Comparison of Planning Quality and Efficiency Between Conventional and Knowledge-Based Algorithms in Nasopharyngeal Cancer Patients Using Intensity Modulated Radiation Therapy

Abstract

Intensity modulated radiation therapy (IMRT) is being widely used for its capability of achieving highly conformal dose and improving locoregional control. However, treatment plan quality and planning time are institute- and planner-dependent, and there is no standardized tool to recognize an optimal plan. A knowledge-based algorithm (KBA) can generate automated constraints to assist optimization and produce planner-independent, high-quality plans within a short time. This report compares the efficiency and quality of IMRT plans generated with and without KBA-assisted IMRT (KBA_IMRT) for nasopharyngeal carcinoma (NPC) patients. Seventy-nine NPC patients treated with radical IMRT at our institution between the years of 2013 and 2014 were used to configure the KBA_IMRT system. Ten consecutive NPC cases indicated for radical radiation therapy were then recruited to evaluate its performance. For these 10 cases, both the IMRT plan quality and planning time with and without using KBA_IMRT were compared using paired t test. Ten consecutive NPC patients treated with radical intent between October 2014 and January 2015 were recruited into the study. Their stage distributions were stage I (1), II (0), III (7), IVA (1), and IVB (1). The treatment gave 70 Gy to PTV1 (for gross tumor volume [GTV]), 63 Gy to PTV2 (for adjacent high-risk structures and bilateral retropharyngeal, levels IB, II, III, VA cervical lymph nodes), and 52.5 Gy to PTV3 (for prophylactic nodal irradiation), over 35 fractions. The results were shown in Table 1. There were no significant differences in dose coverage to GTVP (primary tumor), GTVN (involved nodes), and PTV1. The conformity index to PTV1 was comparable but the homogeneity index with KBA_IMRT was slightly worse. KBA_IMRT achieved a better sparing to brainstem (54.0 Gy vs 56.4 Gy, P=.027) and parotid glands (left: 36.2 Gy vs 43.4 Gy; right: 40.3 Gy vs 45.0 Gy, P<.05), but gave a higher dose to spinal cord (45.6 Gy vs 43.9 Gy, P=.037). The planning time was reduced almost by half using KBA_IMRT (144.5 minutes vs 261.3 minutes, P<.001). This study demonstrates that KBA_IMRT can significantly improve planning efficiency and produce quality IMRT plans for NPC patients. Further fine-tuning of KBA_IMRT parameters and other clinical applications including adaptive radiation therapy are being tested in our center.Poster Viewing Abstracts 2795; Table 1Manual planKBA_IMRTP valueGTVP (min dose)67.6 Gy66.7 Gy.071GTVN (min dose)70.2 Gy70.0 Gy.556% PTV1 < 100% dose2.7%5.4%.192Conformity index PTV11.1501.134.079Homogeneity index PTV10.1220.196.005*Brainstem (max dose)56.4 Gy54.0 Gy.027*Spinal cord (max dose)43.9 Gy45.6 Gy.037*Chiasm (max dose)54.3 Gy54.2 Gy.65Lt Parotid Gland (mean dose)43.4 Gy36.2 Gy.001*Rt Parotid Gland (mean dose)45.0 Gy40.3 Gy.039*Planning time261.3 minutes144.5 minutes<.001**P<.05 was considered statistically significant Open table in a new tab