Reporting and handling of equivocal imaging findings in diagnostic studies of bone metastasis in prostate cancer

Abstract

Background

Equivocal scanning results occur. It remains unclear how these results are presented and their management influence diagnostic characteristics.

Purpose

To investigate the reporting and handling of equivocal imaging findings in diagnostic studies of bone metastases, and to assess the impact on diagnostic performance of the methods used to analyze equivocal findings. The conceptual issue was reified based on two actual observations.

Material and Methods

A recent meta-analysis of bone metastases in prostate cancer was conducted and data were obtained from a large clinical trial with a true reference of bone metastasis, where diagnostic characteristics were calculated with equivocal scans handled by: removal; considered malignant; considered benign; and intention-to-diagnose.

Results

The meta-analysis included 18 trials where the median proportion of reported equivocal results was 27%. Eleven (61%) studies reported an equivocal option for the index test, 42% reported equivocal results and described how these were analyzed. The clinical trial included 583 prostate cancer patients with 20% equivocal results. The different methods of managing equivocal findings resulted in highly variable outcomes: sensitivity = 85%–100%; specificity = 78%–99%; and positive and negative predictive values = 44%–94% and 97%–100%, respectively. The diagnostic performances obtained using the four methods were differentially susceptible to the proportion of equivocal imaging findings and the prevalence of bone metastases.

Conclusion

Reporting of equivocal results was inadequate in bone imaging trials. The handling of equivocal findings strongly influenced diagnostic accuracy.