onc brain

The longer read

Contouring guidelines are easy to undervalue because they report no survival curve, but for an emerging indication they determine whether the outcome data that follow are interpretable at all. Renal SABR arrived at its evidence base through a route that made this problem worse than usual: IROCK's pooled international series and the FASTRACK-II phase 2 established safety and local control across centers that were each using their own dose, fractionation, immobilization and target conventions. The efficacy signal survived that heterogeneity, which is reassuring, but it also means nobody can say which delineation practice the reported local control rates actually reflect. This document is the correction, and its most useful output is not the consensus contour itself but the map of where experienced readers disagree.

That map is the substantive finding. Agreement was not uniform. The two scenarios where the panel diverged, the >10 cm tumor with IVC thrombus and the post-RFA cavity, are precisely the two where the target's boundary is a judgment rather than an observation: how far superiorly to chase thrombus, and how much of an ablation defect is at risk. Case 4 fell to a median DSC of 0.75 with an IQR reaching 0.60, which for a single-organ target contoured by people selected for having done at least ten of these is a wide spread. The ANOVA finding sharpens it: DSC and HD varied by case but not by participant, so this is difficulty inherent to the anatomy, not a few outlier readers. Only MDA varied by participant, at P = .03. A reader should take from this that the guideline's value concentrates in exactly the cases that are hardest, and that its Case 2 and Case 3 statements mostly ratify what people were already doing.

Both contested statements resolve toward a larger volume, and the panel's justification is asymmetry of consequence rather than evidence of subclinical extension. Missing cavity risks recurrence outside the treated field that is difficult to re-treat; including it costs dose to adjacent bowel. The guideline handles that tension by ranking OAR constraints above target coverage and endorsing deliberately heterogeneous coverage with a steep gradient, with the hotspot placed inside the iGTV. This is a coherent position, and it is worth noticing that it hands the actual compromise back to the planner rather than specifying it. Two centers can both follow this guideline and deliver materially different coverage to the same large post-RFA cavity abutting large bowel.

The renal substructure decision is the most interesting piece of reasoning here, and it argues against the reflex that more contours are safer. Ipsilateral parenchymal dose correlates with renal function; renal cortex as a distinct substructure and the hilum do not, separately. The panel therefore defines the parenchyma for use and explicitly refuses to make the hilum dose-limiting, on the grounds that optimizing away from an unvalidated structure would push dose into the one that does matter. AQuOS-II may settle it.

Two caveats bound how far this transfers. The 5 mm PTV margin is conditioned on a specific workflow, supine vacuum-cushion immobilization, 4D-CT, IV contrast, daily CBCT, and is not portable to a center missing any of those. And the recommendations are consensus opinion, unvalidated against outcomes, with the panel itself noting that CT-only images likely inflated the variance it measured and that participants never recontoured afterward, leaving the guideline's own effect on agreement untested.