Longitudinal accumulation of glial activation measured by TSPO-PET predicts later brain atrophy in multiple sclerosis

Abstract

In multiple sclerosis (MS), accumulation of disability is driven by CNS-compartmentalized inflammation. This inflammatory process involves activated microglia and astrocytes, which contribute to neuroaxonal damage which in turn accelerates disease progression. Activated glial cells express 18-kDa translocator protein (TSPO), and TSPO-binding radioligands and positron emission tomography (PET) imaging can be used to quantitate glial activation in vivo. The aim of this study was to evaluate the longitudinal evolution of glial activation in untreated cohorts of relapsing remitting MS (RRMS) and secondary progressive MS (SPMS) patients over one-year follow-up, and to explore how a change in glial activation associates with later imaging and clinical outcomes. Eighteen untreated MS patients (RRMS n = 8, SPMS n = 10) were studied. Expanded disability status scale (EDSS), brain MRI and TSPO-PET scans using [¹¹C]PK11195 were performed at baseline and one year later. Distribution volume ratio (DVR) of [¹¹C]PK11195-binding, and the proportion of TSPO-high voxels at baseline in the normal appearing white matter (NAWM) and other regions of interest were compared to the respective parameters in follow-up scans. Chronic lesions were phenotyped at baseline and at follow-up according to their TSPO-PET-binding patterns, and TSPO-expressing lesions were further characterized using postmortem immunopathological staining. Extended follow-up was obtained after 4-11 years with EDSS available for 18 patients and MR imaging available from 13 patients. TSPO-signal was higher among SPMS compared to RRMS patients at baseline. During one-year follow-up, TSPO uptake remained stable in RRMS patients in all regions of interest. Among the SPMS patients, the proportion of active voxels in the NAWM increased significantly over one-year follow-up. A greater proportion of lesions acquired a rim-active phenotype among SPMS compared to RRMS. According to forward-type stepwise multiple linear regression, change in the proportion of active voxels in the NAWM over one year and baseline body-mass-index were best predictors of later development of brain atrophy (R2 = 0.69). Our study provides novel information about the natural evolution of CNS-compartmentalized inflammation and demonstrates an important link between NAWM TSPO-signal and later adverse outcomes among MS patients, supporting the notion that diffuse glial activation in the NAWM contributes to disease progression.

Keywords: Glia; Microglia; Multiple sclerosis; Positron emission tomography; TSPO.

Fig. 1

Correlations between TSPO-availability (DVR) and T1 lesion volume, thalamic volume and EDSS at baseline. Higher TSPO-binding in the NAWM, thalamus and at the perilesional area correlated significantly with higher T1 lesion volume and smaller thalamic volume at baseline (A-F). NAWM TSPO-binding correlated with EDSS (G). Spearman correlation was used for statistical analyses

Fig. 2

Examples of chronic lesion phenotypes based on TSPO-PET and postmortem neuropathological immunohistochemical staining. A) An automated TSPO-PET-based method classifies lesions into rim-active, overall-active, and inactive categories based on the proportion of active voxels in the perilesional rim relative to the lesion core. B) The upper panel provides an overview of a mixed active/inactive lesion with a complete loss of MBP-positive myelin sheaths and a prominent rim of HLA-DR and TSPO positive microglia, likely corresponding to a TSPO-rim-active lesion. The lower panel shows higher magnifications of the rim stained for TSPO, HLA-DR and GFAP. GFAP staining shows strong fiber gliosis inside and outside of the lesion and single GFAP positive astrocytes at the rim. Turnbull staining demonstrates the presence of iron positive myeloid cells at the rim. C) The upper panel displays a completely demyelinated active lesion with a dense myeloid cell infiltrate (immunohistochemistry for MBP and HLA-DR). Within the active lesion, numerous TSPO positive cells are present. The lower panel shows higher magnifications of the immunohistochemical stainings demonstrating the presence of TSPO-expressing myeloid cells (black arrows) and astrocytes (white arrows). D) The upper panel depicts a completely demyelinated hypocellular inactive lesions with low numbers of HLA-DR positive myeloid cells that shows some TSPO-immunopositivity. Higher magnifications in the lower panel suggest that also GFAP-positive astrocytes express TSPO in inactive lesion. The inactive lesion shows a strong GFAP positive fiber gliosis, GFAP positive astrocytes are indicated by black arrows

Fig. 3

Longitudinal changes in TSPO-binding among RRMS and SPMS patients over the one-year follow-up. No significant change was observed in NAWM DVR-values during the one-year follow-up among RRMS or SPMS patients (A). Among SPMS patients the proportion of active voxels increased significantly in the NAWM, which indicates an enhancement in glial activation during the observation period (B). SPMS patients had significantly more lesions that evolved into rim-active phenotype during the one-year follow-up compared to RRMS (C). Statistical analyses were performed with Wilcoxon signed-rank test (A-B) and Wilcoxon rank-sum test (C)

Fig. 4

Baseline TSPO-binding associates with later clinical progression. Higher DVR at the perilesional area at baseline associated with greater EDSS increase during both one-year (C) and the extended 4–11-year follow-up periods (G). The associations between a greater baseline NAWM DVR and EDSS increase during either follow-up period were not significant (A and E). Patients with confirmed disability progression during the one-year follow-up had greater TSPO-binding (DVR) at the perilesional area but not in the NAWM compared to stable patients (B and D). Patients with confirmed disability progression during the extended 4–11-year follow-up had greater TSPO-binding (DVR) in the NAWM but not at the perilesional area compared to stable patients (F and H). Statistical analyses were performed with Spearman correlation and Wilcoxon rank-sum test

Fig. 5

A greater increase in TSPO-binding during the one-year follow-up associates with more pronounced brain atrophy outcome over 4–11 years. A greater increase in the TSPO-binding in the NAWM and whole brain during the one-year follow-up associate with later brain atrophy measured as annualized PBVC (A-D). According to multiple linear regression modelling the increase in active voxels in the NAWM and baseline BMI explain 69% of the variance in later brain atrophy (E). Statistical analyses were performed with Spearman correlation (A-D) and multiple linear regression (E)