Résumé
3515 Background: Hepatic arterial infusion (HAI) of chemotherapy (CTx) has been proposed as a potential treatment to decrease the risk of hepatic recurrence after resection of colorectal liver metastases (CRLM). Randomized controlled trials (RCTs) led to controversial results and meta-analyses are inconclusive. We conducted a comparative phase II RCT to assess the efficacy of adjuvant HAI oxaliplatin with concomitant intravenous (IV) CTx in pts at high risk of hepatic recurrence (≥4 resected CRLM). Methods: Pts who underwent curative-intent surgery of at least 4 CRLM after preoperative IV CTx were randomly assigned (1:1) to receive adjuvant IV fluorouracil/leucovorin (LV5FU2) combined with oxaliplatin (85mg/m2) by HAI (HAI-IV arm) or IV route (IV arm) every 2 weeks, for at least 3 months to achieve a minimal duration of perioperative CTx of 6 months. The primary endpoint was the hepatic recurrence free survival (h-RFS). Secondary endpoints included the feasibility of delivering at least 4 CTx cycles, toxicity (NCI-CTCAE 4.0) including HAI catheter-related complications, RFS, overall survival (OS) and recurrence pattern. 108 randomized pts were required to detect a gain in 18-month h-RFS from 30% to 50% with a one-sided α risk of 10% and 95% power. The primary endpoint was analyzed with a Cox regression model, with adjustment for stratification factors (tumor response to preoperative CTx and number of resected CRLM). The study was prematurely stopped after randomization of 99 pts due to recruitment issues and HAI catheters market withdrawal. Results: Between Jun 2015 and Dec 2020, 99 pts (median age, 62; median number of CRLM, 5) were randomly assigned in HAI-IV arm (50 pts) or IV arm (49 pts). After a median follow-up of 56 months, the intent-to-treat analysis showed a significantly longer h-RFS in the HAI-IV arm compared to the IV arm (median 25 [16-37] vs 12 months [8-21]; HR 0.598 [95%CI 0.379-0.944], p = 0.027). 5-year h-RFS was 32% [19-45] in the HAI-IV arm vs 13% [4-27] in the IV arm, respectively. Median and 5-year OS were 74 months [51-74] and 60% [43-74] in the HAI-IV arm vs 54 months [37-69] and 46% [29-61] in the IV arm (HR 0.551 [0.299-1.015], p = 0.056). The median number of postoperative CTx cycles was 6 in both groups with 86% and 78% of patients having received at least 4 cycles in HAI-IV and IV arms (p = 0.7), respectively. Grade 3-4 toxicity rate was 58% in the HAI-IV arm vs 31% in the IV arm (p = 0.01). Overall, nine pts (18%) experienced HAI-related complications. No toxic death occurred in both arms. Conclusions: Despite a higher but acceptable toxicity, combination of HAI oxaliplatin with IV LV5FU2 significantly improved h-RFS after curative resection of CRLM in high-risk patients. This combined treatment should be considered as a valid option for these patients. Clinical trial information: NCT02494973 .