In vitro morphological profiling of T cells predicts clinical response to natalizumab therapy in patients with multiple sclerosis

Despite the efficacy of natalizumab, which targets the integrin VLA-4, in treating multiple sclerosis (MS), approximately 35% patients with MS present evidence of disease activity two years after treatment initiation. Individual heterogeneity of leukocyte response to VLA-4 on natalizumab-mediated blockade may underlie disparities in treatment efficacy. Here we use a high-content cell imaging (HCI) pipeline to profile the in vitro effects of natalizumab on VLA-4-stimulated PBMCs from MS patients prior to natalizumab treatment. Unsupervised clustering of image data partially discriminates non-responder MS patients based on morphology, F-actin organization and signaling-related features in CD8+ T cells. Furthermore, through a random forest approach, treatment response can be predicted with a performance of 92% for a discovery cohort and 88% for a validation cohort. Unfavorable treatment response is associated with a distinct actin remodeling response of natalizumab-exposed CD8+ T cells and a residual ability of these cells to spread on VCAM-1. Our study thus unveils that CD8+ T cells from individual MS patients display heterogeneous susceptibility to natalizumab in vitro and highlights the potential of HCI-based pretreatment monitoring to assist individualized treatment prescription. Targeting the integrin VLA-4 with natalizumab has shown efficacy in treating multiple sclerosis, but relapse still occurs in some patients. Here the authors use a high-content cell imaging (HCI) pipeline and machine learning to assess the morphology of T cells from MS patients prior to natalizumab regime for defining features associated with treatment responses.