Abstract

BCR-ABL measurement by real-time quantitative PCR (RQ-PCR) has become an essential component for assessing treatment response for CML. A major molecular response (MMR) has prognostic significance and can be used to guide therapeutic decisions. However, the various methods are not standardized and the value representing MMR varies, which may lead to misinterpretation of molecular response. To align data, an international reporting scale (IS) was proposed where MMR is 0.10%. Conversion to the IS is achieved by applying laboratory (lab) specific conversion factors (CF). We aimed to

  1. calculate CF for diverse RQ-PCR methods by reference of patient BCR-ABL values to those generated in a reference lab with an established CF;

  2. validate the CF by subsequent patient sample exchange;

  3. examine the concordance of BCR-ABL values after IS conversion;

  4. determine if manufactured reference material is suitable for CF calculation. 34 labs from 13 countries (Australia/New Zealand 11, North/South America 9, Asia 8, Europe 6) sent 615 patient samples to the Adelaide reference lab to determine their specific CF.

The RQ-PCR methods varied by the control gene (ABL 17, BCR 12, GUSβ 4, G6PDH 2, β2M 1, GAPDH 1; 3 labs used 2 controls therefore 37 methods), instrument, probe technology and standards. The CF for each method was calculated from the bias of patient BCR-ABL values between the originating lab and the reference lab, providing the bias was consistent across the dynamic range (

Bland and Altman,
Lancet
,
1986
;
1
:
307
). CF were determined for 33 methods, 1 failed due to inconsistencies in the bias and 3 labs sent insufficient samples. CF were validated by sending subsequent sets of patient samples to the reference lab. The validation process is complete for 12 methods using 384 samples. The specific CF remained valid for each method. The mean bias between the reference and originating lab values was negligible after conversion. The limits of agreement indicated that 95% of values were within ±4.6-fold of the reference value. In contrast, prior to conversion 95% of values were within ±13-fold. Importantly after conversion the concordance in the range representing MMR was 87% (154/178 samples). In the future, conversion to the IS will be achieved using certified reference material, however this is currently not available. In order to mimic the patient bias CF calculation we prepared prototype reference material using BCR-ABL positive cells diluted to 4 levels using volunteer cells. The material was distributed to 29 labs and analysis completed for 24 methods. For 12 of the 24 the CF calculated using the reference material was consistent with the patient bias CF. This indicates that CF calculation is achievable using manufactured reference material but optimization is required before widespread distribution. In summary, alignment of BCR-ABL data generated from diverse methods is achievable using an international standardisation approach, and differences between laboratories are small enough to allow consistent interpretation of results and clinical decision-making.

Author notes

Disclosure:Employment: KL is employed by Novartis. Consultancy: SB Novartis. Ownership Interests:; KL Novartis. Research Funding: SB, TH, GS, FP, MM, AH, NC, DWK, JR Novartis, Bristol Myers Squibb. Honoraria Information: SB, TH, AH, GS, DWK, JR, NC, JM - Novartis, Bristol Myers Squibb. Membership Information: TH, AH, GS, JM: Novartis, Bristol-Myers Squibb.