Advantages and limitations of navigation‐based multicriteria optimization (MCO) in selectively sparing pharyngeal constrictor muscles in head and neck radiotherapy treatment planning

TL;DR

This study explores using MCO in radiotherapy to spare pharyngeal muscles while managing trade-offs in treatment coverage.

Contribution

The paper introduces integrating KBP with MCO for selective PCM sparing in head and neck radiotherapy.

Findings

MCO reduced mean doses to PCM and OARs but increased HI and CI.

Most MCO plans maintained acceptable PTV54 coverage, though some were slightly compromised.

Dose metrics between navigated and deliverable plans showed minimal differences.

Abstract

Sparing pharyngeal constrictor muscles (PCMs) during radiotherapy improves patient‐reported swallowing function. This study aimed to explore the feasibility of integrating knowledge‐based planning (KBP) with multicriteria optimization (MCO) in Eclipse v18.0 to selectively spare PCM, quantify the required trade‐off in prophylactic planning target volume (PTV54) coverage, and to evaluate MCO performance.

Ten patients previously planned with KBP for oropharyngeal cancer (65, 60, and 54 Gy in 30 fractions) were retrospectively re‐planned. Clinical plans were further optimized using trade‐off exploration in MCO, with a priority order: spinal cord and brainstem sparing, high‐dose and intermediate‐dose target coverage, PCM sparing, low‐dose target coverage, parotids sparing, remaining organs at risk (OAR). Plans were evaluated based on planning target volumes dose metrics (D50%, D98%, and D2%), homogeneity index (HI), conformity index (CI), and maximum and mean doses to OARs, and paired t‐tests were performed. Differences between navigated and deliverable plans were analyzed. One patient underwent 10 identical repeat plan generations.

MCO reduced the average mean dose to the superior and middle PCM, inferior PCM, contralateral parotid, and larynx by 2.0, 3.4, 2.6, and 3.9 Gy, respectively (p < 0.05) but at the expense of HI and CI. No difference was observed in average PTV54 D98% between techniques; however, all clinical plans and seven MCO plans achieved D98% ≥ 95%, with three MCO plans modestly compromised (D98% 93.7%–94.6%). Dose metrics between navigated and deliverable plans differed by ≤0.7 Gy for mean doses and ≤1.8 Gy for maximum doses. Pareto surface generation was not repeatable.

MCO effectively balances the trade‐off between PCM sparing and low‐dose target coverage. It may be a valuable tool in the context of personalized care.

Figures