Pembrolizumab for the treatment of progressive multifocal leukoencephalopathy following anti-CD19 CAR-T therapy: a case report

Abstract

Progressive multifocal leukoencephalopathy (PML) is an opportunistic brain infection with few treatment options and poor survival when reversal of the underlying immune dysfunction is not achievable. JC polyomavirus reactivation resulting in PML can rarely complicate chimeric antigen receptor T-cell (CAR-T) therapy. We describe successful treatment of PML with Programmed death-1 (PD-1) blockade using pembrolizumab, 4 months following axicabtagene ciloleucel. Radiological features of immune reconstitution inflammatory syndrome without clinical deterioration were seen. Evidence of anti-viral immune reconstitution by in vitro detection of JC-specific T-cells and sustained neurological recovery in this patient suggest PD-1 blockade may be an effective treatment approach for PML post-CAR-T.

Keywords: checkpoint inhibition; chimeric antigen receptor T‐ cell; herald lesion; immune reconstitution inflammatory syndrome; pembrolizumab; progressive multifocal leukoencephalopathy.

FIGURE 1

(A) Magnetic resonance imaging of the brain. Axial, gadolinium‐enhanced images showing T2‐weighted (T2W), diffusion‐weighted (DWI) and contrast‐enhanced (T1W+C) sections at 1st episode of immune effector cell‐mediated neurological syndrome (ICANS; A–C), progressive multifocal leukoencephalopathy (PML) diagnosis (D–F), 1 month following 1st pembrolizumab dose (D1; G–I), and after 3rd dose (D3) of pembrolizumab (J–L), which was 5 months after 1st infusion. Panels A and B show a small focus of T2‐weighted signal abnormality in the right deep parietal white matter, demonstrating mild restriction of diffusion on DWI (B) at ICANS diagnosis. After the first pembrolizumab dose, note the relative increase in the size of PML lesions (G) with the progression of restricted diffusion (H) including new contralateral areas (red arrows) and new contrast enhancement (I), which had improved (J, K) or resolved (L) on interval imaging a month after D3. (B) Timeline of clinical and radiological events post CAR‐T infusion (x‐axis) with serial mini‐mental state examination (MMSE) scores (y‐axis). At presentation, MMSE score was reduced (25/30), declined further after 2 weeks (22/30), but improved over 4 months (to 29/30) following pembrolizumab infusion. Image references refer to (A)

FIGURE 2

(A–E) Histopathology of brain biopsy from the right parietal lobe. Subcortical white matter demonstrated the classic PML diagnostic triad of 1. ‘bizarre’ astrocytes ((A), inset i arrowhead, hematoxylin & eosin (H&E) stain) 2. enlarged oligodendrocyte nuclei (inset ii) which contained basophilic viral inclusions (arrowhead), and 3. demyelination ((B), SMI94 immunohistochemistry). The intense macrophage infiltrate ((D) CD68 immunohistochemistry) with relatively preserved axons (not shown) were suggestive of inflammatory demyelination. SV40 immunohistochemistry for JC virus (C) showed numerous intranuclear inclusions (arrowheads) also consistent with PML. Perivascular inflammation was visible on the H&E‐stained sections ((A), inset iii) and PD‐1‐positive cells were present in the intravascular and perivascular spaces ((E), PD‐1 immunohistochemistry), and throughout the parenchyma (not shown). PD‐1 positive cells made up 20% of total nucleated cells. Scale bars = 50 μm. (F, G) In vitro detection of JCV‐specific T cells. After the 3rd dose of pembrolizumab, following exposure to JCV antigen peptides, intracellular IFN‐γ secretion was detected in a small population of T‐cells by flow cytometry (1.85% [of live CD3+ cells] in the patient (G) compared to <0.1% in a healthy control (F)), suggesting a cellular anti‐JCV immune response post‐pembrolizumab. Briefly, peripheral blood mononuclear cells were pulsed with a JCV peptide pool of VP1, VP2, VP3, ST and LT (Miltenyi Biotec) each at a concentration of 1μg/ml, cultured in media supplemented with 5% human AB serum, and IL‐7 & IL‐15 at 10 ng/ml, over 7 days