Association of urinary concentrations of phthalate metabolites with quinolinic acid among women: A potential link to neurological disorders.

A total of 758 urine samples were provided by 126 women in this analysis. Those included 273 pregnancy samples (by 115 women) and 485 non-pregnancy samples (by 117 women). Women provided 6 samples on average (range of 1 to 23) with 71% of the women providing 3 to 7 samples. Women’s mean (standard deviation) age was 34.2 (4.0) years and BMI of 23.5 (3.7) kg/m 2 . Women were mainly Caucasian (92%), none were current smokers with 27% as past smokers, and nearly all had college or higher degrees (98%) ( Table 1 ). All women’s fertility treatments followed the same in vitro-fertilization (IVF) protocol (luteal phase agonist). The proportion of samples with detectable concentrations of the different phthalate metabolites ranged from 74% for MEHP to 100% for MEP ( Table 2 ). Phthalate metabolite concentrations were comparable to the concentrations observed in the whole sample of women in EARTH study ( Nassan, Williams, et al., 2019 ) and in females of the general population in the USA (National Health and Nutrition Examination Survey, NHANES) ( CDC, 2019 ). All urine samples had detectable concentrations for QA, with a geometric mean of 3.20 (SD=4.91) μg/mL and 5.10 (SD= 5.67) μg/mg for the creatinine-normalized QA. The normal reference range for the latter is between 0 and 6.3 μg/mg creatinine. To allow for comparison between our data on QA and reference range we normalized QA by creatinine using formula QA (ng/mL) / creatinine(mg/dL)*10 ( Table 2 ). After urine dilution adjustment, the geometric means of QA but not phthalate metabolites nor SG or urinary creatinine, showed an increasing trend from non-pregnancy through the pregnancy trimesters ( Supplemental Figure 1 ) Pairwise Spearman correlations between concentrations for phthalate metabolites and QA measured in the same urine samples were all significantly positive ranging from 0.36 for MEHP to 0.68 for DBP metabolites (P-value for all <0.0001) ( Supplemental Figure 2 ). The SG-adjusted ICC ranged from 0.13 (0.08, 0.21) for MCPP to 0.49 (0.41, 0.57) for MEP. The SG-adjusted ICC for QA was 0.18 (0.12, 0.26) ( Supplemental Table 1 ). On average, among all women, the adjusted percent changes in the urinary QA concentrations associated with each doubling of urinary phthalate metabolite concentration were all positive except for MEHP and statistically significant except for MCPP, MEHP and MCOP ( Figure 2 and Supplemental Table 2 ). The adjusted percent changes in the urinary QA concentrations associated with each doubling of urinary phthalate metabolite concentration ranged from 1.98% (95%CI: 0.03, 3.97) for MEHHP to 13.7% (95%CI: 10.6, 16.9) for ∑DBP metabolites. The interaction term between pregnancy status and the association of phthalate metabolites with QA concentration was statistically significant for all except for MEP, DBP metabolites, and MBzP ( Supplemental Table 2 ) but not between the trimesters except for MCOP and MCNP where the third trimester was significantly different than during non-pregnancy ( Supplemental Figure 3 ). The observed associations between the low molecular weight phthalates (MEP, DBP metabolites, and MBzP) and QA tended to be stronger among pregnancy samples than non-pregnancy samples ( Figure 2 and Supplemental Tables 3 and 4 ). However, the associations of high molecular weight phthalates and QA tended to be stronger among the non-pregnancy samples (DEHP metabolites, MCOP, and MCNP) ( Figure 2 and Supplemental Tables 3 and 4 ). The results were consistent when analysis was restricted to one sample per woman ( Supplemental Figure 4 ). Our results in Figure 3 demonstrate that specific gravity was an essential covariate (Li et al. 2013) which suggests that a portion of the urinary phthalate metabolite and QA association was due to a similar excretion pathway. For instance, after adjustment for SG the associations were either attenuated (i.e., MnBP, MiBP, MEP, MBzP) or did not remain (i.e., DEHP metabolites). Based on these results, it was deemed most appropriate to adjust for SG as a covariate rather than normalizing phthalate metabolites by SG.

The Environment and Reproductive Health (EARTH) study is a prospective cohort study started in 2004 that aimed to identify environmental and dietary determinants of fertility among couples from an infertility treatment ( Hauser, Meeker, Duty, Silva, & Calafat, 2006 ). Women who came to the MGH Fertility Center for treatment and were between 18–45 years of age were eligible to participate. Research nurses contacted eligible women and around 55% of those referred by physicians enrolled. At enrollment, participants completed questionnaires about sociodemographic, lifestyle, medical, family, and reproductive history. The nurse recorded the height and weight of the participants from which body mass index (BMI) was calculated. Women provided urine samples at enrollment and during assisted reproductive treatments (referred to as non-pregnancy samples, and also during each trimester of pregnancy. Participants with singletons, twins, or triplets who were born between 2005 and 2015 and were at least 2.5-year-old in 2014 were invited with their child to participate in a neurobehavioral child follow-up study ( Messerlian et al., 2017 ). Therefore, the present analysis included 126 female participants from the EARTH study who had a singleton infant and completed the neurodevelopmental surveys and for whom we had quantified phthalate metabolite concentrations and QA in at least one urine sample (2005–2012) before the index pregnancy ( Messerlian et al., 2017 ). The Institutional Review Boards of the Harvard T.H. Chan School of Public Health, the Massachusetts General Hospital (MGH) approved the study, and the Centers for Disease Control and Prevention (CDC). All participants provided written informed consents. At each visit, using standard procedures, women provided a spot urine sample in a sterile polypropylene cup. Study staff recorded the collection time and measured the specific gravity (SG) of the urine samples using a handheld refractometer (National Instrument Co. Inc.). Urine samples were divided into aliquots (approximately 4ml of urine in a 5ml cryovial), frozen, and stored at −80°C before overnight shipment on dry ice to the CDC laboratory (Atlanta, GA) for quantification of the urinary concentrations of the phthalate metabolites. Briefly, the analytical techniques for quantification of the phthalate metabolites involved enzymatic deconjugation of the target analytes followed by solid-phase extraction, separation by high performance liquid chromatography, and detection by isotope-dilution tandem mass spectrometry ( Dwivedi, Zhou, Powell, Calafat, & Ye, 2018 ; Silva, Jia, Samandar, Preau, & Calafat, 2013 ; Zhou, Kramer, Calafat, & Ye, 2014 ). CDC staff quantified total (free plus conjugated) urinary concentrations (μg/L) of 11 phthalate metabolites (monoethyl phthalate (MEP), mono-n-butyl phthalate (MnBP), mono-isobutyl phthalate (MiBP), mono-3-carboxypropyl phthalate (MCPP), monobenzyl phthalate (MBzP), mono-2-ethylhexyl phthalate (MEHP), mono-2-ethyl-5-oxohexyl phthalate (MEOHP), mono-2-ethyl-5-hydroxyhexyl phthalate (MEHHP), mono-2-ethyl-5-carboxypentyl phthalate (MECPP), mono carboxyisooctyl phthalate (MCOP), and mono carboxy isononyl phthalate (MCNP)).The limits of detection (LOD) ranged from 0.2 for to 1.2 μg/L. We created summary measures after weighting with molecular weight for metabolites of Di (2-ethylhexyl) phthalate (DEHP= ∑(MEHP+MEHHP+MEOHP+MECPP)) and Dibutyl phthalate (DBP = ∑(MnBP+MiBP)). All sample aliquots (approximately 4ml of urine in a 5ml cryovial) were shipped on dry ice in a single batch for analysis to the laboratories at Ethos Research & Development, Newport KY for quantifying QA in urine. As described previously ( Nassan, Gunn, et al., 2019 ), Calibrators and internal standards for QA were created from solid standards. Bio rad lyphocheck quantitative urine levels 1 and 2 were used as quality control (QC) samples. Calibrators, QC samples, and patient samples were all prepared in the same manner. QA was analyzed using LC-MS/MS (Agilent 1260 Infinity HPLC and Agilent 6410B triple quadrupole mass spectrometer). QA samples were resolved on Agilent Poroshell 120 EC-C18, 2.7 μm, 3.0 × 100 mm column. QA was quantified using Mass Hunter Quantitative Analysis Version B.08.00 Software. We calculated descriptive statistics for demographic and time-varying characteristics among women in the study population. We examined the distribution of all phthalate metabolites and QA concentrations, and due to skewness we modeled the different exposures and the outcome as continuous variables after natural-log transformation. We first examined the pairwise correlation between the different phthalate metabolites (exposure) and QA (outcome) by scatter plots and then calculated the Spearman correlation coefficients and 95% confidence intervals (95%CI). Because women gave multiple urine samples over time, we estimated the intra-class correlation coefficient (ICC) for urinary phthalate metabolites and QA adjusted for specific gravity as an indicator for the reliability of the concentrations over time. Participants provided 6 samples on average (range of 1 to 23) and to improve power all samples were included in the analysis. We used linear mixed effect models with a random intercept for each woman to account for the within-woman correlation among the repeated measures. We selected the covariates a priori based on previous knowledge ( Cho et al., 2017 ; Nassan, Gunn, et al., 2019 ). The final model included specific gravity of the urine (continuous), age (continuous), race (Caucasian or not), BMI (continuous), smoking (ever/never), season (warm: April through September/cold) and pregnancy status (pregnancy/non-pregnancy samples). We estimated the percent changes in the QA concentration for each doubling in the different phthalate metabolites. We also examined if these associations were modified by pregnancy status or by trimester by adding an interaction term. In addition, we estimated the percent changes among the pregnancy and non-pregnancy samples as well as the different trimesters separately. Finally, as a sensitivity analysis, we repeated the analyses after restricting the sample to one sample (first one) per woman using linear regression. We then assessed the percent changes in the first samples regardless of pregnancy status as well as only among the first pregnancy samples and the first non-pregnancy samples. We considered two-sided alpha <0.05 as statistically significant. We conducted all statistical analyses using SAS version 9.4 (SAS Institute Inc., Cary, NC).

In the present analysis, background urinary concentrations of phthalate metabolites were all significantly positively moderately (r >0.3) to highly (>0.6) correlated with urinary QA concentration measured in the same urine samples. The multivariable models also showed that higher concentrations of several urinary phthalate metabolite concentrations were associated with higher urinary QA concentrations (with strongest associations for DBP metabolites). Furthermore, the associations between low molecular weight phthalates and QA were stronger among the samples collected during pregnancy while the associations were stronger for the high molecular weight phthalates among the non-pregnancy samples, suggesting effect modification by pregnancy status which could be explained by differences in exposure, metabolism, and/or excretion. In addition, based on molecular docking experiments by Singh et al. ( Singh et al. 2018 ), it was expected that DBP and DEHP and their respective metabolites would have stronger associations given that they have strong binding affinities with ACMSD compared to its biological substrate. Although the associations of urinary phthalate metabolites and QA did not significantly differ by pregnancy trimester, there was a positive trend of higher QA concentrations over pregnancy trimesters. QA is a neuroactive metabolite that is produced during tryptophan metabolism i.e., the Kynurenine Pathway (KP) ( A. A. B. Badawy, 2017 ; Nassan, Gunn, et al., 2019 ) ( Figure 1 ). QA is produced by spontaneous cyclization of 2-amino-3-carboxymuconate-6-semialdehyde (ACMS), however, ACMS can also be enzymatically processed by amino-β-carboxymuconatesemialdehyde-decarboxylase (ACMSD) to form Picolinic acid (PA), a neuroprotective compound. Once formed, QA then is catabolized to the essential coenzyme and redox cofactor (NAD+) by quinolinate phosphoribosyl transferase (QPRT) ( A. A. B. Badawy, 2017 ). Therefore, ACMSD limits QA formation by competing with its precursor (ACMS) and QPRT is essential for QA clearance. Accordingly, QA homeostasis is determined by the activities of ACMSD and QPRT which control the formation and clearance of QA ( Braidy, Guillemin, & Grant, 2011 ; G. J. Guillemin, 2012 ), respectively. Decreased activity of ACMSD or QPRT can lead to QA accumulation. Whether due to overproduction or impaired clearance, QA accumulation has been implicated in the pathogenesis of several neurological disorders including but not limited to autism ( Lim et al., 2016 ), epilepsy ( Heyes et al., 1994 ), familial cortical myoclonic tremor and epilepsy ( Marti-Masso et al., 2013 ), MS ( Lim et al., 2017 ), ALS ( Chen et al., 2010 ), Alzheimer’s disease ( G. J. Guillemin & Brew, 2002 ; Gilles J. Guillemin et al., 2003 ), Parkinson’s disease ( Chang et al., 2018 ), Huntington disease ( Schwarcz et al., 1988 ), major depressive disorder ( Savitz et al., 2015 ) and suicidality ( Brundin et al., 2016 ; Serafini et al., 2017 ). While the accumulation of QA appears to drive neuro-inflammation and excitotoxicity it is important to reinforce that QA is also a vital intermediate in the de novo synthesis of the critical cofactor NAD. NAD plays a central role in energy metabolism and the synthesis of macromolecules ( Shi et al., 2017 ). It is therefore reasonable to hypothesize that any disruption in the delicate equilibrium of QA synthesis and subsequent metabolism to NAD could have significant, system-wide effects on health. Humphreys and colleagues recently showed that a disruption in NAD synthesis during gestation led to NAD deficiency and subsequent congenital malformations. The authors identified genetic mutations in enzymes of the kynurenine pathwayKP as causes of NAD deficiency ( Shi et al., 2017 ). Our research aimed to determine if inhibition of KP enzymes by environmental toxicants such as phthalates could induce similar disruptions in KP function. The trend toward higher QA levels (after accounting for urinary dilution) in pregnant samples is consistent with published literature indicating enhanced tryptophan catabolism during pregnancy ( A. A. Badawy, 2015 ; Schrocksnadel et al., 2003 ). Changes in tryptophan metabolism are observed in the early stages of pregnancy and directly attributed to indoleamine 2,3-dixoygenase (IDO) expression in trophoblasts and placenta. IDO expression is correlated with placental development and this rate limiting enzyme of the KP plays an essential role in ensuring immune tolerance between the mother and the fetus via immunosuppressive kynurenine metabolites. Increased flux of tryptophan down the KP is also expected to be observed during pregnancy due to the well documented decrease in albumin levels in pregnant women.. Due to the increase in available tryptophan substrate and heightened expression of IDO, increases in levels of KP metabolites are anticipated during gestation. Accelerated catabolism of tryptophan via the KP is essential for maternal and fetal health but there is evidence to suggest that this may only be true if all enzymes of the KP are functioning appropriately. If increased catabolism of tryptophan is accompanied by inhibition of certain KP enzymes, it may lead to the accumulation of neuroactive metabolites and significantly affect fetal outcomes ( A. A. Badawy, 2015 ). For example, QA accumulation has been observed in cases of preeclampsia and toxemia ( Tamura, Okatani, & Sagara, 1990 ; Taniguchi, Okatani, & Sagara, 1994 ). Malik et al. (2014) reported that because of the structural similarity between phthalates and QA, phthalic acid is a potent QPRT inhibitor (QA clearance pathway) ( Malik, Patterson, Ncube, & Toth, 2014 ). Furthermore, recently, Singh et al. (2018) used molecular docking simulations to examine the inhibitory effect of several phthalates and their metabolites on human ACMSD. They reported that because of their structural similarity to tryptophan metabolites, several phthalates and their metabolites exhibited a strong binding affinity with the ACMSD active site and formed stable complexes leading to inhibition of ACMSD activity and thus QA accumulation ( Singh et al., 2018 ). Our results in this analysis are consistent with our previously reported results among men with high DBP exposure from medication coating ( Nassan, Gunn, et al., 2019 ). We reported that high-DBP exposure increased the urinary concentrations of QA, which was largely reversed after removal of the high-DBP exposure for four months. The present study however is the first to examine the association of low (background) phthalate exposures and QA. In addition, we examined this association longitudinally among women who were trying to get pregnant and also while they were pregnant. A recent systematic review reported that prenatal exposure to phthalates was associated with adverse behavioral and cognitive outcomes in children, including lower IQ, and attention problems, hyperactivity, and poorer social communication ( Ejaredar, Nyanza, Ten Eycke, & Dewey, 2015 ). A more recent meta-analysis reported significant associations between prenatal DEHP exposures and psychomotor development outcomes in children ( Lee, Kim, Lim, Lee, & Hong, 2018 ). Because of the growing evidence of the association between phthalate exposure and neurodevelopmental disorders, the association we found between urinary phthalate metabolites and QA warrants further investigation into whether QA could be in the pathway between phthalate exposure and adverse neurodevelopmental outcomes. Our study had several potential limitations, because of the relatively short half-lives of the target biomarkers the collection of spot urine samples and unknown time of last urination could have led to misclassification of phthalate exposure and QA. However, this would likely result in non-differential misclassification attenuating our results. We also acknowledge that we did not have available information about dietary tryptophan intake that could have contributed to QA concentrations. Because the analysis included cross-sectional assessment of exposure and outcome (though repeated measures), causality inference is limited. Similarly, because this was an observational study, residual confounding cannot be excluded. However, our results are consistent with our previously reported prospective data in men where we reported mainly within person change of QA that followed changes of DBP exposure from DBP-containing mesalamine medications ( Nassan, Gunn, et al., 2019 ). Those within person changes over time accounted for observed and unobserved non-time-varying confounding ( Nassan, Gunn, et al., 2019 ). Finally, it is unclear whether these results are generalizable to women from the general population given that our study participants were recruited from a single fertility clinic in Boston, Massachusetts and mostly white, nonsmokers, and highly educated. However, biomarker concentrations for this population were similar to women in of the U.S. (NHANES data) ( CDC, 2019 ). Our study also had several strengths including that this is the first study to investigate the association between background phthalate exposure and QA in women. We included pregnancy and non-pregnancy samples with repeated urine samples from each woman. We also examined eleven phthalate metabolites and two summary measures. Finally, we confirmed this promising novel hypothesis that should warrant new research that may link phthalate exposure to neurological disorders through QA as a plausible biomarker for the neurotoxicity.

Our results suggest that background phthalate exposure is associated with significantly higher urinary concentrations of QA among women, and with stronger associations of low molecular weight phthalates with QA during pregnancy. This novel hypothesis and the results warrant more research investigating the potential link of maternal phthalate exposure with adverse neurodevelopmental outcomes among their offspring.

Quinolinic acid (QA) is an excitotoxin that is produced in the primary pathway of tryptophan degradation i.e., the Kynurenine Pathway (KP) ( A. A. B. Badawy, 2017 ; Nassan, Gunn, Hill, Coull, & Hauser, 2019 ) ( Figure 1 ). Tryptophan is found in most protein-based foods and utilized for protein synthesis and as a serotonin precursor. Although QA formation from dietary tryptophan is essential for the synthesis of essential coenzymes e.g., nicotinamide adenine dinucleotide (NAD+), QA is neuroactive ( A. A. B. Badawy, 2017 ). QA accumulation due to overproduction or impaired clearance has been implicated in the pathogenesis of many neurological disorders that are of increasing prevalence such as autism ( Lim et al., 2016 ), epilepsy ( Heyes, Saito, Devinsky, & Nadi, 1994 ), familial cortical myoclonic tremor and epilepsy ( Marti-Masso et al., 2013 ), multiple sclerosis (MS) ( Lim et al., 2017 ), amyotrophic lateral sclerosis (ALS) ( Chen et al., 2010 ), Alzheimer’s disease ( G. J. Guillemin & Brew, 2002 ; Gilles J. Guillemin et al., 2003 ), Parkinson’s disease ( Chang et al., 2018 ), Huntington disease ( Schwarcz, Okuno, White, Bird, & Whetsell, 1988 ), major depressive disorder ( Savitz et al., 2015 ) and suicidality ( Brundin et al., 2016 ; Serafini et al., 2017 ). Given the increased incidence of neurodevelopmental disorders in children, there is increased interest and concern about the potential role of environmental exposures ( P. Grandjean & Landrigan, 2006 ; Philippe Grandjean & Landrigan, 2014 ). One class of chemicals for which there is concern are ortho-phthalates (hereto referred to as phthalates). Furthering this concern is that exposure to phthalates is ubiquitous in the general population ( CDC, 2019 ) given their wide spread use in many consumer and personal care products ( Braun et al., 2014 ; Nassan et al., 2017 ). The impetus for the current study was the finding in rats that oral administration of diets containing phthalates (di- (2 ethylhexyl) phthalate (DEHP)) led to increased production and excretion of QA ( Fukuwatari, Ohsaki, Fukuoka, Sasaki, & Shibata, 2004 ). In addition, due to the structural similarities, several phthalates are capable of inhibiting two primary enzymes responsible for limiting QA synthesis and clearance ( Singh, Dalal, & Kumar, 2018 ). Inhibition of these enzymes results in increased production and decreased clearance of QA, hence accumulation of QA that can drive neuro-pathologies. Therefore, these studies suggest that it is plausible that phthalates may be associated with neurotoxicity ( Fukuwatari et al., 2004 ) possibly through modulation of the neurotoxic branches of the KP responsible for QA synthesis and clearance. In the first human study that we are aware of, we reported that very high dibutyl phthalate (DBP) exposure from certain mesalamine medication coatings increased urinary QA concentrations among men ( Nassan, Gunn, et al., 2019 ). In addition, this increase was largely reversed after removal of the high-DBP exposure for four months ( Nassan, Gunn, et al., 2019 ). The current study was designed to further this investigation by exploring whether background (low) general population phthalate exposures are associated with QA concentrations.