Background
The majority of breast-cancers are sporadic, however, 20-30% are considered familial (occur in the context of a significant family-history of the disease). Women suspected of being at higher risk of breast-cancer than the general-population based on their family-history are frequently referred to a ‘family-risk’ clinic for risk assessment and management (Evans et al, 1996). Reliable estimation of an individual’s risk is important as it aids clinicians in recommending particular risk-management strategies and assists individuals themselves in reaching decisions about which, if any, risk-reducing measures (risk-reducing surgery, chemoprevention, surveillance etc.) to opt for. It is known that the greatest benefit from breast-cancer prevention strategies is obtained through treating women who are at high-risk of the disease (Powles 2002). This further highlights the need for accurate risk assessment so that women can be appropriately stratified for breast-cancer prevention strategies according to their estimated-risk. A risk-prediction model is a statistical-tool for estimating the probability that an individual with specific risk factors (e.g. strong family-history) will develop a future-condition such as breast-cancer. Several such models have been developed to assess breast cancer risk. Understanding the strengths and weaknesses of each model, as well as the clinical scenarios in which predictions from one model vary markedly from others, facilitates accurate breast cancer risk-assessment (Domchek et al, 2003). The performance of risk-prediction models can vary according to the population that they are applied to, for example a model may have good accuracy in a high-risk population but not in a low-risk population and vice versa (Meads, et al, 2012). Risk-prediction models are frequently used as an adjunct to clinical-judgement in the breast-cancer family-risk setting. Currently, in clinical practice different risk-prediction models are used in different centres and geographical differences exist (e.g. Between USA and Europe) with regard to preference of one model over another.
Question to be addressed & Rationale
The proposed review will attempt to answer the following question: Which breast-cancer risk-prediction model or models are the most reliable and appropriate for use in the breast cancer family risk setting?
Rationale
This question is important as it is unclear which of the currently available breast-cancer risk prediction models is most accurate for use in the family-risk setting. Risk prediction models are used to estimate an individual’s risk of developing breast-cancer which aids clinicians and patients in reaching decisions about risk management. Therefore, accurate risk-prediction is extremely important. Furthermore, the ability to accurately predict risk in this ‘family-risk’ population can provide information about future disease burdens and future numbers of individuals that will require risk reducing interventions. This information will benefit health policy makers. To our knowledge, only one other comprehensive systematic-review of breast-cancer risk-prediction models has been published (Meads et al, 2012). Importantly, this systematic-review excluded studies in high risk groups of women such as those with specific genetic-mutations and those who had close family relatives with breast-cancer; as such it differs from our proposed systematic-review in which these studies will be included and will be central to meeting the review objectives.
Objectives:
· To identify all available breast-cancer risk-prediction models that are applicable to the family-risk setting (i.e. models that include family-history of breast-cancer as a component)
· To qualitatively summarise the characteristics of these risk-prediction models
· To appraise the predictive-accuracy (validation) of these models
· To compare model performance by meta-analysing model performance statistics across studies if
feasible
· To make recommendations for breast-cancer risk-prediction models that may be useful for patient
management in the breast-cancer family-risk setting.
Methods of the review:
Participants / population:
Women only (pre- and post-menopausal) including women recruited to:
· cohort studies (prospectively or retrospectively) used in breast-cancer risk prediction model
development or validation
· case-control studies used in breast-cancer risk prediction model development or validation
Intervention & Comparators: The proposed review falls into the category of ‘prognostic review’ individual risk-prediction models will be appraised and models will be compared by meta-analysing model performance statistics across studies if feasible. It is anticipated that the review will be conducted in a similar manner to the exemplar Cochrane prognostic review by Pace et al. (Pace et al, 2014).
Outcomes: Primary: Breast-Cancer occurrence
Secondary: None
Subgroup analyses: If possible we will compare the predictive-accuracy of risk-prediction models in women at increased risk of developing breast-cancer versus those in the general female population.
Types of study:
Included
· Studies describing development of a Breast-cancer risk-prediction model that includes family-history of breast-cancer as a component with or without external validation by independent data
· External model-validation with or without model updating
Excluded
· Models predicting genetic mutations rather that cancer
Anticipation of difficulties:
· A potential problem in conducting this (and any) systematic-review is that there may not be a sufficient number of primary studies on which to base the review. Scoping the literature we found at least 20 studies that are likely to be suitable for inclusion in the review.
· Prognostic reviews present unique challenges in terms of study selection, assessment of bias
and analysis- to overcome these challenges we will engage with both the Cochrane Breast-Cancer
and Prognosis-Methods review groups for advice and guidance at all stages of the review process.
Importantly, the Cochrane Breast-Cancer review group have offered to provide us with a mentor/co-author with extensive experience in prognosis research methods if required. Furthermore, we believe that we possess ample methodological and statistical support through Dr. Kathleen Bennett (local supervisor and co-author) to deal with any analysis issues that may arise.
Training Plan
During this fellowship the following training-courses will be attended by the lead-applicant at the times indicated: Table 1: Training-plan
Training-Event Time Attended
Systematic-review initiation course (Encompassing workshops RA1 & RA2) entitled “Developing a Protocol for a Review”
Year 1- shortly after accepting fellowship Workshop RA3 – Analyses for Systematic-Reviews
End of Year 1 – following protocol completion, after study selection, prior to commencing analysis
Workshop RA4 – Advanced Topics in the Analysis and Reporting of a Systematic review
First half of Year 2 – After data collection, Prior to data analysis.
Team Advisory meetings
with local Supervisor Monthly
with co-authors Monthly with Y. Hanhauser, Quarterly with others
with mentor/co-author supplied by breast-cancer review group At start of fellowship and as required thereafter (via teleconference) with Cochrane Breast-cancer review group and/or Prognosis-Methods group At start of fellowship, prior to study selection, prior to data analysis
Why this review topic is relevant, important and timely on the island of Ireland
Relevant: Breast-cancer family-risk clinics are run regularly at several hospitals throughout the island of Ireland. In a four year period (2010-2013 inclusive), 3880 patients attended the family-risk clinic at St. James’s hospital (SJH) with number of attendees increasing in recent years. Ireland has a strong legacy in the area of familial breast-cancer; for instance, samples from patients attending the first Irish high-risk screening-clinic were used to help identify the BRCA2 gene for the first time in 1994 (Stratton et al, 1994). This legacy continues with further development of breast-cancer family-risk services. This review will further strengthen Ireland’s international reputation as an important player in the field of familial breast-cancer. Thus the proposed review is highly relevant for Ireland.
Important: As is the case internationally, In Ireland, management of women with a family-history of breast cancer relies heavily on assessment of their individual-risk so that they can be stratified for interventions. As such, the findings from this review will be extremely important for Irish health-care providers in the family-risk setting and their patients as it will allow decisions to be made regarding which risk-prediction model to use in their clinics based on trusted evidence. This will instil confidence in Irish healthcare providers that they are using the most appropriate risk-assessment model for their patient population. This review has the potential to be practice-changing both in Ireland and worldwide.
Timely: We are entering a new era in the field of familial breast-cancer in Ireland. Developments have been made in recent years for example, in 2011 the first Irish dedicated cancer-genetics clinics were established at SJH and the Mater Private Hospital (MPH) by Dr. David Gallagher (co-author) for individuals with a strong family-history of cancer. New initiatives include the establishment of a charity through the SJH foundation to help fund support services for high-risk individuals. Furthermore, options are currently being explored in relation to establishing an all-Ireland registry of individuals deemed ‘high-risk’ for breast-cancer. In this context the proposed review is timely and complements the exciting developments that are evolving in the Irish family risk/ cancer-genetics field.
