Pharmacogenomic (PGx) Analysis of Bortezomib-Associated Peripheral Neuropathy in the Phase 3 VISTA Trial of Bortezomib Plus Melphalan–Prednisone Versus Melphalan–Prednisone in Multiple Myeloma.

Abstract 3875

Poster Board III-811

We investigated genotyping of 172 candidate gene loci (2016 SNPs) related to hereditary neuropathy, energy production and fast axonal transport, nociception and pain transmission, mitochondria, neurogenesis, neuroprotection, immune function, and bortezomib (Velcade^®) mechanism of action. Our aim was to identify predictive classifiers for peripheral neuropathy (PN) in multiple myeloma (MM) following treatment with bortezomib in the phase 3 VISTA trial of bortezomib plus melphalan–prednisone (VMP, N=344) vs MP (N=338) in previously untreated MM patients ineligible for high-dose therapy (median age 71 years). We collected whole blood samples from 323 patients who provided consent for this analysis, from 99/151 participating clinical centers. Samples were transferred to a central laboratory for DNA extraction. After quality control for DNA integrity and quantity, DNA from 139 patients on the VMP arm (VISTA PGx subset) was included in this case-control candidate gene study. A custom array was developed by Illumina (San Diego, CA) for this analysis. Patient characteristics of the VISTA PGx subset were similar to those of the overall population. Of the PGx subset, 51.8% had reported PN events (36.0% grade ≥2). We sought to confirm VISTA findings in an independent cohort of patients in the IFM2005-01 phase 3 study of bortezomib–dexamethasone (Vel/Dex) vs VAD in previously untreated MM patients aged ≤65 years eligible for high-dose therapy (median age 57 years); 215 samples were collected from consenting patients on the Vel/Dex arm, of whom 46% had reported PN events (23.7% grade ≥2) including peripheral neuropathies, dysesthesia, and paresthesia. The VISTA analysis included cases and matched controls with result correlations to clinical adverse event descriptions of “peripheral sensory neuropathy”, “peripheral neuropathy NEC”, and/or “neuralgia”, or any occurrence of these three adverse events. Multiple covariates were included in single-marker association analyses using logistic regression in SAS (PROC LOGISTIC, SAS, v9.1) under three different genotypic models (additive, dominant, recessive) followed by FDR correction. No SNPs had significant associations with the terms above. However, analysis for time to onset of PN using Cox regression and the log rank test in SAS (PROC PHREG PROC TPHREG and PROC LIFETEST, SAS v9.1) using the same models and multiple testing corrections methods identified two SNPs significantly associated with shorter time to onset of any PN (rs4553808; CTLA4; FDR=0.002) and time to onset of grade ≥2 PN (rs1474642; PSMB1; FDR =0.014). Both of these associations were found in the recessive model so that 8.3% of patients with PN following bortezomib treatment have the homozygous recessive genotype of CTLA4 and 6% of patients with grade ≥2 PN have the recessive PSMB1 genotype. Both of these associations were also significant under Bonferroni's correction method. Two additional SNPs were identified that had significant association with grade ≥2 PN (rs12568757; CTSS, FDR=0.027) and with grade ≥3 PN (rs11974610; GJE1; FDR=0.041) but these were not significant under Bonferroni's method. These results suggest an immune-related influence for PN development involving genes associated with immune function (CTLA4, CTSS) and reflexive coupling within Schwann cells (GJE1). Mutations in the target (PSMB1) subunit suggest an influence with drug binding as well. When we investigated associations with these SNPs in the IFM2005-01 dataset, no significant associations were found in the analysis of time to onset of PN following multiple testing corrections; however, rs4553808 (CTLA4) had the same trend of association with time to onset of grade ≥2 PN in the recessive model (p=0.105). The differences in association were likely due to differences between the VISTA and IFM2005-01 study populations (e.g. neuropathy status at entry, rate of onset of PN, age, and baseline characteristics including β₂-microglobulin, albumin, and diabetes status). This is the first report of an association between genes associated with immune function (e.g. CTLA4) and time to onset of bortezomib-associated PN. Although the clinical utility of this study is limited due to the low number of genes investigated and the exploratory nature of this analysis, further genome-wide studies may result in a predictive classifier for bortezomib-associated PN applicable to the clinic. This work is currently ongoing.