Inter-assay variation and reproducibility of progesterone measurements during ovarian stimulation for IVF

In recent years there is increasing evidence that elevated progesterone levels during ovarian stimulation for IVF / ICSI have a negative impact on the ART-outcome. However, different progesterone assays were used in the previous studies and different assays might produce varying results. This retrospective study evaluated the reproducibility and reliability of different progesterone assays with a special focus on progesterone levels below 1.5 ng/ml, as this range is crucial for early detection of progesterone rise during ovarian stimulation for IVF. A total of 413 blood samples were categorized in different progesterone ranges and whether they were retrieved on the day of final oocyte maturation and the results were compared regarding their reproducibility and reliability. To compare the reproducibility between the different progesterone assays, the Intraclass Correlation Coefficient (ICC) was calculated and interpretation of the ICC results was done according to Cicchetti, ranging from poor to excellent. The correlation of the assays was excellent when all samples were compared including samples retrieved on day of final oocyte maturation, however in the ranges of progesterone levels 1.0 ng/ml to < 1.5 ng/ml, 0.8 ng/ml to < 1.0 ng/ml and < 0.8 ng/ml, the ICC varied between poor and excellent. The assays "gen III" and "Architect" showed an excellent reproducibility of progesterone results throughout all ranges of progesterone levels. This analysis demonstrates, that different progesterone assays have a limited reproducibility and that the results depend on the assay used and the range of progesterone level. This fact leads to two important conclusions. Firstly the limited reproducibility might lead to substantially different treatment decisions in ovarian stimulation treatment for IVF and secondly critical interpretation of thresholds, provided by meta-analysis, is crucial despite the risk that the so far gained clinical experience might become irrelevant and has to be adjusted to the results, obtained by each assay.