Autor:innen:
S. Anker (Heidelberg, DE)
J. Morgenstern (Heidelberg, DE)
J. Adler (Magdeburg, DE)
M. Brune (Heidelberg, DE)
S. Brings (Heidelberg, DE)
T. Fleming (Heidelberg, DE)
E. Kliemank (Heidelberg, DE)
M. Zorn (Heidelberg, DE)
A. Fischer (Göttingen, DE)
J. Szendrödi (Heidelberg, DE)
L. Kihm (Heidelberg, DE)
J. Zemva (Heidelberg, DE)
Aims
For the quantification of steroids, mass spectrometry offers a higher degree of specificity and sensitivity than conventional immunoassays. However, as reference intervals have not been adjusted accordingly [1], laboratories are required to determine their own values which is time- and cost-consuming due to the recruitment of a sufficient number of healthy reference subjects [2]. In this study, we addressed the question whether reference intervals for serum steroids can be established by an indirect approach. Furthermore, we wanted to contribute to the methodological harmonization of reference intervals.
Materials and Methods
We used data of 9801 individuals, of whom only age and gender were recorded, to establish post-hoc reference intervals for androstenedione, dehydroepiandrosterone, testosterone, progesterone, dihydrotestosterone, 17α-hydroxyprogesterone/-pregnenolone, corticosterone, 11-deoxycorticosterone, 11- and 21-deoxycortisol, aldosterone, cortisol, and cortisone. Analyses were performed on a Waters® Acquity UPLC class I system coupled to a Waters® XEVO TQ-S LC-MS using a MassChrom® Steroids kit (Chromsystems). The indirect reference interval algorithm [3,4], includes three robust quantile-based steps which were executed in R (Version 4.1.1): 1. Bowley’s quartile skewness predicts whether a normal or a lognormal distribution should be assumed. 2. An iterative boxplot method is applied to remove obvious outliers. 3. A normal quantile-quantile plot provides the 2.5th and 97.5th percentiles, calculated from the intercept and slope of the linear regression line according to the formula RI = intercept±1.96∙slope [4]. Post-hoc reference intervals were compared to reference intervals derived from studies using a direct approach.
Results
The majority of reference intervals generated by the indirect method showed a very good overlap with those derived from a direct approach. Significant deviations were seen for sex hormones such as progesterone, 17α-hydroxypregnenolone/-progesterone, and testosterone. This was reflected by skewed quantile-quantile-plots, which indicated an inhomogeneous distribution of the underlying data. The indirect approach yielded complete reference intervals also in cases, in which the lower reference limit was not provided by the direct method, e. g. in the case of aldosterone, 17-OH-pregnenolon and 11-Deoxycorticosterone.
Discussion/Conclusion
Our results suggest that it is a valid approach to verify and establish reference intervals by an indirect method. It should, however, be noted that reference intervals for widely varying sex hormones may differ significantly between direct and indirect methods due to missing background information on female cycle, menopause and stages of puberty when the indirect approach is applied to routine laboratory results. Further, this study contributes a substantial set of data to the methodological harmonization of steroid reference ranges using mass-spectrometry.