Survival estimates stratified by the Nottingham Prognostic Index for early breast cancer: A systematic review and meta-analysis of observational studies

Ewan Gray*, Anna Donten, Katherine Payne, Peter S. Hall

*Corresponding author for this work

Research output: Contribution to journalReview articlepeer-review

Abstract

Background: Estimates of survival for women diagnosed with early staged breast cancer are available based on stratification into prognostic categories defined using the Nottingham Prognostic Index (NPI). This review aimed to identify and summarize the estimated survival statistics from separate sources in the literature and to explore the extent of between-study heterogeneity in survival estimates. Methods: Observational studies in women diagnosed with early and locally advanced breast cancer reporting overall survival by NPI category were identified using a systematic literature search. An exploratory meta-analysis was conducted to describe survival estimates and assess between-study heterogeneity. Results: Twenty-eight studies were identified. Nineteen studies with sufficient data on overall survival were included in meta-analysis. A high level of heterogeneity in survival estimates was evident with I 2 values in the range of 90 to 98%. Conclusions: The substantial differences between studies in the relationship between NPI categories and survival at 5 and 10 years poses challenges for use of this prognostic score in both clinical settings and in decision-analytic model-based economic evaluations.

Original languageEnglish
Article number142
JournalSystematic Reviews
Volume7
Issue number1
Early online date15 Sept 2018
DOIs
Publication statusPublished - 2018

Keywords

  • Breast cancer
  • Meta-analysis
  • Nottingham Prognostic Index
  • Precision medicine
  • Prognosis

Fingerprint

Dive into the research topics of 'Survival estimates stratified by the Nottingham Prognostic Index for early breast cancer: A systematic review and meta-analysis of observational studies'. Together they form a unique fingerprint.

Cite this