Skip Navigation
AHRQ--Agency for Healthcare Research and Quality: Advancing Excellence in Health Care
  • Home
  • Search for Research Summaries, Reviews, and Reports
 
 

EHC Component

  • EPC Project

Related Products for this Topic

Related Links for this Topic



Research Review - Final – Feb. 13, 2012

Noninvasive Diagnostic Tests for Breast Abnormalities: Update of a 2006 Review

Formats

Archived: This report is greater than 3 years old. Findings may be used for research purposes, but should not be considered current.

People using assistive technology may not be able to fully access information in these files. For additional assistance, please contact us.

Structured Abstract

Objectives

To systematically review the literature on the diagnostic accuracy of noninvasive imaging technologies proposed to be useful as part of the workup after recall of women with suspicious breast abnormalities identified on routine screening. This report is an update of a Comparative Effectiveness Review originally published in 2006.

Data Sources

We searched the medical literature, including PubMed and Embase, from December 1994 through September 2010. We included diagnostic cohort studies that enrolled the patient population of interest and used current generation scanners and protocols of the noninvasive imaging technologies of interest. We excluded case-control studies, meeting presentations, and very small (<10 patients) studies.

Review Methods

We abstracted data from the included studies and used a bivariate mixed-effects binomial regression model for meta-analysis. We used the summary likelihood ratios and Bayes' theorem to calculate the post-test probability of having a benign or malignant lesion. We explored heterogeneity in the data with meta-regressions using standard methodology. We graded the strength of evidence supporting each major conclusion as high, moderate, low, or insufficient. The grade was developed by considering four important domains: the risk of bias in the evidence base (internal validity, or quality of the studies), the consistency of the findings, the precision of the results, and the directness of the evidence.

Results

We identified 41 studies of magnetic resonance imaging (MRI). The summary sensitivity of MRI was 91.7 percent (95% CI: 88.5 to 94.1%) and the summary specificity was 77.5 percent (95% CI: 71.0 to 82.9%). The estimate of accuracy was judged to be supported by a moderate to low strength of evidence (low for the estimate of specificity due to the lack of precision as reflected in the wide confidence interval). Bayes' theorem and the summary estimates of accuracy suggest that only women with a pre-MRI suspicion of malignancy of 12 percent or less will have their post-MRI suspicion of malignancy change sufficiently to suggest that a change in patient management may be appropriate.

We identified seven studies of positron emission tomography (PET). The summary sensitivity of PET was 83.0 percent (95% CI: 73.0 to 89.0%) and the summary specificity was 74.0 percent (95% CI: 58.0 to 86.0%). The estimate of accuracy was judged to be supported by a Low strength of evidence. Bayes' theorem and the summary estimates of accuracy suggest that only women with a pre-PET suspicion of malignancy of 5 percent or less will have their post-PET suspicion of malignancy change sufficiently to suggest that a change in patient management may be appropriate.

We identified 10 studies of scintimammography. The summary sensitivity of scintimammography was 84.7 percent (95% CI: 78.0 to 89.7%) and the summary specificity was 77.0 percent (95% CI: 64.7 to 85.9%). The estimate of accuracy was judged to be supported by a Low strength of evidence. Bayes' theorem and the summary estimates of accuracy suggest that only women with a pre-scintimammography suspicion of malignancy of 5 percent or less will have their post-scintimammography suspicion of malignancy change sufficiently to suggest that a change in patient management may be appropriate.

We identified 21 studies of B-mode grayscale ultrasound, six studies of color Doppler ultrasound, and seven studies of power Doppler ultrasound. For B-mode grayscale, summary sensitivity was 92.4 percent (95% CI: 84.6 to 96.4%) and the summary specificity was 75.8 percent (95% CI: 60.8 to 86.3%); for color Doppler, summary sensitivity was 88.5 percent (95% CI: 74.4 to 95.4%) and summary specificity was 76.4 percent (95% CI: 61.7 to 86.7%); for power Doppler, summary sensitivity was 70.8 percent (95% CI: 47 to 86.6%) and summary specificity was 72.6 percent (95% CI: 59.9 to 82.5%). These estimates of accuracy were all judged to be supported by a Low strength of evidence. Bayes' theorem and the summary estimates of accuracy suggest that only women with a pre-ultrasound suspicion of malignancy of 10 percent or less will have their post-ultrasound suspicion of malignancy change sufficiently to suggest that a change in patient management may be appropriate.

Conclusions

The use of noninvasive imaging, in addition to standard workup of women recalled for evaluation of an abnormality detected on breast cancer screening, may be clinically useful for diagnostic purposes only for women with a low (less than 12%) pretest suspicion of malignancy. When choosing which noninvasive imaging technology to use for this purpose, the evidence appears to suggest that diagnostic B-mode grayscale ultrasound and MRI are more accurate than PET, scintimammography, or Doppler ultrasound. The utility of these findings, however, depend on whether clinicians can identify women with a pretest suspicion of malignancy in the ranges necessary for the tests to affect management. Several of the expert reviewers of this report did not think this is currently possible.