Interrelationship between perivascular spaces and white matter hyperintensities: A latent growth curve analysis Authors: Inga Menze, Jose Bernal Moyano, Pinar Kaya, Cagla Aki, Malte Pfister, Jonas Geisendörfer, Renat Yakupov, Roberto Duarte, Maria del C. Valdes Hernandez, Michael T. Heneka, Frederic Brosseron, Matthias Schmid, Wenzel Glanz, Enise I Incesoy, Michaela Butryn, Ayda Rostamzadeh, Dix U. Meiberth, Oliver Peters, Lukas Preis, Dominik Lammerding, Daria Gref, Josef Priller, Eike Jakob Spruth, Slawek Altenstein, Andrea Lohse, Stefan Hetzer, Anja Schneider, Klaus Fliessbach, Okka Kimmich, Ina R Vogt, Jens Wiltfang, Claudia Bartels, Björn H. Schott, Niels Hansen, Peter Dechent, Katharina Buerger, Daniel Janowitz, Robert Perneczky, Boris‐Stephan Rauchmann, Stefan Teipel Published: 2025 ABSTRACT AbstractBackgroundInadequate glymphatic clearance through perivascular spaces (PVS) is hypothesized to contribute to the formation of white matter hyperintensities (WMH). However, longitudinal evidence for such a mechanistic link in aging remains limited. Using multivariate modelling, we investigated the interrelationship between PVS and WMH over time to elucidate potential cascades of early cerebrovascular alterations and tested whether AD‐biomarkers and inflammatory markers associated with vascular disease can explain individual variability in their occurrence and progression.MethodsWe quantified PVS and WMH using T1w MPRAGE and T2w FLAIR imaging of 439 cognitively unimpaired participants from the DELCODE study (52.85% females; meanage = 69.88±5.72), who underwent annual scans over a four‐year period and attended at least three visits (n observations = 1790; meannumber of visits = 4.08±0.79). We employed latent growth curve modelling to assess reciprocal connections between PVS and WMH, focusing on their initial volumes (latent intercepts) and their rates of change over four years (latent slopes). We used log10‐transformed total PVS and WMH volumes, and controlled for age, sex, years of education, total cardiovascular risk score, and total intracranial volume. We then derived interindividual latent factor scores and tested their relation to CSF‐derived AD‐biomarkers (Aβ42/40, pTau181; available for n = 195; z‐scored) and inflammatory markers (CRP, IL‐6; available for n = 125; Box‐Cox‐transformed) via Spearman’s correlation (FDR‐corrected).ResultsThe model showed good model fit (CFI = 0.997; RMSEA = 0.021; SRMR = 0.017; Fig. 1A). WMH and PVS volumes increased over time (interceptWMH‐slope = 0.068, SE = 0.004, Z = 16.490, p<0.001; interceptPVS‐slope = 0.036, SE = 0.007, Z = 4.927, p<0.001; Fig. 1B). Participants with higher baseline PVS volumes not only had higher baseline WMH volumes (covariancePVS‐intercept&WMH‐intercept = 0.120, SE = 0.040, Z = 2.936, p = 0.003; Fig. 1C) but also tended to exhibit faster WMH volume increase over time (covariancePVS‐intercept&WMH‐slope = 0.007, SE = 0.004, Z = 1.796, p = 0.072; Fig. 1C). In this sample of cognitively unimpaired participants, biomarkers of AD and inflammation did neither relate to individual baseline differences nor progression rates (Table 1).ConclusionOur findings are consistent with the notion that PVS dysfunction might contribute to and precede WMH progression (Fig. 1D). However, the individual variability requires further investigation to elucidate mechanisms driving PVS dysfunction in the first place. Unraveling the interrelationships and further factors contributing to cerebrovascular alterations will be crucial to understand pathological cascades in aging that could inform targeted treatment strategies.