Autism spectrum disorder (ASD) is a type of neurodevelopmental disorder characterized by impaired social communication and interactions, as well as repetitive behavior and limited interests (1). Decades of research have shown that the prevalence of ASD has increased dramatically. According to the Centers for Disease Control and Prevention, one out of every 59 children in the United States is diagnosed with ASD among 8-year-olds in 2018, with boys being four times more likely to be diagnosed than girls (2). ASD's etiology is complex, and it may be due to the interaction of genetic and environmental factors (3, 4). Its development is also heavily influenced by genetic factors (5, 6). Pathogenesis is linked to metabolic disorders, gut microbiota, viral and bacterial infections, chemical influences on the mother's body during pregnancy, as well as neurological and immunological factors (3, 7–9).

There are no clinical biomarkers for ASD because the disorder's etiology and pathogenesis are unknown (3, 10). ASD is diagnosed based on an autism-specific history and clinical observation (3, 10, 11). This could lead to a delay in diagnosis. Although early signs of ASD can be observed and diagnosed as early as 15–18 months of age, the average age of diagnosis is about 4.5 years, and it is not even possible to diagnose ASD before this age (3). Currently, there are no effective pharmaceutical treatments for ASD's fundamental symptoms (12). Early behavioral therapies, on the other hand, have been found to be beneficial in lowering disability and making a significant impact on the outcomes for children with ASD, but they are most effective when started early (13, 14). It has been suggested that interventions initiated before 3 years of age may have a stronger favorable impact than those initiated after the age of five (14). Thus, early diagnosis is essential for ASD, prompting researchers to look for ASD biomarkers. In addition to early diagnosis, reliable ASD biomarker groups are beneficial in clinical practice because they measure the risk of birth in “baby siblings” of children with ASD (15). It also reflects pathogenic processes, assesses treatment and intervention results, and identifies a physiologically homogeneous cohort of ASD patients (16). Moreover, it reveals unknown causes and offers a better knowledge of the disease's underlying pathophysiological processes (17).

However, ASD is a genetically diverse disorder. More than 1,000 genes have been linked to ASD, but none have been found to account for more than 1% of cases (6, 18). Meanwhile, intellectual disability, trouble coordinating movement, sleep difficulties, seizures, and gastrointestinal (GI) issues are also common comorbidities associated with ASD (19). Therefore, identifying biomarkers for ASD has been challenging. Despite this, the evidence suggests that immunological dysregulation, inflammation, oxidative stress, mitochondrial dysfunction, and excitotoxicity are key components in ASD pathogenesis (3, 10, 20, 21). The biomarker related to them have been detected in the blood and urine, and these abnormalities have also been observed in the brain of individuals with ASD, indicating that they could be used to reduce diagnostic heterogeneity and enhance treatment response prediction. So far, many studies have reported increased oxidative stress in individuals with ASD, including decreased enzymatic antioxidants, and increased DNA, lipid, and protein oxidation products both in the brain and peripheral circulation (22–28). Increased oxidative stress markers have been found in peripheral body fluids and have been linked to ASD severity (29).

This article reviews the current state of research on oxidative stress in ASDs, focusing on the mechanism of oxidative stress, biological analysis of oxidative stress biomarkers, and antioxidant-based therapy methods. The literature focused on the last 10 years and was collected from PubMed, Web of Science, and Google Scholar.

The concept of oxidative stress was initially introduced in 1985 with the publication of the book “Oxidative Stress” (30). Reduction–oxidation (redox) reaction is a type of indispensable reaction in the cellular physiological process of cells, during which ROS are generated. ROS is typically produced either intentionally (to kill invading pathogens or as intermediates in enzymatic reactions, etc.) or accidentally (via electron leakage from electron transport chains, metabolism of drugs, exposure to chemicals, pollutants, and radiation, etc.) during normal physiological processes of cells. The sources of ROS contain many enzymes. Nicotinamide adenine dinucleotide phosphate oxidase (NOX) isoforms are major sources for endogenous ROS, multifarious NOX isoforms are localized to various cellular membranes and involved in many physiological or pathological events (31, 32). Myeloperoxidase (MPO) is primarily located in immune cells and plays an important role in our immune system, which produces some ROS, particularly hypochlorous acid (HOCl) to kill invading pathogens (33). NO synthases (NOSs) are the most important NO source in both physiological and pathological conditions (34). Interestingly, low concentrations of NO generated by neuronal NOS or endothelial NOS have a physiological neuroprotective function and are involved in signaling pathway, while higher concentrations of NO synthesized by inducible NOS (iNOS) are neurotoxic (35, 36). Beside the sources of ROS mentioned above, there are also many ROS-generating enzymes include succinate dehydrogenase (SDH) (37), dihydroorotate dehydrogenase (DHOH) (38), mitochondrial glycerol-1-phosphate dehydrogenase (mGPDH) (39), cytochrome b5 reductase (40), monoamine oxidases (MAOs) (41), aconitase (ACO) (42), xanthine oxidoreductase (XOR) (43), alpha-Ketoglutarate dehydrogenase complex (KGDHC) (44), and so on. Some of these enzymes were shown to produce ROS at appreciable rates in studies with either isolated enzymes or mitochondria (45). Additionally, it should be noted that ROS is a general term but not some specific molecule (46–48). It contains a group of molecules that come from molecular oxygen, such as superoxide (O2•-), hydrogen peroxide (H₂O2•), hydroxyl radical (OH^•−), and peroxyl radical (RO2•-), and the chemical reactivity of each ROS molecule is quite different (46, 47, 49).

ROS are eliminated by the antioxidant defense of cells in normal physiological processes, and the body is in a state of physiological balance. This balance, however, will be disrupted if there is an increase in ROS production or a decrease in cell antioxidant capacity, resulting in oxidative stress (46, 47). When there is mild oxidative stress, a low level of ROS stimulates the cellular defense mechanism to produce a proper response to ROS, while ROS can also induce cell apoptosis as a signal molecule. This phenomenon is known as “eustress,” and it is beneficial to the maintenance of cellular ROS defense and tissue renewal (48, 50, 51). When cells are subjected to severe oxidative stress, ROS that is out of balance with antioxidant capacity damage biomolecules such as proteins, lipids, and DNA, as well as some biological structures such as bio-membrane structure. This is known as “distress.”

As the energy factories of cells, mitochondria are the main sites for the generation of ROS, in which the electron transport chain (ETC) is a prime source for ROS (38). Both endogenous and exogenous oxidative stress can cause a deficit in mitochondrial ETC complexes, resulting in mitochondrial dysfunction. Dysfunctional mitochondria produce more ROS, which can further impair mitochondrial function. This is a vicious cycle in which more severe oxidative stress and mitochondrial dysfunction occur if ROS is not eliminated promptly due to decreased antioxidant capacity. Glutathione (GSH) has been shown in numerous studies to play an important role in mitochondrial ROS elimination (24, 26, 52, 53).

The antioxidant capacity of various cellular defense systems is based on enzymatic antioxidants such as superoxide dismutase (SOD), catalase (CAT) and glutathione peroxidase (GSH-Px), and non-enzymatic antioxidants such as ascorbic acid (vitamins C), uric acid, tocopherol (vitamins E), quinols, carotenoids, and polyphenols. The interaction between some common antioxidants and ROS has been shown in Figure 1 (54–65). It is worth noting that some enzymatic antioxidants have multiple isoforms, and the different isoforms have different functions, such as SOD (66), GSH-Px (67), etc. Furthermore, as a complement to defense systems, some repair systems, such as methionine sulfoxide reductases, disulfide reductases/isomerases, phospholipases, and DNA repair enzymes, will repair structures and biomolecules that have been damaged or modified by residual ROS. As a result, when cells are subjected to mild oxidative stress, these mechanisms can effectively protect them from oxidative damage (46–48). Although there are numerous mechanisms in cells to combat oxidative stress, numerous studies show that oxidative damage to biological structures and biomolecules continues to accumulate in cells in many related diseases, including ASD (46, 47, 68–72).

The human brain is the largest oxygen-consuming organ in the body. It only accounts for 2% of the body mass but consumes 20% of the oxygen. It has a high content of oxidizable polyunsaturated fatty acids as well as redox-active metals (copper and iron). As a result, the human brain is particularly vulnerable to oxidative stress (47, 73–75). Children are more vulnerable than adults to oxidative stress because of their naturally low glutathione levels from conception to infancy (72). In the brains of children with ASD, low levels of mitochondrial glutathione and mitochondrial dysfunction have been reported (24, 26, 52, 53). In particular, increased oxidative stress has been observed in the brains of children with ASD (24). Oxidative stress causes oxidative damage to lipids, proteins, and DNA in cells. It makes a variety of reversible and irreversible damages in ASD which mainly involves various post-translational modifications of proteins such as 3-nitrotyrosine (3NT) and protein carbonyl formation, abnormal metabolism such as lipid peroxidation, and accumulation of toxic such as ROS. The relationship between oxidative stress and ASD has recently been thoroughly reviewed (76). Many markers of oxidative stress, such as lipid peroxide (LOOH) (77), malondialdehyde (MDA) (78), a marker of oxidative DNA damage 8-hydroxy-2'-deoxyguanosine (8-OH-dG) (24), protein carbonyl (28, 79), and 3-nitrotyrosine (3-NT), are elevated in children with ASD. The increased oxidative stress markers have been observed to be correlated with ASD severity (29).

Furthermore, several studies have shown that oxidative stress causes an inflammatory response as an upstream component in the signaling cascade (80, 81). ASD patients have been shown to have systemic immunological abnormalities as well as an inflammatory response (82, 83). In fact, oxidative stress is often detected alongside inflammation in the brains of people with ASD, and some studies have demonstrated a link between the two in specific brain regions associated with ASD (24, 84, 85). Even though it is difficult to know whether the connection is unique to specific brain regions or not due to the limitations of the brain tissue sample, this has revealed more about the role of oxidative stress in the etiology of ASD. Other studies in peripheral blood cells have found evidence of inflammation and oxidative stress in a variety of cell types, including T cells (86, 87), B cells (88), monocytes (89, 90), neutrophils (90), and lymphocytes (91). In these studies, in vitro induction experiments were also used to demonstrate the link between inflammation and oxidative stress in peripheral cells. Peripheral cells may be useful in studying systemic neurochemical changes in ASD.

In general, oxidative stress is involved in the pathogenesis of ASD. As a result of the interaction of genetic and environmental factors, people with ASD have excessive ROS production, decreased antioxidant capacity, and mitochondrial dysfunction (55). All of these physiological abnormalities have the potential to cause oxidative stress (55, 92, 93). And oxidative stress can cause epigenetic dysregulation (55, 93), neurodevelopment disorder (94), neuro-inflammation (95), cerebral injury (55, 92, 95), and neuro-dysfunction (55, 92, 95), which finally leads to ASD (94–96). Figure 2 depicts the potential mechanisms of oxidative stress in the pathogenesis of ASD.

The studies of potential oxidative stress biomarkers for ASD in the past 10 years are shown in Table 1. It focuses on proteins and metabolites related to oxidative stress in peripheral body fluids such as blood, urine, and saliva. These potential biomarkers include enzymatic antioxidants, non-enzymatic antioxidants, proteins, and lipids damaged by oxidation (28, 29, 97–140). It is critical to verify whether the changes reported in different research are consistent as a potential biomarker. Individual variances must be adapted by effective biomarkers, especially in the case of disease as heterogeneous. Most of these potential biomarker changes in ASD patients are consistent without dissenting reports. Here, we will focus on a few classic oxidative stress biomarkers related to ASD and introduce them according to their classification.

To learn more about the pathophysiology and diagnostic biomarkers of ASD, researchers are currently studying effective drugs and treatments (228). Increased oxidative stress is a common feature in ASD individuals, despite the fact that ASD is heterogeneous. Intervening and treating oxidative stress is one of the most effective techniques for improving the pathogenetic status of ASD patients. Therefore, various antioxidants, including sulforaphane (229), resveratrol (230–233), N-acetylcysteine (NAC) (234, 235), hesperidin (236), flavonoid (237, 238), leptin (239), minocycline, and doxycycline (240), selenium supplements (241), docosahexaenoic acid (DHA) (242), curcumin (243), agmatine (244), and sulindac (245), etc., have been reported to be employed in ASD treatment animal model experiments. In these studies, all of these antioxidants have shown positive therapeutic effects, indicating that they could be useful in the treatment of ASD.

Antioxidant therapy for ASD has only a few clinical investigations. Sulforaphane (246), resveratrol (247), coenzyme Q10 (248), NAC (249), omega-3 fatty acids (250), arachidonic acid, and DHA (251) are some of the antioxidant supplements used in these studies to treat ASD. All of these antioxidants, with the exception of resveratrol, are beneficial. Although resveratrol plays a beneficial role in the treatment of ASD animal model, its clinical study was still in its infancy. Currently, it has only one clinical study and the result is negative (252). Interestingly, a systematic review of treatments based on antioxidants reported that NAC appears to be the most effective antioxidant therapy of ASD currently (253). Furthermore, supplementing micronutrients for redox metabolism has been demonstrated to be helpful in certain children with autism (254). Treatment of ASD patients with antioxidant-rich food, on the other hand, is also a viable option. Several studies have evaluated the effectiveness of antioxidant-rich foods including broccoli (255), camel milk (256), and dark chocolate for ASD (257).

Overall, the results of these studies are positive. But as expected, some of the treatment groups in these clinical studies showed strong individual differences, reflecting the heterogeneity of ASD. It is important to note that ROS is a general term, not a specific molecule. As mentioned above, it contains a set of molecules derived from molecular oxygen, and the chemical reactivity of various ROS molecules varies widely as far as antioxidants are concerned, there are many types of antioxidants, and their specific antioxidant functions are also different. Antioxidants always have a goal that can only handle one type of ROS and not another (46, 47, 49). The causes of oxidative stress in ASD patients may differ due to genetic differences and the diversity of antioxidant defenses against oxidative stress. Using biomarkers to determine the types of antioxidants taken by each ASD patient and then supplementing them might be more successful. Antioxidants have been demonstrated to enhance behavior in persons with ASD in numerous research, however, these effects are generally transient, and only a few studies have shown a long-term behavioral reversal in people with ASD (228). Therefore, effective biomarkers for monitoring the efficacy of antioxidative therapy in ASD patients should be considered.

Some of the antioxidants mentioned above, such as sulforaphane, resveratrol, naringenin, curcumin, and agmatine, work as both antioxidants and Nrf2 activators (252). Nrf2 is a transcription factor implicated in immunological dysregulation/inflammation, oxidative stress, and mitochondrial dysfunction. Nrf2 is generally coupled to Kelch-like ECH-associated protein 1 (Keap1) in an inactive form, and the ubiquitin-proteasome system destroys the complex, allowing cells to maintain a steady low level of Nrf2 (258). The complex dissociates when subjected to oxidative stress, and Nrf2 translocates to the nucleus. Before binding to specific DNA locus antioxidant response elements (AREs), Nrf2 will heterodimerize with Maf or Jun proteins in the nucleus (259, 260). NRF2-ARE binding can regulate the expression of hundreds of cytoprotective genes including antioxidant proteins and phase II enzymes (261). Furthermore, the Nrf2/ARE pathway interacts with the NF-κB (nuclear factor kappa-light-chain-enhancer of activated B cells) pathway. The p65 subunit of NF-κB inhibits the Nrf2/ARE pathway by depriving CREB binding protein (CBP), allowing HDAC3 to recruit to MafK and interact with Keap1 (262, 263). Alternatively, free Keap1 can inhibit the NF-κB pathway by regulating the activity of the inhibitor of nuclear factor-κB kinase subunit beta (IKK-β) (264). NF-κB is a key player in the regulation of inflammation (265), as is involved in the release of pro-inflammatory cytokines such as IL-1, IL-6, IL-12, and TNF-α (266). Several studies have also shown that Nrf2 can directly regulate the availability of mitochondrial respiratory substrates, resulting in mitochondrial depolarization, reduced ATP levels, and impaired respiratory function. Furthermore, the aforementioned negative phenomena can be reversed by activating the Nrf2 pathway (267, 268).

Moreover, when induced with lipopolysaccharide (LPS), Nrf2-deficient mice have a more pronounced release of ROS, microglial activation, and neuro-inflammatory response than normal mice (269). Some studies of BTBR mice (a model of ASD) indicated that the Nrf2 system plays an important role in the regulation of neuroinflammation and oxidative stress in the brain (229, 270). A study in monocytes from people with ASD found a positive result by regulating the Nrf2 system in an in vitro LPS-induced inflammatory model (271) and many other studies have reported the abnormalities of the Nrf2 system in ASD individuals (272, 273). Therefore, the Nrf2 system is one of the important ways of antioxidant therapy. A systematic review of treatments based on the Nrf2 system shows a potentially beneficial result, but also explains that these treatments still lack sufficient evidence for their efficacy and safety (252). Better design and more rigorous research are needed before the treatments can be used.

Disease progression, including ASD, is often accompanied by dramatic changes in the levels of various proteins and metabolites. Biomarkers can be used to comprehensively monitor the physiological status of ASD patients during diagnosis, intervention, and treatment, which can aid in understanding the condition, judging the treatment strategy, and monitoring efficacy and prognosis (3, 10). To this purpose, the efficacy of biomarkers and biological detection systems must meet stringent requirements. Traditional detection methods, such as Western blot analysis and enzyme-linked immunosorbent assay (ELISA), can not detect many markers at once, making comprehensive control difficult. The sensitivity of biological detection technology has improved, and more detection scenarios have been implemented, allowing for more comprehensive monitoring. Mass spectrometry has seen tremendous advancements in recent years, particularly in terms of reproducibility, performance, resolution, precision, and analytical quality.

Currently, mass spectrometry-based targeted proteomic or metabolomic approaches can effectively monitor multiple disease markers simultaneously (3, 10). Several targeted metabolomic techniques to oxidative stress markers have been developed (274) and utilized (275). Methionine, homocysteine, vitamins B6, B12, B9, and their metabolites have been accurately measured in several matrices, including breast milk, plasma, and neonatal mouse brain, using a novel approach (274). Concomitant vitamin B6, B9, and B12 deficits, as well as lower levels of methionine, GSH, SAM, and a lower SAM/SAH ratio, as well as Hcy, SAH, and 5-methyltetrahydrofolate (5-methyltetrahydrofuran) in children's urine samples, have all been linked to autism (275).

At the same time, because the brain is a part of the central nervous system and is susceptible to oxidative stress, numerous physiological abnormalities generated by oxidative stress in the brain would play a role in the development of autism. For the diagnosis and treatment of ASD, identifying the oxidative stress that occurs in the peripheral or brain is beneficial. Brain-derived neurotrophic factor (BDNF) (276), brain-derived exosomes (277), and other plasma brain-derived ASD biomarkers, have been discovered in numerous research. However, though some studies also reported some plasma biomarkers for brain oxidative damage by analyzing different kinds of samples including F4-Neuroprostanes and F2-Dihomo-Isoprostanes (278), biomarkers in peripheral body fluid samples are still insufficient to identify brain or peripheral oxidative stress at the moment. Although cerebrospinal fluid (CSF) samples can detect oxidative stress in the brain, they are not appropriate for patients with ASD due to the risk of injury during the sampling process. Because of their ease of collection, peripheral bodily fluid samples are always the best option.

Furthermore, little research has been done on peripheral blood cells, which play an important role in the immune system. Since a relationship has been demonstrated between oxidative stress and systemic inflammation (82, 83), peripheral oxidative stress and inflammation in ASD patients cannot be ignored. There is a lot of evidence that peripheral immune cells like T cells and B cells can affect brain neurons and can contribute to brain inflammation in some neural diseases (279).

With significant advances in biomedical detection technology, the limitations and defects of previous studies will be improved. Otherwise, some significant topics closely related to oxidative stress of ASD such as brain-derived factors and peripheral blood cells are worthy and promising.

Many studies have demonstrated that oxidative stress plays a crucial part in the disease process of ASD because ASD cases have greater levels of oxidative stress and decreased antioxidant capability. The active use of biomarkers to monitor ASD patients' physiological status is helpful for disease diagnosis, intervention, and treatment. We mainly summarize the most recent research progress in the field of ASD oxidative stress biomarkers in this review. Many possible oxidative stress markers have been discovered in ASD, however, attempts to monitor the oxidative stress status of children with ASD are still difficult to meet clinical application standards, and additional study is needed. At the same time, we present a review of recent studies on antioxidant interventions. Several clinical investigations have found significant individual differences in some therapy groups, indicating that ASD is heterogeneous. With the development of mass spectrometry technology, mass spectrometry-based proteomics, and metabolomic methods have gradually become powerful tools for exploring biomarkers. These methods make ASD biomarker research easier and help to expand the depth and breadth of biomarker research.

XL and LS contributed to conception and design of the study. JL, HZ, JZ and XT collected related literature and tabulated it. XL wrote the first draft of the manuscript. XL, NK, XC and LS wrote sections of the manuscript. All authors contributed to manuscript revision, read, and approved the submitted version.

This article was supported by the National Natural Science Foundation of China (Grant No. 31870825), the Shenzhen Bureau of Science, Technology, and Information (Grant Nos. JCYJ20170412110026229 and JCYJ20200812122708001), and the Shenzhen-Hong Kong Institute of Brain Science-Shenzhen Fundamental Research Institutions (Grant Nos. 2019SHIBS0003 and 2021SHIBS0003).

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.