onc brain

The longer read

RTOG 1112 asked whether liver SBRT extends survival when added to first-line sorafenib in locally advanced HCC that local-regional therapy could not address, a population in which intrahepatic progression is the usual way the disease declares itself. On its own primary test the answer is no: OS HR 0.77 with a 90% CI of 0.59-1.01 and a one-sided P=.06. Every other efficacy read points the same way, though. Median OS moved from 12.3 to 15.8 mo, the stratification-adjusted OS HR was 0.72 (95% CI 0.52-0.99), and the PFS HR of 0.55 had an upper bound of 0.75. A trial that misses its primary endpoint while its secondary estimates cohere is more plausibly underpowered than null, and here there is a concrete reason for underpowering: accrual closed early when the systemic standard changed.

That reading still has limits a skeptic would press. The adjusted OS analysis is a secondary lens, and a reader should not promote it to the primary result because it crossed 0.05. The analysis covers 177 eligible pts out of 193 randomized, a departure from intention-to-treat whose direction of bias the source does not let us judge. The source also does not state how progression was assessed or whether imaging review was blinded, and post-SBRT liver changes can complicate response calls, which matters because PFS carries the strongest signal.

For a radiation oncologist the population is the point. Macrovascular invasion was present in 74.0%, the phenotype in which tumor thrombus threatens liver function and where systemic therapy alone has historically struggled to hold intrahepatic disease. The PFS gain is consistent with SBRT controlling the dominant liver burden, but local control and patterns of failure are not reported in source text, so that mechanism is inferred rather than shown. Dosing was personalized across 27.5 to 50 Gy in 5 fractions, which speaks to feasibility across a range of liver function but prevents any dose-response read. Toxicity did not obviously rise: grade 3+ treatment-related AEs were 47% vs 42% (P=.52), with 1 treatment-related death after SBRT plus sorafenib against 2 with sorafenib alone. The quality-of-life difference, 35% vs 10% improved at 6 mo, sits on 17 and 20 evaluable pts and should be read as hypothesis only.

Against prior liver-directed radiation data, the result lines up with START, which tested TACE plus RT against sorafenib in macrovascular invasion and favored the radiation-containing arm, and departs from the radioembolization trials SARAH, SIRveNIB and SORAMIC, none of which showed an OS gain against or alongside sorafenib. RTOG 1112 is a multicenter phase 3 test of stereotactic external-beam delivery in this setting, so its direction deserves attention even with the missed P value.

The open problem is transfer. The increment was measured against sorafenib, and the change in systemic standard that ended accrual means its size on a newer backbone is unknown. A more active systemic regimen could shrink the benefit by controlling liver disease itself, or enlarge it by holding extrahepatic disease long enough that intrahepatic control decides survival. That question, more than this trial's P value, determines whether liver SBRT becomes routine in first-line advanced HCC.