Readers who are considering having their ovarian reserve tested will be interested in the study by Ji, et al. [1] in this issue of Annals of Laboratory Medicine. Anti-Müllerian hormone (AMH) or Müllerian-inhibiting substance is a glycoprotein of the transforming growth factor-β superfamily [2]. Following secretion by Sertoli cells in a male fetus, AMH induces the regression of the Müllerian ducts during development [3]. In women, AMH is produced by granulosa cells during early follicle development. As its level correlates with the number of antral follicles, it may be used as a marker of ovarian reserve [4]. For clinical judgment of ovarian function based on serum AMH levels, a reference interval for comparison is required. The current AMH reference interval was established using ELISA [5]. Recently, automated assays have been developed, and these yield lower AMH levels than ELISA [6, 7]. Therefore, it is necessary to establish a reference interval suitable for the new test method [8-10].

As the serum AMH level decreases with age, reference intervals should be determined according to narrow age bins. Therefore, establishing reference intervals is more cumbersome for AMH than for other analytes. Ji, et al. [1] calculated 5-year age-specific reference intervals, with all age bins comprising more than 120 reference individuals. They measured AMH levels in serum samples collected in the Korea National Health and Nutrition Examination Survey (KNHANES), a nationwide population-based study [11, 12], and cryopreserved in the National Biobank. In total, 1,450 participants in KNHANES between 2013 and 2016 were enrolled. Had the study not used KNHANES data, procuring such a large number of participants to calculate reference intervals would have been challenging. A previous study that established a reference interval for an automated assay had potential for selection bias because it included women who visited an infertility center [13]. The strategy used by Ji, et al. [1] of including participants in a population-representative study likely is more reliable.

Given that KNHANES targets the general population, most examinees are likely to be healthy. Healthy participants for reference were selected by applying inclusion criteria. The authors carefully considered factors that may potentially affect the reference AMH levels by measuring serum progesterone levels and performing bone morphogenetic protein-15 genetic testing [1]. The Roche assay used by the authors has a relatively large market share among automated assays [14]. Therefore, many researchers will be able to reference the study results.

The greatest significance of this study is that the authors considered ethnic differences in serum AMH levels [9, 15, 16]. The reference interval mentioned in the package insert of the AMH assay reagent may not be suitable for the target population. This is evident from the reference interval calculated by the authors, which was wider than that mentioned in the package insert, exceeding the limits at both ends of the range. The median serum AMH level in the 25–34-year-old Korean population tended to be slightly higher than that in the Caucasian population [1].

More extensive use of KNHANES participant data or samples in studies to establish reference intervals for analytes would be beneficial; however, it has some limitations. First, depending on the analyte, the information required to identify reference individuals may be lacking. Second, for an analyte that does not have a premeasured value, it may be impossible to measure it because of the limited amount of sample stored. In the study by Ji, et al. [1], the AMH levels of 21 participants could not be measured because of insufficient sample volumes. A multicenter study can be a robust alternative to reference interval studies [17].