When sleep, not mood, drives the damage: demyelination maps differently in long COVID insomnia versus depression
- In 76 long COVID patients and 22 healthy controls, quantitative myelin imaging (macromolecular proton fraction mapping) showed that the patients with isolated insomnia – not those with depression – carried the most widespread demyelination, spanning commissural, projection and association white-matter pathways plus the thalamus and midbrain.
- The anatomy of the damage tracked the symptom: insomnia mapped onto broad fibre-tract and juxtacortical demyelination, whereas depression mapped onto a focal pattern centred on the basal ganglia – globus pallidus, putamen and external capsule.
- Cognition followed the same split. The insomnia group performed worst on global screening and on inhibitory control, while the depression group was most impaired on processing speed and set-shifting, and the comorbid group on working memory.
- Demyelination severity correlated with serum levels of anti-PLP, an autoantibody against a myelin protein, which were highest in the insomnia group – pointing to an autoimmune-mediated injury rather than a purely vascular one.
Long COVID has settled into clinical practice as a vague constellation of fatigue, brain fog and mood disturbance, and the temptation is to treat the whole picture as one diffuse condition. This study from a Tomsk-led group argues the opposite: the persistent neuropsychiatric symptoms partition into distinct biological lesions, and the partition is visible on quantitative imaging.
The team studied 76 patients with long COVID and 22 controls, sorting the patients into four groups – isolated insomnia, isolated depression, comorbid insomnia plus depression, and a residual long COVID group without either. Every participant underwent neuropsychiatric scaling, a cognitive battery (MoCA, Stroop, working-memory and trail-making tests), a panel of blood biomarkers, and MRI with macromolecular proton fraction mapping, a technique that estimates myelin content quantitatively rather than inferring it indirectly.
What the data shows
The headline is counterintuitive. Clinicians tend to read depression as the more "neurological" of the two complaints, yet here insomnia was the marker of the heaviest white-matter injury. The insomnia group showed the most extensive demyelination – across the corpus callosum and tapetum, the corona radiata, internal capsule and corticospinal tract, several association bundles, and most juxtacortical regions, the thalamus and the midbrain – and the severity scaled with how bad the insomnia was. Depression produced a smaller, regionally different footprint, concentrated in the globus pallidus, putamen and external capsule. The comorbid group preferentially lost myelin in the inferior fronto-occipital fasciculus, the uncinate fasciculus and the cingulum.
The cognitive profile mirrored the anatomy. Diffuse fibre-tract demyelination in the insomnia group went with global and inhibitory deficits, the kind of impairment one expects when long-range connectivity is degraded across many bundles at once. Basal-ganglia involvement in the depression group went instead with slowed processing and impaired switching, consistent with disruption of fronto-striatal loops. The comorbid group, losing myelin along the uncinate and cingulum, was selectively worst on working memory. Tying the picture together, anti-PLP myelin autoantibodies were highest where demyelination was greatest, and correlated both with imaging severity and with the persistence of the insomnia and depressive symptoms over time. That correlation is the study's most provocative thread: it suggests the white-matter injury is not a static scar left by the acute infection but an ongoing, immunologically active process that keeps pace with how long the symptoms last.
Why this matters for the clinic
For a practitioner, the lesson is to stop treating post-viral insomnia as a downstream nuisance of low mood. In this cohort it was the louder signal of an active, possibly autoimmune, demyelinating process – and the cognitive complaints a patient brings are not interchangeable. A long COVID patient whose chief problem is unrefreshing sleep and poor inhibitory control is, on this evidence, a different neurobiological case from one whose problem is anhedonia and mental slowing, and may warrant different surveillance. The study cannot yet license a treatment change, but it should sharpen the history: ask about sleep as carefully as about mood, and take persistent post-viral insomnia as a sign to look harder, not to reassure.
In long COVID it was insomnia, not depression, that flagged the most widespread myelin loss – and the highest levels of an anti-myelin autoantibody.
This is a cross-sectional case-control study with small subgroups (12 to 32 patients each), so the associations cannot establish that demyelination causes the symptoms rather than accompanying them. The autoantibody correlation is suggestive of an autoimmune mechanism but does not prove it. Findings need replication in a larger, longitudinal sample before they change practice.