Primary CNS Lymphomas (PCNSL) Patient-Derived Xenograft (PDX) Models Capture the Biological and Molecular Characteristics of the Disease

INTRODUCTION

PCNSL are a rare form of diffuse large B-cell lymphoma (DLBCL) with a pejorative prognosis. Alike nodal DLBCL, PCNSL is a heterogeneous disease with molecular heterogeneity and different responses to standard treatments and evolution. Unlike their nodal counterpart, the role of the brain tumor environment appears to be involved in the mechanisms underlying the cerebral tropism of these lymphomas, and the effectiveness of treatment also depends on their bioavailability in the brain parenchyma. Findings regarding nodal lymphoma, and responses to innovative drugs cannot be transposed directly in clinical practice in PCNSL patients. Preclinical models that reflect the molecular characteristics and functional heterogeneity of PCNSL are needed. No PCNSL PDX models have been published to date, and no primary cell line of PCNSL is commercially available.

We present the development of human PCNSL-PDX models.

METHODS

PCNSL patients specimen (biopsy samples: n = 3; cerebrospinal fluid (CSF): n = 4; at time of diagnosis: n = 3; during treatment: n = 1; at relapse: n = 3) were obtained from 2 French centers with written informed patient consent. Immunodeficients Swiss nude female mice were used. Before implantation, cells from CSF underwent centrifugation and cell counts, and tumor samples were mechanically dissociated. The tumoral samples were then grafted by stereotaxic surgery in the caudate nucleus of mice. The characterization of the PDX included immunohistochemistry, mutational profiles, RNA-seq, VDJ recombination, and plasmatic measurement of human interleukin-10 (IL-10).

RESULTS

Tumoral graft was successful in 5 cases out of 7 (71%) and created a diffuse infiltration of the brain parenchyma in 30 to 90 days. The human cell load, quantified in the brain of mice by qPCR, was approximately 30 % at the ethical endpoint. The EBV status was negative in the patient samples and in the PDX models. We have performed 3 paired (tumor and PDX) NGS analysis of a targeted panel of approximately 100 genes found recurrently mutated in PCNSL / DLBCL. In addition, we have also analyzed the transcriptome profile of paired tumor/PDX by RNA-seq analysis. Interestingly, PCNSL's PDX recapitulate the mutational and transcriptional landscape of this rare entity. Stability of the mutational profile was observed through successive tumoral passages. Samples obtained from the PDX models were frozen in nitrogen liquid and were successfully grafted into mice at least 3 months later. Human IL-10 was measurable in the plasma of mice with CNS PDX. Correlation between IL-10 levels and tumoral mass is being evaluated. In addition, the B-cell origin and clonality assessed by the rearrangement of the heavy chain immunoglobulin locus is undergoing.

CONCLUSIONS

We established a panel of 5 human PCNSL-PDX models that capture the histological and molecular characterization of the disease. These models can also be frozen for long term storage while sparing a living animal maintenance. Additional models will be generated with this approach in order to constitute a bank of the different molecular profiles of PCNSL encountered in patients. These models will be available for preclinical studies as well as for cognitive research.