Introduction
Reliable quantitative results in clinical studies are highly affected by selecting the optimal weighting factor for linear and quadratic calibration curves. Although an unweighted linear regression model has been commonly used, appropriate weighting factors such as 1/x, 1/x2, 1/y, and 1/y2 could reduce a percent relative error (%RE) versus concentration, thus improving the precision in quantification.1 Dynamic ranges in bioanalytical assays are usually more than one order of magnitude with data points at each concentration having unequal variances (σ2). The accuracy of instrumental responses decreases in the low range while it increases proportionally to sample concentration.1-3 These heteroscedastic errors could lead to incorrect quantitative information in biomedical applications.
Preeclampsia is characterized by the new onset of pregnancy-induced hypertension and proteinuria after 20 weeks of gestation with complications in 3~4% of pregnancies that may lead to about 50, 000 maternal deaths worldwide a year.4, 5 Physiological changes of maternal hypothalamic-pituitary-adrenal (HPA) axis in pregnancy are associated with maternal and fetal outcomes.6 Dysregulation of maternal HPA-axis also can affect metabolic signatures of adrenal steroids.7, 8 Preeclampsia patients show decreased plasma levels of mineralocorticoids.8, 9 High cortisol levels induced by distress conditions could be associated with preeclampsia.10 In addition, maternal cortisol-to-cortisone ratio is decreased in serum but increased in placenta of patients with preeclampsia.11-13 However, metabolic changes of adrenal steroids of preeclampsia have not been extensively investigated to date.
Biologically active steroid hormones are mostly presented at trace levels in various biological specimens including blood and tissue samples. Due to similar chemical structures, their analytical selectivity can be hampered by matrix complexity derived from endogenous interference and isobaric components.14 Therefore, levels of steroids should be calculated by accurate calibration curves with the correct weighting factor in addition to proper sample pretreatment protocols. In particular, a weighting factor of 1/x2 has been recently recommended for mass spectrometry-based bioanalytical methods2, 3 as it produces lower b value for lower background signal and better stability against unweighted linear calibration equation (y = ax + b).
Here, a comprehensive profiling of adrenal steroids using liquid chromatography-mass spectrometry (LC-MS) was applied to both serum and placenta samples obtained from 15 preeclampsia patients and 17 age-matched pregnant women. In conjunction with verification of quantitative results in accuracy by comparing two different linear regression models with weighting factor of 1 and 1/ x2, altered metabolic signatures of preeclampsia were identified and their physiological roles were discussed.
Experimental
Reagents
Reference standards of adrenal steroids (Table 1) were purchased from Sigma (St. Louis, MO, USA) and Steraloids (Newport, RI, USA). Internal standards (9, 11, 12, 12, -d4-F for eleven corticoids; 2, 2, 4, 6, 6, 17α, 21, 21, 21, - d9-Prog and 2, 2, 4, 6, 6, 21, 21, 21, -d8-17α-OHProg for eight progestogens; 16, 16, 17-d3-T for three androgens; 2, 2, 3, 4, 4, 6-d6-DHEA for DHEA; 16, 16, 17-d3-TS for DHEA-S and Preg-S; 17α, 21, 21, 21, -d4-Preg for Preg; and 21, 21, 21-d3- 17α-OHPreg for 17α-OHPreg) were obtained from C/D/N isotopes (Pointe-Claire, QC, Canada).
Table 1. Comparison of percent bias* at four concentrations with different weighting factors.
*Values are expressed as mean of four experiments.
Stock solutions of all reference standards were prepared at a concentration of 1mg/mL in a mixture of high performance liquid chromatography (HPLC)-grade methanol and chloroform (9:1, v/v; Burdick & Jackson, Muskegon, MI, USA). Working solutions were prepared at concentrations ranging from 0.02 to 1μg/mL. All standard solutions were stored at -20oC until used.
Study subjects and sample collection
Samples were obtained at 34-36 weeks of gestation. This study was approved (KUGH1644-002) by the medical ethics committee of Korea University Guro Hospital, Seoul, South Korea). Preeclampsia was diagnosed based on a blood pressure of at least 140/90mmHg on two or more separate occasions and the development of proteinuria of at least 300mg in a 24h collection or the presence of greater than 2+ of protein on a catheterized urine specimen.
Biospecimens and data used in this study were provided by the Biobank of Korea University Guro Hospital, a member of the Korea. Blood samples were allowed to clot before centrifugation at 2, 000 g for 10 min. All aliquots of serum were stored at -80oC. Biopsies were taken from the central region (1 cm3) of placenta tissue on the maternal area. After washing maternal blood from the sample with saline, biopsy samples were immediately frozen in liquid nitrogen and stored at -80oC until use.
Analytical procedure
Quantitative profiling of adrenal steroids was performed based on a previous method.15 In brief, serum sample (200μL) was spiked with 20μL of an internal standard mixture (d4-F, d8-17α-OHProg, and d4-Preg, 0.2 μg/mL; d9- Prog and d3-17α-OHPreg, 0.1 μg/mL; d3-TS, 1 μg/mL; d3- T, 0.02μg/mL; d6-DHEA, 0.5 μg/mL). After dilution with 1.8 mL phosphate buffer (0.2 M, pH 7.2), the sample was incubated with 50 μL of β-glucuronidase at 55°C for 1 h. To reduce matrix interference including water soluble peptides and proteins, the hydrolyzed sample was loaded onto an Oasis HLB cartridge preconditioned with 4 mL of methanol and water, respectively. After washing the cartridge twice with 0.7 mL of 10% methanol, the sample was eluted twice with 1mL of absolute methanol. Combined eluates were evaporated under a stream of nitrogen at 40oC. The dried extract was reconstituted with 50μL of methanol and centrifuged at 14, 000rpm for 5 min using an Ultrafree-MC centrifugal filter. Then 50 µL of 10% dimethyl sulfoxide (DMSO) was added to the Ultrafree-MC filter and centrifuged at 14, 000 rpm for 5 min. Finally, an aliquot (5μL) was injected into the LC-MS system in both MRM and SIM modes.
Placental steroids were extracted according to a previous protocol.16 Then 2 mg of placenta sample was spiked with 20μL of IS mixtures (d4-F, d8-17α-OHProg, and d4-Preg, 0.2μg/mL; d9-Prog and d3-17α-OHPreg, 0.1 μg/mL; d3- TS, 1μg/mL; d3-T, 0.02 μg/mL; d6-DHEA, 0.5μg/mL). After adding 0.6mL of methanol, the mixture was sonicated for 20 min. The sample was then homogenized using a TissueLyser (Qiagen, Hilden, Germany) at 25Hz for 5min each with 3 zirconia beads twice and centrifuged at 12, 000rpm for 10 min. After transferring the supernatant to a glass tube, methanol was removed in a nitrogen stream at 40 °C and enzymatic hydrolysis step was conducted followed by the same procedures used for the serum assay.
Calibration and validation sets
Calibration sets of serum and placenta samples were prepared at nine and seven different concentrations, respectively, using negative control samples spiked with 23 steroids. Steroid free serum and placenta samples were prepared as described previously16, 17 with minor modifications. Placental tissue samples (50mg) were added to 1mL of chloroform: methanol (1:1, v/v) and pulverized using a TissueLyser at 25Hz for 5min with 4 zirconia beads. Then, the sample was centrifuged at 12, 000 rpm for 10 min and the solution was discarded twice. The sample was then mixed with 1 mL of chloroform: 0.6 M methanolic HCl (1:1, v/v), centrifuged at 12, 000 rpm for 10 min, and the organic solvent was discarded twice. To remove the remaining HCl, the sample was washed with 1 mL of 20% ethanol five times and the sample was frozen at -80oC until needed.
Statistical analysis
All data analyses were performed using SPSS version 22.0 (SPSS Inc., Chicago, IL, USA) and Prism (v. 8; GraphPad Software Inc., San Diego, CA, USA). Parameters (slope and y-intercept) for both unweighted (wi = 1) and weighted (wi = 1/x 2 ) least squares (y = ax + b) were obtained with the following formulas on Excel 2016 spreadsheets (Microsoft Corp., Seattle, WA, USA):
\(\text { slope }\left(a_{w}\right)=\frac{\Sigma w_{i} \cdot \Sigma\left(w_{i} \cdot x_{i} \cdot y_{i}\right)-\Sigma\left(w_{i} \cdot x_{i}\right) \cdot \Sigma\left(w_{i} \cdot y_{i}\right)}{\Sigma w_{i} \cdot \Sigma\left(w_{i} \cdot x_{i}^{2}\right)-\left(\Sigma\left(w_{i} \cdot x_{i}\right)\right)^{2}}\)
\(\mathrm{y}-\text { intercept }\left(b_{w}\right)=\frac{\Sigma\left(w_{i} \cdot x_{i}^{2}\right) \cdot \Sigma\left(w_{i} \cdot y_{i}\right)-\Sigma\left(w_{i} \cdot x_{i}\right) \cdot \Sigma\left(w_{i} \cdot x_{i} \cdot y_{i}\right)}{\Sigma w_{i} \cdot \Sigma\left(w_{i} \cdot x_{i}^{2}\right)-\left(\Sigma\left(w_{i} \cdot x_{i}\right)\right)^{2}}\)
Percent bias was calculated by dividing the bias by the theoretical value and multiplying by 100. Quantitative results are expressed as mean ± standard deviation (SD). Group differences were compared using the MannWhitney U test. P values of less than 0.05 were considered statistically significant.
Results
To validate the applicability of different linear regression models between unweighted (wi = 1) and weighted (wi = 1/ x2) factors, quantitative results from LC-MS based profiling of adrenal steroids were compared with four test sample sets (Table 1). Percent biases in the unweighted model were inversely proportional to concentrations (-739.4– 852.9% for LOQ; -137.7–173.7% for low; -11.4–12.6% for medium; -12.7–9.8% for high), while those in the weighted model were < 18% for all concentrations.
When both linear regression models were applied to clinical samples, the unweighted linear regression showed significantly increased serum levels for six adrenal steroids of Preg, 21-deoxyF, THE, 18-OHF, 6β-OHF, and DHEA (P < 0.01 for all, except THE and DHEA, P < 0.04) in preeclampsia patients, whereas DOC was remarkably decreased (P < 0.0001). In contrast, the 1/x2 regression model resulted in altered levels of only two steroids (Preg, 21-deoxyF and THE; P < 0.05 for all) among seven in the unweighted linear regression (Table 2).
Table 2. Quantitative results* of serum adrenal steroids with different weighting factors.
* Results are expressed as ng/mL in concentration.
Metabolic ratios were also calculated based on quantities of individual steroids corresponding to related enzyme activities. Altered metabolic ratios of both 6β-OHF/F and 18-OHF/F were found (P < 0.001 for both) in only the unweighted linear regression (Table 2). Other metabolic ratios of Preg-S/Preg (P = 0.003 for wi = 1 and P = 0.027 for wi = 1/x2), DOC/Prog (P < 0.0001 vs. P = 0.018), F/21- deoxyF (P = 0.006 vs. P = 0.011), and (THF+allo-THF)/ THE (P = 0.011 vs. P < 0.005) were significantly changed in both regression models.
Seven adrenal steroids were detected in placenta. They showed no statistically significant differences in either regression model except that levels of Adione were decreased (P = 0.030) in preeclamptic patients with the weighted linear regression model (Table 3). Metabolic ratio of Adione/17α-OHProg was also decreased in preeclampsia (P = 0.002), indicating 17, 20-lyase activity.
Table 3. Quantitative results* of placental adrenal steroids with different weighting factors.
* Results are expressed as ng/g in concentration.
Discussion
Functional steroidogenesis is essential in pregnancy and its preeclamptic changes have been extensively studied.7, 8, 18, 19 Some discrepancies in quantitative findings could be linked to sampling and methodological issues. The devised weighted linear regression with weighting factor of 1/x2 in LC-MS based quantitative profiling suggests that metabolic signatures of adrenal steroids in serum and placenta might provide reliable physiological indicators for preeclampsia at late pregnancy weeks of 34~36 in the third trimester.
The unweighted linear regression showed unacceptable percent biases for most steroids analyzed at low concentrations (Table 1). In particular, both 6β-OHF and 18-OHF are relatively high polar compound with fast elution order and these molecules could be easily interfered chromatographically by matrix effects during elution, which cause increase background signal (higher b value) and produce inaccurate quantitative outcomes in unweighted linear calibration equation (y = ax + b) at low concentrations. In general the variance of each data point may be quite different when the range in x-values is large, but the simple regression model considers that all the y-values have equal variances. Larger deviations at larger levels tend to affect the regression line more than smaller deviations correlated with smaller concentration (heteroscedasticity) resulting in the inaccuracy in the lower end of the calibration range.1 In method validation in this study (Table 4), the calibration linearities with the unweighted model were better than those calculated based on the weighted. The linear regression equation usually produces high calibration linearity (r > 0.097) and lower accuracy than 5%, but the straight-line model was systematically rejected at the 95% confidence level on the basis of the Lack-of-fit and Mandel’s fitting test. The quantitative results obtained from the quadratic regression model did not differ significantly from the theoretical values, whereas those derived from the linear regression model were systematically biased.20
Table 4. Method validation for 27 adrenal steroids in placental tissue.
a The limit of quantification was measured according to an S/N ratio > 10.
b Precision and accuracy are expressed as the mean values of data obtained from four different concentrations based on the weighted calibration linearities.
c Recovery is expressed as the mean values of data obtained from three different concentrations.
Serum results obtained between two different weighting factors were comparable in quantitative outcomes (Table 2). DOC is one intermediate in aldosterone biosynthesis. Its excessive production may lead to mineralocorticoidderived hypertension that can occur in patients with 11β- hydroxylase or 17α-hydroxylase deficiencies, congenital adrenal hyperplasia, DOC-producing adrenal tumors, or in patients taking 11β-hydroxylase inhibitors.21–23 However, plasma levels of DOC were not associated with pregnancy induced hypertension.8, 24 Urinary DOC levels are not significantly different between preeclampsia and control.25 Here, DOC serum level was remarkably decreased in preeclampsia patients, leading to a decreased DOC/Prog ratio known to be responsible for 21-hydroxylase activity. However, those were not significant in the weighted model (Table 2).
Serum levels of both 6β-OHF and 18-OHF in preeclampsia were significantly increased in the unweighted model, but not changed in the weighted model (Table 2). In our previous study, serum 6β-OHF in the third trimester pregnant women with preeclampsia was found to be increased than that in the control.18 Such discrepancy in results might also be caused by different linear regression models used. As one of hybrid adrenal steroids, 18-OHF showed increased serum level in patients with primary aldosteronism, a leading cause of secondary hypertension. It can also be used to differentiate subtyping of adrenal diseases.26 However, no experimental findings with maternal 18-OHF have been reported in preeclampsia to date. According to metabolic enzymes of both hydroxylated cortisols to cortisol, 6β- and 18-hydroxylases may not also remarkable with preeclampsia at 34~36 gestational weeks in the weighted model.
Preeclampsia patients presented significantly higher serum levels of Preg, 21-deoxyF, and THE than those of controls in both linear regression models (Table 2). Blood levels of Preg in preeclampsia patients did not differ from those in controls, 8, 27 consistent with our data. However, previous studies used blood samples collected at the second trimester of pregnancy 24~29 weeks, 8 while we collected samples at the third trimester. They also measured levels of serum Preg using a competitive enzyme immunoassay known to cause overestimation by cross reactivity with endogenous steroids.27 Consistent with a previous finding showing increased Preg in maternal blood, 7 increased serum levels of Preg in preeclampsia (P < 0.05 for both unweighted and weighted models) in the present study affected the metabolic ratio of Preg-S/Preg which was remarkably decreased regardless of weighting factors (P < 0.03 for both linear regression models).
Although serum cortisol levels were not significantly different between patients with preeclampsia and controls in the present study, consistent with results in maternal plasma, 8 its precursor 21-deoxyF and one of 5β-reduced metabolites, THE, were remarkably increased in patients with preeclampsia (P < 0.02 for both weighting factors; Table 2). These findings may reveal metabolic signatures of cortisol in preeclampsia. Levels of cortisol 21-hydroxylase as a metabolic enzyme represented by F/21-deoxyF ratio were decreased in preeclampsia patients, while levels of another cortisol biosynthetic pathway from 11-deoxyF were not significantly different between patients with preeclampsia and controls (P = 0.455 in unweighted model and P = 0.370 in weighted model). In general, maternal regulation of 11β- hydroxysteroid dehydrogenase (11β-HSD) can protect the fetus from excessive glucocorticoids and its type 2 (11β- HSD2) can convert active cortisol to inactive cortisone.28 The activity of 11β-HSD2 can be also expressed by the metabolic ratio of (THF+allo-THF)/THE.29 Normotensive pregnant women maintain inactivation of placental cortisol. The metabolic ratio of (THF+allo-THF)/THE in maternal sera of preeclampsia patients was decreased in the present study. Such result is expected for the placenta.28, 29
In the placenta, there were no compatible results between unweighted and weighted models (Table 3). In contrast to an increased tendency of placental DOC level in patients with preeclampsia in the unweighted model (P = 0.064), it was not statistically significant in the weighted regression (P = 0.941) which might be due to matrix interference. DOC concentrations between serum and placenta were also not correlated in both patients and controls (r < 0.18, P > 0.6). Due to its curative effect on preeclampsia model, 30 increased levels of placental 17α-OHProg in preeclampsia patients need to be confirmed in further experiment. Adione in rat placenta could be a predominant androgen for ovarian production of estrogens during pregnancy.31 It is mainly produced via Δ4-steroidogenic pathway.32 However, its metabolic activity in human placenta has remained unclear to date, while CYP17 mRNA and protein expression have recently been found in human trophoblasts.33 The CYP17A1 gene regulates two distinct steroidogenic enzymes of 17α-hydroxylase and 17, 20- lyase, although the association between its polymorphism and the risk of preeclampsia has been controversial.34, 35 In addition to decreased placental levels of Adione in preeclampsia in the weighted model (P = 0.030), the placental metabolic ratio of Adione/17α-OHProg indicating 17, 20-lyase was significantly decreased in preeclampsia patients (P = 0.002), while 17α-hydroxylase activity expressed by 17α- OHProg/Prog ratio was not different between patients and controls (Table 3).
Figure 1. Scatter plots with linear fitted curves and Pearson correlations between serum and placental pregnenolone levels. Both (a) unweighted and (b) weighted regression models showed positive correlations between serum and placenta pregnenolone concentrations (P < 0.001 for both), whereas those of preeclampsia patients showed no significant correlations (blue colored).
Correlations of adrenal steroids between serum and placenta were not observed in wither the unweighted or the weighted model (data not shown). In contrast, positive correlations of pregnenolone between serum and placenta levels were clearly found in healthy pregnant women (r = 0.747 and 0.748, P < 0.001 in both linear regression experiments; Fig. 1). This might imply metabolic disturbance in pregnenolone production at late gestational weeks of preeclampsia. Dysfunction of 11β-HSD2 is a hallmark of preeclampsia physiology.28 Maternal serum metabolic ratio of (THF+allo-THF)/THE was significantly decreased in preeclampsia patients. It may indicate that the fetus is not protected from maternal glucocorticoid excess. In the placenta, decreased 17, 20-lyase activity may explain the lack of androgen production in patients with preeclampsia.
This study has some limitations. First, the number of subjects was relatively small. Therefore, covariables such as age and BMI were not included in our statistical analyses. However, quantitative results were compared between age-match groups. In addition, gestational diabetes as one of serious complications of pregnancy known to cause increased blood pressure was excluded from the preeclampsia group. Second, only seven adrenal steroids were detected in placenta against 23 serum steroids monitored. In addition, associations between serum and placenta steroid signatures were not fully investigated. Third, amniotic concentrations of adrenal steroids were not compared for both maternal serum and placenta to fully understand the maternal-placental-fetal axis in preeclampsia physiology.
Despite these limitations, the present work successfully provided reliable quantification of adrenal steroids from both serum and placenta samples with minimized matrix interference. It was achieved by introducing a correct weighting factor of 1/x2 for linear calibration curve combined with comprehensive LC separation and selective MS detection assay. Comparative metabolic signatures of adrenal steroids in both serum and placenta clearly revealed impaired steroidogenesis in preeclampsia at late gestational weeks of 34~36 in the third trimester.
Conclusions
An optimized LC-MS coupled to corrected linear regression approach revealed metabolic signatures of adrenal steroids in preeclampsia based on exact quantitative outcomes. Based on individual quantity of adrenal steroids, altered steroidogenic activity of 11β-HSD2 was found in maternal serum. In addition, 17, 20-lyase deficiency was observed in placenta of preeclampsia patients. However, large-scale randomized trials are needed to confirm these findings.
참고문헌
- Almeida, A. M.; Castel-Branco, M. M.; Falcao, A. C. J. Chromatogr. B 2002, 774, 215, DOI: 10.1016/s1570-0232(02)00244-1.
- Gu, H.; Liu, G.; Wang, J.; Aubry, A. F.; Arnold, M. E. Anal. Chem. 2014, 86, 8959, DOI: 10.1021/ac5018265.
- Kim, J. Y. Yakhak Hoeji 2016, 60, 304, DOI: 10.17480/psk.2016.60.6.304.
- Ghulmiyyah, L.; Sibai, B. Semin. Perinatol. 2012, 36, 56, DOI: 10.1053/j.semperi.2011.09.011.
- Mol, B. W. J.; Roberts, C. T.; Thangaratinam, S.; Magee, L. A.; de Groot, C. J. M.; Hofmeyr, G. J. Lancet 2016, 387, 999, DOI: 10.1016/S0140-6736(15)00070-7.
- Duthie, L.; Reynolds, R. M. Neuroendocrinology 2013, 98, 106, DOI: 10.1159/000354702.
- Hertig, A.; Liere, P.; Chabbert-Buffet, N.; Fort, J.; Pianos, A.; Eychenne, B.; Cambourg, A.; Schumacher, M.; Berkane, N.; Lefevre, G. Am. J. Obstet. Gynecol. 2010, 203, 477, DOI: 10.1016/j.ajog.2010.06.011.
- Berkane, N.; Liere, P.; Lefevre, G.; Alfaidy, N.; Nahed, R. A.; Vincent, J.; Oudinet, J. P.; Pianos, A.; Cambourg, A.; Rozenberg, P.; Galichon, P.; Rousseau, A.; Simon, T.; Schumacher, M.; Chabbert-Buffet, N.; Hertig, A. Placenta 2018, 69, 40, DOI: 10.1016/j.placenta.2018.07.004.
- Uddin, M. N.; Horvat, D.; Jones, R. O.; Beeram, M. R.; Zawieja, D. C.; Perger, L.; Sprague, D. C. C.; Kuehl, T. J. J. Matern. Fetal Neonatal Med. 2015, 28, 1296, DOI: 10.3109/14767058.2014.951627.
- Vianna, P.; Bauer, M. E.; Dornfeld, D.; Chies, J. A. B. Med. Hypotheses 2011, 77, 188, DOI: 10.1016/j.mehy.2011.04.007.
- McCalla, C. O.; Nacharaju, V. L.; Muneyyirci-Delale, O.; Glasgow, S.; Feldman, J. G. Steroids 1998, 63, 511, DOI: 10.1016/s0039-128x(98)00056-7.
- Kosicka, K.; Siemiatkowska, A.; Szpera-Gozdziewicz, A.; Krzyscin, M.; Breborowicz, G. H.; Glowka, F. K. Endocrine 2018, 61, 125, DOI: 10.1007/s12020-018-1586-4.
- Jayasuriya, N. A.; Hughes, A. E.; Sovio, U.; Cook, E.; Charnock-Jones, D. S.; Smith, G. C. S. J. Clin. Endocrinol. Metab. 2019, 104, 2355, DOI: 10.1210/jc.2018-02312.
- Keevil, B. G. Clin. Biochem. 2016, 49, 989, DOI: 10.1016/j.clinbiochem.2016.04.009.
- Lee, C.; Kim, J. H.; Moon, S. J.; Shim, J.; Kim, H. I.; Choi, M. H. J. Steroid Biochem. Mol. Biol. 2020, 198, 105615, DOI: 10.1016/j.jsbmb.2020.105615.
- Moon, J. Y.; McNamara, K. M.; Lee, J. J.; Chung, B. C.; Sasano, H.; Choi, M. H. J. Steroid Biochem. Mol. Biol. 2018, 178, 185, DOI: 10.1016/j.jsbmb.2017.12.012.
- Lee, S. H.; Lee, N.; Hong, Y.; Chung, B. C.; Choi, M. H. Anal. Chem. 2016, 88, 11624, DOI: 10.1021/acs.analchem.6b03183.
- Moon, J. Y.; Moon, M. H.; Kim, K. T.; Jeong, D. H.; Kim, Y. N.; Chung, B. C.; Choi, M. H. J. Steroid Biochem. Mol. Biol. 2014, 139, 182, DOI: 10.1016/j.jsbmb.2013.02.014.
- Perez-Sepulveda, A.; Monteiro, L. J.; Dobierzewska, A.; Espana-Perrot, P. P.; Venegas-Araneda, P.; Guzman-Rojas, A. M.; Gonzalez, M. I.; Palominos-Rivera, M.; Irarrazabal, C. E.; Figueroa-Diesel, H.; Varas-Godoy, M.; Illanes, S. E. PLoS One 2015, 10, e0139682, DOI: 10.1371/journal.pone.0139682. eCollection 2015.
- Loco, J. V.; Elskens, M.; Croux, C.; Beernaert, H. Accred. Qual. Assur. 2002, 7, 281, DOI: 10.1007/s00769-002-0487-6.
- Irony, I.; Biglieri, E. G.; Perloff, D.; Rubinoff, H. J. Clin. Endocrinol. Metab. 1987, 65, 836, DOI: 10.1210/jcem65-5-836.
- Biglieri, E. G.; Kater, C. E. Clin. Chem. 1991, 37, 1843, DOI: 10.1093/clinchem/37.10.1843.
- Marques, P.; Tufton, N.; Bhattacharya, S.; Caulfield, M.; Akker, S. A. Endocrinol. Diabetes Metab. Case Rep. 2019, 18-0164, DOI: 10.1530/EDM-18-0164.
- Walker, B. R.; Williamson, P. M.; Brown, M. A.; Honour, J. W.; Edwards, C. R. W.; Whitworth, J. A. Hypertension 1995, 25, 626, DOI: 10.1161/01.hyp.25.4.626.
- Shah, D. M.; Gomez-Sanchez, C. E.; Knuppel, R. A.; Vaughn, W. K. Am. J. Perinatol. 1987, 4, 102, DOI: 10.1055/s-2007-999749.
- Lenders, J. W. M.; Williams, T. A.; Reincke, M.; Gomez-Sanchez, C. E. Eur. J. Endocrinol. 2018, 178, R1, DOI: 10.1530/EJE-17-0563.
- Shin, Y. Y.; Jeong, J. S.; Park, M. N.; Lee, J. E.; An, S. M.; Cho, W. S.; Kim, S. C.; An, B. S.; Lee, K. S. Mol. Med. Rep. 2018, 17, 2681, DOI: 10.3892/mmr.2017.8165.
- Wang, G.; Huang, Y.; Hu, T.; Zhang, B.; Tang, Z.; Yao, R.; Huang, Y.; Fan, X.; Ni, X. FASEB J. 2020, 34, 15379, DOI: 10.1096/fj.202001003RR.
- Aufdenblatten, M.; Baumann, M.; Raio, L.; Dick, B.; Frey, B. M.; Schneider, H.; Surbek, D.; Hocher, B.; Mohaupt, M. G. Pediatr. Res. 2009, 65, 198, DOI: 10.1203/PDR.0b013e31818d6c24.
- Mohammed, N. A.; Ali, A. H.; El-Tahawy, N. F. G.; Rifaai, R. A. J. Cytol. Histol. 2018, 9, 1, DOI: 10.4172/2157-7099.1000496.
- Jackson, J. A.; Albrecht, E. D. Biol. Reprod. 1985, 33, 451, DOI: 10.1095/biolreprod33.2.451.
- Warshaw, M. L.; Johnson, D. C.; Khan, I.; Eckstein, B.; Gibori, G. Endocrinology 1986, 119, 2642, DOI: 10.1210/endo-119-6-2642.
- Escobar, J. C.; Patel, S. S.; Beshay, V. E.; Suzuki, T.; Carr, B. R. J. Clin. Endocrinol. Metab. 2011, 96, 1385, DOI: 10.1210/jc.2010-2504.
- Lim, J. H.; Kim, S. Y.; Kim, D. J.; Park, S. Y.; Han, H. W.; Han, J. Y.; Lee, S. W.; Yang, J. H.; Ryu, H. M. Pharmacogenet. Genom. 2010, 20, 605, DOI: 10.1097/FPC.0b013e32833df033.
- Hou, B.; Jia, X.; Deng, Z.; Liu, X.; Liu, H.; Yu, H.; Liu, S. Hum. Genom. 2020, 14, 33, DOI: 10.1186/s40246-020-00286-0.