Background: Children with autism spectrum disorder (ASD) have urinary metabolites suggesting impairments in several pathways, including oxidative stress, inflammation, mitochondrial dysfunction, and gut microbiome alterations. Sulforaphane, a supplement with indirect antioxidant effects that are derived from broccoli sprouts and seeds, was recently shown to lead to improvements in behavior and social responsiveness in children with ASD. We conducted the current open-label study to determine if we could identify changes in urinary metabolites that were associated with clinical improvements with the goal of identifying a potential mechanism of action. Methods: Children and young adults enrolled in a school for children with ASD and related neurodevelopmental disorders were recruited to participate in a 12-week, open-label study of sulforaphane. Fasting urinary metabolites and measures of behavior (Aberrant Behavior Checklist—ABC) and social responsiveness (Social Responsiveness Scale—SRS) were measured at baseline and at the end of the study. Pearson’s correlation coefficient was calculated for the pre- to post-intervention change in each of the two clinical scales (ABS and SRS) versus the change in each metabolite. Results: Fifteen children completed the 12-week study. Mean scores on both symptom measures showed improvements (decreases) over the study period, but only the change in the SRS was significant. The ABC improved − 7.1 points (95% CI − 17.4 to 3.2), and the SRS improved − 9.7 points (95% CI − 18.7 to − 0.8). We identified 77 urinary metabolites that were correlated with changes in symptoms, and they clustered into pathways of oxidative stress, amino acid/gut microbiome, neurotransmitters, hormones, and sphingomyelin metabolism. Conclusions: Urinary metabolomics analysis is a useful tool to identify pathways that may be involved in the mechanism of action of treatments targeting abnormal physiology in ASD. Trial registration: This study was prospectively registered at clinicaltrials.gov (NCT02654743) on January 11, 2016. Keywords: Autism, Metabolomics, Antioxidant, Biomarker * Correspondence: Stephen.Bent@ucsf.edu Department of Psychiatry, University of California, San Francisco, 401 Parnassus, LP-119, San Francisco, CA 94143, USA Department of Epidemiology and Biostatistics, University of California, San Francisco, 401 Parnassus, LP-119, San Francisco, CA 94143, USA Full list of author information is available at the end of the article © The Author(s). 2018 Open Access This article is distributed under the terms of the Creative Commons Attribution 4.0 International License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. The Creative Commons Public Domain Dedication waiver (http://creativecommons.org/publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated. Bent et al. Molecular Autism (2018) 9:35 Page 2 of 12 Background and the SRS . Further, the observed benefits disap- Persons with autism spectrum disorder (ASD) have peared and returned to baseline when the treatment was many physiological abnormalities compared to typically stopped. However, the above study did not examine the developing children. These abnormalities cluster into physiological changes in children with ASD undergoing four areas: oxidative stress, inflammation/immune treatment. dysregulation, mitochondrial dysfunction, and environ- We therefore sought to examine changes in physio- mental toxicant exposure . Prior studies examining logical markers that may underlie beneficial treatment physiological abnormalities have examined biomarkers effects from sulforaphane by analyzing changes in urin- for one particular metabolic pathway (such as cytokines ary metabolites. In an open-label clinical trial of 15 , oxidative stress , or mitochondrial dysfunction school-age children with ASD, we performed metabolo- ), and a variety of others has been proposed . How- mics analyses of urine before and after treatment to ever, recent studies have used metabolomics—the study determine if there are changes in the low molecular of all metabolites present in a cell, organism, tissue, or weight products of metabolism associated with treat- sample (e.g., blood, urine, saliva, feces)—to more broadly ment response, which might explain the mechanism of examine differences in multiple, concurrent physio- action of this phytochemical. logical mechanisms in children with ASD. These meta- bolomics studies comparing children with ASD to typical controls have identified many physiological ab- Methods normalities that also cluster into similar categories of Participants dysfunctional redox status, inflammatory status, and The study was approved by the Committee on Human mitochondrial function as well as metabolomic signals Research at the University of California, San Francisco that suggest abnormalities in the gut microbiome (the (UCSF) on November 5, 2015, and the trial was regis- microbial population of an individual’s gastrointestinal tered at clinicaltrials.gov (NCT02654743) on January 11, tract, which is increasingly recognized as an important 2016, prior to performing any study activities. Partici- contributor to health and disease) [6–16]. Together, all pants completed the study on May 9, 2016, and metabo- of these studies strongly suggest that persons with aut- lomics analyses were completed in September 2016. ism have abnormal physiology that may be intermittent Prior to initiating the study, the UCSF investigators or continuous. developed a relationship with a local, non-public school An exciting therapeutic potential in ASD involves the (Oak Hill School, San Anselmo, CA) that specializes in treatment of these identified physiological abnormalities the education of children and young adults with autism when they are active. Correction of these abnormalities and related neurodevelopmental disorders (ages 5–22, may improve behavior, symptoms, and quality of life in grades K-12). This unique academic-school-parent part- ASD. Preliminary evidence from intervention trials sup- nership was created with the goal of improving overall ports this possibility. For instance, children with autism care and communication between caregivers, clinical who were treated with the antioxidant N-acetylcysteine providers, and teachers. All children/families attending (NAC) for 12 weeks had significant improvements in ir- the school (n = 35 at study initiation) were invited to ritability compared to children treated with placebo . participate in the study through e-mail, informational In a recent randomized controlled trial of methylcobala- flyers, and an evening informational session. min (methyl B12), children with ASD who were treated Children and young adults were eligible to partici- with this methyl donor had significant improvements in pate if they were enrolled in the school, had a formal the Aberrant Behavior Checklist (ABC) compared to diagnosis of autism, reported no prior use of those treated with placebo, and improvements were sulforaphane-containing supplements, were willing to correlated with physiological measures that indicated hold other treatments constant for the 12-week study improved antioxidant status . period, had no major medical problems other than A previous, small, randomized controlled trial exam- ASD, were willing to provide urine samples, and par- ined the efficacy of sulforaphane for the treatment of ents were willing to complete online surveys at speci- children with ASD . Sulforaphane is an indirect anti- fied intervals. ASD was defined as being present if oxidant from broccoli sprout and seed extracts . the child had a diagnosis from a medical professional Sulforaphane has many physiological effects, including trained to diagnose autism or if the student was de- upregulation of cytoprotective enzymes and increases in termined by school staff and the study psychiatrist to detoxification and excretion of highly reactive and po- meet Diagnostic and Statistical Manual of Mental tentially damaging toxicants and free radicals. During Disorders, Fourth Edition (DSM–IV) criteria for ASD. the 18-week trial, children receiving oral sulforaphane Informed consent was obtained from the parent/care- demonstrated marked improvements in both the ABC giver of all study participants. Bent et al. Molecular Autism (2018) 9:35 Page 3 of 12 Intervention (ABC) and changes in social function as measured by the The study was open-label, and all clinicians, parents, and Social Responsiveness Scale (SRS); both scales are com- teachers were aware of the treatment initiation and dur- monly used outcome measures in clinical trials of inter- ation of 12 weeks. All enrolled children were provided ventions in ASD. Parents/caregivers were asked to weight-based dosing of sulforaphane (~ 2.5 μmol glucora- complete both measures at baseline, 4 weeks, and 12 weeks phanin (GR)/lb). Avmacol , a sulforaphane-producing diet- using an online and secure platform. ary supplement, was provided at no cost by Nutramax Urine was collected once during the 2-week window Laboratories Consumer Care, Inc. (Edgewood, MD). Each before treatment initiation and again at the end of the tablet provides a proprietary broccoli seed and broccoli 12-week study period. Parents/caregivers were provided sprout blend (ERS92®) delivering GR and active myrosin- a sterile urine cup and were asked to collect a first ase enzyme and was calculated to deliver at least the same morning urine (fasting since dinner the day before) and dose of sulforaphane as in the prior randomized con- bring it immediately to their clinic visit, which was also trolled trial in autism [19, 21]. In the prior randomized arranged in the morning. Parents/caregivers were asked controlled trial, participants were provided pure sulfo- if the child had a medical illness at the time of urine col- raphane in contrast to the current study, where sulforaph- lection, and no children had a current medical illness. ane was delivered as its precursor (GR) along with a Urine was immediately stored in 2-mL aliquots at − 80 °C. conversion enzyme, myrosinase, which converts GR to During the screening visit, children had a brief physical sulforaphane in the body (providing the advantage of examination including height and weight to guide the greater stability and a longer shelf life). Prior bioavailability proper dosing. Sulforaphane study tablets were provided at studies suggest that the GR plus myrosinase is an excellent the screening visit along with dosing instructions. Parents and efficient delivery method, but there is variability from were required to complete the intake forms online before person to person and the prior bioavailability studies were the study coordinator contacted them to inform them to conducted in adults . At present, there have been no start taking the study supplement on January 21, 2016. dose-efficacy trials, so it is unknown what an “optimal” dose might be. Dosing was adjusted in weight categories Safety assessments as follows: 32–41 kg (6 tablets = 222 μmol GR/day), Parents/caregivers and teachers were advised to report 41–50 kg (7 tablets = 259 μmol GR/day), 50–59 kg any concerns about a new medical problem immediately (8 tablets = 296 μmol GR/day), 59–68 kg (9 tablets = to the study investigators, who were available at all times 333 μmol GR/day), 68–77 kg (10 tablets = 370 μmol to receive reports of possible adverse effects. At the 4- and GR/day), 77–86 kg (12 tablets = 444 μmol GR/day), 12-week online questionnaires, parents/caregivers were 86–95 kg (13 tablets = 481 μmol GR/day), and 95–105 kg asked to report any new medical problems or concerns for (15 tablets = 555 μmol GR/day). Caregivers were advised possible side effects. to administer the tablets once a day in the morning. A sim- ple grinding device (www.carex.com/item/70071/Ultra-Pill- Metabolomic analyses Crusher) was provided to all families so that tablets could All urine samples were sent on dry ice in one batch to be ground and mixed into cold food (yogurt, applesauce, Metabolon (Morrisville, NC). Metabolomics analysis was fruit juice or shakes, etc.), and most families provided conducted at Metabolon as previously described . tablets in this manner. All children completed baseline Briefly, samples were subjected to methanol extraction, measures (described below) and provided urine samples then split into five aliquots for analysis by ultrahigh during the same 2-week screening window, and then all performance/mass spectrometry in the positive (two children commenced treatment on the same day (January methods), negative or polar ion modes. Metabolites were 21, 2016) (batch enrollment) and finished treatment on identified by automated comparison of ion features to a the same day (April 21, 2016). reference library of chemical standards followed by visual inspection for quality control . For statistical Objectives and outcomes analyses and data display, any missing values were The primary goal of the study, defined a priori, was to assumed to be below the limits of detection; these values determine if any observed changes in symptoms of ASD were imputed with the compound minimum (minimum were correlated with changes in urinary metabolites. We value imputation). Data was then normalized by mea- hypothesized that sulforaphane treatment would lead to sured osmolality, which is necessary to reduce the changes in markers of oxidative stress (or other physio- variability in metabolomics analyses due to differing logical abnormalities) and that those changes would cor- urine concentrations . A metabolic pathway for a relate with clinical improvements. given metabolite was assigned based on prior designa- The two primary outcome measures were changes in tions in the literature combined with experience from behavior as measured by the Aberrant Behavior Checklist prior datasets at Metabolon. Bent et al. Molecular Autism (2018) 9:35 Page 4 of 12 Statistical methods Table 1 Characteristics of enrolled participants Summary statistics were used to describe the variables. Category Characteristic Subjects (N = 15) Change in both the clinical variables and the metabolites % N was computed as post-test minus pre-test. Pearson’s Gender Male 80 12 correlation coefficient was estimated and tested to index Female 20 3 the association between the change in each of the two Ethnicity White 60 9 clinical scales and the change in each metabolite. Given Asian/Pacific Islander 7 1 the early-stage nature of the research, we did not adjust No response 33 5 for multiple comparisons because we felt it was more important to risk a type II error than to miss a poten- Age 7–10 13 2 tially important signal by being overly conservative, as 11–14 47 7 has been suggested by prior authors . 15–21 40 6 We defined, a priori, that a correlation cutoff of an Mean age 14.7 absolute value of ≥ 0.6 was of potential clinical relevance, Primary diagnosis Autism spectrum disorder 100 15 because, given our sample size of 15, we would have Comorbidities Intellectual disability 40 6 approximately 80% power to detect a correlation of that size and such a correlation accounts for roughly one Language disorder 27 4 third of the variance. Others have suggested that correla- ADHD 19 3 tions with an absolute value of ≥ 0.6 indicate a moderate Pica 7 1 or higher correlation  (and are hence of greatest Global Development Delay 7 1 interest in pointing to a mechanism of action). Learning disability 7 1 We also examined the number of participants who had Current meds Sertraline 13 2 a clinical response, defined a priori as an improvement of four or more points in the ABC. We compared the Lurasidone 7 1 pre-post changes in the ABC and SRS in the responder Risperidone 13 2 and non-responder groups using the Student’s t test. Birth control 13 2 Zonisamide 7 1 Results ADHD attention deficit hyperactivity disorder Twenty-one of 35 students who were enrolled in the school completed the informed consent process for the We also examined the number of participants who study. Six families dropped out prior to completing any had a “clinical response,” defined as an improvement in post-baseline information for the following reasons: two four or more points in the ABC. Eight participants had a subjects withdrew consent prior to starting for unstated clinical response compared to seven who were classified reasons; two subjects did not like the taste of the supple- as non-responders. Responders exhibited a 21.8-point ment; two parents filled out no surveys. The characteristics decrease (improvement) in total ABC (p < 0.001) and a of the 15 children who provided follow-up information are 20.2-point decrease in SRS (p < 0.001), compared to showninTable 1. All children had a diagnosis of ASD, 80% increases of 10 points in ABC (p = 0.001) and 8 points in were male, the mean age was 14.8, and there were no SRS (p = 0.076) for non-responders. known genetic conditions among study participants. Their Figure 1 demonstrates the marked improvement at baseline ABC hyperactivity and irritability scores (Table 2) 1-month for the ABC and a more gradual improvement are similar to those in the prior, seminal study of risperidone from 1 to 3 months. The SRS demonstrates a slower for behavioral problems in children with autism and “serious initial improvement at 1 month with continued improve- behavioral disturbances,” where baseline hyperactivity and ment to a significant change at 3 months. irritability scores were 31.8 (vs. 29.4 in the current study) Parents were asked to provide subjective descriptions and26.2(vs.25.0in the current study), respectively . of the changes they noted in their children, and seven parents provided responses (Table 3). Changes in symptoms The change in ABC and SRS scores over the study period Correlations between metabolite changes and clinical are shown in Table 2. Mean scores on both measures changes showed improvements (decreases) over the 3-month study The metabolite analysis measured 694 different urinary period, but only the change in the SRS was significant. The metabolites. Table 4 shows all urinary metabolites with ABC improved − 7.1 points (95% CI − 17.4 to 3.2), and the correlations with an absolute value ≥ 0.6 with either the SRS improved − 9.7 points (95% CI − 18.7 to − 0.8). ABC or the SRS. Bent et al. Molecular Autism (2018) 9:35 Page 5 of 12 Table 2 Change in outcome scores over the 12-week study period Outcome measure Adjusted mean scores (95% CI) Change from baseline (95% CI) Baseline 1 month 3 months 1 month p 3 months p Aberrant behavior checklist 103.9 (72.0 to 135.9) 94.2 (63.8 to 124.6) 96.9 (65.7 to 128.0) − 9.7 (− 17.6 to − 1.8) 0.02* − 7.1 (− 17.4 to 3.2) 0.18 total score Hyperactivity 29.4 (21.1 to 37.8) 27.7 (19.0 to 36.4) 28.0 (20.1 to 35.8) − 1.7 (− 4.5 to 1.1) 0.24 − 1.5 (− 5.0 to 2.0) 0.41 Irritability 25.0 (12.9 to 37.1) 22.2 (9.3 to 35.2) 22.8 (10.1 to 35.5) − 2.8 (− 6.1 to 0.6) 0.11 − 2.2 (− 5.8 to 1.3) 0.22 Inappropriateness 5.3 (−0.4 to 10.9) 5.0 (− 0.8 to 10.8) 5.0 (−1.0 to 11.1) − 0.3 (− 0.8 to 0.3) 0.35 − 0.2 (− 1.3 to 0.9) 0.72 Social withdrawal 34.3 (22.8 to 45.7) 30.7 (20.3 to 41.0) 31.2 (19.8 to 42.7) − 3.6 (− 5.8 to − 1.4) 0.001* − 3.0 (− 5.6 to − 0.4) 0.02* Stereotypy 9.9 (6.5 to 13.4) 8.4 (4.9 to 11.8) 9.8 (6.5 to 13.1) − 1.5 (− 2.8 to − 0.3) 0.02* − 0.09 (− 1.8 to 1.6) 0.92 Social responsiveness scale 154.1 (106.6 to 201.7) 147.0 (99.9 to 194.1) 144.4 (95.5 to 193.3) − 7.1 (− 16.9 to 2.6) 0.15 −9.7 (− 18.7 to − 0.8) 0.03* total score Awareness 22.2 (15.7 to 28.7) 22.5 (16.2 to 28.7) 22.7 (15.8 to 29.5) 0.3 (− 0.9 to 1.4) 0.65 0.5 (− 1.2 to 2.1) 0.60 Cognition 26.6 (15.3 to 37.9) 25.4 (12.8 to 38.0) 25.1 (14.0 to 36.3) − 1.2 (− 4.2 to 1.9) 0.44 − 1.5 (− 4.1 to 1.1) 0.27 Communication 55.6 (35.3 to 75.9) 52.7 (32.8 to 72.7) 50.6 (30.5 to 70.8) − 2.9 (− 6.5 to 0.8) 0.13 − 5.0 (− 8.4 to −1.5) 0.005* Mannerisms 24.1 (19.1 to 29.1) 22.7 (16.7 to 28.7) 22.8 (17.3 to 28.3) − 1.5 (− 3.3 to 0.4) 0.12 − 1.3 (− 3.7 to 1.1) 0.28 Motivation 27.1 (17.5 to 36.8) 24.6 (15.5 to 33.6) 24.0 (13.6 to 34.3) − 2.6 (− 4.2 to − 0.9) 0.003* − 3.1 (− 5.1 to − 1.2) 0.001* Mean change scores were adjusted for age and gender *Statistically significant change from baseline Positive correlations indicate that, as the urinary me- (but also associated with pica), one stomach flu, one inflam- tabolite increased, the ABC and SRS also increased mation in the esophagus and weight gain, one ruptured (worsened). Negative correlations indicate that, as the appendix, and one weight loss. The one serious adverse urinary metabolite increased, the ABC and SRS de- event (ruptured appendix) required hospitalization and sur- creased (improved). Metabolites that fall into similar, gery but led to no long-term complications. The committee known metabolic pathways are highlighted and an inter- on human research review determined that it was unlikely pretation is provided in the discussion. to be related to the study supplement. Adverse events Discussion Six families reported that their child had a new medical This study builds upon findings from one prior randomized, event during the study period: one nausea and vomiting placebo-controlled trial, which found that the use of Fig. 1 Change in mean outcome scores over time. a Change in mean aberrant behavior (ABC). b Change in mean social responsiveness (SRS). Mean scores were adjusted for sex and age of subjects. Decreasing score indicates clinical improvement Bent et al. Molecular Autism (2018) 9:35 Page 6 of 12 Table 3 Parental descriptions of behavior change including Nrf2-mediated induction of phase 2 detoxification enzymes, suppression of cytochrome P450 enzymes, induc- Subject Parental description tion of apoptotic pathways, and anti-inflammatory activity 1 Increased conversation, asks questions, makes jokes, “more with us,” decreased vocalizations that have been described in detail . Many prior studies have found increased oxidative 2 Engaging, more lucid, relaxed stress in children with ASD, which may be due to 3 More engaged, increased eye contact, more attentive, calmer increased production or decreased clearance of reactive 4 More flexibility oxygen species (also known as ROS, or oxygen 5 Less repetitive behavior, calmer, more language at school free-radicals). Children with ASD have been found to 6 Energy burst after taking supplement, difficult behavior if have lower levels of the metabolites that process oxygen misses a dose, eager to try things, meltdowns improved free radicals (methionine, S-adenosylmethionine, homo- (fewer, easier to “get out of,” shorter duration) cysteine, cystathionine, cysteine, and total glutathione), 7 Estimated 85% decrease in duration, frequency, and intensity higher levels of metabolites that are involved in the of self-injurious behavior body’s mechanism for reducing oxidative stress (oxidized glutathione, adenosine, and S-adenosylhomocysteine), and sulforaphane led to improvements in behavior and social re- markers of protein and DNA oxidative damage [3, 29]. sponsiveness in children and young adults (aged 13–27) Also, two prior randomized controlled trials—one of the with ASD . Our primary goal was to examine changes antioxidant NAC, and one of the methyl donor, methyl in metabolites in children with ASD who were taking sulfo- B12—have found that these treatments improve clinical raphane to determine a possible mechanism of action. We symptoms in children with ASD [17, 18]. In the methyl observed that a group of school-age children (mean age B12 supplementation study, clinical improvements were 14.8) showed a trend towards improvement in behavior correlated with increases in plasma methionine, decreases (ABC) and a statistically significant improvement in social in S-adenosyl-homocysteine (SAH), and improvements in responsiveness after 12 weeks of treatment. The magnitude the ratio of S-adensylmethionine (SAM) to SAH . of improvements in the current study (− 9.7 points for the Interestingly, in the current study, clinical improvements SRS and − 7.1 points for the ABC) were smaller than in the were correlated with two metabolites known to be in- prior study (− 20.4 points for the SRS and − 21.4 points for volved in redox metabolism. The negative correlations the ABC), which may be related to the younger age of these found with γ-glutamylglutamine and methionine sulfone participants or other differences in the study populations. indicate that, as the urinary levels of these metabolites in- One important difference in the study populations is that in creased, the symptoms scores decreased (improved). This the current study, all of the children were attending suggests that sulforaphane may mediate beneficial clinical one specialized school with programs designed for effects through increases in antioxidant capacity, which is children with ASD. The behavioral interventions in one of its well-documented physiological effects . place at the school may have already been producing Abnormalities in amino acid metabolism have been re- positive effects (at baseline) and limited the ability to ported in children with ASD compared to control children detect further improvement from sulforaphane over the [9, 12, 14, 16], and this may be related to altered processing 12-week study (when compared to the prior study, where of amino acids by gut microbiota . We found correla- subjects were not in the same school). tions between clinical improvement and the amino acids In order to determine which metabolites might mediate tryptophan, tyrosine, and assymetric-dimethylarginine clinical improvements, we examined correlations between (ADMA, a derivative of the amino acid, arginine). The in- the change in urinary metabolite levels and change in behav- volvement of altered amino acids in the pathology of ASD ior and social responsiveness over the 12-week study. We is plausible since amino acids are building blocks for many defined, a priori, that a correlation with an absolute value of key neurotransmitters and hormones, including catechol- 0.6 or greater was potentially relevant and might indicate a amines and serotonin . Six prior studies using urinary mechanism of action. Of 694 measured urinary metabolites, metabolomics noted increased urinary tryptophan in chil- 77 had correlations with an absolute value of ≥ 0.6. While it dren with ASD [7, 9, 12, 14, 16, 30], and tryptophan is a is not possible to discuss the implications of all of these key substrate in the serotonergic metabolic pathway. For correlations, several of these metabolite changes cluster into tryptophan, tyrosine, and ADMA, the correlations in the known pathways (Table 4) that have been reported to be current study were all negative, indicating that as the altered in children with ASD. These pathways involve urinary levels of these amino acids increased, symptom metabolites involved in oxidative stress, amino acid/gut scores decreased (improved). We also identified correla- microbiome, neurotransmitters, hormones/stress response, tions between clinical improvement and changes in a num- and sphingomyelin metabolism, among others. Sulforaphane ber of other amino acids that are known to be associated may affect these pathways through a variety of mechanisms with gut microbiota. Six of the eight metabolites in this Bent et al. Molecular Autism (2018) 9:35 Page 7 of 12 Table 4 Metabolite correlations Primary outcome measure Metabolite ABC SRS Metabolic pathway Corr. p value Corr. p value Amino acids (endogenous) Arginine 0.63 0.03 Amino acid Assymetric dimethylarginine (ADMA) − 0.61 0.06 Arginine catabolite (endogenous inhibitor of nitric oxide synthases) N-delta-acetylornithine 0.62 0.06 Arginine catabolite N-acetylputrescine − 0.65 0.04 Arginine catabolite 4-Acetamidobutanoate − 0.62 0.06 Arginine catabolite Tryptophan − 0.02 Amino acid 0.67 Tyrosine − 0.02 Amino acid 0.67 Theanine 0.7 0.02 a.a. deriv. from tea—of food origin Carnosine 0.6 0.07 a.a. deriv.; dietary; antioxidant; CNS functionality β-hydroxyisovalerate − 0.007 a.a. metabolism (leucine) 0.73 α-hydroxyisocaproate − 0.02 a.a. metabolite (leucine) 0.66 4-Methyl-2-oxopentanoate − 0.62 0.06 a.a. metabolite (leucine) α-hydroxyisovalerate − 0.6 0.04 a.a. metabolism (isoleucine) 2-Methylbutyrylcarnitine (C5) 0.6 0.07 a.a. derived (isoleucine) 3-Hydroxyisobutyrate − 0.72 0.02 a.a. metabolism (valine) N-carbamoylalanine 0.73 0.01 a.a. metabolite (alanine) N-methyltaurine − 0.63 0.05 a.a. metabolite (cysteine) Taurine − 0.6 0.04 a.a. metabolite (cysteine); critical for oxidative stress Dimethylglycine − 0.61 0.06 a.a. metabolite (glycine) 1-Methylguanidine 0.7 0.02 a.a. metabolism N-acetylhistamine 0.69 0.01 a.a. metabolite (histidine) Amino acids (microbiome-associated or contributed) Tyramine O-sulfate 0.61 0.03 a.a. metabolite (tyrosine); catecholamine trigger 3-Indoxyl sulfate 0.63 0.03 a.a. metabolite (tryptophan); likely of gut microbiome origin 2-Oxindole-3-acetate 0.61 0.06 a.a. metabolite (tryptophan); likely of gut microbiome origin Indolin-2-one 0.61 0.03 a.a. metabolite (tryptophan); likely of gut microbiome origin Phenyllactate (PLA) − 0.61 0.06 a.a. metabolite (phenylalanine), contribution from gut microbiome Phenylacetylglutamine 0.65 0.02 a.a. metabolite (glutamine); likely of gut microbiome origin Tryptophan betaine − 0.69 0.03 a.a. (contribution from microbiome) N-acetyl-cadaverine 0.65 0.02 a.a. degradation product (lysine); likely of gut microbiome origin Benzene metabolism Hydroquinone sulfate − 0.66 0.04 Benzene metabolite; likely of exogenous origin Caffeic acid derivatives Chlorogenate 0.61 0.06 Caffeic acid deriv.; dietary/food additive Bent et al. Molecular Autism (2018) 9:35 Page 8 of 12 Table 4 Metabolite correlations (Continued) Primary outcome measure Metabolite ABC SRS Metabolic pathway Corr. p value Corr. p value Cholesterol metabolism Cholesterol − 0.005 Cholesterol 0.75 Cholate − 0.73 0.02 Cholesterol (bile acid) 12-Dehydrocholate − 0.69 0.03 Cholesterol (bile acid), microbiome origin 7-Ketodeoxycholate − 0.6 0.07 Cholesterol (bile acid), microbiome origin Glycocholenate sulfate − < Cholesterol (bile acid), microbiome origin 0.86 0.001 Cortisone − 0.8 0.002 Cholesterol (hormone) Cortisol 21-glucuronide − 0.008 Cholesterol (hormone) 0.73 Epiandrosterone glucuronide − 0.01 Cholesterol (hormone) 0.68 17α-hydroxypregnanolone glucuronide − 0.03 Cholesterol (hormone) 0.61 5α-androstan-3β,17α-diol disulfate − 0.04 Cholesterol (hormone) 0.61 Pregnen-diol disulfate − 0.82 0.004 Cholesterol (hormone) Dehydroepiandrosterone glucuronide − 0.78 0.008 Cholesterol (hormone) 11-Ketoetiocholanolone sulfate − 0.76 0.01 Cholesterol (hormone) 5α-pregnan-3β,20α-diol disulfate − 0.69 0.03 Cholesterol (hormone) 5α-pregnan-3(α/β),20β-diol disulfate − 0.68 0.03 Cholesterol (hormone) 21-Hydroxypregnenolone disulfate − 0.65 0.04 Cholesterol (hormone) 3α,21-dihydroxy-5β-pregnane-11,20-dione 21- − 0.61 0.06 Cholesterol (hormone) glucuronide Androstenediol 3β 17β d0 − 0.6 0.07 Cholesterol (hormone) Fatty acids Diglycerol 0.66 0.02 Lipid, likely of exogenous origin Pimelate (heptanedioate) 0.61 0.03 0.61 0.06 Dicarboxylate fatty acid (DFA) Suberate (octanedioate) 0.75 0.01 Dicarboxylate fatty acid (DFA) Azeloylcarnitine (C9-DC) 0.66 0.02 Acylcarnitine-conjugated DFA Microbial polycyclic aromatic hydrocarbon (PAH) degradation Gentisate (2,5-diOH-benzoate) 0.69 0.03 Key intermediate in microbial PAH degradation 4-Hydroxybenzoate − 0.03 Gut microbiome/amino acid 0.63 3-Ethylphenylsulfate − 0.03 Gut microbiome/amino acid 0.63 Monoterpene phenol Thymol sulfate 0.62 0.03 Monoterpene phenol of food origin Neurotransmitters N-methylglutamate 0.65 0.04 Neurotransmitter Glutamine − 0.003 Neurotransmitter 0.77 Homovanillate (HVA) − 0.03 − 0.65 0.04 Neurotransmitter 0.62 Hypoxanthine − 0.03 − 0.64 0.05 Neurotransmitter 0.62 Bent et al. Molecular Autism (2018) 9:35 Page 9 of 12 Table 4 Metabolite correlations (Continued) Primary outcome measure Metabolite ABC SRS Metabolic pathway Corr. p value Corr. p value Serotonin −0.61 0.03 Neurotransmitter Oxidative stress γ-glutamylglutamine − 0.03 Oxidative stress 0.63 Methionine sulfone − 0.03 Oxidative stress 0.62 Polyol 3-Carboxy-4-methyl-5-propyl-2-furanpropanoate (CMPF) − 0.78 0.008 Polyol associated with uremia Sphingomyelin Palmitoyl sphingomyelin (d18:1/16:0) − 0.01 Sphingomyelin 0.69 Stearoyl sphingomyelin (d18:1/18:0) − 0.004 Sphingomyelin 0.77 Lignoceroyl sphingomyelin (d18:1/24:0) − 0.006 Sphingomyelin 0.74 Behenoyl sphingomyelin (d18:1/22:0) − 0.006 Sphingomyelin 0.74 Sphingomyelin (d18:1/20:1, d18:2/20:0) − 0.004 Sphingomyelin 0.76 Sphingomyelin (d18:1/14:0, d16:1/16:0) − 0.005 Sphingomyelin 0.75 Sphingomyelin (d18:1/20:0, d16:1/22:0) − 0.006 Sphingomyelin 0.74 Sugars Arabinose − 0.7 0.02 Sugar frequently associated with intestinal Candida overgrowth TCA cycle Malate − 0.02 TCA cycle intermediate 0.65 Other Gentisic acid-5-glucoside 0.65 0.04 Xenobiotic (chemical) Erythritol − 0.83 0.003 Xenobiotic (diet-derived) 1-Palmitoyl-2-oleoyl-GPE (16:0/18:1) − 0.008 Phospholipid 0.72 a.a amino acid, PAH polycyclic aromatic hydrocarbon category had positive correlations, indicating that decreased N-methylglutamate, the correlations were negative, indicat- urinary levels were associated with improved clinical symp- ing that increases in the urinary levels of these metabolites toms. It is not yet clear how sulforaphane may affect or were associated with lower scores on the ABC and SRS and improve the amino acid abnormalities (including those therefore improved symptoms. This suggests that either associated with the gut microbiome) and lead to clinical increased production or increased elimination of these me- improvements. tabolites is correlated with beneficial effects. Hypoxanthine, We also found associations between clinical improve- which is part of the purine pathway, was previously found ments and changes in five urinary neurotransmitter-related to be elevated in children with ASD . We also found that metabolites, including N-methylglutamate, glutamine, urinary glutamine was correlated with clinical improve- hypoxanthine, serotonin, and homovanillate (HMV), which ments. Glutamatergic dysfunction has been hypothesized to is the normal end product of dopamine degradation and be involved in the pathogenesis of ASD, with several studies was elevated in the urine of children with ASD in a prior reporting abnormal levels of glutamate in various regions in study . For all of the urinary neurotransmitters except the brain . Bent et al. Molecular Autism (2018) 9:35 Page 10 of 12 We found correlations between clinical improvements measures of repetitive behavior or adaptive function, and a large number of hormones, some of them which might have shown clinical changes in other im- stress-related. Prior studies have found higher salivary portant areas. Many factors affect urinary metabolomics, stress hormones in children with ASD and higher hair including diet, environment, stress, sleep, age, and other cortisol levels, suggesting both acute and chronic elevation factors. In the current study, the wide age range (7–21), in stress hormones [31, 32]. Higher hair cortisol levels different gender and pubertal state of subjects and the were associated with more severe autism symptoms and small sample size all limit the strength of the conclu- anxiety . In the current study, both cortisone and sions regarding urinary metabolomic changes. The vari- cortisol-21-glucuronide had negative correlations with ation in these environmental effects may have been ASD-related behavior, indicating that increased urinary minimized by having all the children attending the same levels were associated with improved symptoms. ASD school 5 days a week. However, the findings of this study behavior and social responsiveness were also related to a should be viewed as hypothesis generating and should large number of other hormones, and all of these correla- be confirmed in future studies with larger sample sizes. tions were negative, again indicating that increased urinary Future studies would also benefit from plasma biochem- levels of various hormones were associated with improved ical assessments of antioxidant status pre- and symptoms. It is not clear if this indicates that increased post-treatment since this is likely a key pathway in the production or increased excretion is associated with im- mechanism of action of sulforaphane. We found that provement, but it highlights changes in hormonal function change in the two outcome measures, ABC and SRS, in ASD as an area for further study. were associated with mostly different metabolites. We A novel finding of the current study is that improve- believe this is to be expected since the measures assess ment in behavior was correlated with seven different different components of human behavior and are likely chemical forms of sphingomyelin. Seven sphingomyelin influenced by different metabolic pathways. Finally, we metabolites were each strongly negatively correlated used a different delivery method of sulforaphane than in with behavior, such that increased urinary levels were a prior randomized controlled trial in autism by associated with improved behavior. Sphingomyelin is a providing a precursor, glucoraphanin, along with a con- sphingolipid found in animal cell membranes, especially version enzyme, myrosinase. Although the dosing was in the membranous myelin sheath that surrounds nerve designed to produce a similar level of sulforaphane, it is cells and axons. To our knowledge, there have been no possible that differences in bioavailable sulforaphane prior reports of abnormalities in sphingolipid levels in levels between the two studies could have led to differ- children with ASD, but there are numerous studies ences in clinical results. documenting abnormalities in the size, number, and morphology of dendrites in autism, which is related to altered synapse function . Furthermore, sphingomye- Conclusions lin abnormalities have been noted in a number of other We are not aware of any prior studies that have exam- central nervous system disorders, including depression, ined changes in urinary metabolites during a clinical trial anxiety, Alzheimer’s disease, and amyotrophic lateral of a treatment for autism. In this pilot study, we demon- sclerosis, suggesting that it may have a central role in strated the feasibility of using metabolomics to identify normal brain development and function [34–37]. It is urinary metabolites that are correlated with clinical im- not clear how sulforaphane might alter sphingomyelin provements and might therefore represent a mechanism metabolism or availability and whether this is related to of action. One group of metabolites in particular, the clinical benefits, but if this association is confirmed, it sphingolipid/sphingomyelin group, was highlighted as has important clinical and treatment implications. being significantly associated with improvement. Other The current study has a number of limitations. Im- urinary metabolite changes that were correlated with portantly, this was a pilot study to investigate whether clinical improvement are related to oxidative stress, metabolomics might be a useful tool to suggest pathways amino acid metabolism/gut microbiome metabolites, that may be involved in the mechanism of action of neurotransmitters, stress, and other hormones. These treatments (in this case, sulforaphane) in ASD. The findings suggest that urinary metabolomics may be a study was open label and parent raters may have rated tool to identify important changes in the quest to deter- more positively knowing that their child was taking the mine how certain biological interventions work to re- sulforaphane, although the magnitude of benefit is lower duce specific symptoms of ASD. This may further than in the one prior randomized controlled trial. The provide clues to the underlying, active pathophysiology ratings were also limited to the ABC and SRS to of ASD and may allow for more precise interventions minimize respondent burden, and while these are widely targeted to the unique metabolome of each individual used outcome tools in ASD, we did not include other with this disorder. Bent et al. Molecular Autism (2018) 9:35 Page 11 of 12 Abbreviations Suite 400, Durham, NC 27713, USA. Department of Medicine, UCSF, ABC: Aberrant Behavior Checklist; ADMA: Assymetric dimethylarginine; SFVAMC, 111-A1, 4150 Clement St, San Francisco, CA 94121, USA. ASD: Autism spectrum disorder; HMV: Homovanillate; NAC: N-acetylcysteine; SAH: S-adenosyl-homocysteine; SAM: S-adensylmethionine; SRS: Social Received: 26 June 2017 Accepted: 22 May 2018 Responsiveness Scale; UCSF: University of California, San Francisco Acknowledgements We wish to express our deep appreciation for the participating families, who References graciously volunteered their time and energy to advance research. 1. Rossignol DA, Frye RE. A review of research trends in physiological Sulforaphane study supplement was provided at no cost by Nutramax abnormalities in autism spectrum disorders: immune dysregulation, Laboratories Consumer Care, Inc. (Edgewood, MD). inflammation, oxidative stress, mitochondrial dysfunction and environmental toxicant exposures. Mol Psychiatry. 2012;17(4):389–401. 2. Ashwood P, Wills S, Van de Water J. The immune response in autism: a new Funding frontier for autism research. J Leukoc Biol. 2006;80(1):1–15. This work was supported through funding from the JS Foundation. The JS Foundation had no role in the design of the study, collection, analysis, and 3. James SJ, Cutler P, Melnyk S, Jernigan S, Janak L, Gaylor DW, Neubrander JA. interpretation of data, nor in the writing the manuscript. Metabolic biomarkers of increased oxidative stress and impaired methylation capacity in children with autism. Am J Clin Nutr. 2004;80(6):1611–7. 4. Rossignol DA, Bradstreet JJ. Evidence of mitochondrial dysfunction in autism Availability of data and materials and implications for treatment. Am J Biochem Biotechnol. 2008;4(2):208–17. The datasets used and/or analyzed during the current study are available 5. Liu H, Talalay P, Fahey JW. Biomarker-guided strategy for treatment of from the corresponding author on reasonable request. autism spectrum disorder (ASD). CNS Neurol Disord Drug Targets. 2016; 15(5):602–13. Authors’ contributions 6. National Research Council. Educating Children with Autism. Committee on SB co-conceived the study, wrote the protocol, supervised all study activities Educational Interventions for Children with Autism. Catherine Lord and including data acquisition, analysis, and interpretation, and wrote the manuscript. James P. McGee, eds. Division of Behavioral and Social Sciences and BL, TW, and FW gave input into the protocol design, performed patient assessments, Education. Washington, DC: National Academy Press; 2001. interacted with families and educators, assisted with online database design, 7. Dieme B, Mavel S, Blasco H, Tripi G, Bonnet-Brilhault F, Malvy J, Bocca C, participated in the data analysis and interpretation, and wrote sections of the Andres CR, Nadal-Desbarats L, Emond P. Metabolomics study of urine in manuscript. KD (1st) and KD (2nd) conducted the analyses and interpreted the data autism spectrum disorders using a multiplatform analytical methodology. J and wrote sections of the manuscript related to data analysis. JWF and BC provided Proteome Res. 2015;14(12):5273–82. expert input regarding sulforaphane dosing, bioavailability, safety, and monitoring 8. Emond P, Mavel S, Aidoud N, Nadal-Desbarats L, Montigny F, Bonnet- and provided interpretation of metabolomic data and correlations with clinical Brilhault F, Barthelemy C, Merten M, Sarda P, Laumonnier F, et al. GC-MS- parameters. JMK conducted analyses of metabolomic data and interpretation of based urine metabolic profiling of autism spectrum disorders. Anal Bioanal pathways related to metabolomics and wrote sections of the manuscript related to Chem. 2013;405(15):5291–300. metabolomic methods and results. RLH co-conceived the study, assisted with the 9. Gevi F, Zolla L, Gabriele S, Persico AM. Urinary metabolomics of young supervision of daily study activities, was the primary clinician responsible for assessing Italian autistic children supports abnormal tryptophan and purine patient eligibility and adverse events, participated in the data interpretation, and metabolism. Mol Autism. 2016;7:47. wrote sections of the manuscript. All authors approved the final version of the 10. Kuwabara H, Yamasue H, Koike S, Inoue H, Kawakubo Y, Kuroda M, Takano manuscript. Y, Iwashiro N, Natsubori T, Aoki Y, et al. Altered metabolites in the plasma of autism spectrum disorder: a capillary electrophoresis time-of-flight mass Ethics approval and consent to participate spectroscopy study. PLoS One. 2013;8(9):e73814. The study was approved by the Committee on Human Research at the 11. Mavel S, Nadal-Desbarats L, Blasco H, Bonnet-Brilhault F, Barthelemy C, University of California, San Francisco (UCSF) on November 5, 2015. Informed Montigny F, Sarda P, Laumonnier F, Vourc’h P, Andres CR, et al. 1H-13C consent was obtained from the parent/caregiver of all study participants. NMR-based urine metabolic profiling in autism spectrum disorders. Talanta. 2013;114:95–102. Competing interests 12. Ming X, Stein TP, Barnes V, Rhodes N, Guo L. Metabolic perturbance in RLH reports research grants from Curemark, BioMarin, Roche, Shire, Sunovion, autism spectrum disorders: a metabolomics study. J Proteome Res. 2012; Vitamin D Council and Advisory Boards for Curemark, BioMarin, Neuren, and 11(12):5856–62. Janssen. None of these grants or companies are directly involved in the 13. Nadal-Desbarats L, Aidoud N, Emond P, Blasco H, Filipiak I, Sarda P, Bonnet- subject of the current study. JWF was a co-author on the first randomized Brilhault F, Mavel S, Andres CR. Combined 1H-NMR and 1H-13C HSQC-NMR controlled trial examining the efficacy of sulforaphane and manufactured the to improve urinary screening in autism spectrum disorders. Analyst. 2014; product used in that study, which found a beneficial effect. BC is the Medical 139(13):3460–8. Director and Director of Consumer Product Support for Nutramax Laboratories 14. Noto A, Fanos V, Barberini L, Grapov D, Fattuoni C, Zaffanello M, Casanova Consumer Care, Inc., which supplied the sulforaphane supplement and matching A, Fenu G, De Giacomo A, De Angelis M, et al. The urinary metabolomics placebo for this study at no cost. JMK is a senior scientist for Metabolon, which profile of an Italian autistic children population and their unaffected performed the metabolomic analyses for this study under a paid contract. All siblings. J Matern Fetal Neonatal Med. 2014;27(Suppl 2):46–52. other authors declare that they have no conflict of interest. 15. West PR, Amaral DG, Bais P, Smith AM, Egnash LA, Ross ME, Palmer JA, Fontaine BR, Conard KR, Corbett BA, et al. Metabolomics as a tool for discovery of biomarkers of autism spectrum disorder in the blood plasma of Publisher’sNote children. PLoS One. 2014;9(11):e112445. Springer Nature remains neutral with regard to jurisdictional claims in 16. Yap IK, Angley M, Veselkov KA, Holmes E, Lindon JC, Nicholson JK. published maps and institutional affiliations. Urinary metabolic phenotyping differentiates children with autism from their unaffected siblings and age-matched controls. J Proteome Res. Author details Department of Psychiatry, University of California, San Francisco, 401 2010;9(6):2996–3004. Parnassus, LP-119, San Francisco, CA 94143, USA. Department of 17. Hardan AY, Fung LK, Libove RA, Obukhanych TV, Nair S, Herzenberg LA, Epidemiology and Biostatistics, University of California, San Francisco, 401 Frazier TW, Tirouvanziam R. A randomized controlled pilot trial of Parnassus, LP-119, San Francisco, CA 94143, USA. Departments of Medicine, oral N-acetylcysteine in children with autism. Biol Psychiatry. 2012; Pharmacology and Molecular Sciences, International Health, and Cullman 71(11):956–61. Chemoprotection Center, Johns Hopkins University, 855 N. Wolfe St. Ste. 625, 18. Hendren RL, James SJ, Widjaja F, Lawton B, Rosenblatt A, Bent S. Baltimore, MD 21205, USA. Nutramax Laboratories Consumer Care, Inc, 2208 Randomized, placebo-controlled trial of methyl B12 for children with Lakeside Blvd, Edgewood, MD 21040, USA. Metabolon, Inc, 617 Davis Dr. autism. J Child Adolesc Psychopharmacol. 2016;26(9):774–83. Bent et al. Molecular Autism (2018) 9:35 Page 12 of 12 19. Singh K, Connors SL, Macklin EA, Smith KD, Fahey JW, Talalay P, Zimmerman AW. Sulforaphane treatment of autism spectrum disorder (ASD). Proc Natl Acad Sci U S A. 2014;111(43):15550–5. 20. Fahey JW, Talalay P. Antioxidant functions of sulforaphane: a potent inducer of phase II detoxication enzymes. Food Chem Toxicol. 1999;37(9–10):973–9. 21. Fahey JW, Holtzclaw WD, Wehage SL, Wade KL, Stephenson KK, Talalay P. Sulforaphane bioavailability from glucoraphanin-rich broccoli: control by active endogenous myrosinase. PLoS One. 2015;10(11):e0140963. 22. Long T, Hicks M, Yu HC, Biggs WH, Kirkness EF, Menni C, Zierer J, Small KS, Mangino M, Messier H, et al. Whole-genome sequencing identifies common-to-rare variants associated with human blood metabolites. Nat Genet. 2017;49(4):568–78. 23. Dehaven CD, Evans AM, Dai H, Lawton KA. Organization of GC/MS and LC/ MS metabolomics data into chemical libraries. J Cheminform. 2010;2(1):9. 24. Chetwynd AJ, Abdul-Sada A, Holt SG, Hill EM. Use of a pre-analysis osmolality normalisation method to correct for variable urine concentrations and for improved metabolomic analyses. J Chromatogr A. 2016;1431:103–10. 25. Rothman KJ. No adjustments are needed for multiple comparisons. Epidemiology. 1990;1(1):43–6. 26. Mukaka MM. Statistics corner: a guide to appropriate use of correlation coefficient in medical research. Malawi Med J. 2012;24:69–71. 27. McCracken JT, McGough J, Shah B, Cronin P, Hong D, Aman MG, Arnold LE, Lindsay R, Nash P, Hollway J, et al. Risperidone in children with autism and serious behavioral problems. N Engl J Med. 2002;347(5):314–21. 28. Juge N, Mithen RF, Traka M. Molecular basis for chemoprevention by sulforaphane: a comprehensive review. Cell Mol Life Sci. 2007;64(9):1105–27. 29. Melnyk S, Fuchs GJ, Schulz E, Lopez M, Kahler SG, Fussell JJ, Bellando J, Pavliv O, Rose S, Seidel L, et al. Metabolic imbalance associated with methylation dysregulation and oxidative damage in children with autism. J Autism Dev Disord. 2012;42(3):367–77. 30. Lussu M, Noto A, Masili A, Rinaldi AC, Dessi A, De Angelis M, De Giacomo A, Fanos V, Atzori L, Francavilla R. The urinary 1 H-NMR metabolomics profile of an Italian autistic children population and their unaffected siblings. Autism Res. 2017;10(6):1058–66. 31. Ogawa S, Lee YA, Yamaguchi Y, Shibata Y, Goto Y. Associations of acute and chronic stress hormones with cognitive functions in autism spectrum disorder. Neuroscience. 2017;343:229–39. 32. Putnam SK, Lopata C, Thomeer ML, Voker MA, Rodgers JD. Salivary cortisol levels and diurnal patterns in children with autism spectrum disorder. J Dev Physic Disabil. 2015;27:453–65. 33. Martinez-Cerdeno V. Dendrite and spine modifications in autism and related neurodevelopmental disorders in patients and animal models. Dev Neurobiol. 2017;77(4):393–404. 34. Berkecz R, Tomosi F, Kormoczi T, Szegedi V, Horvath J, Janaky T. Comprehensive phospholipid and sphingomyelin profiling of different brain regions in mouse model of anxiety disorder using online two-dimensional (HILIC/RP)-LC/MS method. J Pharm Biomed Anal. 2017;149:308–17. 35. Blasco H, Veyrat-Durebex C, Bocca C, Patin F, Vourc’h P, Kouassi Nzoughet J, Lenaers G, Andres CR, Simard G, Corcia P, et al. Lipidomics reveals cerebrospinal-fluid signatures of ALS. Sci Rep. 2017;7(1):17652. 36. Dinoff A, Saleem M, Herrmann N, Mielke MM, Oh PI, Venkata SLV, Haughey NJ, Lanctot KL. Plasma sphingolipids and depressive symptoms in coronary artery disease. Brain and behavior. 2017;7(11):e00836. 37. Kosicek M, Zetterberg H, Andreasen N, Peter-Katalinic J, Hecimovic S. Elevated cerebrospinal fluid sphingomyelin levels in prodromal Alzheimer’s disease. Neurosci Lett. 2012;516(2):302–5.
– Springer Journals
Published: May 30, 2018