By Tony Strattner
January 12, 2004 | Diagnostic testing for cancer is the commercial spark igniting a race in 2004 among suppliers of genomics and proteomics technology. On the genomics track, Agendia and Genomic Health have released gene-profiling products and services to allow physicians to assess patients’ likelihood of breast-cancer recurrence. In proteomics, Correlogic Systems recently licensed its protein-pattern blood test to two national clinical laboratories, LabCorp and Quest Diagnostics, for early detection of ovarian cancer.
Amsterdam-based Agendia first announced the availability of a microarray test using customized DNA chips from Agilent Technologies to predict the aggressiveness of breast cancer. The microarrays interrogate the activity of some 70 genes within a tumor sample. The resulting pattern of gene activity is interpreted by software from Rosetta Inpharmatics to predict the risk that the tumor will metastasize.
Agendia is “pressing ahead with clinical trials” to further validate the test’s predictive power, says Rene Bernards, chief scientific officer. “We’re confident that implementing gene-expression profiling in clinical trials will become the industry standard in the near future.”
Genomic Health, based in Redwood City, Calif., is preparing to have its diagnostic assay using formalin-fixed, paraffin-embedded tissue ready for clinical use early this year. The test, on a 21-gene panel, involves extracting RNA from the tissue specimen, then analyzing it using Applied Biosystems’ TaqMan RT-PCR system. A proprietary algorithm calculates a “recurrence score” from the expression data. Genomic Health is setting up its own laboratory and has received certification under the Clinical Laboratory Improvement Act (CLIA) regulations to provide the analysis service.
Correlogic has been working with researchers at the FDA and National Cancer Institute (NCI) since 2002 on a different approach to diagnosing ovarian cancer. Proteins in patient blood samples are separated and analyzed by mass spectrometry. Correlogic’s pattern-recognition software is then applied to the blood samples to create a disease model.
Correlogic’s new licensing partners, LabCorp and Quest Diagnostics, will take blood samples, perform mass-spectrometry and send the spectral-readout data to Correlogic for pattern-recognition analysis. (The Bethesda, Md.-based company has also received CLIA certification.) LabCorp and Quest will pay Correlogic royalties, milestone payments, and development fees for additional refinements to the technology.
But How Accurate?
The question of accuracy does not loom as large for Correlogic’s protein test as it does for the gene-based tests. In a 2002 study published in [I]The Lancet[I], researchers using Correlogic’s technology analyzing 116 blinded blood samples were able to correctly distinguish all 50 cases of ovarian cancer from non-malignant samples.
In contrast, predicting the likelihood of breast cancer recurrence is more problematic. Genomic Health’s test, for example, separates women at high risk of recurrence from those at medium or low risk. The test has been touted as a way to help women decide whether to endure the debilitating effects of chemotherapy.
But there’s an error rate of plus or minus 2 to 3 percent in each risk category, according to Steven Shak, Genomic Health’s chief medical officer. For example, of the control group of 668 women the company studied, nearly 7 percent classified by the test as “low risk” still saw their cancer return within 10 years.
“That’s not good enough for most oncologists to offer [as a means of deciding whether to withhold chemotherapy],” says Larry Norton, deputy physician-in-chief for breast cancer programs at Memorial Sloan-Kettering Cancer Center in New York City. “Can things be done to get to a 1- to 2-percent relapse rate in the good prognosis group? That’s an open question.”
By comparison, patients classified by Agendia’s test as “low risk” have about a 13-percent chance of the cancer returning within 10 years. A big difference, however, is that unlike the patients in Genomic Health’s control group who were all treated with tamoxifen, the patients in Agendia’s control group had not yet received any treatment. (Indeed, Genomic Health says its test is ineffective for women not treated with tamoxifen.)
Bernards notes that the metastases that typically occur in Agendia’s “low-risk” group are not as aggressive as the “high-risk” patients. “Ninety-six percent of all those low-risk patients are alive in 10 years,” he says, “and most of those women have a functional aromatase receptor, so additional treatment with anti-aromatase inhibitors would be beneficial, whereas chemotherapy may not be.”
Bernards stresses that while neither the Agendia or Genomic Health tests are perfect, they are “a hell of a lot better than the previous methods used.”