ESTRO 2024 - Abstract Book

S1479

Clinical - Lower GI

ESTRO 2024

IOV was high amongst radiologists in cases 1-4. Cases 1 and 3 were intraluminal recurrences, impairing radiological visibility and consistency. Case 2 was a recurrence located within an abscess, in which the variation is explained by differing interpretations of GTV (i.e., whole abscess or solid tumour within the abscess), rather than spatial disagreement. Variation in case 4 is explained by a second lesion that some interpreted as GTV. Homogeneity within the lesion was high. Although IOV was low in case 12, a second PET+ lesion was missed by all radiologists. For group comparisons of overall IOV, cases 1-3 were excluded due to the high variation in radiologists, making the gold standard unreliable. A significant improvement in IOV is seen when comparing all GTV- vs GTV+ delineations, although an effect could not be proven in individual cases. Median (IQR) SDSC increased from 0.85 (0.67-0.92) to 0.90 (0.90-1.00) (p=0.008), DSC from 0.72 (0.63-0.80) to 0.77 (0.67-0.82) (p=0.029) and HD98% decreased from 9.5mm (6.0-15.6) to 6.0mm (4.2-12.3) (p=0.018)). In addition, a significantly higher IOV was observed in patients with anastomotic recurrences or tumours located within fibrosis. IOV was lower in lateral or nodal recurrences. No significant differences in IOV were observed amongst individual clinicians.

Examples of high, average and low radiological IOV delineations (green) with respective delineations performed by GTV+ (red) and GTV- (blue) are shown in figure 1.

A clinical difference in GTV delineations was seen in 6/14 cases (1, 5, 8-10, 12), categorized into better coverage of surgical resection margins at risk for irradical resection (5, 8, 10), reduction of GTV volume (1, 5, 9) and improved GTV coverage (5, 8). In case 12 however, a second PET+ lesion was missed by 4/6 ROs in GTV+ (0/5 GTV-), possibly due to GTV miss by radiologists.