The core problem the researchers identify is heterogeneity. Two people diagnosed with Parkinson’s disease on the same day can follow radically different trajectories: one may retain sharp cognition for decades while motor symptoms slowly worsen, whereas another may face rapid cognitive decline within a few years. The perspective attributes this variability to differences in underlying molecular pathways, cellular vulnerabilities, network-level dysfunction, and compensatory mechanisms. Existing staging systems, including the Neuronal Alpha-Synuclein Disease Integrated Staging System and the SynNeurGe research diagnostic criteria, were designed to diagnose and stage the disease but do not fully capture how individual patients differ from one another.

The proposed framework organizes this diversity along six complementary biological axes. The first is the dominant pathological process, meaning the primary molecular and cellular mechanisms driving neurodegeneration in a given patient. The second is the spatial distribution of pathology, describing whether damage concentrates in the brainstem and deep gray nuclei, spreads preferentially through limbic structures, or reaches the neocortex early. The third is propagation architecture, the spatiotemporal trajectory through which pathology appears to travel, including the distinction between brain-first and body-first disease. The fourth is the temporal dynamics of progression, which separates slow from fast progressors and linear from punctuated decline. The fifth is co-pathology burden, the coexistence of Alzheimer-type amyloid and tau pathology, TDP-43 inclusions, and cerebrovascular disease alongside alpha-synuclein. The sixth is system response and resilience, the brain’s capacity to buffer the clinical impact of accumulating damage.

Each dimension can be interrogated with specific MRI techniques, and the authors map these tools explicitly. Quantitative susceptibility mapping quantifies iron accumulation in the substantia nigra, putamen, and dentate nucleus, and a technique called susceptibility source separation can isolate the paramagnetic iron signal to document progressive nigral iron spread. Neuromelanin-sensitive MRI measures pigment-containing neuron integrity in the substantia nigra pars compacta and the locus coeruleus, with regional neuromelanin loss correlating with motor and cognitive severity. Diffusion tensor imaging yields fractional anisotropy and diffusivity measures that index white matter integrity, while T1-weighted volumetry and deformation-based morphometry localize patterns of regional atrophy.

Spatial subtypes emerge clearly from data-driven clustering of these imaging features. The perspective describes brainstem-predominant patterns with early substantia nigra and locus coeruleus involvement, limbic-predominant patterns beginning in the amygdala and temporal cortex, neocortical or cortex-first patterns with early frontal and parietal thinning, and a diffuse-malignant pattern marked by accelerated multi-regional atrophy and worse longitudinal decline in motor function, cognition, and daily living activities. A restricted deep-gray-matter-first pattern shows early nigrostriatal involvement with relative cortical sparing and milder early cortical symptoms. Hemispheric asymmetry in neuromelanin loss and iron deposition contralateral to the side of motor onset has also been documented, reflecting asymmetric nigral involvement.

Propagation architecture is perhaps the most conceptually ambitious dimension. The authors review connectivity-informed epicenter models that combine structural connectivity with regional atrophy to infer where degeneration begins and how it spreads, with cerebellar and midbrain epicenters linked to faster motor progression and cortical or striatal epicenters to milder courses. They also discuss disease reservoir models, in which cortical thinning propagates from a reservoir through regions with strong functional connectivity, and epidemic-style models based on Susceptible-Infected-Removed mathematics that predict spatial and temporal atrophy patterns seeded from the substantia nigra. The authors caution that MRI provides only indirect insights into propagation and cannot independently verify where the pathology originated, so imaging must be interpreted alongside autonomic testing, longitudinal clinical observation, and alpha-synuclein amplification assays.

Temporal dynamics and co-pathology burden round out the picture. Longitudinal neuromelanin MRI and serial quantitative susceptibility mapping can track progressive neuromelanin loss and iron accumulation over time, while repeated volumetry quantifies rates of regional atrophy. Trajectory modeling approaches such as mixed-effects models and latent class growth analyses can distinguish linear from non-linear progression and identify inflection points where decline accelerates. On the co-pathology side, white matter hyperintensities and microbleeds reflect cerebrovascular burden, medial temporal atrophy may suggest concomitant Alzheimer’s pathology, and inferior temporal thinning has been associated with probable REM sleep behavior disorder and higher risk of cognitive impairment. The authors stress that these MRI markers are approximations that require validation with tau and amyloid PET, cerebrospinal fluid biomarkers, and genetic profiling.

The resilience dimension is the most difficult to measure directly, because compensatory mechanisms such as cognitive reserve, synaptic plasticity, and glial support are not visible to any scanner. Instead, the framework relies on imaging-clinical dissociation: patients who maintain function despite substantial atrophy occupy the high-resilience end of the spectrum, while those with severe symptoms despite minimal pathological burden sit at the low end. Resting-state functional MRI can reveal compensatory network reorganization, and cortical thickness in frontal and parietal association cortices is linked to cognitive reserve. The authors argue that resilience-based stratification may be particularly valuable in prodromal patients, where enhancing compensatory capacity before substantial pathology accumulates could delay or prevent clinical onset.

Critically, the authors emphasize that the six dimensions are continuous biological axes rather than discrete categories, and that familiar labels such as cortex-first, body-first, or brain-first should be treated as estimates of recurring patterns rather than distinct disease entities. Each patient occupies a unique position along every axis, generating a multidimensional profile that could be visualized as a radar plot. This framing explains why patients sharing the same diagnosis can experience such different trajectories, and it offers a mechanistic foundation for precision medicine approaches that current diagnostic and staging systems cannot provide.

Turning this vision into clinical practice will require a demanding translational roadmap. The authors call for standardized operational definitions of each dimension, harmonized multimodal datasets spanning imaging, clinical, genetic, and molecular biomarkers, and models that prioritize interpretability over raw predictive performance. Independent multicenter validation must demonstrate stability across imaging platforms and acquisition protocols, and prospective studies must show that stratification genuinely improves prognostic accuracy or patient selection for disease-modifying therapies. Regulatory engagement, ethical safeguards around equitable access and algorithmic transparency, and integration of MRI with fluid, molecular, and PET measures all feature in their roadmap. MRI, the authors conclude, should be viewed as one layer within a broader multimodal framework, but its scalability and accessibility make it the natural foundation for transforming Parkinson’s disease care from a one-size-fits-all approach into truly personalized medicine.

Subject of Research: A multidimensional MRI-based framework for characterizing biological heterogeneity and patient stratification in Parkinson's disease

Article Title: Biological heterogeneity in Parkinson’s disease: implications for MRI-based patient stratification

Article References: Pokotylo, M. M., Brüggemann, N., & Prasuhn, J. (2026). Biological heterogeneity in Parkinson’s disease: implications for MRI-based patient stratification. npj Parkinson's Disease, 12(1), Article 207. https://doi.org/10.1038/s41531-026-01544-z

Image Credits: AI Generated

DOI: 10.1038/s41531-026-01544-z

Keywords: Parkinson's disease, MRI, biological heterogeneity, patient stratification, neuromelanin-sensitive MRI, quantitative susceptibility mapping, diffusion tensor imaging, alpha-synuclein, co-pathology, brain resilience, precision medicine, npj Parkinson's Disease

The measurement at the heart of the study is called the oxygen extraction fraction, or OEF. In simple terms, OEF is the proportion of oxygen carried by the blood that is actually captured by tissue to keep cells functioning. In healthy brains, this value is remarkably stable across gray and white matter and varies little from person to person, which makes deviations potentially meaningful. When cerebral blood flow drops, the brain typically compensates by extracting a larger fraction of the oxygen that does arrive, a relationship governed by basic physiology and Fick’s principle, which links oxygen consumption to the product of blood flow and extraction. A rising OEF can therefore be read as a warning light on the dashboard: the plumbing may be struggling even if total energy use, measured as the cerebral metabolic rate of oxygen, has not yet declined.

What makes this study technically notable is that the team measured OEF without contrast agents or radiation, using quantitative susceptibility mapping, an MRI method that converts tiny magnetic field disturbances caused by deoxygenated hemoglobin into quantitative maps of venous oxygenation. The researchers acquired multiecho gradient-recalled echo images on a 3 Tesla Siemens Prisma scanner, unwrapped the phase data with an open-source algorithm, removed background field contributions, and inverted the local field maps into susceptibility values using a morphology-enabled dipole inversion approach. Cerebrospinal fluid in the ventricles served as a zero reference, and a deep learning model based on an attention-gated 3D U-Net segmented cortical veins automatically, achieving a Dice coefficient of about 0.79 in held-out testing. Partial volume effects were corrected with an orientation-aware cylindrical fitting method, and subject-specific hematocrit values from blood draws were plugged into the biophysical equation linking susceptibility to oxygen saturation.

The study enrolled 32 participants aged 55 to 85, split into 14 older adults with type 2 diabetes and obesity and 18 age- and sex-matched controls. The diabetes group had a body mass index between 30 and 45 kilograms per square meter and hemoglobin A1c values between 6.0 and 10 percent, while controls had normal-range BMI and HbA1c below 5.7 percent. Anyone with prior stroke, major neurological or psychiatric illness, heart disease, or uncontrolled hypertension was excluded, and the two groups were statistically indistinguishable in age, sex, and education. On average, the participants with diabetes had been diagnosed about 5.7 years earlier and were, by clinical standards, fairly well controlled, with a mean HbA1c of 6.7 percent. Their insulin resistance, however, measured by the HOMA-IR index, was markedly elevated at 5.6 compared with 2.2 in controls, well above a commonly cited metabolic syndrome threshold.

The headline result was striking in its consistency. Whole-brain OEF averaged 29.5 percent in the diabetes and obesity group versus 27.3 percent in controls, a relative increase of about 8 percent that survived statistical correction for multiple comparisons. Every lobar region examined, including the frontal, temporal, parietal, occipital, and insular cortices, showed the same elevation. The largest gap appeared in the occipital lobe, where OEF reached 37.0 percent in the metabolic group against 33.3 percent in controls, an 11 percent relative difference. Crucially, cerebral blood flow measured with arterial spin labeling was mostly similar between groups, and the cerebral metabolic rate of oxygen, calculated by combining flow and extraction, did not differ in any region. That combination, higher extraction with preserved consumption, is exactly the signature expected when the brain is compensating for vascular stress rather than suffering outright metabolic failure.

The cognitive testing added a sobering dimension. Although all scores fell within clinically normal ranges, the diabetes and obesity group performed significantly worse on the Repeatable Battery for the Assessment of Neuropsychological Status, with lower scores in list learning, coding, attention, and the global cognition index. They also scored markedly lower on the Color-Word Switching condition of the Delis-Kaplan Executive Function System, a demanding test of inhibitory control and mental flexibility, where the difference was the largest observed. Depression scores on the Geriatric Depression Scale were equivalent between groups, arguing against mood as a confound. The pattern, spanning attention, processing speed, and executive function rather than a single domain, overlaps with what clinicians see in early vascular cognitive impairment, the spectrum of thinking difficulties caused by compromised brain blood supply.

Even more intriguing were the correlations between brain physiology and behavior. Across all participants, higher regional OEF was consistently associated with poorer cognitive performance, with the strongest relationships appearing in the temporal lobe, where OEF correlated at minus 0.62 with executive switching performance, and minus 0.41 with attention. The direction was uniformly negative across every region tested. On the metabolic side, whole-brain OEF rose with blood leptin levels, a hormone produced by fat tissue that is elevated in obesity and linked to adipose dysfunction and neuroinflammation, and fell with higher HDL cholesterol, the lipoprotein widely regarded as cardioprotective. HbA1c and fasting insulin showed positive trends that approached significance. These associations are exploratory and were not corrected for multiple comparisons, so the authors are careful to frame them as hypothesis-generating rather than definitive.

The interpretation favored by the researchers is that elevated OEF represents an early compensatory response to metabolic and vascular stress, a kind of cerebral overdrive that precedes measurable declines in resting brain metabolism. This fits with prior work showing that OEF is elevated in people with heavy vascular risk burdens, particularly those with cognitive impairment, and with evidence that the occipital lobe is a common site of reduced perfusion in type 2 diabetes. It also echoes findings in other conditions, from asymptomatic carotid artery disease to sickle cell anemia and obstructive sleep apnea, where altered oxygen extraction tracks with cognitive difficulty. If confirmed, OEF could serve as a sensitive, quantitative, and entirely noninvasive marker of cerebrovascular vulnerability, potentially distinguishing vascular contributions to dementia risk from purely neurodegenerative ones, something that visually rated markers like white matter hyperintensities cannot do with the same objectivity.

The caveats are real and the authors state them plainly. This was a small, single-center, cross-sectional pilot study, so it cannot establish causation, cannot disentangle the independent effects of diabetes from those of obesity, and cannot rule out that differences in medication use, including higher rates of antihypertensive treatment in the metabolic group, contributed to the imaging findings. The exploratory statistical analyses carry an elevated risk of false positives. What the study does establish is a plausible and testable physiological link: older adults with type 2 diabetes and obesity show diffusely elevated cerebral oxygen extraction, and that elevation tracks with both adverse metabolic markers and weaker attention and executive performance. The next step, the researchers say, is larger longitudinal cohorts that follow OEF, blood flow, and cognition over time to determine whether this compensatory signature predicts who will decline, and whether interventions that improve metabolic health can normalize the brain’s oxygen economy before damage becomes permanent.

Subject of Research: Cerebral oxygen extraction fraction measured by quantitative susceptibility mapping in older adults with type 2 diabetes and obesity and its associations with cognition and metabolic markers

Article Title: Regional oxygen extraction fraction is higher in older adults with type 2 diabetes and obesity: associations with cognition and metabolic markers

Article References: Elsharydah, M. N., Al-Khoury, Y., Ragunatha Rao, S., Schaffert, J., Saha, A., Kaufman, J., Rezaei, A., Leary, J., Jin, A., Almandoz, J. P., Gauthier, C. J., & Thomas, B. P. (2026). Regional oxygen extraction fraction is higher in older adults with type 2 diabetes and obesity: associations with cognition and metabolic markers. GeroScience. https://doi.org/10.1007/s11357-026-02590-4

Image Credits: AI Generated

DOI: 10.1007/s11357-026-02590-4

Keywords: type 2 diabetes, obesity, oxygen extraction fraction, quantitative susceptibility mapping, cerebral blood flow, cognitive impairment, dementia risk, GeroScience, MRI biomarker, insulin resistance, leptin, vascular cognitive impairment

The research, led by Yan Xie and Shaolong Wu of Henan Provincial People’s Hospital and Tongji Hospital in China, together with colleagues including Yi Wang at Weill Cornell Medicine and Meiyun Wang as senior author, compared 119 patients with relapsing–remitting multiple sclerosis, 47 patients with neuromyelitis optica spectrum disorders, and 93 healthy controls. The choice of comparison group matters. Neuromyelitis optica spectrum disorder was once lumped together with multiple sclerosis as a variant of the same disease, but it is now understood to be a fundamentally different condition, driven by antibodies against a water channel protein called aquaporin-4 rather than by the immune-mediated destruction of myelin that characterizes multiple sclerosis. Contrasting the two diseases in the same imaging framework therefore provides a kind of natural experiment: if a signal appears in multiple sclerosis but not in neuromyelitis optica, it likely reflects processes specific to multiple sclerosis pathology.

The technical heart of the study is an advanced magnetic resonance imaging method called susceptibility separation imaging, sometimes referred to as chi-separation. Conventional MRI is superb at revealing lesions, the bright scars that dot the brains of people with multiple sclerosis, but it says little about the underlying chemistry. Quantitative susceptibility mapping, a technique refined over the past decade, goes further by converting subtle distortions in the magnetic field measured by the scanner into maps of magnetic susceptibility, a physical property tied to the concentration of substances such as iron and myelin. Iron, sequestered in ferritin and heme within cells, is strongly paramagnetic and pushes susceptibility in the positive direction, while myelin, with its lipid-rich sheaths, is diamagnetic and pulls susceptibility negative. The trouble is that ordinary susceptibility maps blend these contributions into a single number, making it impossible to say whether a change reflects iron, myelin, or both.

Susceptibility separation imaging tackles this ambiguity head-on. Reconstructed from standard three-dimensional multi-echo gradient echo acquisitions, the method mathematically decomposes the total susceptibility signal into a positive component, denoted chi-pos, and a negative component, chi-neg. Histological validation work by other groups has shown that the positive component corresponds predominantly to iron stores, whereas the negative component is dominated by myelin. In effect, the technique gives researchers two separate dials, one roughly tracking tissue iron and the other tracking myelin integrity, from data that can be collected on conventional clinical scanners in a single session. That combination of chemical specificity and clinical practicality is what makes the approach attractive for large patient studies.

To focus on the U-fibers, the team defined the region of interest as the white matter lying within four millimeters beneath the gray-white matter boundary, and then divided this thin ribbon into a set of subregions across the whole brain. From each subregion they extracted total susceptibility as well as the positive and negative components, and used general linear models to compare the three groups while partial correlation analyses probed links with clinical measures. The results were strikingly asymmetric. In the multiple sclerosis patients, the positive susceptibility component, the putative iron signal, was significantly decreased in multiple temporal and limbic U-fiber regions relative to healthy controls. The negative component, associated with myelin, was significantly increased across a far more widespread set of U-fiber subregions.

The direction of these changes is as informative as their presence. A decrease in the iron-related signal in superficial white matter fits with a growing body of work suggesting that multiple sclerosis involves not just iron accumulation in deep gray matter structures but actual iron loss in certain tissue compartments, possibly reflecting the death of iron-rich glial cells, efflux of iron from damaged tissue, or dilution as tissue shrinks. Meanwhile, an increase in the negative, myelin-related component across broad U-fiber territories is consistent with demyelination and the inflammatory remodeling of myelin sheaths, processes that alter the diamagnetic signature of the tissue even outside visible lesions. That the myelin signal was abnormal across so many regions underscores a point that neuropathologists have made for decades: the damage in multiple sclerosis extends well beyond the discrete scars that radiologists can see on standard scans.

Perhaps the most consequential finding, however, is what the researchers did not find. When they compared the neuromyelitis optica spectrum disorder patients with healthy controls, no U-fiber subregion showed a significant difference in any susceptibility metric. In other words, the iron and myelin chemistry of the superficial white matter appeared essentially preserved in this disease, despite the fact that patients with neuromyelitis optica can carry a substantial burden of brain lesions and neurological impairment. This dissociation strengthens the argument that the susceptibility changes seen in multiple sclerosis are not a generic consequence of having an inflammatory brain disease, but instead reflect disease-specific processes, most plausibly the primary demyelination that defines multiple sclerosis pathology.

The story acquires an important caveat when the two patient groups are compared directly with each other. After the researchers adjusted for potential confounding effects of disease severity and lesion burden, the significant differences in susceptibility metrics between the multiple sclerosis and neuromyelitis optica groups disappeared. This suggests that part of the apparent disease-specificity may be explained by the fact that the multiple sclerosis patients in this cohort had greater lesion loads or more advanced disease. The U-fiber changes, in this reading, scale with the overall amount of tissue injury rather than being uniquely programmed by multiple sclerosis biology. Disentangling whether superficial white matter chemistry is a cause, a consequence, or simply a barometer of lesion burden will require longitudinal studies that follow patients from the earliest stages of disease.

What the study does establish firmly is a link to clinical disability. Within the multiple sclerosis group, the negative susceptibility component in U-fiber subregions correlated significantly with scores on the Expanded Disability Status Scale, the standard clinical measure of neurological impairment, with the strongest association in the left superior frontal region. That region houses short association fibers serving frontal lobe networks involved in executive function and cognitive control, domains that are frequently impaired in multiple sclerosis even when physical disability seems modest. The finding dovetails with prior work showing that U-fiber diffusion and susceptibility abnormalities relate to cognitive deficits and neurodegeneration, and it raises the possibility that susceptibility metrics could serve as surrogate markers, bridging the persistent gap between what radiologists see on conventional scans and what patients actually experience.

The implications reach toward clinical practice. Because susceptibility separation imaging can be reconstructed from gradient echo data already acquired in routine protocols, it could in principle be added to existing MRI examinations without new scan time, offering clinicians a quantitative readout of iron and myelin status in tissue that conventional sequences render as unremarkable. Such markers could eventually help monitor disease progression, evaluate whether experimental remyelinating therapies are working, and distinguish multiple sclerosis from its mimics at earlier stages. For now, the study’s authors are careful to frame the metrics as suggestive surrogate markers rather than diagnostic tests, and the cross-sectional design leaves open the question of whether the U-fiber changes precede visible lesions or follow them. Still, the work adds a compelling piece to a rapidly growing picture in which the brain’s thinnest, most overlooked layer of wiring, the short fibers hugging the cortical surface, turns out to be a sensitive mirror of the disease process, and a potential window into the biology of disability in multiple sclerosis.

Subject of Research: Iron and myelin susceptibility changes in U-fiber white matter in relapsing–remitting multiple sclerosis versus neuromyelitis optica spectrum disorders

Article Title: Comparison of iron and myelin changes in U‑fiber regions of relapsing–remitting multiple sclerosis and neuromyelitis optica spectrum disorders

Article References: Xie, Y., Wu, S., Zhang, Y., Ju, C., Zhang, S., Wang, Y., Zhu, W., & Wang, M. (2026). Comparison of iron and myelin changes in U‑fiber regions of relapsing–remitting multiple sclerosis and neuromyelitis optica spectrum disorders. Journal of Neurology, 273(10), Article 596. https://doi.org/10.1007/s00415-026-14054-y

Image Credits: AI Generated

DOI: 10.1007/s00415-026-14054-y

Keywords: multiple sclerosis, neuromyelitis optica spectrum disorder, U-fibers, susceptibility separation imaging, quantitative susceptibility mapping, iron, myelin, white matter, MRI, Expanded Disability Status Scale, demyelination, superficial white matter