Brain’s Motor Hierarchy Collapses in Early ALS, Imaging Study Finds

People newly diagnosed with amyotrophic lateral sclerosis who are still at the earliest clinical stage of the disease already show a measurable collapse in the brain’s functional hierarchy, according to a study that combined resting-state functional MRI with normative neurotransmitter maps and gene-expression atlases. In 97 patients classified as early-stage ALS, the principal gradient that organises the cortex from primary sensory areas to high-order association networks was compressed: gradient values rose in the somatomotor network and fell in the default mode network, and the magnitude of those shifts tracked with scores on a standard measure of daily functioning.

The study, published in the Journal of Cachexia, Sarcopenia and Muscle, recruited newly diagnosed patients who met the Gold Coast criteria for ALS and were classified as King’s stage 1, the definition of early-stage disease used in the analysis. They were compared with 94 age- and sex-matched healthy controls. The two groups were closely matched: mean age was 57.12 years in the patient group versus 56.14 years in controls, a difference that was not statistically significant, and the sex distribution was also comparable, with 32 men and 65 women among patients and 35 men and 59 women among controls. Twenty-three patients, or 23.7 percent, had bulbar onset. Median disease duration was 15 months, with an interquartile range of 12 to 18 months, and the median score on the revised Amyotrophic Lateral Sclerosis Functional Rating Scale was 40, in an interquartile range of 37 to 44.

Functional gradients are an imaging measure that describes how cortical regions are arranged along axes of functional integration, running from unimodal sensory and motor areas at one end to transmodal association cortices at the other. The researchers computed the first two gradient components at the voxel level from resting-state fMRI data preprocessed with a standard pipeline, then compared the 400 cortical regions of the Schaefer atlas and the seven canonical networks between groups using ordinary least squares models adjusted for age and sex, with Bonferroni false discovery rate correction for multiple comparisons. The first two gradients together explained just over 10 percent of the total variance in the data.

The group-level maps showed the expected hierarchical organisation in both groups, with the first gradient spanning unimodal to transmodal networks and the second capturing the transition from visual and somatomotor cortices to the default mode network. That fundamental topology was preserved in the patients, but its spatial extent was contracted, and the contraction was specific to the first gradient. Patients showed a significantly reduced gradient range, with a p value of 0.020, a reduced standard deviation, at p = 0.007, and a reduced variance ratio, at p = 0.008. Every metric for the second gradient was preserved, with all p values above 0.05, indicating no significant group differences on that axis.

At the regional level, after controlling for age and sex, patients showed decreased first-gradient values in nodes of the default mode network, including prefrontal and temporal regions, and increased values in bilateral somatomotor regions. These effects were mirrored at the network level: the default mode network showed a significant reduction in gradient values, with a t statistic of −4.124 and a false discovery rate corrected p value below 0.001, while the somatomotor network showed a significant increase, with t = 3.670 and a corrected p value of 0.001. No significant differences appeared for the second gradient at either the regional or network level after correction. The authors describe the pattern as a disrupted functional hierarchy, characterised by higher somatomotor and lower default mode Gradient-1 values.

Within the patient group, the gradient shifts were associated with clinical status. Higher somatomotor Gradient-1 values were linked to lower scores on the functional rating scale, a correlation of −0.274 with a corrected p value of 0.015, whereas higher default mode network values were linked to higher scores, with a correlation of 0.344 and a corrected p value of 0.004. Because the measure is cross-sectional, the association indicates that patients with more pronounced gradient disruption reported worse day-to-day functioning at the same time point; it does not establish that the imaging changes predict future decline. The primary network findings remained significant after additional adjustment for head motion, anxiety and depression scores, and after excluding patients with bulbar onset, and no group-by-sex interaction survived correction across the seven networks.

To give the imaging pattern a neurochemical context, the team used partial least squares regression to relate the case-control gradient map to a positron emission tomography atlas of 19 neurotransmitter receptors and transporters derived from healthy reference datasets. The first component accounted for 24.81 percent of the variance in the cortical alterations and correlated positively with the gradient differences, at r = 0.51. The regional alteration map showed a positive spatial correspondence with the normative distribution of the serotonin 5-HT4 receptor, at r = 0.353, and a negative correspondence with the norepinephrine transporter, at r = −0.314. The authors are careful about interpretation: the correspondence with the 5-HT4 map provides a topographic context for the gradient pattern but does not establish serotonergic involvement in ALS, and the inverse relationship with the norepinephrine transporter map should not be read as evidence of noradrenergic dysfunction, since the PET maps are normative templates rather than measurements made in patients.

Meta-analytic cognitive decoding against the NeuroSynth database added a functional annotation. Gradient alterations correlated positively with transmodal, default-mode-related terms such as theory of mind, medial prefrontal cortex and autobiographical memory, with correlations of 0.36, 0.36 and 0.32 respectively, and negatively with unimodal sensorimotor terms including secondary somatosensory cortex, somatosensory processing and tactile processing, at −0.45, −0.44 and −0.42. The same analysis applied to the second gradient produced a similar but much weaker pattern, with a top positive correlation of only 0.31 and a key negative correlation of only −0.14, underscoring that the strong spatial alignment is specific to the first gradient. These decoding results are hypothesis-generating and do not demonstrate individual-level cognitive impairment in the cohort.

The transcriptomic arm of the study linked the same alteration map to gene expression data from the Allen Human Brain Atlas, derived from six postmortem brains and mapped to 200 left-hemisphere cortical regions. Partial least squares regression identified a significant first component explaining 23.50 percent of the gradient’s variance, correlated with the case-control map at r = 0.51. The analysis yielded 2097 positively weighted and 1881 negatively weighted genes surviving Bonferroni correction, with CHCHD6 the top positive gene and NCOA3 the strongest negative one; a right-hemisphere sensitivity analysis gave concordant results. Downstream annotation split the two gene sets sharply. The positively weighted set was enriched in synaptic components, mapped by SynGO analysis to 318 genes in specific synaptic compartments, and was associated with reduced functional connectivity strength and reduced cerebral metabolic rates of oxygen and glucose, as well as positive spatial associations with neuroimaging meta-analyses of dementia, bipolar disorder, frontotemporal dementia and Alzheimer’s disease. The negatively weighted set was enriched in RNA splicing and nucleocytoplasmic transport, associated with increased connectivity and elevated metabolism, and showed negative associations with the same disease maps. Cell-type enrichment placed the positive set in oligodendrocytes and the negative set in astrocytes and GABAergic neurons, and chromosome-level analysis showed over-expression of the positive set on chromosomes 8 and 1 and of the negative set on chromosomes Y, 16 and 5, the latter notable because FUS on chromosome 5 and CHMP2B on chromosome 16 are established ALS genes.

The authors frame the work as a multi-scale, hypothesis-generating framework rather than a set of established mechanisms, and they list the limitations plainly. The single-centre cohort was modest in size, predominantly female, and may not generalise; the cross-sectional design precludes causal or prognostic inference, so the gradient alterations should not be treated as diagnostic or prognostic biomarkers. The transcriptomic analyses rest on normative data from six donors with limited right-hemisphere coverage and cannot quantify ALS-specific gene-expression changes, and the normative PET maps annotate spatial chemistry rather than measuring receptor availability in patients. Even so, the core observation stands on its own: in the earliest clinical stage of ALS, before substantial functional loss, the cortex’s principal axis of functional organisation is already compressed, with motor and default mode systems pulled toward each other in a pattern that parallels the severity of patients’ functional status.

Subject of Research: Cortical functional gradient disruption and its neurochemical and transcriptomic correlates in early-stage amyotrophic lateral sclerosis

Article Title: Cortical Motor Hierarchy Disruption Is Associated With Functional Status in Early Amyotrophic Lateral Sclerosis

Article References: Luan, J., Yun, Y., Jiao, Y., Wang, Y., Ma, M., Shan, D., Wang, H., Ji, X., Tang, Y., Li, J., Zhan, Z., Sun, X., Gao, N., Lin, P., Yan, C., Sun, X., Yu, D., Liu, S., & Liu, F. (2026). Cortical Motor Hierarchy Disruption Is Associated With Functional Status in Early Amyotrophic Lateral Sclerosis. Journal of Cachexia, Sarcopenia and Muscle, 17(5), Article e70411. https://doi.org/10.1002/jcsm.70411

Image Credits: AI Generated

DOI: 10.1002/jcsm.70411

Keywords: ALS, functional gradients, resting-state fMRI, default mode network, somatomotor network, imaging transcriptomics, neurotransmitter PET, 5-HT4 receptor, Allen Human Brain Atlas, King's stage 1, ALSFRS-R, cortical hierarchy