†(E L Legacki is now at Hollings Marine Laboratory, National Institute of Standards & Technology, Charleston, South Carolina, USA)
Hormone secretion by the maternal ovaries, trophoblast/placenta and fetus occurs sequentially, creating distinct steroid metabolomic ‘signatures’ in systemic blood of pregnant mares that vary with gestational stage. Algorithms were developed to predict the gestational day (GD) from the maternal steroid metabolome (nine steroids; pregnenolone (P5), progesterone (P4), 5α-dihydroprogesterone (DHP), 17α-hydroxyprogesterone, allopregnanolone, 20α-hydroxy-DHP, 3β,20α-dihydroxy-DHP, DHEA and androstenedione) determined by liquid chromatography-tandem mass spectrometry (LC-MS/MS) of eight thoroughbred mares sampled longitudinally throughout pregnancy. A physiologically based model was developed to infer rates of steroid secretion during chorionic gonadotropin secretion, the luteo-placental shift and by the equine feto-placenta unit, demonstrating more variability in P5 and DHP than P4. The average of four empirical models, using nine steroids to predict GD, was calibrated (five mares, R2 = 0.94, RMSE = 20 days) and validated (three mares, R2 = 0.84, RMSE = 32 days). Validation performance was improved using paired samples taken 14 or 30 days apart (RMSE = 29 and 19 days, respectively). A second validation used an independent dataset (single serum samples from 56 mixed breed mares, RMSE = 79 days) and an additional longitudinal subset from the same population sampled monthly throughout gestation (seven mares, RMSE = 42 days). Again, using paired samples improved model performance (RMSE = 32.5 days). Despite less predictive performance of the mixed breed than the thoroughbred datasets, these models demonstrate the feasibility and potential for using maternal steroid metabolomic algorithms to estimate the stage of gestation in pregnant mares and perhaps monitor fetal development.
Journal of Endocrinology is committed to supporting researchers in demonstrating the impact of their articles published in the journal.
The two types of article metrics we measure are (i) more traditional full-text views and pdf downloads, and (ii) Altmetric data, which shows the wider impact of articles in a range of non-traditional sources, such as social media.
More information is on the Reasons to publish page.
| Sept 2018 onwards | Past Year | Past 30 Days | |
|---|---|---|---|
| Full Text Views | 1020 | 20 | 2 |
| PDF Downloads | 258 | 25 | 1 |
Online ISSN: 1479-6805
Print ISSN: 0022-0795
Strengthening biomedical communities
to advance science and health
CONTACT US
Bioscientifica Ltd | Starling House | 1600 Bristol Parkway North | Bristol BS34 8YU | UK
Bioscientifica Ltd | Registered in England no 3190519
