Dosimetric Comparison of Helical Tomotherapy and Intensity-Modulated Proton Therapy in Hippocampus- and Scalp-Sparing Whole Brain Radiotherapy.

Objective: Cognitive decline and alopecia after radiotherapy are challenging problems. We aimed to compare whole brain radiotherapy (WBRT) plans reducing radiation dose to the hippocampus and scalp between helical tomotherapy (HT) and intensity-modulated proton therapy (IMPT).
Methods: We conducted a planning study of WBRT for 10 patients. The clinical target volume was defined as the whole brain excluding the hippocampus avoidance (HA) region. The prescribed dose was 30 Gy in 10 fractions to cover 95% of the target. Constraint goals were defined for the target and organs at risk (OAR).
Results: Both techniques met the dose constraints for the target and OAR. However, the coverage of the target (dose covering 95% [D95%] and 98% [D98%] of the volume) were better in IMPT than HT (HT vs IMPT: D95%, 29.9 Gy vs 30.0 Gy, P < .001; D98%, 26.7 Gy vs 28.1 Gy, P = .002). The homogeneity and conformity of the target were also better in IMPT than HT (HT vs IMPT: homogeneity index, 1.50 vs 1.28, P < .001; conformity index, 1.30 vs 1.14, P < .001). IMPT reduced the D100% of the hippocampus by 59% (HT vs IMPT: 9.3 Gy vs 3.8 Gy, P < .001) and reduced the Dmean of the hippocampus by 37% (HT vs IMPT: 11.1 Gy vs 7.0 Gy, P < .001) compared with HT. The scalp IMPT reduced the percentage of the volume receiving at least 20 Gy (V20Gy) and V10Gy compared with HT (HT vs IMPT: V20Gy, 56.7% vs 6.6%, P < .001; V10Gy, 90.5% vs 37.1%, P < .001).
Conclusion: Both techniques provided acceptable target dose coverage. Especially, IMPT achieved excellent hippocampus- and scalp-sparing. HA-WBRT using IMPT is a promising treatment to prevent cognitive decline and alopecia.

Introduction

Whole brain radiotherapy (WBRT) remains the standard treatment modality for patients with brain metastases because it palliates symptoms, improves intracranial control, and reduces central nervous system death.[1-3] However, 10 to 20% of patients with brain metastases after WBRT develop clinical cognitive decline. Advances in systemic chemotherapy including immunotherapy have prolonged the survival time of the patients with distant metastasis. Thus, cognitive decline due to WBRT is a serious problem.[1,4-6] A prospective observational study of patients with benign or low-grade brain tumors treated with fractionated stereotactic radiation therapy (SRT) reported a relationship between dose to the bilateral hippocampi and likelihood of long-term memory impairment. To overcome this problem, a hippocampus avoidance (HA) technique for reducing the radiation dose to the hippocampus in WBRT has been developed. Intensity-modulated radiation therapy (IMRT) with HA technique delivers therapeutic doses to the whole brain (WB). NRG CC001, a prospective multi-institutional randomized phase III trial, investigated the role of WBRT plus memantine with or without HA technique in brain metastases to validate the hypothesis that conformal avoidance of the hippocampal neural stem cells using IMRT improves cognitive outcomes. HA-WBRT plus memantine preserved cognitive function and patient-reported symptoms, with no difference in intracranial control and survival time. Thus, HA-WBRT is a novel and promising treatment for brain metastases.

Radiation-induced alopecia after WBRT has a significant negative impact on patients’ quality of life (QOL).[10,11] A dose response relationship between the follicle and the occurrence of permanent alopecia after cranial irradiation was reported. A recent study has provided new constraints for the scalp for IMRT that may help in reducing the probability of hair-loss. Helical tomotherapy (HT) is an excellent treatment for delivering IMRT.[14,15] In recent years, proton therapy has gradually spread and intensity-modulated proton therapy (IMPT) enables an excellent dose distribution. Our hypothesis was that IMPT could achieve comparable target coverage and hippocampus- and scalp-avoidance compared with IMRT. To investigate this hypothesis, HA-WBRT plans using HT and IMPT were generated and a planning study comparing dosimetric data of the two techniques was conducted.

Methods

Patients

This study was approval by the institutional review board (Nagoya City University Hospital Ethics Committee, approval number: 60-21-0045), with waivers for informed consent. We selected 10 consecutive patients with brain metastases previously treated with SRT. They were not selected in any particular way that might influence the results of the study. All plans for this pilot study were not actual clinical treatment plans. This study conformed to the Strengthening the Reporting of Observational Studies in Epidemiology guidelines. We had de-identified patient information not to be ascertained in any way.

Treatment Volumes and Normal Structures

A shell was used for immobilization in a supine position. Computed tomography (CT) images were acquired with a 1.25 mm slice thickness. The patients underwent three-dimensional spoiled gradient-recalled axial magnetic resonance imaging (MRI) scans with axial T2-weighted and three-directional gadolinium contrast-enhanced T1-weighted sequence acquisitions with 1.0 mm slice thickness (thin-slice MRI). The CT simulation and MRI images were fused semi-automatically, and the target and the organs at risk (OAR) were contoured using the RayStation treatment planning system (RaySearch Medical Laboratories AB, Stockholm, Sweden).

The contouring and planning directives were generated based on the previous studies.[8,9] The gross tumor volume (GTV) was not defined. The clinical target volume (CTV) included the WB parenchyma, subarachnoid space, meninges, and excluded the HA region. For both the HT and IMPT plans, no margin between CTV and the planning target volume (PTV) was defined. The hippocampus was manually contoured using fused thin-slice MRI. The HA region was defined with a 5-mm margin around the hippocampus. The scalp was defined as the tissue with thickness of 5 mm underlying the skin up to the outer of the skull. It was auto contoured expanding the head contour by 5 mm medially and the skull was subtracted. In general, scalp extends from the superior nuchal lines and occipital turbulences to the supraorbital foramen. Therefore, the forehead area was manually modified. The OAR included the hippocampus, scalp, chiasm, optic nerves, and eyes.

Prescribed Dose and Planning Goal

The prescribed dose was 30 Gy in 10 fractions to cover 95% of the WB target. The grid size for calculation was 2.0 × 2.0 mm for both plans. Dose-volume histogram (DVH) was used to reduce the dose to the OAR as low as achievable, while target coverage was maintained. IMPT plans were optimized assuming a relative biological effectiveness of 1.1 and the same goals for HT plans were used. The modified dose constraints of the NRG CC001 trial were used for this study (Table 1). Deviations from the constraint goal were allowed for the dose constraint for the scalp. In addition, dose to 100% of the hippocampus (D100%) not exceeding 10 Gy and maximal hippocampal dose (Dmax) not exceeding 17 Gy were considered acceptable deviations according to a prior study.

Table 1.

Dose Constraints for Target and Organs at Risk.

Structure	Parameter	Constraint Goal
WB target	D98%	> 25 Gy
	D95%	= 100%
	D2%	< 37.5 Gy
Hippocampus	Dmax	< 16 Gy
	D100%	< 9 Gy
Scalp	V20Gy	< 50%
	V10Gy	< 80%
Optic nerves, chiasm	Dmax	< 33 Gy
Eyes	Dmax	< 33 Gy

Abbreviations: WB, whole brain; Dmax, maximum dose; D2%, D95%, D98%, D100%, dose to 2%, 95%, 98%, 100% volume of the target or organs at risk; V10, 20Gy, percentage of the volume receiving at least 10 or 20 Gy.

HT and IMPT Planning

HT plans were generated using a 6-MV X-ray beam machine dedicated to IMRT (Radixact X9, Accuray, Sunnyvale, USA). The superposition dose calculation algorithm was used in the treatment planning workstation (Precision, Accuray, Sunnyvale, USA). A field width, pitch, and normal modulation factor were 2.5 cm, 0.287, and 2.0, respectively.

IMPT plans were generated using a compact proton beam machine dedicated to pencil beam scanning. (Proteus® One, Ion Beam Applications S.A., Belgium). The Monte Carlo dose calculation algorithm was used in the RayStation. One noncoplanar and four coplanar beams were used considering the target coverage and the dose reductions of OARs. A setup robustness of ±2 mm was used and the range uncertainty of ±3.5% was assumed. The methods of robust setting and worst-case evaluation were previously described in detail.

Comparison of Dose-Volume Indices

The evaluated dosimetric parameters of the WB target were D2%, D95%, D98%, mean dose (Dmean), homogeneity index (HI), and conformity index (CI). The evaluated dosimetric parameters of all OARs were Dmax. In the analysis of hippocampus, D100% and Dmean were added. In the analysis of scalp, V20Gy (percentage of the volume receiving at least 20 Gy), V10Gy, and Dmean were added, respectively. HI was defined as a value calculating D1% divided by D99%. The definition of CI was previously described in detail. Differences of dosimetric parameters between the two techniques were analyzed by Student's t-test. Statistical analyses were carried out with EZR, a graphical user interface for R Version 3.4.1 (The R Foundation for Statistical Computing, Vienna, Austria). P < .05 was considered to indicate a significant difference.

Results

Target Coverage, Homogeneity, and Conformity

The median volume of the WB target was 1486 cm3 (range, 1158-1601 cm3). The median volume of hippocampus, HA region, scalp was 3.95 cm3 (range, 2.50-4.37 cm3), 26.6 cm3 (range, 20.3-28.0 cm3), and 321 cm3 (range, 238-388 cm3), respectively. A representative example of dose distribution is shown in Figure 1. The results of quantitative analysis of the WB target in the two techniques are shown in Table 2. DVH comparison of the WB target between HT and IMPT is shown in Figure 2A. The D2% (HT vs IMPT: 35.2 ± 0.2 Gy vs 33.8 ± 0.91 Gy, P = .001), D95% (HT vs IMPT: 29.9 ± 0.07 Gy vs 30.0 ± 0.05 Gy, P < .001), D98% (HT vs IMPT: 26.7.1 ± 0.95 Gy vs 28.1 ± 0.34 Gy, P = .002) and Dmean (HT vs IMPT: 32.8 ± 0.13 Gy vs 32.0 ± 0.60 Gy, P = .008) were all significantly better in IMPT than HT. The HI (HT vs IMPT: 1.50 ± 0.07 vs 1.28 ± 0.05, P < .001) and CI (HT vs IMPT: 1.30 ± 0.09 vs 1.14 ± 0.02, P < .001) were significantly better in IMPT than HT.

Figure 1.

A representative example of dose distribution in the transverse, sagittal, and coronal planes.

Table 2.

Target Coverage, Homogeneity, and Conformity.

Parameter	HT		IMPT		P-value
	Mean	SD	Mean	SD
D2% (Gy)	35.2	.42	33.8	.91	.001
D95% (Gy)	29.9	.07	30.0	.05	< .001
D98% (Gy)	26.7	.95	28.1	.34	.002
Dmean (Gy)	32.8	.13	32.0	.60	.008
HI	1.50	.07	1.28	.05	< .001
CI	1.30	.09	1.14	.02	< .001

Abbreviations: CI, conformity index; D2, D95%, D98%, dose to 2%, 95%, 98% volume of the target; Dmean, mean dose to the target; HI, homogeneity index; HT, helical tomotherapy; IMPT, intensity-modulated proton therapy; SD, standard deviation.

Figure 2.

Purple lines represent helical tomotherapy. Orange lines represent intensity-modulated proton therapy. A: Dose-volume histogram of the whole brain target. B: Dose-volume histogram of the hippocampus. C: Dose-volume histogram of the scalp.

OAR Sparing

The dose to the OAR of the two techniques is shown in Table 3. DVH comparison of the hippocampus (Figure 2B) and scalp (Figure 2C) between HT and IMPT is shown in Figure 2B and 2C. The D100% (HT vs IMPT: 9.3 ± 0.57 Gy vs 3.8 ± 0.41 Gy, P < .001) and Dmean (HT vs IMPT: 11.1 ± 0.51 Gy vs 7.0 ± 0.26 Gy, P < .001) of hippocampus were significantly reduced in IMPT compared with HT. The V20Gy (HT vs IMPT: 56.7 ± 13.4% vs 6.6 ± 1.9%, P < .001), V10Gy (HT vs IMPT: 90.5 ± 3.7% vs 37.1 ± 3.6%, P < .001) and Dmean (HT vs IMPT: 19.9 ± 1.9 Gy vs 10.0 ± 0.5 Gy, P < .001) of the scalp were significantly reduced in IMPT compared with HT.

Table 3.

Dose to Organs at Risk.

Structure Parameter		HT		IMPT		P-value
		Mean	SD	Mean	SD
Hippocampus	Dmax (Gy)	14.7	.67	15.4	.15	.013
	D100% (Gy)	9.3	.57	3.8	.41	< .001
	Dmean (Gy)	11.1	.51	7.0	.26	< .001
Scalp	Dmax (Gy)	31.7	1.7	30.8	2.0	< .001
	V20Gy (%)	56.7	13.4	6.6	1.9	< .001
	V10Gy (%)	90.5	3.7	37.1	3.6	< .001
	Dmean (Gy)	19.9	1.9	10.0	.5	< .001
Optic nerves	Dmax (Gy)	30.2	1.1	32.8	.2	< .001
Chiasm	Dmax (Gy)	31.8	1.2	32.8	.2	.029
Right eye	Dmax (Gy)	32.0	2.5	26.7	2.7	< .001
Left eye	Dmax (Gy)	31.8	2.3	26.7	2.7	< .001

Abbreviations: Dmax, maximum dose; Dmean, mean dose; D100%, dose to 100% volume of organs at risk; HT, helical tomotherapy; IMPT, intensity-modulated proton therapy; SD, standard deviation; V10, 20Gy, percentage of the volume receiving at least 10 or 20 Gy.

Discussion

This was the first pilot study comparing IMRT and IMPT for HA-WBRT. WBRT is the standard treatment modality for patients with brain metastases. However, cognitive decline after WBRT is a serious problem. Clinical studies reported that a low radiation dose to the hippocampus induced cognitive decline. Radiation-induced cognitive decline was also confirmed in vivo.[22,23] Although the mechanism of radiation-induced cognitive decline is complicated, apoptosis in the neurogenesis zone is considered an important cause.[24,25] Single-dose radiation at 2 to 10 Gy induced dose-dependent apoptosis in the neurogenesis zone of the hippocampus in vivo. Accordingly, reducing the low radiation dose to the hippocampus is essential to prevent cognitive decline. A phase Ⅱ trial reported that HA-WBRT preserved memory function and QOL. A randomized phase III trial (NRG CC001) demonstrated that HA-WBRT plus memantine (a neuroprotective agent) significantly reduced cognitive decline compared with WBRT plus memantine, while no differences were observed in intracranial progression-free survival and overall survival. In assessing organ at risk dose volume analysis score, per protocol and acceptable variation cases were 237 (91.5%) of the HA-WBRT plus memantine arm (n = 261). In contrast, a randomized phase III trial (NCT 01780675) did not demonstrate a reduction of cognitive decline between HA-WBRT and WBRT for prophylactic cranial irradiation of small cell lung cancer. In NCT 01780675, a dose of 6 Gy to the hippocampus without neuroprotective agents may have been too high to prevent the neuron cell death. So far, the role of HA-WBRT using photons without neuroprotective agents in reducing cognitive decline is controversial. However, HA-WBRT has a potential to replace the standard technique of WBRT for brain metastases.

Proton therapy using pencil beam scanning improves the target coverage, homogeneity, conformity, and irradiated regions of normal tissues compared with IMRT and proton therapy using traditional passive scattering. Furthermore, proton therapy reduces the radiation dose to normal tissues. In the present study, D2%, D95%, D98%, Dmean, HI, and CI of the target were significantly improved in IMPT than HT. Moreover, the D100% and Dmean of the hippocampus were significantly reduced in IMPT (HT vs IMPT: D100%, 9.3 Gy vs 3.8 Gy, P < .001; Dmean, 11.1 Gy vs 7.0 Gy, P < .001). Thus, HA-WBRT using IMPT has a superiority of reducing low radiation dose to the hippocampus over HA-WBRT using HT. In the previous study, the Dmax of hippocampus was higher in 9-field IMRT than HT (HT vs IMRT: 12.8 Gy vs 15.3 Gy, P = .001). In the present study, the Dmax of hippocampus was slightly higher in IMPT than HT (HT vs IMPT: 14.7 Gy vs 15.4 Gy, P = .013). A similar tendency of Dmax was observed and the optimization in IMPT should be done carefully.

In recent years, the QOL of patients with cancer has been an increasing concern. Hair loss due to chemotherapy and radiotherapy has a significant negative impact on QOL, even if temporarily.[29,30] Therefore, several studies were conducted to prevent radiation-induced alopecia. Although the tolerance dose to the hair follicles was unclear, previous studies reported the estimated tolerance dose to the scalp for preventing alopecia. A dosimetric analysis of six patients preserving the hairs after WBRT using IMRT reported that the mean dose and V24Gy were 16.3 Gy and 9.0 cm3. Studies of scalp dose analysis related to WBRT were summarized in Table 4.[31-34] Although the DVH comparison among different studies should be done carefully because the definition of scalp was slightly different for each study, a mean dose of 10.0 Gy and V24Gy of 5.4 cm3 in HA-WBRT using IMPT were a favorable outcome. It was reported that skull thickness varies with age, and significant aging-related reduction of skull thickness was observed in females. The reduction of skull thickness means the clinical target is more closed to the scalp. Scalp-sparing WBRT using IMPT may be more useful for elderly female patients. A psychological study demonstrated that chemotherapy-induced alopecia dramatically worsened QOL in women with breast cancer. Considering this outcome, preventing radiation-induced alopecia may be meaningful for women with brain metastases from breast cancer. Also, the application to cranio-spinal treatment may be beneficial. Further studies are needed to establish which patients benefit the most by the use of scalp-sparing WBRT. In general, proton therapy is not employed in palliative treatment at present; however, the benefit of keeping QOL should be considered in future treatment strategies.

Table 4.

Studies of Scalp Dose Analysis Related to Whole Brain Radiotherapy.

Author	n	Pathology	Treatment	Prescribed Dose	Rate of Decrease in Scalp Dose
Mahadevan et al 31	6	metastases	11-field IMRT	30 Gy/10fxs	46%, mean dose 16.3 Gy
Mancini et al 32	9	metastases	4-field IMRT	37.5 Gy/15fxs	62%, mean dose 14.3 Gy
			7-field IMRT		58%, mean dose 15.6 Gy
			13-field IMRT		61%, mean dose 14.5 Gy
			VMAT		54%, mean dose 17.2 Gy
Kao et al 33	17	metastases	3-field IMRT	37.5 Gy/15fxs	56%, mean dose 16.4 Gy
De Puysseleyr et al 34	10	metastases	VMAT	20 Gy/5fxs	37%, mean dose 12.6 Gy

Abbreviations: IMRT, intensity-modulated radiation therapy; VMAT, volumetric-modulated arc therapy; fxs, fractions. Note: Rate of decrease in scalp dose was calculated by subtracting the prescribed dose with the scalp mean dose and then divided it with the prescribed dose.

This study had several limitations. First, it was a preliminary study and the number of cases was small. Second, the comparison among other IMRT techniques such as volumetric modulated arc therapy and fixed-field IMRT was not conducted. Third, the correlations between the dose constraints for hippocampus and scalp and the reductions of symptoms have not yet been fully proven.

Conclusions

In conclusion, both techniques provided acceptable target dose coverage with sufficient hippocampus- and scalp-sparing. Especially, HA-WBRT using IMPT is a promising treatment to prevent cognitive decline and alopecia. Additional dosimetric and clinical studies are required to prove the feasibility and efficacy of HA-WBRT using IMPT.