Preprints.org logo
Preprint
Hypothesis

This version is not peer-reviewed.

The Energy-Deficit Hypothesis of Autism: Multi-Cytokine Convergence on Mitochondrial Dysfunction: A Cerebral Energy-Deficit Hypothesis of Autism Spectrum Disorder

Submitted:

17 September 2026

Posted:

17 September 2026

You are already at the latest version

Abstract
Background: Autism spectrum disorder (ASD) is a neurodevelopmental condition with substantial genetic contributions but incompletely understood biological mechanisms. Family history of autoimmune disease is associated with elevated ASD likelihood (pooled OR 1.28), while maternal immune-mediated conditions show additional associations, suggesting that inherited immune susceptibility and the gestational inflammatory environment may contribute to risk. A mechanism linking these observations to the characteristic developmental trajectory of ASD remains unresolved.Hypothesis: I propose an immune-metabolic framework in which multiple pro-inflammatory cytokines—including TNF-α, IL-6, IL-1β, and IFN-γ—act through distinct molecular pathways yet converge on mitochondrial dysfunction and an ATP supply that cannot meet the developing brain's energy needs. Within this framework, cumulative prenatal inflammatory burden, rather than any single cytokine, drives the energy deficit. Potential contributors include inherited immune susceptibility and fetal exposure to maternal inflammatory signaling arising from a clinically silent but biologically consequential chronic low-grade state. The resulting energy constraint may disrupt synaptic refinement, real-time social processing, synaptic protein synthesis, and flexible predictive inference—the last offering a bioenergetic interpretation of restricted repetitive behaviors and insistence on sameness as compensation for chronic energy limitation. Crucially, the mitochondrial dysfunction is proposed to persist beyond birth, with the gap between cerebral energy demand and supply widening during the rapid brain growth of the first postnatal years. This trajectory may help explain the emergence of clinical symptoms between 12 and 24 months of age, the selective vulnerability of metabolically demanding brain regions, and the regressive pattern seen in a substantial subset of children. The framework further proposes that the reported firstborn association may partly reflect incomplete maternal immune adaptation during first pregnancies.Implications: By uniting prenatal initiation with postnatal persistence into a single developmental trajectory, this framework offers an integrative account of otherwise disparate observations. It also yields directly testable predictions—prospective measurement of maternal cytokine and infant mitochondrial trajectories, registry analysis of partner change and birth order, and cerebral energy metabolism in relation to insistence-on-sameness severity—that could distinguish the proposed mechanism from competing explanations.
Keywords: 
autism spectrum disorder
;  
multi-cytokine convergence
;  
mitochondrial dysfunction
;  
chronic low-grade inflammation
;  
cerebral energy deficit
;  
neurodevelopment
Subject: 
Biology and Life Sciences  -   Neuroscience and Neurology

1. Introduction

Autism spectrum disorder (ASD) is a neurodevelopmental condition characterized by deficits in social communication and interaction, restricted interests, and repetitive behaviors. Despite decades of research, the fundamental biological mechanisms underlying autism remain elusive. While genetic factors contribute substantially to autism risk, environmental and immunological factors increasingly appear to play critical roles [1].
A growing body of evidence indicates that family history of autoimmune disease is associated with elevated autism prevalence. A meta-analysis found a 28% (95% CI 12–48%) higher risk of autism in children with such a history [2]. Familial associations may reflect inherited or shared immune susceptibility, whereas maternal autoimmune disease may additionally expose the developing fetus to an altered inflammatory and metabolic milieu during pregnancy.
In this paper, I propose that autism may be conceptualized as an immune-metabolic disorder in which prenatal pro-inflammatory cytokine signaling contributes to mitochondrial dysfunction and cerebral energy deficiency. The resulting energy constraint may impair four critical processes: synaptic pruning during neurodevelopment, real-time social cognitive processing, synthesis of synaptic scaffolding proteins, and flexible hierarchical predictive inference underlying restricted repetitive behaviors. This framework may help explain core autism symptoms from a unified energetic perspective and generate testable predictions.
A central feature of this framework is convergence: multiple pro-inflammatory cytokines—including TNF-α, IL-6, IL-1β, and IFN-γ—act through distinct molecular pathways but are proposed to converge on mitochondrial dysfunction and an inadequate ATP supply relative to cerebral energy demand. Within this model, cumulative prenatal inflammatory burden, rather than any single cytokine, drives the energy deficit. The model therefore moves beyond previous proposals centered on a single mediator (such as TNF-α or IL-6 alone) and instead treats multiple cytokine pathways as potentially interacting contributors to a common bioenergetic failure.
High heritability does not contradict this hypothesis. Genetic susceptibility can contribute to chronic low-grade immune dysregulation during pregnancy. I propose that such a state may cause ASD through fetal mitochondrial dysfunction, even in the absence of major-effect ASD variants.
A central tenet of this hypothesis is that autism-related neural circuit abnormalities originate during fetal brain development and are sustained postnatally through persistent mitochondrial dysfunction, rather than arising de novo after birth. In this view, the autistic brain is not a typically developed brain that subsequently malfunctions; it is a brain whose foundational architecture—synaptic density, connectivity patterns, and microglial programming—has been shaped by an adverse intrauterine inflammatory and metabolic environment, with the resulting mitochondrial dysfunction persisting through the early postnatal period of rapid brain growth and high energy demand. The clinical manifestations observed postnatally thus reflect both the developmental consequences of altered prenatal trajectories and the ongoing impact of persistent cerebral energy deficit.
The framework presented here builds on the maternal immune activation (MIA) hypothesis, which proposes that inflammatory perturbations in utero can influence fetal neurodevelopment and increase the risk of neurodevelopmental disorders, including ASD [3], and on the broader neuro-immunometabolic framework that integrates fetal neuroinflammation, prenatal stress, and glial-neuronal interplay in ASD developmental origins [4]. The MIA hypothesis has accumulated substantial epidemiological support; recent critical reviews and large-scale familial-confounding analyses, however, have highlighted persistent questions about causation, the identity of the most pathogenically relevant immune mediators, the gestational timing windows of vulnerability, and why only a subset of MIA-exposed offspring develops ASD [5,6]. The present framework extends these prior accounts by proposing a specific molecular convergence point: multiple maternal pro-inflammatory cytokines, despite their distinct upstream effects on neurons, ultimately converge on fetal mitochondrial dysfunction, producing a cerebral energy-deficit state that initiates during the prenatal period and persists postnatally to shape neurodevelopmental trajectories.
A central implication of the present hypothesis is that ASD may arise not only from overt maternal inflammatory disease, but also from a “clinically silent, chronic low-grade pro-inflammatory cytokine state” that is insufficient to endanger maternal or fetal survival, yet sufficient to disrupt fetal brain development during sensitive gestational windows. This distinguishes the present model from acute, clinically obvious conditions such as preeclampsia or cytokine storm: in the cases addressed here, pregnancy may proceed to term with outwardly healthy neonates, while the cumulative impact on fetal brain bioenergetics has already shaped neurodevelopmental trajectories.

2. Epidemiological Evidence: Familial and Parental Autoimmune Disease and Autism Risk

2.1. Large-Scale Studies

Multiple large-scale epidemiological studies have identified associations between familial or parental autoimmune disease and offspring autism risk. Table 1 distinguishes estimates based on family history from those specific to maternal disease during pregnancy.

2.2. A Pro-Inflammatory Cytokine Common Denominator

A critical observation is that several familial or maternal conditions associated with elevated offspring ASD risk are immune-mediated conditions involving broader pro-inflammatory dysregulation—implicating cytokine pathways such as TNF-α, IL-6, and IL-1β—rather than a single isolated immune factor. Epidemiological studies link these familial or maternal immune-mediated conditions to elevated offspring ASD risk:
  • Psoriasis: an immune-mediated inflammatory disease associated with elevated offspring ASD risk in family-history meta-analyses [2]
  • Type 1 Diabetes: family history of T1D is associated with elevated autism risk [2,9], while maternal T1D shows an additional association in large cohort studies [7,8]. The broader familial association may reflect inherited or shared susceptibility, whereas maternal disease additionally provides a potential route for direct gestational inflammatory and metabolic exposure. A TNF-α-targeting biologic (golimumab) has demonstrated disease-modifying effects in new-onset T1D [10], illustrating that TNF-α-related immune pathways can be therapeutically modulated in at least some autoimmune contexts
  • Rheumatoid Arthritis: family and parental history is associated with increased offspring ASD risk [2,11]
Studies of parental rheumatoid arthritis also suggest a possible role for maternal disease around pregnancy. Sun et al. [12] reported an association between maternal rheumatoid arthritis and offspring ASD, but no statistically significant association with paternal rheumatoid arthritis. Zhu et al. [13] found an association with maternal disease classified as pre-delivery, but no statistically clear association with later diagnoses; their pre-delivery category included diagnoses within 90 days after birth.
Yin et al. likewise found an association with maternal rheumatoid arthritis diagnosed before delivery (adjusted HR 1.43, 95% CI 1.11–1.84). However, maternal arthralgia, used as a non-inflammatory comparison condition, showed a similar association with offspring ASD [14]. These findings support further investigation of gestational influences, but do not establish inflammation as the specific explanation.

3. Sources of Prenatal Pro-Inflammatory Cytokine Exposure

Multiple pathways can lead to elevated prenatal pro-inflammatory cytokine exposure during critical periods of neurodevelopment. This section examines four distinct sources or susceptibility contexts: familial autoimmune susceptibility and maternal autoimmune disease, maternal obesity, maternal immune maladaptation during first pregnancies, and endogenous mitonuclear incompatibility.

3.1. Familial Autoimmune Susceptibility and Maternal Autoimmune Disease

As detailed in Section 2, family history of autoimmune diseases including psoriasis, type 1 diabetes, and rheumatoid arthritis is associated with elevated autism prevalence. Such familial associations may index inherited or shared immune susceptibility. Maternal autoimmune disease may contribute through the same inherited susceptibility and, additionally, through direct gestational exposure to altered maternal inflammatory and metabolic signaling. Paternal or other nonmaternal family history should therefore be interpreted as evidence of familial susceptibility rather than direct prenatal cytokine exposure.

3.2. Maternal Obesity: An Additional Source of Prenatal Pro-Inflammatory Cytokine Exposure

Beyond familial autoimmune susceptibility and maternal autoimmune disease, maternal obesity is also associated with elevated offspring ASD risk in population-based studies.

3.2.1. Epidemiological Evidence

Meta-analyses reveal that maternal obesity (BMI ≥30) is associated with increased risk of offspring autism, with Li et al. reporting a pooled adjusted OR of 1.47 (95% CI 1.24–1.74) [15] and Wang et al. reporting a pooled RR of 1.36 (95% CI 1.03–1.78) [16]. Wang et al. also identified a linear dose–response relationship between maternal BMI and ASD risk (pooled RR 1.16 [1.01–1.33] for each 5 kg/m2 increment) [16].
More recently, Morin et al. studied 2,072,445 children in Sweden and Denmark and found a J-shaped association across the full maternal BMI range. Associations with overweight and obesity attenuated toward the null in comparisons of full siblings. For maternal BMI ≥35 versus normal BMI (18.5–24.9), the adjusted HR fell from 1.94 (95% CI 1.87–2.01) in the population analysis to 1.04 (95% CI 0.90–1.20) in the family-stratified analysis [17]. These findings point to shared familial influences and leave the independent contribution of gestational exposure unresolved.

3.2.2. Obesity as a Chronic Inflammatory State

Obesity is fundamentally a state of chronic low-grade inflammation [18]. Adipose tissue is not merely an energy storage organ but an active endocrine tissue that produces pro-inflammatory cytokines:
  • Adipose tissue is a major source of pro-inflammatory cytokine production, including TNF-α; obese individuals show elevated TNF-α expression in adipose tissue compared to lean controls [19]
  • Pro-inflammatory cytokine production is elevated both locally in adipose tissue and systemically, with TNF-α, IL-6, and other mediators contributing to a chronic inflammatory milieu [18,19]
  • Other inflammatory markers (IL-6, CRP, leptin) are also elevated
  • Inflammation correlates with degree of adiposity [18]

3.2.3. Mechanism of Fetal Exposure

During pregnancy, maternal obesity creates multiple pathways for fetal pro-inflammatory cytokine exposure [20]:
  • Direct and indirect signaling: Maternal inflammatory signaling may affect fetal development through a combination of direct cytokine exposure, altered placental function, and secondary placental cytokine production
  • Placental inflammation: The placenta itself becomes inflamed in obese pregnancies, producing additional local cytokines
  • Metabolic stress: Maternal hyperglycemia and insulin resistance further compromise fetal mitochondrial function
  • Oxidative stress: Obesity-associated oxidative stress damages both maternal and fetal mitochondria

3.2.4. Convergence with the Energy-Deficit Model

Within this framework, obesity-associated inflammation is proposed as another route to mitochondrial dysfunction in the developing fetal brain. Whether gestational inflammatory signaling originates from maternal autoimmune disease, maternal obesity, or both, the downstream consequences—impaired synaptic pruning, compromised social cognition, and protein synthesis deficits—may remain the same. Familial autoimmune history may additionally identify inherited or shared susceptibility that modifies this pathway. I propose that maternal obesity and maternal autoimmune conditions could have additive effects when co-occurring.

3.3. The Birth Order Effect: Maternal Immune Maladaptation

Several studies report higher autism risk among firstborn children than among later-born siblings. Reproductive stoppage—parents having no further children after an autistic child—can contribute to this pattern and remains important when interpreting birth-order associations.

3.3.1. Epidemiological Evidence for the Firstborn Effect

Kramer et al. found lower odds of autism diagnosis among second-born than firstborn children in both between-family and within-family analyses of a large US insurance-claims cohort [21]. The within-family result strengthens the evidence for a birth-order association, but does not establish maternal immune adaptation as its cause. Restriction to two-child families and birth-order differences in diagnostic timing may still influence the findings.
Table 2. Birth Order and Autism Risk: Evidence Summary.
Table 2. Birth Order and Autism Risk: Evidence Summary.
Finding Effect Size Reference
Firstborn autism risk (Utah) OR 1.8 Bilder et al. (2009) [22]
Second-born versus firstborn autism diagnosis (US; within-family comparison) OR 0.804 (95% CI 0.786–0.823) Kramer et al. (2026) [21]
Preeclampsia in nulliparous women Significantly higher Robillard et al. (1994) [23]
Partner change between pregnancies Reduced parity-associated protection reported; interpretation debated Tubbergen et al. (1999), Basso et al. (2001), and Zhang (2007) [24,25,26]
Note: OR = odds ratio; CI = confidence interval. In the Kramer et al. row, firstborn children are the reference group; an OR below 1 indicates lower odds in second-borns. Preeclampsia data are included as parallel evidence for primigravid immune maladaptation.

3.3.2. Primigravid Immune Maladaptation Mechanism

The maternal immune system must achieve tolerance to semi-allogeneic fetal antigens. Aluvihare et al. [27] showed that pregnancy induces a systemic expansion of maternal CD25+ regulatory T cells (Tregs) that is alloantigen-independent and functionally essential for fetal tolerance; depletion of these cells caused immune-mediated pregnancy failure in mice. In mouse pregnancy, however, Kallikourdis et al. [28] found that accumulation of highly suppressive CCR5+ Tregs in the gravid uterus is enhanced by alloantigen, indicating a locally antigen-responsive component distinct from systemic expansion. In a complementary mouse model, fetal-antigen-specific maternal Tregs accumulated during primary pregnancy and persisted after parturition; on re-exposure to fetal antigen during a subsequent pregnancy they re-expanded rapidly, providing a form of regulatory memory [29]. This distinction—systemic pregnancy-associated expansion, local alloantigen-enhanced accumulation, and fetal-antigen-specific regulatory memory—provides the immunological mechanism invoked here for a possible parity-related difference in maternal tolerance.

3.3.3. Preeclampsia as Parallel Paradigm

Preeclampsia—characterized by placental inflammation and elevated pro-inflammatory signaling—has been called “the disease of primigravidae” since 1902. Nulliparous women have significantly higher preeclampsia risk than multiparous women [23]. Some studies report that this parity-associated protection is weakened when a subsequent pregnancy is with a new partner [24,30]. This pattern is consistent with the regulatory-memory mechanism described above, suggesting that maternal immune adaptation may be partly specific to paternally derived fetal antigens rather than to pregnancy alone. However, studies disagree on whether partner change itself or the longer interpregnancy interval that often accompanies it better explains the association [25,26]. I propose that an analogous parity-related difference in maternal immune tolerance may contribute to the autism birth-order effect.

3.4. Mitonuclear Incompatibility: An Endogenous Source of Pro-Inflammatory Signaling

While the preceding sections describe how pro-inflammatory cytokine-mediated mitochondrial dysfunction may contribute to autism, an important question remains: what about cases without diagnosed maternal autoimmune disease or known familial autoimmunity? I propose that mitonuclear incompatibility may represent an endogenous source of pro-inflammatory signaling that activates the same pathogenic pathway. This mechanism remains highly speculative and is presented as a hypothesis-generating extension of the broader cytokine-mitochondrial framework; it requires independent empirical validation.

3.4.1. The Gap in the Cytokine-Mitochondrial Hypothesis

The present energy-deficit hypothesis addresses autism risk associated with familial autoimmune susceptibility and maternal autoimmune diseases such as psoriasis, type 1 diabetes, and rheumatoid arthritis. However, autism also occurs in families with no known history of autoimmune disease. Additionally, diagnosed mitochondrial disease has been estimated in approximately 5% of autistic individuals, whereas abnormal mitochondrial biomarkers are reported in a substantially larger subset [31,32]. This discrepancy suggests that mitochondrial dysfunction extends beyond cases meeting conventional diagnostic criteria for mitochondrial disease, leaving its origin unresolved in many individuals.
This raises a critical question: if pro-inflammatory cytokine-mediated mitochondrial dysfunction is central to autism pathophysiology, what is the source of inflammatory signaling in cases without known familial or maternal autoimmune disease?

3.4.2. The Unique Inheritance Pattern of Mitochondria

In humans, mitochondrial DNA (mtDNA) is normally inherited from the mother, whereas nuclear DNA is inherited from both parents. Most proteins required for mitochondrial function are nuclear-encoded and imported into mitochondria. Several oxidative phosphorylation (OXPHOS) complexes depend on coordinated interactions between subunits encoded by the two genomes [33].
The nuclear DNA that a mother passes to her child is itself a recombined mixture inherited from both of her parents. It is not a fixed set of genes transmitted intact with her maternal mtDNA lineage. Mitonuclear compatibility must therefore be evaluated at the level of specific mitochondrial and nuclear variants, not inferred from parental origin alone.
An evolutionary example comes from archaic human ancestry. Sharbrough et al. noted that Neanderthal-derived nuclear DNA persists in modern humans without detectable Neanderthal mtDNA. However, their analysis suggested that mitonuclear incompatibility played, at most, a small role in shaping genome-wide introgression patterns [34].

3.4.3. The Incompatibility Hypothesis: Paternal Nuclear Genes and Maternal Mitochondria

I hypothesize that, in some individuals, paternally inherited nuclear alleles encoding mitochondrial proteins may be suboptimally matched to maternally inherited mtDNA, producing mild mitonuclear incompatibility. Because efficient electron transport depends on precise interactions between nuclear- and mtDNA-encoded subunits, such mismatches could reduce OXPHOS efficiency and promote mitochondrial dysfunction even in the absence of a classical mitochondrial-disease mutation [33,35].
In this account, inflammation arises from mitochondrial stress rather than from immune recognition of mismatched genomes. Damaged mitochondria release components—including mtDNA, cardiolipin, and N-formyl peptides—that act as damage-associated molecular patterns (DAMPs) and activate innate immune receptors, including Toll-like receptor 9 and formyl peptide receptors [36]. Released mtDNA and mitochondrial stress can also activate NF-κB signaling and the NLRP3 inflammasome [37]. Together, these pathways could sustain low-grade pro-inflammatory cytokine production and create a self-reinforcing cycle: mitochondrial dysfunction promotes DAMP release, DAMP signaling drives inflammation, and inflammation further impairs mitochondrial function. This process would not require an external inflammatory exposure associated with maternal autoimmune disease.
Experimental hybrid systems demonstrate that mismatched nuclear and mitochondrial genomes can impair respiration: disrupted mitonuclear coadaptation produces measurable organelle dysfunction and reduced fitness [35]. Whether analogous but subtler mismatches contribute to mitochondrial dysfunction in autism has not been directly tested. This mechanism therefore remains a speculative extension of the framework.

3.4.4. Two Pathways to the Same Outcome

The mitonuclear incompatibility hypothesis does not replace the familial and maternal autoimmune pathway—it complements it by providing a second pathway to pro-inflammatory cytokine-mediated mitochondrial dysfunction:
Table 3. Two Pathways to Pro-Inflammatory Cytokine-Mediated Mitochondrial Dysfunction.
Table 3. Two Pathways to Pro-Inflammatory Cytokine-Mediated Mitochondrial Dysfunction.
Pathway 1: External Pathway 2: Internal
Source of Inflammatory Signaling Maternal inflammatory signaling, potentially shaped by familial immune susceptibility Mitonuclear incompatibility
Mechanism Inherited or shared susceptibility and, when maternal inflammation is present, gestational exposure Mismatch between bi-parentally inherited nuclear OXPHOS genes and maternally inherited mtDNA, impairing mitochondrial function
Diagnosed parental disease required? No; susceptibility or inflammation may be subclinical No
Final common pathway Pro-inflammatory cytokine signaling → Mitochondrial dysfunction → Energy deficit → Autism
Note: Both pathways converge on the same final mechanism of pro-inflammatory cytokine-mediated mitochondrial dysfunction.
This framework is consistent with the following observations:
  • Familial autoimmune history and maternal autoimmune disease are associated with elevated autism prevalence (Pathway 1)
  • Autism also occurs without known familial or maternal autoimmune disease (Pathway 2)
  • Only a subset of children with familial or maternal autoimmune risk factors develop autism (variable mitonuclear compatibility may be protective or additive)

3.4.5. Testable Predictions and Preliminary Evidence

The mitonuclear incompatibility hypothesis generates testable predictions:
  • Anti-mitochondrial antibodies or mitochondria-targeted immune markers may be elevated in autistic individuals without known familial or maternal autoimmune history
  • Inflammatory cytokines may be elevated even in autism cases without known familial or maternal autoimmune disease
  • Specific combinations of maternal mtDNA haplogroup and paternal nuclear-genetic background may show associations with autism risk, with greater mitonuclear mismatch conferring greater risk
Notably, this prediction is consistent with existing findings, though the available evidence speaks to the broader premise of mitochondrial damage rather than to mitonuclear incompatibility specifically. Zhang et al. [38] reported that serum from young autistic children contains significantly elevated levels of anti-mitochondrial antibody Type 2 (AMA-M2; p = 0.001) and extracellular mitochondrial DNA (cytochrome B, p = 0.0002) compared to neurotypical controls. AMA-M2 is classically associated with primary biliary cholangitis (PBC)—an autoimmune disease characterized by immune-mediated destruction of mitochondria-rich biliary epithelial cells. The presence of this antibody in autistic children is consistent with the release of, and a secondary immune response to, mitochondrial components in autism—the expected downstream consequence of mitochondrial damage in the mechanism proposed here. Whether parental PBC is associated with elevated offspring autism risk remains untested and represents an additional testable prediction of this hypothesis.

3.5. The Decidual-Placental Transmission Pathway: How Pro-Inflammatory Signals Reach the Fetal Brain

The preceding sections have identified multiple sources of prenatal pro-inflammatory signaling relevant to autism risk. However, a critical spatial question remains insufficiently addressed: through what anatomical pathway do maternal or endogenous inflammatory signals reach the developing fetal brain? I propose that the decidua basalis—the specialized maternal tissue forming the immunological core of the placenta—represents a key intermediate locus in this transmission chain.

3.5.1. The Decidual Immune Tolerance Circuit

The decidua basalis harbors a high density of immune cells, with uterine natural killer (uNK) cells representing the dominant leukocyte population, accompanied by decidual macrophages and regulatory T cells (Tregs) [39,40]. In normal pregnancy, these populations maintain a coordinated tolerogenic circuit: Tregs suppress maternal immune responses against paternally derived fetal antigens, macrophages adopt an anti-inflammatory M2 phenotype producing IL-10 and TGF-β, and uNK cells promote spiral artery remodeling through controlled IFN-γ secretion rather than cytotoxic activity [40]. This circuit ensures that the local immune environment at the maternal-fetal interface remains anti-inflammatory and supportive of placental development.

3.5.2. Disruption of the Tolerogenic Circuit

I hypothesize that, in pregnancies affected by maternal autoimmune disease, chronic inflammation, obesity, or other predisposing conditions described in Section 3.1, Section 3.2 and Section 3.3, this tolerogenic circuit may become dysregulated through three converging mechanisms [39,40]. First, Treg insufficiency may weaken suppression of pro-inflammatory responses at the maternal–fetal interface. Second, a shift in macrophage polarization from M2 (anti-inflammatory) toward M1 (pro-inflammatory) may increase local production of TNF-α, IL-6, IL-1β, and IFN-γ. Third, uNK cell dysfunction may shift controlled IFN-γ-mediated vascular remodeling toward excessive pro-inflammatory cytokine release, impairing spiral artery remodeling and contributing to chronic placental hypoperfusion. These mechanisms may reinforce one another: Treg insufficiency can promote M1 macrophage polarization, further destabilizing the local immune environment.

3.5.3. Placental Amplification and Fetal Exposure

A critical feature of this pathway is that the placenta does not function as a passive filter. Placental trophoblast cells express receptors for TNF-α and IL-6, and respond actively to inflammatory stimulation by increasing their own cytokine production [41]. This means that a relatively modest inflammatory signal originating in the decidual compartment can be amplified through placental relay, resulting in disproportionate cytokine exposure at the fetal level. This amplification mechanism may explain a clinically important observation: why ASD occurs in offspring of mothers who appear outwardly healthy during pregnancy.

3.5.4. Subclinical Decidual Inflammation: Why Healthy-Appearing Mothers Can Have Affected Offspring

A central implication of this model is that clinically significant immune dysregulation at the decidual level need not manifest as overt maternal illness. The decidual immune environment is relatively compartmentalized from the systemic maternal circulation [39,40]. Its immune cell composition and cytokine profile are shaped by local factors, including progesterone signaling, trophoblast-derived HLA-G, and paracrine interactions among resident immune cells. Consequently, a mother may exhibit normal or subclinical systemic inflammatory markers while harboring a substantially dysregulated decidual immune environment. This local–systemic dissociation differs from conditions such as preeclampsia or cytokine storm, in which maternal systemic inflammation is clinically apparent. The model therefore proposes that “quiet” decidual inflammation—below the threshold of clinical detection—may be amplified through placental relay and impair fetal brain mitochondrial function during critical developmental windows.
The timing of this effect may also matter. Although immune dysregulation at the decidual–placental interface may exert effects throughout gestation, the second trimester may represent a particularly important convergence window. During this period, placental transport function matures and decidual immune activity is high [39,40,41]. Microglia also accumulate prominently in the fetal cerebral cortex [42], while synaptogenesis in cortical layer I is established and increasing [43]. However, the timing of maximal vulnerability likely varies across individuals and may depend on the nature, severity, and duration of the maternal immune perturbation.
This framework helps resolve a persistent puzzle in autism epidemiology: why the majority of mothers who deliver autistic children have unremarkable pregnancy histories. If the operative locus of immune dysregulation is the decidua rather than the systemic maternal compartment, then conventional prenatal screening—which relies on systemic markers—would be expected to miss these cases.

3.5.5. Microglial Reprogramming: The Bridge from Placental Inflammation to Aberrant Synaptic Pruning

The pathway from decidual-placental inflammation to disrupted fetal brain development requires a cellular intermediary within the fetal brain itself. Microglia—the resident immune cells of the central nervous system—serve this role. During human embryonic and fetal development, microglia arise from yolk-sac-derived macrophage lineages [44] and enter the developing cerebral wall from early gestation, with prominent regional accumulation during the second trimester [42]. Across developmental models, microglia perform essential functions including synaptic remodeling, neuronal circuit refinement, and clearance of apoptotic cells [45,46]. These functions are activity-dependent and require precise calibration of microglial state.
In mouse maternal immune activation models, prenatal inflammatory exposure produces persistent changes in microglial inflammatory and synapse-regulatory states [47,48]. Within the present framework, analogous fetal microglial reprogramming could have two consequences directly relevant to autism. First, dysregulated microglia may fail to execute synaptic pruning adequately, disrupting the normal refinement of neural circuits during critical developmental windows. This aligns with the excess synaptic density observed in postmortem autism brain tissue (Section 5.1). Second, and central to the present hypothesis, reprogrammed microglia may become a sustained local source of pro-inflammatory cytokines within the brain. They may thereby relay and prolong signaling that constrains neuronal mitochondrial function (Section 4). In this role, microglia are not proposed as the origin of the immune insult but as a potential intracerebral amplifier of prenatal inflammatory signaling.
This amplifier sits where it can do the most harm. Microglia are the main resident immune cells of the central nervous system parenchyma and are confined to the brain and spinal cord. The cytokine load they may sustain is therefore concentrated in the tissue least able to tolerate an energy shortfall, reinforcing the brain-specific vulnerability developed in Section 4.4. Microglia are maintained largely by local self-renewal rather than by replacement from the bone marrow [49], so a functionally altered state acquired during a critical prenatal window may be maintained after the maternal signal has resolved. Recent synthesis confirms that microglial involvement in autism is reproducible yet heterogeneous, varying with developmental stage, brain region, and sex rather than reducing to a single activation state [50]. Within the present framework the salient microglial contribution is specifically this: maintenance of a local cytokine environment that constrains neuronal energy metabolism.
Such low-grade immune dysregulation—below the clinical threshold of conditions such as preeclampsia or cytokine storm, yet above the level compatible with normal fetal neurodevelopment—may define an “autism-risk immune activation zone”in which clinically silent decidual inflammation remains biologically consequential for fetal neurodevelopment. The precise immunological parameters defining this zone, including cytokine thresholds, Treg counts, and decidual macrophage polarization ratios, remain to be established through future investigation.

5. Consequences of Cerebral Energy Deficit

The present model posits that chronic prenatal pro-inflammatory signaling, regardless of its precise upstream composition, leads to persistent mitochondrial dysfunction and cerebral energy deficiency that extends from fetal life into the postnatal period of rapid brain growth (Section 6). As established in Section 4.4, the brain’s heavy dependence on oxidative phosphorylation renders it particularly vulnerable to this energy deficit, which manifests in five critical domains that explain core autism symptoms and associated features.

5.1. Impaired Synaptic Pruning

Synaptic Refinement and Energy Supply: Synaptic pruning refines developing circuits by removing selected connections. In the human prefrontal cortex, dendritic spine density decreases from elevated childhood levels through adolescence and into the third decade of life [76]. The present framework proposes that persistent energy limitation disrupts this refinement by restricting the cellular work needed to remove and degrade synaptic material.
Human Evidence: Tang et al. found a smaller age-associated decline in dendritic spine density in ASD temporal cortex, alongside altered mTOR and autophagy-related markers (Table 7). Their mouse experiments established a role for neuronal autophagy in spine elimination, rather than demonstrating an energy deficit in microglia [77].
Prenatal Immune Activation and Synaptic Development: Experimental evidence implicates both synapse formation and elimination. Smith et al. identified IL-6 as a mediator of offspring abnormalities in a mouse MIA model [63]. Mirabella et al. showed that prenatal IL-6 elevation increases glutamatergic synapse density through a neuronal STAT3-dependent synaptogenesis program [78]. This provides an alternative route to excess synapses: increased formation rather than reduced pruning.
Other MIA studies link altered spine density to changes in microglial CX3CR1 expression and morphology [47,79]. Microglial depletion with a CSF1R inhibitor followed by repopulation corrected synaptic and behavioral abnormalities in one mouse model [80]. Liu et al. reported sex-specific microglial dysfunction during prefrontal development [81]. The direction of synaptic change is not uniform: Yan et al. found decreased spine density after MIA [48]. These studies show that prenatal immune activation can disrupt synaptic refinement, while ATP shortage remains the proposed mediator within the broader energy-deficit framework.
A Metabolic Link to Synapse Clearance: Monsorno et al. provide a more direct connection between microglial metabolism and synaptic refinement. Deleting the lactate transporter MCT4 impaired lysosomal acidification and synaptic cargo degradation in microglia, and disrupted hippocampal synapse refinement in developing mice [82]. This establishes a role for lactate transport and lysosomal function, not mitochondrial ATP depletion. Microglia also have metabolic flexibility: Bernier et al. showed that they can use glutamine to maintain surveillance when glucose is unavailable [83].
The Energy-Mediated Prediction: Within this framework, an ATP supply that cannot meet microglial demands limits synaptic engulfment and degradation during development. Testing this link requires measuring ATP availability and production as microglia take up and degrade a defined amount of synaptic material. Restoring ATP availability should improve clearance, with controls for direct effects on inflammatory signaling or lysosomal pH. Persistently adequate ATP availability during clearance, or failure of verified ATP restoration to improve clearance, would weaken this proposed mechanism.
Proposed Circuit Consequences: In this account, incomplete refinement leaves excess connections that increase local circuit noise, reduce the signal-to-noise ratio, and contribute to sensory overload. The framework also predicts long-range under-connectivity when energy limitation impairs axonal maturation and maintenance (Section 6.3).

5.2. Impaired Social Cognition and Gaze Avoidance

The Energy Demands of Social Processing: Social cognition—including face recognition, gaze processing, and emotion interpretation—is among the most computationally and energetically demanding brain functions. It requires simultaneous activation of:
  • Fusiform Face Area (FFA): Face identity processing
  • Superior Temporal Sulcus (STS): Gaze direction and biological motion
  • Amygdala: Emotional salience and threat detection
  • Prefrontal Cortex: Social context integration and decision-making
Eye Contact as Energy Conservation: This framework suggests that gaze avoidance in autism may represent an adaptive energy conservation strategy. The following first-person reports are phenomenological rather than direct measurements of metabolic cost, but they are consistent with this interpretation:
Table 8. Self-Reported Experiences of Eye Contact in Autism.
Table 8. Self-Reported Experiences of Eye Contact in Autism.
Experience Category Representative Quote
Energy Exertion “For me [eye contact] feels like I’m using up a lot of energy. The longest I can stare at someone in the eye is from less than 2 to 6 seconds at the most. Then it gets tiring.”
Audiovisual Integration “I can’t concentrate while making eye contact, particularly if I need to listen to what the other person is saying to me. It’s like I need to shut off the visual input in order to completely process the aural input.”
Source: Trevisan et al. [84], PLOS ONE - Qualitative analysis of first-hand accounts.
Neural Evidence: Functional neuroimaging studies show that eye contact elicits greater amygdala activation in autism, consistent with heightened neural engagement and hyperarousal [85]. Gaze avoidance may therefore reduce the processing burden of eye contact. The present framework predicts that this response is accompanied by greater metabolic demand, but direct metabolic measurements are needed to test this prediction (Section 5.5).

5.3. Excitatory–Inhibitory Imbalance: Epilepsy and Retinal Findings

The high comorbidity between autism and epilepsy provides additional support for the pro-inflammatory cytokine-mediated energy deficit hypothesis. Meta-analytic data indicate that approximately 21.5% of autistic individuals with intellectual disability experience epilepsy, compared to approximately 8% in autistic individuals without intellectual disability—both substantially elevated relative to the 1-2% prevalence in the general population [86].
Inflammatory Signaling and Excitation/Inhibition Balance: TNF-α can directly shift synaptic receptor trafficking toward excitation. In cultured hippocampal neurons, Stellwagen et al. [87] showed TNFR1-dependent exocytosis of GluA2-lacking AMPA receptors together with GABA_A receptor endocytosis, increasing excitatory and reducing inhibitory synaptic strength.
Mitochondrial Impairment and Seizure Vulnerability: Experimental evidence provides a complementary bioenergetic route. In mice with conditional loss of mitochondrial pyruvate carrier 1 in adult glutamatergic neurons, reduced oxidative phosphorylation and mitochondrial membrane potential produced marked seizure susceptibility under mild GABAergic inhibition [88]. Clinically, Whittaker et al. [89] found epilepsy in 23.1% of a prospective cohort of 182 adults with genetically defined mitochondrial disease, with prevalence varying substantially by genotype.
The autism–epilepsy comorbidity is therefore compatible with two convergent mechanisms in the model: inflammatory signaling can bias excitation/inhibition balance, while impaired mitochondrial energy production can reduce the energetic reserve needed to maintain stable network function.
Retinal Electrophysiology: A peripheral correlate of excitatory–inhibitory imbalance may be detectable in the retina, which shares both its developmental origin and its core neurochemistry with the brain. The electroretinogram (ERG) b-wave, a response generated predominantly by ON-bipolar cells and shaped by glutamatergic and GABAergic transmission, is reduced in ASD, a finding reported since the pilot study of Ritvo and colleagues [90] and observed in subsequent studies [91,92]. Because the b-wave reflects inner-retinal synaptic transmission, its attenuation is consistent with the excitatory–inhibitory imbalance predicted downstream of the proposed cerebral energy deficit, and provides a non-invasive, in vivo window into that imbalance. These findings remain preliminary, deriving from small samples and showing less consistency in adults, but they motivate a direct empirical test (Section 8.1).

5.4. Impaired Protein Synthesis: The Critical Energy Bottleneck

Protein synthesis is among the cellular processes most sensitive to energy limitation: in concanavalin A-stimulated thymocytes, it showed the greatest control over ATP flux [93]. In addition, each amino acid incorporation requires ~4 ATP equivalents [94], making the synthesis of large synaptic proteins extraordinarily energy-demanding. During neurodevelopment, when neurons must produce vast quantities of synaptic proteins, any ATP deficit creates a critical bottleneck.

5.4.1. Mitochondrial Protein Synthesis as Rate-Limiting Step

The electron transport chain requires both nuclear- and mitochondrial-genome-encoded subunits. Critically, mitochondrial protein synthesis by the 55S ribosome is the rate-limiting step in ETC synthesis [95]. Of the 230 genes in the Mitochondrial Central Dogma, 59 are associated with neurodevelopmental delay, representing a 2-fold enrichment (p < 8.95E-9) [95]. Disorders associated with these mitochondrial dysfunction-related genes often manifest clinically during the period of peak glutamatergic synapse density, consistent with a developmental link between energy consumption and brain maturation.
Table 9. Representative Synaptic Proteins and Translational Regulators Relevant to ASD.
Table 9. Representative Synaptic Proteins and Translational Regulators Relevant to ASD.
Protein Function ASD Evidence Relevance to Protein-Synthesis Bottleneck
SHANK3 Synaptic scaffold Rare SHANK3 variants/deletions contribute to ASD; meta-analytic estimates place SHANK3 mutations in approximately 0.5-2% of cases [96] Large postsynaptic scaffold requiring extensive protein synthesis
NRXN/NLGN Trans-synaptic adhesion Rare human NRXN1 disruptions and NLGN3/NLGN4 mutations have been reported in ASD [97,98] Synthesis and membrane trafficking of synaptic adhesion proteins
FMRP Neuronal mRNA translation regulator Loss of FMRP causes Fragile X syndrome with autistic features; FMRP binds mRNAs linked to synaptic function and ASD [99] Direct regulation of ribosome translocation and neuronal protein synthesis
Note: Examples are illustrative; the table does not quantify ATP expenditure for individual proteins.

5.4.2. The SHANK3-Mitochondria Connection

Deletions involving SHANK3 in Phelan-McDermid syndrome vary in size and may include other genes with mitochondrial functions in the broader 22q13 region. These include SCO2 (cytochrome c oxidase assembly), NDUFA6 (Complex I), TYMP (mitochondrial DNA metabolism), TRMU (mitochondrial tRNA modification), CPT1B (fatty acid metabolism), and ACO2 (TCA cycle) [100].
Frye et al. found respiratory-chain enzyme activities outside the control range in buccal cells from 30 of 51 participants (59%). Abnormalities were most prominent in Complexes I and IV, with Complex I changes in both directions. Specific gene deletions were not significantly associated with these abnormalities in the subgroup with genetic data [100]. Within this framework, impaired mitochondrial energy supply would compound the synaptic consequences of SHANK3 loss, creating a dual synaptic-bioenergetic vulnerability.

5.5. Restricted Repetitive Behaviors and Insistence on Sameness: Behavioral Compensation under Energy Constraint

The second core diagnostic domain of ASD under DSM-5—restricted and repetitive behaviors (RRBs), including insistence on sameness, ritualistic ordering, and resistance to environmental change—has historically been explained by frameworks distinct from the immune-metabolic mechanisms developed above. Longitudinal research supports a distinction between insistence on sameness (IS) and repetitive sensorimotor behaviors across development. In a cohort followed from ages 2 to 19, both types of behavior generally became less severe, although some participants showed increases between ages 2 and 9 [101]. I propose that the same chronic cerebral energy constraint described in Section 5.1, Section 5.2, Section 5.3 and Section 5.4 contributes to IS by limiting flexible predictive inference.
Cortical computation is metabolically expensive. Synaptic transmission alone consumes on the order of 104 ATP molecules per bit of transmitted information, and action potentials together with postsynaptic potentials account for the majority of the brain’s energy budget [102]. Predictive coding accounts of cortical function describe perception as hierarchical inference: top-down predictions meet bottom-up sensory input, and the discrepancy—the prediction error—propagates upward to update the internal model [103,104]. In practical terms, predictable input generates less error signaling and requires less model updating, whereas unexpected input increases neural signaling, model updating, and energy demand. The link is not merely conceptual: variational free energy (the quantity that bounds the surprise of sensory observations) and the ATP cost of neural signaling are closely linked [105].
A 2024 bioRxiv preprint by Hechler and colleagues provided preliminary direct support for this prediction in the human brain, using multiparametric quantitative BOLD (mqBOLD) to measure cerebral oxygen consumption (CMRO2) during visual sequences of varying predictability [106]. Processing of predictable input cost less metabolically as subjective confidence in the pattern increased, with a cortex-wide reduction of approximately 12% (95% CI 4–19%)—approximately 118 µmol O2 per minute across cortical grey matter. Critically, the saving was conditional on confidence rather than on objective predictability: low-confidence subjects showed no benefit even when the input was, in fact, predictable. Subjective precision—the brain’s confidence in its own prior—therefore appears to be the operative variable linking prediction to metabolic efficiency. Independently, computational modeling has shown that predictive-coding-like architecture emerges spontaneously in recurrent neural networks trained under an energy-minimization objective, without that connectivity being specified in advance [107]. Together, these results suggest that predictive processing is not merely a computational strategy but a metabolic one: the brain reduces its energy expenditure by ensuring that incoming input matches a confidently held internal model.
This metabolic view of prediction has direct implications for the cerebral energy deficit proposed here. Multiple computational accounts of autistic perception converge on a common feature: an atypically high or inflexible weighting of sensory prediction errors relative to prior expectations [108,109,110]. Active-inference modeling has recently refined this picture. What is atypical in autism may not be chronically elevated prediction-error precision—the weight assigned to prediction errors—but a failure to adjust that weighting flexibly in response to environmental volatility [111]. On the present account, this inflexibility is a consequence rather than a primitive feature. Flexible adjustment requires gain control and hierarchical message passing [112], while neural signaling carries substantial metabolic costs [102,105]. A brain under chronic ATP constraint—the downstream consequence of sustained mitochondrial dysfunction (Section 4)—may therefore have reduced capacity for moment-to-moment recalibration. Inflexible precision weighting may thus be the inferential expression of cerebral energy deficit.
On this view, restricted repetitive behaviors and insistence on sameness can be reinterpreted as a behavioral compensation. Rather than recalibrating internal predictions to track a changing environment—a metabolically expensive process—an individual may instead constrain the environment to match an existing internal model. Lining objects up, keeping strict routines, preferring familiar stimuli, and resisting changes in context all reduce the entropy of incoming sensory input and thereby suppress prediction error at its source [113]. Śliwiński [113] formalizes this environmental-constraint strategy in information-theoretic terms; the present hypothesis extends that account by proposing chronic mitochondrial energy limitation as a biological mechanism that may increase the relative cost of continual model updating. When cerebral energy is plentiful, such constraint is unnecessary: the brain can afford to update its model in real time. Under chronic energy constraint, by contrast, shaping the environment behaviorally becomes the cheaper option. This account does not displace existing explanations of RRBs as anxiety- or sensory-regulatory behaviors [114]; it provides a bioenergetic foundation that unifies them.
Direct empirical investigation of the metabolic cost of predictive processing has only recently become possible. Most prior work has used standard BOLD fMRI, which cannot disentangle oxygen consumption from confounding effects of blood flow and oxygen supply; its signal changes therefore do not directly index energy metabolism [115]. Quantitative methods that assess cerebral metabolism—mqBOLD for CMRO2 and functional FDG-PET for glucose utilization—have matured over the past decade [116,117]. The empirical literature directly testing the energy–prediction link is therefore nascent, with the first direct human demonstration emerging only in 2024 [106]. Application of these methods to autism remains essentially unexplored and represents a high-priority direction for testing the present framework empirically.
This account generates several directly testable predictions. First, autistic individuals with high IS scores should differ from neurotypical controls in cerebral metabolic rate during periods of routine or predictable input, and should show disproportionately elevated metabolic responses to environmental disruption. The methodologies cited above [106,117] provide a direct experimental pathway. Second, animal models of maternal immune activation that show both mitochondrial dysfunction and stereotyped phenotypes should exhibit altered metabolic dynamics specifically during behavioral transitions between routine and novel conditions. Third, interventions that improve mitochondrial bioenergetic capacity (Section 7.4) should attenuate IS-domain symptoms more than repetitive sensorimotor symptoms, since IS is hypothesized to reflect inferential-level energy compensation specifically. Direct empirical evidence linking RRB or IS severity to cerebral metabolism does not yet exist; this prediction therefore stands as one of the most directly verifiable claims of the present framework, and a priority target for future investigation.

6. Postnatal Persistence of Mitochondrial Dysfunction

Although the present hypothesis locates the initiating insult in prenatal pro-inflammatory exposure, the resulting mitochondrial dysfunction does not resolve at birth. This section addresses a question essential to the clinical coherence of any developmental-origin model of ASD: why, if the maternal prenatal exposure has ended, do the neurodevelopmental consequences persist and even progressively manifest during the early postnatal period? Four interrelated aspects of this framework—molecular persistence, postnatal energy demand, regional vulnerability, and clinical regression—together explain why a prenatal bioenergetic insult produces lifelong neurodevelopmental consequences that often become most clinically apparent well after birth.

6.1. Mechanisms of Postnatal Persistence

Within this framework, mitochondrial dysfunction initiated prenatally becomes self-sustaining through a feedback loop in which ongoing damage outpaces mitochondrial repair and renewal. Long-lived neurons are largely post-mitotic, but their mitochondria remain subject to quality control. Basal mitophagy has been demonstrated in adult mouse neurons [118], and cell studies show that oxidatively damaged mtDNA can be repaired or degraded [119]. This model therefore requires either impaired quality control or a damage burden that overwhelms otherwise functional repair, clearance, and replacement. Evidence from selected ASD-related experimental models implicates altered mitophagy, but its contribution to persistent dysfunction in human ASD remains uncertain [120].
In this proposed loop, mitochondrial stress and local inflammatory signaling sustain one another. Damaged mitochondria generate reactive oxygen species (ROS), which cause further cellular damage, while mitochondrial danger signals activate inflammatory pathways (Section 4.3.2). Continued cytokine signaling, in turn, places additional stress on mitochondria. Microglia reprogrammed by prenatal inflammatory exposure are proposed to sustain local cytokine signaling in response to these mitochondrial danger signals (Section 3.5.5). This feedback maintains the imbalance between injury and renewal after the maternal prenatal exposure has ended. The resulting dysfunction can therefore persist despite ongoing mitochondrial turnover, without requiring the same mitochondria damaged before birth to survive.
Clinical studies establish that mitochondrial and bioenergetic abnormalities are detectable after birth. Giulivi et al. [121] reported impaired mitochondrial NADH oxidase activity (p = 0.001), Complex I deficits (in 6 of 10 cases), and mtDNA abnormalities in peripheral lymphocytes of autistic children aged 2–5 years. Complementing this early-childhood peripheral evidence, Goh et al. [122] detected in-vivo brain lactate elevations across several regions in a subset of individuals with ASD, more frequently in adults than children.
Experimental animal evidence further substantiates the postnatal persistence proposed here. Cieślik et al. [123], using a rat maternal immune activation (MIA) model, demonstrated developmental stage-dependent changes in brain mitochondrial function in offspring, with persistent alterations extending well beyond the prenatal exposure period. Schneider Gasser et al. [124] reported that prenatal poly(I:C)-induced MIA in mice produces long-term alterations in adult offspring brain mitochondrial function, including changes in respiratory capacity, mitochondrial mass, and uncoupling protein expression in the prefrontal cortex and amygdala—regions central to social cognition and core ASD phenotypes. Beyond the immune activation paradigm, Stier et al. [125] experimentally demonstrated prenatal programming of mitochondrial aerobic metabolism that persists into adulthood, establishing that prenatal mitochondrial programming can produce lifetime metabolic consequences. The underlying mechanistic cascade—maternal inflammation → cytokine release → oxidative stress and altered mitochondrial biogenesis (PGC-1α, NRF2, TFAM) → disrupted fetal and postnatal mitochondrial function—has been comprehensively reviewed by Zawadzka et al. [126] in the context of MIA-induced autism pathogenesis.
Human evidence, while necessarily more limited than animal studies, points in the same direction. Gyllenhammer et al. [127] reviewed the role of mitochondria in linking maternal inflammation during pregnancy to offspring brain development, identifying mitochondrial dysfunction as a plausible mediator. In a prospective human cohort, Gyllenhammer et al. [128] found that maternal allostatic load during pregnancy was associated with altered mitochondrial content and bioenergetic capacity in peripheral blood mononuclear cells of children years after birth, linking prenatal maternal allostatic burden to measurable later-childhood mitochondrial differences. Complementing this developmental evidence, Frye et al. [32], in a systematic review and meta-analysis, established that biomarkers of mitochondrial dysfunction are robustly associated with ASD across multiple tissue types and assay platforms, providing strong meta-analytic support for the link between mitochondrial dysfunction and the autistic phenotype that the present framework predicts.

6.2. The Postnatal Brain’s Energy Demand Trajectory

The early postnatal period is the most metabolically demanding phase of brain development. As noted in Section 4.4, the developing brain’s energy use climbs after birth to a peak in early childhood of approximately 40% of total body energy [72], far exceeding the adult proportion. This elevated demand reflects intense neural activity during synaptogenesis, axonal growth, myelination initiation, and—beginning in mid-childhood—extensive synaptic pruning. Each of these processes is profoundly ATP-dependent. Direct measurements bear out this escalating demand: positron emission tomography shows that the cortical metabolic rate for glucose rises steeply through infancy to surpass adult levels in early childhood [129], while the developing brain shifts from neonatal reliance on ketone bodies to adult-like glucose dependence by around two years of age [130].
Critically, this is the period during which the gap between energy demand and energy-producing capacity may become most clinically consequential. In a typically developing brain, mitochondrial biogenesis and function rise to meet the surging postnatal demand. In a brain whose mitochondria were compromised prenatally, the supply cannot scale appropriately. The shortfall is not constant but widens precisely during the windows when developmental processes are most sensitive—a dynamic that may help explain why ASD symptoms typically become clinically recognizable between 12 and 24 months of age [131], well after the maternal prenatal exposure has ended. The model thus predicts that the postnatal manifestation of ASD reflects not a new pathological event, but the moment at which a pre-existing energy-supply deficit can no longer meet a rapidly escalating developmental energy demand.

6.3. Selective Vulnerability of High-Metabolism Brain Regions

Cerebral energy deficit is unlikely to affect all brain regions uniformly. Regions characterized by high basal metabolic rate, dense synaptic activity, or computationally intensive processing should be disproportionately vulnerable to ATP shortfall. This regional gradient may help explain the specific cognitive and behavioral profile of ASD, in which certain domains are characteristically affected while others remain relatively preserved. The prefrontal cortex (PFC) sustains high baseline metabolic activity to support executive function, working memory, and social cognitive processing—domains commonly affected in ASD—while the hippocampus exhibits exceptionally active synaptic plasticity in support of memory formation and contextual learning, processes that are similarly disrupted in ASD.
A particularly informative case is the left inferior frontal region encompassing Broca’s area, whose involvement in ASD is suggested by converging clinical and neuroimaging observations. Clinically, a substantial subset of autistic children exhibit a disproportionate impairment of grammatical morphology—the closed-class function morphemes that mark tense, agreement, and grammatical relations—relative to lexical or content-word knowledge [132,133]. Critically, this disproportionate vulnerability has been documented across typologically diverse languages, including English (tense and agreement marking [132]), French (production of pronominal clitics, where errors closely parallel those seen in specific language impairment [134]), and Greek (clitic pronouns, definite articles, and broader morphosyntactic structures [135]). The convergence of evidence across languages with fundamentally different grammatical systems—where the affected morphemes carry very different surface forms and functional loads—indicates that the vulnerability lies not in any specific surface form but in the underlying computational machinery that handles rapid, rule-governed processing of grammatical markers.
Neuroimaging studies converge on this same region. Just et al. [136] reported reduced relative activation of Broca’s area together with reduced functional connectivity between Broca’s and Wernicke’s areas in ASD during sentence comprehension, supporting an underconnectivity account of autism’s language network. Harris et al. [137] independently reported substantially reduced Broca’s area activation with increased left temporal (Wernicke’s) activation during semantic processing in ASD, leading the authors to characterize Broca’s area as a region of abnormal neurodevelopment in ASD. Subsequent fMRI work has documented further patterns of atypical Broca’s area engagement, including altered activation magnitude and reduced left lateralization during semantic processing in adolescents with ASD [138]. These studies report different directional patterns, but they converge on Broca’s area as a recurring site of atypical functional engagement in ASD. Whether this pattern reflects a shared bioenergetic vulnerability remains unknown.
The present framework offers a testable bioenergetic account. At the lobe level, developmental PET shows that frontal glucose utilization rises by approximately 6–8 months, while cortical metabolic rates reach adult values by about 2 years and exceed them in early childhood [129]. These findings do not establish that Broca’s area is exceptionally energy-demanding; rather, they motivate testing whether the atypical engagement and connectivity described above reflect a regional bioenergetic vulnerability during language development.
The same bioenergetic hypothesis extends to the long-range white-matter connections of the language network. Whereas grey-matter signaling consumes a large share of the brain’s energy budget [139], the white-matter tracts that integrate distant cortical regions impose their own sustained ATP burden, because building myelin and maintaining the oligodendrocyte resting potential are themselves energetically costly [140]. Myelin enables rapid conduction between distant regions, so an underbuilt tract would degrade the timing on which Broca’s and Wernicke’s areas depend even when both regions are intact. The arcuate fasciculus is the principal dorsal-stream tract relaying information between Broca’s and Wernicke’s areas [141,142]. As a long-range, late-myelinating connection, its construction and maintenance would be expected to suffer under a chronic energy deficit.
Diffusion-imaging studies in ASD are consistent with this expectation. The arcuate fasciculus shows increased mean diffusivity, driven mainly by increased radial diffusivity, together with diminished left-hemisphere lateralization [143]. These abnormalities are especially pronounced in toddlers with language regression [144]. Notably, altered lateralization of the dorsal language tracts is detectable in infants at elevated familial likelihood of autism as early as six weeks of age [145]—well before overt language emerges. The trajectory of white-matter development across infancy predicts later language ability in ASD [146]. The very early detection of this structural asymmetry is consistent with a developmental process beginning prenatally or in early infancy, complementing the postnatal-persistence evidence developed below.
Beyond these functional and clinical observations, direct biochemical and in vivo imaging studies provide convergent evidence of mitochondrial dysfunction in several high-metabolism brain regions in ASD. A positron emission tomography study reported reduced availability of mitochondrial complex I in the anterior cingulate cortex of autistic adults, with the magnitude of the deficit correlating with the severity of social-communication impairment [147]. Postmortem analysis of BA21 temporal cortex has shown decreased complex I and IV activity together with elevated oxidative damage [148]; region-specific reductions in electron-transport-chain complex activity have been documented across the frontal, cerebellar, and temporal cortices [149]; and elevated brain lactate has been detected across several regions in a subset of individuals with ASD [122].
Direct evidence of mitochondrial respiratory-chain dysfunction in Broca’s area is currently lacking. The bioenergetic interpretation advanced here rests on three indirect lines of evidence: the developmental rise in frontal cortical metabolism, the functional and connectivity abnormalities reported in Broca’s area, and mitochondrial deficits demonstrated elsewhere in the ASD brain. The framework therefore makes a specific, falsifiable prediction: region-targeted studies should detect reduced complex I availability or other mitochondrial respiratory-chain abnormalities in Broca’s area, with larger abnormalities in autistic individuals with greater language impairment. This prediction could be tested using high-resolution complex I PET, such as [18F]BCPP-EF, or postmortem biochemical assays.

6.4. Regressive Autism as a Clinical Signature of Cerebral Energy Crisis

A substantial subset of autistic children exhibit a regressive phenotype in which apparently typical early development is followed by loss of previously acquired skills. A recent systematic review and meta-analytical update reported a pooled prevalence of approximately 30%, with a weighted average age of onset close to 20 months and language and social domains most frequently affected [150]. The mechanistic question is not simply why development slows, but why previously acquired skills are lost.
Within this framework, regression occurs when ATP supply becomes insufficient to sustain the circuits supporting previously acquired skills. A mildly compromised mitochondrial supply may suffice to support early developmental milestones, but as synaptic density, neural activity, and circuit complexity escalate during the second year of life, the cumulative energy demand may exceed the impaired supply capacity. The resulting bioenergetic failure could disproportionately affect the most metabolically demanding circuits—precisely the language and social cognitive networks where regression is clinically observed. In this view, regressive autism is not a categorically distinct subtype, but a phenotypic expression of the same underlying mechanism under conditions in which the energy gap crosses a critical threshold during a specific developmental window. This interpretation predicts that mitochondrial markers and developmental energy demand should covary with regression timing and severity, a hypothesis that is empirically testable in prospective longitudinal cohorts.

7. Therapeutic Implications

The energy-deficit hypothesis suggests several therapeutic and research directions spanning both prenatal initiation and postnatal persistence. The interventions considered here target the inflammatory and bioenergetic injury itself and its associated medical burden — mitochondrial dysfunction, epilepsy, and gastrointestinal comorbidity.

7.1. Anti-Inflammatory Interventions

Among available anti-inflammatory biologics, anti-TNF agents provide the most relevant precedent for evaluating the therapeutic implications of the present hypothesis. Existing anti-TNF biologics (etanercept, infliximab, golimumab, adalimumab) have proven efficacy in TNF-α-mediated diseases. In type 1 diabetes, golimumab preserved β-cell function in a phase 2 trial [10]. These observations raise the possibility that anti-inflammatory strategies—including but not limited to TNF-α-targeted agents—could eventually be evaluated in carefully defined pregnancies at elevated likelihood, although substantial safety, ethical, and regulatory issues would first need to be resolved.
Preclinical evidence supports this therapeutic rationale. Liu et al. (2023) demonstrated in a preeclampsia mouse model that maternal TNF-α elevation drives ASD-like phenotypes in offspring through fetal NFκB signaling, and that TNF-α neutralization during pregnancy ameliorated these ASD-like behaviors and restored NFκB activation to normal levels [151]. These findings support a causal contribution of maternal TNF-α signaling to offspring ASD-like phenotypes in this preeclampsia model and show that the pathway is pharmacologically modifiable.
In humans, pregnancy-exposure studies primarily inform drug safety. In the PIANO registry, women with inflammatory bowel disease (IBD) receiving biologics, thiopurines, or both were compared with women with IBD unexposed to either drug class. Biologics included anti-TNF and other agents, and the comparator group could receive mesalamine, corticosteroids, or antibiotics. Adjusted analyses found no increased risk of congenital malformations, spontaneous abortion, preterm birth, low birth weight, or infant infections during the first year [152].
Nørgård et al.’s Danish anti-TNF analysis compared 493 exposed children with 728,055 unexposed children from the general birth population, not a cohort restricted to the same maternal inflammatory disease. Several categories of first-year infections were more frequent among exposed children. A composite of psychiatric diagnoses, ASD, and ADHD yielded an adjusted HR of 0.53 (95% CI 0.20–1.41), based on four events among exposed children [153]. This was not an ASD-specific estimate.
Separately, etanercept administered directly to preterm fetal sheep reduced inflammation-associated white matter gliosis [154]. This is preclinical evidence that TNF blockade can modify fetal neuroinflammation. Human safety observations and these experimental findings motivate further investigation, but do not establish an ASD-preventive effect. Testing the proposed benefit requires ASD-specific follow-up in clinically comparable maternal disease groups, accounting for disease activity and treatment timing (Section 8.4).
While anti-TNF agents provide the most developed precedent, the multi-cytokine character of the proposed mechanism implies that other convergent cytokines are equally plausible therapeutic targets. Preclinical evidence supports this broader view: a systematic review of immune-based perinatal neuroprotection found that all 10 included studies investigating anti-IL-1 therapies reported improved outcomes, with anti-IL-1 among the most consistently protective of all interventions examined [155]; in parallel, maternal IL-17A has been shown to cross the placenta and drive communicative and other neurodevelopmental deficits in offspring, identifying IL-17A signaling as a further candidate target [156]. Critically, however, this evidence remains almost entirely preclinical: no randomized controlled trial has tested whether anti-cytokine therapy during pregnancy alters offspring neurodevelopmental outcomes, and pregnant individuals continue to be excluded from trials of these agents. Closing this gap between robust animal data and the near-absence of human efficacy trials represents the central translational challenge for the therapeutic prediction advanced here [155].

7.2. Early Identification

If validated, familial autoimmune history and maternal inflammatory disorders (including psoriasis, T1D, and RA) could serve as clinically accessible markers of pregnancies at elevated autism likelihood, potentially enabling earlier surveillance and stratified research.

7.3. Toward Future Risk Stratification: Prenatal Cytokine Monitoring

A core implication of the present hypothesis is that pregnancies producing autistic offspring may involve subclinical, low-grade inflammatory states that escape detection by conventional prenatal screening, which relies on systemic markers and clinical symptomatology. If supported by sufficient subsequent evidence, prenatal cytokine monitoring—currently not part of routine obstetric care—may merit consideration by the medical community as a candidate strategy.
Such monitoring would face substantial methodological challenges. These include: (1) establishing population-specific reference ranges for pregnancy-relevant cytokines, given that some degree of pro-inflammatory signaling is physiologically necessary across gestation; (2) distinguishing physiologically necessary cytokine fluctuations from pathogenic ones, particularly across the implantation, mid-gestation, and parturition windows in which immune profiles normally shift; and (3) identifying which cytokines, or cytokine combinations, carry the strongest predictive value for offspring neurodevelopmental outcomes. Furthermore, the appropriate clinical response to detected elevations remains undefined, as targeted anti-inflammatory interventions during pregnancy require careful consideration of fetal safety.
Nevertheless, the cumulative evidence reviewed here suggests that systematic investigation of this approach—beginning with longitudinal cohort studies linking maternal cytokine profiles across gestation to offspring neurodevelopmental outcomes—is warranted as a foundation for any future clinical translation. Such studies would also help establish whether the hypothesized “autism-risk immune activation zone” (Section 3.5.5) can be empirically defined.

7.4. Postnatal Interventions

The postnatal persistence component of this framework opens an additional therapeutic window beyond pregnancy. If mitochondrial dysfunction continues to operate during the first years of life as proposed in Section 6, several postnatal candidate strategies merit longitudinal investigation.
First, postnatal mitochondrial assessment: prospective monitoring of mitochondrial biomarkers—including lactate, lactate-to-pyruvate ratio, plasma carnitine profiles, and emerging mtDNA copy-number assays—in infants at elevated likelihood (for example, those with relevant familial autoimmune history or maternal inflammatory disease) could identify children whose mitochondrial trajectories deviate from typical development before behavioral symptoms become apparent.
Second, postnatal anti-inflammatory and mitochondrial-supportive strategies: the reciprocal loop proposed in Section 6.1 identifies both inflammatory signaling and mitochondrial dysfunction as treatment targets. The model predicts that interrupting either component could weaken the loop. It therefore supports testing both approaches rather than assigning anti-inflammatory treatment priority solely because inflammation initiated the process.
Safety must be evaluated for each agent, age group, indication, and treatment duration. Postnatal treatment cannot be assumed to have a more favorable risk-benefit profile than treatment during pregnancy. Anti-inflammatory biologics have been studied in young children with defined inflammatory diseases, including adalimumab in juvenile idiopathic arthritis [157]. Such disease-specific evidence does not establish the benefit-risk balance of prolonged treatment for the low-grade inflammatory state proposed here.
Mitochondrial-supportive nutrients also warrant investigation. Small placebo-controlled L-carnitine trials and open-label coenzyme Q10 studies have reported behavioral improvements in children with ASD, largely in cohorts not selected for confirmed mitochondrial dysfunction [158]. The framework predicts that benefit would be most detectable in participants with relevant immune or mitochondrial abnormalities. Biomarker-stratified trials should test that prediction, measure both biological and developmental outcomes, and examine combined approaches where safety data justify them.
Third, early behavioral intervention coupled with bioenergetic monitoring: existing early behavioral interventions for ASD demonstrate variable efficacy; pairing them with mitochondrial biomarker assessment could clarify whether bioenergetic state modulates intervention response and whether mitochondrial support enhances behavioral outcomes.
These postnatal strategies complement the prenatal research directions in Section 7.1, Section 7.2 and Section 7.3. Prenatal initiation provides a rationale for investigating intervention before developmental injury occurs, whereas postnatal persistence identifies a potential opportunity to interrupt ongoing dysfunction. Because the developmental trajectory may already be altered by birth, postnatal studies must distinguish improvement in current bioenergetic function from reversal of earlier circuit changes. The model does not establish which intervention window offers the greatest clinical benefit or the lowest risk. Longitudinal studies should determine which processes remain modifiable, at what stage, and whether biological improvement translates into better developmental outcomes.

8. Future Directions

The framework advanced here yields several concrete, falsifiable predictions. I outline them not as settled claims but as an invitation to the research community, in the hope that they will be tested directly.

8.1. Retinal Electrophysiology

The reduced ERG b-wave described in Section 5.3 should be examined in a large, multi-site cohort spanning the full range of age and symptom severity, using standardized (ISCEV) protocols. I hope future work will establish whether b-wave attenuation is a developmentally stable peripheral marker of excitatory–inhibitory imbalance and whether it tracks symptom severity.

8.2. Therapeutic Trials

The model implies that interventions restoring mitochondrial function or lowering pro-inflammatory cytokine load (Section 7) could alter the developmental trajectory. I hope anti-inflammatory and mitochondrial-support strategies, such as NAD+ precursors, will be evaluated in carefully designed trials attentive to timing and to biomarker-defined subgroups.

8.3. Mitochondrial and Cerebral Bioenergetic Trajectories

I hope future work will chart these bioenergetic trajectories directly. Prospective studies measuring mitochondrial function in infants at elevated autism likelihood, integrating prenatal markers such as placental and cord-blood mtDNA copy number with longitudinal postnatal assessment, would test whether the prenatal deficit persists across the first years of life (Section 6). I also hope cerebral energy metabolism will be measured in vivo, for example by mqBOLD CMRO2 or functional FDG-PET, in individuals stratified by insistence-on-sameness severity and contrasting predictable input with environmental disruption, to test the account advanced in Section 5.5.

8.4. Mechanistic and Epidemiological Studies

Finally, the proposed mechanisms should be probed more directly.
  • Prenatal inflammation and synaptic bioenergetics. Animal models could clarify how prenatal pro-inflammatory states alter synaptic pruning and neuronal bioenergetics.
  • Chronic low-grade exposure. The effects of chronic low-to-moderate cytokine elevation on fetal brain mitochondria are unmapped, since the mechanistic literature rests on acute exposure paradigms. Animal models sustaining low-grade elevation across gestation would close this gap.
  • Regional bioenergetics of the language network. Developmental PET evidence shows a maturational rise in frontal glucose utilization but does not resolve Broca’s area separately [129]. Region-resolved CMRO2 using non-ionizing MRI-based methods, supplemented by secondary analysis of clinically acquired pediatric FDG-PET data, could test whether Broca’s area and the broader language network are unusually metabolically demanding during typical language development.
  • Registry analysis. Disease-severity-adjusted registry analyses could compare offspring outcomes across pregnancies with differing inflammatory burdens and immune-modifying treatments, and link maternal cytokine trajectories to offspring outcomes.
  • Partner-specific maternal immune adaptation. If partner-specific maternal immune adaptation contributes to the birth-order effect, ASD incidence among second-born children should be higher following a change in paternity than among same-partner second-borns. Population-based registries linking each child to both biological parents allow this comparison, with interpregnancy interval and parental age considered explicitly.
  • Maternal- versus paternal-line autoimmune history. Multigenerational registry studies should compare ASD risk associated with autoimmune disease, including type 1 diabetes, in maternal and paternal relatives of the same degree. If maternal immune susceptibility contributes through both inheritance and the gestational environment, maternal-line history may show a stronger association than equivalently related paternal-line history.
  • Decidual immune parameters. Targeted study of macrophage polarization, regulatory T-cell density, and local cytokine concentrations in prospectively followed pregnancies could test the maternal–fetal transmission pathway.

9. Limitations

This theoretical synthesis presents no new experimental data. The epidemiological associations are compatible with the proposed pathway, but do not establish causation or how much of ASD it explains. Shared genetic, environmental, and ascertainment factors may contribute; family-history definitions vary, and within-family comparisons remain limited for several conditions reviewed here. The effect estimates in Table 1 concern different exposures and populations. Elevated cytokines in ASD likewise do not establish whether inflammation is a cause, a consequence, or both.
The mitochondrial evidence is context-dependent. Cytokine effects on respiration vary across experimental systems, and results from peripheral cells cannot be assumed to describe fetal brain metabolism (Section 4.3). The central prediction—ATP supply insufficient for developmental demand—requires direct testing; altered respiration alone does not demonstrate this shortfall. The relevant cytokine combinations, exposure thresholds, dose–response relationships, and gestational windows remain unresolved, especially under chronic low-grade exposure. Mitonuclear incompatibility is a separate, untested extension: circulating mtDNA and anti-mitochondrial antibodies reported by Zhang et al. [38] do not establish a mismatch between nuclear and mitochondrial genomes.
The decidual–placental and microglial links also require more direct evidence. The studies reviewed here do not separate local decidual immune activity from systemic maternal inflammation in prospective human ASD cohorts. Experimental IL-6 and MIA studies demonstrate changes in synapse formation or microglial function [47,48,63,78,79,80]. These findings do not establish mitochondrial energy limitation as the mediator of impaired pruning. Testing that mechanism requires linking maternal cytokine exposure, fetal microglial bioenergetics, and synapse removal within the same developmental model.
Animal MIA and other developmental models show lasting changes in offspring mitochondrial function [123,124,125]. Human evidence includes associations between maternal allostatic load and later child bioenergetics [128] and cross-sectional ASD biomarkers [32]. These observations support further investigation of the developmental pathway reviewed by Gyllenhammer et al. [127], but do not establish continuous cerebral mitochondrial dysfunction beginning before birth. Repeated measurements spanning prenatal and postnatal development are needed. Temporal continuity must also be distinguished from the mechanism proposed in Section 6.1: continuing injury must outpace mitochondrial repair, clearance, and replacement. The cited studies do not establish that imbalance or the inflammatory feedback proposed to maintain it.
The predictive-inference extension (Section 5.5) and regional-vulnerability account (Section 6.3) also require direct tests. Theoretical work [105,107] and the metabolic findings reported in the cited 2024 preprint [106] motivate the energy–prediction link, but do not demonstrate that energy limitation produces insistence on sameness in ASD. Testing this account requires metabolic measurements during predictable input and routine disruption, related to behavioral severity. The proposed selective bioenergetic vulnerability of Broca’s area likewise requires region-specific metabolic measurements rather than inference from functional activation alone.
Finally, the therapeutic implications remain research proposals rather than established clinical strategies. Pregnancy-exposure studies mainly inform drug safety and broad outcomes, not ASD prevention (Section 7.1). The model does not determine whether anti-inflammatory or mitochondrial-supportive treatment should take priority, or whether prenatal or postnatal intervention offers the better benefit–risk balance (Section 7.4). Trials must establish whether modifying the proposed pathway improves developmental outcomes, which effects remain reversible, and the safety of each intervention in the intended population.

10. Conclusion

I propose that autism spectrum disorder may be understood as an immune-metabolic disorder in which multiple pro-inflammatory cytokines—each acting through distinct molecular mechanisms—converge on a common pathogenic pathway of mitochondrial dysfunction and cerebral energy deficiency. The hypothesis does not assign primacy to any single cytokine; rather, it proposes that cumulative prenatal pro-inflammatory signaling, regardless of its precise composition, may impair fetal brain bioenergetics during sensitive developmental windows. This energy deficit may impair synaptic pruning during development and compromise real-time social cognitive processing, potentially explaining core autism symptoms from a unified mechanistic perspective.
This hypothesis therefore locates the origin of pathology in disrupted prenatal brain construction, while recognizing that the resulting mitochondrial dysfunction persists across the early postnatal period and continues to shape neurodevelopment as energy demands escalate during rapid brain growth. Aberrant immune signaling during pregnancy may compromise the formation of neural circuits before birth, after which the established bioenergetic deficit may continue to influence circuit maturation, regional vulnerability, and the timing of clinical symptom emergence.
If supported by future studies, this framework would imply that autism originates during fetal brain development, with both prenatal initiation and postnatal persistence shaping its developmental trajectory. Under this account, neural circuit abnormalities underlying autistic traits—including excess synaptic density, altered connectivity, and impaired real-time social processing—would reflect both prenatal exposure to an adverse immune-metabolic environment and persistent mitochondrial dysfunction during early postnatal life. This reframing suggests that future research and any eventual risk-modification strategies should encompass both windows: identification of pregnancies at elevated likelihood based on familial autoimmune history and maternal immune profiles, and prospective investigation of postnatal mitochondrial and neurodevelopmental trajectories during the first years of life.

Funding

This research received no external funding.

Institutional Review Board Statement

Not applicable. This is a hypothesis paper that synthesizes published literature and does not involve new studies on humans or animals.

Use of Artificial Intelligence

The core hypothesis and central concepts of this paper originated with me. I used ChatGPT (OpenAI) and Claude (Anthropic) to assist with literature searches, critical discussion of the framework and published evidence, and manuscript drafting, restructuring, and language editing. I reviewed and revised the resulting material, made all final decisions on the manuscript’s content, and take full responsibility for the accuracy of its text, references, and interpretations.

Conflicts of Interest

The author declares no conflicts of interest.

References

  1. Meltzer, A.; Van de Water, J. The role of the immune system in autism spectrum disorder. Neuropsychopharmacology 2017, 42, 284–298. [Google Scholar] [CrossRef] [PubMed]
  2. Wu, S.; Ding, Y.; Wu, F.; et al. Family history of autoimmune diseases is associated with an increased risk of autism in children: A systematic review and meta-analysis. Neurosci. Biobehav. Rev. 2015, 55, 322–332. [Google Scholar] [CrossRef] [PubMed]
  3. Han, V.X.; Patel, S.; Jones, H.F.; Dale, R.C. Maternal immune activation and neuroinflammation in human neurodevelopmental disorders. Nat. Rev. Neurol. 2021, 17, 564–579. [Google Scholar] [CrossRef] [PubMed]
  4. Frasch, M.G.; Yoon, B.-J.; Helbing, D.L.; Snir, G.; Antonelli, M.C.; Bauer, R. Autism Spectrum Disorder: A Neuro-Immunometabolic Hypothesis of the Developmental Origins. Biology 2023, 12, 914. [Google Scholar] [CrossRef] [PubMed]
  5. Gardner, R.M.; Brynge, M.; Sjöqvist, H.; Dalman, C.; Karlsson, H. Maternal Immune Activation and Autism in Offspring: What Is the Evidence for Causation? Biol. Psychiatry 2025, 97, 1127–1138. [Google Scholar] [CrossRef] [PubMed]
  6. Khachadourian, V.; Arildskov, E.S.; Grove, J.; O’Reilly, P.F.; Buxbaum, J.D.; Reichenberg, A.; Sandin, S.; Croen, L.A.; Schendel, D.; Hansen, S.N.; Janecka, M. Familial confounding in the associations between maternal health and autism. Nat. Med. 2025, 31, 996–1007. [Google Scholar] [CrossRef] [PubMed]
  7. Xiang, A.H.; Wang, X.; Martinez, M.P.; et al. Maternal Type 1 Diabetes and Risk of Autism in Offspring. JAMA 2018, 320, 89–91. [Google Scholar] [CrossRef] [PubMed]
  8. Persson, M.; et al. Maternal type 1 diabetes, pre-term birth and risk of autism spectrum disorder. Int. J. Epidemiol. 2023, 52, 377–385. [Google Scholar] [CrossRef] [PubMed]
  9. Atladóttir, H.Ó.; Pedersen, M.G.; Thorsen, P.; Mortensen, P.B.; Deleuran, B.; Eaton, W.W.; Parner, E.T. Association of family history of autoimmune diseases and autism spectrum disorders. Pediatrics 2009, 124, 687–694. [Google Scholar] [CrossRef] [PubMed]
  10. Quattrin, T.; Haller, M.J.; Steck, A.K.; et al. Golimumab and Beta-Cell Function in Youth with New-Onset Type 1 Diabetes. N. Engl. J. Med. 2020, 383, 2007–2017. [Google Scholar] [CrossRef] [PubMed]
  11. Keil, A.; Daniels, J.L.; Forssen, U.; et al. Parental autoimmune diseases associated with autism spectrum disorders in offspring. Epidemiology 2010, 21, 805–808. [Google Scholar] [CrossRef] [PubMed]
  12. Sun, C.K.; Cheng, Y.S.; Chen, I.W.; Chiu, H.J.; Chung, W.; Tzang, R.F.; Fan, H.Y.; Lee, C.W.; Hung, K.C. Impact of parental rheumatoid arthritis on risk of autism spectrum disorders in offspring: A systematic review and meta-analysis. Front. Med. 2022, 9, 1052806. [Google Scholar] [CrossRef] [PubMed]
  13. Zhu, E.H.; Yip, B.H.K.; Fyfe, C.; Merzon, E.; Kodesh, A.; Askling, J.; Reichenberg, A.; Yin, W.; Levine, S.Z.; Sandin, S. Maternal rheumatoid arthritis and the risk of offspring autism spectrum disorder: two national birth cohorts and a meta-analysis. Mol. Autism 2025, 16, 61. [Google Scholar] [CrossRef] [PubMed]
  14. Yin, W.; Norrbäck, M.; Levine, S.Z.; Rivera, N.; Buxbaum, J.D.; Zhu, H.; Yip, B.; et al. Maternal rheumatoid arthritis and risk of autism in the offspring. Psychol. Med. 2023, 53, 7300–7308. [Google Scholar] [CrossRef] [PubMed]
  15. Li, Y.M.; Ou, J.J.; Liu, L.; Zhang, D.; Zhao, J.P.; Tang, S.Y. Association between maternal obesity and autism spectrum disorder in offspring: a meta-analysis. J. Autism Dev. Disord. 2016, 46, 95–102. [Google Scholar] [CrossRef] [PubMed]
  16. Wang, Y.; Tang, S.; Xu, S.; Weng, S.; Liu, Z. Maternal body mass index and risk of autism spectrum disorders in offspring: a meta-analysis. Sci. Rep. 2016, 6, 34248. [Google Scholar] [CrossRef] [PubMed]
  17. Morin, M.; Yin, W.; MacLean, H.; Devlin, B.; Reichenberg, A.; Swan, S.H.; Buxbaum, J.D.; et al. Maternal body mass index in early pregnancy and autism in offspring: a population-based cohort study in Sweden and Denmark. BMC Med. 2025, 23, 620. [Google Scholar] [CrossRef] [PubMed]
  18. Hotamisligil, G.S. Inflammation and metabolic disorders. Nature 2006, 444, 860–867. [Google Scholar] [CrossRef] [PubMed]
  19. Kern, P.A.; Saghizadeh, M.; Ong, J.M.; Bosch, R.J.; Deem, R.; Simsolo, R.B. The expression of tumor necrosis factor in human adipose tissue: regulation by obesity, weight loss, and relationship to lipoprotein lipase. J. Clin. Invest. 1995, 95, 2111–2119. [Google Scholar] [CrossRef] [PubMed]
  20. Pantham, P.; Aye, I.L.; Powell, T.L. Inflammation in maternal obesity and gestational diabetes mellitus. Placenta 2015, 36, 709–715. [Google Scholar] [CrossRef] [PubMed]
  21. Kramer, B.; Kushner, S.A.; Rzhetsky, A. Birth order and disease risk across the human phenome. Nature Health Published online August 3, 2026. 2026. [Google Scholar] [CrossRef]
  22. Bilder, D.; Pinborough-Zimmerman, J.; Miller, J.; McMahon, W. Prenatal, perinatal, and neonatal factors associated with autism spectrum disorders. Pediatrics 2009, 123, 1293–1300. [Google Scholar] [CrossRef] [PubMed]
  23. Robillard, P.Y.; Hulsey, T.C.; Perianin, J.; Janky, E.; Miri, E.H.; Papiernik, E. Association of pregnancy-induced hypertension with duration of sexual cohabitation before conception. Lancet 1994, 344, 973–975. [Google Scholar] [CrossRef] [PubMed]
  24. Tubbergen, P.; Lachmeijer, A.M.A.; Althuisius, S.M.; Vlak, M.E.J.; van Geijn, H.P.; Dekker, G.A. Change in paternity: A risk factor for preeclampsia in multiparous women? J. Reprod. Immunol. 1999, 45, 81–88. [Google Scholar] [CrossRef] [PubMed]
  25. Basso, O.; Christensen, K.; Olsen, J. Higher risk of pre-eclampsia after change of partner: An effect of longer interpregnancy intervals? Epidemiology 2001, 12, 624–629. [Google Scholar] [CrossRef] [PubMed]
  26. Zhang, J. Partner change, birth interval and risk of pre-eclampsia: A paradoxical triangle. Paediatr. Perinat. Epidemiol. 2007, 21 (Suppl. 1), 31–35. [Google Scholar] [CrossRef] [PubMed]
  27. Aluvihare, V.R.; Kallikourdis, M.; Betz, A.G. Regulatory T cells mediate maternal tolerance to the fetus. Nat. Immunol. 2004, 5, 266–271. [Google Scholar] [CrossRef] [PubMed]
  28. Kallikourdis, M.; Andersen, K.G.; Welch, K.A.; Betz, A.G. Alloantigen-enhanced accumulation of CCR5+ ’effector’ regulatory T cells in the gravid uterus. Proc. Natl. Acad. Sci. U. S. A. 2007, 104, 594–599. [Google Scholar] [CrossRef] [PubMed]
  29. Rowe, J.H.; Ertelt, J.M.; Xin, L.; Way, S.S. Pregnancy imprints regulatory memory that sustains anergy to fetal antigen. Nature 2012, 490, 102–106. [Google Scholar] [CrossRef] [PubMed]
  30. Dekker, G.; Robillard, P.Y. Pre-eclampsia: Is the immune maladaptation hypothesis still standing? J. Reprod. Immunol. 2007, 76, 8–16. [Google Scholar] [CrossRef] [PubMed]
  31. Rossignol, D.A.; Frye, R.E. Mitochondrial dysfunction in autism spectrum disorders: a systematic review and meta-analysis. Mol. Psychiatry 2012, 17, 290–314. [Google Scholar] [CrossRef] [PubMed]
  32. Frye, R.E.; Rincon, N.; McCarty, P.J.; Brister, D.; Scheck, A.C.; Rossignol, D.A. Biomarkers of mitochondrial dysfunction in autism spectrum disorder: A systematic review and meta-analysis. Neurobiol. Dis. 2024, 197, 106520. [Google Scholar] [CrossRef] [PubMed]
  33. Sloan, D.B.; Warren, J.M.; Williams, A.M.; et al. Cytonuclear integration and co-evolution. Nat. Rev. Genet. 2018, 19, 635–648. [Google Scholar] [CrossRef] [PubMed]
  34. Sharbrough, J.; Havird, J.C.; Noe, G.R.; Warren, J.M.; Sloan, D.B. The mitonuclear dimension of Neanderthal and Denisovan ancestry in modern human genomes. Genome Biol. Evol. 2017, 9, 1567–1581. [Google Scholar] [CrossRef] [PubMed]
  35. Burton, R.S.; Pereira, R.J.; Barreto, F.S. Cytonuclear genomic interactions and hybrid breakdown. Annu. Rev. Ecol. Evol. Syst. 2013, 44, 281–302. [Google Scholar] [CrossRef]
  36. Zhang, Q.; Raoof, M.; Chen, Y.; et al. Circulating mitochondrial DAMPs cause inflammatory responses to injury. Nature 2010, 464, 104–107. [Google Scholar] [CrossRef] [PubMed]
  37. Zhou, R.; Yazdi, A.S.; Menu, P.; Tschopp, J. A role for mitochondria in NLRP3 inflammasome activation. Nature 2011, 469, 221–225. [Google Scholar] [CrossRef] [PubMed]
  38. Zhang, B.; Angelidou, A.; Alysandratos, K.D.; et al. Mitochondrial DNA and anti-mitochondrial antibodies in serum of autistic children. J. Neuroinflammation 2010, 7, 80. [Google Scholar] [CrossRef] [PubMed]
  39. Moffett, A.; Colucci, F. Co-evolution of NK receptors and HLA ligands in humans is driven by reproduction. Immunol. Rev. 2015, 267, 283–297. [Google Scholar] [CrossRef] [PubMed]
  40. Erlebacher, A. Immunology of the maternal-fetal interface. Annu. Rev. Immunol. 2013, 31, 387–411. [Google Scholar] [CrossRef] [PubMed]
  41. Mor, G.; Aldo, P.; Alvero, A.B. The unique immunological and microbial aspects of pregnancy. Nat. Rev. Immunol. 2017, 17, 469–482. [Google Scholar] [CrossRef] [PubMed]
  42. Monier, A.; Adle-Biassette, H.; Delezoide, A.-L.; Evrard, P.; Gressens, P.; Verney, C. Entry and distribution of microglial cells in human embryonic and fetal cerebral cortex. J. Neuropathol. Exp. Neurol. 2007, 66, 372–382. [Google Scholar] [CrossRef] [PubMed]
  43. Zecevic, N. Synaptogenesis in layer I of the human cerebral cortex in the first half of gestation. Cereb. Cortex 1998, 8, 245–252. [Google Scholar] [CrossRef] [PubMed]
  44. Bian, Z.; Gong, Y.; Huang, T.; Lee, C.Z.W.; Bian, L.; Bai, Z.; Shi, H.; et al. Deciphering human macrophage development at single-cell resolution. Nature 2020, 582, 571–576. [Google Scholar] [CrossRef] [PubMed]
  45. Paolicelli, R.C.; Bolasco, G.; Pagani, F.; et al. Synaptic pruning by microglia is necessary for normal brain development. Science 2011, 333, 1456–1458. [Google Scholar] [CrossRef] [PubMed]
  46. Cunningham, C.L.; Martínez-Cerdeño, V.; Noctor, S.C. Microglia regulate the number of neural precursor cells in the developing cerebral cortex. J. Neurosci. 2013, 33, 4216–4233. [Google Scholar] [CrossRef] [PubMed]
  47. Fernández de Cossío, L.; Lacabanne, C.; Bordeleau, M.; Castino, G.; Kyriakakis, P.; Tremblay, M.-È. Lipopolysaccharide-induced maternal immune activation modulates microglial CX3CR1 protein expression and morphological phenotype in the hippocampus and dentate gyrus, resulting in cognitive inflexibility during late adolescence. Brain Behav. Immun. 2021, 97, 440–454. [Google Scholar] [CrossRef] [PubMed]
  48. Yan, S.; Wang, L.; Samsom, J.N.; Khan, A.; Wong, A.H.C.; Liu, F. PolyI:C Maternal Immune Activation on E9.5 Causes the Deregulation of Microglia and the Complement System in Mice, Leading to Decreased Synaptic Spine Density. Int. J. Mol. Sci. 2024, 25, 5480. [Google Scholar] [CrossRef] [PubMed]
  49. Ajami, B.; Bennett, J.L.; Krieger, C.; Tetzlaff, W.; Rossi, F.M. Local self-renewal can sustain CNS microglia maintenance and function throughout adult life. Nat. Neurosci. 2007, 10, 1538–1543. [Google Scholar] [CrossRef] [PubMed]
  50. Lian, A.; He, M.; Zhang, H.; Yang, Y. Microglia in autism spectrum disorder: heterogeneity, immunometabolism, and synapse-related pathways. Front. Immunol. 2026, 17, 1783755. [Google Scholar] [CrossRef] [PubMed]
  51. Vargas, D.L.; Nascimbene, C.; Krishnan, C.; Zimmerman, A.W.; Pardo, C.A. Neuroglial activation and neuroinflammation in the brain of patients with autism. Ann. Neurol. 2005, 57, 67–81. [Google Scholar] [CrossRef] [PubMed]
  52. Li, X.; Chauhan, A.; Sheikh, A.M.; et al. Elevated immune response in the brain of autistic patients. J. Neuroimmunol. 2009, 207, 111–116. [Google Scholar] [CrossRef] [PubMed]
  53. Ashwood, P.; Krakowiak, P.; Hertz-Picciotto, I.; et al. Elevated plasma cytokines in autism spectrum disorders provide evidence of immune dysfunction and are associated with impaired behavioral outcome. Brain Behav. Immun. 2011, 25, 40–45. [Google Scholar] [CrossRef] [PubMed]
  54. Xie, J.; Huang, L.; Li, X.; et al. Immunological cytokine profiling identifies TNF-α as a key molecule dysregulated in autistic children. Oncotarget 2017, 8, 82390–82398. [Google Scholar] [CrossRef] [PubMed]
  55. Estes, M.L.; McAllister, A.K. Immune mediators in the brain and peripheral tissues in autism spectrum disorder. Nat. Rev. Neurosci. 2015, 16, 469–486. [Google Scholar] [CrossRef] [PubMed]
  56. Che, X.; Hornig, M.; Bresnahan, M.; Stoltenberg, C.; Magnus, P.; Surén, P.; Mjaaland, S.; Reichborn-Kjennerud, T.; Susser, E.; Lipkin, W.I. Maternal mid-gestational and child cord blood immune signatures are strongly associated with offspring risk of ASD. Mol. Psychiatry 2022, 27, 1527–1541. [Google Scholar] [CrossRef] [PubMed]
  57. Kathuria, A.; Lopez-Lengowski, K.; Roffman, J.L.; Karmacharya, R. Distinct effects of interleukin-6 and interferon-γ on differentiating human cortical neurons. Brain Behav. Immun. 2022, 103, 97–108. [Google Scholar] [CrossRef] [PubMed]
  58. Matelski, L.; Morgan, R.K.; Grodzki, A.C.; Van de Water, J.; Lein, P.J. Effects of cytokines on nuclear factor-kappa B, cell viability, and synaptic connectivity in a human neuronal cell line. Mol. Psychiatry 2021, 26, 1622–1635. [Google Scholar] [CrossRef] [PubMed]
  59. Stadler, J.; Bentz, B.G.; Harbrecht, B.G.; Di Silvio, M.; Curran, R.D.; Billiar, T.R.; Hoffman, R.A.; Simmons, R.L. Tumor necrosis factor alpha inhibits hepatocyte mitochondrial respiration. Ann. Surg. 1992, 216, 539–546. [Google Scholar] [CrossRef] [PubMed]
  60. Samavati, L.; Lee, I.; Mathes, I.; Lottspeich, F.; Hüttemann, M. Tumor necrosis factor alpha inhibits oxidative phosphorylation through tyrosine phosphorylation at subunit I of cytochrome c oxidase. J. Biol. Chem. 2008, 283, 21134–21144. [Google Scholar] [CrossRef] [PubMed]
  61. Doll, D.N.; Rellick, S.L.; Barr, T.L.; Ren, X.; Simpkins, J.W. Rapid mitochondrial dysfunction mediates TNF-alpha-induced neurotoxicity. J. Neurochem. 2015, 132, 443–451. [Google Scholar] [CrossRef] [PubMed]
  62. Tarasenko, T.N.; Jestin, M.; Matsumoto, S.; Saito, K.; Hwang, S.; Gavrilova, O.; Trivedi, N.; et al. Macrophage derived TNFα promotes hepatic reprogramming to Warburg-like metabolism. J. Mol. Med. (Berl.) 2019, 97, 1231–1243. [Google Scholar] [CrossRef] [PubMed]
  63. Smith, S.E.; Li, J.; Garbett, K.; Mirnics, K.; Patterson, P.H. Maternal immune activation alters fetal brain development through interleukin-6. J. Neurosci. 2007, 27, 10695–10702. [Google Scholar] [CrossRef] [PubMed]
  64. Wegrzyn, J.; Potla, R.; Chwae, Y.J.; et al. Function of mitochondrial Stat3 in cellular respiration. Science 2009, 323, 793–797. [Google Scholar] [CrossRef] [PubMed]
  65. Xu, J.; Wakai, M.; Xiong, K.; et al. The pro-inflammatory cytokine IL6 suppresses mitochondrial function via the gp130-JAK1/STAT1/3-HIF1α/ERRα axis. Cell Rep. 2025, 44, 115403. [Google Scholar] [CrossRef] [PubMed]
  66. Sarieva, K.; Kagermeier, T.; Khakipoor, S.; et al. Human brain organoid model of maternal immune activation identifies radial glia cells as selectively vulnerable. Mol. Psychiatry 2023, 28, 5077–5089. [Google Scholar] [CrossRef] [PubMed]
  67. Xu, W.; Huang, Y.; Zhou, R. NLRP3 inflammasome in neuroinflammation and central nervous system diseases. Cell. Mol. Immunol. 2025, 22, 341–355. [Google Scholar] [CrossRef] [PubMed]
  68. Vallese, A.; Cordone, V.; Ferrara, F.; et al. NLRP3 inflammasome-mitochondrion loop in autism spectrum disorder. Free Radic. Biol. Med. 2024, 225, 581–594. [Google Scholar] [CrossRef] [PubMed]
  69. Jessop, F.; Buntyn, R.; Schwarz, B.; Wehrly, T.; Scott, D.; Bosio, C.M. Interferon gamma reprograms host mitochondrial metabolism through inhibition of complex II to control intracellular bacterial replication. Infect. Immun. 2020, 88, e00744-19. [Google Scholar] [CrossRef] [PubMed]
  70. Bae, H.R.; Shin, S.K.; Lee, J.Y.; et al. Chronic low-level IFN-γ expression disrupts mitochondrial complex I activity in renal macrophages: An early mechanistic driver of lupus nephritis pathogenesis. Int. J. Mol. Sci. 2025, 26, 63. [Google Scholar] [CrossRef] [PubMed]
  71. Ball, A.B.; Jones, A.E.; Nguyễn, K.B.; et al. Pro-inflammatory macrophage activation does not require inhibition of oxidative phosphorylation. EMBO Rep. 2025, 26, 982–1002. [Google Scholar] [CrossRef] [PubMed]
  72. Kuzawa, C.W.; Chugani, H.T.; Grossman, L.I.; et al. Metabolic costs and evolutionary implications of human brain development. Proc. Natl. Acad. Sci. U. S. A. 2014, 111, 13010–13015. [Google Scholar] [CrossRef] [PubMed]
  73. McElhanon, B.O.; McCracken, C.; Karpen, S.; Sharp, W.G. Gastrointestinal symptoms in autism spectrum disorder: a meta-analysis. Pediatrics 2014, 133, 872–883. [Google Scholar] [CrossRef] [PubMed]
  74. Klusek, J.; Roberts, J.E.; Losh, M. Cardiac autonomic regulation in autism and fragile X syndrome: a review. Psychol. Bull. 2015, 141, 141–175. [Google Scholar] [CrossRef] [PubMed]
  75. Lopez-Espejo, M.A.; Nuñez, A.C.; Moscoso, O.C.; Escobar, R.G. Clinical characteristics of children affected by autism spectrum disorder with and without generalized hypotonia. Eur. J. Pediatr. 2021, 180, 3243–3246. [Google Scholar] [CrossRef] [PubMed]
  76. Petanjek, Z.; Judaš, M.; Šimić, G.; Rašin, M.R.; Uylings, H.B.M.; Rakic, P.; Kostović, I. Extraordinary neoteny of synaptic spines in the human prefrontal cortex. Proc. Natl. Acad. Sci. USA 2011, 108, 13281–13286. [Google Scholar] [CrossRef] [PubMed]
  77. Tang, G.; Gudsnuk, K.; Kuo, S.H.; et al. Loss of mTOR-dependent macroautophagy causes autistic-like synaptic pruning deficits. Neuron 2014, 83, 1131–1143. [Google Scholar] [CrossRef] [PubMed]
  78. Mirabella, F.; Desiato, G.; Mancinelli, S.; Fossati, G.; Rasile, M.; Morini, R.; Markicevic, M.; Grimm, C.; Amegandjin, C.; Termanini, A.; Peano, C.; Kunderfranco, P.; di Cristo, G.; Zerbi, V.; Menna, E.; Lodato, S.; Matteoli, M.; Pozzi, D. Prenatal interleukin 6 elevation increases glutamatergic synapse density and disrupts hippocampal connectivity in offspring. Immunity 2021, 54, 2611–2631.e8. [Google Scholar] [CrossRef] [PubMed]
  79. Fernández de Cossío, L.; Guzmán, A.; van der Veldt, S.; Luheshi, G.N. Prenatal infection leads to ASD-like behavior and altered synaptic pruning in the mouse offspring. Brain Behav. Immun. 2017, 63, 88–98. [Google Scholar] [CrossRef] [PubMed]
  80. Ikezu, S.; Yeh, H.; Delpech, J.-C.; Woodbury, M.E.; Van Enoo, A.A.; Ruan, Z.; Sivakumaran, S.; You, Y.; Holland, C.; Guilarte, T.R.; Maeda, J.; Suhara, T.; Higuchi, M.; Tanaka, M.; Sata, M.; Ikezu, T. Inhibition of colony stimulating factor 1 receptor corrects maternal inflammation-induced microglial and synaptic dysfunction and behavioral abnormalities. Mol. Psychiatry 2021, 26, 1808–1831. [Google Scholar] [CrossRef] [PubMed]
  81. Liu, C.; Li, T.; Ji, J.-J.; Li, Z.; Samsom, J.N.; et al. Maternal immune activation causes sex-specific impairment of microglial function during prefrontal cortex development. Brain Behav. Immun. 2026, 136, 106770. [Google Scholar] [CrossRef] [PubMed]
  82. Monsorno, K.; Ginggen, K.; Ivanov, A.; Buckinx, A.; Lalive, A.L.; Tchenio, A.; et al. Loss of microglial MCT4 leads to defective synaptic pruning and anxiety-like behavior in mice. Nat. Commun. 2023, 14, 5749. [Google Scholar] [CrossRef] [PubMed]
  83. Bernier, L.-P.; York, E.M.; Kamyabi, A.; Choi, H.B.; Weilinger, N.L.; MacVicar, B.A. Microglial metabolic flexibility supports immune surveillance of the brain parenchyma. Nat. Commun. 2020, 11, 1559. [Google Scholar] [CrossRef] [PubMed]
  84. Trevisan, D.A.; Roberts, N.; Lin, C.; Birmingham, E. How do adults and teens with self-declared Autism Spectrum Disorder experience eye contact? PLoS ONE 2017, 12, e0188446. [Google Scholar] [CrossRef] [PubMed]
  85. Dalton, K.M.; Nacewicz, B.M.; Johnstone, T.; et al. Gaze fixation and the neural circuitry of face processing in autism. Nat. Neurosci. 2005, 8, 519–526. [Google Scholar] [CrossRef] [PubMed]
  86. Amiet, C.; Gourfinkel-An, I.; Bouzamondo, A.; et al. Epilepsy in autism is associated with intellectual disability and gender: evidence from a meta-analysis. Biol. Psychiatry 2008, 64, 577–582. [Google Scholar] [CrossRef] [PubMed]
  87. Stellwagen, D.; Beattie, E.C.; Seo, J.Y.; Malenka, R.C. Differential regulation of AMPA receptor and GABA receptor trafficking by tumor necrosis factor-alpha. J. Neurosci. 2005, 25, 3219–3228. [Google Scholar] [CrossRef] [PubMed]
  88. De La Rossa, A.; Laporte, M.H.; Astori, S.; Marissal, T.; Montessuit, S.; Sheshadri, P.; Ramos-Fernández, E.; et al. Paradoxical neuronal hyperexcitability in a mouse model of mitochondrial pyruvate import deficiency. eLife 2022, 11, e72595. [Google Scholar] [CrossRef] [PubMed]
  89. Whittaker, R.G.; Devine, H.E.; Gorman, G.S.; Schaefer, A.M.; Horvath, R.; Ng, Y.; Nesbitt, V.; et al. Epilepsy in adults with mitochondrial disease: A cohort study. Ann. Neurol. 2015, 78, 949–957. [Google Scholar] [CrossRef] [PubMed]
  90. Ritvo, E.R.; Creel, D.; Realmuto, G.; Crandall, A.S.; Freeman, B.J.; Bateman, J.B.; Barr, R.; Pingree, C.; Coleman, M.; Purple, R. Electroretinograms in autism: a pilot study of b-wave amplitudes. Am. J. Psychiatry 1988, 145, 229–232. [Google Scholar] [CrossRef] [PubMed]
  91. Constable, P.A.; Gaigg, S.B.; Bowler, D.M.; Jägle, H.; Thompson, D.A. Full-field electroretinogram in autism spectrum disorder. Doc. Ophthalmol. 2016, 132, 83–99. [Google Scholar] [CrossRef] [PubMed]
  92. Lee, I.O.; Skuse, D.H.; Constable, P.A.; Marmolejo-Ramos, F.; Olsen, L.R.; Thompson, D.A. The electroretinogram b-wave amplitude: a differential physiological measure for Attention Deficit Hyperactivity Disorder and Autism Spectrum Disorder. J. Neurodev. Disord. 2022, 14, 30. [Google Scholar] [CrossRef] [PubMed]
  93. Buttgereit, F.; Brand, M.D. A hierarchy of ATP-consuming processes in mammalian cells. Biochem. J. 1995, 312, 163–167. [Google Scholar] [CrossRef] [PubMed]
  94. Storch, D.; Pörtner, H.O. The protein synthesis machinery operates at the same expense in eurythermal and cold stenothermal pectinids. Physiol. Biochem. Zool. 2003, 76, 28–40. [Google Scholar] [CrossRef] [PubMed]
  95. Bülow, P.; Patgiri, A.; Faundez, V. Mitochondrial protein synthesis and the bioenergetic cost of neurodevelopment. iScience 2022, 25, 104920. [Google Scholar] [CrossRef] [PubMed]
  96. Leblond, C.S.; Nava, C.; Polge, A.; et al. Meta-analysis of SHANK mutations in autism spectrum disorders: a gradient of severity in cognitive impairments. PLoS Genet. 2014, 10, e1004580. [Google Scholar] [CrossRef] [PubMed]
  97. Kim, H.-G.; Kishikawa, S.; Higgins, A.W.; Seong, I.-S.; Donovan, D.J.; Shen, Y.; Lally, E.; et al. Disruption of Neurexin 1 Associated with Autism Spectrum Disorder. Am. J. Hum. Genet. 2008, 82, 199–207. [Google Scholar] [CrossRef] [PubMed]
  98. Jamain, S.; Quach, H.; Betancur, C.; Råstam, M.; Colineaux, C.; Gillberg, I.C.; Soderstrom, H.; et al. Mutations of the X-linked genes encoding neuroligins NLGN3 and NLGN4 are associated with autism. Nat. Genet. 2003, 34, 27–29. [Google Scholar] [CrossRef] [PubMed]
  99. Darnell, J.C.; Van Driesche, S.J.; Zhang, C.; Hung, K.Y.S.; Mele, A.; Fraser, C.E.; Stone, E.F.; et al. FMRP stalls ribosomal translocation on mRNAs linked to synaptic function and autism. Cell 2011, 146, 247–261. [Google Scholar] [CrossRef] [PubMed]
  100. Frye, R.E.; Cox, D.; Slattery, J.; et al. Mitochondrial dysfunction may explain symptom variation in Phelan-McDermid Syndrome. Sci. Rep. 2016, 6, 19544. [Google Scholar] [CrossRef] [PubMed]
  101. Masjedi, N.; Clarke, E.B.; Lord, C. Development of restricted and repetitive behaviors from 2–19: Stability and change in repetitive sensorimotor, insistence on sameness, and verbal behaviors in a longitudinal study of autism. J. Autism Dev. Disord. 2025, 55, 2254–2271. [Google Scholar] [CrossRef] [PubMed]
  102. Laughlin, S.B.; de Ruyter van Steveninck, R.R.; Anderson, J.C. The metabolic cost of neural information. Nat. Neurosci. 1998, 1, 36–41. [Google Scholar] [CrossRef] [PubMed]
  103. Rao, R.P.N.; Ballard, D.H. Predictive coding in the visual cortex: a functional interpretation of some extra-classical receptive-field effects. Nat. Neurosci. 1999, 2, 79–87. [Google Scholar] [CrossRef] [PubMed]
  104. Friston, K. The free-energy principle: a unified brain theory? Nat. Rev. Neurosci. 2010, 11, 127–138. [Google Scholar] [CrossRef] [PubMed]
  105. Sengupta, B.; Stemmler, M.B.; Friston, K.J. Information and efficiency in the nervous system—a synthesis. PLoS Comput. Biol. 2013, 9, e1003157. [Google Scholar] [CrossRef] [PubMed]
  106. Hechler, A.; de Lange, F.P.; Riedl, V. Energy savings across the cortex for confidently predicted visual input. bioRxiv 2024, preprint v2. [Google Scholar] [CrossRef]
  107. Ali, A.; Ahmad, N.; de Groot, E.; van Gerven, M.A.J.; Kietzmann, T.C. Predictive coding is a consequence of energy efficiency in recurrent neural networks. Patterns 2022, 3, 100639. [Google Scholar] [CrossRef] [PubMed]
  108. Pellicano, E.; Burr, D. When the world becomes “too real”: a Bayesian explanation of autistic perception. Trends Cogn. Sci. 2012, 16, 504–510. [Google Scholar] [CrossRef] [PubMed]
  109. Van de Cruys, S.; Evers, K.; Van der Hallen, R.; Van Eylen, L.; Boets, B.; de-Wit, L.; Wagemans, J. Precise minds in uncertain worlds: predictive coding in autism. Psychol. Rev. 2014, 121, 649–675. [Google Scholar] [CrossRef] [PubMed]
  110. Lawson, R.P.; Rees, G.; Friston, K.J. An aberrant precision account of autism. Front. Hum. Neurosci. 2014, 8, 302. [Google Scholar] [CrossRef] [PubMed]
  111. Arthur, T.; Vine, S.; Buckingham, G.; Brosnan, M.; Wilson, M.; Harris, D. Testing predictive coding theories of autism spectrum disorder using models of active inference. PLoS Comput. Biol. 2023, 19, e1011473. [Google Scholar] [CrossRef] [PubMed]
  112. Bastos, A.M.; Usrey, W.M.; Adams, R.A.; Mangun, G.R.; Fries, P.; Friston, K.J. Canonical microcircuits for predictive coding. Neuron 2012, 76, 695–711. [Google Scholar] [CrossRef] [PubMed]
  113. Śliwiński, P. An entropic explanation of insistence on sameness in autism. Front. Comput. Neurosci. 2026, 19, 1714428. [Google Scholar] [CrossRef] [PubMed]
  114. Wigham, S.; Rodgers, J.; South, M.; McConachie, H.; Freeston, M. The interplay between sensory processing abnormalities, intolerance of uncertainty, anxiety and restricted and repetitive behaviours in autism spectrum disorder. J. Autism Dev. Disord. 2015, 45, 943–952. [Google Scholar] [CrossRef] [PubMed]
  115. Drew, P.J. Vascular and neural basis of the BOLD signal. Curr. Opin. Neurobiol. 2019, 58, 61–69. [Google Scholar] [CrossRef] [PubMed]
  116. Christen, T.; Schmiedeskamp, H.; Straka, M.; Bammer, R.; Zaharchuk, G. Measuring brain oxygenation in humans using a multiparametric quantitative blood oxygenation level dependent MRI approach. Magn. Reson. Med. 2012, 68, 905–911. [Google Scholar] [CrossRef] [PubMed]
  117. Villien, M.; Wey, H.Y.; Mandeville, J.B.; Catana, C.; Polimeni, J.R.; Sander, C.Y.; Zürcher, N.R.; Chonde, D.B.; Fowler, J.S.; Rosen, B.R.; Hooker, J.M. Dynamic functional imaging of brain glucose utilization using fPET-FDG. NeuroImage 2014, 100, 192–199. [Google Scholar] [CrossRef] [PubMed]
  118. McWilliams, T.G.; Prescott, A.R.; Allen, G.F.G.; Tamjar, J.; Munson, M.J.; Thomson, C.; Muqit, M.M.K.; Ganley, I.G. mito-QC illuminates mitophagy and mitochondrial architecture in vivo. J. Cell Biol. 2016, 214, 333–345. [Google Scholar] [CrossRef] [PubMed]
  119. Shokolenko, I.; Venediktova, N.; Bochkareva, A.; Wilson, G.L.; Alexeyev, M.F. Oxidative stress induces degradation of mitochondrial DNA. Nucleic Acids Res. 2009, 37, 2539–2548. [Google Scholar] [CrossRef] [PubMed]
  120. Kovacheva, E.; Gevezova, M.; Mehterov, N.; Kazakova, M.; Sarafian, V. The intersection of mitophagy and autism spectrum disorder: A systematic review. Int. J. Mol. Sci. 2025, 26, 2217. [Google Scholar] [CrossRef] [PubMed]
  121. Giulivi, C.; Zhang, Y.F.; Omanska-Klusek, A.; Ross-Inta, C.; Wong, S.; Hertz-Picciotto, I.; Tassone, F.; Pessah, I.N. Mitochondrial dysfunction in autism. JAMA 2010, 304, 2389–2396. [Google Scholar] [CrossRef] [PubMed]
  122. Goh, S.; Dong, Z.; Zhang, Y.; DiMauro, S.; Peterson, B.S. Mitochondrial dysfunction as a neurobiological subtype of autism spectrum disorder: evidence from brain imaging. JAMA Psychiatry 2014, 71, 665–671. [Google Scholar] [PubMed]
  123. Cieślik, M.; Zawadzka, A.; Czapski, G.A.; Wilkaniec, A.; Adamczyk, A. Developmental Stage-Dependent Changes in Mitochondrial Function in the Brain of Offspring Following Prenatal Maternal Immune Activation. Int. J. Mol. Sci. 2023, 24, 7243. [Google Scholar] [CrossRef] [PubMed]
  124. Schneider Gasser, E.M.; Schaer, R.; Mueller, F.S.; Bernhardt, A.C.; Lin, H.-Y.; Arias-Reyes, C.; Weber-Stadlbauer, U. Prenatal immune activation in mice induces long-term alterations in brain mitochondrial function. Transl. Psychiatry 2024, 14, 289. [Google Scholar] [CrossRef] [PubMed]
  125. Stier, A.; Monaghan, P.; Metcalfe, N.B. Experimental demonstration of prenatal programming of mitochondrial aerobic metabolism lasting until adulthood. Proc. R. Soc. B Biol. Sci. 2022, 289, 20212679. [Google Scholar] [CrossRef] [PubMed]
  126. Zawadzka, A.; Cieślik, M.; Adamczyk, A. The Role of Maternal Immune Activation in the Pathogenesis of Autism: A Review of the Evidence, Proposed Mechanisms and Implications for Treatment. Int. J. Mol. Sci. 2021, 22, 11516. [Google Scholar] [CrossRef] [PubMed]
  127. Gyllenhammer, L.E.; Rasmussen, J.M.; Bertele, N.; Halbing, A.; Entringer, S.; Wadhwa, P.D.; Buss, C. Maternal Inflammation During Pregnancy and Offspring Brain Development: The Role of Mitochondria. Biol. Psychiatry Cogn. Neurosci. Neuroimaging 2022, 7, 498–509. [Google Scholar] [CrossRef] [PubMed]
  128. Gyllenhammer, L.E.; Picard, M.; McGill, M.A.; Boyle, K.E.; Vawter, M.P.; Rasmussen, J.M.; Buss, C.; Entringer, S.; Wadhwa, P.D. Prospective association between maternal allostatic load during pregnancy and child mitochondrial content and bioenergetic capacity. Psychoneuroendocrinology 2022, 144, 105868. [Google Scholar] [CrossRef] [PubMed]
  129. Chugani, H.T.; Phelps, M.E.; Mazziotta, J.C. Positron emission tomography study of human brain functional development. Ann. Neurol. 1987, 22, 487–497. [Google Scholar] [CrossRef] [PubMed]
  130. Erecińska, M.; Cherian, S.; Silver, I.A. Energy metabolism in mammalian brain during development. Prog. Neurobiol. 2004, 73, 397–445. [Google Scholar] [CrossRef] [PubMed]
  131. Tanner, A.; Dounavi, K. The emergence of autism symptoms prior to 18 months of age: a systematic literature review. J. Autism Dev. Disord. 2021, 51, 973–993. [Google Scholar] [CrossRef] [PubMed]
  132. Roberts, J.A.; Rice, M.L.; Tager-Flusberg, H. Tense marking in children with autism. Appl. Psycholinguist. 2004, 25, 429–448. [Google Scholar] [CrossRef]
  133. Eigsti, I.M.; Bennetto, L. Grammaticality judgments in autism: Deviance or delay. J. Child Lang. 2009, 36, 999–1021. [Google Scholar] [CrossRef] [PubMed]
  134. Prévost, P.; Tuller, L.; Zebib, R.; Barthez, M.A.; Malvy, J.; Bonnet-Brilhault, F. Pragmatic versus structural difficulties in the production of pronominal clitics in French-speaking children with autism spectrum disorder. Autism Dev. Lang. Impair. 2018, 3, 1–21. [Google Scholar] [CrossRef]
  135. Terzi, A.; Marinis, T.; Kotsopoulou, A.; Francis, K. Grammatical abilities of Greek-speaking children with autism. Lang. Acquis. 2014, 21, 4–44. [Google Scholar] [CrossRef]
  136. Just, M.A.; Cherkassky, V.L.; Keller, T.A.; Minshew, N.J. Cortical activation and synchronization during sentence comprehension in high-functioning autism: evidence of underconnectivity. Brain 2004, 127, 1811–1821. [Google Scholar] [CrossRef] [PubMed]
  137. Harris, G.J.; Chabris, C.F.; Clark, J.; Urban, T.; Aharon, I.; Steele, S.; McGrath, L.; Condouris, K.; Tager-Flusberg, H. Brain activation during semantic processing in autism spectrum disorders via functional magnetic resonance imaging. Brain Cogn. 2006, 61, 54–68. [Google Scholar] [CrossRef] [PubMed]
  138. Knaus, T.A.; Silver, A.M.; Lindgren, K.A.; Hadjikhani, N.; Tager-Flusberg, H. fMRI activation during a language task in adolescents with ASD. J. Int. Neuropsychol. Soc. 2008, 14, 967–979. [Google Scholar] [CrossRef] [PubMed]
  139. Attwell, D.; Laughlin, S.B. An energy budget for signaling in the grey matter of the brain. J. Cereb. Blood Flow Metab. 2001, 21, 1133–1145. [Google Scholar] [CrossRef] [PubMed]
  140. Harris, J.J.; Attwell, D. The energetics of CNS white matter. J. Neurosci. 2012, 32, 356–371. [Google Scholar] [CrossRef] [PubMed]
  141. Catani, M.; Mesulam, M. The arcuate fasciculus and the disconnection theme in language and aphasia: history and current state. Cortex 2008, 44, 953–961. [Google Scholar] [CrossRef] [PubMed]
  142. Friederici, A.D. The brain basis of language processing: from structure to function. Physiol. Rev. 2011, 91, 1357–1392. [Google Scholar] [CrossRef] [PubMed]
  143. Fletcher, P.T.; Whitaker, R.T.; Tao, R.; et al. Microstructural connectivity of the arcuate fasciculus in adolescents with high-functioning autism. NeuroImage 2010, 51, 1117–1125. [Google Scholar] [CrossRef] [PubMed]
  144. Zhang, L.; Li, K.; Zhang, C.; et al. Arcuate fasciculus in autism spectrum disorder toddlers with language regression. Open Med. 2018, 13, 90–95. [Google Scholar] [CrossRef] [PubMed]
  145. Liu, J.; Tsang, T.; Jackson, L.; et al. Altered lateralization of dorsal language tracts in 6-week-old infants at risk for autism. Dev. Sci. 2019, 22, e12768. [Google Scholar] [CrossRef] [PubMed]
  146. McFayden, T.C.; Rutsohn, J.; Cetin, G.; et al. White matter development and language abilities during infancy in autism spectrum disorder. Mol. Psychiatry 2024, 29, 2095–2104. [Google Scholar] [CrossRef] [PubMed]
  147. Kato, Y.; Yokokura, M.; Iwabuchi, T.; et al. Lower availability of mitochondrial complex I in anterior cingulate cortex in autism: a positron emission tomography study. Am. J. Psychiatry 2023, 180, 277–284. [Google Scholar] [CrossRef] [PubMed]
  148. Tang, G.; Gutierrez Rios, P.; Kuo, S.-H.; et al. Mitochondrial abnormalities in temporal lobe of autistic brain. Neurobiol. Dis. 2013, 54, 349–361. [Google Scholar] [CrossRef] [PubMed]
  149. Chauhan, A.; Gu, F.; Essa, M.M.; et al. Brain region-specific deficit in mitochondrial electron transport chain complexes in children with autism. J. Neurochem. 2011, 117, 209–220. [Google Scholar] [CrossRef] [PubMed]
  150. Tan, C.; Frewer, V.; Cox, G.; Williams, K.; Ure, A. Prevalence and age of onset of regression in children with autism spectrum disorder: A systematic review and meta-analytical update. Autism Res. 2021, 14, 582–598. [Google Scholar] [CrossRef] [PubMed]
  151. Liu, X.; Liu, H.; Gu, N.; Pei, J.; Lin, X.; Zhao, W. Preeclampsia promotes autism in offspring via maternal inflammation and fetal NFκB signaling. Life Sci. Alliance 2023, 6, e202301957. [Google Scholar] [CrossRef] [PubMed]
  152. Mahadevan, U.; Long, M.D.; Kane, S.V.; et al. Pregnancy and neonatal outcomes after fetal exposure to biologics and thiopurines among women with inflammatory bowel disease. Gastroenterology 2021, 160, 1131–1139. [Google Scholar] [CrossRef] [PubMed]
  153. Nørgård, B.M.; Nielsen, J.; Friedman, S. In utero exposure to thiopurines/anti-TNF agents and long-term health outcomes during childhood and adolescence in Denmark. Aliment. Pharmacol. Ther. 2020, 52, 829–842. [Google Scholar] [CrossRef] [PubMed]
  154. Galinsky, R.; Dhillon, S.K.; Dean, J.M.; Davidson, J.O.; Lear, C.A.; Wassink, G.; Nott, F.; Kelly, S.B.; Fraser, M.; Yuill, C.; Bennet, L.; Gunn, A.J. Tumor necrosis factor inhibition attenuates white matter gliosis after systemic inflammation in preterm fetal sheep. J. Neuroinflammation 2020, 17, 92. [Google Scholar] [CrossRef] [PubMed]
  155. Kelly, S.B.; Tran, N.T.; Polglase, G.R.; et al. A systematic review of immune-based interventions for perinatal neuroprotection: closing the gap between animal studies and human trials. J. Neuroinflammation 2023, 20, 241. [Google Scholar] [CrossRef] [PubMed]
  156. Andruszewski, D.; Uhlfelder, D.C.; Desiato, G.; et al. Embryo-restricted responses to maternal IL-17A promote neurodevelopmental disorders in mouse offspring. Mol. Psychiatry 2025, 30, 1585–1593. [Google Scholar] [CrossRef] [PubMed]
  157. Kingsbury, D.J.; Bader-Meunier, B.; Patel, G.; Arora, V.; Kalabic, J.; Kupper, H. Safety, effectiveness, and pharmacokinetics of adalimumab in children with polyarticular juvenile idiopathic arthritis aged 2 to 4 years. Clin. Rheumatol. 2014, 33, 1433–1441. [Google Scholar] [CrossRef] [PubMed]
  158. Frye, R.E. Mitochondrial Dysfunction in Autism Spectrum Disorder: Unique Abnormalities and Targeted Treatments. Semin. Pediatr. Neurol. 2020, 35, 100829. [Google Scholar] [CrossRef] [PubMed]
Table 1. Familial and Parental Autoimmune Disease and Offspring Autism Risk.
Table 1. Familial and Parental Autoimmune Disease and Offspring Autism Risk.
Familial or Parental Exposure Effect Estimate 95% CI Key Reference
Family history of Psoriasis OR 1.59 1.28-1.97 Wu et al. (2015) [2]
Family history of Type 1 Diabetes OR 1.49 1.23-1.81 Wu et al. (2015) [2]
Maternal Type 1 Diabetes HR 2.36 1.36-4.12 Xiang et al. (2018) [7]
Family history of Rheumatoid Arthritis OR 1.51 1.19-1.91 Wu et al. (2015) [2]
Family history of Hypothyroidism OR 1.64 1.07-2.50 Wu et al. (2015) [2]
Family history of Any Autoimmune Disease pooled OR 1.28 1.12-1.48 Wu et al. (2015) [2]
Note: OR = odds ratio; HR = hazard ratio; CI = confidence interval. Wu et al. is a systematic review and meta-analysis; its family-history definitions vary across the included studies and its estimates pool multiple primary cohort and case-control studies. The maternal T1D estimate is shown separately because it is specific to disease in the gestational parent. Additional cohort evidence supports an association among maternal type 1 diabetes, preterm birth, and ASD risk [8].
Table 4. Illustrative TNF-α–Mitochondrial Effects Across Experimental Contexts.
Table 4. Illustrative TNF-α–Mitochondrial Effects Across Experimental Contexts.
Experimental Observation Bioenergetic Consequence
Complex I activity reduction in rat hepatocytes [59] Suppressed mitochondrial respiration, with reactive oxygen intermediates implicated in the effect
Cytochrome c oxidase (Complex IV) inhibition in bovine and murine liver/hepatocyte models [60] Reduced oxidative phosphorylation, mitochondrial membrane potential, and cellular ATP
Rapid respiratory suppression in HT-22 cells and primary neurons, with membrane depolarization characterized mechanistically in HT-22 cells [61] Reduced basal/maximal respiration and ATP production, with cytochrome c release
Warburg-like reprogramming of hepatocytes driven by macrophage-derived TNF-α [62] Increased glycolytic programming with reduced reliance on oxidative metabolism in the hepatic model
Note: These examples are context-specific; they illustrate distinct experimentally observed routes by which TNF-α can impair or reprogram mitochondrial metabolism and should not be assumed to occur simultaneously in the same cell type.
Table 6. Energy Metabolic Profiles: Brain vs. Skeletal Muscle.
Table 6. Energy Metabolic Profiles: Brain vs. Skeletal Muscle.
Parameter Brain (Neurons) Skeletal Muscle
Primary ATP source Predominantly OXPHOS OXPHOS + glycolysis (variable)
Glycolytic capacity Very limited High (type II fibers)
Glycogen reserves Minimal Substantial
Tolerance to OXPHOS impairment Very low Moderate to high
Note: OXPHOS = oxidative phosphorylation. Brain energy consumption is even higher in early childhood (~40% of total body energy) [72], coinciding with peak vulnerability to mitochondrial dysfunction.
Table 7. Dendritic Spine Density and Molecular Markers in Human Temporal Cortex.
Table 7. Dendritic Spine Density and Molecular Markers in Human Temporal Cortex.
Measure Controls ASD
Childhood spine density (spines per 10 μm) 11.37 ± 0.68 12.32 ± 0.60
Adolescent spine density (spines per 10 μm) 6.24 ± 0.59 10.33 ± 0.74
Adolescent mTOR pathway markers (p-mTOR, p-S6) Reference group Higher
Autophagy-related markers (LC3-II, p62) Reference group Lower LC3-II; higher p62
Source: Tang et al. [77]. Spine densities are means ± SD from basal dendrites of layer V pyramidal neurons in temporal cortex (BA21), with five donors per diagnosis and age group. These cross-sectional differences are not longitudinal measurements of synapse elimination. The molecular markers do not directly measure autophagic flux.
Disclaimer/Publisher’s Note: The statements, opinions and data contained in all publications are solely those of the individual author(s) and contributor(s) and not of MDPI and/or the editor(s). MDPI and/or the editor(s) disclaim responsibility for any injury to people or property resulting from any ideas, methods, instructions or products referred to in the content.
Copyright: This open access article is published under a Creative Commons CC BY 4.0 license, which permit the free download, distribution, and reuse, provided that the author and preprint are cited in any reuse.