Use of biopsy detail to identify subgroup risk in high-risk patients with prostate cancer.

Abstract

169 Background: High Risk (HR) prostate cancer (CaP) patients (pts) are a heterogeneous group. A recursive partitioning analysis (RPA) was performed to determine if tissue involvement could identify patients at risk for biochemical failure (BF). Methods: Between 1992-2009, 608 HR CaP pts with complete prostate biopsy detail data (% overall tissue involvement (%TI), GP45 (weighted average of %TI for Gleason Pattern 4 or 5), Gleason Score (GS)) underwent RT with/out ADT. Primary endpoint was freedom from biochemical failure (FFBF). RPA based on time to BF was used to determine cutpoints for each prognostic variable. Multivariable (MVA) RPAs were used to assess interactions of significant variables (p<0.05) from univariate analysis (UVA). Results: Median follow up for PSA measurement was 45 months (3-193 mo). UVA RPA revealed significant cutpoints for FFBF for Tstage (higher risk of BF(HRBF) for T2c,T3b,T3c,T4,Tx vs lower risk of BF(LRBF) for T1,T2a,T2b,T3a, (p<0.001)); GS(HRBF 3+4/5, 4+5 vs LRBF 2-6, 2+3/4, 5+3/4/5 (p=0.011); iPSA (LRBF ≤37 vs HRBF >37 ng/ml (p<0.001)); dose (HRBF ≤74.9 vs LRBF >74.9 Gy); RT type(LRBF IMRT vs HRBF 3D-CRT, p=0.032), %TI (LRBF ≤53% vs HRBF >53% (p<0.001)); GP45 (LRBF ≤33% vs HRBF >33%(p<0.001)). Overall, FFBF at 5 yr was 79.2% (95%CI 75.0-82.8). For MVA RPA, splits are same as UVA unless noted. MVA RPA excluding GP45 and %TI showed 5 groups (5 yr FFBF 45.9%, 66.8%, 74.2%, 84.2% and 89.5%; N 16, 195, 69, 20 and 308, respectively). Tstage, iPSA, GS and RT type had significant interactions, with highest risk group including PSA >44ng/ml, low Tstage and 3D-CRT, and lowest risk group including PSA ≤44 ng/ml, lower risk Tstage, and lower risk GS. Including %TI in the RPA, significant interactions were between %TI (2 splits), iPSA and GS (5 yr FFBF 50.0%, 62.5%, 71.9%, 80.4% and 92.7%; N 80, 24, 65, 233, and 206, respectively). Highest risk group included pts with %TI >53%; lowest risk included pts with %TI <17%, iPSA <40 ng/mL, and GS lower risk. Using GP45 instead of %TI, 3 groups were found: high GP45; low GP45 and iPSA ≥40; low GP45 and iPSA<40 ng/ml, (5 yr FFBF 58.9%, 67.7% and 85.5%; N 103, 63 and 442, respectively). Conclusions: Pts with > 53% TI, > 33% GP45 and iPSA > 40 ng/ml are at highest risk of failure, and should be considered for most aggressive treatment.