Concept of Running a Clinical Trial Without Genetic Profiling May Soon Be Unthinkable

March 1, 2003

SAN ANTONIO, Texas-The genetic signature of breast tumors seems to be a powerful predictor of aggressiveness and metastatic potential, outperforming the individual clinical parameters that have traditionally been used, according to several presentations at the San Antonio Breast Cancer Symposium.

SAN ANTONIO, Texas—The genetic signature of breast tumors seems to be a powerful predictor of aggressiveness and metastatic potential, outperforming the individual clinical parameters that have traditionally been used, according to several presentations at the San Antonio Breast Cancer Symposium.

"Molecular profiling will soon impact the treatment of breast cancer," predicted Stephen Friend, MD, PhD, vice president of basic research at Merck Laboratories in West Point, Pennsylvania. Molecular profiling, or gene expression profiling, involves sophisticated pattern recognition that reveals the activities within a tumor cell. The analysis of a host of genes, as opposed to one biological marker such as estrogen receptor status, yields a true "richness of data" and can be used to predict clinical outcome. "It allows for categories of disease that are not simply defined as black or white," Dr. Friend pointed out.

Studies at Merck/Rosetta Laboratories were able to identify patients at risk for distant metastases and to use this information to guide clinical decision-making, specifically, to direct adjuvant therapy only to patients who truly require it.

Individualized Therapy

Gene expression profiling subtypes each breast cancer tumor by its genetic defects, a process that in the future should make individualized therapy possible. The format for gathering this information is cDNA microarray technology, used to rapidly screen and immunohistochemically stain tissue for the presence of multiple genetic markers.

In a study using microarray cell lines, Swedish investigators identified a list of new and partly uncharacterized genes associated with excellent vs poor prognosis, and others associated with chemotherapy resistance and sensitivity.

At the Karolinska Institute, Stockholm, 524 patients underwent surgery for breast cancer from 1994 to 1996, and 186 of these patients provided material for RNA expression profiling. Of these, 37 were profiled on a small gene chip with 10,000 genes and 149 were analyzed on a larger chip involving 33,000 genes (some patients were excluded). Altogether, the microarrays yielded information on 12,625 genes. Distant disease-free survival was assessed relative to the array profiles in 134 patients.

The analysis identified the 50 genes most positively associated with distant disease-free survival and 50 most negatively associated with clinical outcome. Among the top ten best-prognosis genes were a number of ribosomal proteins along with less familiar types of proteins and cyclin-dependent kinase inhibitor.

Among the genes negatively linked to distant disease-free survival were several proteins associated with cell metabolism. The hierarchical clustering of genes was similar to genetic patterns identified in human breast cancer cell lines that are associated with drug sensitivity and resistance, according to Judith Bjohle, MD, who reported the study.

Size Doesn’t Matter

Dr. Friend related his findings to clinical outcomes in a large and unselected cohort of breast cancer patients. The analysis confirmed the predictive power of the 70 prognostic genes identified in his laboratory. The study matched the genetic fingerprint of the 295 tumors with the clinical outcome of the patient with a mean follow-up of 8 years. Among the 114 patients with "good prognosis genes," 15 developed distant metastasis. Among the 181 patients with "poor prognosis genes," 86 developed metastasis.

Furthermore, since metastatic potential is programmed at an early stage, size doesn’t matter, he continued. Small tumors with a poor-prognosis genetic profile are also likely to metastasize, the study showed.

Genetic Fingerprints

In another report, Lajos Pusztai, MD, PhD and colleagues from M.D. Anderson Cancer Center, Houston, identified a novel group of genetic markers for response to chemotherapy. The five-gene set predicted, with 81% accuracy, a complete response to treatment neoadjuvant chemotherapy with paclitaxel plus fluorouracil, doxorubicin, and cyclophosphamide (Cytoxan, Neosar). Genes that were identified are involved in a variety of cellular functions, including signal transduction, cytoskeletal organization, and cell cycle regulation.

Investigators also described the development of a model that will use gene profiling to test the response of tumor cells to drugs, in this case, letrozole (Femara). "Currently we have no predictive model to determine who will respond to hormonal therapies and who won’t, so we prescribe chemotherapy as a backup measure to ensure the cancer’s demise," said Matthew Ellis, MD, PhD, director of the breast cancer program at Duke University in Durham, North Carolina. Duke University is one of the sites in the multicenter trial, which will involve 140 women and use gene expression profiling to measure changes in 16,000 genes as they react, or fail to react, to letrozole.

Dr. Ellis estimated that 50% of patients receive chemotherapy unnecessarily. "The gene chip allows us to measure levels of various genes that give rise to drug resistance. With such genetic fingerprints, we can develop new drugs that target the cellular signaling pathways that have malfunctioned," he said.

Looking to the Future

Dr. Friend predicted that genetic profiling will be most useful in patients with ‘early disease’ without lymph node metastasis, who would not traditionally receive adjuvant therapy but who will benefit from chemotherapy if they have the metastatic phenotype.

The new information has great relevance for clinical trial design and for individualizing therapy. "It is likely that, within a year or two, the concept of running a clinical trial without profiling the tumors will be a thing of the past," he said. In terms of future therapies, it is now clear that "programmed aggressiveness should dictate therapeutic strategy," he added. "Depending on the portions of the markers that are shifting, you would add certain therapies directed toward the altered pathways…. There is a way to use this information to define the disease biology and disease targets, and to direct therapy to the subtype of disease that is really there for the individual patient."