Research

For a complete version of this figure with gene names, click on the image.

Human carcinomas show great diversity in their morphologies, clinical histories and in their responsiveness to therapy. This wide tumor diversity poses the main challenge to the effective treatment of cancer patients. The focus of my lab is to characterize the biological diversity of human tumors using genomics, molecular genetics, and cell biology, in order to develop improved therapies that are specific for each tumor subtype. Our genomic characterization of human breast tumors identified at least five biologically distinct subtypes including Luminal A, Luminal B, Basal-like, HER2+ and Normal-like (Perou et al. 2000, Sørlie et al. 2001 and 2003, Hu et al. 2006); these Intrinsic Subtypes are predictive of relapse-free and overall survival times, and predictive of responsiveness to chemotherapy (Rouzier et al. 2005 and Carey et al. 2007). We have also shown that the Intrinsic Subtype classification is adding new clinical information beyond current clinical parameters (Hu et al. 2006) and is concordant with other gene expression based predictors for breast cancer patients (Fan et al. 2006). In addition to breast carcinomas we are also studying Head and Neck Squamous Cell Carcinomas (Chung et al. 2004), Lung carcinomas (Hayes et al., 2006), Glioblastomas and Ovarian Carcinomas.

Concurrent with our tumor profiling studies are animal model and cell line projects that are aimed at determining the molecular function of the genes that define the Intrinsic Subtypes (Finlin et al. 2001, Usary et al. 2004, and Moyana et al. 2006). As an example, my lab has shown that GATA3 is somatically mutated in some ER-positive breast tumors (Usary et al. 2004). One of the somatic mutations we identified was identical to a germline GATA3 variant that is present in two kindreds with HDR syndrome (OMIM #146255), which is an autosomal dominant syndrome caused by the haplo-insufficiency of GATA3. The conditional knock-out of GATA3 in mouse mammary tissue greatly inhibits mammary gland development and caused a complete loss of luminal/ER+ epithelial cell formation. We have also determined that there are human germline variants of GATA3 that predispose to developing breast tumors (Garcia-Closas et al., 2007).

Although numerous mouse models of human breast carcinomas have been developed, we do not know the extent to which any faithfully represent clinically significant human tumor phenotypes. To address this need, we characterized mammary tumors from over 20 different murine models using DNA microarrays and identified many similarities to human breast tumors including proliferation and Intrinsic Subtype signatures (Herschkowitz et al., 2007); tumors of several models displayed characteristics of human Basal-like tumors (Brca1-/- and T-antigen) while other models showed Luminal characteristics (MMTV-Neu and MMTV-Pymt). Our mouse model studies are an ongoing project where we are focused on “humanizing” existing models using additional genetic engineering to obtain the combinations of genetic alterations seen in human tumors, and then we use these models to empirically test new therapeutics.

Our genomic studies also depend upon the utilization of many computational tools. In addition to maintaining the UNC Microarray Database (https://genome.unc.edu/), our Lineberger Comprehensive Cancer Center Bioinformatics Group develops tools for the analysis of microarray data including Distance Weighted Discrimination (DWD), which is a powerful classification method that allows one to correct for systematic biases that are present across microarray data sets (Benito et al. 2004). DWD has allowed us to combine microarray data sets together from different groups or organisms, and to use the combined data to validate the prognostic and/or predictive importance of a given gene set (Hu et al. 2006, Herschkowitz et al., 2007).

A final project in the lab is the integration of genomics and epidemiology, which is being done in collaboration with researchers at the NCI, Norway, and the UNC Department of Epidemiology. Recently, we determined using an immunohistochemical surrogate for the breast intrinsic subtypes and the population based Carolina Breast Cancer Study, that the Basal-like tumors were twice as frequent in young African Americans (AA) versus Caucasians, which may contribute to the higher mortality rates seen in AA (Carey et al. 2006). In addition, our most recent analyses of epidemiological variables show that the risk factor profile for Basal-like tumor patients is different than the profile for Luminal/ER+ patients, with multiple pregnancies and no lactation being risk factors only for developing Basal-like tumors (Millikan et al., 2007).

In summary, my lab utilizes a multi-disciplinary approach to characterize tumor diversity, and we then use this information to understand more about tumor biology and to design new clinical trials for cancer patients that are based upon tailored therapies (see http://clinicaltrials.gov/show/NCT00232505 = Cetuximab Alone and Cetuximab With Carboplatin in ER/PR-Negative, HER2 Non-overexpressing Metastatic Breast Cancer). I am actively seeking new graduate students, medical fellows and postdocs and have opportunities available that utilize genomics, genetics, molecular and cellular biology, computational biology and human population genetics.

Those interested should write to Dr. Perou at cperou@med.unc.edu