Assessment of subthreshold group differences in rs-fMRI by higher criticism: adult phenylketonuria

874 

Abstract Submission 

Benedikt Sundermann^1,2,3, Reinhold Feldmann⁴, Christian Mathys^1,3,5, Johanna Rau⁴, Stefan Garde⁶, Anke McLeod¹, Josef Weglage⁶, Bettina Pfleiderer⁷

¹Evangelisches Krankenhaus Oldenburg, Medical Campus University of Oldenburg, Oldenburg, Germany, ²University of Münster, Münster, Germany, ³Research Center Neurosensory Science, University of Oldenburg, Oldenburg, Germany, ⁴University Hospital Münster, Münster, Germany, ⁵University of Düsseldorf, Düsseldorf, Germany, ⁶University Hospital Münster, Münster, Germany, Münster, Germany, ⁷University of Münster, Münster, Germany, Münster, Germany

Benedikt Sundermann  
Evangelisches Krankenhaus Oldenburg, Medical Campus University of Oldenburg|University of Münster|Research Center Neurosensory Science, University of Oldenburg
Oldenburg, Germany|Münster, Germany|Oldenburg, Germany

Reinhold Feldmann  
University Hospital Münster
Münster, Germany

Christian Mathys  
Evangelisches Krankenhaus Oldenburg, Medical Campus University of Oldenburg|Research Center Neurosensory Science, University of Oldenburg|University of Düsseldorf
Oldenburg, Germany|Oldenburg, Germany|Düsseldorf, Germany

Johanna Rau  
University Hospital Münster
Münster, Germany

Stefan Garde  
University Hospital Münster, Münster, Germany
Münster, Germany

Anke McLeod  
Evangelisches Krankenhaus Oldenburg, Medical Campus University of Oldenburg
Oldenburg, Germany

Josef Weglage  
University Hospital Münster, Münster, Germany
Münster, Germany

Bettina Pfleiderer  
University of Münster, Münster, Germany
Münster, Germany

Phenylketonuria (PKU) is an inherited metabolic disorder affecting neurocognitive development (van Spronsen et al., 2021). Despite early treatment adults with PKU exhibit minor cognitive impairments (Romani et al., 2022). Knowledge about the integrity of brain networks and their functional connectivity (FC) in PKU measured by resting-state functional magnetic resonance imaging (rs-fMRI) is limited, including initial evidence of altered default mode network (DMN) and prefrontral cortex FC (De Giorgi et al., 2023).
The rarity of PKU limits fMRI sample sizes. A multi-scale testing approach using "higher criticism" (HC), a technique to detect rare and weak effects in high-dimensional data (Donoho & Jin, 2015), at the level of sub-networks could recently provide evidence of group differences widely distributed across cognition-related networks in fetal alcohol syndrome, which conventional mass-univariate testing failed to detect (Sundermann et al., 2023). The purpose of this analysis was therefore to assess the ability of this HC-based approach to detect FC alterations in cognitive networks in adults with PKU.

Patients with PKU (n = 11, age: 27.2 ± 3.7 years, all female) and healthy controls (n = 11, age: 25.9 ± 3.8 years, all female) were included. MRI data acquisition at 3 Tesla: 9:45 min rs-fMRI (234 volumes, repetition time: 2500 ms, echo time: 35 ms, spatial resolution 3.6 x 3.6 x 3.6 mm), 3D T1 (inversion-prepared turbo field echo). Preprocessing with denoising was based on fmriprep (Esteban et al., 2019). FC was calculated as linear correlation between 243 atlas regions in 10 cognition-related sub-networks (Schaefer et al., 2018). Subsequent modelling was based on multiple linear regression models comparing FC between groups (one test per pair of regions, including age and head motion as covariates). Resulting main effects of group were subjected to HC-based multiscale testing, first determining whether there is any alteration of FC in PKU in the full cognitive connectome, and subsequently resolving these findings to the level of either FC within each network or between-network connectivity (Sundermann et al., 2023). HC statistics test a joint hypothesis of an excess of low p-values in such multiple primary tests as evidence of non-zero effects (Donoho & Jin, 2015). Finally, individual connections were assessed using conventional mass-univariate testing with multiple comparison correction.

Global analysis (HC, omnibus test): FC among cognition-related brain regions was significantly altered in PKU compared with controls. Within-network analysis (HC): Significant group effects in the all 4 attention-related networks (dorsal and ventral/salience), and 2 out of 3 DMN sub-networks. No group effects in the fronto-parietal control network (3 sub-networks). Between-network analysis (HC): No significant group effects. Conventional mass-univariate analysis of single connections: No significant group effects (global multiple comparison correction).

Group differences in the DMN are in line with limited previous research on functional networks in PKU (De Giorgi et al., 2023). Alterations within attention-related systems are compatible with the clinical phenotype (i.e. including attentional deficits) previously observed in this adult age group (Romani et al., 2022). Contrary to findings of frontal-lobe associated performance deficits typically reported in children (Diamond, 2007), we did not observe group differences in the fronto-parietal control network, which is in line with the notion that these deficits might become less important when PKU patients reach adulthood (Weglage et al., 1999). Despite the relatively small sample, this HC-based analysis could detect group effects at the network level, which were missed by a conventional mass-univariate analysis. Thus, HC-based global hypothesis testing might be an informative complementary analysis approach in rs-fMRI studies of rare disorders.

Neurodevelopmental/ Early Life (eg. ADHD, autism)

Neurogenetic Syndromes ¹

Early life, Adolescence, Aging

fMRI Connectivity and Network Modeling ²

Perception and Attention Other

Cognition

FUNCTIONAL MRI

Statistical Methods

Other - phenylketonuria