Research Grants Awarded
Time-Dependent Patterns of Tumor Recurrence and Other Events After Early Stage Breast Cancer
Detection, Diagnosis and Prognosis
Background: Recurrence risk remains elevated for an extended period after tumor removal for early stage breast cancer, obviating some type of adjuvant treatment for many patients. The recurrence hazard has been reported in several recent studies to exhibit a bimodal ‘double-peaked’ pattern over time. It has been hypothesized that this pattern has biologic significance and that, in light of evolving therapeutic options such as anti-angiogenesis agents, could influence treatment choice if confirmed. However, the data sources and the analytic methods applied in these studies are inadequate to validate and further interpret these apparent patterns. We propose a systematic investigation of patterns of recurrence hazard in breast cancer, using large high-quality datasets from randomized trials and modern statistical analysis methodology. Objectives/Hypotheses: The principal goal of this research is to determine whether the recurrence hazard exhibits non-monotonic patterns consistent with earlier reports, and to identify factors over which recurrence patterns differ. Objectives thus include determination of a) whether the recurrence hazard exhibits the bimodal pattern b) whether and how axillary lymph node status, menopausal status, and tumor estrogen receptor content influence the hazard (all are implicated as modifiers of the bimodal pattern in previous studies), c) whether and how failure patterns are influenced by adjuvant therapy, and d) how recurrence at specific anatomic sites and competing risks of second cancers and non-cancer mortality contribute to patterns in the recurrence hazard. Specific Aims: 1) Perform estimation and tests to confirm the existence of bimodal recurrence hazard among axillary node-negative and node-positive patients; to compare patterns of recurrence hazard by nodal status (to address the previous assertion that the pattern is absent in node-negative patients) 2) Perform tests of whether any such pattern depends on other factors including tumor estrogen receptor status and menopausal status 3) Compare recurrence hazards by treatment, with particular emphasis on pre-operative and multi-agent chemotherapy 4) Perform competing risks analyses to determine whether specific anatomic failure sites show time-dependent patterns and to characterize outcomes with respect to second cancers and deaths from non-cancer causes over long-term follow-up. Study Design: This study will utilize patient cohorts from clinical trials of the NSABP, a national multi-center clinical trials group. Data will be included from over 20,000 participants across a spectrum of therapeutic strategies. Statistical methodology will involve recently developed simultaneous inference methods for smoothed hazards, which accounts for variability in hazard function estimates over time. Previous analyses presented only empirical arguments for the existence of patterns in the hazard, without accounting for variability or appeal to statistical inference procedures. Potential Outcomes and Benefits of the Research: It is anticipated that highly reliable information will be obtained concerning time-dependent patterns in recurrence risk. Benefits of this research include rigorous validation of whether hypothesized failure patterns are reproducible, insights into how traditional and new treatment paradigms alter recurrence risk, and detailed information about how the risk for specific failure types and other events change over time. Knowledge of these issues has significant implications for both treatment and appropriate follow-up surveillance.
Background: After treatment for early stage breast cancer, women may initially remain free of tumor recurrence for several years, but their recurrence risk remains elevated for a substantially longer period of time. Recent studies have proposed that this risk increases around the period 1.5-3 years following surgery and then decreases before increasing again around 5-8 years. It has been suggested that new treatment and follow-up strategies could alter this course and improve outcomes. However, so far the patterns have only been noted in limited numbers of patients, and many additional questions remain before this concept could be used to guide treatment choice and follow-up care. We propose an investigation of time-varying patterns of breast cancer recurrence, using modern analysis methods on high quality data from breast cancer clinical trials. Objectives/Hypotheses: The primary objective of this research is to determine whether the recurrence risk demonstrates the pattern reported in recent studies, and whether this pattern depends on other patient and tumor characteristics and treatment type. Several questions will be addressed: Does the risk of recurrence resemble the pattern of previous studies? Does the risk depend on other factors such as lymph node status or menopausal status? More importantly, how do different treatments alter the risk pattern? Finally, because recurrence can involve different organs and events such as other cancers or deaths from other causes can occur, how do different event types contribute to changes in recurrence risk? Specific Aims: Specific aims addressing the above questions are as follows: 1) Confirm or refute the presence of a ‘double-peaked’ recurrence risk, 2) Determine whether any such pattern depends on factors such as lymph node status, tumor estrogen receptors, and menopausal status, 3) Compare patterns in the recurrence hazard for different treatments, including pre-operative chemotherapy and chemotherapy regimens that include newer agents such as taxanes, 4) Determine whether specific types of recurrence (such as bone metastases) show time-dependent patterns, and estimate long-term risks for other cancers and non-cancer deaths. Study Design: This research will use a unique data source consisting of treatment and follow-up history from over 20,000 women who participated in clinical trials of the National Surgical Adjuvant Breast and Bowel Project, a national multi-center cancer clinical trials group. Detailed information is available on types of recurrence and other events observed over a period of up to 25 years. In addition to this wealth of available data, modern analytic methods not previously applied to this question will be used. Potential Outcomes and Benefits of the Research: This research will provide the most reliable evidence to date regarding this provocative concept, determining whether breast cancer recurrence risk may follow more predicable patterns than previously appreciated. Understanding this process has significant implications for both treatment and follow-up monitoring.