A systematic review of generic and breast cancer specific life expectancy models in the elderly.

ABSTRACT

INTRODUCTION: The use of primary endocrine therapy (PET) in managing breast cancer in the elderly has become common practice. Whilst there appears to be no difference in overall survival in comparison with surgery, PET has been found to be inferior in local disease control with a limited duration of efficacy (2-3 years). The International Society of Geriatric Oncology (SIOG) state that PET may be considered in patients with a short life expectancy (<2 years) or considered unfit for surgery. Frequently, decision making for PET allocation is a subjective process by the clinician. METHOD: A systematic literature review was performed to establish what prediction models are available for all-cause mortality in the elderly, and what breast-specific models have been produced. RESULTS: 18 prognostic models were deemed eligible from 15 papers. 1 breast-specific model was found, 2 nursing home related and 15 for community-dwelling elders. Accuracy (as defined by discrimination; c-statistic or AUROC) ranged from 0.69 (moderate) to 0.86 (very good). CONCLUSIONS: This review highlighted a variety of validated prognostic indexes. Several models with very good accuracy were identified but most were validated in US-populations and relied on information from administrative datasets. One breast specific model by Stotter et al. was identified, specifically to aid treatment planning for frail elderly patients but had limited accuracy. The strength of an index will ultimately be on its clinical impact and influence on treatment decisions rather than its accuracy and as of yet no trials exploring this have been carried out.