Targeted Treatments: Pre-Clinical Evaluation Of Efficacy By Tracking Clonal Diversity In CLL Xenograft Models

Progressive CLL with DNA damage response defects caused by ATM or TP53 inactivation is still a therapeutic problem. There is currently a growing appreciation of the subclonal complexity in this type of leukaemia, mostly supported by whole genome analyses of CLL samples such as SNP arrays and next generation sequencing, both highlighting the role of subclonal variations in progression and therapeutic response. Although revealing, these analyses do not inform about the clonal make-up of individual proliferating cells and such information might be necessary to determine effectiveness of novel targeted treatments. In order to determine and follow the clonal evolutionary millieu in individual CLL cases we developed multi-colour fluorescence in situ hybridization (MCF) to analyse CLL samples at the single cell level. We screened 134 samples for the presence of 11q and 17p deletions of which 25 were identified with one of these cytogenetic defects and these were screened alongside 28 samples with normal 17p and 11q loci for 13q and 6q deletion as well as trisomy 12. We found that all but four of the samples with 11q or 17p deletions had at least one other genomic abnormality, whilst only two of the 28 samples with normal 17p and 11q loci harboured two genomic abnormalities. We subsequently performed MCF on the 27 samples with multiple genetic abnormalities and generated evolutionary trees for each of the samples. Two types of clonal evolution were identified: linear and branched, with the latter being the more common. We were able to analyse eleven of these samples post-treatment and found that whilst the clonality of some samples was largely unaffected by treatment, others showed treatment-induced differences in the subclonal make-up. Furthermore, some samples also exhibited signs of evolution with the generation of novel subclones upon treatment. Pre-clinical testing of novel therapeutic agents in xenograft models requires that the subclonal architecture of engrafted samples is representative of the donor sample. To this end, modifications were made to two primary CLL xenograft models. Firstly, samples from three different CLLs with a complex karyotype and multiple subclones were engrafted with autologous ex-vivo stimulated T-cells into NOG mice. All CLL subclones displayed a capacity to engraft and proliferate in this xenograft model. Furthermore, the MCF protocol was implemented to assess subclonality in vivo . Mice were randomised to Rituximab or control saline treatment three times over a period of five days. Analysis of engrafted cells in the spleen a week later displayed a good response to Rituximab with a significantly lower number of hCD45 + CD19 + CD5 + cells. However, upon isolation of CLL cells by FACS and assessment of clonal architecture using MCF it was shown that Rituximab treatment affected all subclones with a differential subclonal response. This resulted in a greater proportion of the CLL subclones with the greatest genetic complexity harbouring both 11q and 6q deletions still remaining. Finally, to recapitulate patient response to therapy, T-cell depleted, pre-treatment PBMC from a patient poorly responsive to bendamustine + Rituximab were engrafted in humanised mice. As in the patient, bi-weekly therapy for 3 weeks resulted in a limited, poor response to therapy with a corresponding small reduction of CFSE-labelled CLL PBMC and hCD45+ CD19+ cells. We conclude that the majority of CLLs display a branched pattern of evolution and that the subclone dynamic is an important determinant of CLL proliferation and response to treatment. We suggest that novel targeted therapies should be tested in the context of their ability to eradicate CLL subclones with the highest proliferative capacity, as these subclones are most likely to evolve. The development of an MCF protocol, in combination with the xenograft model, provides a powerful tool to help predict overall and subclonal responses to therapy in the patient. Disclosures: No relevant conflicts of interest to declare.