Follow-up PSA testing after surgery for low-risk prostate cancer
Posted Aug 01 2010 12:00am
One of our regular readers has suggested we comment on a recent paper questioning the need for annual, lifelong PSA testing after radical prostatectomy for patients treated for low-risk prostate cancer.
The recent paper by Tollefson et al. offers a retrospective analysis of data on the long-term follow-up of > 2,000 patients treated surgically for low-risk, localized disease between 1994 and 2004 at the Mayo Clinic.
In this study, the authors defined low-risk disease as pathological stage pT2c or less, Gleason score 6 or less, negative lymph nodes, negative surgical margins, and a presurgical PSA level of < 10 ng/ml. Patients who received any form of neoadjuvant or adjuvant therapy were excluded from the analysis. Biochemical failure was defined as a PSA level > 0.4 ng/ml and PSA levels < 0.15 ng/ml were considered to be undetectable.
The results of this study showed the following:
According to the abstract, 2,219 patients met the criteria for low-risk disease, but the full text of the paper states that only 93 percent of these patients actually had a biopsy Gleason score of 6 or less, and 14/2,219 (0.6 percent) had a biopsy score of 8 to 10, implying that about 6.4 percent or about 150 had a Gleason score of 7, so why were these patients included in the data set?
Median age at surgery was 61 years (range 33 to 84).
Approximately half of the patients received their surgery before 2000.
Median follow-up was 75 months (range 3 to 161 months).
Only 142/2,219 patients (6.4 percent) met the specified criterion for biochemical failure during the course of follow-up.
200/2,219 patients (9.0 percent) has a detectable PSA > 0.15 ng/ml but never had a PSA ≥ 0.4 ng/ml (and therefore never met the study criterion for biochemical failure).
2/2,219 patients (0.1 percent) actually died of prostate cancer.
84/2,193 patients (3.8 percent) with an undetectable PSA at 1 year post-surgery ultimately experienced bichemical failure.
59/2,117 patients (2.8 percent) with an undetectable PSA at 2 years post-surgery ultimately experienced biochenical failure.
For patients with a PSA level < 0.4 ng/ml at 5 years after surgery, the biochemical failure rates over the next 1, 3, and 5 years were 0.0, 0.7, and 1.3 percent, respectively.
For patients who had an undetectable PSA for 1, 3, 5, and 10 years after surgery, biochemical failure rates within the following 12 months were 0.2, 0.4, 0.0, and 0.0 percent, respectively.
2,157/2,219 patients (97.2 percent) never needed or received any form of follow-up treatment for prostate cancer after their initial surgery.
Tollefson et al. conclude that, in low-risk patients, the risk of biochemical failure is inversely proportional to time for which the PSA is undetectable after radical prostatectomy. They go on to suggest that taking PSA levels every 2 years should be sufficient to identify the majority of low-risk patients who experience biochemical progression.
The “New” Prostate Cancer InfoLink would certainly agree that annual PSA testing is probably unnecessary in the majority of patients who are treated surgically for low-risk disease if their PSA is undetectable after surgery. However, the key question is how long should annual PSA testing be carried out before the patient can be switched to testing every 2 (or perhaps even every 3) years. Based on this paper, it would seem likely that even low-risk patients should all receive at least annual testing for 3 years after surgery.
We are puzzled by some other issues associated with this paper. Most particularly, the generally accepted definition of an undetectable PSA is < 0.1 ng/ml (not < 0.15 ng/ml) and the usually accepted criterion for biochemical failure after surgery is a PSA level > 0.2 ng/ml (not 0.4 ng/ml). One wonders why the authors felt the need to use different crieria, and what impact the use of the “normal” criteria would have had on their results. It should also be noted that biochemical recurrences are well known to occur as long as 25 years after radical prostatectomy as first-line treatment for prostate cancer, so the follow-up data for this analysis is relatively short.
Perhaps the item that stands out most in consideration of these data, however, is the apparently compelling suggestion that many of these patients may never have needed treatment at all. With less than 7 percent of these patients showing any sign of even biochemical progression, and only 0.3 percent demonstrating systemic disease after their first-line treatment, one is forced to wonder what would have happened if this group of patients had all been carefully monitored according to an active surveillance protocol which suggested that treatment would only be appropriate if the patients exhibited a PSA doubling time of less than (say) 18 months.