Fractionating autism based on neuroanatomical normative modeling

Abstract

Autism is a complex neurodevelopmental condition with substantial phenotypic, biological, and etiologic heterogeneity. It remains a challenge to identify biomarkers to stratify autism into replicable cognitive or biological subtypes. Here, we aim to introduce a novel methodological framework for parsing neuroanatomical subtypes within a large cohort of individuals with autism. We used cortical thickness (CT) in a large and well-characterized sample of 316 participants with autism (88 female, age mean: 17.2 ± 5.7) and 206 with neurotypical development (79 female, age mean: 17.5 ± 6.1) aged 6–31 years across six sites from the EU-AIMS multi-center Longitudinal European Autism Project. Five biologically based putative subtypes were derived using normative modeling of CT and spectral clustering. Three of these clusters showed relatively widespread decreased CT and two showed relatively increased CT. These subtypes showed morphometric differences from one another, providing a potential explanation for inconsistent case–control findings in autism, and loaded differentially and more strongly onto symptoms and polygenic risk, indicating a dilution of clinical effects across heterogeneous cohorts. Our results provide an important step towards parsing the heterogeneous neurobiology of autism.

Federated Multi-Site Normative Modeling using Hierarchical Bayesian Regression

“…12 and 13) with a regional distribution of abnormalities that is largely consistent with the known pathology of each disorder. This is in line with earlier publications (Wolfers et al, 2018;Zabihi et al, 2019;Wolfers et al, 2020;Zabihi et al, 2020;Wolfers et al, 2021), which show that while we observe differences between groups of patients and controls, those differences are not perfect to the extent of complete group separation (Marquand et al, 2019). This has been linked to the heterogeneous nature of these illnesses, which generally show a unique pattern of sub-and supra-normal deviations in individuals even when diagnosed with the same illness.…”

“…Deviations of individuals from the normative range can be quantified as z-scores (Marquand et al, 2019). This approach has recently been used to dissect the heterogeneity of several mental disorders (Wolfers et al, 2018;Zabihi et al, 2019;Wolfers et al, 2020;Zabihi et al, 2020), providing compelling evidence that brain abnormalities of patients with psychiatric disorders cannot be captured in a case-control setting, i.e., by average group differences between patients with a specific disorder and healthy controls. Thus, normative modeling allows us to enhance classical symptom-based diagnostics by incorporating biological and environmental factors in a principled way.…”

Towards robust and replicable sex differences in the intrinsic brain function of autism

“…Overall, the extent to which each sample-related factor affects replicability needs to be systematically examined in future well-powered studies. Only this type of studies will allow for emerging subtyping approaches to dissect heterogeneity by brain imaging features using a range of data-driven methods [ 107 , 108 ], including normative modelling [ 72 , 109 , 110 ].…”

Mental health explains individual deviations from normative range in cognition-associated brain states

“…Deviations from the estimated normative range per individual (i.e., abnormality score) can be then quantified and linked to relevant variables (e.g., individual's mental health). Normative modeling has been increasingly applied to study associations between brain functions and behavior, particularly for clinically relevant conditions where significant heterogeneity may exist that cannot be captured by categorical partitioning of the cohort into traditional patientcontrol groups [20][21][22][23] . Here, we apply normative modeling to a population cohort to investigate whether individual-level abnormality in brain state measures as estimated from cognitive ability can be attributed to overall mental health.…”