Damokosh, Andrew I.

Abstract Introduction: We investigated tumor gene expression profiling (PGx) to identify predictive classifiers of response and elucidate the mechanism of efficacy of proteasome inhibition in multiple myeloma (MM) by bortezomib, a first-in-class proteasome inhibitor. Analysis of 44 samples from the SUMMIT phase 2 trial identified predictive classifiers of response in patients (pts) with refractory MM (ASH 2002, Abstr 1519). We now report on the PGx analysis of the SUMMIT-derived classifier in an independent sample set from the APEX phase 3 trial, a multicenter randomized study of 669 pts that demonstrated the superiority of bortezomib to high-dose dex for relapsed MM following 1 to 3 prior therapies (Richardson NEJM 2005). As well, we report on a new classifier of response built by combining samples from SUMMIT and MLN34101-040 (040) (a study of bortezomib in pts who failed dex in APEX or were not eligible for APEX) and tested in the APEX samples. Methods: 93 participating clinical centers collected pretreatment bone marrow aspirates and enriched tumor cells on site before centralized RNA extraction and microarray analysis. Approximately 85% of pts consented to whole-genome analysis. Quality control for RNA integrity and quantity, microarray data quality, and tumor cell purity resulted in 129 patient samples from APEX evaluable for PGx; 67 were from the bortezomib arm and 62 from the dex arm. 48 samples were evaluable from 040. Patient characteristics of this subset were similar to those of the overall study population. Predictive classifiers were identified and tested via separate training and test datasets. Results: Consistent with the SUMMIT analysis, the APEX and 040 datasets exhibit biologic features expected of MM, including the overexpression of translocation target genes such as cyclin D1, MAF, and MMSET/FGFR3. Classifiers built using only the SUMMIT samples predict the outcome of pts in the APEX bortezomib arm with an overall accuracy of 51% (sensitivity [Sn] = 49%, specificity [Sp] = 55%; P = .81) and in the dex arm with an overall accuracy of 56% (Sn = 64%, Sp = 50%; P = .31). If the classifier was built using data from both the SUMMIT and 040 trials (87 total samples), the prediction of response in the APEX bortezomib arm was statistically significant (66% accuracy, Sn = 89%, Sp = 42%; P = .006). The same classifier only predicts the response of dex-treated pts with an overall accuracy of 46% (Sn = 72%, Sp = 28%; P = .88), suggesting specificity for bortezomib. Many of the overexpressed genes associated with bortezomib insensitivity relate to protein synthesis (EIF4B, RPL5) and RNA binding (SFRS10, CPSF6). Future Directions: This work highlights practical and scientific issues that inform future efforts to implement PGx in clinical trials. Predictive accuracy may be influenced by the clinical end point, limited sample size, and/or technical considerations associated with distinct trials and genomic data. The influence of study differences (relating to pts’ time from diagnosis, number of prior therapies, molecular subtypes, or hybridization characteristics) on the ability to build classifiers and identify biologic pathways related to response are being examined; these and potential insights on drug mechanisms will be presented.

Abstract Purpose: Women with advanced epithelial ovarian cancer are routinely treated with platinum-paclitaxel chemotherapy following cytoreductive surgery, yet only ∼20% achieve long-term disease-free survival. We hypothesized that differences in gene expression before treatment could distinguish patients with short versus long time to recurrence after administration of platinum-paclitaxel combination chemotherapy. Experimental Design: To test this hypothesis, gene expression profiling of 79 primary surgically resected tumors from women with advanced-stage, high-grade epithelial ovarian cancer was done using cDNA microarrays containing 30,721 genes. Supervised learning algorithms were applied in an effort to develop a binary classifier that could discriminate women at risk for early (≤21 months) versus late (>21 months) relapse after initial chemotherapy. Results: A 14-gene predictive model was developed using a set of training samples (n = 51) and subsequently tested using an independent set of test samples (n = 28). This model correctly predicted the outcome of 24 of the 28 test samples (86% accuracy) with 95% positive predictive value for early relapse. Conclusions: Predictive markers for early recurrence can be identified for platinum-paclitaxel combination chemotherapy in primary ovarian carcinoma. The proposed 14-gene model requires further validation.