MRI Index Tied to the Brain’s Waste Clearance System May Be Misleading in Epilepsy

A widely promoted MRI measure that researchers have used as a window into the brain’s waste-clearance system may be telling a very different story in people with epilepsy than previously believed. In a retrospective study of 79 people with drug-resistant focal epilepsy evaluated for surgery at the Medical University of Vienna between September 2019 and March 2023, researchers found that the diffusion tensor image analysis along the perivascular space, or DTI-ALPS index, showed no significant differences between patients and 25 age- and sex-matched healthy controls. Instead, the metric tracked something more mundane and more local: the physical distance between a patient’s epileptic lesion and the small brain region where the measurement is taken, and whether the patient had already undergone surgery.

The findings, published in European Radiology Experimental, add fuel to a growing debate about what the DTI-ALPS index actually measures. The index was originally proposed as a noninvasive surrogate for glymphatic system function, the network of fluid channels along blood vessels that helps clear metabolic waste products from the brain. It exploits a geometric quirk of brain anatomy: at a specific level adjacent to the lateral ventricles, medullary vessels, projection fibers, and association fibers run in distinct, roughly perpendicular directions, allowing diffusivity along each direction to be separated and combined into a single index value.

But the assumptions behind that geometry have come under scrutiny. A growing body of evidence suggests the metric is not specific to glymphatic dysfunction. Recent studies have indicated that the index primarily reflects vascular pathology, and experimental work has shown that crossing fibers systematically inflate the DTI-ALPS index through mathematical bias, undermining its validity as a glymphatic biomarker. Even the authors of the original method have revisited its definition, cautioning that the index should be interpreted carefully as a measure of glymphatic system function. Microstructural degeneration of tissue related to natural aging has also been shown to affect the metric, even in areas perpendicular to the medullary veins.

Against that backdrop, the Vienna team set out to systematically evaluate the index in a large and clinically diverse epilepsy cohort. The 79 patients, with a mean age of 28.6 years and including 47 men and 32 women, all underwent comprehensive presurgical work-up, including prolonged video-electroencephalogram monitoring, a state-of-the-art 3-Tesla MRI protocol, functional MRI, and neuropsychological testing. The etiologies spanned the spectrum of drug-resistant focal epilepsy: MRI-negative cases were the most common at 27 patients, followed by 20 with malformations of cortical development, 15 with lesional temporal lobe epilepsy, and 14 with low-grade epilepsy-associated tumors, with a small remainder attributable to infarction or vascular pathologies. Patients with concomitant neurological diseases, a history of chemotherapy, or prior brain surgery were excluded.

The imaging pipeline was deliberately standardized. Diffusion MRI data were preprocessed with QSIPrep, and regions of interest were automatically defined as 5-millimeter spheres within the bilateral projection and association fibers of the superior corona radiata and the superior longitudinal fasciculus, at the level of the lateral ventricle body, using a validated atlas-based workflow. The team also measured the distance between the DTI-ALPS sampling region on the affected side and the closest border of any MRI-visible lesion, performed manually in consensus by two researchers. Statistical analyses used multivariable linear regression with pre-specified predictors of seizure frequency, age at MRI, and sex, along with linear mixed models for hemispheric and subgroup comparisons, with false discovery rate correction applied across families of tests.

The headline result was a null one. The multivariable model did not reach statistical significance overall, and among the individual predictors, only patients in the high seizure frequency group, defined as experiencing at least monthly seizures, showed a nominally significant association with lower DTI-ALPS values compared to controls, with a beta of -0.102 and an unadjusted p-value of 0.045. That signal did not survive correction for multiple comparisons, which yielded an adjusted p-value of 0.18. Age at MRI and sex showed no significant associations. Comparing affected and unaffected hemispheres within patients also revealed no significant difference, although the affected hemisphere showed slightly lower mean values, at 1.529 versus 1.577 on the contralateral side.

Subgroup analyses told a similar story of fragile trends. A hemispheric difference reached nominal significance only in the frontal lobe group, where affected hemispheres showed a mean of 1.375 versus 1.49 on the unaffected side, but this finding did not survive multiple comparison correction either. Temporal lobe and other lobe groups showed non-significant trends in the same direction, and neither lesional nor non-lesional epilepsy showed significant hemispheric differences. Notably, the absence of an age correlation in this cohort is likely explained by the comparatively young mean age of the patients, which was below 40 years, the age at which DTI-ALPS indices are thought to peak, and the largest prior normative study also found age, not epilepsy status, to be the main predictor of the index.

What did stand out was anatomy. In the subgroup of patients with MRI-visible lesions, the distance between the closest lesion border and the DTI-ALPS region of interest showed a significant positive correlation with the index in the overall cohort, with a beta of 0.007 and a p-value of 0.014, meaning that lesions lying closer to the sampling region were associated with more abnormal values. The researchers interpret this as implicating local white matter disruption as a modifier of the measurement. The pattern was further corroborated by a longitudinal analysis: in a subset of 12 patients with paired pre- and postoperative MRI data, a Wilcoxon signed-rank test revealed a significant reduction in DTI-ALPS scores following surgery, with a p-value of 0.002. Surgery, like the lesions themselves, physically alters the fiber tracts near the measurement region.

Equally telling were the variables that showed no relationship to the index. The researchers found no associations with disease duration, sex, or normalized gray and white matter volumes, the latter assessed by linear regression adjusting for age and sex. The mean duration of epilepsy in the cohort was 12.3 years, yet this clinical burden left no detectable trace on the metric. In the authors’ view, this pattern supports the possibility that local disruptions due to pathologies, or microstructural degeneration associated with aging, rather than glymphatic dysfunction per se, account for at least part of the observed variance in the index. The study cannot definitively determine whether any component of the measure reflects glymphatic function, because complementary noninvasive techniques such as intravoxel incoherent motion imaging or intrathecal gadolinium applications were not available.

The authors also acknowledge the limitations inherent to this kind of research. The study is cross-sectional, preventing assessment of longitudinal trajectories except in the small postoperative subgroup. Quantifying seizure burden is notoriously difficult, complicated by seizure amnesia and underestimation, which is why the team used the seizure frequency score as a pragmatic compromise, though it remains subject to bias. Potential confounding effects of medication load also merit consideration, since prior studies have reported lower DTI-ALPS values in drug-resistant compared to drug-responsive epilepsy, raising the possibility that drug exposure could obscure or amplify disease-related associations. And an inherent problem in all DTI-ALPS studies is the subjective selection of voxels for region-of-interest placement, restricting comparability across studies, while the standard region is agnostic to the location of a patient’s disease focus.

For the epilepsy field, the practical message is one of caution. The DTI-ALPS index, whatever it captures about fluid movement along perivascular spaces, is evidently sensitive to the local tissue architecture through which its measurement fibers pass, and in focal neurological diseases that sensitivity can be dominated by the very lesions clinicians are trying to characterize. The authors conclude that local disruption of fiber tracts may act as an important modifier of the measure and that caution is warranted when interpreting the index as a glymphatic biomarker in focal epilepsy. As interest in noninvasive glymphatic imaging continues to grow, this study serves as a reminder that a metric designed to probe one biological system can be quietly rewritten by another: the tangled, lesion-scarred, surgically reshaped white matter of the epileptic brain itself.

Subject of Research: Confounding effects of lesion proximity and surgery on the DTI-ALPS index as a glymphatic biomarker in focal epilepsy

Article Title: Reassessment of the DTI-ALPS index in focal epilepsy

Article References: Chambers, S., Hofer, L., Tranfa, M., Kienast, P., Shawarba, J., Stepponat, R., Jud, J., Aull-Watschinger, S., Pataraia, E., Fischmeister, F. P. S., Barkhof, F., Hangel, G., Rössler, K., Pontillo, G., Kasprian, G., & Haider, L. (2026). Reassessment of the DTI-ALPS index in focal epilepsy. European Radiology Experimental, 10(1), Article 162. https://doi.org/10.1186/s41747-026-00799-8

Image Credits: AI Generated

DOI: 10.1186/s41747-026-00799-8

Keywords: DTI-ALPS index, focal epilepsy, glymphatic system, diffusion tensor imaging, MRI biomarker, white matter integrity, drug-resistant epilepsy, lesion proximity, epilepsy surgery, perivascular spaces, presurgical evaluation, QSIPrep