Performance evaluation of Siemens Atellica enhanced estradiol assay

Introduction

Estradiol (17β-estradiol) (E2) is the most potent one of naturally occurring estrogens and also the most widely analyzed one in clinical routine and research laboratories. The determination of this C18-steroid hormone in serum provides relevant information in many endocrinological or gynecological conditions. Thus, E2 measurements are useful in assisted reproduction to monitor follicular development and in the clinical evaluation of hypogonadism, hirsutism, polycystic ovary syndrome, amenorrhea, female infertility, menopause, gonadal tumors, and feminization in males or monitoring of aromatase inhibitor therapy in breast cancer, among others [1]. Physiological levels of E2 are low (often <184 pmol/L) in the serum of men, children, and postmenopausal women, but they can range widely (110–2,937 pmol/L) in premenopausal women [2]. These differences in the expected levels of serum E2 are even larger, spanning across five orders of magnitude, when considering the variety of its clinical applications. Thus, for example, serum E2 concentration may reach up to 36,713 pmol/L during testing performed in support of in vitro fertilization programs, whereas aromatase inhibitor therapy can decrease serum E2 concentration to 3.7 pmol/L [3]. In summary, serum E2 presents a wide range of reference intervals and clinical applications that determine a need to maintain analytical quality specifications in a wide analytical range. In this regard, assessment of the analytical capacity of an E2 assay is critical when considering its reliability to answer specific clinical or research questions [2].

High-throughput automated direct (unextracted) immunoassays allow the growing clinical need for E2 quantitation and are convenient to be implemented by most hospital clinical laboratories. However, awareness of loss of analytical specificity and accuracy at low concentrations of direct immunoassays has grown over the last few years [4], [5], [6], [7], [8], [9], [10], [11]. Very recently, two new analyzers (Atellica IM 1300 and Atellica IM 1600, Siemens Healthineers, Erlangen, Germany) are incorporated and expected to expand into many clinical laboratories. In this regard, we evaluated the analytical performance of the direct immunoassay Atellica IM 1600 enhanced estradiol assay (eE2) (Siemens Healthineers, Erlangen, Germany). Particularly, we focused our attention on the assessment of imprecision, accuracy, limit of quantification, assay linearity, and method comparison while paying special attention to eE2 low analytical range.

Materials and methods

Results

Limit of quantification

The imprecision of the assay was also assessed with decreasing concentrations of E2 using serum samples. All serum samples presented an interday imprecision <20%, 53 pmol/L being the lowest concentration evaluated. Minimum concentration of E2 corresponding to CV <10% was estimated to be 125 pmol/L (Figure 1).

Figure 1:

Limit of quantification of Atellica IM 1600 eE2 assay.

Assay linearity

Correlation between measured and expected values was high (R ² > 0.99), indicating good linearity for both sets of serial E2 dilutions. The average percent recovery for dilution series A and B was 89% and 83%, respectively, (Figure 2).

Figure 2:

Atellica IM 1600 eE2 assay linearity studies starting from a serum sample with eE2 concentration of 1,957 pmol/L (A) and 3,748 pmol/L (B).

Method comparison

Comparison of results obtained for serum samples using Atellica IM 1600 and ADVIA Centaur XP is shown in Figure 3. The Atellica IM 1600 assay demonstrated good correlation with ADVIA Centaur XP assay over the entire range (R ² = 0.99) and at E2 concentrations <734 pmol/L (R ² = 0.96). Passing–Bablok regression equation was as follows: for Atellica, it was 0.96 (95% CI: 0.95–0.98) × Centaur + 2.8 (95% CI: −4.9 to 8.7) for the entire range and 0.95 (95% CI: 0.89–0.99) × Centaur + 5.9 (95% CI: −2.76 to 14) for concentrations lower than 734 pmol/L. Bias between both methods was 18 pmol/L (95% CI: −43 to 6.9) for the entire range and −5.3 pmol/L (95% CI: −11 to 0.41) for concentrations lower than 734 pmol/L.

Figure 3:

Method comparison.

Passing–Bablok regression and Bland–Altman analyses comparing the Atellica IM 1600 eE2 and Advia Centaur XP eE2 assays. (A.1) E2 whole range concentration Passing–Bablok regression (n = 119). (A.2) E2 whole range concentration Bland–Altman plot (n = 119). (B.1) E2 concentrations <734 pmol/L Passing–Bablok regression (n = 76). (B.2) E2 concentrations <734 pmol/L Bland–Altman plot (n = 76).

Comparison of results obtained for serum samples using Atellica IM 1600 and DRG Sensitive ELISA is shown in Table 3. The concentration of four serum samples analyzed using the DRG Sensitive ELISA was <11 pmol/L, which also presented undetectable results when analyzed using the Atellica IM 1600 (<43.3 pmol/L). The concentrations of 11 of 16 serum samples analyzed using the Sensitive ELISA were between 11.0 and 43.3 pmol/L, which presented undetectable results (<43.3 pmol/L) when analyzed using the Atellica IM 1600. Conversely, the concentration of eight of the nine serum samples analyzed using the DRG Sensitive ELISA was >43.3 pmol/L, which also presented detectable results when analyzed using the Atellica IM 1600.

Table 3:

Low E2 serum samples comparison between DRG Sensitive ELISA and Atellica IM 1600 eE2 assays.

DRG Sensitive ELISA, pmol/L	n	Atellica IM 1600, pmol/L	n	Agreement, %
<11	4	<43	4	100%
11–43	16	≤43	11	69%
		>43	5
≥43	9	≥43	8	89%
		<43	1

1. n, number of measurements.

Discussion

In this study, we assessed the analytical performance of the Atellica IM eE2 assay. Direct E2 immunoassays offer many advantages that include their commercial availability as automated assays, avoiding of time-consuming sample extraction, high throughput, and fast turnaround times. A method capable of quantifying across a wide dynamic range is an essential tool in clinical diagnostic laboratories, especially those engaged in fertility testing.

The assay performed well overall, demonstrating good linearity and imprecision across a typical clinical range. In this regard, according to Stanczyk et al. [2], we focused the evaluation of analytical imprecision and accuracy on E2 values <3,671 pmol/L. The intraday imprecision (CV from 1.8% to 4.0%) and interday imprecision (CV from 1.0% to 6.6%) were satisfactory at the five different concentrations of quality controls and serum pools and met the desirable specification for imprecision based on biological variation (<11%) [12]. Imprecision results are in agreement with those reported by the manufacturer and comparable with those previously reported for the eE2 assay in the ADVIA Centaur [14]. By means of reference certified materials, accuracy of the assay was evaluated at concentrations of 114, 690 and 1,340 pmol/L. The results obtained by the Atellica IM eE2 assay were compared with the certified values determined by the use of an isotope dilution gas chromatography mass spectrometry (GC–MS) reference method. Accuracy was good (<6.9%) and met the quality requirements of desirable specification for estradiol inaccuracy based on biological variation specifications (<8.3%) [12] at the values of 690 and 1,340 pmol/L. However, accuracy at the low level (114 pmol/L) was 28% and did not meet this requirement, which could be in agreement with the difficulties of direct assays to reliably evaluate low levels of E2 due to cross-reactivity with other steroid molecules or interferents not removed from the serum, which can result in significant bias [6], [7]. In fact, Ketha et al. [13] reported that analysis of data reported in a survey conducted in the College of American Pathologists showed that an E2 concentration of 106 pmol/L measured by LC–MS/MS (Liquid chromatography-tandem mass spectrometry) was overestimated (40% to 300%) by all the evaluated immunoassays platforms. In this survey, different ADVIA Centaur platforms presented an accuracy of approximately 40%.

Repeated measuring of an E2-free sample (eE2 assay diluent, n = 10) did not result in any detectable result. However, determination of the LOD was not possible as the commercial assay does not report values below the company-stated LOD (43 pmol/L). Instead, we were able to evaluate the limit of quantification as interassay CV of decreasing serum levels, obtaining a CV <20% at 53 pmol/L and a CV <10% at 125 pmol/L. This indicates that, in contrast with the difficulty of the assay in achieving good accuracy at low levels, imprecision remains very low in the entire range of the assay, also including low levels. This is also in agreement with results obtained with the ADVIA Centaur eE2 assay by Chen et al. [14].

Overall, results indicate that the method is reliable and appropriate for routine measurements of E2 concentrations beyond 43 pmol/L. However, as with other current direct immunoassays, the assay is not useful when evaluating very low serum E2 levels such as those expected in postmenopausal women receiving aromatase inhibitors for the treatment of breast cancer. In this scenario, methods need to be able to distinguish between suppressed levels of less than 3.7 pmol/L and pretreatment levels that commonly range from 37 to 55 pmol/L [3]. Mass spectrometry and indirect radioimmunoassays are currently the proposed alternatives to achieve this analytical sensitivity. Further improvement of direct immunoassays and E2 standardization is needed to better quantify low serum E2 levels in specific clinical situations.

Sample comparison between Atellica and ADVIA Centaur XP eE2 assays was performed covering all ranges of concentration evaluated and at concentration <734 pmol/L, which is the most common result obtained in our daily routine. No significant bias was observed both at the entire range and at concentration <734 pmol/L, and a high agreement between assays was noted. Finally, Atellica and the DRG E2 sensitive assays also presented good agreement and most of the samples below the LOD of Atellica (<43 pmol/L) were also <43 pmol/L by DRG.

In summary, the Atellica IM eE2 assay demonstrated acceptable imprecision and accuracy above the LOD of the assay and showed good agreement with the ADVIA Centaur XP.