It is recommended for intermediate stage HCC (BCLC B). But there is no consensus concerning treatment modalities. Recently several prognostic scores have been proposed to guide the treatment decision: ART, HAP, ABCR (EASL 2014, abstract A-627-0008-01729). Purpose: To evaluate and compare these three prognostic scores on a multicenter independent cohort treated by TACE. Methods: This retrospective study included Child-Pugh A or B patients with BCLC B HCC, BCLC A HCC (not eligible for curative treatment) and

BCLC C HCC with limited portal vein thrombosis, treated ATR cancer by TACE from 01/2007 to 01/2013, without complementary treatment (RF or graft), not involved in the development of ABCR score. To compare the three scores, we used an independent cohort: 153 patients, median age 68 years, BCLC A 17%, BCLC B 69%, BCLC C 14% treated in Marseille and Nancy. Cirrhosis was viral 40%, related to alcohol 43%, to a fatty liver disease 12%. Median survival in the three scores, overall effect of find more scores on survival time (Wald test). Results: Patients in the independent cohort were treated an average of 2.75 TACE. The response rate (EASL criteria) was 61%. Median follow-up was 19 months [17–23]. HAP score

distinguished four groups: HAP A 31 months [25–37] vs. HAP B 31 months [20–51] vs. HAP C 22 months [17–25] vs. HAP D 18 months [6–32], p = 0.0454, but the risk of death in HAP B and D groups were not significantly different from the reference HAP A group (respectively HR 0.88 [0.52–1.50], p = 0.640, HR 1.56 [0.81–2.99],

p = 0.1820). ART score distinguished two groups with different survival: ART (0–1.5) 27 months [23–37] vs. ART (≥2.5) 19 months [14–25 ], p = 0.0013, but the risk of death of the ART 4 group was not significantly different from the reference ART 0 group (HR 1.61 [0.81–3.21], p = 0.178) conversely ART 1 group (HR 3.26 [1.91–5.55], p < .0001). The ABCR score distinguished three groups with different survival: ABCR ≤ 0: 37 months [27–49] vs. ABCR [1–3]: medchemexpress 17 months [14–20] vs. ABCR ≥ 4: 8 months [6–18], p < 0.0001 . The risk of death of ABCR [1–3] and ABCR ≥ 4 groups was significantly increased compared to the reference ABCR ≤ 0 group (respectively HR 3.85 [2.46–6.02], p < .0001, HR 14.72 [6.57–33], p < .0001). Conclusion: In this multicenter mainly BCLC B HCC series, the distribution of patients according to the ART and HAP scores is inaccurate because it is not correlated with prognosis. The ABCR score better distributes unresectable HCC and therefore optimize treatment: continuation of TACE, systemic therapy or therapeutic trial. Key Word(s): 1.