Mette Hansen Viuff, Jesper Just, Sara Brun, Tine Vrist Dam, Mette Hansen, Lars Melgaard, David M Hougaard, Michael Lappe, and Claus Højbjerg Gravholt. The Journal of Clinical Endocrinology and Metabolism, volume 107, Issue 7 (July 2022).
Abstract.
Context.
Women with Turner syndrome (TS) suffer from hypergonadotropic hypogonadism, causing a deficit in gonadal hormone secretion. As a consequence, these women are treated with estrogen from the age of 12 years, and later in combination with progesterone. However, androgens have been given less attention.
Objective.
To assess sex hormone levels in women with TS, both those treated and those nontreated with hormone replacement therapy (HRT), and investigate the impact of HRT on sex hormone levels.
Methods.
At Aarhus University Hospital, 99 women with TS were followed 3 times from August 2003 to February 2010. Seventeen were lost during follow-up. Control group 1 consisted of 68 healthy age-matched control women seen once during this period. Control group 2 consisted of 28 young, eumenorrheic women sampled 9 times throughout the same menstrual cycle. Serum concentrations of follicle-stimulating hormone (FSH), luteinizing hormone (LH), 17β-estradiol, estrone sulfate, DHEAS, testosterone, free androgen index, androstenedione, 17-OH progesterone, and sex hormone–binding globulin (SHBG) were analyzed.
Results.
All androgens, 17-OH progesterone, and sex hormone–binding globulin (SHBG) were 30% to 50% lower in TS compared with controls (P < 0.01). FSH, LH, and estrone sulfate were more than doubled in women with TS compared with controls (P < 0.02). Using principal component analysis, we describe a positive correlation between women with TS receiving HRT, elevated levels of SHBG, and decreased levels of androgens.
Conclusion.
The sex hormone profile in TS reveals a picture of androgen deficiency, aggravated further by HRT. Conventional HRT does not normalize estradiol levels in TS.
The Abzu QLattice.
To investigate if androgen and androgen precursor levels could predict which group each subject belonged to (TS women or controls), we used the Abzu QLattice (ver. 1.6.2; https://arxiv.org/abs/2104.05417), a supervised machine learning tool. The QLattice aims to add functions together to build understandable and interpretable mathematical models using symbolic regression. We split the sectional sex hormone dataset into a training set (75%) and a test set (25%) and used the training set as input to build a classification model. The model was trained with 50 iterations and a max depth of 2. The best resulting model, based on the loss function, was then used to classify the test dataset to evaluate performance.
