11/29/2004RSNA ACCURACY OF AXILLARY FDG-PET FOR STAGING PRIMARY BREAST CANCER: A META-ANALYSIS BEN A. DWAMENA; GAVIN MACKIE; DIANE THANAGAMI; MARK HELVIE University of Michigan (Departments of Radiology and Nuclear Engineering) & VA Ann Arbor Health Care System (Nuclear Medicine Service)
11/29/2004RSNA BACKGROUND Axillary node staging is important for treatment and prognostication in operable breast cancer Conventional imaging tests such as CT, MR, scintimammography and xray mammography have suboptimal performance
11/29/2004RSNA BACKGROUND Thus, axillary dissection and sentinel node biopsy are standard of care These are invasive/minimally invasive and not without clinically significant complications (in up to 20% of patients) The latter has a steep learning curve and variable false-negatives (0-15%)
11/29/2004RSNA BACKGROUND Several investigators have studied FDG PET as a non-invasive imaging alternative with highly variable results and ongoing lively debate regarding diagnostic operating characteristics We sought to empirically inform this debate by means of a systematic review and meta-analysis of the published diagnostic accuracy studies (vis a vis SLNMB performance)
11/29/2004RSNA RESEARCH QUESTIONS How accurate is FDG-PET in detecting and/or excluding Axillary Lymph Node Metastases (Sensitivity & Specificity)? How informative is a positive/negative axillary FDG-PET scan (Likelihood Ratios)?
11/29/2004RSNA METHODS Exhaustive search of multiple databases Selection of relevant studies with 2X2 data Random effects and SROC meta-analyses to obtain performance indices Testing for heterogeneity and selection bias
11/29/2004RSNA SYSTEMATIC REVIEW 26 datasets(25 studies) of 1743 patient units published eligible for review Study size: Pretest probability of nodal metastases: 48% ( 16%-81%) Reported sensitivity: 56%-100% Reported specificity: 63%-98%
11/29/2004RSNA ACCURACY
11/29/2004RSNA INFORMATIVENESS
11/29/2004RSNA SUMMARY ROC CURVE Threshold SEN: 0.98 ( ) Threshold SPE: 0.75 ( )
11/29/2004RSNA PUBLICATION BIAS
11/29/2004RSNA TECHNOLOGICAL EVOLUTION USING PUBLICATION YEAR ( ) AS SURROGATE MEASURE OF TECHNOLOGICAL EVOLUTION IN A CUMMULATIVE META- ANALYSIS SHOWS THAT EARLIER STUDIES REPORTED MORE OPTIMISTIC RESULTS
11/29/2004RSNA CONCLUSIONS The pooled evidence & corresponding clinically- relevant measures from SROC analysis show excellent specificity and moderate likelihood ratios for axillary FDG-PET The levels of discriminatory performance and informativeness do not obviate surgical staging in a large number of patients Adequately powered, methodologically sound, multicenter (comparative SLNMB-PET) studies using relevant clinical spectrum are warranted.