Disrupted glymphatic function and its relationship with sleep and cognitive impairment in ME/CFS assessed via DTI-ALPS

Thapaliya et al. 2026 — DTI-ALPS in ME/CFS

TL;DR

In a single-center cross-sectional study of 31 adults with ME/CFS and 27 healthy controls, the mean bilateral diffusion tensor imaging along the perivascular space (DTI-ALPS) index was lower in ME/CFS (1.44 ± 0.086 vs. 1.51 ± 0.11; raw p=0.014, false-discovery-rate-adjusted p=0.028). The between-group difference was detectable on the right but not the left; however, the direct between-group asymmetry comparison was null. Within ME/CFS, lower bilateral DTI-ALPS correlated with worse single-item self-reported sleep disturbance (r=−0.47, p=0.013) and impaired concentration (r=−0.43, p=0.026), but not with the separate World Health Organization Disability Assessment Schedule (WHODAS) cognitive score. The results are preliminary associations: DTI-ALPS is a local deep-white-matter directional-diffusion measure with uncertain glymphatic specificity, and the study did not establish whether sleep disturbance, altered brain microstructure, or impaired clearance caused the other findings.¹²

Design

Field	Value
Design	Cross-sectional case–control imaging study
Site	National Centre for Neuroimmunology and Emerging Diseases, Gold Coast, Australia
Enrollment	32 ME/CFS and 29 healthy controls (n=61)
Analyzed	31 ME/CFS and 27 healthy controls (n=58)
Imaging exclusions	One ME/CFS participant and two controls excluded for motion artifact
Age eligibility	18–65 years
ME/CFS ascertainment	Physician diagnosis plus Canadian Consensus Criteria or International Consensus Criteria
Additional case restriction	Self-reported no prior COVID-19 infection
Controls	No chronic illness or current medical condition
Ethics	Griffith University Human Research Ethics Committee 2022/666; written consent
Registration	No prospective study registration or prespecified analysis plan reported

The exclusion screen removed participants with mental illness, malignancy, autoimmune disorders, neurological disease, or cardiovascular conditions; pregnancy and breastfeeding were also exclusionary. This produces a selected ME/CFS cohort with fewer comorbidities than are often encountered clinically.¹

Participant characteristics

Measure	ME/CFS (n=31)	Healthy controls (n=27)	Reported p
Age, years	42.95 ± 13.51	37.92 ± 10.9	0.12
Sex, female/male	21/10	19/8	not reported
Body-mass index	24.43 ± 4.35	24.26 ± 2.56	0.96
Illness duration, years	14.32 ± 11.29	not applicable	not applicable

Age and body-mass index did not differ statistically between groups. Age and sex were included as covariates in the imaging group comparisons and correlation analyses.¹

Clinical, sleep, and cognition instruments

The paper used several self-report instruments; it reports no polysomnography, actigraphy, Pittsburgh Sleep Quality Index assessment, or objective neuropsychological testing.¹

Construct	Instrument and direction	Role in this paper
Pain	SF-36 version 2, 0–100; lower score=worse	Group comparison; correlation with DTI-ALPS reported null
Physical function	SF-36 version 2, 0–100; lower score=worse	Group comparison; correlation with DTI-ALPS reported null
Cognitive impairment	WHODAS 2.0-derived 0–100 score; higher=worse	Group comparison; correlation with DTI-ALPS reported null
Functional disability	Dr Bell’s Chronic Fatigue and Immune Dysfunction Syndrome Disability Scale	Group comparison
Fatigue impact	Modified Fatigue Impact Scale (MFIS)	Group comparison
Fatigue severity	NCNED Research Registry single item, 0–5; higher=worse	Correlation analysis; null
Impaired concentration	NCNED Research Registry single item, 0–5; higher=worse	Headline correlation with bilateral DTI-ALPS
Sleep disturbance	NCNED Research Registry single item, 0–5; higher=worse	Headline correlation with bilateral DTI-ALPS

The title’s “cognitive impairment” association therefore rests on the single impaired-concentration item, not the WHODAS cognitive score or a performance-based cognitive battery. Similarly, “sleep” was measured by one ordinal symptom item rather than an objective or multidimensional sleep instrument. Although the methods call the symptom assessment validated, the citation attached to the three 0–5 registry items is a glymphatic-system review rather than a psychometric validation report; that citation does not establish the items’ validity.¹

Clinical group differences

Measure	ME/CFS	Healthy controls	Reported p
SF-36 pain	51.07 ± 25.09	88.07 ± 18.05	<0.001
WHODAS-derived cognitive impairment	49.9 ± 16.93	5.53 ± 12.42	<0.001
SF-36 physical function	38.57 ± 27.51	publication prints “93.65.0 ± 20.02”	<0.001
Bell disability score	39.13 ± 14.43	97.91 ± 6.58	<0.001
MFIS	64.37 ± 10.76	7.48 ± 12.24	<0.001

Among 29 ME/CFS participants with the 0–5 registry items completed, moderate-to-very-severe symptoms were reported for fatigue by 28/29 (96.6%), impaired concentration by 26/29 (89.6%), and sleep disturbance by 23/29 (79.3%); two cases had missing data for each item. The fatigue row in Table 2 prints “34.4% (10/19),” although 34.4% and the category totals indicate a denominator of 29; this appears to be another source-table typo.¹

DTI acquisition and DTI-ALPS measurement

• Imaging used a 3 T Siemens Prisma scanner with a 64-channel head–neck coil.
• The two-shell acquisition comprised 30 directions at b=1,000 s/mm², 66 directions at b=2,500 s/mm², and nine b=0 scans; only the b=1,000 shell was used for tensor fitting.
• Additional parameters were repetition/echo time 4,100/75 ms, field of view 244×244, matrix 122×122, 2.0-mm isotropic voxels, and 66 slices.
• MRtrix3 was used for denoising and motion/eddy-current correction; FSL dtifit generated diffusion tensors and fractional-anisotropy maps.
• An operator manually placed 3-mm-diameter spherical regions of interest in projection- and association-fiber regions on color-coded fractional-anisotropy maps. Intraclass correlation coefficients between two independent manual delineations were all >0.90.
• For each hemisphere, the DTI-ALPS index was calculated as mean(Dxx-projection, Dxx-association) / mean(Dyy-projection, Dzz-association), following the original DTI-ALPS method.³
• The global index was the mean of left and right indices. The asymmetry index was (left − right) / ((left + right) / 2).

The index operationalizes preferential water diffusion along the presumed perivascular-space direction relative to orthogonal fiber directions. It was interpreted by Thapaliya et al. as a proxy for glymphatic function, but the study did not measure cerebrospinal-fluid flow, interstitial-solute clearance, aquaporin-4 localization, or waste products directly.¹

This construct-validity caveat is substantive. In a separate 2026 validation study of 56 people undergoing work-up for cerebrospinal-fluid disorders, Mossige et al. found that DTI-ALPS had limited correspondence with intrathecal contrast-enhanced MRI measures of brain glymphatic function: there was no association with 48-hour tracer dynamics in any region, and the authors concluded that the index more likely reflects a feature of local white-matter integrity than brain-wide glymphatic clearance.² That study was not conducted in ME/CFS and does not negate the group difference reported here; it changes the defensible label for that difference from demonstrated “glymphatic dysfunction” to an altered DTI-ALPS index of uncertain biological specificity.

Statistical analysis

• General linear models compared global and hemispheric DTI-ALPS indices between groups and left versus right indices within groups.
• Quade nonparametric analysis of covariance was used for non-normally distributed clinical measures.
• Spearman rank correlations tested DTI-ALPS against symptom severity; the authors state that age and sex were covariates, but do not describe the partial-rank implementation in detail.
• Benjamini–Hochberg false-discovery-rate (FDR) correction was applied separately to three test families: DTI-ALPS values, clinical measures, and symptom correlations.
• Imaging results report both raw and FDR-adjusted p-values for significant comparisons. The correlation section reports p=0.013 and p=0.026 without saying whether these are raw or adjusted values and without supplying the corresponding FDR-adjusted values.

Key results

Between-group DTI-ALPS comparisons

Metric	ME/CFS	Healthy controls	Raw p	Reported p-FDR	Reported 95% CI for difference
Global DTI-ALPS	1.44 ± 0.086	1.51 ± 0.11	0.014	0.028	−0.122 to −0.014
Right DTI-ALPS	1.41 ± 0.097	1.49 ± 0.12	0.009	0.028	−0.144 to −0.022
Left DTI-ALPS	1.47 ± 0.11	1.53 ± 0.13	0.116	not reported	−0.119 to 0.013

Thus, the global and right-hemisphere differences survived the paper’s stated correction, while the left-hemisphere comparison was null.¹

Hemispheric and asymmetry analyses

Comparison	Values	Raw p	Reported p-FDR	Reported 95% CI
Left vs. right within ME/CFS	1.47 ± 0.11 vs. 1.40 ± 0.097	0.012	0.028	internally malformed in Table 3
Left vs. right within controls	1.53 ± 0.13 vs. 1.49 ± 0.12	0.261	not reported	−0.030 to 0.111
Asymmetry index, ME/CFS vs. controls	0.047 ± 0.080 vs. 0.026 ± 0.088	0.357	not reported	−0.025 to 0.068

The direct disease-versus-control asymmetry test was null. Consequently, the significant within-ME/CFS left–right comparison, alongside a null within-control comparison, does not by itself demonstrate that hemispheric asymmetry differs by disease status. This limits the paper’s stronger language about a specifically lateralized ME/CFS dysfunction.¹

Associations with symptoms in ME/CFS

DTI-ALPS measure	Symptom measure	Reported association
Global	Single-item sleep disturbance (0–5)	r=−0.47, p=0.013
Global	Single-item impaired concentration (0–5)	r=−0.43, p=0.026
Global	Single-item fatigue severity	null; coefficient and p not reported
Global or unilateral	Illness duration	null; coefficients and p-values not reported
Global or unilateral	SF-36 pain	null; coefficients and p-values not reported
Global or unilateral	WHODAS-derived cognitive impairment	null; coefficients and p-values not reported
Global or unilateral	SF-36 physical function	null; coefficients and p-values not reported
Unilateral	Sleep disturbance and impaired concentration	null; coefficients and p-values not reported

The results sentence also lists an undefined “severity” measure among null associations. It does not provide a full correlation matrix, exact null coefficients, or FDR-adjusted correlation p-values. Separate DTI-ALPS associations with Bell disability and MFIS are not clearly reported.¹

Interpretation

The data support a modest cross-sectional association between ME/CFS status and a lower bilateral DTI-ALPS index, plus within-case associations between that index and two symptom items. They do not establish any of the following directions:

1. that impaired glymphatic clearance causes sleep disturbance or impaired concentration;
2. that sleep disturbance causes the lower DTI-ALPS index;
3. that altered white-matter microstructure or other ME/CFS biology does not account for the diffusion signal; or
4. that waste accumulation or neuroinflammation occurred, because neither was measured.

The authors propose a sleep–clearance–waste–neuroinflammation mechanism, but it remains a mechanistic hypothesis rather than an endpoint of this study. gap/no-mechanism

Relevance to human aging

This is a human study, but not a study of aging: the participants were 18–65 years old, mean ages were approximately 43 and 38, and no longitudinal age effect was tested. Its relevance to this wiki is that sleep, cognition, brain fluid clearance, and neuroinflammation intersect with neurodegenerative aging. The paper should be treated as disease-specific, early evidence for a candidate imaging phenotype, not as evidence that ME/CFS accelerates brain aging or that increasing DTI-ALPS improves healthspan.

Limitations and evidence-quality flags

• Cross-sectional and small: n=58 analyzed at one center; temporal direction and progression cannot be inferred. gap/needs-replication
• First ME/CFS DTI-ALPS report: there is no independent ME/CFS replication, external validation cohort, or longitudinal follow-up. gap/needs-replication
• Uncertain construct validity: DTI-ALPS is sensitive to directional diffusion in selected deep-white-matter regions and is not a direct assay of glymphatic solute clearance. Direct 2026 comparison with intrathecal contrast-enhanced MRI found limited correspondence and suggested the index may reflect local white-matter integrity rather than brain-wide clearance.²
• Manual regions of interest: delineation ICC was >0.90, but placement and algorithm choice can influence the index.
• Sleep and cognition measurement: the two significant symptom associations used single ordinal self-report items; objective sleep and performance-based cognition were absent. The separate WHODAS cognitive score was null.
• Unmeasured modifiers: handedness and sleeping position were not measured; time of MRI, recent sleep state, and objective habitual sleep were not reported as controls.
• Sex analysis underpowered: the sample was predominantly female, and the study could not assess sex-dependent DTI-ALPS effects.
• Selected case population: broad comorbidity exclusions and the no-prior-COVID restriction limit generalizability to typical clinical ME/CFS and post-COVID illness.
• Incomplete statistical reporting: exact coefficients and p-values for null correlations, adjusted correlation p-values, and a complete model specification are absent.
• Source-table errors requiring a journal correction: Table 3 prints the global and right ME/CFS standard deviations as 0.86 and 0.97, whereas the abstract and results text give 0.086 and 0.097; the corroborated values are used on this page. For the within-ME/CFS comparison, the table gives a right index of 1.40 ± 0.097 whereas the between-group text gives 1.41 ± 0.097, an unexplained context-specific discrepancy retained as reported. Its within-ME/CFS confidence interval is malformed (lower 0.015, upper 0.0121), reverses its bounds, and does not contain the displayed mean difference; no corrected interval can be inferred. The column labeled “Cohen’s d” reports 0.107 (global), 0.119 (right), 0.045 (left), 0.66 (within ME/CFS), 0.309 (within controls), and 0.016 (asymmetry). At minimum, the independent-group and asymmetry entries are incompatible with Cohen’s d calculated from the displayed means and standard deviations; the paired entries cannot be checked without the within-person correlation or standard deviation of differences. None of these labeled effect-size values is used as an effect estimate on this page.
• No open analytic package: the data-availability statement says raw data will be made available by the authors, but no public data, code, or analysis plan is linked.

Funding and conflicts

The work was funded by ME Research UK with support from the Fred and Joan Davies Bequest and the Stafford Fox Medical Research Foundation. The authors declared no commercial or financial conflicts of interest.¹

Downstream propagation candidates

• ME/CFS: add the lower bilateral DTI-ALPS association as preliminary single-study imaging evidence, with the null WHODAS cognition and causal caveats adjacent.
• glymphatic-system: add this as the first reported ME/CFS application, explicitly labeling DTI-ALPS a local diffusion measure with uncertain glymphatic specificity.
• dti-alps: propagate the source-table defects, construct-validity caveat, null disease-versus-control asymmetry result, and distinction between a bilateral mean and whole-brain clearance.
• Sleep: mention the within-ME/CFS single-item correlation only as disease-specific observational evidence; it does not test a sleep intervention.
• cognitive-impairment: distinguish the significant impaired-concentration item from the null WHODAS cognitive association.
• neuroinflammation and brain: the paper’s waste-accumulation/neuroinflammation chain is proposed, not measured, and should not be propagated as a demonstrated ME/CFS mechanism.

Citation

Footnotes

1. thapaliya-2026-me-cfs-dti-alps · n=58 analyzed (31 ME/CFS, 27 healthy controls; 61 enrolled) · cross-sectional · model: adults with physician-diagnosed ME/CFS vs. healthy controls · doi:10.3389/fnins.2026.1875420 · PMID 42403482 · single-center, no prospective registration reported. ↩ ↩² ↩³ ↩⁴ ↩⁵ ↩⁶ ↩⁷ ↩⁸ ↩⁹ ↩¹⁰ ↩¹¹

2. doi:10.1148/radiol.252070 · PMID 41631990 · Mossige I et al. · Radiology 2026;318:e252070 · n=56 · prospective-observational secondary analysis · model: humans undergoing work-up for cerebrospinal-fluid disorders · direct comparison of DTI-ALPS with intrathecal contrast-enhanced MRI found limited correspondence; no 48-hour tracer-dynamics association in any region. ↩ ↩² ↩³

3. doi:10.1007/s11604-017-0617-z · Taoka T et al. · Japanese Journal of Radiology 2017;35:172–178 · original DTI-ALPS method paper · observational imaging method · model: humans. ↩