(P 318) DEVELOPMENT OF NOVEL OUTLINING GUIDELINES FOR NORMAL TISSUE STRUCTURES FOR LIMB SOFT TISSUE SARCOMA.

Objective: RT to soft tissue sarcoma of the extremities (STSE) improves local control rates. However, STSE survivors often develop significant toxicities (defined as ≥ grade 2 RTOG; grade 2+), which impact on quality-of-life. Past clinical trials reported grade 2+ subcutaneous fibrosis (48.2% vs 31.5% ), lymphoedema (23.2% vs 15.5%), bone fractures (11% vs 5%) and joint stiffness (23.2% vs 17.8%) associated with post-operative and pre-operative RT, respectively. Traditionally, RT plans are optimised to achieve maximum target coverage, whilst avoiding high doses to weight-bearing bones and intermediate to high doses to the normal tissue (NT) limb corridor. Novel NT outlining guidelines and atlas were developed, hypothesising that using these for RT planning will reduce RT toxicity. This study aims to examine the interobserver variations in lower limb NT outlining following the development of this novel approach.

Methods: Two STSE cases were selected as training cases: Case 1 - myxoid fibrosarcoma of the thigh and knee; case 2 - malignant peripheral nerve sheath tumour of the thigh. Target volumes had previously been outlined as part of their treatment courses. A set of reference volumes was outlined on each case by one radiation therapist/therapeutic radiographer (RTT) and peer-reviewed by a consultant radiation oncologist specializing in STSE. NTs were then outlined by 7 additional observers (5 RTT and 2 clinicians) following the novel guidelines and atlas and re-outlined 6 months later by the reference volume expert to test intra-observer variability. The novel NT for STSE completed as part of this work were the muscle compartments of thigh (MCT), the neurovascular bundle (NVB), joints and bones (figure 1). Dice coefficients (DICE) were calculated for individual NT volumes against the gold-standard. The Kruskal-Wallis test was performed to detect statistically significant differences in DICE between NT volumes. A preliminary analysis of DICE was conducted after 3 observers (1 RTT; 2 clinicians) completed the cases and an amendment to the guidelines was carried out.

Results: Figure 2 shows the boxplot for DICE results above and below. There was a good agreement for the anterior, posterior, gluteal and adductor MCT, femur, femoral NVB and knee. Although there were differences in the DICE scores between RTTs and clinicians, these were not statistically significant (p=0.1). Differences in DICE were found for the lateral rotator MCT (p=0.008) which were corrected after training programme and guidelines refinement. Significant statistical (p=0.008) differences have been highlighted in the deep thigh and sciatic NVB and further corrections to guidelines to improve these 2 structures are currently under discussion.

Conclusion: Our results identify that novel NT outlining guidelines for STSE are reproducible between observers and, in a multi-professional environment, RTT and clinician scores are consistent. This work has also identified areas of refinement and modification required to improve the guidelines, particularly for the deep thigh and sciatic NVBs. Future work will enable us to attain dose-volume constraints for these structures which will also result in further refinements of the guidelines depending on significant areas to apply dose-volume constraints.

The first author is clinical doctoral research fellow CDR-2018-04-ST2-004, funded by HEE/ NIHR.