A Proof-Of Practical Approach for Achieving Equivalent Results From Non-Harmonized Measurement Methods While Awaiting Harmonization: the Ca 125 Example

Abstract

Background: Laboratory test results are crucial for clinical decisions, yet inconsistencies arise when measurements are not harmonized due to the lack of suitable higher-order references. This study introduces an approach to improve result comparability across different measurement systems, applicable until full metrological harmonization of the measurand is achieved. Methods: A linear transformation formula was developed, utilizing the 2.5th and 97.5th percentiles of data from source and target methods, to adjust source method results. The feasibility of this formula was tested using carbohydrate antigen 125 (CA 125) data from two commercial assays provided by Roche Diagnostics and Abbott Diagnostics. Method comparison statistics, including difference plots and Passing-Bablok regression, were used to evaluate the transformation's effectiveness before and after adjustment. A web application, "Result Transformer," was developed to facilitate the application of the transformation process. Results: Prior to transformation, the median relative difference between the Roche and Abbott CA 125 assays was 37.7% (95% CI: 34.5-40.8%), exceeding acceptable bias. Passing-Bablok regression yielded a slope of 1.450 (95% CI: 1.400-1.485) and an intercept of -0.83 kU/L (95% CI: -1.50 to -0.29). After adjustment using the proposed approach, the median relative difference decreased to 6.0% (95% CI: 4.3-7.7%), falling within the desirable acceptable bias goal. The slope and intercept of the regression equation improved to 1.075 (95% CI: 1.039-1.102) and -0.12 kU/L (95% CI: -0.71 to 0.19), respectively. Conclusion: The proposed transformation method effectively improved the comparability of results from different assays, permitting a more consistent test result interpretation during patient follow-up.