died. The different prognostic factors were assessed by univariate analysis with the Mann-Whitney U and Kendall A-B tests. RESULTSOf the evaluated 51 inpatients, eight died (16%) and 43 survived (84%). The median (range) age was 63 (17-85) years and the median time from the onset of the symptoms until the admission to the emergency room was 7.8 (1-60) days. The mean hospital stay was 33 (2-90) days and 17 patients were admitted to the intensive-care unit for a mean of 4.5 days. There was no statistically significant difference between the groups. Body surfaces involved were the scrotum in five patients (10%), the penis and scrotum in 11 (22%), the scrotum and perineum in 30 (59%) and the abdominal wall in five (10%). There was no statistically significant difference in the distribution in those who survived or died ( P = 0.131). The median age of 60 (17-81) years in the survivors was significantly lower than that of 73.5 (50-85) years in those who died ( P = 0.02). There was no significant difference ( P = 0.06) between the number of repeated debridements in the survivors (3.23) and those who died (5.25). The mean (range) FGSI score for survivors was 6.7 (0-14), vs 8.7 (6-13) for those who died ( P = 0.12). The only laboratory variables associated with death were serum bicarbonate ( P = 0.04) and serum sodium ( P = 0.02) levels. CONCLUSIONS

e16541 Background: PSA is a widely used biomarker for monitoring outcome in mCRPC. Current treatment guidelines do not consider PSA progression before at least 12 weeks of treatment, and recommend treatment continuation in pts experiencing progression by PSA only, without radiographic or clinicall progression. We aimed to evaluate the prognostic value of a PSA progression in mCRPC pts treated with first-line therapy. Methods: We analyzed the value of a PSA progression (PSAProg) at cycle 5 in pts treated in the COU-AA-302 trial. PSAProg was defined as an increase ≥ 25% and ≥ 2 ng/mL from baseline, confirmed by a second reading. Radiographic progression (RadProg) was defined as per PCWG2 criteria for bone scan or RECIST progression. Survival and radiographic progression-free survival (rPFS) from the time of PSAProg was calculated using Kaplan-Meier estimates. Cox-regression models were used to evaluate the impact of PSAProg and treatment arm on survival. Results: 1088 pts were randomized in the COU-AA-302 trial. 908 pts (83.5%) had valid baseline and cycle 5 PSA values. Of these, 222 (24.4%) pts experienced PSAProg, which was confirmed in 195 (21.5%) pts; 45/479 (9.4%) of abiraterone and 150/429 (35%) of placebo-treated pts. A confirmed PSAProg was associated with shorter survival (37.8 vs 26m; HR: 1.8; p < 0.001) and rPFS (13.9 vs 5.6m; HR:2.1; p < 0.001). Median survival from the time of PSAProg was 22.2m (95%CI:18.9-25.4). Abiraterone-treated pts had a significantly longer survival from the time of PSAProg (23.1 vs 17.4m; HR: 0.64; p = 0.018). 111 pts (56,9%) experienced PSAprog without RadProg; in pts with PSAprog only, median time to RadProg was 7.4m (95%CI:7.1-12.9). No differences in rPFS from the time of PSAprog were observed in abiraterone vs placebo-treated pts (HR: 0.99; p = 0.963). Conclusions: 9.4% of abiraterone-treated pts in the COU-AA-302 trial experienced PSAProg at 12 weeks (cycle 5 day 1). PSAProg was associated with significantly worse survival. Abiraterone increased survival from the time of PSAProg, which supports its continued use in pts with PSA progression only at 12 weeks. This study was carried out under YODA Project #2018-3011.

No abstract

53 Background: HRQoL is a relevant endpoint in trials in advanced prostate cancer. An association between HRQoL and OS has been reported. The Functional Assessment of Cancer Therapy-Prostate (FACT-P) is a validated HRQoL PRO in mCRPC. Methods: We evaluated the association between FACT-P and OS in the COU-301 and COU-302 trials (abiraterone vs placebo in mCRPC pts). FACT-P scores, sub-scores (physical (PWB), emotional (EWB), functional (FWB), social (SWB) well-being, prostate cancer subscale (PCS) FACT-G and the Trial Outcome Index (TOI) were calculated. A decrease in 3 (PWB, EWB, SWB, FWB, PCS), 9 (FACT-G, TOI) or 10 (FACT-Total) points after 3 cycles was considered clinically relevant. The association between FACT-P and OS was evaluated with Kaplan-Meier, Cox-regression models and c-indices. Results: 2,177 pts (COU-301: 1,121 /COU-302: 1,056) had valid baseline (BL) FACT-P scores. Mean BL score was 106.6 (COU-301) and 122.3 (COU-302). BL total scores were associated with OS in both COU-AA-301 (p<0.001) and COU-302 (p<0.001), independent of treatment. All FACT-P sub-scores except SWB were associated with OS (Table). A decrease in FACTP scores was associated with decreased OS in COU-301 (19.6 vs 14.2m; HR: 1.8; p<0.001) and COU-302 (34.4 vs 27.7m; HR: 1.3; p=0.009) datasets. Conclusions: BL FACTP scores (except SWB subscale) are significantly associated with outcome. Early declines in HRQoL can be observed and are associated with worse outcome. Prospective evaluation of the significance of changes in HRQoL is needed. YODA Project 2018-3745.[Table: see text]

No abstract