Plasma polyunsaturated fatty acids and amino acids in children with attention-deficit hyperactivity disorder: correlations with clinical and cognitive features Authors: Liuyan Zhu, Lei Wang, Tianmiao Gu, Weijun Chen Published: 2026 ABSTRACT BackgroundAttention-deficit hyperactivity disorder (ADHD) is a common neurodevelopmental disorder in children, characterized by inattention, hyperactivity and impulsivity. Metabolic profiles of polyunsaturated fatty acids (PUFAs) and amino acids have been linked to neurodevelopmental processes associated with ADHD. This cross-sectional study aimed to compare plasma levels of PUFAs and amino acids between medication-naive children with ADHD and typically developing (TD) children, after adjusting for multiple confounding factors, and explore their correlations with clinical symptoms and cognitive function.MethodsA total of 102 medication-naive children with ADHD and 91 TD children were enrolled. Plasma concentrations of four PUFAs and seven neuroactive amino acids were measured using liquid chromatography-tandem mass spectrometry. Group comparisons were performed using analysis of covariance adjusted for age, gender, body mass index (BMI), paternal education level, maternal education level, family annual income, only child status, and main caregiver of children. Correlation analyses and multiple testing corrections were also conducted.ResultsAfter adjustment for confounding factors, ADHD group had higher plasma glutamate (Glu) (median: 103.08 vs. 102.25 µmol/L, P=0.043). Tryptophan (Try) was positively correlated with the Swanson, Nolan, and Pelham rating scale version IV (SNAP-IV) inattention scores (r=0.209, P=0.03). γ-aminobutyric acid (GABA) was negatively correlated with full scale intelligence quotient (IQ) (r=−0.198, P=0.04), and eicosapentaenoic acid (EPA) was negatively correlated with Wechsler Intelligence Scale for Children, Fourth Edition (WISC-IV) processing speed (r=−0.212, P<0.05). Glu and arachidonic acid (ARA) levels differed significantly across groups defined by hyperactivity-impulsivity severity (H=6.695, P=0.04; H=8.616, P=0.01). Boys had higher WISC-IV verbal comprehension (P=0.04).ConclusionsThis study provides exploratory evidence of weak but significant correlations among plasma PUFAs, amino acids, clinical symptoms, and cognitive function in medication-naive children with ADHD. These preliminary findings support further investigation of peripheral metabolic profiles as potential phenotypic correlations rather than established biomarkers for ADHD. Highlight box Key findings • Plasma glutamate (Glu) was significantly increased in medication-naive children with attention-deficit hyperactivity disorder (ADHD) after adjusting for confounders. • Tryptophan (Try) positively correlated with inattention symptoms, while γ-aminobutyric acid and eicosapentaenoic acid were negatively correlated with full scale intelligence quotient and processing speed. • Glu and arachidonic acid (ARA) levels changed along with hyperactivity-impulsivity severity. • Plasma amino acids and polyunsaturated fatty acids (PUFAs) show weak yet significant correlations with clinical and cognitive features of ADHD. What is known and what is new? • ADHD relates to abnormal neural metabolism. PUFAs and amino acids affect its progression, yet data for medication-naive children remain limited. • We found elevated plasma Glu in ADHD children. Multiple metabolites correlate with symptoms and cognition. Glu and ARA differed by hyperactivity severity. Metabolic interactions and gender differences were also observed. What is the implication, and what should change now? • These metabolite-phenotype links suggest unique metabolic correlates for ADHD cognitive and behavioral deficits. • Clinicians may combine plasma metabolic indicators with auxiliary ADHD assessment. Glu and ARA can help stratify hyperactivity-impulsivity severity, while Try aids evaluation of inattention symptoms. Wechsler Intelligence Scale for Children, Fourth Edition verbal scores should be interpreted with gender differences in mind. • The associations between PUFAs, amino acids and ADHD phenotypes may provide preliminary evidence for personalized dietary intervention targeting fatty acid balance and Try metabolism. • Given the cross-sectional design and weak correlation magnitudes, these metabolites cannot serve as standalone diagnostic biomarkers. Multi-center longitudinal studies with dietary recording and ADHD subtype grouping are needed to validate and translate these metabolic correlates into clinical practice. Introduction Attention-deficit hyperactivity disorder (ADHD) is a common childhood neurodevelopmental disorder with a global prevalence of 2–7% (1). It is marked by sustained inattention, hyperactivity and impulsivity, severely impairing children’s study, social contact and life quality (2). Its exact cause is unclear, jointly affected by biological, psychological and social factors. Current ADHD diagnosis is primarily based on subjective behavioral evaluations, which is vulnerable to misdiagnosis and missed cases. Objective biomarkers and neurocognitive assessments display variable diagnostic effects but lack sufficient evidence for independent clinical diagnosis, calling for more systematic verification research (3). Nutrients like polyunsaturated fatty acids (PUFAs) and minerals closely relate to neural development. Research shows that vitamin D and docosahexaenoic acid (DHA, C22:6, n-3) supplements combined with routine drugs can relieve ADHD symptoms in children (4). PUFAs fall into ω-3 and ω-6 categories, which cannot be endogenously synthesized in humans and must be obtained from dietary essential fatty acids (EFAs): linoleic acid (LA, n-6) and alpha-linolenic acid (ALA, n-3). Sequential metabolism mediated by desaturase and elongase enzymes drives the conversion of these two essential precursors into long-chain PUFAs (LC-PUFAs) (5). For the n-6 series, LA undergoes stepwise desaturation and carbon chain elongation to generate gamma-linolenic acid and arachidonic acid (ARA, C20:4 n-6) (5). The ω-3 family originates from alpha-linolenic acid, which is further processed by desaturases and elongases to produce eicosapentaenoic acid (EPA, C20:5 n-3) and DHA (6), mainly acting on nervous and cardiovascular systems. EPA exerts neuroprotective effects by reducing pro-inflammatory cytokines, preserving blood-brain barrier integrity, and supporting synaptic transmission (7). Meta-analyses have confirmed modest efficacy of high-dose EPA supplementation in improving ADHD symptoms, particularly when baseline EPA levels are low (8). DHA variations affect neurotransmitter levels, closely linked to ADHD pathogenesis (9,10). As key n-6 lipid mediators, ARA and its derivatives modulate inflammation and sleep-related substances (11). Consistent evidence from preclinical animal experiments, human clinical trials and adult ADHD proteomic research demonstrates that inflammatory signaling pathways are implicated in the pathological mechanisms of ADHD (12,13). Review indicates that diet can modulate the levels of Omega-3 and Omega-6 PUFAs in the human body and influence ADHD through the gut-brain inflammatory pathway. A Western dietary pattern is associated with reduced blood concentrations of EPA and DHA, alongside an increase in pro-inflammatory n-6 ratios, exacerbating inflammation and symptoms of hyperactivity and attention deficits. In contrast, a Mediterranean diet has been shown to elevate anti-inflammatory Omega-3 levels, improve microbiota composition, and ameliorate inflammatory states. There is a prevalent issue of low Omega-3 intake and serum levels among individuals with ADHD. The efficacy of PUFA supplementation is contingent upon baseline concentrations and dosage, higher doses of EPA supplementation are more beneficial for alleviating ADHD symptoms in populations with low EPA levels. Furthermore, an increased n-6/n-3 ratio is positively correlated with the severity of ADHD symptoms (14). Wang et al. utilizing data from the National Cross-Sectional Study of Children and Adolescents in the United States, confirmed a linear positive correlation between dietary n-6/n-3 PUFA intake ratios and the risk of ADHD. The high consumption of LA-rich vegetable oils prevalent in Western diets elevates the proportion of dietary n-6, which, through competitive inhibition of metabolic enzymes, suppresses the endogenous synthesis of EPA and DHA, consequently increasing the likelihood of developing ADHD. Conversely, the intake of n-3 rich deep-sea fish can help balance the fatty acid ratios and improve the composition of blood PUFAs, thereby reducing the risk of ADHD. This association remained robust even after adjusting for various confounding factors; however, it is important to note that this cross-sectional study can only reveal correlations and cannot establish causation (15). Adults with ADHD exhibit dysfunction within fronto-striatal circuits, they fail to mount sufficient task-evoked increases in glutamate + glutamine (Glx) during attention control tasks, which blunts excitatory glutamatergic signaling in this circuitry (16). γ-aminobutyric acid (GABA) maintains neural excitatory-inhibitory balance, and its disorder aggravates attention impairment (17). Skalny et al. (18) found distinct serum amino acid differences between ADHD and typically developing (TD) children. ADHD children have lower glutamine (Gln), proline (Pro) and histidine, higher glutamate (Glu) and hydroxyproline (Hypro), with declined Gln/Glu and Pro/Hypro ratios. These amino acid levels correlate significantly with ADHD clinical scores. Lipid and amino acid metabolism are found to be associated with ADHD pathophysiology. Nonetheless, few studies have systematically explored correlations between PUFA and neuroactive amino acid profiles and neuropsychological phenotypes in treatment-naive children with ADHD. This study performed multivariable adjusted analyses to eliminate the influence of confounders. It compares plasma levels of PUFAs and seven neuroactive amino acids between untreated ADHD children and TD children, and analyzes the associations of these metabolic characteristics with ADHD symptoms and Wechsler Intelligence Scale scores. The study seeks to further explore metabolic factors related to ADHD progression. We propose the following hypotheses, plasma levels of PUFAs and neuroactive amino acids differ between medication-naive children with ADHD and TD children, after controlling for confounders. Plasma PUFA and amino acid levels are significantly correlated with ADHD symptoms and cognitive function. ARA and Glu levels vary with the severity of hyperactivity-impulsivity symptoms in children with ADHD. We present this article in accordance with the STROBE reporting checklist (available at https://tp.amegroups.com/article/view/10.21037/tp-2026-0590/rc). Methods Participants This study included a total of 102 children diagnosed with ADHD, who were initially diagnosed at the outpatient clinic of the department of pediatric health care at the hospital from January to December 2024. The study comprised 74 boys and 28 girls. This study was conducted in accordance with the Declaration of Helsinki and its subsequent amendments. The study was approved by the Medical Ethics Committee of Children’s Hospital, Zhejiang University School of Medicine (approval No. 2024-IRB-0391-P-01), and written informed consent was obtained from the guardians of all participating children. Inclusion criteria All participants met the diagnostic criteria for ADHD as specified in the fifth edition of the Diagnostic and Statistical Manual of Mental Disorders (DSM-5), verified by two attending pediatric psychiatrists. Participants had not received any ADHD-targeted psychostimulant or non-stimulant medications (e.g., methylphenidate or atomoxetine) or nutritional supplements (such as omega-3 fatty acids or amino acid formulations, or other micronutrient additives) within 3 months prior to enrollment in the study. The full scale intelligence quotient (FSIQ), assessed using the Wechsler Intelligence Scale for Children, Fourth Edition (WISC-IV), was required to be greater than 80 points. Participants were aged between 6 and 15 years. Exclusion criteria Children presenting with neurological or developmental disorders, including but not limited to epilepsy, autoimmune encephalitis, cerebral palsy, tic disorders, or leukodystrophy. Those with significant physical illnesses such as hearing impairment, visual impairment, thyroid dysfunction, or obstructive sleep apnea. Children diagnosed with other psychiatric conditions, including schizophrenia, depression, or pediatric bipolar disorder. A documented history of severe brain injury. Children with genetic disorders, including Fragile X syndrome, Turner syndrome, or neurofibromatosis. Any perinatal complications, such as a gestational age of less than 32 weeks, a birth weight below 1,500 g, or an Apgar score of less than 7 at one minute after birth. Additionally, a control group consisting of 91 non-ADHD children undergoing routine health examinations at the same hospital was established, this group included 45 boys and 46 girls. Procedure Clinical assessment of indicators The study employed the Chinese version of the Swanson, Nolan, and Pelham rating scale version IV (SNAP-IV) for ADHD, developed by Professor Shufen Gao from the National Taiwan University College of Medicine (19). This scale contains 26 items, each scored from 0 to 3, with higher scores indicating more severe symptoms. It assesses three domains: inattention (IN, items 1–9), hyperactivity impulsivity (HI, items 10–18), and oppositional defiance (OD, items 19–26). Parents completed the parent version of the scale independently at the outpatient clinic following standardized instructions provided by the research team. The Cronbach’s α coefficient for the complete parent version was 0.95, while the coefficients for the three sub-factors were 0.90, 0.89, and 0.88, respectively. The intraclass correlation coefficients for test-retest reliability ranged from 0.68 to 0.76, which indicates robust psychometric properties. Intelligence assessments were carried out using the WISC-IV (20). This assessment includes four core indices: verbal comprehension index (VCI), fluid reasoning index (FRI), working memory index (WMI), and processing speed index (PSI), along with the FSIQ. All evaluations were performed by licensed psychological assessors who had received official training in WISC-IV administration. Assessments occurred in a quiet, distraction-free environment and lasted approximately 45 to 60 minutes. The evaluation process strictly followed the standardized procedures specified in the WISC-IV manual, and all results were subsequently reviewed and validated by an additional experienced assessor to ensure the accuracy of the data. Blood biochemical analysis All participants provided blood samples between 6:30 and 8:00 AM while fasting. A trained nurse collected 5 mL of venous blood from the antecubital vein into ethylenediaminetetraacetic acid anticoagulant tubes. The samples were transported to the laboratory within 30 minutes at a temperature of 4 ℃, where they were centrifuged at 3,000 rpm (with a radius of 10 cm) for 10 minutes. The upper plasma layer was carefully extracted and aliquoted into enzyme-free eppendorf tubes (200 µL each), which were stored at −80 ℃ in an ultra-low temperature freezer to avoid repeated freeze-thaw cycles. All samples were subsequently analyzed simultaneously after collection was complete. The concentrations of four PUFAs (LA, DHA, ARA, EPA) and seven amino acids (GABA, Glu, glycine (Gly), Serine (Ser), taurine (Tau), tyrosine (Tyr), tryptophan (Try) in the plasma were conducted on the API 4500 liquid chromatography-tandem mass spectrometry (LC-MS/MS) system with fully validated and standardized protocols to ensure reproducibility. These seven amino acids are built into our hospital’s routine clinical Neurotrophic Metabolite LC-MS/MS assay owing to their critical neuroactive functions in neurotransmitter synthesis, synaptic transmission, and the maintenance of neuronal excitatory-inhibitory balance. The four quantified PUFAs correspond to the primary long-chain n-3 and n-6 species closely linked to brain neurodevelopment and peripheral inflammatory regulation. Importantly, these 11 analytes form the full fixed detection panel of this clinical test; no additional metabolites can be incorporated into the assay workflow. To eliminate selective reporting bias, we comprehensively analyzed and reported every metabolite provided by this standardized panel rather than filtering or excluding any measurable analytes. Control of confounding factors To minimize bias from nutritional and lifestyle factors, children with any use of omega-3 fatty acid supplements, amino acid formulas, or other nutritional additives within 3 months before enrollment were excluded. Age, gender, body mass index (BMI), paternal education level, maternal education level, and annual family income, only child status, and main caregiver of children were adjusted as key confounding covariates in statistical analyses. Notably, only pre-study nutritional supplement intake was restricted in this cross-sectional design. Quantitative dietary records assessing habitual total energy, total fat, protein, as well as n-3 and n-6 PUFA intake were not collected, and these dietary variables could not be incorporated into multivariate adjustment models. Statistical analysis Data analysis was conducted using SPSS version 26.0. The Kolmogorov-Smirnov test was applied to evaluate the normality of continuous variables. Normal distributed data were presented as mean ± standard deviation, and between-group comparisons were performed using independent t-tests. Non-normally distributed data were presented as median (interquartile range), between-group comparisons were performed using Mann-Whitney U test. Categorical variables were presented as n (%) and compared using χ2 test. Bivariate tests (t-test, Mann-Whitney U test, χ2 test) were used for exploratory analysis only. To address potential confounding effects, the primary between-group comparisons of metabolites were performed using analysis of covariance (ANCOVA), with group (ADHD/control) as the fixed factor, and the following eight variables as covariates: age, gender, BMI, paternal education level, maternal education level, annual family income, only child status, and main caregiver of children. For comparisons among multiple groups, one-way ANOVA was performed with post hoc analyses conducted via the least significant difference t-test (LSD-t). For non-normally distributed data, values are expressed as medians (interquartile range, Q25, Q75) and analyzed using the Mann-Whitney U test for two-group comparisons and the Kruskal-Wallis H test for multiple group comparisons. Pearson correlation analysis was utilized to investigate relationships among variables. The false discovery rate (FDR) method was used to correct for multiple comparisons to control Type I error. A corrected P value of less than 0.05 was deemed statistically significant. Results Sample demographics In this study, a total of 102 children with ADHD were included, comprising 74 boys (72.5%) and 28 girls (27.5%). Additionally, 91 control children were included, consisting of 45 boys (49.4%) and 46 girls (50.6%). The comparison results between the groups showed that the age of the ADHD group was significantly lower than that of the control group (P=0.001), and the proportion of boys was significantly higher in the ADHD group than in the control group (P=0.002), both of which were statistically significant. There were no statistically significant differences in the distributions of BMI, parental education level, annual family income, only child status, and primary caregivers for children between the two groups (P>0.05 for all), as detailed in Table 1. Table 1Socio-demographic profile in the study populationVariableChildren with ADHD (n=102)Typically developing children (n=91)U/χ2P valueGender9.8990.002 Male74 (72.5)45 (49.4) Female28 (27.5)46 (50.6)Age, months105.60±23.63118.03±30.0334080.001BMI, kg/m217.68±2.1718.20±2.3840390.12Paternal education level0.3060.98 Graduate degree or above18 (17.6)17 (18.7) Bachelor’s degree40 (39.2)35 (38.5) High school diploma or vocational or below44 (43.2)39 (42.8)Maternal education level1.5700.45 Graduate degree or above19 (18.6)19 (20.9) Bachelor’s degree37 (36.3)39 (42.9) High school diploma or vocational or below46 (45.1)33 (36.2)Annual family income, RMB0.7450.69 <200,00031 (30.4)23 (25.3) 200,000–300,00053 (52.0)49 (53.8) >300,00018 (17.6)19 (20.9)Only child status0.0080.93 Yes42 (41.2)39 (42.9) No60 (58.8)52 (57.1)Main caregiver of children0.5480.91 Mother50 (49.0)45 (49.5) Father8 (7.8)7 (7.7) Both parents19 (18.6)20 (22.0) Grandparents25 (24.6)19 (20.8)Data are presented as mean ± standard deviation or n (%). Continuous variables (age, BMI) were compared using the Mann-Whitney U test. Categorical variables (gender, education, income, only child status, main caregiver) were compared using the χ2 test. A P value <0.05 was considered statistically significant. ADHD, attention-deficit hyperactivity disorder; BMI, body mass index. Socio-demographic profile in the study population Data are presented as mean ± standard deviation or n (%). Continuous variables (age, BMI) were compared using the Mann-Whitney U test. Categorical variables (gender, education, income, only child status, main caregiver) were compared using the χ2 test. A P value <0.05 was considered statistically significant. ADHD, attention-deficit hyperactivity disorder; BMI, body mass index. Comparison of blood metabolite levels after adjustment for confounders After adjusting for age, gender, BMI, paternal education level, maternal education level, family income, only child status, and main caregiver, we found the ADHD group exhibited a significantly higher plasma Glu level compared to the control group [median (Q25, Q75): 103.08 (80.44, 151.06) µmol/L vs. 102.25 (77.65, 126.77) µmol/L], indicating statistical significance. However, no significant differences were found for the remaining six amino acids (GABA, Gly, Ser, Tau, Tyr, Try) or the four PUFAs (LA, DHA, EPA, ARA) (all P>0.05). See Table 2 for details. Table 2Comparison of blood metabolite levels between children with ADHD and control groupMetaboliteChildren with ADHD (n=102)TD children (n=91)PSer181.176±42.636177.013±35.0820.60Tau128.471±57.762132.927±45.2660.67Tyr59.806±13.13364.836±12.3970.052LA73.575 (49.145, 100.123)60.710 (39.680, 86.740)0.08GABA0.400 (0.318, 0.512)0.410 (0.340, 0.490)0.76Glu103.075 (80.440, 151.058)102.250 (77.650, 126.770)0.04Gly319.890 (256.080, 380.470)310.390 (259.430, 383.070)0.83Try76.915 (60.872, 97.530)80.490 (68.540, 94.530)0.33EPA0.330 (0.170, 0.692)0.395 (0.177, 0.575)0.83DHA12.490 (8.803, 16.328)10.760 (7.430, 16.720)0.75ARA6.910 (5.740, 8.393)6.230 (5.210, 8.060)0.08Data are presented as mean ± SD for normally distributed variables and median (P25, P75) for non-normally distributed variables. The units of the above metabolites are μmol/L. Between-group comparisons were performed using ANCOVA adjusted for age, gender, BMI, paternal education level, maternal education level, annual family income, only child status, and main caregiver. P values were corrected using the FDR method. P value <0.05 was considered statistically significant. All results were reported after adjustment for confounders. ADHD, attention-deficit hyperactivity disorder; ANCOVA, analysis of covariance; ARA, arachidonic acid; BMI, body mass index; DHA, docosahexaenoic acid; EPA, eicosapentaenoic acid; FDR, false discovery rate; GABA, γ-aminobutyric acid; Glu, glutamate; Gly, glycine; LA, linoleic acid; SD, standard deviation; Ser, serine; Tau, taurine; TD, typically developing; Try, tryptophan; Tyr, tyrosine. Comparison of blood metabolite levels between children with ADHD and control group Data are presented as mean ± SD for normally distributed variables and median (P25, P75) for non-normally distributed variables. The units of the above metabolites are μmol/L. Between-group comparisons were performed using ANCOVA adjusted for age, gender, BMI, paternal education level, maternal education level, annual family income, only child status, and main caregiver. P values were corrected using the FDR method. P value <0.05 was considered statistically significant. All results were reported after adjustment for confounders. ADHD, attention-deficit hyperactivity disorder; ANCOVA, analysis of covariance; ARA, arachidonic acid; BMI, body mass index; DHA, docosahexaenoic acid; EPA, eicosapentaenoic acid; FDR, false discovery rate; GABA, γ-aminobutyric acid; Glu, glutamate; Gly, glycine; LA, linoleic acid; SD, standard deviation; Ser, serine; Tau, taurine; TD, typically developing; Try, tryptophan; Tyr, tyrosine. The impact of gender on amino acids, PUFAs, and clinical indicators in children with ADHD Comparison analyses revealed no significant differences between the two groups concerning the levels of seven amino acids, four PUFAs, WISC (excluding the VCI), or the factors measured by the SNAP-IV scale (all P>0.05). A notable gender difference was observed exclusively in the VCI, where the boys group achieved a score of 100.000 (96.000, 113.000), which was significantly higher than the girls group’s score of 98.500 (81.250, 103.000), yielding a P value of 0.035, thus indicating statistical significance (Table 3). Table 3The impact of gender on amino acids, PUFAs, and clinical indicators in children with ADHDMetaboliteGendertMann Whitney-UMann Whitney-zPMale (n=74)Female (n=28)Ser174.167±35.235193.423±47.385−1.9550.06Tau128.941±54.661135.224±61.657−0.50.62Try59.187±10.77457.837±14.1850.5160.61FRI101.811±12.71899.286±12.9350.8910.38WMI90.811±11.98287.750±12.3071.1430.26HI12.108±5.67513.071±6.694−0.7280.47LA78.250 (49.203, 100.123)69.605 (40.332, 108.582)969−0.5020.62DHA11.820 (8.008, 15.290)13.125 (10.880, 20.075)882.5−1.1510.25EPA0.320 (0.147, 0.643)0.605 (0.283, 0.730)786.5−1.8710.06ARA6.895 (5.570, 8.283)7.050 (6.067, 8.787)926.5−0.8210.41Glu95.070 (76.845, 147.923)112.900 (85.275, 152.475)868−1.260.21Gly311.880 (253.983, 368.947)334.810 (283.480, 423.865)861−1.3120.19Tyr75.375 (61.000, 96.915)79.445 (57.375, 104.063)1,024−0.090.93FSIQ97.000 (87.000, 105.500)93.000 (84.250, 103.000)861.5−1.3090.19VCI100.000 (96.000, 113.000)98.500 (81.250, 103.000)755−2.110.04PSI92.000 (83.000, 102.000)90.500 (86.000, 99.750)1,021−0.1130.91IN15.500 (12.750, 21.000)14.000 (13.000, 18.750)903−10.32OD6.000 (4.000, 8.000)7.000 (4.000, 8.000)870−1.2520.21Data are presented as mean ± SD or median (P25, P75). The units of the above metabolites are μmol/L. Comparisons between male and female were performed using independent t-test or Mann-Whitney U test. P values were corrected using the FDR method. A P<0.05 was considered statistically significant. ADHD, attention-deficit hyperactivity disorder; ARA, arachidonic acid; DHA, docosahexaenoic acid; EPA, eicosapentaenoic acid; FRI, fluid reasoning index; FSIQ, full scale intelligence quotient; GABA, γ-aminobutyric acid; Glu, glutamate; Gly, glycine; HI, hyperactivity impulsivity; IN, inattention; LA, linoleic acid; OD, oppositional defiance; PSI, processing speed index; PUFAs, polyunsaturated fatty acids; SD, standard deviation; Ser, serine; Tau, taurine; Try, tryptophan; Tyr, tyrosine; VCI, verbal comprehension index; WMI, working memory index. The impact of gender on amino acids, PUFAs, and clinical indicators in children with ADHD Data are presented as mean ± SD or median (P25, P75). The units of the above metabolites are μmol/L. Comparisons between male and female were performed using independent t-test or Mann-Whitney U test. P values were corrected using the FDR method. A P<0.05 was considered statistically significant. ADHD, attention-deficit hyperactivity disorder; ARA, arachidonic acid; DHA, docosahexaenoic acid; EPA, eicosapentaenoic acid; FRI, fluid reasoning index; FSIQ, full scale intelligence quotient; GABA, γ-aminobutyric acid; Glu, glutamate; Gly, glycine; HI, hyperactivity impulsivity; IN, inattention; LA, linoleic acid; OD, oppositional defiance; PSI, processing speed index; PUFAs, polyunsaturated fatty acids; SD, standard deviation; Ser, serine; Tau, taurine; Try, tryptophan; Tyr, tyrosine; VCI, verbal comprehension index; WMI, working memory index. Correlation analysis of seven amino acids and four fatty acids with symptoms and cognitive levels in children with ADHD Correlation analysis findings (refer to Table 4) indicated that plasma Try levels in children with ADHD exhibited a positive correlation with the SNAP-IV attention deficit factor score (r=0.209, n=102, P<0.05), suggesting that elevated Try levels are correlate with more pronounced symptoms of inattention. In contrast, plasma GABA levels were found to be negatively correlated with the WISC-IV FSIQ (r=−0.198, n=102, P<0.05) Moreover, plasma EPA levels displayed a significant negative correlation with the WISC-IV PSI (r=−0.212, n=102, P<0.05), indicating that higher plasma EPA concentrations correlated with poorer processing speed scores. Table 4Correlation analysis of seven amino acids and four fatty acids with symptoms and cognitive levels in children with ADHDMetaboliteFSIQVCIFRIWMIPSIINHIODLA−0.01−0.0710.0650.16−0.101−0.1950.085−0.107EPA−0.0190.143−0.013−0.067−0.212*0.0250.073−0.006DHA0.0010.096−0.045−0.057−0.117−0.080.04−0.093ARA−0.168−0.104−0.194−0.168−0.0230.090.161−0.051GABA−0.198*−0.061−0.081−0.126−0.1480.0080.032−0.066Glu−0.0230.132−0.041−0.017−0.04−0.0590.082−0.086Gly0.0390.111−0.019−0.020.053−0.1080.001−0.019Ser−0.0880.04−0.103−0.137−0.0930.0950.147−0.038Tau−0.0320.103−0.044−0.004−0.008−0.086−0.011−0.104Tyr−0.042−0.0190.063−0.001−0.0770.0780.0680.112Try−0.18−0.146−0.105−0.048−0.1220.209*0.0340.024The units of the above metabolites are μmol/L. Correlation coefficients were analyzed using Pearson correlation. P values were corrected using the FDR method. P<0.05 was considered statistically significant. *, P<0.05. ADHD, attention-deficit hyperactivity disorder; ARA, arachidonic acid; DHA, docosahexaenoic acid; EPA, eicosapentaenoic acid; FDR, false discovery rate; FRI, fluid reasoning index; FSIQ, full scale intelligence quotient; GABA, γ-aminobutyric acid; Glu, glutamate; Gly, glycine; HI, hyperactivity impulsivity; IN, inattention; LA, linoleic acid; OD, oppositional defiance; PSI, processing speed index; Ser, serine; Tau, taurine; Try, tryptophan; Tyr, tyrosine; VCI, verbal comprehension index; WMI, working memory index. Correlation analysis of seven amino acids and four fatty acids with symptoms and cognitive levels in children with ADHD The units of the above metabolites are μmol/L. Correlation coefficients were analyzed using Pearson correlation. P values were corrected using the FDR method. P<0.05 was considered statistically significant. *, P<0.05. ADHD, attention-deficit hyperactivity disorder; ARA, arachidonic acid; DHA, docosahexaenoic acid; EPA, eicosapentaenoic acid; FDR, false discovery rate; FRI, fluid reasoning index; FSIQ, full scale intelligence quotient; GABA, γ-aminobutyric acid; Glu, glutamate; Gly, glycine; HI, hyperactivity impulsivity; IN, inattention; LA, linoleic acid; OD, oppositional defiance; PSI, processing speed index; Ser, serine; Tau, taurine; Try, tryptophan; Tyr, tyrosine; VCI, verbal comprehension index; WMI, working memory index. Relationship between metabolite levels and severity of hyperactive-impulsive symptoms in children with ADHD Children with ADHD were classified into three categories based on their total scores from the SNAP-IV Hyperactivity impulsivity factor, symptomatic, mild, moderate severe. Results from the Kruskal-Wallis H test indicated significant differences in Glu and ARA levels across varying hyperactivity impulsivity severity levels (H=6.695, P=0.04; H=8.616, P=0.01). No significant differences were found for other PUFAs (LA, DHA, EPA) and amino acids among the different severity levels of hyperactivity impulsivity (all P>0.05, see Tables 5,6). Table 5Relationship between metabolite levels and severity of hyperactive-impulsive symptoms in children with ADHDMetaboliteSymptomatic (n=48)Mild (n=31)Moderate-severe (n=23)HPDHA10.580 (7.678, 14.658)14.410 (10.960, 21.870)11.990 (10.620, 14.950)5.9110.052EPA0.325 (0.182, 0.623)0.300 (0.100, 0.730)0.590 (0.170, 1.110)4.7880.09ARA6.425 (5.295, 7.845)8.000 (6.380, 8.830)7.530 (6.160, 9.170)8.6160.01*LA67.070 (38.625, 88.775)78.740 (51.010, 118.760)84.780 (52.390, 100.340)2.3290.31Glu87.105 (76.117, 132.650)121.570 (88.260, 172.750)108.140 (78.230, 193.530)6.6950.04*Gly325.820 (271.190, 395.618)318.310 (281.070, 398.060)285.900 (245.770, 353.380)2.7130.26Tyr76.415 (66.245, 96.550)76.580 (57.480, 104.200)78.030 (58.400, 93.690)0.0830.96GABA0.400 (0.323, 0.460)0.410 (0.330, 0.570)0.380 (0.290, 0.580)0.9510.62Data are presented as median (P25, P75). The units of the above metabolites are μmol/L. Group differences were analyzed using Kruskal- Wallis H test. P values were corrected using the FDR method. P<0.05 was considered statistically significant. ADHD, attention-deficit hyperactivity disorder; ANOVA, analysis of variance; ARA, arachidonic acid; DHA, docosahexaenoic acid; EPA, eicosapentaenoic acid; FDR, false discovery rate; GABA, γ-aminobutyric acid; Glu, glutamate; Gly, glycine; LA, linoleic acid; Tyr, tyrosine. Relationship between metabolite levels and severity of hyperactive-impulsive symptoms in children with ADHD Data are presented as median (P25, P75). The units of the above metabolites are μmol/L. Group differences were analyzed using Kruskal- Wallis H test. P values were corrected using the FDR method. P<0.05 was considered statistically significant. ADHD, attention-deficit hyperactivity disorder; ANOVA, analysis of variance; ARA, arachidonic acid; DHA, docosahexaenoic acid; EPA, eicosapentaenoic acid; FDR, false discovery rate; GABA, γ-aminobutyric acid; Glu, glutamate; Gly, glycine; LA, linoleic acid; Tyr, tyrosine. Table 6Relationship between other metabolite levels and severity of hyperactive impulsive symptoms in children with ADHDMetaboliteSymptomatic (n=48)Mild (n=31)Moderate–severe (n=23)FPSer175.650±41.555183.846±40.707181.469±34.7220.4360.65Tau125.733±56.769142.852±56.513124.535±55.4491.0430.36Try59.108±10.86357.330±10.79660.212±14.7450.420.66Data are presented as mean ± SD. The units of the above metabolites are μmol/L. Group differences were analyzed using one-way ANOVA. P values were corrected using the FDR method. P<0.05 was considered statistically significant. ANOVA, analysis of variance; FDR, false discovery rate; SD, standard deviation; Ser, serine; Tau, taurine; Try, tryptophan. Relationship between other metabolite levels and severity of hyperactive impulsive symptoms in children with ADHD Data are presented as mean ± SD. The units of the above metabolites are μmol/L. Group differences were analyzed using one-way ANOVA. P values were corrected using the FDR method. P<0.05 was considered statistically significant. ANOVA, analysis of variance; FDR, false discovery rate; SD, standard deviation; Ser, serine; Tau, taurine; Try, tryptophan. Synergistic associations between seven amino acids and fatty acid metabolism in children with ADHD Two significant positive correlations were identified: Ser with ARA, indicating that increased Ser levels significantly elevate ARA levels, and Ser with Glu, reflecting a moderate positive correlation. Additionally, two trending positive correlations were noted: Ser with DHA, approaching statistical significance and suggesting a weak positive correlation; and Glu with ARA, indicating a potential weak positive correlation between Glu and ARA levels (see Figure 1). Correlation matrix of plasma amino acids, PUFAs and cognitive function scores in children with ADHD. Correlation coefficients (r) are presented for all pairwise comparisons. Correlations were analyzed using Pearson’s correlation test. ADHD, attention-deficit hyperactivity disorder; ARA, arachidonic acid; DHA, docosahexaenoic acid; EPA, eicosapentaenoic acid; FRI, fluid reasoning index; FSIQ, full scale intelligence quotient; GABA, γ-aminobutyric acid; Glu, glutamate; Gly, glycine; HI, hyperactivity impulsivity; IN, inattention; LA, linoleic acid; OD, oppositional defiance; PSI, processing speed index; PUFAs, polyunsaturated fatty acids; Ser, serine; Tau, taurine; Try, tryptophan; Tyr, tyrosine; VCI, verbal comprehension index; WMI, working memory index. Discussion ADHD is a prevalent chronic neurodevelopmental disorder in children. Although the neurobiological basis of ADHD remains incompletely understood, excitatory-inhibitory neurotransmitter imbalance has been associated with ADHD-related neurodevelopmental processes. GABA is the primary inhibitory neurotransmitter involved in modulating dopamine transmission and behavioral regulation, whereas Glu is the main excitatory neurotransmitter that modulates dopamine release. Imbalances between GABA and Glu have been linked to altered sensory gating in striatal-prefrontal pathways (21,22). Postprandial plasma amino acid profiles mirror the inherent properties of consumed dietary protein, serve as key regulators of whole-body tissue protein turnover, and dictate anabolic amino acid utilisation in peripheral tissues including skeletal muscle (23,24). Peripheral amino acid availability may shape ADHD-related cognition and behavior through competitive blood-brain barrier uptake and altered central neurotransmitter synthesis. For instance, Try bioavailability constrains cerebral serotonin production, and kynurenine pathway metabolites modulate NMDA receptor signaling to disrupt neural function (25). The present study found that plasma Glu levels were higher in the ADHD group than in TD children, which is consistent with previous report of elevated serum Glu in untreated ADHD children (18). Glu is the predominant excitatory neurotransmitter in the mammalian brain, circulating Glu is partially derived from the breakdown of dietary protein, and skeletal muscle together with the liver function as major peripheral Glu reservoirs. Once the blood-brain barrier becomes disrupted, surplus central Glu effluxes into peripheral blood, and elevated systemic Glu concentrations can serve as an indicator of aberrant central glutamatergic neurotransmission (26). Our study reports a positive correlation between plasma Try levels and SNAP-IV inattention scores in ADHD children. Try is an essential amino acid wholly dependent on dietary protein intake and competes with other large neutral amino acids for transport across the blood-brain barrier; its availability in plasma directly constrains cerebral serotonin synthesis (27). The observed correlation between Try and inattention symptoms supports that Try metabolic profiles are related to the inattention phenotype of ADHD (28,29). We noted that a negative correlation between plasma GABA levels and FSIQ in children with ADHD. This result seems to contradict earlier findings by Cook et al. (30), who reported a positive correlation between GABA concentrations and specific intelligence domains (in the visual cortex and visual intelligence). Zacharopoulos et al. (31), who noted a similar relationship between GABA in the intraparietal sulcus and fluid intelligence. Several factors may clarify these discrepancies. First, while Cook et al. and Zacharopoulos et al. used proton magnetic resonance spectroscopy to measure localized GABA concentrations in specific brain areas, this study assessed plasma GABA levels. Second, Cook et al. focused on “visual intelligence” via the WISC matrix reasoning test, whereas Zacharopoulos et al. investigated “fluid intelligence”. This study evaluated FSIQ, encompassing various dimensions, including verbal comprehension, working memory, processing speed, and perceptual reasoning. Thus, GABA’s effect on different intelligence dimensions may significantly differ. Zacharopoulos et al. found that the GABA-intelligence association was significant only in mature individuals, such as late adolescents and adults. Our participants, aged 6 to 15, were in a crucial neurodevelopmental phase. These factors may help explain the distinct association observed in this cross-sectional dataset. Further research is needed to clarify the relationship between peripheral GABA and cognitive function in children with ADHD. The findings of this study diverge from the meta-analysis (32), which indicated lower blood levels of DHA and EPA in children with ADHD compared to their neurotypical counterparts. In contrast, this study found no significant differences in the levels of DHA and EPA between the ADHD group and the control group. This discrepancy may be related to three key factors. First, regional dietary habits leading to differences in baseline fatty acid intake are an important consideration. Previous studies have confirmed that coastal residents, whose diets are primarily based on seafood, consume significantly higher amounts of EPA and DHA daily compared to inland residents. An analysis of breast milk fatty acids substantiates this difference (33). A study across eight Chinese cities included 75 breastfeeding mothers postpartum for 6–9 months, collecting 600 breast milk and 75 maternal fingertip blood samples. Oleic acid was the most prevalent fatty acid in breast milk, followed by LA. Inland mothers had a DHA proportion of only 0.26%, significantly lower than 0.57% in coastal mothers, while ALA was higher in inland milk compared to coastal. The study linked these differences to maternal dietary habits but had limitations, including a small sample size and lack of infant blood samples or data on obese mothers. Furthermore, the findings of Harris et al. support this conclusion, as they measured the Omega-3 Index (O3I, defined as the percentage of EPA and DHA in total fatty acids in red blood cells) among volunteers from Newport, Rhode Island and five inland cities (including Indianapolis and Charleston, total n=1,109) (34). The results indicated that Newport volunteers had a significantly higher average O3I than inland city volunteers, supporting the hypothesis that coastal populations have greater access to seafood. However, this study also faced limitations: both groups were convenience samples, lacking data on seafood intake, Omega-3 supplement usage, socioeconomic factors, and weight; additionally, the data collection for inland participants occurred from 2014 to 2015, while coastal data were collected in 2019, introducing potential biases. Overall, local children’s higher seafood intake in Zhejiang Province may correlate with elevated plasma ω-3 fatty acid levels, obscuring differences between ADHD and control groups. Second, the lack of subdivision among ADHD subtypes may contribute to result bias. Previous research by Burgess et al. (35) categorized ADHD children into L-ADHD (ADHD children presenting with abundant EFA deficiency manifestations (including persistent thirst, frequent urination and dry hair) subgroups and others: the L-ADHD subgroup exhibited more symptoms of EFA deficiency, including thirst, frequent urination, and dry hair; whereas other subgroups exhibited fewer EFA deficiency symptoms. The findings revealed that the plasma AA to DHA ratio was significantly lower in the L-ADHD subgroup compared to those with fewer symptoms, whose fatty acid composition did not differ significantly from healthy controls. This suggests that previous studies on fatty acid supplementation in ADHD may have yielded inaccurate results due to sample heterogeneity. The lack of differentiation among ADHD subtypes in this study likely averaged out fatty acid level differences, masking significant variations between the ADHD and control groups. Additionally, this study quantified plasma fatty acid concentrations. Prior research has established that erythrocyte membrane fatty acids reflect long-term fatty acid status over weeks to months, given the 120 day lifespan of red blood cells. In comparison, plasma fatty acid levels only capture recent dietary intake and are more sensitive to short-term dietary fluctuations (36). Importantly, our exclusion criterion only ruled out standardized nutritional supplements within 3 months prior to enrollment, but we lacked records of long-term habitual PUFA intake from natural food sources such as seafood and edible oils. Short-term supplement use generates transient lipid fluctuations that are easy to restrict via screening, whereas lifelong dietary patterns establish stable baseline plasma PUFA levels that cannot be eliminated simply by excluding supplements. As a cross-sectional study without validated food frequency questionnaires, we cannot quantify or adjust for this persistent dietary confounding in multivariate models. Our study sample was recruited from Zhejiang Province. As widely recognized, coastal populations generally have higher seafood consumption rates. However, we did not collect any dietary data from our participants. Should such unmeasured inter-individual dietary differences exist within our cohort, they could attenuate the between-group disparities in n-3 PUFAs observed in meta-analyses focused on inland populations. This potential confounding factor remains merely speculative and lacks empirical support from our dataset. Notably, although we excluded participants using PUFA or amino acid supplements three months before enrollment, we lacked quantitative assessment of daily habitual diet including total energy, fat, protein, and natural food-derived PUFA intake, which restricts deeper interpretation of the present metabolite associations. Prior observational cohort data implied that balanced dietary PUFA ratios could alleviate ADHD-related neurodevelopmental deficits (12,13), while this Cochrane systematic review emphasised that unregulated habitual fatty acid intake serves as a prominent confounding factor, obscuring the genuine correlation between endogenous lipid metabolism and ADHD clinical manifestations and accounting for the high heterogeneity across PUFA intervention trials (37). As highlighted in the meta-analysis by Fu et al., regional disparities in seafood intake driven by diverse dietary customs lead to inconsistent baseline circulating PUFA concentrations across study populations; cross-group analyses lacking dietary covariate adjustment are therefore susceptible to residual confounding that distorts PUFA-ADHD associations (8). Future work integrating validated food frequency questionnaires is required to isolate the independent effects of inherent PUFA metabolism from long-term dietary nutrient exposure in children with ADHD. Interestingly, this study identified a negative correlation between EPA levels and processing speed in children with ADHD, while ARA levels were positively associated with hyperactivity-impulsivity scores. Emerging evidence supports beneficial roles of EPA in mitigating neuroinflammation and alleviating ADHD behavioral deficits, largely by competing with ARA to lower pro-inflammatory lipid mediators and indirectly modulating central monoamine signaling (8,12). However, we observed an inverse association between plasma EPA and processing speed in this Zhejiang cohort, which may be attributable to regional seafood intake patterns, plasma’s susceptibility to short-term diet, and lack of ADHD subtype stratification as elaborated above. Correlations were observed between Ser and ARA, as well as Ser and Glu, suggesting coordinated changes in these metabolites in children with ADHD. Correlation analyses found positive associations between Ser and ARA, as well as Ser and Glu, while Gly showed a trend towards positive correlation with DHA and Glu with ARA. These correlational findings suggest potential coordinated metabolic trends that require replication and mechanistic validation. Conclusions Based on the study hypotheses stated above, the present study confirmed that plasma Glu was significantly higher in children with ADHD than in TD children after adjustment for confounders. Try was positively correlated with inattention symptoms. GABA was negatively correlated with full-scale IQ and EPA was negatively correlated with processing speed. ARA and Glu levels differed significantly across hyperactivity-impulsivity severity groups. These preliminary, weak correlational findings suggest that peripheral plasma PUFAs and amino acids may be associated with ADHD phenotypic and cognitive features. However, due to the cross-sectional design, modest effect sizes, and small number of significant findings, no causal inference can be drawn. Future longitudinal and multi-center studies are needed to validate these exploratory signals. Our study has several limitations. The cross-sectional design precludes causal inference, only associations can be identified. The single-center design and relatively small sample may limit generalizability. ADHD subtypes were not stratified, which may mask subgroup-specific associations. Another key limitation is the fixed composition of the clinical neurotrophic metabolite panel used herein. Only seven amino acids and four long-chain PUFAs are available for quantification via this assay, while numerous other neuroactive, inflammatory and Try-pathway metabolites potentially involved in ADHD pathophysiology remain unmeasured. These unassessed metabolites may serve as unmeasured confounders or independent correlates of ADHD clinical and cognitive phenotypes, reducing the comprehensiveness of our metabolic findings. Future untargeted metabolomics covering a broader spectrum of circulating metabolites is required to address this gap. Additionally, we lacked systematic quantitative dietary assessment. Unmeasured habitual intake of energy, macronutrients and n-3/n-6 PUFAs may introduce residual confounding and alter the cross-sectional metabolite-phenotype correlations identified in this study. Although bivariate tests were used for exploratory purposes, future studies should employ multivariate regression models to further adjust for diet, physical activity, and socioeconomic status. Future research should use longitudinal designs to explore temporal relationships and validate whether these metabolic markers can predict or stratify ADHD phenotypes in larger multi-center cohorts. Supplementary The article’s supplementary files as Available at https://tp.amegroups.com/article/view/10.21037/tp-2026-0590/prf