Ovanome Executive Summary: Part 1
EXECUTIVE SUMMARY OF OVANOME
DNAP has achieved a milestone in the field of chemotherapy. On March 19, 2002, DNAPrint and University of Miami doctors presented the successful results of an ongoing pharmacogenomics study to the Society of Gynecological Oncology. The team identified a set of variant human genes that may be used to predict therapeutic outcomes for Paclitaxel (Taxol), one of the most commonly used chemotherapeutic agents for cancer. The product of this discovery is OVANOME, and it?s introduction is expected to boost the first-line paclitaxel response rate in chemo-niave ovarian cancer patients.
OVANOME is a pharmacogenomics product known as a classifier or chemo-predictive test. OVANOME will be used to predict the likelihood of response of ovarian cancer patients to therapy with Paclitaxel (Taxol). Patients that fail the first round of treatment have substantially lower survival statistics. By matching ovarian cancer patients with the drug most appropriate for their genetic constitution, OVANOME has the potential to save lives through pharmacogenomics testing. OVANOME demonstrated excellent sensitivity, specificity and predictive value in preliminary tests.
DNAPrint scientists used an innovative heuristic method with geometrical projection to identify and model proprietary ?eigengenotypes?' or population based vectors of genome information, that serve as features for variable Paclitaxel response. The team found that certain Paclitaxel response ?eigengenotypes? were strongly predictive for first-line chemotherapy response in Ovarian Cancer patients. Data was shown that suggests that by screening patient genomes for these ?eigengenotypes? prior to the commencement of chemotherapy, most non-responders could be flagged and re-directed towards altered doses and/or alternative chemotherapy more appropriate for their genetic constitution.
The clinical trial for OVANOME will be finished in early 2004. An application for approval to market, if required, will be submitted to the FDA by mid-2004 and approval obtained by early 2005. DNAP has a patent pending for compositions and methods relating to OVANOME. Sales of the OVANOME products to labs in 2005 will result in income of $4 million. This product launch will create a new, emerging US market of over $16 million. OVANOME is forecast to reach a peak-year sales record of $12 million in year three and hold steady for several years as the classifier test of choice in this market niche.
Abstract
Thirty five percent of Ovarian Cancer (OC) patients fail to respond to first-line combination paclitaxel (taxol) and carboplatin (TC) therapy. Because OC patients exhibit wide variability in the TC metabolism, it is possible that some or all of this variable response can be explained in pharmacogenetic terms. To determine whether common polymorphisms are associated with variable TC response, we applied novel analytics and data resources for a candidate gene survey of those genes most likely involved in paclitaxel disposition.
We genotyped 42 ovarian cancer patients (27 clinical responders and 15 non-responders) at 746 SNPs from 41 xenobiotic metabolism and 3 tubulin genes. Given the previous literature on paclitaxel metabolism, we were not surprised to identify haplotype alleles in Confidential (p<0.000) and Confidential (p=0.005) as significantly associated with variable TC response. However we also identified haplotype alleles in the Confidential (p=0.018), Confidential (p=0.022), and Confidential (p=0.035) genes to be significantly associated with outcome, and haplotype alleles in the Confidential (p=0.07) and Confidential (p=0.07) genes to be marginally associated. We applied linear and quadratic discriminate techniques to model these genetic features for predicting patient TC response.
Using an ?overall? clinical response criteria for evaluation of response over the treatment line, the efficiency of classification was 95%, the specificity of the responder classification was 96% and the negative predictive power of the classifier was 93%. We also developed a set of hierarchical rules using a direction-setting algorithm to capture both linear and non-linear effects and obtained similar results. Comparable results were obtained with either method using an ?average? rather than an ?overall? response criteria. These results confirm and extend previous results suggesting that the Confidential and Confidential family of Confidential genes are important determinants of variable paclitaxel response, but they also implicate the Confidential and Confidential family of genes as determinants as well.
Our results suggest that first-line TC response is largely a function of xenobiotic metabolism in OC patients, rather than tumor type or stage, and that OC patients may be pre-screened for xenobiotic metabolism gene sequences in order to individualize TC chemotherapy.
Table 1. Classification of Ovarian Cancer patients as responders or non responders (columns) to first line paclitaxel + carboplatin therapy relative to their actual clinical response (rows). Classification is made based on the presence or absence of haplotype alleles associated with clinical outcome. Inferred responders are considered haplotype (+), while Non Responders are considered haplotype (-). The rows represent actual responders and actual non-responders, whereas the columns represent classifications as haplotype (+) responders (Columns 2-3) or haplotype (-) non-responders (Columns 4-5). Cell entries represent probabilities, which were derived using the Overall and Average TC response criteria as indicated for each column (Materials and Methods). The column is read as follows: Responders (Row 1, Column 1) were correctly classified as responders using the Overall TC response criteria with a probability of 0.96 (Row 1, Column 2), and using the Average TC response criteria with a probability of 0.96 (Row 1, Column 3). Responders (Row 1, Column 1) were incorrectly classified as non-responders using the Overall TC response criteria with a probability of 0.04 (Row 1, Column 4), and using the Average TC response criteria with a probability of 0.04 (Row 1, Column 5).
Figure 1. Graphical depiction of the potential improvement in response rates afforded by the classifier. Without using the classifier, 30-40% of the patients fail to achieve clinical response (A). Applying the classifier, most of the non-responders are flagged (B) and redirected to other therapy (arrow), in this case gemcatibine and cisplatin. Assuming standard response rates for the gemcatibine and cisplatin therapy, the overall response rate achieved by the bifurcated treatment regimen could approach about 85%.
