Department of Biostatistics and Computational Biology
Research
Methodologic research
Faculty members in the Department of Biostatistics and Computational Biology have long been active in extending the boundaries of statistical methods for medical research. All of this work is supported by peer-reviewed grants, usually from the National Cancer Institute (NCI).
In the last year, these grants supported work in bioinformatics, statistical issues in the early detection of disease, methods for analyzing correlated sequential measurements, robust methods for survival data, mechanisms of mortality and morbidity due to air particles, and the development of software for toxicological risk assessment.
Some of our accomplishments include:
- The Computational Biology and Functional Genomics Laboratory has been using laboratory and computational approaches to develop new methods for the integrative analysis of biomedical data. Much of that work has focused on organizing information to drive discovery. Joseph White partnered with Edie Weller to create a data warehouse that links clinical and research data in multiple myeloma, greatly accelerating the pace of research in this disease and creating an infrastructure that can be expanded to other disease areas. Aedin Culhane created a database of gene expression signatures in cancer and used this to demonstrate that many published signatures are confounded by molecular subtype, raising a call for better annotation of data in public databases. Jessica Mar has developed a new “state space” approach to understanding cell fate transitions and has been using this and other methods to explore viral perturbations of cellular networks. John Quackenbush received a prestigious EUREKA award to expand work he and Amira Djebbari have done in developing a seeded Bayesian Network approach to analyzing Gene Expression. And the new Center for Cancer Computational Biology, under the direction of Quackenbush and Mick Correll, opened to bring the methods this and other groups are developing to support the broader community of investigators at DFCI.
- The Cancer Invention Surveillance Modeling Network (CISNET) is an NCI-funded consortium of researchers with a main goal of improving our understanding of the impact of cancer control interventions on population trends in incidence and mortality using modeling. As a part of the CISNET, Drs. Zelen and Lee have developed an analytic theory (Biometrics, 2008) for a general problem of modeling the early detection of chronic disease. Questions such as optimal exam scheduling, over-diagnosis and predicting future mortality of disease in a population can possibly be answered by the proposed model. In collaboration with the CISNET breast investigators, the stochastic model of Zelen and Lee has been applied to assess relative and absolute contributions of screening mammography and adjuvant treatment in the reduction of breast cancer mortality in the U.S. in 1975-2000 (NEJM, 2005; JNCI, 2006). The CISNET breast group also has collaborated with the U.S. Preventive Services Task Force to evaluate the efficacy of annual vs. biennial screening schedules; resulting in updated guidelines that will be published in 2009.
- The lab of Dr. X. Shirley Liu focuses on computational models of transcription and epigenetic regulation by integrating data from genome-wide ChIP-chip / ChIP-Seq, nucleosome occupancy and histone modifications, gene expression microarray/RNA-sequencing, genomic sequence and conservation. Dr. Liu developed a number of algorithms for transcription factor motif discovery from promoters of gene expression clusters or ChIP-chip data. The Liu Lab pioneered ChIP-chip / ChIP-Seq data analysis, and is developing algorithms to study the regulation of transcription factors, chromatin factors, and histone modifications. Her lab has been collaborating with other researchers to study gene regulation in cancers (especially estrogen receptor regulation in breast cancer and androgen receptor regulation in prostate cancer), metabolic disease, aging, and stem cell development.
- Dr Yi Li’s Biostatistical Lab in Population Science examines the technical advances in biomedicine that produce an abundance of high-throughput data and result in a major statistical challenge in variable selection, which is important for ensuring high prediction accuracy and for discovering most relevant variables among many potential candidates. The group collaborates with DFCI investigators on detecting predictive genes, out of thousands of candidates, for myeloma patients' event-free survival and overall survival. They are now investigating a new weighted Dantzig variable selector for regression models when the response variable is subject to censoring. They establish the theoretical properties of the new procedure, while numerically examining its performance.
- During 2008, department members authored or coauthored more than 150 peer-reviewed publications. These articles appeared in a range of journals in both cancer research and biostatistics, including the Journal of Clinical Oncology, Journal of the American Medical Association, Blood, Journal of the American Statistical Association, Cancer, and Biometrics
Collaborative research
Statisticians in the department provide expertise in nearly all aspects of interdisciplinary research in cancer. They prepare or review the statistical designs and analysis plans for all clinical protocols conducted at Dana-Farber and DF/HCC through membership in the newly developed disease programs, the DF/HCC Scientific Review Committees, and the Institutional Review Boards. Because of its many years of experience in clinical research, the department has built an extensive library of software for calculating sample sizes and analyzing data from clinical trials. Approximately 30 percent of the projects of the Cancer Center Support Grant are collaborative efforts with basic laboratory scientists at DF/HCC. Our statisticians also provide assistance with statistical content for all grant submissions. We estimate that each year our statisticians are involved in more than 800 projects, large and small, with DFCI investigators.
Several of the departments' collaborative projects extend beyond DFCI, including:
Cancer Care Outcomes Research and Surveillance Consortium
The Statistical Coordinating Center of CanCORS is headed by David Harrington, PhD. CanCORS is a population-based study of diagnosis, treatment, and outcome in lung and colorectal cancer, with special emphasis on the reasons behind well-documented disparities in treatment by race or ethnicity and age. The study team has completed baseline and follow-up interviews of more than 10,000 patients or their caregivers, and has also conducted a complete and detailed review of these patients' medical records, focusing in particular on all aspects of their cancer care delivery and associated decision making. This project is funded by several grants from the National Cancer Institute.
Center for Cancer Computational Biology (CCCB)
The mission of the Center for Cancer Computational Biology (CCCB) is to further the Dana-Farber Cancer Institute’s commitment to advance the understanding, diagnosis, treatment, cure, and prevention of cancer and related diseases. We aim to do that by providing broad-based support for the analysis and interpretation of ’omic data and in doing so further basic, clinical, and translational research and to conduct research that opens new ways of understanding human cancer.
Center for Cancer Computational Biology Web Site
Eastern Cooperative Oncology Group
Our department serves as the statistical center for the Eastern Cooperative Oncology Group (ECOG), a consortium of over 350 hospitals and smaller treatment centers in the United States funded by the National Cancer Institute (NCI) to conduct multicenter clinical trials in adult malignancies. Now more than 50 years old, ECOG maintains a database of more than 100,000 cancer cases, and is conducting active follow-up of over 20,000 patients treated in clinical trials. Recent ECOG studies have shown that the addition of the anti-VEGF monoclonal antibody bevacizumab to standard chemotherapy improves outcomes in patients with advanced colon, lung, or breast cancers. Other major studies have shown the benefits of rituximab therapy for non-Hodgkin's lymphoma and of thalidomide therapy for multiple myeloma. ECOG is also initiating new studies investigating the use of molecular markers to tailor treatments for patients with breast and colorectal cancer. The ECOG Statistical Center is led by Robert Gray, PhD.
Dana-Farber/Harvard Cancer Center
The Dana-Farber/Harvard Cancer Center (DF/HCC) is an innovative collaboration formed in 1999 from the original Dana-Farber Cancer Institute, which had been designated by the National Cancer Institute (NCI) as a Comprehensive Cancer Center, and has been supported by the NCI for the past 36 years. DF/HCC was formally funded as a 7-institution consortium, including the Beth Israel Deaconess Medical Center, Brigham and Women's Hospital, Children's Hospital Boston, Dana-Farber Cancer Institute, Harvard Medical School, Harvard School of Public Health, and Massachusetts General Hospital, and an NCI-designated Comprehensive Cancer Center in 2000. The cancer research and treatment collaborations sponsored across the participating institutions by the DF/HCC will combine the disciplines of population science, clinical science, and basic research, to facilitate the development of novel and improved modalities for the prevention and treatment of cancer.
Dana-Farber/Harvard Cancer Center Web site
DNA-Chip Analyzer
Developed by Dr. Cheng Li, DNA-Chip Analyzer (dChip) is a software package implementing model-based expression analysis of oligonucleotide arrays and several high-level analysis procedures. The model-based approach allows probe-level analysis on multiple arrays. By pooling information across multiple arrays, it is possible to assess standard errors for the expression indexes. This approach also allows automatic probe selection in the analysis stage to reduce errors due to cross-hybridizing probes and image contamination. High-level analysis in dChip includes comparative analysis and hierarchical clustering.
International Breast Cancer Study Group
For over 25 years, the International Breast Cancer Study Group (IBCSG) has dedicated itself to the development of innovative clinical research to improve the quality of life and survival of women with breast cancer. IBCSG is a consortium of institutions in Europe, South America, Australia and New Zealand, Asia, and South Africa, and is part of Breast International Group (BIG). The Statistical Center, located in our department, works closely with investigators and patients in over 19 countries around the world to achieve excellence in large-scale clinical trials for adjuvant therapy.
Since 1978, IBCSG has conducted large, randomized phase III clinical trials evaluating the timing and duration of chemotherapy and the role of endocrine therapy as adjuvant treatment for breast cancer. IBCSG has been a leader in the field of tailored treatment approaches for specific subpopulations of patients with breast cancer. Members of this department, led by Richard Gelber, PhD, comprise the Statistical Center for the IBCSG. An internationally recognized expert in the field, Dr. Gelber co-organizes the St. Gallen (Switzerland) International Conference on Primary Therapy of Early Breast Cancer, held biannually, and coauthors the conference proceedings.
IBCSG has conducted trials of tailored treatment approaches for younger patients and older patients with endocrine responsive early breast cancer. The premenopausal trials, SOFT and TEXT, were launched in 2003 and are the first studies that IBCSG has undertaken in collaboration with North America. The postmenopausal trial, BIG 1-98, reported on the monotherapy comparison between tamoxifen and letrozole in 2005 and the results of the evaluation of these agents given in sequence will be presented at the San Antonio Breast Cancer Symposium in December 2008.
Harvard School of Public Health Department of Biostatistics
Located on the Harvard Medical Campus, the Department of Biostatistics was one of the first departments in the newly formed Harvard School of Public Health in 1922. The departmental location is central to its mission: to facilitate collaborations between the members of the department and other medical scientists. Now in its 86th year, the department comprises 70 students, 60 faculty members, and 55 research scientists, associates and fellows. Our size contributes to our ability to address a broad spectrum of biostatistical and public health issues.
Department of Biostatistics, Harvard School of Public Health Web site
Lunenburg Lymphoma Biomarker Consortium (LLBC)
Edie Weller, Ph.D., is the lead statistician in the international LLBC effort to standardize the measurement of biomarkers for diffuse large b-cell lymphoma and to validate their prognostic relevance using data from large clinical trials performed by cooperative groups throughout the world. Over 1500 DLBCL patients from six different countries (Canada, England, France, Germany, Netherlands, United States) with clinical data and specimens available will be included in the analysis. Results on the standardization of the markers have been published. Further results regarding the prognostic relevance of these markers were expected to be available in April 2009.
