2350 - Evaluation and Validation of Open Access Knowledge Based Planning (KBP) Model in India Developed for Treatment of Bilateral Head and Neck Cancer

H. Liu¹, S. Chaudhari², R. M. Murugan², M. G. Chowdhury², M. Duhan², M. Damera², G. Upendar², R. Clark³, A. Magliari³, and S. Beriwal³; ¹Department of Radiation Oncology, University of Pennsylvania, Philadelphia, PA, ²American Oncology Institute, Hyderabad, India, ³Varian Medical Systems Inc, Palo Alto, CA

Purpose/Objective(s): There are various open access head and neck KBP models available with variable performance. A new open access RP model (HN-SIB-BPI) was created with 66 imaging and contouring datasets from western population. It was created using a previous best available model as foundation with goal to increase conformality and further reduce dose to OARs including parotid glands with quantification of improvement by evaluation with a dosimetric scorecard. We tested the performance of this model in our clinic in India to evaluate if this KBP model could improve plan quality in our clinic in comparison to existing plans. Materials/

Methods: 18 patients with head and neck cancer (HNC) (9 oropharynx, 5 oral cavity, 1 larynx, and 3 hypopharynx) treated with volumetric modulated arc therapy (VMAT) with planning target volume (PTV) overlapping bilateral parotid glands were selected to be replanned with this new RP model. Each patient had either two or three dose level PTVs. The plans were redone with both C arm (Plan A) and ring gantry (Plan B) delivery systems using four arcs as specified in HN-SIB-BPI model documentation. All plans were created by three physicists with pre-existing clinical goals to determine any meaningful improvement in plan quality with this model.
Results: Plan A showed better coverage for low, intermediate, and high-risk target volume and reduced doses to brain stem, spinal cord, larynx, and bilateral parotid glands compared to the VMAT plan. (Table 1). The mean reduction in left and right parotid glands were 5.73 and 3.19 Gy respectively (p < 0.01). There was increased mean dose to lip by 2.46 Gy with Plan A (p = 0.03). Similarly, Plan B showed better coverage for low, intermediate, and high-risk target volume and reduced dose to brain stem, spinal cord, and bilateral parotid glands compared to the VMAT (Table 1). The mean reduction in left and right parotid glands were 4.19 and 2.72 Gy (p < 0.01). Similarly, there was increased mean dose to lip by 2.24 Gy (p = 0.04) with Plan B.
Conclusion: The HN-SIB-BPI open access KBP model was validated in our patient population showing significant reduction in dose to parotid glands with better coverage of target volume. Application of this KPB model can improve uniformity and quality of plans. We will implement this as our default model in our clinic. Abstract 2350 – Table 1

		Plan A vs VMAT			Plan B vs VMAT
		n	t	p-value	n	t	p-value
PTV1	V70Gy = 97.0 %	15	2.20	0.04	15	1.59	0.13
PTV1	V95% = 95.0 %	15	2.59	0.02	15	3.25	0.01
PTV1	Dmax = 105.0 %	17	1.12	0.28	17	-0.24	0.81
PTV2	V95% = 99.0 %	14	2.40	0.03	14	3.02	0.01
PTV2	V60Gy = 97.0 %	11	3.12	0.01	11	3.64	<0.01
PTV3	V95% = 95.0 %	12	1.58	0.14	12	1.91	0.08
PTV4	V54Gy = 97.0 %	12	2.83	0.02	12	2.57	0.03
Brainstem	Dmax = 54.00 Gy	18	-3.83	<0.01	18	-4.12	<0.01
Larynx	Dmean < 20.00 Gy	16	-3.73	<0.01	16	-1.63	0.12
Lips	Dmean = 20.00 Gy	17	2.46	0.03	17	2.24	0.04
Parotid L	Dmean < 22.00 Gy	18	-5.73	<0.01	18	-4.19	<0.01
Parotid R	Dmean < 22.00 Gy	18	-3.19	0.01	18	-2.72	0.01
Spinal Cord	Dmax = 50.00 Gy	18	-4.39	<0.01	18	-6.29	<0.01
Total Monitor Unit		16	2.23	0.04	16	2.25	0.04