Study of the dose-volume parameters variation in tumor target volumes and organs at risk during nasopharyngeal carcinoma radiotherapy applying deformation registration.

BACKGROUND: During intensity-modulated radiotherapy (IMRT) for nasopharyngeal carcinoma (NPC), the volume of the target volume and the organs at risk (OARs) will change constantly, which may lead to differences between the actual dose received and the initial planned dose. In this study, the cumulative dose of the two plans was obtained by deformable registration. This study provides an approach to evaluate the dose volume of IMRT for the NPC objective.
METHODS: From July 2014 to May 2018, eighteen NPC patients who accepted simultaneous integrated boost IMRT were enrolled. All patients underwent simulation CT (CT1) and replanning CT (CT2) scans after 20-25 fractions of radiation therapy. The treatment plans were designed on CT1 and CT2 with the name of Plan1 and Plan2, respectively. The Planreg and Plandef were obtained after registering from CT2 to CT1 using rigidity and deformation technology by Velocity. Then the dose-volume indices of the tumor target volumes and OARs at Plan1, Plan2, Planrig and Plandef were compared.
RESULTS: The gross tumor volume (GTV) and the left and right parotid gland volumes decreased by 20.8% (P<0.001), 36.8% (P<0.001) and 37.5% (P<0.001), respectively, from CT1 to CT2. There was no significant difference in the dose-volume index on the GTV and plan gross tumor volume (PGTV) between Plan1 and Plan2. The V30 of the left and right parotid gland and the Dmax of the brainstem, left and right eyeballs, left and right lens, and left and right optic nerves were all lower in Plan2 than in Plan1 (the average decrease was 17.0% to 60.1%). The differences in some dose-volume parameters (including Dmean, D99 of the GTV and PGTV, Dmean of the parotid glands, Dmax of the lens and optic nerves) between Plandef and Plan1 were less than 5%. The differences in some dose-volume parameters (including Dmean, D95 of the GTV and PGTV, Dmean, D50 and V30 of the parotid glands, Dmax of lens and optic nerves) between Planrig and Plan1 were less than 10%. The Dyce Similarity Coefficient of the target volume and OARs after deformation registration were higher than that after rigid registration.
CONCLUSIONS: The volume of the GTV and parotid glands were decreased during the IMRT for NPC. The dose-volume indices of the GTV and the OARs in Plandef were similar to those in Plan1. Therefore, the dose-volume indices of Plan1 can be used to evaluate the efficacy of radiotherapy and to predict radioactive damage. 2019 Translational Cancer Research. All rights reserved.

Introduction

Nasopharyngeal carcinoma (NPC) occurs mostly in Southeast Asia and southern China (1,2). Due to the particularity of its anatomical location, NPC is surrounded by many important tissues and organs. The vast majority of NPC are more sensitive to radioactivity, and consequently radiotherapy or radiotherapy-based comprehensive treatment has now become the main method for radical treatment of patients with NPC (3). For patients with early NPC after radical radiotherapy, the 5-year survival rate can reach 80%.

There are many important organs at risk (OARs) around the target volume of NPC, such as the parotid glands, brain stem, spinal cord, eyeballs, optic nerves and lens. The exposure doses by these OARs were closely related to the quality of life of patients after treatment. The key to improving the effect of radiotherapy is to ensure the dose of the target volume and to minimize the dose of normal tissues and OARs. Therefore, it is of great significance to accurately predict the dose of the target volume and OARs during the radiotherapy process for patients with NPC.

Intensity-modulated radiotherapy (IMRT) is a routine clinical radiotherapy technique for NPC. It has a steep dose gradient distribution, which can reduce the exposure dose of OARs around the tumor while ensuring that the tumor target volume can be irradiated with a sufficient dose (4). However, some studies have shown that the volume and location of tumor target volume and OARs are different during the IMRT for NPC (5-7). In particular, the volume of the parotid glands was significantly reduced during IMRT for NPC (8). Differences in the volume of the parotid glands may lead to inadequate coverage of a radiation dose over the tumor target volume or excessive dosage of OARs, which deviates from the original plan. Deformed registration provides a feasible method for accurately accumulating radiation doses to target volume and OARs in NPC patients during radiotherapy. The purpose of deformed registration technology is to accumulate the dose through point-to-point registration, and then track the dose of the target volume and OARs during IMRT for NPC.

In the course of radiotherapy for NPC, the volume reduction in tumor target and OARs may not be synchronized with the changes in their dose volume indices. This study used registration technology to accumulate the dose of two plans of IMRT for NPC by rigid and deformed methods, and to thereby analyze the difference in the dose-volume index of the tumor target volume and OARs. This study provided an approach to evaluate the dose volume of IMRT for the NPC objective.

Methods

Patient characteristics

Eighteen NPC patients (13 males and 5 females) who received simultaneous integrated boost intensity-modulated radiotherapy (SIB-IMRT) from July 2010 to May 2018 were selected. The clinical stage was divided into 10 cases at the T2 stage and 8 cases at the T3 stage according to the AJCC 2010 staging standard T stage.

CT scanning and planning design

All patients underwent scanning and simulation on a Philips large-aperture CT, taking the supine position; fixing the head, neck, and shoulder with a thermoplastic mask; and scanning from the top of the head to the lower 3 cm of the clavicle with a layer thickness of 3 mm. The scanned CT image was transmitted to the Eclipse 13.5 planning system and the SIB-IMRT technique was used. The radiologist outlined the target volume, as well as OARs such as the parotid gland, brain stem, spinal cord, eyeballs, optic nerves, and lens. The dosage limitation conditions for OARs are a 50% parotid volume dose less than 30 Gy, brainstem dose less than 54 Gy, spinal cord dose less than 45 Gy, optic nerve dose less than 54 Gy, eyeball dose less than 50 Gy, lens dose less than 8 Gy, etc. Plan1 was the IMRT plan on simulation CT before radiotherapy and Plan2 was the IMRT plan on repeated simulation CT at the middle and late stages of radiotherapy. The prescribed dose was 66–70 Gy.

Image registration

The patient’s two radiotherapy plans were imported into the velocity image registration software based on B-spline algorithm, and the repeated simulated CT2 was registered into the first simulated CT1 to obtain the transformed registration image CT3. The dose distribution of Plan2 and the dose distribution of Plan1 were separately subjected to rigidity and deformation registration to obtain Planrig and Plandef.

Statistical indicators

The volume of the gross tumor volume (GTV) and bilateral parotid glands; average dose (Dmean), 95% volume dose (D95) and 99% volume dose (D99) of GTV; and plan gross tumor volume (PGTV) were measured and recorded. Indicators of various OARs include the bilateral parotid Dmean, 50% volume dose (D50) and 30 Gy surrounded volume (V30); maximum point dose (Dmax) and 5% volume dose (D5) of the brain stem; Dmax of the bilateral eyeballs, optic nerves and lens; and Dice Similarity Coefficient (DSC) of the GTV, PGTV, parotid gland, brainstem, eyeballs, optic nerves and lens after deformation registration and rigid registration. (DSC=2|A∩B|/(|A|+|B|). The volume of the target volume and OARs in CT1 and CT2 were A and B respectively.)

Statistical methods

The data are expressed in the form (). A paired t-test was performed using SPSS 19.0 statistical software. P<0.05 indicates that the difference is statistically significant.

Results

Change in the volume of the GTV and parotid gland

Compared with the initial simulated CT1, the volumes of the GTV, left parotid gland and right parotid gland in the repeated simulated CT2 decreased by 20.83% (P<0.001), 36.83% (P<0.001) and 37.47% (P<0.001), respectively. After deformation registration, the volume of the GTV decreased by 3.15% (P<0.001) compared with the initial positioning of the GTV, the left parotid volume decreased by 6.56% (P<0.001), and the right parotid volume decreased by 5.89% (P<0.001) ().

Table 1

Volume of the tumor target volume and parotid gland in positioning image CT1, reset image CT2 and registration image CT3 (cm3, )

Parts	CT1	CT2	CT3	CT1 & CT2			CT1 & CT3
				t	P		t	P
GTV	79.28±49.66	64.47±42.39	76.79±48.21	5.477	0.000		4.292	0.000
Parotid-L	16.76±4.74	10.77±4.07	15.61±4.43	9.972	0.000		5.735	0.000
Parotid-R	16.63±4.40	10.46±3.99	15.58±4.03	9.303	0.000		6.116	0.000

Dosimetric comparison in the target volume

There was no significant change in the dose indices of the GTV and PGTV between Plan2 and Plan1 (P>0.05). Compared with Plan1, the Dmean of the GTV in Plandef and Planrig decreased by 0.42% and 0.62%, respectively (), and the difference was statistically significant (P<0.05). The Dmean of the PGTV in Plandef and Planrig decreased by 1.08% and 1.49%, respectively; the D95 of the PGTV in Plandef and Planrig decreased by 2.77% and 5.49%, respectively; the D99 of the PGTV in Plandef and Planrig decreased by 4.29% and 7.39% (); and these differences were statistically significant (P<0.05).

Table 2

Target volume the dose distribution (Gy, )

Target	Parameter	Plan1	Plan2	Plandef	Planrig	t1	P1	t2	P2	t3	P3	t4	P4
GTV	Dmean	73.91±3.59	73.85±3.55	73.60±3.56	73.45±3.54	0.533	0.601	2.714	0.015	3.308	0.004	1.553	0.139
	D95	71.93±3.49	71.79±3.91	71.35±3.78	70.51±4.13	0.853	0.405	1.792	0.091	2.361	0.030	1.432	0.170
	D99	71.21±3.50	70.29±4.96	70.21±3.87	67.60±8.36	1.398	0.180	2.516	0.022	1.948	0.068	1.470	0.160
PGTV	Dmean	73.56±3.42	72.10±6.48	72.78±3.70	72.48±3.71	1.053	0.307	3.935	0.001	3.734	0.002	1.555	0.138
	D95	70.99±3.43	69.14±6.46	69.09±5.12	67.12±7.47	1.313	0.207	3.319	0.004	2.553	0.021	1.441	0.168
	D99	69.08±3.60	65.85±7.75	66.23±6.29	64.03±8.63	1.863	0.080	3.195	0.005	2.819	0.012	1.657	0.116

t1-t4, P1-P4 represent the values of Plan1 and Plan2, Plan1 and Plandef, Plan1 and Planrig, and Plandef and Planrig as paired t-tests, respectively.

DSC comparison between rigid registration and deformation registration

The DSC of the target volume and the OARs after rigid registration and deformation registration, respectively, is shown in . The DSC of the target volume and the OARs after deformation registration was greater than the DSC of the rigid registration target volume and the OARs. The DSC difference between the left and right parotid deformation registration and rigid registration was significant (P<0.05).

Table 3

DSC for the deformation registration and rigid registration

Parts	Deformation DSC	Rigid DSC	t	P
GTV	0.76±0.10	0.73±0.10	1.744	0.099
PGTV	0.77±0.13	0.75±0.12	1.550	0.139
Parotid-L	0.70±0.09	0.56±0.16	5.158	0.000
Parotid-R	0.71±0.10	0.58±0.12	7.808	0.000
Brainstem	0.85±0.04	0.85±0.04	-0.368	0.718
Eye-L	0.87±0.05	0.84±0.07	1.971	0.065
Eye-R	0.87±0.06	0.82±0.09	1.985	0.064
Lens-L	0.50±0.24	0.44±0.24	1.044	0.311
Lens-R	0.51±0.25	0.43±0.26	1.464	0.162
Optic-L	0.61±0.17	0.49±0.22	2.664	0.016
Optic-R	0.57±0.14	0.50±0.19	2.001	0.062
Spinal cord	0.83±0.03	0.68±0.13	4.660	0.000

Dosimetric comparison in the OARs

The dose distribution of each OAR in Plan1, Plan2, Plandef and Planrig is shown in . Compared with Plan1, the V30 of the left parotid gland in Plan2 decreased by 33.5%, the Dmean of the right parotid gland decreased by 31.45%, the V30 decreased by 60.13%, and the D50 decreased by 28.35%. The difference was statistically significant (P<0.05). The Dmean of the left and right parotid glands in Plandef decreased by 1.98% and 2.3%, respectively, compared with Plan1, and the difference was statistically significant (P<0.05). The changes in V30 and D50 of the left and right parotid glands between Plandef and Plan1 were not significant (P>0.05). The Dmean of the left and right parotid in Planrig decreased by 4.21% and 4.93% compared with Plan1, V30 decreased by 7.68% and 10.35%, and D50 decreased by 3.37% and 3.46%, and this difference was statistically significant (P<0.05).

Table 4

Dose distribution of OARs ()

OARs	Parameter	Plan1	Plan2	Plandef	Planrig	t1	P1	t2	P2	t3	P3	t4	P4
Parotid-L	Dmean (Gy)	29.73±6.94	24.06±11.57	29.02±6.55	28.31±6.21	2.089	0.052	2.152	0.046	4.138	0.001	3.814	0.001
	V30 (%)	36.15±17.39	25.52±23.64	34.88±17.00	33.20±15.98	2.247	0.038	2.055	0.056	5.345	0.000	3.204	0.005
	D50 (Gy)	24.75±5.97	20.92±10.22	24.36±5.78	23.83±5.63	1.488	0.155	1.531	0.144	3.709	0.002	4.029	0.001
Parotid-R	Dmean (Gy)	30.07±4.37	20.23±7.65	29.39±4.51	28.65±4.99	4.813	0.000	2.643	0.017	3.587	0.002	1.792	0.091
	V30 (%)	36.2±14.54	14.84±16.37	34.29±14.20	32.90±14.70	5.117	0.000	2.049	0.056	3.489	0.003	1.752	0.098
	D50 (Gy)	25.00±4.69	17.21±6.98	24.62±4.57	24.18±4.97	3.726	0.002	1.466	0.161	2.412	0.027	1.823	0.086
Brainstem	Dmax (Gy)	50.41±11.06	40.80±13.80	49.90±10.84	50.03±10.99	2.578	0.020	0.371	0.715	0.283	0.780	-1.119	0.279
	D5 (Gy)	45.52±10.45	31.82±14.75	44.00±10.00	44.10±10.05	3.374	0.004	3.965	0.001	3.687	0.002	-1.910	0.073
Eye-L	Dmax (Gy)	31.53±16.44	20.35±16.22	30.12±16.00	30.19±15.97	3.738	0.002	5.478	0.000	5.472	0.000	-0.811	0.429
Eye-R	Dmax (Gy)	32.24±18.68	22.45±22.56	31.17±18.46	31.36±18.32	2.908	0.010	3.340	0.004	2.421	0.027	-1.134	0.272
Lens-L	Dmax (Gy)	5.17±1.42	3.83±2.14	5.20±1.68	5.22±1.67	3.180	0.005	-0.201	0.843	-0.392	0.700	-0.868	0.398
Lens-R	Dmax (Gy)	5.24±1.52	3.93±2.25	5.16±1.57	5.16±1.54	2.791	0.013	1.122	0.277	1.258	0.226	-0.218	0.830
Optic-L	Dmax (Gy)	40.77±26.36	20.99±21.59	38.49±24.82	38.66±24.96	3.728	0.002	3.503	0.003	3.638	0.002	-1.422	0.173
Optic-R	Dmax (Gy)	41.28±26.08	22.75±23.82	39.19±24.74	39.41±24.87	3.536	0.003	2.788	0.013	3.071	0.007	-1.110	0.282

t1-t4, P1-P4 represent the values of Plan1 and Plan2, Plan1 and Plandef, Plan1 and Planrig, and Plandef and Planrig as paired t-tests, respectively.

Compared with Plan1, Dmax and D5 of the brainstem in Plan2 were reduced by 17.04% and 27.3%, respectively. The Dmax of the left eyeball, right eyeball, left lens, right lens, left optic nerve and right optic nerve were decreased by 36.43%, 28.48%, 24.95%, 24.29%, 44.69% and 41.1%, respectively. The difference was statistically significant (P<0.05). Compared with Plan1, D5 of the brainstem in Plandef was reduced by 3.3%. The Dmax of the left eyeball, right eyeball, left optic nerve and right optic nerve were decreased by 6.13%, 3.44%, 5.67% and 5.19%, respectively. The difference was statistically significant (P<0.05).

Compared with Planrig, the Dmean, V30 and D50 of the left parotid gland in Plandef increased by 2.26%, 4.36% and 2.1%, respectively, and the difference was statistically significant (P<0.05). The difference in Dmean, V30 and D50 in the right parotid gland was not significant. The Dmax of the OARs, such as the brainstem, eyeballs, lens and optic nerves, were all decreased.

Discussion

In the process of IMRT, although IMRT is more effective than conventional radiotherapy in increasing the target dose, the radiation exposure of normal tissues and OARs is relatively low (9,10). However, for NPC, factors such as changes in the patient weight, volume of tumors and OARs may have a certain impact on dose assessment. Anatomical changes may result in an insufficient target dose and excessive exposure of OARs. The difference between the actual dose received by the patient and the radiotherapy plan designed by simulation CT was 15% (11). Lu et al. (12) compared the repeated simulated CT with the initial simulated CT in 12 patients with NPC and found that the volume of the PGTV was reduced by an average of 16.4%±27.3%. In this study, the tumor target volume of repeated simulated CT was reduced by 20.83% on average compared with that of the initial simulated CT, and the difference between the Plan deformation, Plan2 and Plan1 dose evaluation showed that the Dmean, D95 and D99 of the GTV were not significant. Cheng et al. (13) and Wang et al. (14) showed that in the course of radiotherapy for NPC, the volume of the tumor target was smaller than that before radiotherapy, but the target dose index did not decrease. In this study, it was found that the volume of the target gradually decreased during IMRT for NPC, and this conclusion is consistent with most studies (12-14). The dose index of the target volume was basically the same in the four plans. Therefore, the dose index of target volume in the initial plan can accurately evaluate the efficacy of radiotherapy.

The parotid gland is sensitive to radiation, and the volume of the parotid gland changes significantly during radiotherapy for NPC. Fung et al. (15) transmitted the original plan dose distribution of NPC radiotherapy to the new CT to form a synthetic plan compared with the original plan dosimetry, the dose of the OARs increased, and the Dmean of the left and right parotid glands significantly increased. In this study, we found that the volume of the parotid gland changed significantly compared with the initial location CT, and the left and right parotid glands were reduced by 36.83% and 37.47%, respectively. Compared with Plan1, the V30 of the left parotid gland in Plan2 decreased by 36.15%; the Dmean, V30 and D50 of the right parotid gland decreased by 30.07%, 36.2% and 25%, respectively. The dose-volume evaluation of Plandef and Plan1 showed that the Dmean of the left and right parotid glands decreased by 1.98% and 2.3%, respectively, and the V30 and D50 of the left and right parotid glands were not statistically significant. These studies have shown that the parotid gland volume shrinks as the dose is absorbed during radiotherapy, and the proportion of shrinkage is not synchronized with the shrinkage of the tumor target volume. In this study, it was found that the dose-volume index of parotid glands in Plan1 and Plan2 was significantly different. The dose-volume index of Plan1 and Plan2 may not accurately predict the radiation damage of parotid glands during radiotherapy. However, deformation accumulation is an objective method to eliminate the effect of volume shape change, and the difference in the dose-volume index between the Plan1 and Plandef were less than 5%. Therefore, it can be considered that the use of the initial planned dose-volume index can accurately predict the radiation damage of the parotid gland during IMRT for NPC.

Compared with the initial simulated CT, repeated simulated CT was deformed to different degrees, and therefore, the dose distribution of the two plans cannot be accumulated directly. Image registration method is needed to deform and accumulate the dose. Janssens et al. (16) used the Demons algorithm based on the image grayscale and the Morphons algorithm based on the edge and line to achieve the deformation registration. In this study, deformation registration and rigid registration based on the B-spline algorithm were adopted. The DSC of the target volume and OARs in the deformation registration were larger than that in the rigid registration, especially with respect to the parotid glands (P<0.05), which shows that deformation registration was superior to rigid registration.

The volume changes in the brain stem, eyeballs, optic nerves, and lens during IMRT for NPC were very small. Compared with Plan1, Plan2 showed a significant difference (P<0.05) in the dose index of these OARs, which was attributed to the shrinking of the GTV volume, and the variation in the dose-volume index in Plandef compared with Plan1 were very small.

Conclusions

In summary, during IMRT for NPC, the volume of the tumor target volume and parotid gland were significantly reduced. However, there was no significant difference between the dose-volume index of the tumor target and OARs in the first planning and the dose-volume index accumulated by deformation registration. Therefore, the dose-volume index of the initial plan can be used to evaluate the curative effect of NPC patients in the whole radiotherapy process and to predict the radiation damage.