell lysis buffer (EUTOS protocol), and stored in either TRIzol Reagent (Invitrogen, Carlsbad, CA) or in MagNA Pure LC mRNA lysis buffer (Roche Diagnostics GmbH, Mannheim, Germany). Laboratories C and F received the samples in MagNA Pure LC mRNA lysis buffer (Roche Diagnostics GmbH), as this buffer was used routinely in these laboratories. Laboratories D, G, and H all received the samples in TRIzol reagent (Invitrogen), although in the case of laboratories G and H, this was not fully in accordance with their routine buffer preparation (Perfect Pure buffer (5 PRIME GmbH, Hamburg, Germany) and RNAzol (Tel-Test Inc., Friendswood, Texas, SA), respectively). All samples were shipped on dry ice. Further sample handling and qPCR procedures were done according to local protocols.
Adjustment of local MRD data by the IS CF may be done in the software by using the incorporated possibility to enter a correction factor to the MRD calculation formulae. To test this feature the participants were asked to test generate four reports; an MRD graph and a list of all results before application of the CF, and an MRD graph and a list of all results after inclusion of the CF in the MRD calculations.