January 15, 2005 (Vol. 25, No. 2)
Using Genomic and Proteomic Data To Make More Accurate Predictions
A test just on the market by Genomic Health (www.genomichealth.com) predicts the recurrence risk for individual breast cancer patients and, importantly, which patients may benefit from tamoxifren and which might not.
The information, presented at last month’s “San Antonio Breast Cancer Symposium,” helps resolve the conundrum faced by oncologistswhether to treat breast cancer patients with chemotherapy after lumpectomies. The problem for physicians has been that up to 80% of the patients receiving chemotherapy after lumpectomies don’t need it, but doctors have no reliable way of knowing which patients would benefit from chemo and which would not.
By using the genomic and proteomic data from individual patients’ tumors, Genomic Health and other biotech companies are developing tests that are highly accurate predictors of breast cancer recurrence and response to treatment, allowing physicians to target chemotherapy to those patients who actually can benefit from that therapy.
Genomic Health’s test, called the Oncotype DX Recurrence Score Assay, measures the expression of 21 genes (16 related to cancer development plus five reference genes) to identify a molecular profile of an individual tumor.
“We conducted three independent studies to identify and select genes that appeared to be most important in breast cancer. Breast cancer is a complex disease, and we know that more genes are better than a few,” notes Steve Shak, M.D., CMO.
“We didn’t find that more than 21 genes did better, but fewer genes did worse,” he says. The Oncotype product was developed with tamoxifen-treated patients in mind.
“Physicians,” continues Dr. Shak, “tend to look at the average patient, but breast cancer has different molecular signatures. These can identify and quantify the likelihood of recurrence and point to the magnitude of chemotherapy benefit.
“Based on the trends, women with lower risks do well and have minimal benefit from chemotherapy. The reason is molecular characteristics in which low proliferations of tumor cells or high estrogen receptor levels increase responses to hormonal therapy.”
Conversely, for high recurrence, scored data indicate the largest benefit from chemotherapy. “With the Oncotype DX Re-currence Score Assay, the measure of expression for each gene is precise over a large dynamic range,” Dr. Shak says, and is correlated to clinical data.
Because the test uses tissue stored in paraffin blocks, which are stable over time, Genomic Health was able to correlate its results in a 10-year, 668-patient retrospective study using tissue samples and medical records from a National Surgical Adjuvant Breast and Bowel Project study of tamoxifen. In addition to correlation with actual outcomes, the study also showed that about half of patients are erroneously classified as high or low risk, using the traditional measurements of age, tumor size, and tumor grade.
The test is available now, and is performed in-house by Genomic Health using real-time PCR and fixed paraffin blocks.
At Exagen Diagnostics (www.exagendiagnostics.
com), researchers are developing a test that accurately identified 91% of the hormone receptor-negative and hormone receptor-positive patients from a retrospective study of patients diagnosed with invasive ductal carcinoma between 1986 and 1999. When node negative patients were tested, their results were 100% accurate, based on an average follow-up of 8.9 years.
Exogen’s test offers results in about 30 minutes using the pattern of genomic amplification technique FISH and can be done by virtually any lab. “It’s a single procedure, without needing competitive DNA,” notes Suzanne Z. Mattingly, Ph.D., vp, business development and marketing.
This as-yet-unnamed test is based upon two sets of three gene markers each, which are present in patients’ excised tumors that predict prognosis or drug response. “We look for changes in DNA copy number,” Dr. Mattingly says.
Why only three genes? “We focus on the genes that need to be in the test,” Dr. Mattingly emphasizes. “Other tests start with the literature and use a univarient `p value to make a guess. We’re different, using multivariant analyses and computational science to mine sets of genes in combinations to find those that are truly predictive. Which genes come up in combination are different, from those merely associated with cancer,” she says.
“We started with 25,000 genes,” Dr. Mattingly elaborates, “and identified 17 that were important.” In a 308-patient study, Exogen further defined the relevant gene set, identifying three genes that were predictive for hormone receptor positive patients and three for hormone receptor negative patients, with 91% accuracy. The results have held in subsequent projects.
Exogen is beginning its second validation set this year, and expects the test to be available by 2006. The test appears to be predictive for recurrence in stages I through III.
ViroLogic (www.virologic.com), which recently acquired Aclara Biosciences, is taking a different approach by identifying dimerization, which indicates downstream pathway activation, as a key biomarker for cancer activity.
According to Sharat Singh, Ph.D., CTO, the eTAG system correctly differentiates responders from non-responders to Herceptin treatment. Results from a 13-patient trial indicate it may be more accurate than DakoCytomation’s Hercep-Test (which identifies the presence of the Her2 receptor) in predicting response to Herceptin treatments. A larger, confirmatory study is being planned.
Unlike phosphorylation, which companies initially considered promising, protein complexes are stable in paraffin-embedded tissues, allowing eTAG to identify dimerization and thereby identify activated signaling pathways in tumor tissues. The eTAG System is focusing on the HER and VEGF pathways.
More than 600 tags can be attached to various analytes and then separated and analyzed by capillary electrophoresis.
According to Dr. Singh, “The multi-label eTag System is an extremely sensitive assay method for identifying protein or pathway activation in FFPE samples. Along with identifying responders to Herceptin in breast cancer, the eTag System is being used to identify responders to targeted therapies in non-small cell lung cancer, head and neck cancer, and colorectal cancer. Results are extremely promising.”
Agendia (www.agendia com) used the 20,000 samples and 20 years of research from the Netherlands Cancer Institute/ Antoni van Leeuwen-hoek Hospital in Amsterdam in developing the MammaPrint gene expression profiling service.
“We determined that 231 genes were linked to significant differences in expression related to metastasis,” Bernhard Sixt, Ph.D., president and CEO, says. “Of those we looked at the best predictors a 70 gene set. It was an unbiased way to look at it as the analysis wasn’t tethered to known function. In fact, half the genes had no known function at the time. Science doesn’t really understand the complexity of tumors, and diagnostics should be hanged by it.”
MammaPrint uses customized DNA microarray chips for gene expression profiling to determine the chance of recurrence of breast cancer in women under age 55 with lymph node negative tumors and either positive or negative estrogen receptor status. The chips include several hundred normalization genes, which are used to adjust the signal strength of the two fluorescent labels, and negative control genes to avoid artifacts. Because each chip contains three identical sets of 70 genes, one chip enables three independent measurements.
The prognostic test is for all breast cancer subtypes. It identifies not only high-risk patients who were undetected by conventional diagnostic tests, but can further identify “the very, very high-risk patients,” Dr. Sixt says.
In drug development, Dr. Sixt notes that MammaPrint can identify the population subset for which particular drugs would elicit responses. “We have trials under way for high-risk groups for drug development, involving 5,000 patients,” he says.
MammaPrint has been validated in more than 300 patients and is available globally. It became available in the U.S. in January, through the Molecular Profiling Institute in Phoenix.
AutoGenomics (www.autogenomics.com) is focusing on developing near-patient prognostic tests for breast and cervical cancers, the pharmacogenomics tests CYP 450, 2D6, 2C9, 2C19, and tests for cystic fibrosis, cardiovascular diseases, and bleeding disorders. The company also is developing tests for tumor suppressors, response modulators, apoptotic modulators, and angiogenesis and methylation markers. “Eventually, all these markers could go on one chip,” according to Fareed Kureshy, CEO.
Using BioFilmChip microarrays, up to 48 tests can be performed simultaneously on the Infiniti genomic and proteomic analysis platform. Most proteomic tests can be performed in less than one hour, and genomic tests in less than two hours, Kureshy says.
“We envision a signature of markers that will include DNA, DNA methylation, and proteins and will provide a better profile for diagnosis, prognosis, and management of disease states.”