Diffusion and contrast-enhancement MRI phenotypes associated with immune checkpoint inhibitor exposure and with immune cell infiltration in brain metastases.

Abstract

Background: MRI-derived apparent diffusion coefficient (ADC) and contrast-enhancement (CE) may represent non-invasive biomarkers to evaluate microstructural tissue changes and histopathological immune cell infiltration induced by immune checkpoint inhibitors (ICI).

Methods: 125 lesions across 93 patients were analyzed. ADC (reflecting water diffusivity) and CE T1-subtraction maps (reflecting blood-brain barrier permeability) tumor values were used for bivariate histograms of ADC and CE, and to quantify voxel fractions belonging to highADC-highCE and lowADC-lowCE quadrants. The association between ICI exposure and voxel frequency in ADC-CE quadrants was evaluated with multivariate mixed-effects models accounting for corticosteroid exposure and other variables. In a subset of patients (n=23), the association between immune cell counts (CD45+ cell density) from tissue samples and ADC-CE phenotypes was tested.

Results: ICI and corticosteroid exposure were associated with distinct ADC-CE phenotypes on bivariate histograms. ICI-exposed lesions showed significantly higher voxel frequency in the highADC-highCE quadrant (p<0.001) and lower voxel frequency in the lowADC-lowCE quadrant (p<0.01), compared to ICI-naïve lesions. Multivariate analyses confirmed that ICI exposure was an independent determinant of higher voxel frequency in the highADC-highCE quadrant (p=0.001) and lower voxel frequency in the lowADC-lowCE quadrant (p=0.01) in absence of corticosteroid treatment. Corticosteroid usage significantly influenced the relationship between ADC-CE phenotypes and ICI exposure. Voxel frequency in ADC-CE quadrants was correlated with CD45+ cell density in histopathology (partial ρ=0.56 p=0.01 for highADC-highCE, partial ρ=-0.64 p<0.01 for lowADC-lowCE).

Conclusions: ADC-CE phenotypes are associated with ICI exposure and immune cell infiltration in BM, representing potential non-invasive markers to monitor ICI-induced pathophysiological changes.