The NAG/AG Appetite Subgroup HypothesisThe NAG/AG immunological divergence — locked and dynamic cascade states as two distinct ASD metabolic-immune phenotypes

1. Background and Observational Basis

Autism spectrum disorder encompasses extraordinary biological heterogeneity beneath a unified diagnostic label. Efforts to identify reproducible subgroups have pursued regression history, gastrointestinal comorbidity, immune activation markers, and genetic copy number variants, yet no stratification criterion has achieved consistent replication across independent cohorts. This failure is increasingly attributed not to the absence of meaningful biological subgroups, but to the use of behavioral and cognitive instruments — designed to capture the phenotypic surface of ASD — as the primary classification tool. Instruments that measure social communication deficit and restricted behavior by definition cannot separate individuals whose identical surface phenotype arises from mechanistically distinct underlying biology.

Appetite behavior represents a largely unexplored stratification variable. It is directly observable, reliably reportable by caregivers without specialist training, and does not require neuropsychological assessment. Its potential biological significance has been systematically underestimated because the field has attributed eating differences in ASD to sensory processing variation, food selectivity, or behavioral routine — without investigating whether a deeper immunological divergence is responsible. The present hypothesis proposes that appetite phenotype in ASD is not a behavioral epiphenomenon but a surface readout of upstream cytokine dominance, and that this cytokine dominance is both measurable and mechanistically explanatory of a broader cluster of metabolic, motor, and social differences between subgroups.

A substantial body of clinical observation in community ASD support group settings — spanning more than fifteen years of longitudinal participant contact — reveals a consistent and previously unexplained bifurcation in eating behavior among individuals with ASD. Roughly half of observed participants display persistent limited appetite and apparent absence of hunger drive; the other half display hyperphagia — constant seeking of food, inability to recognize satiety, and tendency toward overweight. These two phenotypes are not distributed along a continuum but appear to represent discrete poles. Critically, the appetite difference is not the only observed divergence. The hyperphagic (AG) individuals tend to display better motor coordination, more normal gait, better balance, and stronger social engagement, and do not present with the characteristic under-methylation phenotype seen in the food-refusing (NAG) group. This co-clustering of appetite, motor, social, and metabolic features suggests a common upstream mechanism separating the two groups — not a collection of independent variables.

2. The Somatostatin Polarity Framework

The Immune-Derived Autism (IDA) cascade positions somatostatin dysregulation as the critical neuroimmune bridge between peripheral immune activation and the core ASD phenotype. Somatostatin is encoded by a single gene but processed into two principal bioactive forms: SST-28, predominating in the gastrointestinal tract as a potent satiety signal, and SST-14, predominating in cortical interneurons as the pacemaker of gamma oscillation — the neural rhythm underlying social cognition, sensory gating, and executive function.

In the established IDA cascade, gut pH failure drives pepsin/DPP-IV disruption, producing opioid peptide excess and LPS translocation. LPS activates IDO1 and NF-κB. NF-κB sequesters the transcriptional co-activator CBP away from CREB, silencing SST-14 interneurons. In parallel, gut SST-28 becomes hyperactive — signaling persistent satiety and eliminating hunger drive.

Brain and peripheral somatostatin signaling exert opposite effects on circulating ghrelin via SST2 receptor subtype^[1]. Central SST2 activation increases ghrelin; peripheral SST2 activation suppresses ghrelin. The direction of SST dysregulation therefore determines the appetite phenotype. In the NAG, gut SST-28 hyperactivation suppresses peripheral ghrelin — no hunger signal results. In the AG, gut SST-28 is suppressed — ghrelin runs high, producing insatiable hunger. The appetite phenotype is a direct downstream readout of SST-28 polarity in the gut, and that polarity is determined by which immune arm is dominant upstream.

3. The Immune Fork: Convergent NF-κB, Divergent Cascade Architecture

A fundamental question this hypothesis must address is whether the NAG and AG pathways truly converge or run in parallel. Both elevate NF-κB and both silence SST-14 through CBP sequestration from CREB. In that narrow mechanistic sense they converge. But the routes to NF-κB elevation are structurally different in ways that produce profoundly different cascade persistence, metabolic consequences, and treatment responsiveness. Understanding this distinction is the key to the entire framework.

The IFN-γ Dominant Route (NAG) — The Locked Cascade

IFN-γ is the primary inducer of IDO1 transcription via interferon-stimulated response elements (ISRE) in the IDO1 promoter^[2]. IDO1 diverts tryptophan away from serotonin synthesis and into the kynurenine pathway. Kynurenine metabolites activate the aryl hydrocarbon receptor (AhR), which transcriptionally induces further IDO1 expression^[3] — creating a self-sustaining positive feedback loop. AhR also directly activates NF-κB, sustaining CBP sequestration independently of the original immune trigger. Critically, IFN-γ directly drives somatostatin upregulation^[4], and elevated gut SST-28 suppresses peripheral ghrelin via gastric D-cell paracrine action^[1]. The full consequence of this locked architecture:

• IDO1 active → tryptophan depleted → serotonin synthesis collapsed → whole-blood serotonin normal or low

• KYN/TRP ratio elevated (>5%) — measurable by existing plasma assay^[7]

• AhR/IDO1 feedback loop sustains NF-κB even after initial immune trigger is reduced

• SST-28 hyperactive → ghrelin suppressed → no appetite

• Methyl donor pool depleted systemically — global SAM/SAH ratio depression

• NF-κB sustained → SST-14 substantially silenced → severe motor, social, cognitive impairment

• Cascade state: locked — removing the immune trigger does not break the loop; the AhR/IDO1 cycle perpetuates NF-κB independently

The TNF-α Dominant Route (AG) — The Dynamic Cascade

TNF-α activates NF-κB through the canonical IKK pathway — binding TNF-R1, phosphorylating IκBα, and releasing p65/p50 for nuclear translocation^[5] — without requiring IDO1 as an intermediate. Crucially, this route does not engage the AhR/IDO1 feedback loop. There is no self-sustaining metabolic cycle. NF-κB activation tracks directly with the ongoing TNF-α signal — when that signal is reduced, NF-κB resolves. The IFN-γ → SST → ghrelin suppression chain is not engaged. The full consequence:

• IDO1 inactive → tryptophan preserved → serotonin synthesis intact → hyperserotonemia unmasked

• KYN/TRP ratio normal (<5%)

• SST-28 suppressed → ghrelin elevated → persistent hyperphagia

• SAM depletion localized to macrophage compartment (TNF-driven one-carbon metabolism) rather than systemic^[6]

• NF-κB via TNF-R1 → SST-14 compromised but less completely than in NAG, and more dynamically

• Better motor coordination and social engagement — consistent with partial SST-14 preservation

• Cascade state: dynamic — NF-κB tracks the inflammatory signal; removing it allows the cascade to resolve without metabolic loop intervention

Why the Distinction Matters Clinically

The locked vs dynamic distinction is not merely academic — it determines what intervention is required and what recovery looks like. A locked cascade requires breaking the AhR/IDO1 feedback loop as a prerequisite to functional recovery, even after immune clearance. A dynamic cascade resolves more directly when the upstream immune signal is addressed. This predicts faster and more complete IMIG response in the AG, and the need for adjunctive IDO1 pathway intervention in the NAG to achieve equivalent outcomes.

4. The cAMP and Adenylyl Cyclase Axis — The Molecular Hinge

The distinction between locked and dynamic cascade states becomes most precise at the level of adenylyl cyclase and cAMP signaling. This is the molecular hinge on which every downstream difference between NAG and AG turns.

The Two Gαi Inputs to Adenylyl Cyclase Suppression

In the established IDA cascade, two Gαi-coupled inputs converge on adenylyl cyclase suppression. First, casomorphin and gliadorphin — opioid peptides produced by incomplete protein digestion under gut pH failure — activate mu-opioid receptors (MOR). All opioid receptors couple exclusively to inhibitory Gi/Go proteins, and MOR activation suppresses adenylyl cyclase by releasing the Gαi subunit^{[8, 9]}. Second, CD26/DPP-IV blockade by these same opioid peptides impairs adenosine metabolism, driving adenosine accumulation and activation of A1/A2A receptors — also Gαi-coupled — providing a second additive suppressive input on adenylyl cyclase. Both inputs together substantially collapse cAMP production, preventing PKA-mediated CREB phosphorylation and therefore SST-14 transcription.

Why the Opioid Peptide Burden Differs Between NAG and AG

The amplitude of MOR-mediated Gαi input depends on the opioid peptide burden, which depends on the severity of gut pH failure and pepsin/DPP-IV disruption. The IDA framework proposes that gut pH failure is the founding event in the cascade. In the NAG, full gut pH dysregulation produces maximum opioid peptide excess — the MOR input to adenylyl cyclase is fully engaged. In the AG, gut pH failure is proposed to be less complete — producing a lower opioid peptide burden that partially engages MOR but does not fully saturate the Gαi input to adenylyl cyclase. Additionally, TNF-α directly suppresses DPP-IV expression, contributing to opioid peptide accumulation in the AG through a different route — but the magnitude of this suppression is likely less than the substrate overwhelm seen in full NAG-type gut pH failure.

The Consequence: Collapsed vs Partially Preserved cAMP

In the NAG, both Gαi inputs are maximally engaged — adenylyl cyclase is substantially suppressed and cAMP is effectively collapsed. PKA cannot activate CREB. SST-14 transcription is near-zero. The neuropeptide cascade — oxytocin, VIP, secretin — dependent on SST-14 interneuron integrity, fails near-completely. This is the locked state at the molecular level: the cAMP axis cannot drive recovery because it has been removed from functional signaling.

In the AG, the lower opioid peptide burden means MOR engagement is partial. Adenylyl cyclase retains residual activity. cAMP is reduced but not collapsed. PKA retains some activity. CREB can still be phosphorylated at a low level. SST-14 transcription is dampened but present. The neuropeptide cascade is impaired but partially functional. This is the molecular basis for the better motor coordination, gait, and social engagement observed in the AG — these are not separate phenomena but direct expressions of partial cAMP/PKA/CREB/SST-14 preservation.

The Forskolin Prediction

Forskolin bypasses both Gαi inputs by directly activating the adenylyl cyclase catalytic subunit, restoring cAMP production regardless of MOR or A1/A2A receptor state^[10]. In the NAG, where adenylyl cyclase is maximally suppressed, forskolin provides a large gain from a deeply depressed baseline — a substantial and clinically observable cAMP restoration. In the AG, where adenylyl cyclase retains partial activity, forskolin still operates but the delta from baseline is smaller because baseline cAMP is already elevated relative to NAG. This predicts that forskolin produces a larger observable effect in NAG than in AG — a testable prediction that could be used clinically as a preliminary subgroup indicator before the full biomarker panel is available.

The Estrogen Compensatory Axis

The estrogen compensatory axis provides an alternative route to SST-14 transcription via Gq-mER → Gαq → PLC → DAG → PKCδ → AC7 → cAMP → PKA → CREB, bypassing the Gαi block entirely^[11]. This axis explains the 4:1 male-to-female ratio in ASD — females have a partial compensatory mechanism that males lack. Critically, this bypass is more effective in the AG than in the NAG: when the Gαi block is only partial (AG), the estrogen-driven AC7 contribution restores a meaningful proportion of total cAMP. When the Gαi block is near-complete (NAG), the AC7 contribution represents a smaller fraction of total suppressed capacity. This predicts that the male-to-female ratio within the AG will be closer to 2:1 or 3:1 rather than the 4:1 seen in the general ASD population — a testable demographic prediction that could be checked against existing epidemiological data stratified by appetite phenotype.

5. Gut-Brain Appetite Axes — CCK, Vagal Transmission, and GLP-1

The appetite difference between NAG and AG is not solely determined by the SST-28/ghrelin axis. Two further, independently important post-ingestive satiety systems — cholecystokinin (CCK) and glucagon-like peptide-1 (GLP-1) — operate in parallel with ghrelin and are differentially affected in the two subgroups. CCK and the vagal satiety reflex are developed in Sections 5.1–5.5. Section 5.6 extends the same D-cell SST-28 brake, by homologous SSTR5 mechanism, to L-cell GLP-1.

CCK as the Primary Post-Ingestive Satiety Signal

Cholecystokinin is released from enteroendocrine I-cells in the duodenum in response to fat and protein arriving from the stomach. CCK binding to CCK-1 receptors (CCK-1R, formerly CCK-A receptors) on vagal afferent nerve endings activates the vagal satiety pathway^{[12, 13]}. This signal travels via vagal afferents to the nucleus tractus solitarius and from there to the hypothalamus, generating the subjective experience of fullness and meal termination. CCK-1R signaling on vagal afferents is therefore the primary peripheral satiety brake after food intake begins.

The NAG: A Double Satiety Lock

In the NAG, the gut SST-28 hyperactivation that suppresses ghrelin also directly inhibits CCK release from I-cells via D-cell paracrine signaling. This creates a double satiety lock: SST-28 provides a constant satiety signal even in the fasted state (via ghrelin suppression), and simultaneously prevents CCK from being released post-ingestion (eliminating the post-prandial satiety signal). The result is a system with no hunger signal and no need for the vagal satiety reflex — food is physiologically invisible to these individuals. The vagal afferent pathway is not defective; it simply has nothing to relay because CCK is not being released.

The AG: CCK Released but Vagal Signaling Disrupted

In the AG, SST-28 suppression removes the inhibitory brake on CCK release from I-cells. CCK is produced and released in response to meals. However, TNF-α directly modulates vagal afferent sensitivity. The vagus nerve is a key component of the cholinergic anti-inflammatory pathway — vagal efferents suppress TNF-α release from macrophages via alpha-7 nicotinic acetylcholine receptors^{[14, 15]}. The relationship is bidirectional: chronic TNF-α elevation in the AG gut environment progressively desensitizes vagal CCK-1R signaling, impairing the post-ingestive satiety reflex^[16]. CCK is released after meals, but its vagal satiety signal is blunted. Ghrelin is chronically elevated. The AG individual thus has two independent mechanisms simultaneously driving hyperphagia — unrestrained ghrelin and impaired CCK/vagal satiety transmission.

Central CCK-B Receptor Implications

CCK also functions as a neurotransmitter centrally via CCK-B receptors, where it is involved in anxiety modulation, satiety integration, and pain signaling. In the AG, where peripheral CCK-1R vagal transmission is impaired, CCK may accumulate in the periphery and gain greater central access via CCK-B, potentially contributing to the anxiety profiles commonly observed in this subgroup. This is a secondary and speculative mechanism at this stage, but it is mechanistically coherent with the central CCK-B pharmacology literature and warrants investigation.

Secretin and the SST-CCK Interaction

Secretin is also released from duodenal S-cells in response to acid and stimulates somatostatin release from antral D-cells^[17]. In the existing IDA framework, secretin IV delivery circumvents the S-cell pH gate and has been shown to produce transient improvement — consistent with the hypothesis that the S-cell gate is closed in vivo by the alkaline gut pH failure. In the AG, where SST-28 is suppressed and CCK is released, the secretin-somatostatin interaction at the antral level may behave differently than in the NAG. This dimension requires further investigation as part of the observational study proposed in Section 11. Secretin, CCK, and GLP-1 all sit in the same D-cell somatostatin neighborhood developed in Sections 5.2–5.3 and 5.6: secretin stimulates D-cell somatostatin release, which brakes I-cell CCK and, via SSTR5, L-cell GLP-1. The NAG/AG behavior of the secretin–somatostatin interaction should be read together with Section 5.6, not as an isolated finding.

GLP-1 as a Third Gut-Brain Appetite Axis

The SST-28/D-cell brake described in Sections 5.2 and 5.3 for CCK extends, by homologous mechanism, to a second gut hormone: glucagon-like peptide-1 (GLP-1), secreted by L-cells in response to nutrient ingestion. This section develops that extension. Somatostatin, acting preferentially through SSTR5, is a well-documented paracrine and endocrine inhibitor of L-cell GLP-1 secretion, established across rodent culture, colonic tissue, and pharmacologic models spanning more than two decades. Whether this loop is anatomically as tight at human ileal L-cells as the D-cell/I-cell arrangement already invoked for CCK is inferred from animal and culture data rather than demonstrated directly in ASD or human meal-satiety preparations, and is stated at that confidence level throughout.

Somatostatin-28 as the Operative Inhibitor of L-Cell GLP-1

Somatostatin-28 (S-28) inhibits GLP-1 secretion via somatostatin receptor subtype 5 (SSTR5) far more potently than SST-14, with an EC50 roughly 500-fold lower for S-28 across a panel of receptor-selective analogs in fetal rat intestinal cultures^[23]. The same study identified a reciprocal feedback loop: GLP-1 itself stimulates S-28 secretion, which is then autoregulated by SSTR5 activation. This specificity for S-28 over S-14 is directly relevant to the cascade architecture developed in Sections 2–4, since S-28 is the gut-resident isoform this framework already identifies as the driver of NAG/AG bifurcation, while SST-14 is reserved for the distinct downstream CNS interneuron silencing mechanism.

Direct in vivo confirmation comes from a mouse study combining SSTR2/SSTR5 antagonism with perfused intestine preparations, which found that somatostatin exerts strong tonic inhibition of GLP-1 secretion preferentially through SSTR5, that antagonizing SSTR2 and SSTR5 together increased both GLP-1 and somatostatin secretion, and that the reciprocal somatostatin response was itself GLP-1 receptor-dependent^[24]. This is the closest available analog to the D-cell/I-cell paracrine loop already invoked for CCK in Section 5.3 — an intact-tissue, receptor-antagonism demonstration of tonic SST-mediated restraint on L-cell GLP-1 output, in the same species and a comparable intestinal preparation.

A separate line of pharmacologic and colonic tissue evidence corroborates the SSTR5 mechanism from a different angle. In rat proximal colon, selective SSTR5 antagonism lowers the threshold at which GLP-1 and short-chain fatty acids accelerate peristalsis, and GLP-1-positive epithelial cells were found to coexpress SSTR5^[25]. This is colonic motility data, not a duodenal or ileal meal-satiety preparation, and should be read as further functional and anatomical support for an SSTR5 brake on GLP-1-producing cells generally, rather than as direct evidence of the same satiety-signaling circuit CCK uses. In parallel, a review of L-cell secretory regulation notes that paracrine somatostatin restrains L-cell GLP-1 output and identifies somatostatin receptor antagonism as a candidate route to therapeutically increase endogenous GLP-1 secretion^[26] — independent confirmation, from a pharmacology-development perspective, that the SST brake on L-cells is treated as physiologically real and clinically actionable in the broader endocrinology literature.

The molecular mechanism reinforces the cAMP-collapse architecture developed in Section 4. In an L-cell line, SSTR5 activation by the SSTR5-preferring agonist pasireotide suppressed GLP-1 secretion through Gi-dependent inhibition of cAMP, sufficient to functionally block the Gs-coupled secretory pathway in that model^[27]. This demonstrates the mechanism at the receptor-signaling level; it does not by itself establish that endogenous gut SST-28 achieves the same degree of suppression in human L-cells in vivo. The same Gi/cAMP logic Section 4 applies to interneuron SST-14 transcription applies, by homology, to L-cell hormone output — a second tissue in which the proposed NAG/AG cAMP axis would be expected to produce an observable secretory consequence, pending direct measurement in this population.

Applying the NAG/AG Fork

In the NAG, gut SST-28 hyperactivation would be expected to engage the SSTR5 brake on L-cells as it does on I-cells, suppressing GLP-1 secretion alongside CCK. Together with the orexigenic suppression already described for ghrelin, this yields concordant silencing across all three gut-brain appetite axes — the orexigenic signal (ghrelin) and both major post-ingestive satiety hormones (CCK, GLP-1) — consistent with the existing framing that food is physiologically invisible in this subgroup at the level of gut-hormone signaling generally, not only at the CCK-specific mechanism described in Section 5.2.

In the AG, SST-28 suppression would be expected to release the SSTR5 brake on L-cells as it does on I-cells, allowing GLP-1 secretion in response to meals — paralleling the CCK release proposed in Section 5.3. However, GLP-1 receptor is expressed on vagal afferent terminals subject to the same chronic TNF-α-driven desensitization mechanism proposed for CCK-1R in Section 5.3. The local paracrine route by which L-cell GLP-1 signals satiety via adjacent vagal afferents would therefore be expected to be blunted in parallel with CCK, by the same proposed mechanism.

A Pharmacologic, Not Endogenous, Asymmetry: The Area Postrema Route

Long-acting GLP-1 receptor agonists (semaglutide-class) reach the area postrema and adjacent brainstem structures directly, independent of vagal afferent integrity, because they are protease-resistant and circulate at concentrations sufficient to act at circumventricular organs that lack a complete blood-brain barrier. This is well established and is the accepted pharmacologic rationale for the central appetite effects of this drug class.

This bypass should not be extended to endogenous, meal-released GLP-1 without qualification. Native GLP-1 has a circulating half-life of roughly one to two minutes and is understood in the broader literature primarily as a local paracrine signal acting on adjacent vagal afferents, not as a hormone that reliably reaches the area postrema at physiological postprandial concentrations. If TNF-α desensitizes vagal GLP-1R and CCK-1R terminals in the AG as proposed in Section 5.6.2, endogenous GLP-1 does not automatically retain a working bypass route through the area postrema — that bypass is best understood as a property of the pharmacologic drug class at supraphysiological exposure, not a general property of the hormone at meal-released levels. Whether sufficiently elevated endogenous L-cell output could partially engage this route in a subset of AG individuals is a testable prediction, not a mechanism already established, and should not be used to explain within-AG severity variability until postprandial GLP-1 is measured in this population.

GLP-1 Receptor Agonism as a Candidate AG-Specific Intervention

If the pharmacologic area-postrema route in Section 5.6.3 is engaged at the receptor-agonist exposure levels used clinically, GLP-1 receptor agonist pharmacology would be expected to produce a satiety effect in AG individuals through a route that does not depend on vagal afferent integrity — a mechanistically distinct rationale from CCK-directed intervention, worth flagging as a candidate direction for the AG-specific priorities in Section 12.4. GLP-1 receptor agonists have documented anti-inflammatory effects on macrophage populations, including suppression of NF-κB-driven cytokine output; if this generalizes to the AG's TNF-α-dominant gut and vagal compartment, GLP-1 agonism could plausibly act on the upstream inflammatory driver in addition to the appetite symptom, though this transfer from systemic macrophage biology to the AG's specific tissue compartment is a second, unconfirmed hypothesis layered on the first. This proposal requires the same prospective validation standard already applied to the CCK-B central anxiety mechanism (Section 5.4) and the ghrelin neuroprotection hypothesis (Section 7.2). Fasting and postprandial GLP-1, stratified by appetite phenotype, has not been measured in this population and is added to the biomarker panel in Section 9 and the observational study in Section 11.

Predicted Harm in NAG

GLP-1 receptor agonism is predicted to be directionally harmful in the NAG and mechanistically non-contributory to loop-breaking. As established in Section 5.6.2, GLP-1 secretion, CCK release, and ghrelin signaling are all expected to be suppressed in this subgroup by the SST-28/SSTR5 brake — food is already physiologically invisible before any intervention is introduced. Introducing a GLP-1 receptor agonist into this state would not correct a deficit; it would layer a pharmacologically forced satiety signal, plus slowed gastric emptying, onto a system in which appetite suppression is already at or near its ceiling. The predicted clinical consequence is a worsening of the presenting NAG symptom and increased risk of the nutritional and growth complications food refusal already produces in this subgroup. This is the same directional category of subgroup-discordant risk as tryptophan/5-HTP administration in the AG (Section 12.4): an intervention appropriate for one subgroup and predicted to be harmful in the other.

This prediction is not yet supported by the same evidential weight as the tryptophan/5-HTP contraindication, which rests on an already-established AG biomarker (hyperserotonemia). No GLP-1 measurements by appetite phenotype exist in this population, and no direct evidence of GLP-1-receptor-agonist exposure or outcome in NAG-classified individuals exists. Until appetite-stratified GLP-1 data are available, GLP-1 receptor agonism should not be used empirically in biomarker-classified NAG, and any off-label or consumer exposure should be treated as a safety flag to note at intake rather than as a confirmed contraindication carrying the same evidentiary status as Section 12.4.

A second, independent consideration is mechanistic futility rather than harm. The NAG's core lesion is the self-sustaining IDO1/kynurenine/AhR feedback loop and the near-complete Gαi-mediated collapse of adenylyl cyclase (Sections 3.1, 4.3) — mechanisms upstream of and structurally distinct from GLP-1 signaling. A GLP-1 receptor agonist, whatever its effect on appetite, would not be expected to engage IDO1, AhR, or the Gαi block on adenylyl cyclase, and should not be substituted for, or assumed to contribute toward, the NAG-specific priorities already outlined in Section 12.3.

6. The Methylation Intersection

The clinical observation that AG individuals do not present with the characteristic under-methylation phenotype of NAG individuals is explained by mechanistically different routes to methyl donor depletion in each group.

In the NAG, IDO1 activation continuously diverts tryptophan into kynurenine, depleting the substrate pool available for serotonin and melatonin synthesis and drawing methionine cycle intermediates into compensatory reactions. The result is systemic SAM/SAH ratio depression — a global methylation deficit visible across cell compartments and measurable in peripheral blood. Jon Pangborn's framing — that methylation in the brain is the most important biochemical function in the human body — is particularly relevant here: the NAG's systemic methyl donor depletion impairs brain methylation at the most fundamental level.

In the AG, IDO1 is not activated and tryptophan metabolism is preserved. However, TNF-α-driven macrophage activation requires and consumes SAM for histone methylation at inflammatory gene promoters to sustain IL-1β and TNF production^[6]. This creates compartment-specific SAM depletion within the macrophage without producing the systemic deficit visible in peripheral measurements. The AG individual therefore appears relatively normal on standard methylation assessment — not because the methylation system is healthier, but because the depletion mechanism is localized rather than global. Whole-blood SAM/SAH ratio is predicted to cleanly separate the two subgroups and to correlate with KYN/TRP ratio within each group.

7. Differential Hormonal Profiles

The two immune arms produce not just different metabolic states but different hormonal landscapes — each with its own secondary consequences for neurological function.

NAG Hormonal Profile

In the NAG, IDO1 activation depletes tryptophan — the precursor to serotonin, melatonin, and a substrate for the methionine cycle. The hormonal consequence is systemic:

• Serotonin — depleted peripherally and centrally; hyperserotonemia is masked by IDO1 competition for tryptophan

• Melatonin — impaired synthesis; disrupted circadian regulation

• Oxytocin — suppressed via hypothalamic serotonin depletion (IDO1 reduces the serotonin input to oxytocin neurons)

• Ghrelin — low, consistent with SST-28 hyperactivation

• GLP-1 — predicted low (fasting and postprandial), consistent with SST-28/SSTR5 brake on L-cells (Section 5.6)

• VIP and secretin — substantially reduced via SST-14 silencing of the interneuron cascade

AG Hormonal Profile

In the AG, the hormonal picture is fundamentally different — and produces paradoxes that require mechanistic explanation:

• Serotonin — elevated (hyperserotonemia unmasked by absent IDO1 competition). Elevated peripheral serotonin drives GI dysmotility, potentially worsening gut permeability and sustaining LPS translocation that feeds TNF-α production — a secondary feedback loop specific to the AG

• Oxytocin — paradoxically suppressed despite elevated serotonin. Acutely, 5-HT1A receptor activation promotes oxytocin release. However, chronic serotonin elevation desensitizes 5-HT1A autoreceptors^[18] — the very receptors that would drive oxytocin release. The AG individual therefore ends up with oxytocin deficiency through a completely different mechanism than the NAG, without the serotonin depletion that drives it in NAG.

• Ghrelin — chronically elevated. Beyond driving hyperphagia, acyl ghrelin has documented neuroprotective and neurogenic properties: it crosses the blood-brain barrier, binds GHS-R1a in the hippocampus, promotes hippocampal neurogenesis via BDNF-dependent mechanisms, enhances long-term potentiation, and modulates dopamine signaling in the substantia nigra^{[19, 20]}. Chronically elevated ghrelin in the AG may therefore be providing partial neuroprotective compensation for SST-14 interneuron dysfunction — complicating the interpretation of simply normalizing it downward post-IMIG.

• GLP-1 — predicted normal or elevated secretion once the SST-28/SSTR5 brake is released, with the behavioral satiety response predicted to be blunted by the same TNF-α-driven vagal desensitization proposed for CCK-1R (Section 5.6.2). This is a secretion/response dissociation, not a secretory failure.

• VIP and secretin — reduced via SST-14 suppression, but less completely than in NAG due to partial SST-14 preservation

8. Brain SST-14 in NAG versus AG — Same Mechanism, Different Severity

A critical question is whether SST-14 is disrupted identically in both subgroups. The immediate mechanism — NF-κB-mediated CBP sequestration away from CREB, preventing SST-14 gene transcription — is shared. But the degree and reversibility of that disruption differ substantially.

In the NAG, the AhR/IDO1 feedback loop amplifies NF-κB continuously and independently of the original immune trigger. CBP sequestration is sustained and deep. SST-14 transcription is near-zero. The interneurons that gate gamma oscillation are substantially silenced — producing the severe motor dyspraxia, gait disturbance, sensory processing failures, and social cognition deficits characteristic of the more severely affected ASD population.

In the AG, NF-κB is elevated via TNF-R1 but tracks the inflammatory signal rather than running autonomously. CBP sequestration is present but fluctuates with inflammatory state. SST-14 transcription is dampened but not zero. This produces a different interneuron phenotype — gamma oscillation integrity is impaired but partially preserved, yielding better motor coordination, more normal gait, and stronger social engagement. The SST-14 disruption is mechanistically identical in type but different in severity and — critically — in reversibility. Reducing TNF-α signal in the AG allows NF-κB to resolve and SST-14 to recover without requiring the metabolic loop to be broken first.

This distinction also explains why the same NF-κB-targeting interventions produce better outcomes in the AG — not because the mechanism is different, but because the cascade has not locked, and therefore retains the capacity to self-correct when the upstream driver is removed.

9. Proposed Biomarker Panel

The following panel is proposed for stratifying ASD individuals into NAG and AG subgroups. All measurements are achievable with existing clinical laboratory infrastructure; no novel assays are required.

Footnote. The AG postprandial-GLP-1 pattern — normal/robust secretion paired with a predicted blunted behavioral response — mirrors the secretion/vagal-response dissociation already described in the CCK/vagal row. Both reflect the AG core lesion proposed in Section 5.3: the hormone is released, but TNF-α-driven vagal desensitization is proposed to blunt its downstream signal. This is a family of related findings under the AG mechanism, not an isolated anomaly. Postprandial GLP-1 is a table prediction; it is not part of the primary single-draw study in Section 11 (see 11.1 follow-on sub-study).

Primary separators (highest confidence): KYN/TRP ratio and whole-blood serotonin. A KYN/TRP ratio threshold of 5% has been validated in published ASD cohorts with IDO1 activation present in 58.7% of the ASD population studied^[7] — a figure remarkably consistent with the NAG prevalence suggested by this framework. These two measurements alone should correctly classify the majority of individuals.

Secondary confirmation: IFN-γ vs TNF-α dominance direction, fasting acyl ghrelin (not total ghrelin — acyl form is the biologically active appetite signal), SAM/SAH ratio, and fasting GLP-1 (SST-28-consistent polarity relative to ghrelin).

Functional indicators: Adenylyl cyclase activity (indirectly assessed via cAMP challenge response), vagal tone (heart rate variability), motor coordination assessment, and appetite phenotype caregiver report.

10. Testable Predictions

11. Proposed Observational Study

The hypothesis generates a critical experiment that does not yet exist in the published literature: a cross-sectional cohort study co-measuring fasting acyl ghrelin and KYN/TRP ratio in an appetite-stratified ASD cohort. This study requires no intervention, no novel assays, and no specialized infrastructure beyond standard clinical laboratory access.

Study Design

• Design: Cross-sectional observational; single fasting blood draw per participant in the primary cohort

• Sample size: 40–60 ASD-diagnosed individuals; powered for bimodal biomarker distribution detection

• Stratification: Caregiver-reported appetite phenotype using a standardized appetite assessment instrument at intake; NAG and AG classifications made prior to laboratory analysis

• Primary measurements: Fasting acyl ghrelin (plasma); KYN/TRP ratio (plasma); whole-blood serotonin

• Secondary measurements: IFN-γ and TNF-α (plasma); SAM/SAH ratio; IDO1 activity proxy (neopterin or kynurenine); plasma tryptophan; fasting GLP-1 (plasma, same draw)

• Phenotype variables: Motor coordination assessment; gait observation; Social Responsiveness Scale score; sex/gender

• Analysis: Blinded laboratory analysis; primary outcome is co-segregation of KYN/TRP threshold status with appetite group assignment and fasting ghrelin direction

Postprandial GLP-1 — Deferred Follow-On Sub-Study

Postprandial GLP-1 response cannot be captured within the existing single-fasting-draw design and is not included in the primary 40–60-participant cohort. It is proposed as a deferred follow-on sub-study: a standardized mixed-meal tolerance test with GLP-1 sampled at a fixed post-meal interval (e.g., 30 minutes), conducted in a subset of participants already classified as NAG or AG from the primary cohort's fasting biomarkers. This keeps the primary study's single-draw feasibility intact while preserving the postprandial prediction as a testable secondary claim.

Primary Outcome

KYN/TRP ratio above 5% threshold will co-segregate with caregiver-classified NAG status and low fasting acyl ghrelin. KYN/TRP below threshold will co-segregate with AG status and elevated fasting acyl ghrelin. Confirmation of this co-segregation would validate the central mechanistic claim of the hypothesis.

As a secondary measure, fasting GLP-1 is predicted to show SST-28-consistent polarity across the same NAG/AG split already defined by ghrelin and KYN/TRP: low fasting GLP-1 in the NAG (same direction as low ghrelin — both silenced by the SST-28 brake), and normal-to-elevated fasting GLP-1 in the AG (opposite polarity to elevated ghrelin, because ghrelin and GLP-1 sit at opposite ends of the orexigenic/anorexigenic axis, but the same SST-28 brake is released in both cases). A positive finding would strengthen confidence in the primary ghrelin/KYN-TRP outcome without functioning as an independent claim. Postprandial GLP-1 response, tested separately in the follow-on sub-study, would further test the AG-specific secretion/response dissociation predicted in Section 5.6.

Secondary Outcomes

• IFN-γ/TNF-α dominance direction will track with KYN/TRP-defined subgroup assignment

• Whole-blood serotonin will be inversely associated with KYN/TRP ratio across the cohort

• Male-to-female ratio will differ between NAG and AG subgroups in the predicted direction

• Motor coordination and SRS severity will be worse in the KYN/TRP-elevated subgroup

• Fasting GLP-1 will show SST-28-consistent polarity with the ghrelin/KYN-TRP split (low in NAG; normal or elevated in AG)

Why This Study Is Feasible Now

No intervention is required. Existing ASD clinical registries and support group networks provide ready access to potential participants. Caregiver appetite assessment requires no specialist training and can be completed at intake. The laboratory panel is available at standard clinical reference laboratories. Fasting GLP-1 uses the same draw. The study could be initiated within existing academic pediatric neurology or developmental pediatrics infrastructure without dedicated research funding beyond standard laboratory costs. The mixed-meal GLP-1 sub-study is optional and does not condition primary-cohort feasibility.

12. Treatment Implications

The Core Principle: Biomarker Stratification Before Intervention

The NAG/AG framework establishes that standard ASD supplement and intervention protocols applied uniformly across both subgroups are likely misdirected in the AG and potentially harmful. The KYN/TRP ratio and whole-blood serotonin, as a two-marker preliminary screen at clinical intake, would allow intervention to be directed appropriately before any protocol is initiated.

Shared Interventions — Relevant to Both Groups

• Luteolin: NF-κB suppression via upstream mast cell stabilization and PKC-θ/AP-1/JNK inhibition. Relevant regardless of which arm activated NF-κB.

• Forskolin: Adenylyl cyclase bypass of the Gαi block. Relevant to both groups; effect magnitude will be larger in NAG (more depressed baseline) than AG.

• Sulforaphane: Nrf2 activation reduces oxidative stress driving NF-κB in both groups; additional IDO1-modulating properties are a bonus in NAG.

• Low-dose naltrexone: Reduces mu-opioid receptor activation by casomorphin/gliadorphin, partially relieving Gαi suppression of adenylyl cyclase in both groups; independently, at low dose, naltrexone also antagonizes TLR4 on glial cells, an anti-inflammatory action distinct from and additive to its opioid-receptor effect^[22].

• Hydroxocobalamin: Methylation support in both groups (methylcobalamin contraindicated in slow COMT individuals regardless of subgroup).

NAG-Specific Priorities

• IDO1 pathway modulation: EGCG, resveratrol, and 1-methyltryptophan analogs have documented IDO1 inhibitory properties. Breaking the IDO1/KYN/AhR feedback loop is a prerequisite for cascade resolution in NAG — without it, NF-κB continues to be driven even after immune clearance.

• Tryptophan support: Once IDO1 activity is reduced, tryptophan restoration via dietary sources or supplementation may support serotonin and melatonin synthesis recovery. Timing is critical — premature supplementation before IDO1 modulation may simply drive more kynurenine production.

• Aggressive systemic methylation support: SAM precursors, methylfolate, hydroxocobalamin — addressing the global methyl donor depletion that is characteristic of the IDO1-active state.

• GLP-1 receptor agonism — predicted-harm flag, not a protocol item: Do not use empirically in biomarker-classified NAG. Off-label or consumer exposure should be noted at intake. This is not a confirmed contraindication at the evidential weight of Section 12.4 tryptophan/5-HTP in AG (see Section 5.6.5).

AG-Specific Priorities

• TNF-α modulation: Omega-3 fatty acids (EPA/DHA), curcumin, and low-dose naltrexone reduce TNF-α via NF-κB pathway suppression — for naltrexone, via TLR4 antagonism on glial cells rather than its opioid-receptor activity^[22]. These are the primary upstream targets in AG.

• Tryptophan and 5-HTP — CONTRAINDICATED in AG: These are commonly used in ASD supplement protocols. In the AG, where tryptophan metabolism is intact and serotonin is already elevated, tryptophan or 5-HTP supplementation will worsen hyperserotonemia. This risks increased GI dysmotility, potential serotonin syndrome interactions, and behavioral dysregulation. This contraindication should be stated explicitly at clinical intake for any individual identified as AG on biomarker screening.

• Caution regarding ghrelin normalization: Chronically elevated ghrelin in the AG may be providing partial neuroprotective and neurogenic compensation via hippocampal GHS-R1a signaling. Post-IMIG ghrelin normalization should be monitored carefully, and cognitive function tracked alongside metabolic normalization to ensure the neuroprotective contribution of elevated ghrelin is not abruptly lost.

• GLP-1 receptor agonism — candidate only, not a protocol item: Pharmacologic area-postrema access provides a satiety route that does not depend on vagal afferent integrity (Section 5.6.3–5.6.4). This is a mechanistically distinct rationale worth considering after appetite-stratified GLP-1 measurement, not an intervention to initiate from this document.

IMIG Response and Recovery Trajectories

IMIG operates by clearing the chronic immunoglobulin load that sustains NF-κB activation. Both subgroups share the NF-κB/SST-14 mechanism and both should respond to IMIG — but the response trajectories will diverge.

• NAG recovery — sequential steps: IMIG clears immune load → IDO1 activation subsides → AhR loop gradually breaks → NF-κB resolves → CBP releases to CREB → SST-14 transcription recovers → neuropeptide cascade rebuilds. Adjunctive IDO1 pathway intervention accelerates the loop-breaking step. Appetite emerges as SST-28 normalizes downward and ghrelin recovers. Methylation status improves systemically. This is a multi-step recovery process with each step dependent on the prior one completing.

• AG recovery — spring release: IMIG reduces TNF-α load → NF-κB resolves more directly without loop to break → CBP releases to CREB → SST-14 transcription recovers → neuropeptide cascade rebuilds. The partial cAMP/PKA axis already present amplifies the recovery. Satiety regulation emerges — the individual begins to recognize fullness for the first time. This is a more elastic recovery; the system has enough residual signaling capacity to rebound toward normal function relatively quickly.

Appetite normalization direction — emergence of hunger in NAG versus emergence of satiety regulation in AG — is proposed as a secondary metabolic endpoint in any IMIG trial that includes both subgroups. If KYN/TRP-classified NAG individuals develop appetite post-IMIG and KYN/TRP-classified AG individuals develop satiety regulation post-IMIG, the hypothesis is confirmed at the clinical level without requiring additional mechanistic measurement.

Supplement Table — Shared vs Subgroup-Specific

13. Limitations and Open Questions

This is a hypothesis-generating document. The following elements require prospective validation:

• SST-14 differential silencing: The claim that AG individuals have meaningfully better SST-14 function is inferred from motor, gait, and social behavioral observations. Direct measurement — gamma oscillation coherence via EEG or CSF SST-14 levels in appetite-stratified cohorts — has not been performed.

• IFN-γ / TNF-α mutual exclusivity: Published ASD literature reports both cytokines elevated across the ASD population without stratifying them as inversely dominant within subgroups. The hypothesis requires that IFN-γ and TNF-α are mutually dominant across individuals — a pattern not yet explicitly tested.

• Gut pH failure amplitude as founding variable: The proposal that NAG individuals have more severe gut pH failure than AG individuals is a foundational assumption that requires direct measurement of gastric pH, pepsin activity, and opioid peptide levels across appetite-stratified cohorts.

• CCK-B central anxiety mechanism: The proposal that peripheral CCK accumulation produces CCK-B mediated central anxiety effects in the AG is mechanistically coherent but speculative. Direct CCK plasma measurement in appetite-stratified ASD cohorts would be required to test it.

• Ghrelin neuroprotection in AG: The proposal that chronically elevated ghrelin provides partial hippocampal neuroprotection in the AG is based on published ghrelin neurobiology but has not been tested in the ASD context. Post-IMIG cognitive monitoring is required to assess the clinical significance of ghrelin normalization.

• GLP-1 polarity by appetite group: No fasting or postprandial GLP-1 measurements stratified by NAG/AG (or by KYN/TRP) exist in this population. Section 5.6 is inferred from rodent culture, perfused-intestine, colonic-motility, and L-cell-line data plus the existing SST-28 architecture; it is not yet constrained by ASD GLP-1 data.

• GLP-1 receptor agonist risk in neurodevelopment: Pediatric and neurodevelopmental risks of GLP-1 receptor agonist exposure (lean mass, linear growth, GI tolerability) fall outside the scope of this paper and cannot be inferred from the adult obesity literature in which this drug class was developed. The AG-candidate and NAG-harm flags in Sections 5.6.4–5.6.5 are mechanistic predictions, not prescribing guidance.

• Observational basis: The appetite phenotype observations underlying this hypothesis derive from longitudinal clinical contact in a community ASD support setting, not a structured epidemiological study. Formal phenotyping with standardized appetite assessment instruments in a clinical cohort is required.

• Acyl vs total ghrelin: Published ASD ghrelin studies have generally measured total ghrelin without separating acyl (appetite-active) and des-acyl forms. Future studies must specify acyl ghrelin measurement.

14. Conclusion

We propose that appetite phenotype in ASD is a surface readout of upstream cytokine dominance, and that this dominance determines the architecture of the entire downstream cascade. Two immunologically distinct subgroups — NAG (IFN-γ dominant, IDO1-active, locked cascade) and AG (TNF-α dominant, IDO1-inactive, dynamic cascade) — both produce ASD phenotype through convergent NF-κB-mediated SST-14 silencing, but diverge at every upstream and parallel level: the degree of adenylyl cyclase suppression, the character of somatostatin dysregulation, the direction of ghrelin modulation, the completeness of neuropeptide cascade failure, the presence or absence of a self-sustaining metabolic feedback loop, the methylation deficit character, and the hormonal landscape including the paradoxical oxytocin deficiency mechanism in the AG and the homologous CCK/GLP-1 satiety-axis fork developed in Section 5.

The locked vs dynamic cascade distinction is not merely descriptive — it predicts differential treatment requirements, different IMIG response trajectories, different adjunctive intervention priorities, and at least one important supplement contraindication (tryptophan/5-HTP in the AG) that uniform treatment protocols currently ignore. A second, lower-weight directional warning applies to GLP-1 receptor agonism in NAG; an AG-specific candidate rationale exists via pharmacologic area-postrema access, contingent on prospective GLP-1 measurement.

The proposed biomarker panel — anchored by KYN/TRP ratio and whole-blood serotonin — is measurable now, using existing clinical infrastructure, in any ASD cohort. The critical experiment is a single cross-sectional study co-measuring fasting acyl ghrelin and KYN/TRP ratio in an appetite-stratified ASD cohort, with fasting GLP-1 on the same draw as a secondary polarity check. That study does not exist. Its execution represents the first direct test of the SST-28/ghrelin/cytokine subgroup hypothesis and could be initiated within existing academic clinical infrastructure without dedicated research funding beyond standard laboratory costs.

If validated, the NAG/AG framework transforms appetite phenotype from an unexplained behavioral variable into a diagnostic stratification criterion with direct implications for immune-targeted treatment design, trial endpoint selection, adjunctive intervention prioritization, supplement protocol safety, and personalized intervention strategy across the ASD population.

Related reading

• The Anatomy of Autism — the foundational cascade model this hypothesis builds from.
• Restoring the Somatostatin Signal in Immune-Derived Autism — the primary SST-14 restoration framework.
• L1-79 and the Three-Intervention Model — where forskolin's differential NAG/AG response prediction connects to the broader intervention framework.
• The IMIG Clinical Trial — where NAG/AG stratification would inform trial design and endpoint selection.
• Autism and Alzheimer's — Two Sides of a Plasticity Problem — the shared plasticity biology underlying why SST-14 recovery differs by cascade state, relevant to both NAG and AG.

References

1. [1]Schalla MA, Stengel A. Activation of somatostatin 2 receptors in the brain and the periphery induces opposite changes in circulating ghrelin levels: functional implications. Front Endocrinol. 2013;3:178.
2. [2]Robinson CM, Hale PT, Carlin JM. The role of NF-κB and STAT1 in the induction of IDO1 by interferon-γ. J Interferon Cytokine Res. 2005;25(8):508–516.
3. [3]Platten M, Nollen EAA, Röhrig UF, Fallarino F, Opitz CA. Tryptophan metabolism as a common therapeutic target in cancer, neurodegeneration, and beyond. Nat Rev Drug Discov. 2019;18(5):379–401.
4. [4]Stutte S, Ruf J, Kugler I, et al. Type I interferon-mediated induction of somatostatin leads to suppression of ghrelin and appetite thereby promoting viral immunity in mice. Brain Behav Immun. 2021;95:429–443.
5. [5]Liu T, Zhang L, Joo D, Sun SC. NF-κB signaling in inflammation. Signal Transduct Target Ther. 2017;2:17023.
6. [6]Liu PS, Wang H, Li X, et al. α-ketoglutarate orchestrates macrophage activation through metabolic and epigenetic reprogramming. Nat Immunol. 2017;18(9):985–994.
7. [7]Delorme R, Ey E, Toro R, et al. Impact of IDO activation and alterations in the kynurenine pathway on hyperserotonemia, NAD+ production, and AhR activation in autism spectrum disorder. Transl Psychiatry. 2023;13(1):393.
8. [8]Frontiers in Neuroscience. Molecular and cellular basis of mu-opioid receptor signaling: mechanisms underlying tolerance and dependence development. Front Neurosci. 2025;19:1597922.
9. [9]Casomorphin and gliadorphin as ligands for opioid receptors: inhibitory G-protein coupling and adenylyl cyclase suppression. In: Diet and Autism. 2013. [Review]
10. [10]Seamon KB, Padgett W, Daly JW. Forskolin: unique diterpene activator of adenylate cyclase in membranes and in intact cells. Proc Natl Acad Sci USA. 1981;78(6):3363–3367.
11. [11]Qiu J, Bosch MA, Tobias SC, et al. Rapid signaling of estrogen in hypothalamic neurons involves a novel G-protein-coupled estrogen receptor that activates protein kinase C. J Neurosci. 2003;23(29):9529–9540.
12. [12]Rehfeld JF. Cholecystokinin — from local gut hormone to ubiquitous messenger. Front Endocrinol. 2017;8:47.
13. [13]Cummings DE, Overduin J. Gastrointestinal regulation of food intake. J Clin Invest. 2007;117(1):13–23.
14. [14]Bonaz B, Bazin T, Pellissier S. The vagus nerve at the interface of the microbiota-gut-brain axis. Front Neurosci. 2018;12:49.
15. [15]Breit S, Kupferberg A, Rogler G, Hasler G. Vagus nerve as modulator of the brain-gut axis in psychiatric and inflammatory disorders. Front Psychiatry. 2018;9:44.
16. [16]de Lartigue G, Barbier de la Serre C, Espero E, Lee J, Raybould HE. Leptin resistance in vagal afferent neurons inhibits cholecystokinin signaling and satiation in diet-induced obese rats. PLoS One. 2012;7(3):e32967.
17. [17]McIntosh CH, Pederson RA, Koop H, Brown JC. Effect of cholecystokinin and secretin on somatostatin release from cultured antral cells. Gastroenterology. 1993;104(5):1320–1327.
18. [18]Piñeyro G, Blier P. Autoregulation of serotonin neurons: role in antidepressant drug action. Pharmacol Rev. 1999;51(3):533–591.
19. [19]Nass R, Farhy LS, Liu J, et al. The neurocognitive effects of ghrelin-induced signaling on the hippocampus: a promising approach to Alzheimer's disease. PMC. 2018;PMC6235652.
20. [20]Gibb FW, Littlejohn CM, Walker BR, et al. Ghrelin acylation — a post-translational tuning mechanism regulating adult hippocampal neurogenesis. PMC. 2022;PMC8909677.
21. [21]Fourie PR, Armstrong C. Intravenous immunoglobulin in autism spectrum disorder: a case series. Med Res Arch. 2024;12(10).
22. [22]Younger J, Parkitny L, McLain D. The use of low-dose naltrexone (LDN) as a novel anti-inflammatory treatment for chronic pain. Clin Rheumatol. 2014;33(4):451–459. DOI: 10.1007/s10067-014-2517-2. PMID: 24526250.
23. [23]Chisholm C, Greenberg GR. Somatostatin-28 regulates GLP-1 secretion via somatostatin receptor subtype 5 in rat intestinal cultures. Am J Physiol Endocrinol Metab. 2002;283(2):E311–E317. DOI: 10.1152/ajpendo.00434.2001.
24. [24]Jepsen SL, Grunddal KV, Wewer Albrechtsen NJ, et al. Paracrine crosstalk between intestinal L- and D-cells controls secretion of glucagon-like peptide-1 in mice. Am J Physiol Endocrinol Metab. 2019;317(6):E1081–E1093. DOI: 10.1152/ajpendo.00239.2019.
25. [25]Nakamori H, Hosoi F, Hashitani H. Inhibitory effects of somatostatin on glucagon-like peptide-1-mediated acceleration of peristalsis in the rat proximal colon. Am J Physiol Gastrointest Liver Physiol. 2026;330(3):G256–G269. DOI: 10.1152/ajpgi.00189.2025.
26. [26]Holst JJ, Jepsen SL, Modvig I. GLP-1 - Incretin and pleiotropic hormone with pharmacotherapy potential. Increasing secretion of endogenous GLP-1 for diabetes and obesity therapy. Curr Opin Pharmacol. 2022;63:102189. DOI: 10.1016/j.coph.2022.102189. PMID: 35231672.
27. [27]Sato J, Manaka K, Horikoshi H, et al. Insights into GLP-1 and insulin secretion mechanisms in pasireotide-induced hyperglycemia highlight effectiveness of Gs-targeting diabetes treatment. Sci Rep. 2025;15(1):9494. DOI: 10.1038/s41598-025-90896-2.