Dear Editor,

Determining serum estradiol (E2) levels has been a long-standing challenge in laboratory medicine [1]. This is partly because E2 levels may vary widely, ranging from <1 to >2,000 pg/mL, depending on age, sex, and treatment (e.g., fertility hormones and aromatase inhibitors) [2].

For high E2 levels, high-speed automated immunoassays show sufficient sensitivity and specificity and are routinely used for follow-up of women undergoing ovulation induction. These assays have an advantage of fast processing time, underscoring their appropriateness for ovarian induction [2]. However, these assays have limitations for low E2 levels [3]. A 2013 Position Statement by the Endocrine Society mentioned that assays available at that time were neither sufficiently sensitive nor specific to track the onset of pubertal development, especially for E2 levels <30 pg/mL [2].

For years, E2 has been measured in clinical laboratories using RIAs with a measurement range of 2.2–545 pg/mL. Recently, many laboratories have begun replacing traditional RIA with liquid chromatography–tandem mass spectrometry (LC–MS/MS). However, LC–MS/MS-based measurement of low E2 levels remains difficult as the method requires either derivatization or multiple extractions, is time-consuming, and is not applicable at routine laboratory scale [2, 4, 5]. As an ultrasensitive alternative to RIA, the Italian biotechnology company DiaSorin recently developed an automated direct competitive immunoassay with a chemiluminescence-based detection system. The kit indicates a limit of quantification (LOQ) of 16.2 pg/mL.

We aimed to compare the performance of the Estradiol II Gen Assay (DiaSorin, Saluggia, Italy) and ESTR-US-CT RIA (Cisbio Bioassays, Codolet, France) to confirm the LOQ of the former and evaluate its concordance with clinicopathological data.

A total of 117 serum samples were selected among samples planned to be discarded following routine laboratory activities at the University Hospital of Nantes, France, between January and December 2018. We selected aliquots from subjects who were followed up for suspicion of early puberty, early ovarian failure, infertility, or gynecomastia. To determine the LOQ, dilutions were performed with a pool of serum with undetectable E2 levels obtained from prepubescent boys (under 8 years of age). According to French law, which follows the Declaration of Helsinki, our study does not fall within the category of research involving humans. Therefore, we did not seek approval from either the local Research Ethics Committee (Groupe Nantais d’Ethique dans le Domaine de la Santé) or the National Center for Computing and Liberties. All samples were anonymized. After centrifugation, samples were frozen at –20°C until measurement.

The analytical performance of the DiaSorin assay (DA) was in concordance with the manufacturer’s guidelines for repeatability (CVs for 35, 100, and 400 pg/mL: 2.9%, 1.8%, and 1.5%, respectively) and reproducibility (CVs for 100 and 400 pg/mL: 4.7% and 6.3%, respectively). The LOQ was 13.5 pg/mL, which was lower than the value reported by the manufacturer. Although the correlation between the DA and RIA was excellent when all values were compared (Fig. 1A), it was poor for values below 30 pg/mL (Fig. 1B). Bland–Altman analysis showed a negative bias for the DA of –3.1 pg/mL (95% limit of concordance: –41.02 to 34.82) for all values (Fig. 1C) and –3.7 pg/mL (95% limit of concordance –19.03 to 11.67) for values under 30 pg/mL (Fig. 1D).

We compared the concordance of serum E2 levels with the clinicopathological characteristics in 41 young girls (median age: 8 years; age range: 3–17 years). Each subject was classified into one of the following groups: prepuberty, puberty, or pubertal start, determined according to clinical information (Tanner stages, onset of period, and ultrasound data). We checked the clinical concordance with biological values for both assays (Fig. 2). The DA results were concordant with the clinicobiological data of 100% of the “puberty” group subjects (E2>20 pg/mL), whereas concordance of only 88.9% was found for the RIA results. A threshold of 20 pg/mL was selected based on previous studies [6, 7]. The DA and RIA results were concordant with the clinicobiological data of 87.5% and 62.5% of the “prepuberty” group subjects (E2<20 pg/mL). Taken together, these results suggest that the DA performs better than the RIA at tracking the onset of puberty.

This study reports an alternative to traditional RIA methods that is sufficiently sensitive to evaluate most clinical situations, including early puberty. This technology allows for the quick and easy measurement of E2 levels that is also appropriate for routine laboratory activities, although it cannot determine E2 levels below 13 pg/mL. Until a new LC–MS/MS method is developed that is sufficiently rapid, sensitive, and simple to permit low E2 level measurement in a routine laboratory setting, the Estradiol II Gen assay developed by DiaSorin is a viable alternative.