LAUSR

2 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 Dear Editors: It is with great interest that we read the publication written by Salem et al: “Y90 Radioembolization Significantly Prolongs Time to Progression Compared With Chemoembolization in Patients With Hepatocellular Carcinoma.” This article gave results of the first randomized study comparing 2 different (radiovs chemo-) embolization approaches, with the primary objective significantly in favor of radioembolization, with a median time to progression (TTP) of 26 months versus 6.5 months after chemoembolization, with a hazard ratio of 0.122 (95% confidence interval, 0.027-0.557; P 1⁄4 .007). It should be noted, however, that overall survival (OS), although censored at the time of liver transplantation, did not differ between the treatment groups. We acknowledge that censoring at the time when a new therapeutic agent is introduced is a valid methodologic approach for evaluating a new agent, especially in the palliative setting, with physicians offering different lines of treatment after disease progression. However, the current situation completely differs from that scenario, given that both arms are using an approach able to downstage HCC for resection or serve as a bridge before liver transplantation. Both are curative treatments associatedwith highly prolonged OS in comparison to either radioembolization or chemoembolization given alone. In contrast with drugs that are introduced in the event of disease progression, resection and transplantation are typically performed before occurrence of progression, and this was indeed the case for 18 of 20 transplanted patients in the published study. In this case, transplantation is an event related to the therapeutic effect. Therefore, censoring at the time of transplantation introduces a major bias when assessing both TTP and OS. Most important, transplantation is currently proposed to patients likely to obtain the best results from either radioembolization or chemoembolization. Therefore, censoring at transplantationmeans that the best achievable results in terms of OS have not yet been obtained, which is clearly the opposite of what we want to achieve, and evaluation of the real OS of patients undergoing transplantation is not accurate. This bias may be high, given that median OS after transplantation postbridging can reach 76.6 months versus only 26.9 and 17.9 months after radioembolization administered alone for respectivelyBCLCCstageAand17.2 andBLCCstageBdisease. Analyzing OS not censored at transplantation is another means of evaluating chemoembolization or radioembolization outcomes in the neoadjuvant setting, as