Serum folate and brain-derived neurotrophic factor in pediatric autism spectrum disorder and their predictive role in illness severity

Abstract

BACKGROUND

Autism spectrum disorder (ASD) involves social and neurological impairment, and affected individuals have an elevated risk of bullying.

AIM

To clarify serum folate (SF) and brain-derived neurotrophic factor (BDNF) expression in ASD-affected children and evaluate their prediction value for illness severity.

METHODS

From February 2023 to February 2025, 53 ASD-affected children and 50 healthy controls visiting Fuzhou University Affiliated Provincial Hospital were enrolled as the research and control groups, respectively. SF and BDNF levels were measured in all children. The Childhood Autism Rating Scale (CARS) was used to assess ASD symptom severity. In ASD cases, SF and BDNF expression differences were compared across illness-severity subgroups and before versus after treatment. Pearson r was used to assess correlations between SF/BDNF and CARS in the research group. Receiver operating characteristic (ROC) curves were used to assess their predictive value for ASD severity. Univariate and multivariate binary Logistic models were used to identify ASD progression determinants.

RESULTS

ASD children showed significantly lower SF and higher BDNF higher than controls. Severe cases had lower SF and higher BDNF than mild-to-moderate cases. SF correlated inversely with the CARS score, whereas BDNF correlated positively. For predicting ASD severity, the area under the ROC curve (AUC) of SF and BDNF was 0.700-0.750, and their combined use increased the AUC to 0.823. Both markers were confirmed to be independent determinants of ASD aggravation.

CONCLUSION

SF is down-regulated and BDNF is up-regulated in ASD-affected children, SF correlates negatively with ASD severity and BDNF correlates positively. Low SF and high BDNF are risk factors for ASD deterioration in children.

Key Words: Serum folate; Brain-derived neurotrophic factor; Autism spectrum disorder; Children; Illness severity; Receiver operating characteristic curve

Core Tip: We selected 53 autism spectrum disorder (ASD)-affected children and 50 healthy controls. Serum folate (SF) was down-regulated and brain-derived neurotrophic factor (BDNF) was up-regulated in ASD children. Both indices related closely to illness severity and could reflect therapeutic effectiveness. Moreover, low SF and high BDNF levels held predictive value for ASD aggravation in children.

INTRODUCTION

Autism spectrum disorder (ASD) is a neurodevelopmental disorder arising in the fetus and very young children[1], yet often detected only preschool years[2]. Its manifestations include impaired social interaction and communication and restricted behaviors, interests, and activity patterns. These heterogeneous neurodevelopmental and functional disorders frequently persist into adulthood, imposing substantial burden on individuals, their families, the education system, and society[3,4]. ASD pathogenesis involves the downregulation of synaptic-related genes, astrocyte and microglia activation, gut microbiota-induced neuroinflammation and immune dysregulation, and functional disturbances in brain regions and neural circuits[5]. Relevant epidemiological data show that ASD-affected children face higher bullying risks, which increase with the severity of social impairments[6]. Few studies have predicted ASD severity based on serum markers[7]. This study therefore analyses serum folate (SF) and brain-derived neurotrophic factor (BDNF) expression to explore more effective diagnostic and treatment strategies and reduce the overall burden on patients and society.

SF-vitamin B9 is an essential micronutrient for fetal neurodevelopment, supporting normal DNA synthesis and complete neural tube closure[8], and this dependence continues after birth. Due to dietary problems, children with ASD have a significantly increased risk of multiple nutrient deficiencies (SF, vitamin C, iron, etc.)[9,10]. Children with ASD typically have lower SF levels than normally developing children, suggesting that ASD pathogenesis may be related to deficient brain SF[11,12]. Conversely, BDNF is a central nervous system neurotrophin involved in the pathophysiology of various mental diseases (bipolar, major depressive, obsessive-compulsive disorders, etc.) and regulates neuronal development, synaptic plasticity, and neural network function[13]. In children, BDNF regulates early normal neuronal development and functional establishment; its dysregulation can induce long-term structural/functional changes and abnormally elevated peripheral BDNF in ASD-affected children[14]. A study has linked ASD-like behaviors to impaired conversion of BDNF precursor to mature BDNF in the brain, suggesting that BDNF may also mediate ASD pathogenesis[15]. A basic experiment also associated autistic-like behaviors with low BDNF and glial fibrillary acidic protein (GFAP) methylation, increased oxidative stress, increased inflammatory-cell infiltration, and frontal-cortex structural change; folic acid intervention helps reverse these pathological changes and improve autistic-like behaviors, suggesting that BDNF and SF may co-mediate ASD pathophysiology[16].

Few studies have examined SF and BDNF in pediatric ASD cases and their severity-prediction value. This study therefore evaluates whether SF and BDNF may serve as predictors of ASD severity in children through correlation analyses. The 5-10-year age selection was based primarily on: (1) More stable and reliable ASD diagnosis in this age group, effectively reducing early-childhood atypical-symptom interference[17]; (2) A relatively stable plateau in brain development, reducing age-related physiological fluctuations in serum SF and BDNF levels[18]; and (3) Better cooperation and reliable the reliability of the assessment by the Childhood Autism Rating Scale (CARS) scoring. Intervention during this critical period also improves clinical utility.

MATERIALS AND METHODS

Case selection

This study involved 53 ASD-affected children (research group) and 50 healthy children (control group) who visited Fuzhou University Affiliated Provincial Hospital (February 2023-February 2025). Inclusion criteria: A confirmed ASD diagnosis[19]; age 5-10 years; intact clinical data; no immune/chronic gastrointestinal disorders or cranial injuries; treatment-naive status. Exclusion criteria: Cerebral palsy, mental retardation, or other significant/acute neurodevelopmental disorders; significant heart, lung, brain, or kidney impairments; genetically determined diseases (Rett syndrome, fragile X syndrome); serious infectious diseases (acute/chronic); schizophrenia, obsessive-compulsive disorder, or epilepsy; moderate-to-severe anemia; administration of vitamin D-suppressing drugs or calcium preparations within the last three months.

Intervening methods

The research team employed the Applied Behavior Analysis method, targeting cognition, fine/gross motor skills, language, imitation, social interaction, and daily living abilities. Cognitive and fine motor training involved bead threading, shape matching, and puzzle-solving. Gross motor training aimed to enhance the child’s vestibular and visual senses through body rotation, vertical jumps, slide descents, and balance beam walking. Language training included single-word pronunciation, daily expressions, answering questions, and imitation of teacher speech. Imitation training involved action and verbal replication. Social skills were cultivated through interactive games, while daily living training focused on eating, toileting, and dressing. Training content was divided into several small behavioral units, starting from the simplest to the most complex. One-on-one instruction was used until mastery. Finally, the children were taught to gradually apply the learned content to their daily lives. Training was conducted once daily for 60 minutes, 5 days a week, with 2 days of rest. The training lasted for 2 months continuously.

Outcome measurement

SF and BDNF determination. All the children underwent morning fasting blood collection, with 5 mL of venous blood taken from the elbow; serum was separated via centrifugation. SF level was determined using chemiluminescent micro-particle immunoassay on an immunoassay system. This technique integrates multiple advanced technologies and enables highly sensitive and specific detection[20]. BDNF was determined via enzyme-linked immunosorbent assay, which uses two specific antibodies recognizing different BDNF epitopes for sandwich detection[21]. Strict adherence to blood-handling procedures was ensured to avoid invalid measurements.

ASD severity assessment. The CARS[22] assessed ASD severity across 15 domains: Near-sight sensory response, verbal communication, nonverbal communication, anxiety response, activity level, intellectual function, imitation, interpersonal relationship, motor coordination, emotional response, adaptation to environmental changes, interaction with inanimate objects, auditory response, visual response, and general impression. Using the 4-point rating method, a score of 1 denotes typical age-appropriate behavior, whereas 4 represents marked abnormalities. Total scores (15-60) were categorized as mild-to-moderate (30-36) or severe (≥ 37). CARS demonstrates excellent reliability (Cronbach’s α = 0.935).

Statistical analysis

SPSS 26 was used to process all data. Following Kolmogorov-Smirnov testing, nominally distributed measurement data are expressed as mean ± SD and compared using t-tests. Categorical data are expressed as % and compared using χ² tests. Pearson r was used to investigate associations between SF, BDNF, and CARS. Receiver operating characteristic (ROC) curves quantified SF and BDNF ASD severity-prediction value. Determinants of ASD were identified by univariate screening followed by logistic regression. P < 0.05 indicated statistical significance.

RESULTS

Patients’ demographic information

As shown in Table 1[23], there were no significant between-group differences in gender, age, disease duration, family genetic history, or parental personality (P > 0.05), confirming good comparability.

Table 1 Patients’ demographic information, n (%).

Data	Control group (n = 50)	Research group (n = 53)	t/χ2	P value
Gender			0.918	0.338
Male	32 (64.00)	29 (54.72)
Female	18 (36.00)	24 (45.28)
Age (years)	7.34 ± 1.57	7.06 ± 1.26	1.001	0.319
Disease duration (months)	5.88 ± 2.58	6.04 ± 2.85	0.298	0.766
Family genetic history			2.443	0.118
No	43 (86.00)	39 (73.58)
Yes	7 (14.00)	14 (26.42)
Severity
Mild to moderate	-	28 (52.83)
Severe	-	25 (47.17)
Father's personality			0.240	0.887
Introverted	15 (30.00)	17 (32.08)
Ambiverted	10 (20.00)	12 (22.64)
Extroverted	25 (50.00)	24 (45.28)
Mother's personality			0.404	0.817
Introverted	13 (26.00)	15 (28.30)
Ambiverted	20 (40.00)	18 (33.96)
Extroverted	17 (34.00)	20 (37.74)

For reference, parental personalities were evaluated through the Eysenck’s Personality Questionnaire-revised designed for adults[23]. In this assessment, the E (extraversion-introversion) dimension grouped participants into introverted, ambiverted, or extroverted categories.

SF and BDNF levels

ASD-affected children exhibited significantly lower SF and markedly higher BDNF levels than controls (both P < 0.001; Figure 1).

Figure 1 Comparative analysis of serum folate and brain-derived neurotrophic factor concentrations. A: Serum folate levels across groups; B: Comparative brain-derived neurotrophic factor measurements. ^cP < 0.001. SF: Serum folate; BDNF: Brain-derived neurotrophic factor.

SF and BDNF concentrations by ASD severity

Severe cases had lower SF (P < 0.01) and higher BDNF (P < 0.001) than mild-to-moderate patients (Figure 2).

Figure 2 Serum folate and brain-derived neurotrophic factor levels across different severity grades in the research group. A: Serum folate concentrations in mild, moderate, and severe patients; B: Brain-derived neurotrophic factor concentrations in cases with mild, moderate, or severe illness. ^bP < 0.01; ^cP < 0.001. SF: Serum folate; BDNF: Brain-derived neurotrophic factor.

Pre- and post-treatment changes in SF and BDNF concentrations in ASD-affected children

In ASD cases, SF concentration increased from baseline (P < 0.01) while BDNF decreased (P < 0.01) following treatment (Figure 3).

Figure 3 Treatment-induced alterations in serum folate and brain-derived neurotrophic factor levels in pediatric subjects in the research group. A: Pre-treatment vs post-treatment serum folate concentrations in the research group; B: Brain-derived neurotrophic factor level fluctuations in pre- and post-therapy. ^bP < 0.01. SF: Serum folate; BDNF: Brain-derived neurotrophic factor.

Correlation between SF, BDNF, and CARS scores

As shown in Table 2, SF correlated inversely with CARS (r = −0.345, P = 0.011), whereas BDNF correlated positively with it (r = 0.389, P = 0.004).

Table 2 Correlation between serum folate, brain-derived neurotrophic factor levels and Childhood Autism Rating Scale scores in the research group.

Indicators	CARS (points)
	r	P value
SF (nmol/L)	-0.345	0.011
BDNF (pg/mL)	0.389	0.004

SF: Serum folate; BDNF: Brain-derived neurotrophic factor; CARS: Childhood Autism Rating Scale.

Predictive value of SF and BDNF in childhood ASD

SF showed an area under the ROC curve (AUC) of 0.703 (95%CI: 0.561-0.845) while BDNF had an AUC of 0.744 (95%CI: 0.607-0.882) for predicting ASD severity. Combined use improved the AUC to 0.823 (95%CI: 0.711-0.934) (Table 3 and Figure 4).

Figure 4 Receiver operating characteristic curves for serum folate and brain-derived neurotrophic factor in predicting the severity of autism spectrum disorder in children. AUC: Area under the receiver operating characteristic curve; SF: Serum folate; BDNF: Brain-derived neurotrophic factor.

Table 3 Predictive potential of serum folate and brain-derived neurotrophic factor in childhood autism spectrum disorder.

Indicators	AUC	95%CI	P value	Sensitivity (%)	Specificity (%)	Cutoff
SF (nmol/L)	0.703	0.561-0.845	0.011	68.00	71.43	18.34
BDNF (pg/mL)	0.744	0.607-0.882	0.002	84.00	67.86	16.62
Joint detection	0.823	0.711-0.934	< 0.001	84.00	71.43	0.40

SF: Serum folate; BDNF: Brain-derived neurotrophic factor; AUC: Area under the receiver operating characteristic curve.

Analysis of key determinants influencing ASD severity

When screening variables impacting pediatric ASD severity, non-significant associations were observed for gender, age, family genetics, and parental personality traits (P > 0.05). However, disease duration, SF, and BDNF levels correlated significantly with ASD severity (P < 0.05; Table 4). Logistic regression identified SF ≥ 20 nmol/L as protective (OR = 0.175) and elevated BDNF ≥ 18 pg/mL as a risk factor (OR = 10.147; P < 0.05; Table 5).

Table 4 Univariate assessment of autism spectrum disorder severity determinants in pediatric cases, n (%).

Variable	Mild-to-moderate (n = 28)	Severe (n = 25)	χ2	P value
Gender			0.533	0.465
Male	14 (50.00)	15 (60.00)
Female	14 (50.00)	10 (40.00)
Age (years)			2.672	0.102
< 7	14 (50.00)	18 (72.00)
≥ 7	14 (50.00)	7 (28.00)
Disease duration (months)			5.306	0.021
< 5	21 (75.00)	11 (44.00)
≥ 5	7 (25.00)	14 (56.00)
Family genetic history			0.759	0.384
No	22 (78.57)	17 (68.00)
Yes	6 (21.43)	8 (32.00)
Father's personality			5.731	0.057
Introverted	5 (17.86)	12 (48.00)
Ambiverted	7 (25.00)	5 (20.00)
Extroverted	16 (57.14)	8 (32.00)
Mother's personality			2.237	0.327
Introverted	6 (21.43)	9 (36.00)
Ambiverted	9 (32.14)	9 (36.00)
Extroverted	13 (46.43)	7 (28.00)
SF (nmol/L)			4.567	0.033
< 20	12 (42.86)	18 (72.00)
≥ 20	16 (57.14)	7 (28.00)
BDNF (pg/mL)			10.194	0.001
< 18	19 (67.86)	6 (24.00)
≥ 18	9 (32.14)	19 (76.00)

SF: Serum folate; BDNF: Brain-derived neurotrophic factor.

Table 5 Multivariate regression modeling for childhood autism spectrum disorder prognostic factors.

Variable	β	SE	Wald	P value	OR	95%CI
Disease duration (months)	-0.465	0.688	0.457	0.499	0.628	0.163-2.420
SF (nmol/L)	-1.740	0.760	5.244	0.022	0.175	0.040-0.778
BDNF (pg/mL)	2.317	0.749	9.574	0.002	10.147	2.338-44.031

SF: Serum folate; BDNF: Brain-derived neurotrophic factor.

DISCUSSION

Analysis of 53 pediatric ASD cases and 50 healthy children revealed significantly lower SF and higher BDNF levels in ASD cases. Zou et al[24] similarly found reduced SF in ASD and emphasized dietary management, nutritional supplementation, and behavioral interventions for symptom improvement. In a randomized controlled trial, Panda et al[25] observed ASD symptom amelioration with oral folinic acid administration, particularly in children with high folate receptor autoantibody titers, suggesting SF as a therapeutic target. A rodent study also showed that folate alleviated autism-like behaviors induced by neonatal separation in rats, partly via epigenetic modifications of the BDNF and GFAP promoters[16]. Farmer et al[26] reported the highest BDNF concentrations in ASD compared with normal-developing and developmentally delayed peers, aligning with our results. Mahmoud et al[27] confirmed BDNF mRNA overexpression in ASD rodent models, suggesting that abnormal BDNF up-regulation in ASD may involve the cross-linking between the MeCP2/BDNF-CREB signaling transduction.

When stratified by illness severity, this study found lower SF and higher BDNF levels in severe ASD. Following treatment, SF significantly increased, while BDNF markedly decreased in ASD-affected children. Correlation analysis revealed that CARS scores correlated inversely with SF but positively with BDNF. In the study by Cui et al[28], which included both children and adolescents, ASD participants showed significantly higher serum BDNF concentrations than those with intellectual disabilities, with BDNF positively correlating with CARS (r = 0.330, P = 0.004), supporting our findings. Similarly, Li et al[29] reported lower SF in severe ASD and a significant negative correlation between CARS and SF, consistent with our results. These findings suggest that SF and BDNF jointly influence ASD progression and may exhibit an antagonistic relationship.

We further assessed the predictive performance of both markers for ASD severity prediction. SF and BDNF exhibited AUCs of 0.703 and 0.744, respectively, with combined use improving the AUC to 0.823. SF showed the highest specificity (71.43%; cut-off: 18.34 nmol/L), whereas BDNF showed superior sensitivity (84.00%; cut-off: 16.62 pg/mL). Combined analysis achieved higher sensitivity (84.00%) and specificity (71.43%) (cut-off = 0.40). In Jasim Tuama Ali et al[30], BDNF was identified as a potential biomarker for CARS diagnosis (AUC = 0.821), re-confirming its predictive value in ASD. Our regression analysis indicated that SF acted as a protective factor, whereas elevated BDNF increased ASD risk. Thus, low SF and BDNF overexpression may accelerate disease progression. DeVilbiss et al[31] also associated maternal folic acid status with ASD risk, complementing our results.

This study has several limitations. First, all participants were recruited from a single center with a most sample size (n = 103), which may introduce certain information bias. Second, the absence of a 3-5-year follow-up limits assessment of long-term predictive value; future follow-ups could clarify the prognostic potential of SF and BDNF. Lastly, the lack of mechanistic studies on SF and BDNF pathogenesis in ASD constrain interpretation. Future in vitro and in vivo investigations are warranted to elucidate underlying mechanisms and identify therapeutic targets.

CONCLUSION

Low SF and high BDNF levels, characteristic of ASD-affected children, correlate with illness severity and treatment response. Collectively, these biomarkers may serve as potential indicators for disease assessment and represent key components in the complex pathophysiology of ASD.