A critical analysis of the interpretation of biochemical failure in surgically treated patients using the American Society for Therapeutic Radiation and Oncology criteria

Abstract

Purpose: The use of prostate specific antigen (PSA) to indicate biochemical failure has become an accepted procedure to measure the effectiveness of therapy. Because long-term randomized studies comparing radiation modalities to radical prostatectomy are not available, use of biochemical recurrence as a surrogate measure of efficacy is increasing. Unfortunately, the definition of failure is not uniform among therapies. We evaluate how the American Society for Therapeutic Radiation and Oncology (ASTRO) criteria affect the interpretation of failure when applied to radical prostatectomy.

Materials and methods: We retrospectively reviewed data from 2,691 men who underwent anatomical radical prostatectomy for localized disease between 1985 and 2000. All patients had regular followup visits ranging from 6 months to 15 years (mean 6). No patients were treated with radiation or hormonal therapy preoperatively or postoperatively until clinical recurrence. Biochemical failure was defined as any measurable PSA 0.2 ng./ml. or greater. We evaluated how elements of the ASTRO criteria affected the failure rate when applied to this series. We looked at 1) backdating the failure date to the midpoint between nadir and first PSA greater than 0.2 ng./ml., 2) early censoring if only 1 or 2 increasing values were available and 3) defining failure after 3 consecutive PSA increases and backdating failure time (midpoint of nadir and first PSA increase).

Results: Using actuarial analysis of the data defining failure as the first PSA 0.2 ng./ml. or greater, biochemical freedom from failure at 5, 10 and 15 years was 85%, 77% and 68%, respectively. In contrast, when backdating was used in this series, almost all failures occurred early with rare late failures (freedom from failure 82%, 80% and 80% at 5, 10 and 15 years, respectively). The difference in failure became even more pronounced when ASTRO criteria were applied requiring 3 consecutive increases, and backdating failure to the midpoint between nadir and first PSA (freedom from failure 90%, 90% and 90% at 5, 10 and 15 years, respectively).

Conclusions: The application of ASTRO criteria to a mature series of surgically treated patients with localized prostate cancer produced an apparent improvement in the probability of being biochemically free of disease at 15 years from 68% to 90%. Until prospective trials comparing these different therapies become available, caution should be exercised when interpreting outcomes between series due to the inherent differences in definition of biochemical failure.