Background

CNS manifestations of aggressive non-Hodgkin Lymphoma are associated with serious morbidity and adverse prognosis. Primary CNS lymphomas (PCNSL) exhibit a dichomatous growth pattern, either dissemination within brain, typical at presentation, and/or leptomeningeal spread, common at relapse. Elucidation of the mechanistic basis of CNS lymphoma progression as well as drug resistance requires preclinical models that recapitulate their pathogenesis.

Methods

We developed a novel method to derive cell lines of CNS lymphoma that recapitulate disease phenotypes upon intracranial implantation into mice. We are applying genomics, in vitrochemotaxis, preclinical testing of targeted therapies and neuroimaging to evaluate mechanisms of invasion and resistance.

Results

We developed 7 CNS lymphoma cell lines; 6 DLBCL (all ABC-type), 1 Burkitt; 5 from secondary CNS lymphoma (SCNSL), and 2 from PCNSL, of which 1 was treatment naïve. Intracranial implantation of lymphoma cells from these tumours within NSG mice provides a reproducible model to dissect the pathogenesis of CNS lymphomas. PCNSL specimens were 10X more efficient in CNS dissemination than SCNSL. High resolution array-CGH demonstrated that intracranial tumour growth was associated with retention of genomic aberrations of the original tumours (e.g. del 6q, gains on 12, etc) and that these were maintained with serial passage in vivo. CNS-infiltrative lymphomas expressed significantly increased levels of MMP-7 and RGS-13 transcripts compared to lymphomas that did not infiltrate brain, while osteopontin and cathepsin D expression by lymphoma cells did not correlate with CNS invasion. Targeted shRNA-mediated knockdown of RGS-13 was performed using lentiviral infection and resulted in significant delay of CNS lymphoma growth in vivo in a xenograft model but had no effect on lymphoma proliferation in culture. Therapeutic response to lenalidomide, minus and plus rituximab, was recapitulated in RAG-/- mice, despite deficient T-cell function and correlated with baseline relative cereblon expression, as quantified using a highly specific immunohistochemical assay. The emergence of resistance to lenalidomide in human CNS lymphoma xenografts also correlated with loss of cereblon protein expression, supporting a role for cereblon in the efficacy of lenalidomide in CNS lymphomas. Notably, significant cereblon protein expression by lymphoma cells was detected by immunohistochemistry in 12/22 diagnostic specimens of aggressive CNS lymphoma. Metabolic imaging of model CNS lymphomas using magnetic resonance spectroscopy demonstrated significant intratumoural lactate production in the microenvironment, detectable before evidence of aberrant T2 signal and reduced diffusion. Lenalidomide reduced tumour expression of lactate dehydrogenase and lactate, as well as RGS-13, consistent with anti-proliferative as well as anti-invasive effects.

Conclusions

To the best of our knowledge we have developed the first panel of patient-derived CNS lymphoma cell lines. We have used these to generate intracranial xenografts that provide a highly reproducible model system to dissect key elements of CNS lymphoma pathogenesis, leading to the elucidation of mechanisms of CNS lymphoma growth and invasion as well as resistance. Our results support a direct, T-cell independent effect of lenalidomide on CNS lymphoma growth and invasion which may be cereblon-dependent. Additional studies are needed to define the role of cereblon as a biomarker and mediator of lenalidomide efficacy in CNS lymphomas. In addition, we are using these models to identify novel genomic and metabolic aberrations predictive of early resistance to lenalidomide and other targeted therapies. Supported by the Lymphoma Research Foundation, Leukemia and Lymphoma Society, and by NIH R01CA139-83-01A1.

Disclosures:

Heise:Celgene: Employment, Equity Ownership. Rubenstein:Celgene: Research Funding; Genentech: Research Funding. Off Label Use: use of lenalidomide in CNS lymphoma.

Author notes

*

Asterisk with author names denotes non-ASH members.

Sign in via your Institution