Toward Personalized Medicine

From OncoLog, January 2007, Vol. 52, No. 1

Toward Personalized Medicine

by Dawn Chalaire

For years, clinicians have noted that patients with the same type of cancer can have wildly different responses to the same treatment. Some of that variation is attributed to differences in individual physiology, such as the way patients metabolize drugs. Recently, however, molecular studies have been providing more and more evidence that genetic and proteomic differences lie at the heart of the question of who will respond to treatment and who will not.

Think of it as going beyond finding a needle in a haystack to being able to predict the kind of hay that will be found in the vicinity of a particular type of sifting through approximately 25,000 genes, 300,000 single-nucleotide polymorphisms (SNPs), and 1.5 million proteins to identify the molecular signatures that are associated with certain types or stages of cancer, prognoses, and responses to treatment. Identifying these gene- or protein-based biomarkers in patients with cancer could lead to the prevention or earlier diagnosis of disease and to the selection of more effective treatments for individual patients.

“This type of research has the potential to revolutionize how we manage patients,” said Gordon Mills, M.D., Ph.D., professor and chair of the Department of Systems Biology at The University of Texas M. D. Anderson Cancer Center. “The idea is to develop and implement personalized molecular medicine.” Dr. Mills is the co-director of the institution’s Robert J. Kleberg, Jr. and Helen C. Kleberg Center for Molecular Markers.

The center, established two years ago as part of a major research initiative at M. D. Anderson, comprises clinicians, translational scientists, and basic scientists, who—like other scientists around the country—are working to identify molecular markers that have potential applications in cancer prevention, detection, and treatment. Dr. Mills predicts that clinicians will see some tangible progress in these areas within the next five years. Ultimately, the goals are to be able to identify people at high risk for specific cancers, diagnose cancers at earlier stages of development, and better specify which patients should be treated with a particular therapeutic drug. To achieve these goals, the center supports the development of basic science research through in-house programs in leukemia and breast, ovarian, and lung cancers and facilitates the translation of basic science research performed outside the center. Programs are also being developed in glioma, prostate cancer, and renal cell carcinoma.

Finding predictors of treatment outcomes in glioblastoma

The discovery of a clinically useful marker for a certain type of cancer is the result of a long, multistep process. Ken Aldape, M.D., an associate professor in the Department of Pathology, and his colleagues have been addressing the question of why some patients with glioblastoma who receive the standard treatment (chemoradiation with temozolomide followed by adjuvant temozolomide) survive significantly longer than other patients given the same treatment.

It was recently found that the methylation status of the gene O6-methylguanine-DNA methyltransferase (MGMT) was a predictor of outcome to standard therapy in glioblastoma. Patients whose tumors were methylated at MGMT had a better outcome than patients whose tumors were not. However, the test was not specific enough to dictate what therapy an individual patient should receive, so Dr. Aldape and his colleagues began looking for other markers to complement MGMT status.

“We did some high-throughput analyses of glioblastoma tumor samples in patients with known survival data, asking the question, ‘What were the genes that differed between these favorable versus nonfavorable outcomes?’ One of the genes that came out of that analysis was YKL40, but there are other genes, too,” Dr. Aldape said.

Genes that tend to be overexpressed in patients with poor survival represent therapeutic targets. If those genes can be neutralized, survival rates could be improved. Alternatively, patients who are genetically identified to have disease that does not respond well to the standard treatment could be selected for different treatment approaches.

An array analysis such as the one Dr. Aldape and his colleagues performed shows only an association between a certain marker and, in this case, patient outcome. In tests, such as microarrays, that use a relatively small number of samples to search for tens of thousands of markers, some of the associations found will be due only to chance. To address this multiple comparisons problem, Dr. Aldape and his colleagues began looking at other researchers’ profiling array findings to see if they identified the same genes.

“We were able to obtain four independent data sets from glioblastoma samples, and by comparing the data, we found a set of 38 genes that seem to be reproducibly predictive,” Dr. Aldape said. The researchers are now trying to validate the gene markers using independent tumor samples from an ongoing Radiation Therapy Oncology Group clinical trial. Dr. Aldape said that M. D. Anderson’s Center for Molecular Markers can help with these samples by looking at additional platforms, including DNA (SNP array) and microarray analyses.

Genetic signatures for tumor response

Investigators with the Breast Cancer Pharmacogenomic Program are applying genomic technology to the problem of improving breast cancer patients’ responses to adjuvant chemotherapy. Several adjuvant chemotherapy regimens—including various combinations of fluorouracil, doxorubicin, cyclophosphamide (FAC); paclitaxel (T); docetaxel; and capecitabine—are virtually equal in terms of patient outcomes, and so the choice of which regimen to use is usually made on the basis of physician preference.

According to Lajos Pusztai, M.D., Ph.D., an associate professor in the Department of Breast Medical Oncology, about 25% of patients given any of the best preoperative chemotherapy regimens for breast cancer have no residual disease after six months of treatment and will have excellent longterm survival. By using the results of molecular analyses to match gene expression profiles of each patient’s cancer to different regimens, the researchers hope to improve the pathologic complete response rate to 35% to 40%.

Fraser Symmans, M.D., an associate professor in the Department of Pathology who leads a clinical gene expression profiling laboratory, is working with Dr. Pusztai on the breast cancer study. By performing microarray analyses of fine-needle aspiration biopsy samples, the researchers have discovered genetic signatures for tumor response to different adjuvant chemotherapy regimens. The discovery of and first validation results for a 30-gene predictor of response to T-FAC have recently been reported by their group. The paper by Ken Hess, Ph.D., an associate professor in the Department of Biostatistics, showed the genomic test to be as accurate as any of the other diagnostic tests currently in use. Researchers in the Breast Cancer Pharmacogenomic Program are now performing validation studies on the predictors for FAC and FAC plus docetaxel and capecitabine and are also finalizing the design of a prospective clinical trial in which patients will be assigned to receive treatments on the basis of their tumors’ genetic signatures.

In the course of their research, the investigators also identified an important mechanism of drug resistance, in which the gene that most influences response to T-FAC therapy is TAU, which blocks the binding site of paclitaxel, a key ingredient in T-FAC, and reduces the effect of the drug.

“So out of the midst of all this mathematics from microarray experiments comes a key gene that is functionally and biochemically responsible for a good part of the failure to get a complete response,” Dr. Symmans said. “On its own, TAU is a decent predictor of treatment success. But combining TAU with the other genes that were identified from the microarrays into a 30-gene signature provides a much better predictor of response to T-FAC chemotherapy than we had before.”

Drs. Symmans and Pusztai are also working on identifying a predictor for response to endocrine treatment in patients with estrogen receptor-positive breast cancer using molecular and biostatistical methods to identify the genes that represent estrogen receptor activity in biopsy samples. In 260 patients treated with tamoxifen only for five years, the researchers saw a clear and strong association between the expression of certain genes and how much the patients benefited from endocrine treatment.

According to Dr. Symmans, the methods and standards being established in their RNA-based microarray studies are relevant for both proteomic and DNA-based studies being conducted by other investigators through the center. “Others are investigating at the protein level what we’re learning from RNA measurements of gene expression in breast cancer trials,” Dr. Symmans said. “So there’s good synergy occurring at scientific and clinical levels.”

Bringing molecular research to the clinic

But the identification of molecules that play a role in the development and progression of cancer or in its response to treatment is only the first step in the process of developing clinically relevant applications. The road from laboratory association to clinical applicability is often long and treacherous, and many researchers simply choose not to take it.

“A lot of studies of biomarkers going back many decades show that we can measure something quite reliably and show that a particular marker is associated with a specific outcome, but that’s pretty much where 95% of the literature ends,” Dr. Pusztai said. “We need to translate these into clinically useful assays—to move from showing an association to developing actual tests that can be used for therapeutic decision making.”

The Center for Molecular Markers works with other M. D. Anderson researchers on the design of the translational aspects of clinical trials. Dr. Mills said that the center is working with the Department of Biostatistics on a new clinical trial design that will integrate molecular marker identification and validation into clinical trials.

“In essence,” Dr. Mills said, “the goal of the Center for Molecular Markers is to realize the promise of personalized molecular medicine, which is developing a treatment plan specific to the molecular makeup of each patient.” It’s a lofty goal, but a promising one, and progress toward it is well under way.

For more information on this topic or for questions about M. D. Anderson’s treatments, programs, or services, call askMDAnderson at (877) MDA-6789.

TOP

©2008 The University of Texas M. D. Anderson Cancer Center
1515 Holcombe Blvd., Houston, TX 77030
1-877-MDA-6789 (USA) / 1-713-792-3245
Patient Referral Legal Statements Privacy Policy