Application of quantitative DTI metrics in sporadic CJD

Abstract

Diffusion Weighted Imaging is extremely important for the diagnosis of probable sporadic Jakob-Creutzfeldt disease, the most common human prion disease. Although visual assessment of DWI MRI is critical diagnostically, a more objective, quantifiable approach might more precisely identify the precise pattern of brain involvement. Furthermore, a quantitative, systematic tracking of MRI changes occurring over time might provide insights regarding the underlying histopathological mechanisms of human prion disease and provide information useful for clinical trials. The purposes of this study were: 1) to describe quantitatively the average cross-sectional pattern of reduced mean diffusivity, fractional anisotropy, atrophy and T1 relaxation in the gray matter (GM) in sporadic Jakob-Creutzfeldt disease, 2) to study changes in mean diffusivity and atrophy over time and 3) to explore their relationship with clinical scales. Twenty-six sporadic Jakob-Creutzfeldt disease and nine control subjects had MRIs on the same scanner; seven sCJD subjects had a second scan after approximately two months. Cortical and subcortical gray matter regions were parcellated with Freesurfer. Average cortical thickness (or subcortical volume), T1-relaxiation and mean diffusivity from co-registered diffusion maps were calculated in each region for each subject. Quantitatively on cross-sectional analysis, certain brain regions were preferentially affected by reduced mean diffusivity (parietal, temporal lobes, posterior cingulate, thalamus and deep nuclei), but with relative sparing of the frontal and occipital lobes. Serial imaging, surprisingly showed that mean diffusivity did not have a linear or unidirectional reduction over time, but tended to decrease initially and then reverse and increase towards normalization. Furthermore, there was a strong correlation between worsening of patient clinical function (based on modified Barthel score) and increasing mean diffusivity.

Keywords: Atrophy; Creutzfeldt–Jakob disease; DWI; MD; Prion disease; sCJD.

Fig. 1

Average mean diffusivity (MD) cross-sectionally in sCJD cohort versus controls per brain volumes of interest (VOIs) (after correction for multiple comparisons (FDR)). The colored areas are regions with significantly reduced (or a strong trend towards) mean MD in sCJD versus controls superposed on the Montreal Neurological Institute (MNI) atlas. White VOIs have a trend towards reduced average mean MD with p-values ≥ 0.05 and < 0.1. Teal VOIs have significantly reduced average MD with p values ≥ 0.01 and < 0.05. Blue VOIs have significantly reduced average MD with p values ≥ 0.001 and < 0.01. Sporadic CJD had VOIs with reduced average MD compared to controls primarily in temporal and parietal lobes and deep nuclei. No VOI had increased MD in sCJD compared to controls. Orientation is radiologic (right brain is left side of image).

Fig. 2

Inverse correlation of change in Barthel and change in MD over two months. Distribution of Spearman Rank Correlation Coefficients of change in the modified Barthel and average MD over two months for all cortical VOIs. The graph shows the number of cortical VOIs (y axis) per each Spearman Rank Correlation Coefficient (x axis: values from 1 to − 1). The green dotted line shows the limit for statistically significant values without FDR; the red dash line shows the limit for statistically significant values with FDR. At whole brain, hemispheric, lobar and the VOI level, all areas (except the right parahippocampus) showed an inverse (negative) relationship between the change in the (delta) modified Barthel and the change in the (delta) average MD with 15 VOIs (13 cortical and 2 subcortical; outside red dotted line) reaching statistical significance with FDR (p < 0.05)—as modified Barthel decreased/worsened, MD increased. For many brain VOIs, there was at least a trend towards increased MD with worsening modified Barthel.

Fig. 3

Cross-sectional and longitudinal (change) in Mean Diffusivity in sCJD vs. Barthel score. (A–B) Cross-sectional data. Using splines, we estimated group-wise average MD values at (A) whole brain and (B) at lobar/region level (frontal, limbic, temporal, parietal, occipital, deep nuclei), as a function of baseline modified Barthel scores for the 26 sCJD subjects. Circles in the figures represent average of MD values within each lobe/region for each subject studied either cross-sectionally (A–B) or longitudinally (C–D). In A, we show the 95% confidence interval (black solid lines) and estimate of average tendency in MD (red dotted line). In B, we show estimates of average tendency in MD at lobar levels (colored solid lines). (C–D) Longitudinal data. The average change over 2 months in MD in seven sCJD subjects in (C) whole brain and (D) in each lobes/region independently (frontal, limbic, temporal, parietal, occipital, deep nuclei) as a function of baseline Barthel score was estimated using splines. In C, we show the 95% confidence interval (black solid lines) and estimate of average tendency in change of MD (red dotted line). In D, we show estimates of average tendency in change of MD at the lobar level (colored solid lines). We obtained an identical graph to Fig. 3A even by excluding 7 patients with longitudinal scans (not shown). Deep nuclei = thalamus, pallidum, caudate and putamen; limbic lobe = cingulate cortex and insula.

Supplemental Fig. 1.

Average longitudinal mean diffusivity change in 7 sCJD subjects. Each subject was scanned twice, approximately 2 months apart. VOIs with reduced averaged mean MD (Z-scores) in the second compared to the first MRI are shown in blue colors; VOIs with increased averaged mean MD are shown in orange-red colors. In most VOIs, there was increased averaged MD between the serial MRIs. MD decreased between serial MRIs only in the left pars opercularis, left bankssts, right rostral middle frontal, right superior frontal, frontalpole and temporal pole. None of the averaged MD changes, in either direction, however, were statistically significant. Average decline (reported as mean ± SD) in the modified Barthel score was 8.6 ± 13.4 points (from average 76.4 ± 23.0 to 67.9 ± 32.5); average decline for MMSE was 1.9 ± 4.0 points (from average 14.4 ± 11.0 to 12.6 ± 11.5; average change (worsening) for the NPI was 5.7 ± 9.6 points (from average 19.4 ± 8.7 to 25.1 ± 16.8). Orientation is radiologic.