Research Grants Awarded
Development Of A Methylation Panel To Determine Breast Cancer Recurrence Risk: Using The Breast Cancer Hypermethylome To Identify Highly Promising Biomarkers
Investigator Initiated Research
Recurrence of breast cancer after curative surgery for early stage breast cancer is devastating. There is a critical need for development of prognostic biomarkers that can predict the risk of recurrence for patients. In recent years, hypermethylation of promoter-associated CpG islands has been found to be a frequent mechanism of inactivation of tumor suppressor genes in breast cancer.
Recently, Sjoblom et al. sequenced 13,023 human genes in breast and colon cancer and showed that 189 CAN (candidate cancer genes) genes were likely selected for during tumorigenesis, albeit most with a low rate of mutation frequency. However, the true extent of hypermethylated genes in breast cancer is, as yet, unknown. To this end, we have recently described an exciting gene expression array strategy to identify the spectrum of genes that are hypermethylated and silenced in cancers, the hypermethylome. Our data suggests that, on the average, 150-500 genes are hypermethylated in each tumor. This work was recently published using colorectal cancer. We have now used this expression array strategy to characterize the breast cancer hypermethylome using a series of breast cancer cell lines with varying malignant potential and hormone receptor status. The hypermethylome represents the cluster of re-expressed genes following treatment with the DNA methyltransferase inhibitor 5-deoxyazacytidine but not following treatment with the HDAC I/II inhibitor trichostatinA. Following filtering of genes with no basal expression on the microarrays, and choosing genes with promoter CpG islands, this approach identified 3433 genes for further analysis. In-depth analysis in breast cancer cell lines (n=124 genes) showed a validation rate of >70% of the hypermethylated genes in cell lines with a false negative rate of 9%. Analysis of 73 genes in a large series of primary human breast cancer samples identified 44 genes that are methylated and silenced in primary breast cancers without methylation in normal breast tissues or surrounding lymphocytes.
Furthermore, the breast hypermethylome was compared to the consensus list of mutated genes identified by Sjoblom et al. Most remarkably, a subset of these CAN genes (n=18) were also found within the breast cancer hypermethylome and they represent our most promising set of biomarkers that have been derived from this hypermethylome strategy, since these genes are common targets of both mutation and methylation (CTMM). Preliminary data suggests that some of the CTMM genes are preferentially methylated in advanced tumors. Furthermore, correlation with published breast cancer microarray databases shows that silencing of certain CTMM genes appears to predict a more aggressive phenotype of breast cancer associated with decreased 5-year overall survival and death. Thus, these novel genes represent a treasure trove of biomarkers which deserve further in-depth study. We now propose to test if these novel promising genes (including the highly promising CTMM genes) can be used to predict breast cancer recurrence using two independent case control cohorts with long-term follow-up.
We hypothesize that methylation profiles of a panel of genes in the primary can provide molecular biological staging of breast cancer. Furthermore, we hypothesize that the recent identification of the breast hypermethylome by our laboratory now provides us with novel candidates that may serve as promising methylation biomarkers. We now propose the following studies:
Aim 1: To identify a panel of best candidate genes that may predict breast cancer recurrence from our candidate pool within the breast hypermethylome.
Aim 2: To perform, in our first large initial cohort at Memorial Sloan Kettering, a nested case control study to determine if methylation of a panel of genes can predict recurrence in early stage breast cancer.
Aim 3: To validate, in a second large cohort in New Mexico, the HEAL cohort, the utility of our biomarker panel to predict recurrence in an independent cohort of patients.
We will perform a series of coordinated studies to systematically test the methylation of status of these novel genes as a candidate gene discovery study in Aim1 using a pilot set of patients with recurrence and controls without recurrence to narrow our list of potential biomarkers to a testable number of 20 genes for the proposed studies in Aims 2 and 3. We will then conduct a blinded nested case control study of DNA methylation in primary breast cancer and regional lymph nodes in a large well-characterized, independent cohort, the MSKCC cohort (Memorial Sloan Kettering Cancer Center) of 450 patients with an average follow-up of 8.5 years of the most promising candidates from Aim1. Furthermore, validation studies of the biomarker panel will be performed using another independent cohort, the HEAL (Health, Eating, Activity and Lifestyle) cohort, a prospective population multicenter study that enrolled 750 women from stage 0-IIIA between 1996-1999. Methylation analysis will be performed using MSP which was developed in our laboratory.
Breast cancer remains the second leading cause of cancer death in women in United States. Due to improvements in screening, the majority of women now present with early-stage breast cancer. Mortality from breast cancer is now primarily due to development of recurrence or metastatic disease after undergoing curative surgery for breast cancer. Thus, understanding and preventing the development of distant metastases is one of the most important aims in research and treatment of breast cancers. However, our currently accepted prognostic and predictive markers fall short and there is a critical need to identify novel biomarkers that predict breast cancer recurrence. This is illustrated by the fact that approximately 25% of patients with primary breast cancer without lymph node involvement who are younger than 50 years of age will still develop distant metastases after 5 years. These patients will benefit from adjuvant chemotherapy. Conversely the majority of women with node negative disease will remain free from recurrence and do not benefit from adjuvant therapy. The identification of biomarkers that can individualize a patient?s risk of recurrence accurately will help stratify patients and has the potential to profoundly impact treatment selection for individual patients.
Aberrant DNA methylation is a frequent and early event during carcinogenesis and serves as a mechanism for silencing tumor suppressor genes. Methylation-associated silencing targets genes in multiple cellular pathways during breast carcinogenesis including cell-adherence and metastasis, DNA repair, and cell cycle control. Analyses of methylation profiles-which are cancer-specific, widespread in all cancer types and easily studied- are, thus, thought to have enormous prognostic and predictive value. Current applications of methylated genes as biomarkers have been limited, so far, to a handful of genes that have been identified to be methylated in breast cancer such as HIN1, Twist, Cyclin D and AP2?. However, the true extent of the number of genes hypermethylated in cancer is, as yet, unknown and this has hampered development of powerful biomarkers.
We have recently developed an exciting strategy to define the true extent of hypermethylated genes in breast cancer. This strategy has identified novel genes that are aberrantly methylated and silenced in breast cancer. Our data shows that there are, on average, 150-500 genes hypermethylated in any given breast cancer. We have performed detailed studies of 73 of these genes in a large subset of breast cancer of varying stages and grades and show that 44 of these genes are cancer- specific events. Most importantly, we demonstrate that a subset of these genes are not only methylated but also mutated in breast cancer. Inactivation of these genes by multiple mechanisms during carcinogenesis highlights their functional importance in breast cancer and these are our most promising biomarkers. As proof of principle, we have already seen that a subset of these highly promising genes predict strongly for poor clinical outcomes including overall survival, disease-free survival, higher stages and grades in a small cohort. These genes now need to be tested in an unbiased approach for their prognostic potential to predict breast cancer recurrence.
We hypothesize that methylation profiles of a panel of genes in the primary can provide molecular biological staging of breast cancer. Furthermore, we hypothesize that the recent identification of the breast hypermethylome provides us with novel candidates that may serve as promising methylation biomarkers.
We now propose to test these highly promising, novel genes in two independent prospective cohorts to determine if they can serve as biomarkers for breast cancer recurrence. During this grant period, we hope to develop a methylation panel that can provide a risk profile for each patient. The proposed studies are eminently feasible. The PI and Co-PI have recently worked together on identifying the breast cancer hypermethylome and are now primed to translate these findings into this next exciting level. Importantly, methylation biomarkers can be easily translated into the clinical arena within a short time span. In fact, a methylation test for early detection of prostate cancer has recently been commercialized. Methylation analysis can be performed on archived, paraffin-embedded tissues which are available on all patients, can be performed for all breast cancer types and can be performed with low-cost using small amounts of DNA. Finally, a key reason for our optimism for the proposed studies is our recent success in using methylation profiles to perform molecular staging in lung cancer.
The ability to develop individualized biomarkers that predict recurrence risk for early stage breast cancer will have a huge potential impact in reducing breast cancer mortality. The ability to identify high risk patients at the time of diagnosis will allow the treating oncologist to deliver aggressive adjuvant therapy to such patients. Equally important, the ability to identify patients who are not at high risk will allow us to spare toxic effects of chemotherapy to our patients who are deemed low risk. In addition, identification of such predictive factors for distant metastases will lead to more insight in the biological processes leading to the development of distant metastases which may allow development of more targeted therapies.