Gene Expression Profiling for Molecular Features of Response in a Phase I Trial of Alisertib Plus Romidepsin for Relapsed/Refractory Aggressive B- and T-Cell Lymphomas

Background: The histone deacetylase inhibitor (HDACi) romidepsin (Coiffier 2010) and other HDACi such as belinostat (O'Connor 2015) are management options for relapsed peripheral T-cell lymphoma (PTCL), and HDACi have also shown benefit for B-cell lymphomas (Younes 2012). The aurora A kinase (AURKA) inhibitor alisertib has shown promising results in aggressive lymphomas including Burkitt lymphoma (BL), diffuse large B cell lymphoma (DLBCL), and PTCL (Friedberg 2011, Barr 2014), and a registration trial in PTCL has recently completed enrollment. Given preclinical evidence for synergy when combining an AURKA inhibitor plus an HDACi (Zullo 2015, Muscal 2011, Kretzner 2008), we initiated a Phase I trial (NCT01897012, supported by CTEP-NCI) combining alisertib plus romidepsin in patients with multiple lymphoma subtypes.

Methods: 12 patients have been enrolled to date, 3 at each of 4 escalation dose levels with patients treated with alisertib on days 1 to 7 and romidepsin on days 1 to 8 of 21 day cycles. Of the eligible histologies there was 1 patient with BL, 3 with DLBCL and FISH positive for translocations of MYC and BCL2 (double-hit lymphoma, DHL), 4 with DLBCL with other unfavorable features (FISH-positive for c-myc alone; 90% Ki-67 positive; high protein staining for MYC and BCL2, but not FISH-positive; and transformed from follicular lymphoma), 3 with PTCL, and 1 with composite PTCL and DLBCL (Fanale 2014). Tumor core biopsies were performed at baseline and at the end of 1 cycle of therapy, part of which was collected in RNAlater for gene expression profiling (GEP) using Illumina HT12v4.0 arrays. Whole peripheral blood was also collected at these and other time points in PAXgene tubes for future GEP.

Results: Dose escalation and safety data have been previously presented (Fanale 2014). Responses to date are CR (PTCL, dose level 1), SD (PTCL, dose level 3, composite PTCL/DLBCL, dose level 4), PD (3 DHL, 1 HG DLBCL, 1 DLBCL with c-Myc, 1 PTCL). 4 of the patients with PD have died from continued refractory disease and 1 has been transitioned to hospice. The CR patient received 7 prior lines of treatment and had a duration of remission of 10 month. The PTCL patient with SD underwent an allogeneic stem cell transplant (SCT) but developed further PD 5 months after completion of SCT.

Technically-satisfactory GEP results were obtained for 9 pairs of baseline (pre-Rx) and post-cycle 1 (post-Rx) tumor biopsies. Hierarchical clustering of highly-variant genes across all samples, including additional samples from 2 patients (pre-Rx only), showed appropriate grouping of samples according to B- or T-cell lineage. For all but one sample pair, pre-Rx and post-Rx samples from the same patient clustered as nearest neighbors, indicating that treatment had less effect on GEP data than did patient origin, and that "subtraction" (comparing post-Rx to pre-Rx samples) would be necessary for comparing treatment effects. To identify treatment-induced GEP changes associated with response to alisertib plus romidepsin, log2-transformed and subtracted data for the lone patient (with PTCL) who reached CR were compared to those for the other 8 pairs. There were 160 gene probes that were upregulated by log2 >=1.5 in the lone responder, but by >=0.75 in 2 or fewer of the other patients; similarly, 163 probes were downregulated by log2 <=-1.5 in the lone responder, but by <=-0.75 in 2 or fewer of the other patients. The hypergeometric distribution test showed that upregulated genes were enriched with high significance for genes involved in cell cycle progression, including both AURKA and AURKB (more than 4-fold); interpretation is uncertain, but could indicate compensation for functional inhibition of aurora kinases. Individual, uniquely-upregulated genes suggested induction of an anti-tumor immune response: >5-fold induction of TMSB15A, repressed by TGF-beta, and BATF3, essential for CD8a+ dendritic cells in animal models of anti-tumor immune response; and >3-fold induction of TNFRSF18 (GITR), upregulated by T-cell activation.

Conclusions: Based on toxicity and response profile thus far, an amendment is under review to modify dosing schema for alisertib and romidepsin, and once the MTD is reached a potential cohort expansion for PTCL patients is planned. GEP changes suggest that induction of an antitumor response may underlie clinical responses, and will be investigated further by other studies of tumor biopsies and whole-blood GEP.