Long Term PSA Stability and Predictive Factors of Failure after Permanent Seed Prostate Brachytherapy

Audrey Tetreault-Laflamme, MD, FRCPC, Juanita Crook, MD, FRCPC, Jeremy Hamm, PhD, Tom Pickles, MD FRCPC, Mira Keyes, MD, FRCPC, Michael McKenzie, MD, FRCPC, Howard Pai, MD, FRCPC, Francois Bachand, MD, FRCPC, James Morris, MD
FRCPCPP07 Presentation Time: 9:54 AM


The Phoenix definition of biochemical failure (BF) (nadir+2) may overestimate cure rates after low dose rate prostate brachytherapy (LDR-PB). The purpose of this study is to assess long term PSA stability after LDR-PB and predictive factors of eventual BF for those with a slowly rising PSA.

2772 low or intermediate risk prostate cancers underwent Iodine-125 LDR-PB monotherapy between 1998 and 2010. 49.7% received androgen deprivation (ADT) prior to LDR-PB (treatment policy: 6 months). Patients with less than 36 months follow-up were excluded (n=433). Clinical characteristics, dosimetric parameters and sequential PSA readings were retrieved from a prospective provincial database. A rising PSA was considered to be PSA ≥ 0.2 ng/mL with an increase ≥ 0.1 ng/mL over previous 2 years. The Phoenix definition was used to identify BF. Patients were classified as (1) stable PSA (cured), (2) rising PSA (without BF) or (3) BF. The three groups were compared according to clinical, dosimetric and post-treatment parameters. Multivariate analysis was performed on the cured and failed groups to determine variables predicting for failure. Logistic regression model was applied with cross validation to test for model accuracy. ROC curves were obtained for patients with and without ADT to determine predictive cut-offs for BF.

Median follow-up is 89 months (37-199); median age at implant 66 years(43-84). Majority of patients (80.7%) had clinical stage T1-T2a, 55% had Gleason score ≤ 6 and median baseline PSA was 6.5 ng/mL (0.3-40 ng/mL). 59% were intermediate risk. Among the 2339 patients analyzed, 2004 (85.7%) had a stable PSA and were considered cured [median PSA at 60 months (PSA-60): 0.04ng/mL], 145 (6.2%) had a rising PSA (PSA-60: 0.27ng/mL) and 190 (8.1%) had BF. PSA nadirs for the 3 groups were respectively 0.03 (cured), 0.16 (rising PSA) and 0.51ng/mL (BF) (p<0.0001). For patients with no prior ADT, the variables associated with failure are PSA nadir (OR: 20.6 p<0.0001) and PSA-60 (OR: 18.3 p<0.0001). If the model is applied to the rising PSA group, using a PSA-60 cut-off of 0.3ng/mL (sensitivity:85%, specificity: 98.1%), the risk of failure is 9.8% (8/82) for patients not having received ADT. For patients who received ADT, the predictive factors of failure are PSA-60 (OR:53.9 p<0.0001) and T-stage (OR:0.25 p=0.0008). Using this model and a PSA-60 cut-off of 0.1ng/mL (sensitivity: 85%, specificity: 92.9%), the predicted risk of BF in the rising PSA group is 53.7% (36/56). Taking into account the two predictive models, the anticipated cure rate for the entire cohort is 89.7%.

Patients treated with LDR-PB monotherapy and whose PSA-60 is ≤ 0.3ng/mL are highly likely to be cured even if they experienced a slight PSA rise. However, for patients who also received ADT, a stricter cut-off of 0.1ng/mL may be appropriate.