Ethical approval and study population
The local Institutional Review Board approved the study (CIRB/2011/404/D) and written informed consent was obtained from each participant. A case-control study was conducted to evaluate the metabolic and cytokine effect of DHEA administration in women, below the age of 42 starting their IVF treatment who met one of the two following features of POR (an abnormal ovarian reserve test and/or a previous poor response to ovarian stimulation in an IVF cycle) were assessed for eligibility.(3)The ESHRE working group on Poor Ovarian Response Definition of diminished ovarian reserves (AMH <1.0 ng/mL or Day 2 or 3 FSH >10 IU/L), or women with fewer than four oocytes retrieved with either standard long or antagonist protocols was used to defined POR in this study(3).Inclusion criteria included women with diminished ovarian reserves (anti-müllerian hormone<1.0 ng/mL or D2/3 follicle stimulating hormone>10 IU/L), or women with fewer than four oocytes retrieved with either standard long or antagonist protocols. The study excluded women with previous or current DHEA supplementation, use of corticosteroids within the past three months, major systemic illnesses, and allergy to DHEA. 28 eligible patients received DHEA (Pharma Natural, USA) at the dose of 75 mg/day for three to eight months prior starting their controlled ovarian stimulation (COS), herein known as DHEA+, and 24 patients received no treatment (DHEA-). Table 1summarizes the baseline characteristics of the patients in this study.Average age (DHEA-, 36 years; DHEA+, 37 years) and body mass index (DHEA-, 24 kg/m2; DHEA+, 23 kg/m2) were similar in both groups (p>0.05). Baseline hormones between the two groups were comparable.
All individuals received the same stimulation protocol, same starting dose of gonadotropin, and fertilization technique. Briefly, the IVF/ICSI treatment cycle involved an antagonist-based COS protocol consisting of daily sub-cutaneous injections of recombinant-FSH (Puregon, Follitropin beta, 300iu; MSD, USA) and highly-purified human menopausal gonadotropin (Menopur; Menotropin, 150 IU; Ferring Pharmaceuticals, Germany) with initiation of gonadotropin releasing hormone antagonist (Ganirelix, Orgalutan, 0.25 mg s/c; MSD, USA) on day 5 of COS. The dose of Menopur and Puregon could be further increased depending on individual ovarian response. All patients had thisstandardized antagonist (short) protocol: no agonist (long) protocol was used. Human chorionic gonadotropin (i.m 10,000IU hCG; Pregnyl; MSD, USA) was administered when at least one follicle measured ≥17 mm in diameter (averaged orthogonal measurements). The endometrial thickness, peak estradiol and progesterone levels were assessed on the day of human chorionic gonadotropin (hCG) trigger. Ultrasound-guided trans-vaginal oocyte retrieval was performed 36 hours after hCG administration. The effect of DHEA supplementation on the markers of ovarian reserve (anti-müllerian hormone; AMH), and follicular function (IGF-1)(36), as well as ovarian follicular levels of estradiol, testosterone, and DHEA collected from the lead follicle at the time of OPU were assessed through ELISAas previously described(37).
Embryo transfer was performed on day 2 or day 3 of embryo-culture, and luteal phase support was achieved with vaginal progesterone (micronized progesterone, Utrogestan, 200 mg three times a day, Besins-International, France). Pregnancy was established by serum beta-hCG seventeen days post embryo transfer. Clinical pregnancy will be established by a transvaginal ultrasound four weeks after embryo transfer.
FF (DHEA+, N=18 and DHEA-, N=16)were divided for metabolomics and cytokine analyses. For untargeted metabolomics analysis, sample preparation followed previously published reports with some modifications.(38,39) A volume of 50 µL from each FF sample was thawed at 4°C, and FF proteins were precipitated with 200 µL ice-cold methanol. After vortexing, the mixture was centrifuged at 16,000 rpm for 10 min at 4°C and the supernatant was collected and evaporated to dryness in a speedvacuum evaporator. The dry extracts were then re-dissolved in 200 µL of water/methanol (98:2; v/v) for liquid chromatography-tandem mass spectrometry (LC-MS/MS) analysis.
A pooled quality control (QC) sample was generated to allow comparison of analytic behavior over long periods of time. The pooled reference samples were for the purposes of quality control (i.e., to ensure relative consistency among identical samples within days) and for quality assurance (i.e., to ensure consistent results between days). They did not contribute data to downstream statistical analysis.
Liquid Chromatography-Tandem Mass Spectrometry-based Metabolomics
The supernatant fraction from sample preparation step was analyzed using Agilent 1290 ultra-high pressure (performance) liquid chromatography system (Waldbronn, Germany) equipped with a Bruker impact II Q-TOF mass spectrometer with its normal electrospray ionization (ESI) ion source (Bruker Daltonics). 2.5 μL of samples was injected and were separated using Waters Acquity HSS T3 (2.1 mm i.d. x 100 mm, 1.8 µm) at a flow rate of 0.2 mL/min.The oven temperature was set at 50°C. The gradient elution involved a mobile phase consisting of (A) 0.1% formic acid in water and (B) 0.1% formic acid in methanol. The initial condition was set at 5% B. A 5.5 min linear gradient to 60% B was applied, followed by a 13 min gradient to 98% B (total 24 min including wash and re-equilibration) at a flow rate of 0.4 ml/min.The ion spray voltage was set at 4,500V, and the DryTemperature was maintained at °C. The drying nitrogen gas flowrate and the nebulizer gas pressure were set at 8.0 L/min and 26 psi, respectively. Calibration of the system was performed using sodium formate clusters before data acquisition. The stability of the LC-MS method was examined and evaluated by sodium formate clusters (1 mM NaOH, 0.1% formic acid, 50% 2-propanol) infused into the system.
The ESI mass spectra were acquired in positive ion mode. Mass data were collected between m/z 100 and 1000 at a rate of three scans per second. Auto MS/MS was triggered at 8 Hz with duty cycle of 1.5 s. Threshold was set at 1500 counts, with active exclusion activated after 3 spectra, released after 0.3 min and overwritten if the current or previous intensity changes. MS/MS spectra were acquired at collision energyof 20–50 eV automatically varied by the charge states and the intensities of the selected precursors. Fragment spectra acquisition was carried out at a scan rate dependent on the MS precursor intensities - MS/MS spectra for high-intensity precursors were acquired for a shorter time (90000 counts, 12 Hz) than low-intensity precursor ions (10000 counts, 6 Hz) thus allowing for a balancing of maximal scan time and MS/MS spectral quality. As shown in Figure S1, the eight pooled quality control samples clustered in Principal Component Analysis (PCA) scores plots, and together with retention time CV% < 0.1 min, peak m/z values 3 mDa, and relative standard deviations of peak areas < 20%, there was good system stability, mass accuracy and reproducibility of the chromatographic separation during the whole LC-MS/MS sequence. In addition, intensity CV% of the identified compounds in pooled quality control samples are low (average 6%). PCA hotelling (T2) revealed one DHEA+ subject as an outlier (D4) and was removed from further analysis (Figure S1).
Structure identification was achieved via the following in MetaboScape (version 2.0): elemental composition was predicted via isotopic pattern following the rules (i) mSigma of MS1: 20 with tolerance of 5 ppm and (ii) MS2: 50 with tolerance of 2 mDaof the differential metabolites was searched against Bruker HMDB (Human Metabolome Database) using a precursor match of ±10 mDa, minimum score of 400 and minimum match score of 250. Progesterone, glycerophosphocholine, linoleic acid and valine were structurally confirmed using chemical standards.
Multiplex immunoassay analysis
45 cytokines were detected and measured using ProCartaplex (EBioscience, CA, USA) as previously reported [BDNF, EGF, Eotaxin (CCL11), FGF-2 (FGF basic), GM-CSF, CXCL1 (GROα), HGF, IFNγ, IFNα, IL-1RA, IL-1β , IL-1α, IL-2, IL-4, IL-5, IL-6, IL-7, CXCL8 (IL-8), IL-9, IL-10, IL-12 p70, IL-13, IL-15, IL-17A, IL-18, IL-21, IL-22, IL-23, IL-27, IL-31, CXCL10 (IP-10), LIF, CCL2 (MCP-1), CCL3 (MIP-1α), CCL4 (MIP-1β), βNGF, PDGF-BB, PLGF, CCL5 (RANTES), SCF, CXCL12 (SDF1α), TNFα, LTA (TNFβ), VEGF-A, VEGF-D].(40)Briefly, 5 μL of FFs were diluted with 5 μL Universal Dilution Buffer, and mixed with 50 μL of antibody-conjugated, magnetic beads in a 96 well DropArray plate (Curiox Biosystems, Singapore) and rotated at 450 rpm for 120 min at 25°C while protected from light. Beads were internally dyed with different concentrations of two spectrally distinct fluorophores and covalently conjugated to antibodies against the 45 cytokines, chemokines and growth factors. The plate was washed three times with wash buffer (PBS, 0.05% Tween-20) on the LT210 Washing Station (Curiox) before adding 10 μL of secondary antibody and rotating at 450 rpm for 30 min at 25°C protected from light. Subsequently, the plate was washed three times with wash buffer, and 10 μL of streptavidin-phycoerythrin added and rotated at 450 rpm for 30 min at 25°C protected from light. The plate was again washed thrice with wash buffer; 60 μL of reading buffer was then added and the samples read using the Bio-Plex Luminex 200 (BioRad). The beads were classified by the red classification laser (635 nm) into its distinct sets, while a green reporter laser (532 nm) excites the phycoerythrin, a fluorescent reporter tag bound to the detection antibody. Quantitation of the 45 cytokines was then determined by extrapolation to a six or seven-point standard curve using five-parameter logistic regression modelling. Calibrations and validations were performed prior to runs and on a monthly basis respectively.
GraphPad Prism 6 (GraphPad Software Inc.) was used for performing all statistical analyses. Data were checked for normal distribution using Kolmogorov-Smirnov test. Unpaired or paired t-test was performed, as appropriate, to determine statistical significance between groups form normally distributed data. Mann-Whitney U test was used for non-normally distributed data. For comparing more than three groups, the data were analyzed using ANOVA test, followed by the t-test with Bonferroni adjustment. P<0.05 was considered significant.Metabolomic data was further analyzed by Principal Component Analysis (PCA) and Partial Least Squares Regression (PLSR) modelling (Unscrambler X version 10.1) after the normalization of data by first centering the data to the median and scaling it by division with the standard deviation. Full cross-validation was applied in PLSR to increase model performance and for the calculation of β-coefficient regression values.Metabolites with β-coefficient regression values≥1 are considered to have contribute significantly to the PLSR model. In this study, metabolites fulfilling both PLSRβ-coefficient regression values>1.2 and Student t-test p<0.05 was considered as differential.