Overview for analysis pipeline.

(A) The pipeline for computing glymphatic dynamics and transcriptional analysis. Gradients of cortical glymphatic influx and clearance were derived using BrainSpace. Utilizing Abagen toolbox, the glymphatic patterns were further analyzed with transcriptional data from Allen Human Brain Atlas (AHBA). Both gene ontological and cell type analyses were conducted. (B) The relationship between glymphatic clearance function and spontaneous neural activity measured as fALFF was explored in a subgroup with resting-state fMRI (n = 15). (C) The mismatch index between glymphatic clearance function and spontaneous neural activity was defined to investigate its association with amyloid-β (Aβ) using public dataset.

Population characteristics

Regional glymphatic dynamics across the cortex.

(A-C) Glymphatic signal percentage change at 4.5h (A), 15h (B), and 39h (C) across cortex based on Schaefer-400 atlas. (D-E) T statistics from paired t-tests comparing the signal percentage change between 15h and 4.5h (D), as well as 39h and 15h (E) across cortex. (F) Line graph of 4.5, 15, and 39h percentage change (PC) in 400 ROIs across cortex. Dot represents mean, and bar represents standard error of the mean. (G, I) The first gradient map across cortex based on Schaefer-400 atlas for 4.5h (G) and 39h percentage change (I). The scatter plots show Pearson correlations between first gradient score and percentage change at 4.5h (H) and 39h (J). The translucent area around the regression lines represents the 95% confidential interval for the regression estimate, with fit line (blue) for significant association. The blue histogram represents the distribution of variables in x-axis, whereas the red histogram represents the distribution of variables in y-axis. L, left; R, right. ***, p < 0.001.

Comparison of glymphatic dynamics between patients with peripheral neuropathy (PN) and neurodegenerative diseases (NGD).

(A-C) Glymphatic signal percentage change (PC) at 4.5h, 15h, and 39h in PN (n = 49). (D-F) Glymphatic signal PC at 4.5h, 15h, and 39h in NGD (n = 17). (G-I) T statistics from independent two sample t tests of PC at 4.5h, 15h, and 39h between NGD and PN without false discovery rate (FDR) correction. (J) Violin plot of paired two sample t test of 4.5h PC between PN and NGD. (K) Scatter plot of Pearson correlation of 4.5h PC between PN and NGD. (L) Violin plot of paired two sample t test of 39h PC between PN and NGD. (M) Scatter plot of Pearson correlation of 39h PC between PN and NGD. In K and M, the translucent area around the regression lines represents the 95% confidential interval for the regression estimate, with fit line (blue) for significant association. The blue histogram represents the distribution of variables in x-axis, whereas the red histogram represents the distribution of variables in y-axis. L, left; R, right. ***, p < 0.001.

Association between first gradient of glymphatic flow and gene expression profiles.

(A-E) shows the result for the first gradient of 4.5h percentage change. (A) The top 20 GO biological processes linked with gene sets positively correlated with the first gradient of 4.5h percentage change, and (B) shows the gene enrichment network. (C) The top 20 GO biological processes linked with gene sets negatively correlated with the first gradient of 4.5h percentage change, and (D) shows the gene enrichment network. (E) Cell type enrichment analysis of the first gradient map. (F-J) shows the result for the first gradient of 39h percentage change. (F) The top 20 GO biological processes linked with gene sets positively correlated with the first gradient of 39h percentage change, and (G) shows the gene enrichment network. (H) The top 20 GO biological processes linked with gene sets negatively correlated with the first gradient of 39h percentage change, and (I) shows the gene enrichment network. (J) Cell type enrichment analysis of the first gradient map. Colors indicate directions of association (red = positive, blue = negative). *, p < 0.05; #, false discovery rate (FDR) p < 0.05.

Association between fractional ALFF (fALFF) and 39h percentage change (PC) across cortex.

(A) The 39h PC across cortex based on Schaefer-400 atlas in patients with fMRI (n = 15). (B) The scatter plot of Pearson correlation between 39h PC in whole cohort (n = 96) and subgroup with fMRI (n = 15). (C) The fALFF across cortex based on Schaefer-400 atlas in patients with fMRI (n = 15). (D) The scatter plot of Pearson correlation between fALFF and 39h PC and across cortex based on Schaefer-400 atlas in patients with fMRI (n = 15). In B and D, the translucent area around the regression lines represents the 95% confidential interval for the regression estimate, with fit line (blue) for significant association. The blue histogram represents the distribution of variables in x-axis, whereas the red histogram represents the distribution of variables in y-axis.

Association between fALFF-39h percentage change (PC) mismatch and amyloid-β (Aβ) burden.

(A) fALFF-39h PC match (orange), mismatch1 (high activity with low clearance, light blue) and mismatch2 (low activity with high clearance, dark blue) regions. (B) Open-source 11C-PiB SUVR for Aβ PET imaging based on Schaefer-400 atlas. (C) Independent one-way ANOVA test of PiB SUVR among regions in match (orange), mismatch1 (light blue) and mismatch2 (dark blue) pattern. p values are calculated using FDR-controlled multiple comparison: two-stage linear step-up procedure of Benjamini, Krieger and Yekutieli. (D-F) The scatter plot of Spearman correlation between z score of fALFF and 39h PC in match (D), mismatch1 (E) and mismatch2 (F) regions. (G-I) The scatter plot of Spearman correlation between z score of fALFF and PiB SUVR in match (G), mismatch1 (H) and mismatch2 (I) regions. (J-L) The scatter plot of Spearman correlation between fALFF-39h PC mismatch index and PiB SUVR in match (J), mismatch1 (K) and mismatch2 (L) regions. In D-L, the translucent area around the regression lines represents the 95% confidential interval for the regression estimate, with fit line for significant association; r indicates the Spearman correlation coefficient, and partial r indicates the Spearman partial correlation coefficient when adjusting for z scores of fALFF and 39h PC. FDR, false discovery rate. *, p < 0.05. L, left; R, right.

Flowchart of patient selection.

fALFF-39h percentage change (PC) pattern based on different clustering methodology.

(A-B) The spatial location (A) and scatter plot (B) of clusters based on fALFF-39h PC mismatch index. (C-D) The spatial location (C) and scatter plot (D) of data-driven k-means clustering based on z-scores of fALFF and 39h PC. Orange, high neural activity and fast clearance; pink, high neural activity but slow clearance; purple, low neural activity and slow clearance; green, low neural activity but fast clearance.

Association between ReHo-39h percentage change (PC) mismatch and amyloid-β (Aβ) burden.

(A) fALFF-39h PC match (orange), mismatch1 (high activity with slow clearance, light blue) and mismatch2 (low activity with fast clearance, dark blue) regions. (B) ReHo-39h PC match (orange), mismatch1 (high activity with slow clearance, light blue) and mismatch2 (low activity with fast clearance, dark blue) regions. (C) The scatter plot of Pearson correlation between fALFF-39h PC mismatch index and ReHo-39h PC mismatch index. (D-F) The scatter plots of Spearman correlation between z score of ReHo and 39h PC in match (D), mismatch1 (E) and mismatch2 (F) regions. (G-I) The scatter plots of Spearman correlation between z score of ReHo and PiB SUVR in match (G), mismatch1 (H) and mismatch2 (I) regions. (J-L) The scatter plots of Spearman correlation between ReHo-39h PC mismatch index and PiB SUVR in match (J), mismatch1 (K) and mismatch2 (L) regions. In D-L, the translucent area around the regression lines represents the 95% confidential interval for the regression estimate, with fit line for significant association; r indicates the Spearman correlation coefficient, and partial r indicates the Spearman partial correlation coefficient when adjusting for z scores of ReHo and 39h PC. FDR, false discovery rate. L, left; R, right.