Biostatistics Training Initiative
Dr. Richard Cook, University of Waterloo
Dr. Gregory Pond, McMaster University
Recent technological advances have led to far more complex and much larger datasets in cancer research. There is a critical need for biostatisticians who can work with these data, ensuring they can be used to inform on the next generation of cancer treatments. Since 2010, the Biostatistics Training Initiative (BTI), previously known as the Oncology Research Training and Methods Program (ORTMP), has placed Master’s students from the University of Waterloo in eight-month internships with mentors at Ontario cancer centres. In doing so, the BTI is training the next generation of biostatisticians in the province, ensuring they are equipped with the technical knowledge, skill set and experience to deal with the changing landscape of cancer treatment and clinical trials.
Based on the success of the program in its first five years and feedback from the Ontario research community regarding its continued need for training, expertise and placement, the BTI expanded in 2016. The new expanded BTI engages a broader group of trainees, including PhDs and postdoctoral fellows, and fosters a sense of community amongst participants through a visiting lecture and monthly seminar series.
The BTI is funded through the Ontario Institute for Cancer Research.
The BTI Internship Program is a continuation of the earlier OICR-funded ORMTP in which selected top Master’s students in Biostatistics from the University of Waterloo are placed in eight-month internship positions in cancer research centres across Ontario with joint mentorship by a biostatistician and an oncology researcher. Up to four fully funded placements will be made each year from 2016-2020.
For further information and instructions to apply, please visit https://uwaterloo.ca/biostatistics-training-initiative/internship
The BTI Fellowship Program is designed to support doctoral and postdoctoral training in biostatistics for cancer research. The primary objective of the Program is to support the training of the next generation of biostatisticians in the province, ensuring they are equipped with the technical knowledge, skill set and capability for the application of rigorous quantitative methods for cancer research. Through the receipt of a BTI Studentship/Fellowship Award, biostatistics or statistics doctoral students and postdoctoral fellows at an Ontario university will engage in interdisciplinary, collaborative cancer research with investigators located at an Ontario Host Institution.
Any questions about these awards or how to apply may be emailed to email@example.com.
A monthly seminar series will be held by WebEx/teleconferencing on advanced issues in clinical and population health advances, as well as methodological challenges arising in cancer research. The aim is to engage clinicians, statisticians, interns and fellows. The focus of the seminars will include:
- recently completed high impact trials in oncology within the group and internationally,
- recent developments in the design and analysis of cancer clinical trials,
- biostatistical and more general methodological challenges arising in the planning or conduct of trials, and
- reports of ongoing methodological research conducted by engaged faculty, students and interns.
BTI Fellowship Program
The BTI Fellowship Program is a new and exciting initiative designed to support doctoral and postdoctoral training in biostatistics for cancer research. The primary objective of the Program is to support the training of the next generation of biostatisticians in the province, ensuring they are equipped with the technical knowledge, skill set and capability for the application of rigorous quantitative methods for cancer research. BTI Fellows will engage in interdisciplinary, collaborative cancer research with investigators located at an Ontario Host Institution.
BTI Fellowship Program
The BTI Fellowship Program is designed to support doctoral and postdoctoral training in biostatistics for cancer research. The primary objective of the Program is to support the training of the next generation of biostatisticians in the province, ensuring they are equipped with the technical knowledge, skill set and capability for the application of rigorous quantitative methods for cancer research. Through the receipt of a BTI Studentship / Fellowship Award, biostatistics or statistics doctoral students and postdoctoral fellows at an Ontario university will engage in interdisciplinary, collaborative cancer research with investigators located at an Ontario Host Institution.
BTI Studentship / Fellowship Award
For the February 2017 Call for Applications, one award valued at $15,000 per year for each of two years will be made available. All funds from this award must be used to support a doctoral student or postdoctoral fellow conducting biostatistical or statistical research which has an application to a problem in oncology.
Eligible applicants include:
- Students currently enrolled in the second or higher year of a Biostatistics or Statistics doctoral program at an Ontario university; or
- Current Postdoctoral Fellows in Ontario or individuals beginning a postdoctoral position in Ontario within six months of the application deadline; fellows must have completed doctoral training in biostatistics or statistics; or
- Biostatistics or statistics faculty member at an Ontario university, with a strong background in cancer research, who plan to supervise a biostatistics or statistics doctoral student or postdoctoral fellow.
Applications, composed of completed Form I, letters of support, biosketch and curriculum vitae (CV), must be submitted to firstname.lastname@example.org by 5 p.m. EST on April 14, 2017. Additional details are available in the Applicant Guide.
The award is expected to be announced in early May 2017 with funds available by June 1, 2017
For any questions please contact Teresa.email@example.com
BTI Seminar Series
October 24 2017
Clement Ma, PhD,
Dana-Farber/Boston Children’s Cancer and Blood Disorders Center, Harvard Medical School
Dr. Clement Ma is Lead Biostatistician at Dana-Farber/Boston Children’s Cancer and Blood Disorders Center and Instructor of Pediatrics at Harvard Medical School. He provides statistical leadership in collaborative clinical research with pediatric hematologists and oncologists. His current methodological research focuses on the design of early-phase clinical trials for children with cancer. During his graduate and post-graduate training at the University of Michigan, he developed statistical methods for analyzing low-frequency genetic variants in single and multi-center genetic association studies of complex diseases. He completed his bachelor’s (biomedical engineering) and master’s (biostatistics) degrees at the University of Toronto, and was a former biostatistician at Princess Margaret Cancer Centre.
Dual-agent dose escalation methods for pediatric oncology clinical trials
Phase 1 clinical trials aim to identify the optimal dose for the therapeutic agent that balances patient safety and potential efficacy. Cancer therapies that include two or more agents may increase efficacy as well as toxicity. Many adaptive dose-escalation designs have been proposed for trials of combination therapies. These designs can better assign dose combinations near the maximum tolerated dose combination (MTDC) to enrolled patients but require significant resources to design and monitor. Hence, relatively few adult oncology trials have used these designs, and to our knowledge, none have been used in pediatric trials. To motivate the use of adaptive designs in pediatric oncology, we performed a simulation study to compare the performance of dual-agent dose-escalation methods in a pediatric oncology framework.
We selected four Bayesian methods, and the commonly used 3+3 rule-based design (assuming a prespecified set of dose combinations) for our study. We designed 7 simulation scenarios with a restricted number of dose combinations and low total sample size (N=24) to reflect the realities of pediatric trials. We performed 2,000 simulated trials per scenario for each method and compared the methods across six metrics. Overall, all adaptive methods had a similar performance across all metrics. The average recommendation rates for the true dose combination ranged from 37% to 43%. The average proportion of patients receiving a dose greater than the MTDC ranged from 29% to 33%, which is near our target toxicity level of 30%. As expected, the conservative 3+3 design had lower recommendation rates than the adaptive designs. Adaptive designs represent a safe and effective way for dual-agent dose escalation trials in children.
- Learn about different dose-escalation designs for clinical trials of combination therapies; and
- Understand the use of statistical simulation to evaluate the operating characteristics of clinical trial designs.
- A comparative study of adaptive dose-finding designs for phase I oncology trials of combination therapies
- Competing designs for drug combination in phase I dose-finding clinical trials
Ontario Institute for Cancer Research
Boardroom 5-20/5-21, MaRS Centre (NOTE ROOM CHANGE)
661 University Avenue, Suite 510, Toronto, Ontario
Juravinski Hospital and Cancer Centre
G Wing, 2nd floor boardroom
711 Concession St, Hamilton, Ontario
University of Waterloo
M3 3001, 200 University Avenue West
1. Register your name and email:
a) Go to https://cc.callinfo.com/r/1e6o6cjhzi2qa&eom (not currently compatible with Microsoft Edge);
b) Register your first and last name, email address, and click “Register Now”. Registration can be done several minutes in advance of the webinar. No password is required.
2. Join the teleconference:
a) Dial-in using: 1.800.503.2899, access code: 2484428;
b) Important: Please mute your phone during the presentation, unmuting it for question period.
In advance of the seminar, we strongly recommend testing the WebEx application to ensure your computer is compatible. To do so, use https://www.webex.com/test-meeting.html to launch the test. You will be required to enter your name and email address. It will then run a test WebEx meeting advising if it was successful. Should you experience technical difficulty, please contact your IT administrator.
We look forward to your participation.
January 23, 2018
Speaker: Jack Lee, PhD, MS, DDS, MD Anderson,
Presentation topic: Development/application of innovative Bayesian methods for cancer clinical trials
February 27, 2018
Speaker: Bingshu Chen, PhD, Queen’s University,
Presentation topic: A hierarchical Bayes model for biomarker subset effects in clinical trials
Past Seminar Series
September 26 2017
Speaker: Viswanath Devanarayan, PhD, FAAPS, Charles River Laboratories
Dr. Devanarayan is currently the Executive Director and Head of Global Statistics at Charles River Laboratories. He has over 21 years of combined pharmaceutical research experience from Eli Lilly, Merck, and AbbVie. His statistical & data-analytic contributions span a wide range of applications across drug discovery and development, such as target identification, high-throughput-screening, genomics, proteomics, bioanalytical methods, precision medicine, and exploratory clinical research. He has filed 10 patent applications, given over 100 invited talks at scientific meetings, and co-authored over 55 publications that includes several white-papers with regulatory, academic and industry scientists. He is an elected Fellow of the American Association of Pharmaceutical Scientists (AAPS), and is also serving as an Adjunct Professor at the University of Illinois in Chicago. He is currently volunteering as the AAPS Task Theme Chair on Predictive Modeling.
Title: Subgroup identification algorithms for precision medicine
Abstract: Causal mechanism of relationship between the clinical outcome (efficacy or safety endpoints) and putative biomarkers, clinical baseline, and related predictors is usually unknown and must be deduced empirically from experimental data. Such relationships enable the development of tailored therapeutics and implementation of a precision medicine strategy in clinical trials to help stratify patients in terms of disease progression, clinical response, treatment differentiation, and so on. These relationships often require complex modeling to develop the prognostic and predictive signatures. For the purpose of easier interpretation and implementation in clinical practice, defining a multivariate biomarker signature in terms of thresholds (cutoffs/cut points) on individual biomarkers is preferable. In this talk, we will propose some methods for developing such signatures in the context of continuous, binary and time-to-event endpoints. Results from simulations and case study illustration are also provided.
Speaker: Eleanor Pullenayegum, Sick Kids
Title: Multiple outputation for longitudinal data subject to irregular observation
Abstract: Observational cohort studies often feature longitudinal data subject to irregular observation. Moreover, the timings of observations may be associated with the underlying disease process and must thus be accounted for when analysing the data. Multiple outputation, which consists of repeatedly discarding excess observations, can be a helpful way of approaching the problem.
- Explain why irregular observation can be a problem
- Understand different visiting patterns
- Choose an appropriate analysis for a given visiting pattern
Lunenfeld Mount Sinai
A novel region-based Bayesian approach for genetic association with next generation sequencing (NGS) data coverage.
The discovery of rare genetic variants through Next Generation Sequencing (NGS) is becoming a very challenging issue in the human genetic field. We propose here a novel region-based statistical test based on a Bayes Factor (BF) approach to assess evidence of association between a set of rare variants located on this region and a disease outcome. Marginal likelihood is computed under the null and alternative hypotheses assuming a binomial distribution for the rare variants count in the region. A Beta distribution or a mixture of Dirac and Beta distribution is specified for the prior distribution. The hyper-parameters are determined to ensure the null distribution of BF does not vary across genes with different sizes. A permutation test or False Discovery Rate (FDR) statistic are used for inference. Our simulations studies showed that the new BF statistic outperforms standard methods under most situations considered. Our real data application to a lung cancer study found enrichment for rare variants in novel genes.
Tuesday March 28, 2017
Luyao Ruan – Juri Reimand, OICR
Neural Network Model for Copy Number Data
In my presentation, I will mainly discuss the project that I worked on during my co-op at the Ontario Institute for Cancer Research. The project focuses on using convolutional neural networks to classify
copy number variation data of Medulloblastoma cancer samples into different subgroups. I will also mention some future directions for the project and discuss some of my other experiences and achievements
during the co-op.
Qihuang Zhang – Wei Xu, UHN
The Influence of Genetic Variation on the Association Between Statin and Prostate Cancer Risk: A Genome-wide Association Study
During my eight-month internship at Princess Margaret Hospital, I had picked up experience in real oncology clinical studies, data quality control, data analysis, scientific report and manuscript writing. In this talk, I will mainly share my experience in a Genome-wide Association Study (GWAS), cooperated with principal investigator Dr. Rob Hamilton. Genome-wide association study is a statistical examination of a genome-wide set of genetic variants (represented by Single-nucleotide polymorphism, or SNPs) in different individuals to see if any variation is associated with a trait. In the initial study, 3651 prostate cancer patients were included. 533,631 SNPs were collected, and the number was extended to 6,873,954 after doing genotype imputation. A genetic quality control was conducted to remove the individuals and SNPs that were not eligible for the study. The removing criteria included the missing minor allele frequency, outlying heterozygosity rate and correlations among individuals. We constructed additive interaction model and applied it to case-control outcomes by scanning through the whole genome. The results of the study can provide clinical practitioners with a guidance of personalized medicine when giving treatment to prostate cancer patients.
Ri Tong (Rick) Wang – Melania Pintilie, UHN
Therapeutic Somatostatin Analogs Effect on FGFR4 Polymorphic Alleles
In my presentation, I will elaborate on my general experiences as a biostatistics intern at PMH. In particular, I will focus on the work and challenges leading up to the publication of the paper “FGFR4 polymorphic alleles modulate mitochondrial respiration: A novel target for somatostatin analog action in pituitary tumors”; a paper which compares effects of different somatostatin analogs for FGFR4 codon 388 SNPs.
Erle Holgersen – Paul Boutros, OICR
Prostate Cancer Heterogeneity and Assorted Chi-Squared Tests
In this presentation I will talk about my work as a biostatistics intern at OICR. I will reflect on learning outcomes of the internship and challenges associated with working as a statistician in a bioinformatics lab. The talk will include examples from a study on the heterogeneity of radioresistant prostate tumours.
All speakers are Master’s students in the Biostatistics graduate program at the University of Waterloo
Tuesday February 28, 2017
Dr. David Berman, MD, PhD
Director, Queen’s Cancer Research Institute, Queen’s University
Biomarkers for Treatment Stratification in Early Prostate Cancer
One in 8 men will be diagnosed with prostate cancer in his lifetime. The vast majority of these men will suffer no harm from this disease, but a small minority (ca. 15%) will die of metastatic cancer spread. The field of prostate cancer care has shifted from over-diagnosis and over-treatment of harmless prostate cancers to a less aggressive stance that de-emphasizes broad screening and offers deferred treatment (active surveillance) for men with cancers that are deemed to have a low risk of metastasis. Unfortunately, risk assessment is highly inaccurate, due to challenges in prostate cancer imaging, biopsy, and pathologic evaluation. This presentation will discuss clinical, pathologic, and molecular features associated with low risk prostate cancer, and attempts to improve risk assessment using molecular features.
Tuesday, January 24, 2017
Dr. Rinku Sutradhar
Senior Scientist, Institute for Clinical Evaluative Sciences
TExamining cancer screening adherence in Ontario using multistate transitional models.
Understanding disparities among women in breast cancer screening adherence is of considerable interest, however prior work has been methodologically limited. Our work longitudinally examines adherence to screening, and determines factors associated with becoming adherent. The cohort consisted of 2, 537, 960 women age 50-74 from Ontario, Canada. Using age as the time scale, a relative rate multivariable regression was implemented under the 3-state transitional model to examine the association between covariates (all time-varying) and the rate of becoming adherent. Individual- and physician-level characteristics played an important role in a woman’s adherence to screening. Our research improves the quality of evidence regarding disparities among women in adherence to breast cancer screening, and provides a novel methodological foundation to investigate adherence for other types of cancer screening, including cervix and colorectal cancer screening.
Tuesday December 13, 2016
Methodology to Identify Environmental and Host Genomic Association on Human Microbiome Data
Dr. Wei Xu
Assistant Professor, University of Toronto, Dalla Lana School of Public Health
Technological advances in genomic sequencing have enabled researchers to unveil the wide variability of bacteria presented within different locations of the body, i.e. the microbiome, and how it relates to disease. However, our understanding of how microbiomes affect diseases is still unclear. It is necessary to better understand both environmental and host genetic factors impact the composition of the microbiome to improve disease management. Powerful statistical and bioinformatics tools are needed to overcome these knowledge gaps.
Friday November 25, 2016
Phase II Study Design in Oncology Drug Development
Dr. Wendy Parulekar, MD
Senior Investigator, Cancer Clinical Trials Group, Queens University
The Phase II trial has a pivotal role in drug development since the decision to proceed with further evaluation of a drug/ drug combination is based on the efficacy and safety data generated from this type of study. Important additional generated from the Phase II trial may include elucidation of the mechanism of action of a new therapy and delineation of the target population for administration. The objectives of this webinar are to provide the clinician perspective on Phase II trial design and conduct including classification of designs and statistical framework. Examples of actual clinical trials will be drawn from the casebook of the Canadian Cancer Trials Group.
Friday September 23, 2016
The Challenges of Making Clinically Useful Biomarkers
Dr. Paul Boutros, PhD
Principal Investigator, Informatics and Bio-computing Platform, OICR
Assistant Professor, Department of Medical Biophysics, University of Toronto
Biomarkers are a core part of personalized medicine. They give patients and their clinicians insight about possible future, increasing their confidence in the decisions they make. They also provide regulators and health-care funders confidence that care is being delivered cost-effectively and maximizing outcomes for the entire population. Creating such biomarkers from the many large emerging genomic datasets is surprisingly difficult. There are problems in optimizing experimental design, in fully leveraging large biological data, and in ultimately validating and generalizing the models. We will talk about all these issues, using recent successful genomic biomarkers in prostate cancer as a guiding example.
June 24 2016
Life as a biostatistician
Bethany Pitcher, MSc
Biostatistician, Princess Margaret Cancer Centre
The role of a biostatistician is not limited to the analysis of data. As with many careers, schoolwork lays the foundation for your skills, but does not fully prepare you for the varied ways you will apply these skills in the workplace. I will discuss some of the unique challenges and unexpected aspects of working as a biostatistician that I encountered during my internship and after graduation. I hope to give current students an idea of what they can expect from a career in Biostatistics. I will also discuss my experience as an intern in the Biostatistics Training Initiative and highlight one project I worked on throughout my placement at Princess Margaret Cancer Centre. For this project we performed a meta-analysis of control-arm data from multiple clinical trials available on the Project Data Sphere database. The goal was to determine if any concomitant medications were associated with prognosis in patients with metastatic prostate cancer.
May 20 2016
Patient Reported Outcome Measures (PROMs) for use with children and adolescents: a view across disciplines.
Gillian Lancaster, PhD
Lancaster University, United Kingdom
April 22 2016
How clinical trial design pitfalls slow progress against cancer
David J. Stewart, MD
Head, Division of Medical Oncology, University of Ottawa/The Ottawa Hospital
April 1 2016
The analysis of progression-free survival, overall survival and markers in cancer clinical trials
Richard Cook, PhD
Professor of Statistics, Department of Statistics and Actuarial Science, University of Waterloo
Cancer clinical trials are routinely designed on the basis of event-free survival time where the event of interest may represent a complication, metastasis, relapse, or progression. This talk is concerned with a number of statistical issues arising from the use of such endpoints including the interpretation of results based on composite endpoints, the consequences of naïve analysis of event times subject to dual censoring schemes, and the causal interpretation of treatment effects. Related issues in the joint analysis of longitudinal and survival data will also be highlighted.