BackgroundSelective internal radiation therapy (SIRT) using yttrium-90 resin microspheres was compared to sorafenib 400mg bid in a phase 3 randomised trial (SARAH) in the treatment of intermediate or advanced hepatocellular carcinoma (HCC) not amenable to curative treatment. The trial did not show a survival benefit of SIRT over sorafenib in the intention to treat (ITT) population. The effectiveness of SIRT may depend on the tumour-absorbed dose, which can be predicted with the analysis of SPECT/CT imaging during each patient work-up, before the administration of SIRT. In this post hoc analysis we explored the comparative effectiveness of SIRT and sorafenib in a subgroup of patients defined by their predicted tumour-absorbed dose. MethodsCox proportional hazards regressions were conducted in the ITT population of the SARAH trial. Since predicted tumour-absorbed dose was only available for SIRT, the comparisons between SIRT at a given dose and sorafenib were not randomised. Inverse probability of treatment weighting (IPTW) using propensity scores was used to account for potential confounding by differences in prognostic factors between the treatment arms, with the sorafenib sample reweighted to match the SIRT patients. A cut-off value of 100Gy was used for tumour absorbed dose, approximating the median dose, with 120Gy used in a sensitivity analysis. ResultsFor patients with a predicted dose ≥100Gy, the hazard ratio (HR) for overall survival (OS) from the unweighted sample was 0.70 (95% CI: 0.50-0.98, p=0.04). After reweighting, the HR was 0.74 (95% CI: 0.51-1.04, p=0.09). Predicted mean OS was 22.5 months (mos) for SIRT vs 17.9 mos for sorafenib. Results were similar with a 120Gy cut-off: the HR for OS was 0.76 (95% CI 0.52-1.10, p=0.14). Among patients who received subsequent curative therapy post-SIRT, 11/12 were alive at the end of follow-up (median 26.6 mos, range 16.0-34.8) and only one had a predicted dose <100Gy. ConclusionsThe analysis suggests that HCC patients may derive a meaningful benefit from treatment using SIRT with a predicted dose ≥100Gy compared to sorafenib. This may inform personalised treatment selection and clinical trial design. Clinical trial identificationNCT01482442. Legal entity responsible for the studySirtex Medical UK Ltd. FundingSirtex Medical UK Ltd. DisclosureN.S. Hawkins: Honoraria (institution), Advisory / Consultancy, Research grant / Funding (institution): Sirtex. P.J. Ross: Honoraria (self), Travel / Accommodation / Expenses: Amgen; Honoraria (self), Advisory / Consultancy, Travel / Accommodation / Expenses: Bristol-Myers Squibb; Advisory / Consultancy: Celgene; Travel / Accommodation / Expenses: Bayer; Honoraria (self), Advisory / Consultancy: Pierre Fabre; Honoraria (self): Roche; Honoraria (self), Travel / Accommodation / Expenses: Servier; Research grant / Funding (institution): Sanofi; Honoraria (self), Advisory / Consultancy: Shire; Honoraria (self), Advisory / Consultancy, Travel / Accommodation / Expenses: Sirtex. D.H. Palmer: Advisory / Consultancy, Travel / Accommodation / Expenses: Bayer; Advisory / Consultancy, Travel / Accommodation / Expenses: Eisai; Advisory / Consultancy, Travel / Accommodation / Expenses: Bristol-Myers Squibb; Advisory / Consultancy, Travel / Accommodation / Expenses: MSD; Advisory / Consultancy, Travel / Accommodation / Expenses: AstraZeneca. V. Vilgrain: Honoraria (self), Advisory / Consultancy, Research grant / Funding (institution), Travel / Accommodation / Expenses: Sirtex. All other authors have declared no conflicts of interest.