Tools for Identifying Pathologically Insignificant Prostate Cancer Are Inaccurate In Unscreened Men


Key Points

  • Proposed tools for identifying clinically unimportant prostate cancer were found to be inaccurate in unscreened populations.
  • The tools have low accuracy in screened populations, as well.

In a study reported in British Journal of Cancer, Shaw et al assessed the accuracy of several reported criteria for identifying insignificant prostate cancer for active surveillance in a population of unscreened men. None of the examined tools provided sufficient discrimination of insignificant cases.

Study Details

The study involved a prospectively collected case series of 848 unscreened patients treated with radical prostatectomy between July 2007 and October 2011 at an English tertiary care center. Tumor volume was assessed by pathologic examination. Five criteria for insignificant prostate cancer in screened populations (Tosoian, Adamy, van den Bergh, Whitson, and Soloway) were assessed according to three definitions of insignificant cancer: a “classical” definition (organ-confined tumors of < 0.5 cm3 Gleason 3+3 with no Gleason 4 or 5); the European Randomised Study of Screening for Prostate Cancer (ERSPC) definition (organ-confined Gleason 3+3 tumors with no Gleason grade 4 or 5, index tumor volume ≤ 1.3 cm3, and a total tumor volume of ≤ 2.5 cm3); and an “inclusive” definition (organ-confined Gleason 3+3 tumors). Performance was compared with the D’Amico low-risk criteria of prostate-specific antigen ≤ 10 ng/mL, Gleason 3 +3, and cT1-2a.

Of the 848 patients, 415 had Gleason 3+3 disease on biopsy. Of these, 32.0% had extraprostatic extension and 50.2% were upgraded, and one patient had positive lymph nodes. A total of 206 (24%) were D’Amico low risk; of these, 143 had more than two biopsy cores involved.

Poor Accuracy

Receiver operator characteristic area under the curve (AUC) values for the five tools in addition to any and all criteria ranged from 0.42 to 0.60 for the “classical” definition, 0.45 to 0.57 for the ERSPC definition, and 0.50 to 0.59 for the “inclusive” definition. As noted by the investigators, these findings must be considered in light of the fact that the AUC value for a random act, such as a coin toss, is 0.50. No significant difference in discriminative power was found for any tool vs the D’Amico low-risk criteria.

Poor performance was also found when the five criteria, any criteria, and all criteria were assessed according to insignificant cancer definitions in a screened U.S. cohort. Receiver operator characteristic AUCs ranged from 0.61 to 0.68 for the “classical” definition, 0.58 to 0.64 for the ERSPC definition, and 0.55 to 0.62 for the “inclusive” definition.

The investigators concluded, “None of the tools evaluated has adequate discriminative power in predicting insignificant tumour burden. Accuracy is low in PSA-screened and -unscreened populations…. Detection of a wider size range of prostate tumors in the unscreened may contribute to relative inaccuracy.”

G.R. Shaw, MD, of Cambridge University Hospitals NHS Foundation Trust, is the corresponding author for the British Journal of Cancer article.

The study was supported by the National Institute for Health Research and National Cancer Research ProMPT Collaborative.

The content in this post has not been reviewed by the American Society of Clinical Oncology, Inc. (ASCO®) and does not necessarily reflect the ideas and opinions of ASCO®.