Session Item

A national guideline for treatment planning in head and neck cancer has continuously been developed, improved and implemented . Due to differences in local TPS, linacs, and specific planning practices, plan quality may still vary between individual planners following the guideline. We analyze inter-planner variations for a group of planners following the guideline, and suggest use of quantitative metrics for further plan exploration. .

Planners from seven clinics were asked to create RT plans for the same patient (right-sided oropharyngeal cancer patient with bilateral lymph node target). All were given identical CT and structure data. Plans (66/60/50 Gy) were made locally according to national guidelines (although not under clinical conditions, i.e. reviewing processes and clinical approval of plans was not performed). RT plans were either 1-3 arc VMAT or 9-field co-planar IMRT (PTV margin 3mm). Clinics used different treatment machines and planning systems. Fifteen RT plans were available for analysis. We calculated the following metrics for all plans: clinical dose constraints, D5%, VXGy (X = 10,20,…,70Gy), dose standard deviation for each structure (target or OAR), dose conformality to targets, maximum dose at 1-3cm from targets, dose gradient index, size/number of contiguous cold volumes in targets.

All plans met the criteria for target coverage and critical OAR (spinal cord, brainstem) dose. For target coverage the plans were very similar, with a max-min span of <3.1Gy (CTV D99% and PTV D98%). There were notable variations in several metrics not related to critical constraints (see examples in fig. 1). DVHs for high-risk PTV (small variation) and contralateral parotid (larger variation) are shown in figure 1. When considering non-critical OAR (mean doses) we saw a group of very similar plans that were notably better than the others (fig.2). These plans achieved lower mean doses for most OAR - it did not seem to be only a case of prioritizing differently between OARs. There were no significant differences in target metrics between the two groups (Mann-Whitney U-test p<0.05). Differences in OAR mean dose correlated with differences in a wide range of dose levels (fig 2 (3)), indicating that constraints were not given equal time and weight by all planners during optimization. Number of fields had no influence.

The national guidelines generally resulted in high plan quality, meeting all clinical constraints. Using the additional quantitative metrics, we saw notable inter-planner variation. . Although plans were not made under strict audit-conditions, this study demonstrates the potential for interplanner variations within clinical acceptability and under a guideline, and methods for quantifying these variations. Systematic application of quantitative metrics during planning should be investigated as a method of supporting dissemination of best practices in treatment planning.