A Danish PET Study Flips the Autism Dopamine Story

Autistic adults in a 60-person Danish PET study showed extra dopamine D2 binding in the thalamus, reversing the direction of two earlier receptor scans. The same deep nuclei also used more glucose, and that energy use tracked social and communication scores on a trait questionnaire.

Laust Vind Knudsen, a postdoc at the Research Unit for Psychiatry at Odense University Hospital and the University of Southern Denmark, led the work as part of his PhD, with Professor Tanja Maria Michel as research leader. The team is careful: the scans do not explain why autism develops and cannot diagnose it. They do change what “a dopamine difference” can mean, because extra D2 binding here is modest, sex-split, and locked to energy use and network talk.

Autistic Adults Carried More D2 Binding in Deep Nuclei

Between November 2022 and October 2024, 30 autistic adults and 30 neurotypical adults, matched on age, sex, body mass index, and IQ, completed a dual-tracer PET and MRI study on a 3T SIGNA PET/MR scanner. Each autistic group and each control group included 15 women. Mean age was 24.59 years in the autistic group and 25.99 years in controls. Mean IQ was 105.6 versus 107.4. Autism-Spectrum Quotient totals were 32.57 versus 12.87.

Everyone first received a 60-minute [¹¹C]raclopride scan, a tracer that binds D2 receptors. After about 102 minutes, they received [¹⁸F]FDG, which tracks glucose use. Resting-state fMRI ran during the raclopride session. Five autistic participants were on stable medicines that do not target dopamine. Stimulant use was an exclusion, so this was not a medicated ADHD sample.

THREE SCANS IN ONE SESSION

Knudsen said the three measures had not been combined before in autistic people, which is the design claim the paper rests on. D2 availability is not a raw dopamine count. Raclopride binding potential reflects receptor density, how much native dopamine is already occupying those sites, and some D3 signal, a limit the authors put in the paper.

The Pooled D2 Cluster Measured 72 Cubic Millimeters

Across all 60 scans, extra D2 binding in autism landed in a left-thalamic cluster of 9 voxels, or 72 mm³ (peak t = 3.36, p-FWE = 0.044). That is the group result the headlines folded into “more D2 receptors.” Region-of-interest tests trended the same way and did not clear the bar.

Split the sample by sex and the map grows. Autistic women showed higher bilateral thalamic binding than neurotypical women (103 voxels, 824 mm³, peak t = 3.59, p-FWE = 0.032). Autistic men showed higher binding than neurotypical men in bilateral thalamus, left nucleus accumbens, and right putamen (434 voxels, 3472 mm³, peak t = 4.89, p-FWE = 0.004). Inside the autistic group, women still ran higher than men in thalamus (58 voxels, 464 mm³). Control women also ran higher than control men (51 voxels, 408 mm³).

CLUSTER SIZES ON THE D2 MAP

• Pooled autism vs control: 72 mm³ in left thalamus, the only whole-group D2 hit.
• Autistic men vs control men: 3472 mm³ spanning thalamus, accumbens, and putamen.
• Autistic women vs control women: 824 mm³ in bilateral thalamus.
• Autistic women vs autistic men: 464 mm³ in thalamus, a sex effect inside the diagnosis.

The authors call the pooled D2 increase modest and limited in extent. They argue sex-specific thalamic subregions diluted the mixed-sex contrast, which is why the male-only and female-only maps look larger than the headline cluster. AQ scores did not correlate with D2 binding at all.

Lower D2 Binding in Two Earlier PET Papers

Human D2 PET in autism was already pointing down. A 2022 [¹¹C]FLB457 study in 22 unmedicated autistic men and 24 typical men found lower extrastriatal D2/3 availability, with the largest effect in the posterior thalamus. A Turku raclopride study in 16 autistic men and 24 controls, published in 2025 in the same journal family, found lower striatal D2 receptor availability, including about 13% less in nucleus accumbens and 19% less in globus pallidus.

THREE D2 PET SAMPLES

Knudsen’s group lists the clash in the discussion and offers the usual mix of reasons: tracer affinity (FLB457 is built for cortex and thalamus; raclopride is the striatal workhorse that still binds thalamus), scanner setup (one PET/MR with the same pipeline for everyone), sample size, and sex mix. Prior D2 PET in autism was almost all men. This protocol forced a 1:1 sex split even though autism is diagnosed more often in males.

The new map does line up with other tissue. The paper cites higher striatal D2 mRNA in postmortem autism samples and higher accumbens D2 density across several male mouse models. In vivo human PET had not been telling that story until this cohort.

Social and Communication Scores Rose With Thalamic Energy Use

Glucose was the broader signal. Autistic adults showed higher FDG uptake in bilateral thalamus and right globus pallidus, a cluster of 825 voxels, or 6600 mm³ (peak t = 4.06, p-FWE = 0.004). That volume is more than 90 times the pooled D2 cluster. The extra burn held in women (825 voxels, 6600 mm³) and in men (4533 voxels, 36264 mm³). Autistic men and women did not differ from each other on glucose.

D2 binding and glucose use rose together across caudate, globus pallidus, accumbens, thalamus, and putamen, a 3948-voxel coupling (31584 mm³) that remained when the autistic group and the control group were tested on their own. The authors treat that lockstep as a shared subcortical feature, not an autism-only artifact.

AQ SCORES AND GLUCOSE, NOT D2

• Communication: Left thalamus r = 0.474, right thalamus r = 0.394, right globus pallidus r = 0.390, all FDR-significant.
• Social skills: Left thalamus r = 0.413, right thalamus r = 0.384.
• Total AQ: Left thalamus r = 0.366, right thalamus r = 0.314.
• D2 binding: No FDR-significant link to any AQ scale.

On fMRI, left-thalamic D2 binding related to left-caudate connectivity differently in autism than in controls (134-voxel caudate cluster, p-FDR = 0.0004). From that caudate seed, higher thalamic D2 went with weaker caudate-cortical coupling, the largest follow-up cluster in right cuneus (371 voxels). That D2-to-network pattern also differed between autistic men and women, with a 1492-voxel follow-up in left superior frontal gyrus. Control men and women did not show a sex split on that coupling (peak p-FDR = 0.183).

Why Autism and ADHD Keep Showing Up Together

Knudsen has been explicit about the next question. Dopamine is already central to ADHD biology, and ADHD is the comorbidity everyone in clinic already sees. A 2021 meta-analysis of 63 papers put the lifetime ADHD prevalence of 40.2 percent among autistic people (95% CI 34.9 to 45.7), with current prevalence at 38.5 percent. The university briefing on the scans also put one co-occurring diagnosis at about 70 percent of autistic people, and two or more at about 40 percent.

Brain research can help us understand autism better, not in order to change autistic people, but to create greater understanding of neurodiversity and better conditions in society. Our findings also raise new questions about why autism and ADHD so often occur together, which we would like to investigate further.

Laust Vind Knudsen, postdoc, Research Unit for Psychiatry, University of Southern Denmark briefing

This cohort cannot answer that question. Stimulant users were kept out. Psychiatric comorbidity in the sample was low by the authors’ own limit list. What the scans do show is a subcortical D2-and-glucose pattern in cognitively able autistic adults who were not on dopamine drugs, with network coupling that already differs by diagnosis and by sex. That is a plausible shared machine for attention and reward, not proof that autism and ADHD are one condition.

The Two D2 Drugs Clinics Already Use

Clinics did not wait for this PET map. The only U.S. drugs approved for autism target D2. Risperidone, a D2 antagonist, was approved in 2006 for children 5 to 16. Aripiprazole, a D2 partial agonist, was approved in 2009 for ages 6 to 17. Both are labeled for irritability associated with autistic disorder, including aggression, self-injury, tantrums, and quickly changing moods, not for core social-communication features. The paper notes those medicines are used for agitation and irritability, which the authors say affect about 90% of autistic children.

WHAT THIS COHORT CAN AND CANNOT SHOW

• Who was scanned: Adults 20 and older, full-scale IQ above 80, equal numbers of men and women, mostly White (28 of 30 autistic participants).
• Who was not: Children, people with intellectual disability, stimulant users, and a sample that matches male-biased clinic prevalence.
• What binding means: Raclopride BPnd is a proxy for D2 density at rest, still pushed around by native dopamine and receptor affinity.
• What the authors want next: Larger groups, including people the dual-tracer protocol is harder to run on.

Funding came from the Psychiatric Research Fund of the Region of Southern Denmark, grant R31-A1616. The paper went online on 6 July 2026. The university posted its briefing on 19 August. Replication is still the test, because a 72 mm³ pooled D2 cluster and a 6600 mm³ glucose cluster are not the same kind of finding, and the last two raclopride papers in autistic men did not agree on the sign of the D2 change.

Frequently Asked Questions

Can a dopamine PET scan diagnose autism?

No. Diagnosis still rests on developmental history and DSM-5 behavior criteria, and the Odense team says these scans cannot be used to diagnose. This protocol also required IQ above 80, no nicotine, no stimulants, and enough compliance to complete two PET sessions, so it sampled cognitively able adults rather than the full clinical range.

What does raclopride PET actually measure?

Non-displaceable binding potential (BPnd) for D2, estimated with a simplified reference tissue model using cerebellar grey matter. Higher BPnd can mean more receptors, less competition from dopamine already in the synapse, some D3 contribution, or a mix, which is why the paper treats BPnd as a proxy for density rather than a direct dopamine assay.

How often do autism and ADHD occur together?

In the 2021 Rong meta-analysis, current ADHD among autistic people was 38.5 percent (95% CI 34.0 to 43.2) across 56 studies, while lifetime prevalence was 40.2 percent across 13 studies. Age, intellectual disability, recruitment setting, and diagnostic rules all shifted the current rate, which is why clinic quotes bounce around that 40 percent band.

Are risperidone and aripiprazole approved for core autism traits?

No U.S. drug is approved for core social-communication features. Risperidone’s label covers irritability in autistic disorder for ages 5 to 16, and aripiprazole covers the same indication for ages 6 to 17. Efficacy in the risperidone trials was scored on the Aberrant Behavior Checklist irritability subscale, not on diagnostic social-communication instruments.

Why did this PET study enroll equal numbers of women?

No prior D2 PET study in autism had tested sex differences, even though D2 availability and autism presentation both vary by sex and autistic women are often underdiagnosed. Recruitment was forced to 15 women and 15 men in each group, which the authors say does not match male-biased prevalence and was still required if women were going to appear in the maps at all.

Disclaimer: This article is news reporting on a published imaging study and is for information only. It is not medical advice, a diagnostic tool, or a recommendation to start, stop, or change any medicine, including risperidone, aripiprazole, or ADHD stimulants. Readers who have questions about autism, ADHD, or brain-scan research should talk with a qualified physician or psychiatrist who knows their history before making health decisions. Figures, eligibility rules, and drug labels reflect the papers and agency documents cited as of 2 September 2026 and may change with new studies or label updates.