Autism spectrum disorder: difficulties in diagnosis and microRNA biomarkers

Abstract

We performed a PubMed search for microRNAs in autism spectrum disorder that could serve as diagnostic biomarkers in patients and selected 17 articles published from January 2008 to December 2023, of which 4 studies were performed with whole blood, 4 with blood plasma, 5 with blood serum, 1 with serum neural cell adhesion molecule L1-captured extracellular vesicles, 1 with blood cells, and 2 with peripheral blood mononuclear cells. Most of the studies involved children and the study cohorts were largely males. Many of the studies had performed microRNA sequencing or quantitative polymerase chain reaction assays to measure microRNA expression. Only five studies had used real-time polymerase chain reaction assay to validate microRNA expression in autism spectrum disorder subjects compared to controls. The microRNAs that were validated in these studies may be considered as potential candidate biomarkers for autism spectrum disorder and include miR-500a-5p, -197-5p, -424-5p, -664a-3p, -365a-3p, -619-5p, -664a-3p, -3135a, -328-3p, and -500a-5p in blood plasma and miR-151a-3p, -181b-5p, -320a, -328, -433, -489, -572, -663a, -101-3p, -106b-5p, -19b-3p, -195-5p, and -130a-3p in blood serum of children, and miR-15b-5p and -6126 in whole blood of adults. Several important limitations were identified in the studies reviewed, and need to be taken into account in future studies. Further studies are warranted with children and adults having different levels of autism spectrum disorder severity and consideration should be given to using animal models of autism spectrum disorder to investigate the effects of suppressing or overexpressing specific microRNAs as a novel therapy.

Introduction

Autism spectrum disorder (ASD) is a neurological dysfunction emerging during childhood and persisting throughout the whole of life (Lord et al., 2018; Richards et al., 2020). Onset typically begins in infancy, with gradual development noticeable by 18 months of age, and diagnosis attainable by 24 months (Zhang et al., 2020). ASD is mainly related to genetic factors, maternal pregnancy, perinatal factors, neurobiological factors, infection and immune factors, and nutritional factors (Alvarez-Arellano et al., 2020). The primary symptoms of children with ASD are social interaction disorder, communication disorder, and stereotyped repetitive behaviors and interests (Kodak and Bergmann, 2020; Liu et al., 2022). In 2020, the estimated prevalence of ASD in 8-year-old children was 1 in 36, with a male-to-female ratio in ASD of approximately 4:1. These estimates are higher than previous estimates from 2018 to 2020 (Maenner et al., 2021, 2023). ASD diagnosis is based on behavioral manifestations and developmental situations (Smith et al., 2019). Early behavior reinforcement intervention can significantly influence development, particularly in behavior, adaptability, and communication aptitude (Landa, 2018). Promoting early detection, verification, and treatment is recommended to improve children’s socially acceptable behavior and reduce or eradicate negative behavior.

Although symptoms of ASD appear in infancy, the average age of children diagnosed with ASD is approximately 4–5 years old (National Collaborating Center for Mental Health, UK, 2012), and the effectiveness of treatments is directly related to the time of diagnosis (Elder et al., 2017). Currently, clinical diagnosis of ASD is based on the Diagnostic and Statistical Manual of Mental Disorders-5 (DSM-5) (American Psychiatric Association, 2013). Such diagnosis focuses on assessing patients’ behaviors, but lacks quantifiable indicators and cannot accurately identify mild or non-typical autism or autism in very young patients (Weiss et al., 2009). Standardized tests such as the Autism Diagnostic Observation Schedule Second Edition or Autism Diagnostic Interview-Revised (Lord et al., 2012) are often used for the diagnosis of ASD. However, the high diversity of the clinical symptoms in ASD makes it difficult to diagnose, particularly in early childhood [National Collaborating Centre for Mental Health (UK), 2012]. A psychotherapist, a neurologist or a psychiatrist assesses the patient using an available test. This assessment is performed on the basis of observations and questioning of the patient or relatives. Using such an approach, the assessments of the same patient by different doctors can vary considerably. Children without delay in language development, or who are females, often receive later diagnoses (Maenner et al., 2020). A substantial number of cases of ASD are overlooked, even until adulthood (Lai and Baron-Cohen, 2015). At present, there are no dependable biomarkers of ASD. Identifying the cause of ASD could result in beneficial medical advancements, expert advice for both medical professionals and parents, and, most importantly, early detection and support for affected children to promote their optimal development.

There is an urgent need to find useful and reliable biomarkers to facilitate the diagnosis of autism. In the past decade, searching for genetic biomarkers in ASD has received considerable attention. Numerous related genes have been reported, including NRXN1, SHANK3, SHANK2, MECP2, SNC2A, CHD8, DYRK1A, POG2, GRIN2B, KATNAL2, NLGN3, NLGN4, CNTN4, CDH10, CDH9, and SEMA5A (Sanders et al., 2012; Wakefield, 2016; Krämer et al., 2020). Unfortunately, only 10%–30% of ASD cases have been reported with known genetic deficits (Vorstman et al., 2017; Turner and Eichler, 2019; Lin et al., 2021). In the great majority of patients, ASD is thought to be due to complex interactions of genetic and environmental factors during critical periods of brain development. With the diverse effects of such genetic and environmental factors, disease outcomes and therapeutic options are highly heterogeneous in ASD populations (Meek et al., 2013). At present, there are few biomarker(s) for assessing clinical outcomes, responses to intervention, or the underlying effects of genetic/environmental factors in ASD. Many individuals with ASD respond poorly to the first-line behavioral and pharmacological interventions, and the identification of objective biomarkers will be especially important for such individuals.

Epigenetic modifications may directly contribute to the pathogenesis of several neurodevelopmental disorders including ASD (Grafodatskaya et al., 2010; Siniscalco et al., 2013; Huang et al., 2015). Acting at the interface of genes targeting different mechanisms: histone modifications, DNA methylation, chromatin remodeling or noncoding RNA (microRNA), epigenetic modifiers control heritable changes in gene expression without changing the DNA sequence (Delcuve et al., 2009; Abdul et al., 2017). Measuring microRNAs (miRNAs) as diagnostic or treatment biomarkers in body fluids is attractive, and several studies have reported that miRNA expression in the peripheral blood has the potential to be a biomarker in psychiatric disorders (Weber et al., 2010; Gibbons et al., 2020; Khavari and Cairns, 2020). Several miRNAs in the peripheral blood were found to be relevant to ASD (Mundalil Vasu et al., 2014; Hicks and Middleton, 2016). MiRNAs are small, single-stranded non-coding RNA molecules that regulate gene expression of their complementary mRNA targets (Bartel, 2004; Mattick and Makunin, 2006). In the case of ASD, it has been proposed that differentially upregulated miRNAs with high expression levels mostly repress genes associated with neuronal and synaptic dysfunction, while differentially downregulated miRNAs lead to abnormal activation of genes involved in inflammatory and compensatory processes (Wu et al., 2016). The intricate etiology of ASD is often additionally complicated by the occurrence of comorbid disorders (e.g., gastrointestinal, and immune alterations, sleep disturbances, and other neurological disorders such as epilepsy and attention deficit hyperactivity disorder) (Geschwind, 2009; Grandin, 2009; Klintwall et al., 2011), which intervenes with the precision of the diagnosis (Constantino and Charman, 2016). Therefore, novel and easily reproducible molecular tests, which complement the standard diagnostic procedures and/or provide reliable results before the ASD symptoms manifest themselves, are urgently needed. In this review, we have performed a PubMed literature search of miRNAs in blood-based samples of ASD patients to identify potential biomarkers of the disorder. Many other studies had examined miRNA expression in saliva and postmortem brain tissues of ASD patients. Blood-based samples are easily obtainable and would allow possible identification of diagnostic biomarkers at an early stage of the disease, thereby enabling therapeutic intervention.

Search Strategy

We performed a PubMed search for miRNA biomarkers in ASD patients compared to healthy control subjects using the search terms “microRNA biomarker” and “autism spectrum disorder”. The total number of articles found was 70 and they were published from January 2008 to December 2023. Of 50 of these articles that were selected, 17 had involved computational network/signaling pathway analysis, bioinformatics, and saliva analysis, 2 were animal studies, and 10 were reviews. Also, 4 studies were excluded as they involved patients having Fragile X syndrome, tuberous sclerosis complex, attention deficit hyperactivity disorder, or analysis of postmortem cerebellar cortex. For the present review, 4 studies were performed with whole blood, 4 with blood plasma, 5 with blood serum, 1 with serum neural cell adhesion molecule L1 (L1CAM)-captured extracellular vesicles (LCEVs), 1 with blood cells, and 2 with peripheral blood mononuclear cells (PBMCs), giving a total of 17 articles (Figure 1). ASD in adults was diagnosed in accordance with the Diagnostic and Statistical Manual of Mental Disorders 4^th or 5^th edition criteria (American Psychiatric Association, 2013). Other tests that were used in adults included the Wechsler Adult Intelligence Scale, the Autism Diagnostic Observation Scale (ADOS), and the Social Responsiveness Scale. Diagnosis of ASD in children was made after interviews with the parents by certified psychiatrists, and clinical examinations with the use of ADOS and/or the Autism Diagnostic Interview-Revised (Steiner et al., 2012), Childhood Autism Rating Scale (CARS) (Schloper et al., 1980), and Gilliam Autism Rating Scale, adhering to DSM-5 criteria. The Aberrant Behavior Checklist (Aman et al., 1985) and the Children’s Sleep Habits Questionnaire (Owens et al., 2000), were used to evaluate ASD children for their behavioral symptoms and sleep habits, respectively. Information on cognitive ability and adaptive skills was obtained from previous school evaluation records performed within 1 year of enrolment in the study, using standard measures such as the Woodcock-Johnson III test (for cognitive ability) (Vought and Dean, 2011) and Vineland Adaptive Behavior Scale (for adaptive skills) (Sparrow and Chicetti, 1985). The Structured Clinical Interview for DSM-4 (Kübler, 2013) was used to exclude participants having comorbid psychiatric illnesses.

Figure 1.

Flow diagram to indicate how the articles were obtained for the review.

ASD: Autism spectrum disorder; LCEVs, L1CAM-captured extracellular vesicles; PBMCs: peripheral blood mononuclear cells.

MicroRNAs in Autism Spectrum Disorder

The studies reviewed are summarized as follows.

Whole blood

Hosokawa et al. (2023) recruited a discovery cohort of 6 ASD adult patients and 6 healthy adults as controls, together with a replication cohort of 20 ASD adults and 20 adult controls. The ASD and control subjects were Japanese and had no blood relationship. Total RNA was extracted from whole peripheral blood samples. No significant differences were found in the demographic data between the ASD and control subjects in the replication cohort. Based on previous reports of miRNAs in psychiatric illnesses (Gibbons et al., 2020), 10 miRNAs were chosen for measurement in the discovery cohort using qRT-PCR. Of those miRNAs, the expression of miR-15b-5p level was higher in ASD subjects than in controls but was not statistically significant. No significant differences were found in the expression level of the other miRNAs: miR-15a-5p, -19b-3p, -27a-3p, -106b-5p, -320-5p, -320a-3p, -451a, -494-5p, and -494-3p. In the replication cohort by qRT-PCR, miR-15b-5p expression levels were significantly higher in ASD subjects than in the controls. No significant differences were found in the expression levels of the other miRNAs. TGFBR3 and MYBL.1 mRNA expression levels were measured and no significant differences were found between the ASD and control subjects. A significant correlation between gender and miR-15b-5p occurred in the ASD and control subjects.

Seven ASD children and four healthy children were recruited by Vaccaro et al. (2018). The ASD subjects were in the same autism severity group using two well-validated clinical tests: the CARS and Autism Diagnostic Observation Schedule-Generic (ADOS-G) (Lord et al., 1999). Clinical diagnosis of the healthy children excluded the presence of psychiatric disorders, and the presence of ASD subjects in the family. The 26 candidate miRNAs evaluated in this study were based on previous studies with patients having sleep disturbances (Davis et al., 2007) or some neurodegenerative diseases (Barbato et al., 2009; Eacker et al., 2009). By RT-qPCR, of the 26 miRNAs evaluated, 7 were statistically altered in ASD patients compared to controls: 4 were upregulated (miR-34c-5p, -145-5p, -92a-2-5p, and -199a-5p) and 3 downregulated (miR-19b-1-5p, -27a-3p, and -193a-5p).

Peripheral blood samples were obtained from 30 ASD adult subjects and 30 healthy adults by Nakata et al. (2019). In addition to DSM-5 criteria, all participants were assessed using the Wechsler Adult Intelligence Scale (Ryan et al., 2003), the ADOS, and the Social Responsiveness Scale-Second Edition (Constantino et al., 2003). All participants were Japanese and took no medication for at least 3 months prior to collection of blood samples. No participants exhibited intellectual disability although there were small significant differences in full and performance intelligence quotient (IQ) scores between ASD patients and controls. The ADOS and the Social Responsiveness Scale-Second Edition were significantly higher in ASD patients compared to controls. By Agilent microarray analysis, 2 miRNAs were significantly upregulated (miR-328-3p and -4515) and 16 significantly downregulated (miR-6126, -6780a-5p, -1227-5p, -3156-5p, -4716-3p, -144-3p, -3127-5p, -5581-5p, -6756-5p, -6767-5p, -7977, -4486, -6734-5p, -4653-3p, -6085, and -874-3p) in ASD patients compared to controls. In particular, miR-6126, -3156-5p, -1227-5p, -6780a-5p, and -328-3p were the most significantly dysregulated in ASD. By qRT-PCR, miR-6126 was significantly downregulated in ASD compared to controls. The expression of miR-6126 was significantly negatively correlated with Social Responsiveness Scale-Second Edition total score, but with no significant correlation between the expression of miR-6126 and IQ scores. These results suggested that miR-6126 expression is associated with social severity of ASD, independent from IQ.

Hirsch et al. (2018) recruited 7 ASD children and 7 healthy children. Autistic individuals who presented secondary autism or autism as an associated feature of an identified genetic condition (Fragile X Syndrome, Rett Syndrome, Angelman Syndrome, Prader-Willi Syndrome, Smith-Lemli-Opitz Syndrome, and Tuberous Sclerosis) were excluded. By RT-PCR assay, there were significant increases in the expression of miR-134-5p and miR-138-5p in peripheral blood samples of ASD children compared with control subjects. There were no differences in the levels of the 11 remaining miRNAs between the two groups.

Blood plasma

Mehmetbeyoglu et al. (2023) obtained blood plasma samples from 45 ASD children and 21 healthy children, as well as from 33 healthy siblings (1–20 years). All participants were of Turkish origin. By PCR array, 189 miRNAs were statistically significantly different among the study groups. A predefined threshold was introduced to identify miRNAs that exhibited downregulation by at least 90% in comparison with the control group. Using this stringent selection criterion, the study identified 6 miRNAs that were significantly downregulated in ASD patients and their families compared to healthy controls: miR-3613-3p, -150-5p, -19a-3p, -361-5p, -499a-5p, and -126-3p. Receiver operating curve (ROC) analysis of the individual six miRNAs showed an area under the curve (AUC) values ranging from 0.894 to 0.998. The miRNA with the highest AUC score, miR-361-5p, had greater potential in distinguishing between ASD children and controls compared to other miRNAs.

Ivanov et al. (2022) developed a “Silicon-On-Insulator”-based nanosensor and used it to test selected miRNA levels in a small group of 4 ASD children and 3 healthy adults. Upon the analysis of miRNA samples isolated from the plasma of ASD patient (male, 20 years), a decrease occurred in the current flowing through the nasnosensor (NS) sensitized with probe oDNA #4 (complementary to the synthetic analog of miR-494-5p); and a decrease in signal from the NS sensitized with probe oDNA #6 (complementary to the synthetic analog of miR-15b-5p) was observed upon analysis of miRNA samples isolated from the plasma of an ASD patient (male, 6 years). By contrast, in the control experiments with the miRNA isolated from the sample of healthy control subjects, the change in the signal was insignificant (female, 33 years; male, 5 years). The analysis of data obtained indicated that upon repeated analysis of one and the same sample from plasma of ASD patient (male, 9 years), changes in the signal from the NS sensitized with probe oDNA #1 (complementary to the synthetic analog of miR-106a-5p) made up approximately 40%, while the signal was virtually unchanged in the case with NS sensitized with probe oDNA #4 (complementary to the synthetic analog of miR-494-5p). The lower limit of miRNA detection was 1 × 10^–17 M.

Thirty-eight ASD children and 28 typically developing children as controls were recruited by Kichukova et al. (2021). The children in the control group were matched to the age and gender of the ASD children, and the absence of ASD symptoms was determined after a clinical examination and CARS assessment. The participants had not received any medication treatments before blood sampling. Blood plasma was obtained from subjects in a fasting state (> 3 hours without food consumption). A pooled study was performed in which a total of 42 miRNAs were assessed using qRT-PCR, and in which 29 were found to be downregulated, 11 upregulated, and 2 unchanged in ASD children compared to healthy controls. A confirmation step evaluating the expression in samples from individual patients was performed, with the 4 most dysregulated miRNAs selected for such an analysis. By qRT-PCR, a significantly altered expression of miR-500a-5p, -197-5p, -424-5p, and -664a-3p was found in ASD subjects compared with controls. MiR-197-5p (fold change 9.5), -664a-3p (fold change 5.8), and -424-5p (fold change 4.3) were upregulated in ASD children compared to controls, while miR-500a-5p (fold change > 5) was downregulated. By ROC analysis, the diagnostic AUC values for miR-197-5p, miR-500a-5p, miR-664-3p, and miR-424-5p were 0.825, 0.796, below 0.7, 0.756; sensitivities were 86.1%, 77.8%, 25%, and 88.9%; and specificities 78.6%, 92.9%, 100%, and 75%, respectively. Thus, miR-197-5p, miR-500a-5p, and miR-424-5p were good ASD predictors. For the combination miR-197-5p and miR-500a-5p, the AUC value increased to 0.975 and the sensitivity and specificity values were > 90%.

In an earlier study, Kichukova et al. (2017) determined miRNA expression in the blood plasma of 30 ASD children and 30 healthy controls who were age- and gender-matched to the ASD group. All participants were Bulgarians. None of the participants had received any medications before blood sampling. By qRT-PCR pooling assay, in preliminary expression experiments, 42 miRNAs were selected as candidate biomarkers. 29 miRNAs were found to be downregulated (miR-589-3p, -6849-3p, -3135a, -15a-5p, -328-3p, -183-5p, -3674, -96-5p, -3687, -6799-3p, -587-3p, -504-5p, -576-5p, -486-3p, -3909, let-7i-3p, -29c-5p, -301a-3p, 3064-5p, -145-5p, -424-5p, -193b-3p, -487b-3p, -197-5p, -500a-5p, -664b-3p, -20b-3p, -671-3p, and -199a-5p) and 11 were upregulated (miR-4489, -8052, -106b-5p, -142-3p, -3620-3p, -365a-3p, -664a-3p, -374b-5p, -18b-3p, -619-5p, and -210-5p) while 2 (miR-134-5p and -128-3p) did not show any expression changes in ASD children compared with controls. Using qRT-PCR pooled analysis, the expression of the 8 most dysregulated miRNAs was confirmed. The levels of only 3 miRNAs (miR-365a-3p, -619-5p, and -664a-3p) in ASD patients were markedly higher than in controls, and the levels of the other 5 (miR-3135a, -328-3p, -197-5p, -500a-5p, and -424-5p) in ASD patients were lower than in controls.

Blood serum

Jiang et al. (2023) recruited 60 ASD children and 58 healthy children as controls. Patients were of Han nationality. Patients with other neuropsychiatric diseases such as Wright’s syndrome, fragile X syndrome, epilepsy, schizophrenia, obsessive-compulsive disorder, affective disorder, or congenital heart disease, trisomy 21 syndrome, and other congenital diseases were excluded from the study. Peripheral blood samples were collected and the serum was obtained by centrifugation. By qPCR, the relative expression level of lncRNA PVT1 in the ASD group was decreased compared to controls. MiR-21-5p was overexpressed in ASD patients. By Pearson analysis, the increased content of miR-21-5p was inversely proportional to the level of PVT1 in ASD patients. Luciferase reporter gene assay indicated that PVT1 directly targeted miR-21-5p. By ROC analysis to determine the predictive value of PVT1 and miR-21-5p in ASD patients, the AUC for PVT1 was 0.848 with a sensitivity of 85.0% and specificity of 79.3%. For miR-21-5p, the AUC was 0.921, sensitivity 78.3%, and specificity 91.4%. When PVT1 was combined with miR-21-5p, the AUC increased to 0.954, suggesting that miR-21-5p and PVT1 combination had high accuracy in predicting ASD patients.

Blood serum samples were obtained from 45 ASD children and 21 healthy children by Ozkui et al. (2020). Also, serum samples from 33 healthy siblings and 74 parents were included in the study. The ASD group included 27 children with autism spectrum disorder and 18 children with pervasive developmental disorder-not otherwise specified (PDD-NOS). Twenty-two of the children were diagnosed with intellectual disability, four of whom were diagnosed with epilepsy with EEG abnormalities, and two of them also exhibited attention deficit hyperactivity disorder (ADHD) among the autistic cases. Eleven children were diagnosed with mental retardation, 5 of who were diagnosed with ADHD among the pervasive developmental disorder-not otherwise specified cases. All the participants were of Turkish origin. MiRNA detection and validation were performed by qPCR. While most of the miRNAs tested exhibited normal to lower levels in the patients, expression levels in the range of 1% to 10% of the healthy controls were found for 6 of them (miR-3613-3p, -150-5p, -126-3p, -361-5p, -19a-3p, and -499a-5p). Intermediate levels for the same 6 miRNAs occurred in the serum of the healthy siblings, fathers, and mothers of the affected children. Although the levels of these miRNAs were higher than in the affected children, they were significantly lower by 40 to 50% compared with those in genetically unrelated controls.

Jyonouchi et al. (2019) recruited 105 ASD children/young adults and 35 healthy children/young adults. Venous blood samples were collected for serum separation. Only one sample was obtained for control subjects. For selected ASD subjects (n = 10), samples were obtained at two time points to assess the variability of serum miRNA levels. By high-throughput miRNA sequencing, 4 upregulated and 14 downregulated miRNAs with > 2-fold differences of expression (DE) were found between ASD and control serum samples. The levels of 27 miRNAs in the ASD sera in comparison to controls were analyzed when expressed as TMM normalized readouts (CPM) (trimmed mean of M-values). 18 miRNAs revealed significant DEs in the total ASD samples compared to controls (upregulated: miR-206, -184, -223-5p, -4732-5p, and -193b-5p; downregulated: miR-576-3p, -193a-5p, -27a-5p, -379-5p, -134-5p, -574-3p, -382-5p, -7-5p, -103a-3p, -378a-3p, -3614-5p, -873-3p, and -433-3p). An additional 9 miRNAs revealed significant DEs in one or more ASD subgroups compared to controls with readouts (> 7.0) ASD high IL-1β/IL-10 ratio group, upregulated: miR-423-5p, -483-5p, -320b, and -320d; downregulated: miR-20a-5p. ASD normal IL-1β/IL-10 ratio group, downregulated: miR-370-3p. ASD low IL-1β/IL-10 ratio group, downregulated: miR-100-5p, -99b-5p, and -93-5p. In these 27 miRNAs, serum miRNAs of 2, 4, 2, 2, and 3 differed according to gender, use of neuroleptics, ADHD medications, anti-epileptic drugs (AEDS), and selective serotonin receptor inhibitors (SSRIs), respectively. Only 2 miRNAs (miR-193b-5p, miR-320b) revealed significant associations between miRNA readouts and ages in both ASD subjects and controls. The clinical parameters did not affect differences between ASD subjects and controls by co-variance analysis except for miR-206 which revealed the effects of the use of AEDs: ASD subjects with AEDS (n = 30) had lower miR-206 levels compared to ASD subjects without AEDs (n = 86). In 7 of the 27 miRNAs there were nominally significant differences in serum miRNA levels with ASD severity (miR-27a-5p, -382-5p, -223-5p, -103a-3p, -20a-5p, -370-3p, and -93-5p).

Thirty ASD children and 30 typically developing children, gender- and age-matched to the ASD group, were recruited by Popov et al. (2018). Inspection of all children in the control group for the absence of autistic features was done by clinical examination and CARS. Children with known infections, and oncological, metabolic, or genetic conditions were excluded from the study. No children were receiving any drug therapy when they were recruited. Peripheral blood was collected from fasting participants (> 3 hours without a meal) and the serum was obtained. By qRT-PCR, the relative expression levels of miR-3135a and miR-328-3p were markedly lower in ASD patients than in controls. Using ROC analysis, for miR-3135a and miR-328-3p the AUC values were 0.828 and 0.858, sensitivities 76.3% and 78.9%, and specificities 88.9% and 88.9%, respectively. The combined ROC analysis had a greater diagnostic value than individual miRNAs in ASD with AUC 0.858.

Blood serum samples were analyzed by Mundalil Vasu et al. (2014) from 55 ASD children and 55 healthy children, with age- and gender-matched to ASD. All of the participants were Japanese and had not received any drug treatment for ASD. Venous blood samples were collected between 11.00 am and noon from each subject and used for serum separation. In preliminary microarray screening, 14 miRNAs had altered expression in the ASD samples compared to control samples: miR-151a-3p, -181b-5p, -320a, -328, -433, -489, -572, and -663a (all downregulated), miR-101-3p, -106b-5p, -19b-3p, -195-5p, -130a-3p, and -27a-3p (all upregulated). In validation by qPCR, consistent results were observed for all miRNAs except miR-27a-3p. MiR-151a-3p, -181b-5p, -320a, -328, -433, -489, -572, and -663a were downregulated, while miR-101-3p, -106b-5p, -19b-3p, -195-5p, and -130a-3p were upregulated in ASD subjects compared to controls. The differences in the expression of miRNAs between the ASD and control groups remained significant even after adjusting for age and gender. ROC analysis showed significant diagnostic values of the 13 differentially expressed miRNAs for ASD. High values of AUC, sensitivity, and specificity were observed for 5 miRNAs: miR-181b-5p, -320a, -572, -130a-3p, and -19b-3p.

Cell adhesion molecule L1–captured extracellular vesicles

Qin et al. (2022) recruited 100 ASD children and 60 typically developing children, who were from the same region as the ASD children to minimize the influence of different environments. The severity degree of ASD was mild 17%, moderate 64%, and severe 19%. Those children who had any types of infection or disease within 2 weeks before the time of examination or taking any medicines were excluded. Venous blood samples were collected and the serum was separated. 25 µL of each serum sample was collected and every 20 samples were pooled into one subgroup, giving 5 ASD subgroups and 3 control subgroups for lncRNA microarray detection and RNA sequencing. To avoid bias caused by gender differences, the proportion of males in each subgroup was the same (90%). L1CAM-captured EVs in the pooled sera in each group were isolated. Nanoparticle tracking analysis showed a higher LCEVs concentration in the ASD group than in the control group. By qRT-PCR, a total of 4310 mature miRNAs were examined in the serum LCEVs, among which 150 were present in all subgroups. The miRNA with the highest concentration across all the subgroups was hsa-miR-21-5p_R+1. Unpaired two-tailed Student’s t-test identified 10 miRNA sequences with significant differences between ASD and controls. The sequence GATTTCTTCCCAGTGCTCTGA was aligned to two pre-miRNAs and was given two names: mmu-miR-6240-p3_1ss8GT and mmu-miR-6240-p5_1ss8GT. Of these 11 miRNAs, the one with the largest variation was PC-3p-38497_124, which was markedly upregulated (FC = 20.3) in ASD compared to controls. The miRNA with the most significant change was PC-5p-139289_26 which was absent in the control subgroups. Two other miRNAs (PC-3p-275123_15 and PC-5p-149427_24) were upregulated in ASD, and another 7 miRNAs (e.g., hsa-miR-193a-5p and mmu-miR-6240-p3_1ss8GT) were downregulated.

Blood cells

Fifty ASD children and healthy controls were recruited by Eftekharian et al. (2019). Peripheral blood samples were collected from all participants between 10 a.m. and noon in EDTA tubes. Total blood cells were subjected to RNA extraction, and the relative transcript levels of genes were compared between study groups using RT-PCR. The expression levels of genes HPRT1 and RNU68 were similar between ASD and controls. Transcript levels of apoptosis-related genes BCL2, CASP8, and hsa-miR-29c-3p were significantly lower in ASD patients compared with normal children. When the gender of participants was considered, the difference in transcript levels was significant only in male subjects. No significant correlation was found between the relative expression of genes and the age of ASD subjects. However, expression of hsa-miR-17-5p and hsa-miR-20a-5p were inversely correlated with age in controls. BCL2 expression was significantly correlated with an expression of both CASP8 and CASP2 genes in both ASD and controls. CASP8 and CASP2 expressions were correlated only in controls. hsa-miR-29a-3p expression was correlated with expression levels of other miRNAs only in controls. The most significant pairwise correlation was between hsa-miR-20a-5p and hsa-miR-17-5p. As transcript levels of BCL2, CASP8, and hsa-miR-29c-3p were significantly different between ASD patients and controls, the diagnostic power of these genes in differentiating disease status was evaluated by ROC curve analysis. BCL2 and hsa-miR-29c-3p had 100% sensitivity and 92% specificity in ASD diagnosis. The diagnostic power of the combination of transcript levels of these three genes was estimated to be 78% based on the calculated AUC value of 0.78.

Peripheral blood mononuclear cells

Atwan et al. (2020) recruited 37 ASD children and 40 healthy children. ASD patients were from two cities in Iran, Amol, and Tehran, while controls were from Amol. Peripheral blood samples were collected in EDTA tubes and PBMC isolation was performed using Ficol (Tayebi et al., 2017). Total RNA extraction was performed by a precipitation method (Tayebi et al., 2017). Over 85% of brain development occurs in the first 5 years, after which it can be considered as a new period in brain function and development. For ASD patients 29.7% were aged < 5 years, 70.3% > 5 years, while for controls 12.5% were aged < 5 years, 87.5% > 5 years. By RT-PCR, no statistically significant differences were found between the two groups in the pattern of gene expression of BCL-2, IL-6, miR-16, miR-181b-5p, and miR-23a-3p. Also, the expression of the target genes was not significantly different between the genders. The expression of the miR-181b-5p target gene was significantly different between Amol and Tehran, with it being downregulated in Tehran (n = 20) compared to Amol (n = 17). Patients were separated into three groups (mild, moderate, severe) based on their symptoms and their scores in diagnostic tests. It was shown that the expression of the IL-6 target gene increased with the severity of the disorder in ASD patients (n = 37). IL-6 target gene expression was significantly increased in ASD patients compared to controls. ROC analysis showed for IL-6 target gene expression, AUC was 0.681, sensitivity 70.3%, and specificity 67.5%. For miR-23a target gene expression, AUC was 0.703, sensitivity 70.3%, and specificity 65.0%.

Sixty ASD children/adults and 27 children/adults as controls were recruited by Jyonouchi et al. (2017). Venous blood samples were obtained. Only 1 blood sample was obtained for control subjects. For ASD subjects with fluctuating behavioral symptoms and varying GI symptoms, it was attempted to obtain two samples, one when behavioral symptoms were what was considered their baseline and another when parents reported exacerbation of behavioral symptoms. PBMCs were isolated by Ficoll-Hypaque density gradient centrifugation. In selected ASD subjects (n = 23), PBMC samples were obtained 2–3 times. Each sample was analyzed for both cytokine production and miRNA expression, along with an evaluation of behavioral symptoms using the Aberrant Behavior Checklist checklist. Control cells revealed the similar tight ranges of IL-1β/IL-10 ratios that have been reported previously. ASD monocytes examined in this study revealed a much higher frequency of high/low IL-1β/IL-10 ratios that were either higher or lower than control cells. High IL-1β/IL-10 ratio > +2SD than control cells and at least one culture condition and/or > +1SD under more than two culture conditions. Normal IL-1β/IL-10 ratio fall into -1SD < IL-1β/IL-10 ratios < +1SD under all the culture conditions or +1SD < IL-1β/IL-10 ratios < +2SD under only one culture condition. Low IL-1β/IL-10 ratio fall into < -1SD under at least one culture condition.

For miRNA sequencing, ASD cells submitted included cells with a high ratio group (n = 43), low ratio (n = 18), and normal ratios (n = 47). ASD cells with high IL-1β/IL-10 ratios revealed upregulated expression of multiple miRNAs as compared to other ASD groups and control cells. ASD cells with normal IL-1β/IL-10 ratios revealed little differences from controls. ASD cells with low IL-1β/IL-10 ratios revealed upregulation of one miRNA and downregulation of two miRNAs. ASD cells high IL-1β/IL-10 ratios vs. normal group: upregulated miR-181a-1, -181a-2, -93, -342, -425, -223, -181b-2, -660, -1248, -21, -191, -126, -484, -409, -451a, -17, -99b, -30a, and -181b-1; downregulated none. ASD cells high IL-1β/IL-10 ratios vs. low group: upregulated miR-342, -93, -223, -181a-1, -181a-2, -21, -484, -181b-2, -126, -30a, -660, -30d, -26a-1, let-7a-1, -17, -181b-1, -26a-2, -19b-1, -19b-2, -451a, and let-7a-2; downregulated none. ASD cells high IL-1β/IL-10 ratios vs. control cells: upregulated miR-181a-1, -181a-2, -1248, -93, -181b-2, -223, and -342; downregulated none. ASD cells normal IL-1β/IL-10 ratios vs. low group: upregulated none; downregulated none. ASD cells normal IL-1β/IL-10 ratios vs. control cells: upregulated none; downregulated miR-342. ASD cells low IL-1β/IL-10 ratios vs. control cells: upregulated miR-1248; downregulated miR-425, -342. ASD all combined vs. control cells: upregulated none; downregulated none.

Those miRNAs having altered expression in whole blood, blood plasma, blood serum, serum neural cell adhesion molecule L1-captured extracellular vesicles, blood cells, and peripheral blood mononuclear cells in ASD children, and in whole blood in ASD adults are summarized in Additional Table 1. In addition, miRNAs having altered expression in blood serum and peripheral blood mononuclear cells in children/adults are included. The group sizes, gender distribution, and mean ages are shown.

Additional Table 1.

Alterations of miRNA expression in whole blood, blood plasma, blood serum, serum neural cell adhesion molecule L1 (LlCAM)-captured extracellular vesicles, blood cells, peripheral blood mononuclear cells of patients with autism spectrum disorder

References	Analysis method	Comparison; number, gender and mean age of subjects			Altered miRNA expression in ASD
Whole blood
Hosokawa et al., 2023	RT-PCR	ASD 13M/7F 30.0±8.2 yr	vs.	Control 13M/7F 30.9±8.3 yr	Upregulated: miR-15b-5p
Vaccaro et al., 2018	RT-PCR	ASD 7M/0F 7.5±2.5 yr	vs.	Control 4M/0F 7.5±2.5 yr	Upregulated: miR-34c-5p, -145-5p, -92a-2-5p, -199a-5p Downregulated: miR-19b-1-5p, -27a-3p, -193a-5p
Nakata et al., 2019	RT-PCR	ASD 18M/12F 28.4±7.3 yr	vs.	Control 18M/12F 28.4±8.4 yr	Downregulated: miR-6126
Hirsch et al., 2018	RT-PCR	ASD male [5-15 yr]	vs.	Control [5-15 yr]	Upregulated: miR-134-5p, -138-5p
Blood plasma
Mehmetbeyoglu et al., 2023	PCR	ASD 31M/14F [2-13 yr]	vs.	Control 10M/11F [3-16 yr]	Downregulated: miR-3613-3p, -150-5p, -19a-3p, -361-5p, -499a-5p, -126-3p
Ivanov et al., 2022	Nanosensor 3 individual ASD samples	ASD 4M/0F 10.5±6.5 yr	vs.	Control 1M/2F 29.0±22.3 yr	Upregulated: miR-494-5p, -15b-5p, -106a-5p
Kichukova et al., 2021	RT-PCR	ASD 30M/8F children	vs.	Control 22M/6F children	Upregulated: miR-197-5p, -424-5p, -664a-3p Downregulated: miR-500a-5p
Kichukova et al., 2017	RT-PCR	ASD 24M/6F 6.9 yr [3-11 yr]	vs.	Control 30 gender/age matched to ASD	Upregulated: miR-365a-3p. -619-5p, -664a-3p Downregulated: miR-3135a, -328-3p, -197-5p, -500a-5p, -424-5p
Blood serum
Jiang et al., 2023	qPCR	ASD 60 7.1±2.6 yr	vs.	Control 58 7.8±2.7 yr	Upregulated: miR-21-5p gene expression which is suggestive of downregulated expression of miR-21-5p
Ozkui et al., 2020	qPCR	27ASD/18 PDD-NOS 31M/14F [2-13 yr]	vs.	Control 10M/11F [3-16 yr]	Downregulated: miR-3613-3p, -150-5p, -126-3p, -361-5p, -19a-3p, -499a-5p
Jyonouchi et al., 2019b	miRNA sequencing	ASD 88M/17F 11.3±5.4 yr [2-22 yr]	vs.	Control 27M/8F 15.5±7.8 yr [4-30 yr]	Upregulated: miR-206, -184, -223-5p, -4732-5p, -193b-5p Downregulated: miR-576-3p, -193a-5p, -27a-5p, -379-5p, -134-5p, -574-3p, -382-5p, -7-5p, -103a-3p, -378a-3p, -3614-5p, -873-3p, -433-3p
Jyonouchi et al., 2019b	miRNA sequencing	ASD high IL-1β/IL-10	vs.	Control	Upregulated: miR-423-5p, -483-5p, -320b, -320d Downregulated: miR-20a-5p
Jyonouchi et al., 2019b	miRNA sequencing	ASD normal IL-1β/IL-10	vs.	Control	Downregulated: miR-370-3p
Jyonouchi et al., 2019b	miRNA sequencing	ASD low IL-1β/L-10	vs.	Control	Downregulated: miR-100-5p, -99b-5p, -93-5p
Popov et al., 2018	RT-PCR	ASD 24M/6F [3-11yr]	vs.	Control 30 gender/age matched to ASD	Downregulated: miR-3135a, -328-3p
Mundalil Vasu et al., 2014	qPCR	ASD 48M/7F 11.3±2.5 yr [6-16 yr]	vs.	Control 41M/14F 11.3±2.4 yr [6-16 yr]	Upregulated: miR-101-3p, -106b-5p, -19b-3p, -195-5p, -130a-3p Downregulated: miR-151a-3p, -181b-5p, -320a, -328, -433, -489, -572, -663a
Serum neural cell adhesion molecule L1 (LlCAM)-captured extracellular vesicles (LCEVs)
Qin et al., 2022	RT-PCR	ASD 90M/10F 3.5 yr [3-6 yr]	vs.	Control 54M/6F 3.5 yr [3-6 yr]	Upregulated: PC-3p-38497_124, PC-5p-139289_26, PC-3p-275123_15, PC-5p-149427_24 Downregulated: miR-193a-5p
Blood cells
Eftekharian et al., 2019	RT-PCR	ASD 38M/12F 6.0±1.5 yr [4-9 yr]	vs.	Control 37M/13F 6.0±1.7 yr [4-13 yr]	Downregulated: transcript level of miR-29c-3p only in male children which is suggestive of upregulated expression of miR-29c-3p in male children
Peripheral blood mononuclear cells (PBMCs)
Atwan et al., 2020	RT-PCR	ASD Amol 11M/6F Tehran 15M/5F 7yr [3-15 yr]	vs.	Control Amol 27M/13F 9yr [4-12 yr]	Downregulated: miR-181b-5p target gene in ASD children from Tehran (n=20) vs. Amol (n=17) which is suggestive of upregulated expression of miR-181b-5p
Jyonouchi et al., 2017	miRNA sequencing	ASD 52M/16F median 11.8 yr [3-27 yr]	vs.	Control 16M/11F median 10.1 yr [4-27 yr]	No change in miRNA expression for total ASD cells vs. control cells
Jyonouchi et al., 2017	miRNA sequencing	ASD high IL-1β/IL-10 43	vs.	Control 27	Upregulated: miR-181a-1, -181a-2, -1248, -93, -181b-2, -223, -342
Jyonouchi et al., 2017	miRNA sequencing	ASD highIL-1β/IL-10 43	vs.	ASD normal ratio 47	Upregulated: miR-181a-1, -181a-2, -93, -342, -425, -223, -181b-2, -660, -1248, -21, -191, -126, - 484, -409, -451a, -17, -99b, -30a, -181b-1
Jyonouchi et al., 2017	miRNA sequencing	ASD high IL-1β/IL-10 43	vs.	ASD low ratio 18	Upregulated: miR-342, -93, -223, -181a-1, -181a-2, -21, -484, -181b-2, -126, -30a, -660, -30d, - 26a-1, let-7a-1, -17, -181b-1, -26a-2, -19b-1, -19b-2, -451a, let-7a-2
Jyonouchi et al., 2017	miRNA sequencing	ASD normal ratio 47	vs.	Control 27	Downregulated: miR-342
Jyonouchi et al., 2017	miRNA sequencing	ASD low ratio 18	vs.	Control 27	Upregulated: miR-1248 Downregulated: miR-425, -342

Ages are given as means (± SD, in most cases) unless otherwise specified or age range [in brackets]. All the miRNAs listed are human miRNAs (hsa-miRs). ASD: Autism spectrum disorder; F: female; IL: interleukin; M: male; miRNA: microRNA; PDD-NOS: pervasive developmental disorder not otherwise specified; qPCR: quantitative polymerase chain reaction; RT-PCR: real time polymerase chain reaction; yr: year(s);

Discussion

ASD in children is difficult and challenging to diagnose accurately, and delayed or inappropriate treatment can have undesirable life-long consequences for most of the affected children (Hus and Segal, 2021). To be diagnosed with ASD, the individuals must first have persistent problems in social communication and interaction. This includes deficits in empathy, body language, facial expression, and eye contact; and difficulties or lack of interest in social relationships and making friends. Second, they must exhibit restricted, repetitive patterns of behavior, interests, or activities such as insistence on rigid routines, fixation on certain topics, sensory hypersensitivities such as noise sensitivity, and sometimes hyposensitivities such as a high pain threshold. The problems must be severe enough to cause impairment in their everyday life.

The rate of ASD diagnosis has increased significantly in recent years and may be being over-diagnosed. In addition, there are still children, particularly girls, whose ASD diagnosis is missed until late in their development and schooling. Moreover, ASD-like symptoms can occur to a mild degree in children who are stressed, anxious, and depressed. This could be due to trauma in their early life or insecure attachment, where neglect or abuse has hindered early bonding with parents, or because their mother is suffering from depression or anxiety. Eye contact and a sense of care and safety that allows them to develop their natural social skills may be lacking. ASD symptoms such as social avoidance and restricted interests can also be coping mechanisms for children with other learning difficulties, including intellectual disabilities, dyslexia, and speech and language disorders. The Autism Diagnostic Observation Schedule is a specialized ASD diagnostic test that helps to standardize the diagnostic process, but it is still dependent at some level on the subjective perspectives of clinicians (No author listed, 2016). Furthermore, ASD patients frequently have comorbid conditions (e.g., ADHD, epilepsy, anxiety, depression, and bipolar disorder) and it is difficult to separate whether abnormal behaviors are related to ASD or physical discomfort caused by a co-occurring condition (Bennett, 2017). An objective test based on validated laboratory-based biomarkers is urgently required to improve ASD diagnosis.

MiRNA biomarkers, having been identified for many neurological diseases and disorders (Taguchi and Wang, 2018), are promising drug targets for neurological diseases (Wen, 2016; Titze-de-Almeida et al., 2020). The studies reviewed here had performed miRNA profiling in whole blood (two studies), blood plasma (four studies), blood serum (four studies), serum neural cell adhesion molecule L1-captured extracellular vesicles (one study), blood cells (one study), and peripheral blood mononuclear cells (one study) in children with ASD; and in whole blood (two studies) and blood serum (one study) in adults with ASD. MiRNA profiling was performed in blood serum (one study) and peripheral blood mononuclear cells (one study) in children and adults combined (Additional Table 1). Similar findings were found in two or more separate studies for some miRNAs. With regard to findings in children with ASD, miR-3613-3p, -150-5p, -19a-3p, -361-5p, -499a-5p, and -126-5p were downregulated in blood plasma (Mehmetbeyoglu et al., 2023) and blood serum (Ozhui et al., 2020). Additionally, miR-3135a was downregulated in blood plasma (Kichukova et al., 2017) and blood serum (Popov et al., 2018). Upregulation of miR-664a-3p and downregulation of miR-500a-5p, but conflicting results for miR-197-5p and -424-5p, in blood plasma were found in two studies (Kichukova et al., 2017, 2021). MiR-328 was downregulated in blood serum in two studies (Mundalil Vasu et al., 2014; Popov et al., 2018). There was a potential conflict in that miR-181b-5p was downregulated in blood serum (Mundalil Vasu et al., 2014) but the level of miR-181b-5p target gene was downregulated in peripheral blood mononuclear cells of children from Tehran versus Amol (Atwan et al., 2020) which is suggestive of upregulated expression of miR-181b-5p. For adults with ASD, no similarly altered miRNAs were found in two previous studies (Nakata et al., 2019; Hosokawa et al., 2023). It was noted that miR-15b-5p was upregulated in whole blood of ASD adults (Hosokawa et al., 2023) and in blood plasma of ASD children (Ivanov et al., 2022). Also, miR-433 was downregulated in the blood serum of ASD children/adults (Jyonouchi et al., 2019) and ASD children (Mundalil Vasu et al., 2014). MiR-193a-5p was downregulated in the blood serum of ASD children/adults (Jyonouchi et al., 2019) and in LCEVs of ASD children (Qin et al., 2022). A large number of miRNAs were dysregulated in the blood serum of ASD children/adults (Jyonouchi et al., 2019), but with no similar changes in miRNA expression for peripheral blood mononuclear cells of ASD children/adults (Jyonouchi et al., 2017). Receiver operating curve analysis indicated that several miRNAs had good diagnostic potential for distinguishing ASD patients from control subjects. Included among these in blood plasma of ASD children were miR-3613-3p, -150-5p, -19a-3p, -361-5p, -499a-5p, and -126-3p (with AUC values ranging from 0.894 to 0.998), with miR-361-5p having greater discriminating potential than the other miRNAs (Mehmetbeyoglu et al., 2023). Also, miR-500a-5p, -197-5p, and -424-5p in blood plasma had high diagnostic power in distinguishing ASD children from control subjects (AUC values 0.756 to 0.825) (Kichukova et al., 2021). In blood serum, miR-3135a and -328-3p (AUC values 0.828 and 0.858, respectively) (Popov et al., 2018) and miR-181b-5p, -320a, -572, -130a-3p, and -19b-3p (for miR-181b-5p AUC 0.868, miR-320a AUC 0.906, miR-572 AUC 0.822, miR-130a-3p AUC 0.852, miR-19b-3p AUC 0.822) (Mundalil Vasu et al., 2014) could discriminate between ASD children and controls with high accuracy.

In comparing the results of the studies, a marked difference was noticeable in the sample sizes of some studies, with some being very small. For instance, miRNA expression was measured in whole blood of seven ASD children and four healthy children as controls (Vaccaro et al., 2018). Moreover, miRNA analysis of blood plasma was performed with samples from 45 ASD children and 21 healthy children (Mehmetbeyoglu et al., 2023). MiRNA expression in blood serum was measured in 45 ASD/pervasive developmental disorder-not otherwise specified children and 21 healthy children (Ozkui et al., 2020). Similarly, miRNAs in blood serum were analyzed in 105 ASD children/adults and 35 healthy children/adults as controls (Jyonouchi et al., 2019). For the determination of miRNAs in LCEVs, samples from 100 ASD children and 60 healthy children were used (Qin et al., 2022). MiRNA analysis of PBMCs was measured in 68 ASD children/adults and 27 healthy children/adults as controls (Jyonouchi et al., 2017). Furthermore, some studies had selected only a very small subsample for miRNA profiling or validation. For example, 10 miRNAs in whole blood were selected for measurement in the discovery cohort and replication cohort using qRT-PCR (Hosokawa et al., 2023). Also, 26 miRNAs were evaluated in whole blood using RT-qPCR (Vaccaro et al., 2018), and 13 miRNAs were chosen for measurement in whole blood by RT-PCR (Hirsch et al., 2018). For blood plasma, the 4 most dysregulated miRNAs were selected from 42 for measurement in a validation step (Kichukova et al., 2021), while in a previous study, the expression of the 8 most dysregulated miRNAs of 42 in blood plasma was confirmed by RT-PCR (Kichukova et al., 2017). Only two miRNAs were analyzed in blood serum by qRT-PCR (Popov et al., 2018).

Alterations in the expression levels of miRNAs that were confirmed in a validation set of children with ASD compared with control subjects using RT-PCR assay were miR-500a-5p, -197-5p, -424-5p, -664a-3p, -365a-3p, -619-5p, -664a-3p, -3135a, -328-3p, and -500a-5p in blood plasma (Kichukova et al., 2017, 2021), and miR-151a-3p, -181b-5p, -320a, -328, -433, -489, -572, -663a, -101-3p, -106b-5p, -19b-3p, -195-5p, and -130a-3p in blood serum (Mundalil Vasu et al., 2014). In adults with ASD, miRNAs validated by PCR were miR-15b-5p in whole blood (Hosokawa et al., 2023), and miR-6126 in whole blood (Nakata et al., 2019). These miRNAs can be considered the most promising potential candidate biomarkers for ASD in children or adults. The potential target gene sets of miR-328-3p and miR-619-5p include some related to mediation of the cellular influx of calcium ions upon membrane polarization as well as other genes involved in epigenetic processes such as dicer 1, ribonuclease type III (DICER). MiR-424-5p target genes include drosha, ribonuclease type III nuclear (RNASEN) (Kichukova et al., 2017). Many of the validated target genes of the dysregulated miRNAs in blood plasma are involved in synaptic pathways (in cholinergic, dopaminergic, GABAergic, or glutamatergic synapses). Among them are four separate genes that encode G-protein subunits and play an important role in intracellular signal integration: GNAL, GNAQ, GNB1, and GNG12. Three other target genes GRIN2B, GABRG2, and GABARAPL1 encode different subunits of receptors for neuroactive ligands and receptor-associated proteins. Two other target genes SCLC1A2 and SCLC1A1 participate in glutamate reuptake (Kichukova et al., 2021). MiR-664a-3p has numerous target genes affecting a large array of biological processes that are not directly related to the central nervous system or key aspects of ASD behavior. Somewhat similar is the situation with miR-197-5p. However, miR-424-5p and miR-500a-5p have a pronounced involvement in nerve tissue–specific pathways, including enrichment of target genes affecting different types of synapses, axon guidance, neuroactive ligand-receptor interactions, and some pathologies (e.g., cancer) (Kichukova et al., 2021). The predicted target genes of the dysregulated miRNAs in blood serum could be involved in axon guidance, transforming growth factor β signaling, mitogen-activated protein kinase signaling, adherens junction, regulation of actin cytoskeleton, oxidative phosphorylation, hedgehog signaling, focal adhesion, mTOR signaling, and Wnt signaling (no specific pathways for miR-572) (Mundalil Vasu et al., 2014). For target gene prediction concerning miR-15b-5p, involvement in cell development, including axonogenesis, and the Wnt signaling pathway was found (Hosokawa et al., 2023). Target genes of miR-6126 were related to axons, neuron guidance, and oxytocin signaling pathways (Nakata et al., 2019). The functions of target genes of the dysregulated miRNAs in ASD subjects are summarized in Additional Table 2.

Additional Table 2.

Potential candidate miRNAs in whole blood, blood plasma and blood serum of patients with autism spectrum disorder

References	MicroRNA	Target genes and functions
Dysregulated miRNAs in children
Blood plasma, upregulated miRNAs
Kichukova et al., 2021	miR-197-5p, -424-5p, -664a-3p	miR-424-5p target genes related to epigenetic processes e.g., DICER, RNASEN; affecting different types of synapses, axon guidance, neuroactive ligand-receptor interactions; and some pathologies (e.g., cancer). miR-664a-3p target genes are not related to key aspects of ASD behavior.
Kichukova et al., 2017	miR-365a-3p, -619-5p, -664a-3p	miR-619-5p target genes related to cellular influx of Ca2+ ions upon membrane polarization and to epigenetic processes e.g., DICER, RNASEN. miR-664a-3p target genes are not related to key aspects of ASD behavior.
Blood plasma, downregulated miRNAs
Kichukova et al., 2021	miR-500a-5p	miR-500a-5p target genes affecting different types of synapses, axon guidance, neuroactive ligand- receptor interactions; and some pathologies (e.g., cancer).
Kichukova et al., 2017	miR-3135a, -328-3p, -500a-5p, -197-5p, -424-5p,	miR-328-3p target genes related to cellular influx of Ca2+ ions upon membrane polarization and to epigenetic processes e.g., DICER, RNASEN.
Blood serum, upregulated miRNAs
Mundalil Vasu et al., 2014	miR-101-3p, -106b-5p, -19b-3p, -195-5p, -130a-3p	miRNA target genes related to axon guidance, TGF-β signaling, MAPK signaling, adherens junction, regulation of actin cytoskeleton, oxidative phosphorylation, hedgehog signaling, focal adhesion, mTOR signaling, and Wnt signaling.
Blood serum, downregulated miRNAs
Mundalil Vasu et al., 2014	miR-151a-3p, -181b-5p, -320a, -328, -433, -489, -572, -663a	miRNA target genes related to axon guidance, TGF-β signaling, MAPK signaling, adherens junction, regulation of actin cytoskeleton, oxidative phosphorylation, hedgehog signaling, focal adhesion, mTOR signaling, and Wnt signaling.
Dysregulated miRNAs in adults
Whole blood, upregulated miRNA
Hosokawa et al., 2023	miR-15b-5p	miR-15b-5p target genes related to involvement in cell development, including axonogenesis, and Wnt signaling pathway.
Whole blood, downregulated miRNA
Nakata et al., 2019	miR-6126	miR-6126 target genes related to axons, neuron guidance, and oxytocin signaling pathways; includes ASD-related genes ANK3, CACNA2D1, NRXN3, PCDH9.

From the studies of Kichukova et al. (2017,2021), an inconsistency is apparent in the dysregulated levels of miR-197-5p and miR-424-5p. The reasons for this are unclear. ASD: Autism spectrum disorder; MAPK: mitogen-activated protein kinase; miRNAs: microRNAs; TGF : transforming growth factor.

A recent review article draws attention to the mechanisms and therapeutic implications of signaling pathways in ASD (Jiang et al., 2022). The Wnt signaling pathway has been implicated in the etiology of autism (Kwan et al., 2016). Molecular, cellular, electrophysiological, and behavioral abnormalities in accordance with autism-like phenotypes have been found in several Wnt signaling-related knockout mouse models (Durak et al., 2016; Katayama et al., 2016). Wnt signaling in the brain involves two primary pathways: (1) β-catenin-dependent “canonical” signaling and (2) β-catenin-independent “noncanonical” signaling (Salinas and Zou, 2008). Many key proteins in both signaling pathways are located at synapses and play important roles in synaptic growth and maturation (Oliva et al., 2013; Caracci et al., 2016). Canonical Wnt signaling acts directly on β-catenin to increase its stability, enabling it to translocate from the cell surface to the nucleus, and linking extracellular signaling to nuclear gene expression (Mullins et al., 2016). ASD-associated Met receptor tyrosine kinases (e.g., CDH8) release β-catenin to bind to surface calcium (Judson et al., 2011). Multiple Wnt molecules, including Wnt2, transmit signals at the surface membrane (MacDonald and He, 2012). It was hypothesized that increased canonical Wnt signaling contributes to the hyperproliferation of embryonic neural progenitor cells in the brain, and may partially be responsible for macrocephaly in individuals with autism (Bernier et al., 2014; Sugathan et al., 2014). The levels of synaptic proteins can be influenced directly by neuronal activity–dependent synaptic mRNA translation pathways, thereby controlling synaptic strength and number (Buffington et al., 2014). The extracellular mTOR and FMRP signaling pathways are the two primary regulators of mRNA translation (de la Torre-Ubieta et al., 2016). Glutamate and brain-derived neurotrophic factors can induce a cascade of mTOR and FMRP pathways, causing an increase in mRNA translation (Bourgeron, 2015). Increased levels of glutamate and brain-derived neurotrophic factors were present in the blood of ASD patients (Kasarpalkar et al., 2014; Al-Otaish et al., 2018). Activation of cell surface receptors such as NMDARs, AMPARs, mGluR, IGFR, and TrKB is linked to activation of the extracellular-signal-regulated kinase/mitogen-activated protein kinase and phosphatidylinositol 3-kinase/mammalian target of rapamycin pathways (Jiang et al., 2022).

Several important limitations were identified in the studies reviewed: (1) there was a marked difference in the sizes of the ASD and control groups in some studies (Jyonouchi et al., 2017, 2019; Ozkui et al., 2020; Qin et al., 2022; Mehmetbeyoglu et al., 2023) and none of the studies had provided a power-size calculation; (2) gender and/or age of ASD and/or control groups were not reported in three of the studies (Hirsch et al., 2018; Kichukova et al., 2021; Jiang et al., 2023); (3) ASD subjects had not been treated with medications prior to collection of blood samples in some of the studies (Mundalil Vasu et al., 2014; Kichukova et al., 2017, 2021; Popov et al., 2018; Nakata et al., 2019; Qin et al., 2022); in the other studies it was not reported whether subjects had received medications prior to blood collection; (4) blood samples were collected while participants were fasting in some of the studies (Popov et al., 2018; Kichukova et al., 2021); in the other studies, it was not indicated whether subjects had fasted prior to blood collection; (5) normalization of miRNA expression levels was not reported in three studies (Jyonouchi et al., 2017; Hosokawa et al., 2023; Jiang et al., 2023); (6) some of the studies are best regarded as discovery studies as no validation study was subsequently performed to confirm the preliminary findings; (7) some studies had used a very small subsample for miRNA profiling or validation; (8) only one study had examined the effects of ASD severity on miRNA levels (Jyonouchi et al., 2019); in most studies IQ and Autism Diagnostic Interview-Revised and ADOS scores of ASD subjects were not reported, so severity levels of ASD subjects are not known (Additional Table 3); (9) only one study reported the effect of AEDs on miRNA levels in ASD individuals (Jyonouchi et al., 2019).

Additional Table 3.

Age, IQ and intellectual disability, ADI-R scores, and ADOS scores of patients with autism spectrum disorder.

References	Age of ASD subjects	IQ and intellectual disability	ADI-R scores	ADOS scores
Hosokawa et al., 2023	Discovery cohort: 31.7±8.5yr Replication cohort: 30.0±8.2 yr	Average IQ of 14 ASD subjects among the discovery and replication cohorts 88.6±24.1	NR	NR
Nakata et al., 2019	28.4±7.3 yr	Full scale IQ (WAIS) 104.1±17.3 No participants exhibited intellectual disability	NR	ADOS total score 7.6±5.6, SRS-2 total score 88.3±28.8
Jyonouchi et al., 2019	11.3±5.4 yr [2-22 yr]	NR	NR	NR
Jyonouchi et al., 2017	11.8 yr [2.8-27.0 yr]	NR	NR	NR
Vaccaro et al., 2018	7.5±2.5 yr	NR	The ASD subjects were enrolled in the same autism severity group by two well-validated clinical tests: Childhood Autism Rating Scale and Autism Diagnostic Observation Schedule-Generic
Hirsch et al., 2018	5-15 yr	NR	NR	NR
Mehmetbeyoglu et al., 2023	2-13 yr	35 of 45 ASD subjects (78%) had intellectual disability	NR	NR
Kichukova et al., 2021	Children, ages NR	NR	NR	NR
Kichukova et al., 2017	Children, ages NR	NR	NR	NR
Eftekharian et al., 2019	6.0±1.5 yr	NR	NR	NR
Jiang et al., 2023	7.1±2.6 yr	IQ 68.1±15.9 i.e., mild intellectual disability	NR	Communication score 7.2±2.0, social interaction 6.8±1.8, repetitive and restricted behaviors 3.7±1.8
Ozkui et al., 2020	2-13 yr	22 of 45 ASD/PDD-NOS children (49%) had intellectual disability	NR	NR
Popov et al., 2018	3-11 yr	NR	NR	NR
Mundalil Vasu et al., 2014	11.3±2.5 yr [6-16 yr]	NR	Social score 17.7±7.6, communication score 12.8±5.6, stereotype score 4.1±2.8	NR
Qin et al., 2022*	3.5 yr	NR	NR	Communication score 5.1±2.8, social interaction 11.4±3.2, repetitive and restricted behaviors 1.6±1.2
Atwan et al., 2020	7 yr	NR	NR	NR

Intellectual disability (ID) indicated by IQ < 70 (DSM-4). Values are means (± SE in most cases). *In the study by Qin et al. (2022), severity degree of ASD children was mild 17%, moderate 64%, severe 19%. It would probably be similar for ASD children in the study by Jiang et al. (2023). In the ADI-R, for Social Interaction Threshold is 10, Maximum 30; Communication Threshold 8 (7 if non-verbal) Maximum 26; Patterns of Behaviors Threshold 3, Maximum 12. Threshold scores indicate autism. https://www.veritasassessments.or2/understandin2-your- veritas-autism-diagnostic-assessment-report/. ADI-R: Autism Diagnostic Interview-Revised; ADOS: Autism Diagnostic Observation Schedule; ASD: autism spectrum disorder; IQ: intelligence quotient; NR: not reported; PDD-NOS: pervasive developmental disorder-not otherwise specified; SRS: social responsiveness scale; WAIS: Wechsler Adult Intelligence Scale; yr: year(s).

In summary, possible candidate miRNA biomarkers in blood plasma and blood serum have been identified mainly in ASD children who had not received medication prior to blood collection. Further studies are needed in ASD adults to evaluate miRNAs as diagnostic biomarkers, and also studies in ASD male and female children to determine if any of the miRNAs are gender-dependent. A significant correlation was found between gender and miR-15b-5p in whole blood of both control and ASD adults (Hosokawa et al., 2023). In addition, it would be helpful to determine if ASD severity or medication treatment is a significant confounding variable for miRNA expression changes. Studies using animal models of ASD could be performed to support and extend miRNA expression changes in human studies. Several animal models have been described (Li et al., 2021; Berg and Silverman, 2022) and include environment-induced models (related to environmental pollution, exposure to chemical or toxic substances, viral infection, and repeated cold temperature stress) and genetic models. Two of the human studies reviewed also used animal models. Female Wistar rats were mated overnight and received a single intraperitoneal injection of 600 mg/kg sodium valproate or saline on embryonic day 12.5 (E12.5) and daily subcutaneous injections of trans-resveratrol 3.6 mg/kg or DMSO from E6.5 to E18.5. MiR-134-5p was significantly increased in blood samples from male offspring 30 days after birth and resveratrol was able to prevent this alteration (Hirsch et al., 2018). Development of the ASD-like phenotype was followed in B6D2 and Balb/c male mice that had received an intraperitoneal injection of 500 mg/kg valproic acid at 2 weeks of age. The six serum miRNAs (miR-3613-3p, -150-5p, -126-3p, -361-5p, -19a-3p, and -499a-5p) that were decreased in ASD human patients were altered to a comparable extent in the blood of both mouse models compared to controls and with altered behavioral traits characteristic of the ASD phenotype (Ozkui et al., 2020). This research has the potential to markedly impact translational and clinical practices, particularly in early ASD diagnosis among children.

Additional files:

Additional Table 1: Alterations of miRNA expression in whole blood, blood plasma, blood serum, serum neural cell adhesion molecule L1 (L1CAM)-captured extracellular vesicles, blood cells, peripheral blood mononuclear cells of patients with autism spectrum disorder.

Additional Table 2: Potential candidate miRNAs in whole blood, blood plasma, and blood serum of patients with autism spectrum disorder.

Additional Table 3: Age, IQ and intellectual disability, ADI-R scores, and ADOS scores of patients with autism spectrum disorder.

Data availability statement:

All relevant data are within the manuscript and its Additional files.