First Trimester Maternal Cortisol Signatures in Small-for-gestational Age

Background: Abnormal maternal hypothalamus-pituitary-adrenal axis is associated with fetal growth, and we hypothesized that the alteration in metabolic signatures of cortisol might be detectable during early pregnancy. The objective of this study was to identify predictable maternal serum signatures of cortisol metabolism during the rst trimester of women who are expected to deliver small-for-gestational age (SGA) neonates. Methods: This prospective cohort study included 112 pregnant women (with and without SGA, n = 56 each). Maternal serum samples were collected at 10~14 gestational weeks to quantify the levels of cortisol and its precursors and metabolites by liquid chromatography-mass spectrometry. Results: Increased maternal serum levels of tetrahydrocortisol (THF, 11.82 ± 8.16 ng/mL vs. 7.51 ± 2.90 ng/mL, P < 0.005) and decreased 21-deoxycortisol (21-deoxyF, 2.98 ± 1.36 ng/mL vs. 4.33 ± 2.06 ng/mL, P < 0.0001) were observed in pregnant women carrying SGA fetus. In conjunction with individual steroid levels, metabolic ratios corresponding to the activity of related enzymes were calculated. In addition to increased THF/cortisol ratio (P < 0.006), the SGA group showed a signicant increase in the two metabolic ratios including cortisol/11-deoxycortisol (F/11-deoxyF; P < 0.03) and cortisol/21-deoxycortisol (F/21-deoxyF; P < 0.0003) indicating cortisol biosynthesis. The ROC curve generated in combination with three variables of 21-deoxyF concentration and two metabolic ratios of F/21-deoxyF and THF/F resulted in AUC = 0.824 (95% condence interval, 0.713 ~ 0.918). Conclusions: A signicant decrease in maternal serum levels of 21-deoxyF and an increase in two metabolic ratios of F/21-deoxyF and THF/F, indicating cortisol biosynthetic rate, represent a reliable biomarker for the prediction of SGA in the rst trimester.


Introduction
Small-for-gestational age (SGA) neonates are de ned as small babies with birth weights below the 10th percentile for babies in the same gestational condition. The SGA is associated with not only an increased risk of stillbirth and neonatal mortality, but also adverse lifelong health-related outcomes [1,2]. Pregnant women carrying SGA fetus warrant clinical attention such as intensive fetal surveillance to reduce the risk of fetal compromise. Therefore, the accurate prediction of SGA is essential for clinical management. Several predictable biomarkers have been suggested, but they provide poor sensitivities and/or speci cities [3]. It is di cult to distinguish between pathological and constitutional SGA, although it is an important issue in the management of SGA condition [4].
The maternal hypothalamic-pituitary-adrenal (HPA) axis is activated during pregnancy, resulting in increased cortisol production. The intense activity of placental 11β-hydroxysteroid dehydrogenase type 2 (11β-HSD2), which metabolizes cortisol into cortisone, ensures the control of biologically active glucocorticoid levels in the fetus compared with those of the pregnant female, and protects the fetus from excessive cortisol levels [5,6]. Despite the altered adrenal function of newborns with intrauterine growth restriction (IUGR) within the rst several weeks of life [7], increased cord blood levels of cortisol were detected in SGA fetus and newborn [8,9], while low-birth weight preterm infants had decreased cortisol levels compared with full-term normal babies [10,11]. Cortisol metabolism in pregnant women carrying SGA fetus was altered, but the increased levels of maternal serum cortisol were not statistically signi cant [12,13].
Since antenatal overexposure to glucocorticoids is one of the key factors underlying early life programing of disease [14], maternal abnormalities involving the HPA axis can in uence fetal development and birth outcomes. Among the currently available diagnostic procedures [4], non-invasive serum analysis for predictable detection of cortisol levels at an early stage of pregnancy enables intrauterine homeostasis of cortisol metabolism. Here, liquid chromatography-mass spectrometry (LC-MS)-based quantitative pro ling [15] was extensively used to identify metabolic signatures of cortisol in maternal serum samples obtained from SGA group in the rst trimester. The main objective of this study was to establish whether altered cortisol metabolism in maternal serum during the rst trimester was an appropriate tool for assessing the risk of SGA infant delivery.

Materials And Methods
Study design and etiology of SGA Current case-control study population included 56 pregnant women who provided fasting serum samples at 10-14 weeks of gestation and subsequently delivered SGA neonates between 2015 and 2018. SGA was de ned as birth weight less than 10 percentile for each gestational age at delivery according to Korean birth weight standard of reference [16]. The control group included 56 pregnant women with a fasting serum sample drawn at 10-14 weeks and who delivered non-SGA neonates. Control subjects were matched for gestational age at sampling and clinical variables, which affected fetal growth (maternal age, parity, maternal height and weight) based on propensity score analysis. Patients with gestational diabetes, preeclampsia and preterm birth were excluded.

Chemicals
Cortisol and its 2 precursors and 7 metabolites (Fig. 1) were purchased from Sigma (St. Louis, MO, USA), Steraloids (Newport, RI, USA), and Wako (Osaka, Japan). An internal standard of 9,11,12,12,-d 4 -cortisol was obtained from C/D/N isotopes (Pointe-Claire, QC, Canada). The stock solutions of 10 reference standards were prepared at a concentration of 1 mg/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 1 µg/mL. All standard solutions were stored at −20°C until required. A commercially available steroid-free serum sample (SCIPAC, Sittingbourne, UK) was further processed for both calibration and quality control (QC) as described in our previous method [17] and used after con rming the lack of endogenous steroids.
All adrenal steroids were analyzed and quanti ed in MRM mode using electrospray ionization (ESI) coupled with polarity switching with a high-speed scan rate of 30,000 u/s. Among the 10 steroids tested, cortisol (F), 6β-hydroxycortisol (6β-OHF) and cortisone (E) were detected in negative ion mode with the MRM screening method, and the remaining 7 steroids were analyzed in the positive ion mode. All peaks were identi ed via comparison of retention times and matching the area ratios of characteristic ions.
The cartridge was washed twice with 10% methanol (0.7 mL). Serum adrenal steroids were then eluted with absolute methanol (1 mL ⋅ 2). Combined eluates were evaporated under a stream of nitrogen at 40°C. The dried extract was reconstituted with 50 µL of methanol and centrifuged using an Ultrafree-MC Centrifugal Filter (polyvinylidene uoride, pore size: 0.1 µm; Millipore, Billerica, MA, USA) for 5 min at 14,000 rpm. Thereafter, 50 µL of 10% dimethyl sulfoxide (DMSO) was added to the Ultrafree-MC lter and centrifuged for 5 min at 14,000 rpm. Finally, an aliquot (5 µL) was injected into the LC-MS system.

Statistical analysis
The data were analyzed using SPSS (v 22.0; SPSS Inc., Chicago, IL, USA) and GraphPad Prism (v. 8.2; GraphPad Software Inc; San Diego, CA, USA). Based on the quantitative results of individual steroids, the ratios of metabolite to corresponding precursor, which indicate enzyme activities, were examined. A nonparametric Mann-Whitney U test was used to evaluate the group differences between sterol signatures and laboratory ndings. The predictive performance was evaluated according to the receiver operating characteristic (ROC) curves for the 50% training/discovery set split. All quantitative results are expressed as means ± SD and statistical signi cance was considered at P < 0.05.

Clinical characteristics
General characteristics of the study population are listed in Table 1. Clinical parameters such as maternal age, parity, and gestational age/BMI at sampling were not different between the two groups analyzed. The two groups had similar gestational age at delivery, but the birth weight varied signi cantly between the two groups of cases (2.7 ± 0.2 kg vs. 3.2 ± 0.3 kg, P < 0.001).  (Fig. 1). Metabolic ratios of cortisol and its precursors and metabolites were also evaluated to re ect enzyme activities. One of cortical reduction metabolic ratios, THF/F, was also remarkably increased in the SGA group (P < 0.006) ( Fig. 2A). Two metabolic ratios, indicating cortisol biosynthesis, of cortisol to 11-deoxycortisol (F/11-deoxyF; corresponding to 11β-hydroxylase, P < 0.03) and F/21-deoxyF (corresponding to 21-hydroxylase, P < 0.0003) were signi cantly increased in the SGA group ( Fig. 2B and 2C). Other metabolic ratios did not differ between groups studied (data not shown).
Based on quantitative signatures, the ROC curves for discriminate accuracies between SGA and non-SGA groups yielded the highest predictive performance in the area under the ROC curve (AUC) if the three statistically signi cant variables of 21-deoxyF and two metabolic ratios of F/21-deoxyF and THF/F were combined (AUC = 0.824, 95% con dence interval, CI, 0.713 ~ 0.918, Fig. 3).

Discussion
The performance of cortisol immunoassays is diminished by allo-THF, 11-deoxyF, 21-deoxyF, 6β-OHF and synthetic glucocorticoids [18]. Therefore, the LC-MS-based pro ling was conducted to generate serum cortisol signatures of SGA at 10 ~ 14 gestational weeks and quantify individual glucocorticoid levels.
Based on the individual quantities, the metabolic ratios of precursors to their corresponding metabolites were also determined for indirect assessment of comparative enzyme activities.
Progesterone slowly increases from week 9 until week 32 of pregnancy, whereas the serum cortisol level increases signi cantly in the rst trimester and peaks in the second week, followed by a decline in the third trimester [19]. Biologically active free cortisol is generally decreased by the increased corticosteroidbinding globulin; however, in pregnant women the cortisol is replaced by a large quantity of progesterone [20]. Maternal stress is one of the major risk factors for spontaneous abortion during the earliest gestational stages and the increased levels of maternal cortisol are correlated with a higher risk of miscarriage during pregnancy [21]. However, the serum levels of cortisol in the SGA group at early stage of the rst trimester were not statistically signi cant (P = 0.337; Fig. 1), consistent with previous studies involving women in the rst and third trimesters [12,13].
However, the metabolic signatures of cortisol, in this study, showed remarkable changes in SGA (Fig. 1).
Among those signatures, 11β-HSD2 metabolizes cortisol to inactive cortisone, which is reactivated by 11β-HSD1. Both cortisol and cortisone are then reduced to dihydro-, tetrahydro-and allotetrahydrometabolites catalyzed by different reductases. In addition to antenatal glucocorticoid treatment [14], the decreased activity of placental 11β-HSD2 is caused by higher levels of maternal cortisol [5,6], which also results in birth weight reduction [22]. However, the maternal blood levels of cortisone and its 5β-reduced metabolite THE remained unchanged in SGA and non-SGA groups (Fig. 1). The metabolic ratios of E/F, indicating 11β-HSD2 activity, and THE/E were also unchanged (data not shown). In contrast, the metabolite of cortisol catalyzed by 5β-reductases, THF, was signi cantly increased (Fig. 1), while its metabolic ratio relative to cortisol, THF/F, was also higher in women carrying SGA fetus ( Fig. 2A), similar to previous ndings in pregnant women in third trimester [13]. Our study demonstrated that these changes may also represent predictive biomarkers in the rst trimester. The level of 5α-reduced metabolite of cortisol, allo-THF, tended to increase in the SGA group (Fig. 1), while its metabolic ratio relative to cortisol (allo-THF/F) was not signi cant (data not shown). Another reduced metabolite of cortisol monitored in this study, 20α-DHF, which is catalyzed by 20α-reductase, was also not signi cant between the groups (Fig. 1).
Cortisol is derived from two different precursors, 11-deoxyF and 21-deoxyF, catalyzed by 11β-and 21hydroxylases, which are key enzymes in cortisol biosynthesis (Fig. 1). In contrast to suppression of cortisol biosynthesis under low 11-deoxyF levels in cord blood obtained from very low birth weight babies [10], maternal 11-deoxyF was decreased (P = 0.098, Fig. 1), which may be comparable to the apparent decrease in intrauterine fetal death and anencephalic fetus [23]. The maternal F/11-deoxyF metabolic ratio was increased (Fig. 2B), indicating activated 11β-hydroxylase in cortisol synthesis, whereas a reduced 11β-hydroxylase was found in low birth weight neonates [10]. The 21-Hydroxylase de ciency results in increased 21-deoxyF, which represents an alternative diagnostic marker in late-onset congenital adrenal hyperplasia [24], with similar activity between preterm and term infants [25].
The activity of 11β-HSD isoenzymes and its metabolic capacity for glucocorticoids play a pivotal role in regulating fetal growth. The placental mRNA level of 11β-HSD 1 is signi cantly increased during the late gestation period (38 ~ 40 weeks), while the levels of 11β-HSD 2 are decreased, corresponding to fetal maturation and labor [26]. In contrast, pregnant women in their rst trimester showed a compartmental distribution of 11β-HSD 1 and 2 at the feto-maternal interface, both of which were upregulated to possibly coordinate the interaction between isoenzymes [27]. Excessive glucocorticoid levels inhibit fetal growth and are expressed by increased cortisol metabolite THF and the THF/F ratio [13,28]. The nding was in accordance with the negative association of maternal serum THF levels with birth weight of babies (Fig. 4).
To the best of our knowledge, this is the rst study to evaluate metabolic signatures of cortisol in the rst trimester. Although the present work was designed to provide detailed information regarding cortisol metabolism in serum obtained from pregnant women with SGA fetus, this prospective cohort study has several limitations. First, the incidence of adverse fetal outcomes may re ect placental and maternal abnormalities, but the cortisol signatures in placenta were not measured in pregnant women in their rst trimester. The quantitative results from mothers could be indirectly evaluated based on those obtained from babies due to protective mechanism against excessive glucocorticoid levels. However, no comparative experiment was performed. Third, the longitudinal data related to metabolic changes during pregnancy in this study were intended to identify metabolic changes in early stages.
In summary, a decreased serum 21-deoxyF combined with increased F/21-deoxyF combined with higher THF/F ratio, indicating fetal growth inhibition [13,28], represents a potentially reliable biomarker for predicting SGA in the rst trimester (Fig. 3). Studies are needed to investigate the molecular mechanism of enzymes related to cortisol metabolism and their association with lipid pro les.

Availability of data and materials
The raw and processed data used and analyzed in the current study available from the corresponding author.

Ethics approval and consent to participate
This study was approved by the Ethics Committees of Seoul National University Hospital. All participants signed an informed consent form prior to the study.

Consent for publication
Figures Figure 1 Cortisol metabolism and comparative serum levels of individual steroids between SGA and non-SGA pregnant women. As a cortisol precursor, 21-deoxycortisol was signi cantly decreased (P < 0.0001), whereas one of the reductive metabolites of cortisol, tetrahydrocortisol (THF) was remarkably increased in the SGA group (P < 0.005). Serum levels are expressed in ng/mL. The horizontal lines represent the mean value and error bars indicate 95% con dence intervals.

Figure 2
Comparative maternal serum metabolic ratios between SGA and non-SGA groups. Metabolic ratios of (A) tetrahydrocortisol to cortisol (THF/F), (B) cortisol to 11-deoxycortisol (F/11-deoxyF), and (C) cortisol to 21-deoxycortisol (F/21-deoxyF) were signi cantly increased in the SGA group compared with the non-SGA group. All data were converted to log values and the horizontal lines represent the mean value and error bars indicate 95% con dence intervals.