Poster Session I: Lymphoma biology & translational research
With an expanding array of novel treatment options in follicular lymphoma (FL), we largely lack effective biomarkers to predict response in this molecularly heterogeneous disease. Activating mutations focused on components of the nutrient-sensing arm of the mTORC1 pathway (RRAGC, ATP6V1B2, ATP6AP1), are particularly enriched in FL, occurring in ˜30% of cases. Meanwhile, clinical mTOR inhibitors (mTORi) have shown some promise, with response rates of 50-60% in relapsed/refractory FL.
We therefore sought to determine whether gene mutation status - particularly in genes regulating the mTOR pathway - may represent biomarkers of mTORi-response in FL.
We retrospectively analyzed pre-treatment tumor samples from relapsed/refractory FL patients treated on two phase II clinical trials with single agent mTOR inhibitors, everolimus (NCT00436618) or temsirolimus (NCT00290472), with well characterized clinical outcomes. In total, 21 patients had available biopsies, consisting of 15 everolimus- and 6 temsirolimus-treated cases.
Tumor DNA extracted from formalin fixed paraffin embedded (FFPE) tissue was subjected to targeted sequencing using a 22 gene panel comprising genes recurrently mutated in FL, and PI3K/AKT/mTOR pathway genes mutated in germinal center lymphomas. Libraries were prepared using capture-based target enrichment (Nonacus), and sequenced on the NextSeq550 (Illumina) with a mean coverage depth of x700. A stringent bioinformatic pipeline was adopted with variants called with VarScan2, and a 10% variant allele frequency (VAF) threshold was employed to filter out sequencing artefacts due to the FFPE material. 100% (15/15) of a selection of variants were validated by Sanger sequencing. Staining for pS6 was undertaken on tissue slides to examine downstream mTOR activation by immunohistochemistry.
Overall response rate (CR and PR) was 48% (10/21). Mutational analysis showed no significant association between clinical response and mTORC1-pathway mutations (mut) taken individually or collectively; although, absolute numbers of mutations in these genes were low, including RRAGC (responder: 0/10 mut vs non-responder: 1/11 mut), ATP6AP1 (responder: 1/10 vs non-responder: 1/11), and ATP6V1B2 (responder: 1/10 vs non-responder 3/11). Notably, no difference in pS6 staining was seen between responders and non-responders.
However, we found mutations in the epigenetic regulator CREBBP to be enriched in the responder group (100% (10/10) responders vs 54.5% (6/11) non-responders, p = 0.04). This effect was more pronounced when considering only mutations within the catalytic histone acetyltransferase (HAT) domain of CREBBP, with HAT mutations seen in 100% (10/10) responders, and 27.3% (3/11) non-responders (p = 0.001).
Mutations in the mTOR pathway did not predict for response to mTOR inhibition in our study. The enrichment of CREBBP mutations in responders, and particularly those affecting the catalytic HAT domain, points to CREBBP as a potential biomarker of response which warrants validation in trials of drugs modulating the PI3K/AKT/mTOR axis. More broadly our data may lend support to a distinct clinical and biological behavior of CREBBP HAT domain-mutant disease that also warrants further evaluation. Taken together, our findings highlight the potential opportunities afforded by undertaking correlative studies, and underlines the need to incorporate prospective biomarker discovery into clinical trials in order to help focus future therapeutic strategies more rationally.