Next-generation biomonitoring of the early-life chemical exposome in neonatal and infant development

Exposure to synthetic and natural chemicals is a major environmental risk factor in the etiology of many chronic diseases. Investigating complex co-exposures is necessary for a holistic assessment in exposome-wide association studies. In this work, a sensitive liquid chromatography-tandem mass spectrometry approach was developed and validated. The assay enables the analysis of more than 80 highly-diverse xenobiotics in urine, serum/plasma, and breast milk; with detection limits generally in the pg-ng mL−1 range. In plasma of extremely-premature infants, 27 xenobiotics are identified; including contamination with plasticizers, perfluorinated alkylated substances and parabens. In breast milk samples collected longitudinally over the first 211 days post-partum, 29 analytes are detected, including pyrrolizidine- and tropane alkaloids which have not been identified in this matrix before. A preliminary estimation of daily toxicant intake via breast milk is conducted. In conclusion, we observe significant early-life co-exposure to multiple toxicants, and demonstrate the method’s applicability for large-scale exposomics-type cohort studies.


Chemicals and Reagents
Supplementary Table 1 provides supplier information and CAS numbers for reagents and chemicals used in this work. Analytical standards were initially dissolved in appropriate solvents (acetonitrile (ACN), methanol, LC-MS-grade water or dimethyl sulfoxide) to yield concentrations of 1 or 2 mg mL -1 . A stock solution containing all compounds at 20-fold the concentration of the highest standard was prepared in ACN. Calibration standards were separately prepared for each validation study from this standard mixture. A separate solution containing all internal standards (IS) at 20-fold the fortified concentrations was prepared in ACN. All chemical stocks and solutions were stored at -20°C. The commercial, analytical-grade perfluorooctanesulfonic acid standard consisted of at least 3 isomers with the linear structure eluting last at the same time as the internal standard. Similarly, a racemic mixture of 2 diastereomers resulted in 2 separated peaks for anisodamine. For both compounds, isomeric peaks were integrated together.

LC-MS/MS method development
The xenoestrogen method developed by Preindl et al. (2019) 1 also incorporated endogenous estrogens. While the objective of this method covered general xenobiotic assessment, the decision was made to also include estrogens in the method development as the base for future methods and to report technical parameters to the scientific community. Moreover, parameters and validation results of xenobiotics that were not successfully validated in any matrix are also included.
Precursor ions, tandem MS fragment masses, and optimized compound-dependent ion optic parameters (declustering potential (DP), collision energy (CE), cell exit potential (CXP)) of the mass spectrometer were determined to achieve maximal ion transmission by injecting single compound standards (ranging from 50 µg L -1 to 5 mg L -1 in 10% ACN) directly into the ion source and utilizing the automatic optimization tool of the system. Final compound-dependent multiple reaction monitoring (MRM) parameters are summarized in Supplementary Table 2. Parameters of the electrospray ionization (ESI) source (curtain gas flow (CUR), sheath gas flow (GS1), drying gas flow (GS2), temperature, spray voltages in positive and negative mode) and the dissociation gas pressure (CAD) were optimized for a selection of the most representative compounds by comparison of ion transmission rates and signal-tonoise ratios (S/N) at different configurations using a flow injection analysis of solvent standards (in ACN) without chromatographic separation as the column was replaced by a connecting piece. Afterwards, full LC-MS/MS measurements of matrix standards were conducted and the S/N ratios of all analytes were evaluated at different configurations. The final method utilized CUR at 30 psi, GS1 at 80 psi, GS2 at 60 psi, a temperature of 500°C, a negative ion spray voltage of -4500 V, a positive ion spray voltage of 5500 V and the CAD pressure setting at "medium". The position of the vertical probe yielded best signal-to-noise ratios at 2 mm instead of the standard setting of 5 mm.

Comparison of plasma and serum matrices
The commercially-available heparinized pooled plasma was heavily contaminated with a number of toxicologically-relevant analytes (bisphenol A (BPA), 2-naphthol, perfluorooctanoic acid (PFOA), perfluorooctanesulfonic acid (PFOS), mono-n-ethylhexyl phthalate (MEHP), cotinine, trans-3-hydroxy cotinine) which resulted in a much higher baseline signal when the extracted matrix (matrix blank) was compared to the pooled serum, thus calibration could not encompass low-level contaminations and would have hampered the quantitation of trace amounts. In a few cases, such as PFOS, the highest matrix-matched standard was of similar intensity to the extracted plasma blank. Fortifying at even higher concentrations would have resolved this issue for highly-contaminated samples, however, the purpose of this method was the accurate quantitation of trace-level contamination. For MEHP, the blank contamination even resulted in detector saturation. Matrix contamination was also present in serum, but fewer analytes were affected. Except for cotinine and MEHP, the contaminations were acceptably low to enable sufficient linear calibration and consequently accurate quantitation after applying baseline correction. This highlights the difficulty of appropriate matrix selection to determine ubiquitous environmental xenobiotics that heavily contaminate commercially-available pooled matrices. Moreover, qualitative features (retention time, ion ratios) were similar for both matrices. Consequently, a decision was made to apply a standard calibration with serum as the matrix of choice for method development and quantitation of plasma samples of prematurely-born babies, as this resembled the infant samples more closely than solvent standards and did not exhibit the heavy background contamination observed with the pooled plasma.
Supplementary Figure 1 depicts chromatograms of selected compounds in the blank serum and plasma matrices and also the fortified standards. As a result of the high contamination, contrary to serum, the plasma blank could not be differentiated from the highest matrix-matched standard for perfluorooctanoic acid ( Supplementary Figure 1 a). Similarly, matrix contamination of the plasma was notably higher for bisphenol A and 2-naphthol when compared to serum ( Supplementary Figure 1 b, c). No quantitation would have been possible for the isomers of PFOS using plasma-matched calibration. Supplementary  Figure 1 d-h compares chromatograms of both mass transitions (quantifier and qualifier) of selected compounds that were identified in the premature infant cohort between both matrices. Figure 1: Comparison between plasma and serum chromatograms, and the presence of nicotine metabolites in breast milk. a-c Comparison of non-spiked plasma and serum (plasma/serum blank) and the highest spiked concentration level of the respective matrix calibration series. d-h Depiction of both mass transitions of chosen compounds detected and quantitated in the premature infant cohort in a serum and a plasma standard at the same fortified concentration. i-j Evidence of the presence of nicotine metabolites in the blank breast milk matrix that was used for method development. Source data are provided as a Source Data file.  Apart from 13 C12-BPA and 13 C18-ZEN, internal standards were not individually tuned and ion optic parameters were adapted from the non-labelled compounds. For some analytes, retention times varied between validation batches and the three matrices urine (U), serum (S) and breast milk (M).

Supplementary Table 3: Spiking levels and extraction recovery results as obtained during in-house validation. Fortified concentrations at low level (LL) and high level (HL), extraction recovery (RE), intermediate precision (RSDR) and repeatability (RSDr) of the investigated xenobiotics
and endogenous estrogens in three matrices. Parameters that could not be determined are displayed as n.d. For a few compounds, the RE and RSDR could not be calculated, while the RSDr was calculated. This was due to no signal detected in at least the low-level fortified samples during the first two validation runs, but successfully-detected peaks during the third validation sequence which were used to assess the repeatability of the method. * Data from breast milk originate from two evaluated validations only, as a retention time shift outside the programmed MRM window was observed in the last validation run. Therefore, no repeatability could be calculated. ** RE and RSDR could not be calculated, while the RSDr was calculated. This was due to no signal being detected in at least one of the low-level fortified samples during the first two validation runs, but successfully-detected peaks during the third validation sequence which was used to assess the repeatability of the method. Table 4: Calibration parameters, limits of quantitation (LOQ), matrix effects and ion ratios as obtained during in-house validation. Average (n=3) regression coefficients, calibrated concentration ranges, limits of quantification (LOQs), signal suppression or enhancement (SSE) and ion ratios (given for the quantitated transition) of the investigated xenobiotics and endogenous estrogens in urine (U), serum (S) and breast milk (M). Parameters that could not be determined are displayed as n.d.