Comparison of two different methods for measuring anti-mullerian hormone in a clinical series

Anti Müllerian Hormone (AMH) is increasingly being used as a biochemical marker for the assessment of the growing ovarian pool and thus as a surrogate marker for the ovarian reserve and female fertility [1]. AMH was originally introduced as a marker with minimal variation during the menstrual cycle [2, 3] in opposition with other markers that have hitherto been used, such as follicle stimulating hormone (FSH) and inhibin. However, recent publications have demonstrated intra-cycle fluctuations, with the highest levels in the early follicular phase and the lowest levels in the late luteal phase [4], as well as substantial circadian variations [5]. This should be interpreted with caution because the results are based on a small number of samples, and they are influenced by assay performance. It must be considered that the total variation associated with the measurement of AMH is a combination of the within-person variation and a variation linked to the assay itself, which is represented as follows: CV _TOTAL = √ [(CV _{Within-person Biological Variation})² + (CV _Analytical)²] [6]. In clinical biochemistry, the setting of analytical goals based on the within-person biological variability is often considered the most appropriate, and the optimal performance (CV _Analytical < 0.25 CV _{Within Biological variation}), should be achieved when the technology and methodology are available [6].

Until recently, AMH has been measured with an enzyme-linked immunosorbent assay (ELISA, Beckmann Coulter), which is a manual technique characterized by large analytical variation. However, a new automated assay (Elecsys Cobas, Roche), which shows good intermediate imprecision, has recently been introduced for measuring AMH [7].

The aim of this study was to compare the utility of the two assays in a clinical setting to evaluate whether the assays meet the goals of analytical performance. We investigated the blood samples collected from women who participated in a study examining the potential changes in ovarian reserve following laparoscopic sterilization as an example of minimally invasive pelvic surgery.

AMH, the total AFC, and OV did not change significantly during the 6-month follow-up period from the baseline levels prior to surgery.

Analyzing controls on consecutive days, we found a large difference in the analytical variation between the two methods. As seen in Table 1, the analytical variance for the Elecsys Cobas method was approximately 3 % in the high and low ranges compared with the ELISA method, which showed a larger variation in the high range and a particularly larger variation in the low range with a CV of more than 10 %.

As depicted in Fig. 1, the concentrations of AMH measured by the Elecsys Cobas assay were, on average, 32 % lower than the concentrations measured by ELISA (Fig. 1a), and there was a tendency of increased bias between the two assays in the high concentration range (Fig. 1b).

The total variance (biological and analytical) in the clinical samples was divided in two levels, < and > 10 pmol/l. The total variance (mean CV%) in the < 10 pmol/l group was 32 % when measured by ELISA and 28 % when analyzed using the Elecsys Cobas method. For concentrations > 10 pmol/l, the CV was approximately 22 % according to both assays. Applying the goals for analysis (CV _Analytical < 0.25 CV _{Within-person Biological Variation}), the analytical variation should not exceed 5.3 % and 7.6 % in the two levels, respectively. This was achieved by the Elecsys Cobas assay throughout the measuring range, whereas the AMH Gen II ELISA failed to meet the goals in the low measuring range.

The limit of quantitation (LOQ) (CV < 15 %) was 3 pmol/l for the ELISA and 0.5 pmol/l for the Elecsys Cobas method. Accordingly, approximately 15 % of the measurements were below the LOQ for the ELISA assay, but only 2 % were below the LOQ for the Elecsys Cobas method.