The radiobiological effects of He, C and Ne ions as a function of LET on various glioblastoma cell lines

Abstract

The effects of the charged ion species ⁴He, ¹²C and ²⁰Ne on glioblastoma multiforme (GBM) T98G, U87 and LN18 cell lines were compared with the effects of 200 kVp X-rays (1.7 keV/μm). These cell lines have different genetic profiles. Individual GBM relative biological effectiveness (RBE) was estimated in two ways: the RBE₁₀ at 10% survival fraction and the RBE_2Gy after 2 Gy doses. The linear quadratic model radiosensitivity parameters α and β and the α/β ratio of each ion type were determined as a function of LET. Mono-energetic ⁴He, ¹²C and ²⁰Ne ions were generated by the Heavy Ion Medical Accelerator at the National Institute of Radiological Sciences in Chiba, Japan. Colony-formation assays were used to evaluate the survival fractions. The LET of the various ions used ranged from 2.3 to 100 keV/μm (covering the depth–dose plateau region to clinically relevant LET at the Bragg peak). For U87 and LN18, the RBE₁₀ increased with LET and peaked at 85 keV/μm, whereas T98G peaked at 100 keV/μm. All three GBM α parameters peaked at 100 keV/μm. There is a statistically significant difference between the three GBM RBE₁₀ values, except at 100 keV/μm (P < 0.01), and a statistically significant difference between the α values of the GBM cell lines, except at 85 and 100 keV/μm. The biological response varied depending on the GBM cell lines and on the ions used.

INTRODUCTION

Glioblastoma multiforme (GBM), a Grade IV malignant astrocytic glioma, is the most common subtype and the most lethal form of primary brain tumour [1]. GBM has the hallmarks of uncontrolled cellular proliferation, diffuse infiltration, propensity for necrosis (and hypoxia), robust angiogenesis, intense resistance to apoptosis and extensive heterogeneous genomic instability [2]. For these reasons, it is difficult to treat GBM successfully. The standard of care for GBM is surgical resection, where possible, and radiotherapy with concomitant Temozolomide [3]. Radiotherapy has been a key treatment for GBM, and conventional radiotherapy consists of 60 Gy of external beam irradiation delivered 5 days per week in fractions of 1.8–2.0 Gy [4, 5]. However, 90% of GBMs usually recur near the original site or within the high-dose region [4–6] after standard radiotherapy treatment. Dose escalation to 70–80 Gy can achieve local control, but relapse then occurs elsewhere in the brain. The prognostic factors for GBM are World Health Organization (WHO) classification, age of patient, tumour location, neurologic performance status, extent of surgical resection, proliferative indices, and genetic alterations [7].

The Cancer Genome Atlas (TCGA) studied GBM as its first cancer type. TCGA aims to catalogue and discover major cancer-causing genome alterations in large cohorts of human tumours through integrated multidimensional analysis. It aims to establish this feasibility and has the power to rapidly expand our knowledge of the molecular basis of cancer and the patterns of mutation that may inform future therapeutic decisions, thus setting the stage for a new era in the discovery of cancer interventions [8]. Besides phenotypes, genotypes of the central nervous system tumours also play an important role in the effective treatment of GBM, such that Louis et al. [9] have incorporated molecular parameters into the 2016 WHO classification of tumours in the central nervous system, including GBMs, with the objective of improved diagnostic accuracy, as well as improved patient management and more accurate determination of prognosis and treatment response. Halperin has predicted, in his historical review of particle therapy and treatment of cancer, that insights into molecular biology might clarify the ideal particles for clinical situations [10]. In addition, Van Meir et al. [5] propose that, due to each individual GBM tumour’s uniqueness in its expression profile, personalized medicine may be indicated for defined therapies and homogenous responses for critical subgroups.

Although conventional radiotherapy has been the mainstay in treatment for GBM, it is one of the most radioresistant tumours due to its persistently hypoxic nature, its ability to repair radiation-induced injury (which is accomplished by aberrant/or amplified growth), and survival signalling pathways [5, 11], as well as changes in the phosphatase and tensin homolog (PTEN) and p53, both tumour suppressor genes that have been linked to radioresistance [5, 12–15].

Charged particle/ion radiotherapy potentially possesses physical and biological advantages over photons (megavoltage X-rays). The physical benefits of heavy, charged particles provided by the Bragg peak allow precise delivery of high radiation doses to tumours while minimizing destructive irradiation to normal tissues and critical organs at risk. Furthermore, its depth–dose distributions can be modulated/shaped (by means of the ‘spread-out Bragg peak’) to cover tumours of different shapes, and the increase in ion density Linear Energy Transfer (LET) [16] can also make it a potentially superior modality of radiation compared with photons. Increasing LET (within the clinical range) increases the Relative Biological Effectiveness (RBE), reduces the Oxygen Enhancement Ratio (OER), produces less variation in cell-cycle–related radiosensitivity and decreases the ability to repair radiation damage [17]. RBE is dependent on many factors, such as cell and tissue type, biological end point, culture condition, dose, dose rate and fractionation, charged particle type, LET, and oxygenation status [18, 19].

Numerous basic in vitro cellular studies on GBM cell lines using photons [20–24] and various heavy ions [25–30] have been carried out. Pilot Phase I and II clinical studies using heavy ions such as helium, neon and carbon ions have been undertaken with the aim of improving radiotherapy for GBM patients [31–38].

Treating GBM with neon ions was first investigated by Castro et al. [32] with 14 GBM patients, and an optimal prescribed dose was not demonstrated, except that there was a trend toward better results with higher doses. Further work could not be done, because the Berkeley accelerator was shut down for budgetary reasons.

Carbon ion radiotherapy has been made available in Japan at the National Institute of Radiological Science (NIRS) since June 1994. Since the successful report of carbon ion radiotherapy by Tsujii et al. from NIRS using the Heavy Ion Medical Accelerator (HIMAC) [39], much attention has been focused on carbon ions. The first GBM Phase I–II trials using carbon ion as a boost for treatment has been reported as successful by Mizoe et al. [34], but unfortunately it did not proceed to a Phase III clinical trial. It should be noted that the RBE used was 3. Combs et al. [37, 38] have reported two randomized Phase II trials to evaluate the use of carbon ion radiotherapy as a boost versus a proton boost for primary GBM patients after surgery and standard chemotherapy with Temozolomide; and using carbon ion versus fractionated stereotactic radiotherapy in patients with recurrent or progressive gliomas.

Due to its invasive, highly infiltrative nature and heterogeneous genetic alteration properties, GBM is known to be one of the most resistant cancers to aggressive multimodality treatments. Moreover, past GBM clinical trials with carbon and neon ion radiotherapy have not attained optimum effectiveness [32, 34]. The key aim of this study was to answer the question: is the same charged ion therapy effective for all GBM patients? To seek to answer this question, it was decided to estimate the RBE₁₀ and RBE_2Gy, α, β and α/β ratio for each individual GBM cell type. The response of these GBM cell types to ⁴He, ¹²C and ²⁰Ne ions as a function of LET and dose was investigated. To our knowledge, this is one of the first studies to use three different GBM cell lines with various charged ions and various LETs for each ion. Recent studies in the literature have mostly concentrated on other cell lines [40–44].

MATERIALS AND METHODS

Cell lines and cell maintenance

Three Grade IV GBM cell lines: T98G, U87 and LN18, with different genetic alterations, doubling times and morphologies were used, as shown in Table 1. The T98G cells were a gift from Mick Woodcock, Gray Institute for Radiation Oncology and Biology, Oxford, UK; the U87 cells were obtained from the Health Protection Agency Culture Collections (HPACC, Wiltshire, UK), and the LN18 cells were obtained from the American Type Culture Collection (ATCC, Middlesex, UK). All the cell lines were confirmed Mycoplasma free with a Lonza MycoAlert® Mycoplasma Detection Assay.

Table 1.

Characteristics of the established glioblastoma cell lines employed

Cell	p53 status/genotype	PTEN	Doubling time	Morphology
T98G	Mutated/homozygous	Mutated	22 ± 3 h	Fibroblastic
U87	Wild type/homozygous	Mutated	35 ± 4 h	Epithelial
LN18	Mutated/heterozygous	Wild	33 ± 4 h	Epithelial

Cell lines were maintained in 75 cm³ plastic flasks (T75 BD FalconTM 353084) in Eagle’s Minimum Essential Medium (MEM: Nissui Pharmaceutical Co. Ltd, Tokyo) supplemented with 10% fetal bovine serum (FBS: Hyclone, Thermo Scientific, USA) in a humidified 95% air/5% CO₂ incubator at 37°C. Cells were subcultured from a T75 plastic flask by rinsing in calcium- and magnesium-free phosphate-buffered saline (PBS), and exposed to 0.2% Trypsin solution containing 0.5 mM EDTA. Cell numbers were determined by Coulter Counter. For all experiments, 3 × 10⁵ cells of each cell line were inoculated into a 25 cm³ plastic flask (T25 BD Falcon 353014) 3 days before irradiation to enable the cells to be at approximately 85–90% confluent stage for all dose points.

Irradiation

Cells were irradiated with ⁴H, ¹²C and ²⁰Ne monochromatic ion beams accelerated by the HIMAC, Japan. The energy and dose average LET for each of the various ions used are shown in Table 2. Lucite absorbers with various thickness were used to change the energy/LET of the beams. The details of the HIMAC beam delivery system, physical characteristics, biological irradiation procedures and dosimetry have been described by Kanai et al. and Torikoshi et al. [45, 46]. The dose average LET values range from 1.7 to 100 keV/μm, as shown in Table 2. The dose rate of all ion beams used was ~3 Gy/min. A 200 kVp X-ray (20 mA) beam filtered with 0.5 mm Cu and 0.5 mm Al by a TITAN 320 irradiator (Ge Inspection Technologies Shimadzu, Japan) at a dose rate of 1.00 Gy/min ± 0.02 was used as the reference radiation. All the irradiations were carried out at room temperature. Two or more independent experiments were performed with each ion species and LET, except for ⁴He and ²⁰Ne (30 keV/μm) for LN18, for which only one experiment was performed, and no ⁴He experiments were carried out for U87 due to beam line unavailability. For dose calculation, fragmentations were taken into consideration and applied to convert particle fluence (Φ) to dose, as described in Suzuki et al. and Matsufuji et al. [27, 47]:

$$\mathrm{Dose}\left(\mathrm{Gy}\right)=\mathrm{1}{\mathrm{.6021}\times \mathrm{10}}^{-9}\times \mathrm{LET}\left(\mathrm{keV}/\mu \mathrm{m}\right)\times \mathrm{\Phi }\left({\mathrm{cm}}^{-2}\right)$$ (1)

Table 2.

Summary of the GBM cell lines’ α, β, α/β ratio, D₁₀, RBE₁₀, RBE_2Gy and the ratio of RBE_2Gy/RBE₁₀ fitted into the LQ model

Cell lines		T98G—p53 mutation homozygous and PTEN mutation
Energy	LET	α	β	α/β	D10	RBE	D2Gy	RBE	Ratio of RBE
MeV/n	keV/μm	(Gy−1)	(Gy−2)	(Gy)	(Gy)	D10	SF ratio	2 Gy	D2Gy/D10
200 kVp	1.70 ± 0.02	0.049 ± 0.001	0.109 ± 0.009	0.45	4.57 ± 0.023	1.00 ± 0.050	0.588 ± 0.021	1.00 ± 0.035	1.00
4He 150	2.30 ± 0.04	0.045 ± 0.001	0.116 ± 0.001	0.39	4.63 ± 0.026	1.00 ± 0.006	0.579 ± 0.004	1.02 ± 0.007	1.02
4He 150	13.0 ± 0.05	0.141 ± 0.004	0.104 ± 0.003	1.36	4.09 ± 0.045	1.12 ± 0.013	0.500 ± 0.007	1.18 ± 0.017	1.05
12C 290	13.3 ± 0.19	0.201 ± 0.002	0.053 ± 0.004	3.79	5.21 ± 0.006	0.88 ± 0.001	0.541 ± 0.009	1.09 ± 0.019	1.24
20Ne 400	30.0 ± 0.45	0.292 ± 0.001	0.141 ± 0.008	2.07	3.14 ± 0.065	1.46 ± 0.030	0.318 ± 0.010	1.85 ± 0.056	1.27
12C 290	85.0 ± 0.32	0.643 ± 0.009	0.075 ± 0.003	8.57	2.71 ± 0.008	1.69 ± 0.005	0.204 ± 0.002	2.88 ± 0.023	1.70
20Ne 400	85.0 ± 0.25	0.885 ± 0.007	0.042 ± 0.001	21.1	2.35 ± 0.006	1.95 ± 0.005	0.144 ± 0.001	4.08 ± 0.042	2.09
12C 135	100 ± 1.77	0.935 ± 0.010	0.025 ± 0.004	37.4	2.32 ± 0.003	1.97 ± 0.002	0.140 ± 0.000	4.20 ± 0.021	2.13
Cell lines		U87—p53 wild type and PTEN mutation
Energy	LET	α	β	α/β	D10	RBE	D2Gy	RBE	Ratio of RBE
MeV/n	keV/μm	(Gy−1)	(Gy−2)	(Gy)	(Gy)	D10	SF ratio	2 Gy	D2Gy/D10
200 kVp	1.70 ± 0.02	0.157 ± 0.006	0.068 ± 0.004	2.31	4.81 ± 0.078	1.00 ± 0.016	0.559 ± 0.002	1.00 ± 0.005	1.00
4He 150	2.30 ± 0.04	N/A	N/A	N/A	N/A	N/A	NA	NA	NA
4He 150	13.0 ± 0.05	N/A	N/A	N/A	N/A	N/A	NA	NA	NA
12C 290	13.3 ± 0.19	0.315 ± 0.001	0.071 ± 0.002	4.44	3.90 ± 0.040	1.23 ± 0.013	0.401 ± 0.004	1.39 ± 0.014	1.13
20Ne 400	30.0 ± 0.45	0.596 ± 0.004	0.029 ± 0.001	20.6	3.31 ± 0.009	1.45 ± 0.004	0.269 ± 0.001	2.08 ± 0.001	1.43
12C 290	85.0 ± 0.32	0.774 ± 0.006	0.111 ± 0.001	6.97	2.25 ± 0.050	2.14 ± 0.048	0.136 ± 0.005	4.11 ± 0.146	1.92
20Ne 400	85.0 ± 0.25	0.913 ± 0.010	0.003 ± 0.001	304	2.53 ± 0.035	1.90 ± 0.026	0.160 ± 0.003	3.49 ± 0.059	1.84
12C 135	100 ± 1.77	1.102 ± 0.101	0.039 ± 0.038	28.3	2.27 ± 0.034	2.12 ± 0.032	0.100 ± 0.034	5.59 ± 0.065	2.64

Cell lines		LN18—p53 mutation heterozygous and PTEN wild type
Energy	LET	α	β	α/β	D10	RBE	D2Gy	RBE	Ratio of RBE
MeV/n	keV/μm	(Gy−1)	(Gy−2)	(Gy)	(Gy)	D10	SF ratio	2 Gy	D2Gy/D10
200 kVp	1.70 ± 0.02	0.316 ± 0.001	0.076 ± 0.001	4.16	3.81 ± 0.010	1.00 ± 0.004	0.392 ± 0.002	1.00 ± 0.001	1.00
4He 150	2.30 ± 0.04	0.271 ± 0.000	0.038 ± 0.000	7.13	4.98 ± 0.000	0.76 ± 0.000	0.499 ± 0.000	0.79 ± 0.000	1.04
4He 150	13.0 ± 0.05	0.374 ± 0.000	0.038 ± 0.000	9.84	4.29 ± 0.000	0.89 ± 0.000	0.406 ± 0.001	0.97 ± 0.001	1.09
12C 290	13.3 ± 0.19	0.277 ± 0.003	0.075 ± 0.001	3.69	4.00 ± 0.010	0.95 ± 0.003	0.426 ± 0.004	0.92 ± 0.008	0.97
20Ne 400	30.0 ± 0.45	0.543 ± 0.000	0.061 ± 0.000	8.90	3.14 ± 0.000	1.21 ± 0.000	0.265 ± 0.000	1.48 ± 0.000	1.22
12C 290	85.0 ± 0.32	0.853 ± 0.014	0.227 ± 0.004	3.76	1.80 ± 0.001	2.11 ± 0.005	0.073 ± 0.001	5.37 ± 0.063	2.53
20Ne 400	85.0 ± 0.25	0.936 ± 0.031	0.039 ± 0.003	24.0	2.25 ± 0.050	1.70 ± 0.037	0.132 ± 0.007	2.97 ± 0.149	1.75
12C 135	100 ± 1.77	1.079 ± 0.002	0.052 ± 0.001	20.8	1.95 ± 0.001	1.95 ± 0.002	0.094 ± 0.001	4.17 ± 0.007	2.14

NA = not available, SF = surviving fraction. Errors indicate standard error of the mean.

Cell-survival assay

Cell inactivation was measured with the colony-formation assay to assess reproductive death. After irradiation, cells were removed from the T25 flask, and appropriate numbers were inoculated into triplicate 60 mm plastic dishes (Falcon 353002) to produce 60–70 colonies per dish. Cells were counted using a Coulter Counter (Coulter Electronics Ltd, Japan, Tokyo). The plating efficiency (PE) for T98G is 0.46 ± 0.007, for U87 0.15 ± 0.002, and for LN18 it is 0.95 ± 0.03, with the stated uncertainty showing the standard error of the mean (SEM). Although U87 has a PE of only 0.15, irradiation with doses of up to 4 Gy still produced ~40 colonies at a LET of 85–100 keV/μm. Fourteen days later, colonies were washed, fixed and stained with 0.2% crystal violet. Triplicate dishes of each dose point colony (consisting of >50 cells) were counted under a stereomicroscope. The surviving fraction (SF) of each dose was determined as the ratio of live colonies in the treated dish relative to the untreated/control dish. The mean values and standard deviations of triplicate samples were calculated.

Data and statistical analysis

Surviving fraction data were obtained from the mean of at least two or more independent experiments and fitted by the least squares Linear Quadratic (LQ) Model equation:

$$\mathrm{S}=\exp \left({-\mathrm{\alpha }\mathrm{D}-\mathrm{\beta }\mathrm{D}}^{\mathrm{2}}\right),$$ (2)

where S is the surviving fraction and D is the dose in gray. Doses were calculated from particle fluence and the LET values, as shown in Equation 1. The parameters α and β are constants describing the linear component and the quadratic part of the curve, respectively. The α and β parameters are determined by minimizing the sum of squares calculated by Equation 2. It is known that there is an almost linear relationship between the high-LET charged particle dose and cell killing for higher doses. However, the linear equation was not used to fit the survival curves due to the inherent low-dose hyper-radiosensitivity (HRS) of some GBM species [21, 22, 48]. Furthermore, low doses of 0.2, 0.25, 0.4, 0.5, 0.6 and 0.75 Gy were used to detect low-dose HRS of GBM cell lines. The highest dose used was 8 Gy for LET ≤85 keV/μm.

RBE is defined as the ratio of a reference photon dose D_γ and a corresponding ion dose D_I yielding the same biological effect:

$${\mathrm{RBE}}_{\mathrm{10}}{=\mathrm{D}}_{\mathrm{\gamma }}{/\mathrm{D}}_{\mathrm{I}}$$ (3)

RBE₁₀ is the dose ratio at a surviving fraction of 10% for X-rays relative to each ion.

RBE_2Gy is the ratio of the surviving fraction at 2 Gy for X-rays in relation to each ion. RBE_2Gy is of particular interest, because some strains of GBM are known to exhibit low-dose HRS after single low doses [21, 22, 48]. Moreover, most of the fractionated doses given clinically are between 1.8 and 2 Gy.

The α/β ratio is the dose, in Gy, when the number of cells killed by the linear component is equal to the cell kill from the quadratic component in the linear quadratic equation. The survival curves versus LET were plotted with KaleidaGraph by Synergy software (version 3.5). Analysis of variance (ANOVA) of the mean value of α and RBE₁₀ values for each individual GBM cell line and ion were analysed using Xcel ANOVA for a single factor (Analysis Tool-kit). P values of <0.01 were considered significant.

RESULTS

A summary of all the GBM results with the various ions at different energies and LETs are shown in Table 2. Uncertainties in LET are expressed as the SEM, with uncertainties for all other parameters expressed as the standard deviation (SD).

D₁₀ is the dose required to inactivate the cells to 10% survival.

RBE₁₀ and RBE_2Gy and the RBE_2Gy/RBE₁₀ ratio and a comparison of the effects of genetic mutation of GBM are presented in Table 2. Figures 1–3 show survival curves of X-rays (1.7 keV/μm) and ions (up to 100 keV/μm). Figure 1A compares the radiosensitivity of the three GBM cell lines to X-rays, Fig. 1B compares their radiosensitivity to ¹²C ions (13.3 keV/μm), and Fig. 1C compares their radiosensitivity to ²⁰Ne (entrance LET of 30 keV/μm). Figure 2 compares T98G and LN18 with ⁴He ions at 2.3 and 13 keV/μm, respectively. Figure 3 compares ¹²C (85 and 100 keV/μm) and ²⁰Ne ions (85 keV/μm). Figure 4 demonstrates T98G, U87 and LN18 trends for α and RBE₁₀ in relation to LET, and Fig. 5 the β parameters as a function of LET.

Fig. 1.

Survival curves of T98G-U87-LN18 with (A) X-rays (1.7 keV/μm), (B) ¹²C 290 MeV/n (13.3 keV/μm) and (C) ²⁰Ne 400 MeV/n (30 keV/μm). Error bars indicate standard deviations.

Fig. 3.

(A and B) Survival curves comparing LETs of 85 keV/μm of ¹²C 290 MeV/n and ²⁰Ne 400 MeV/n, respectively, and (C) ¹²C 135 MeV/n (100 keV/μm). LN18 is more radiosensitive and T98G more radioresistant to ¹²C (85 and 100 keV/μm). U87 is more radioresistant to ²⁰Ne (85 keV/μm). Error bars indicate standard deviations.

Fig. 2.

Survival curves of T98G and LN18 with ⁴He 150 MeV/n (A) at 2.3 keV/μm and (B) at 13 keV/μm. LN18 is more radioresistant than T98G. Error bars indicate standard deviations.

Fig. 4.

Comparison of trends for T98G, U87 and LN18 α parameters and RBE₁₀ as a function of LET.

Fig. 5.

T98G, U87 and LN18 β parameters as a function of LET.

Table 2 shows the RBE₁₀ values of T98G relative to X-rays, ranging from 0.88 to 1.97. The α values ranged from 0.045 to 0.935. The T98G α and RBE₁₀ values increased with LET (Figs 1–4). There was no discernible trend for β values (Fig. 5). The α/β values ranged from 0.39 to 37.4. The α/β values for high LET of ²⁰Ne and ¹²C, with values of 21.1 and 37.4 respectively, have been highlighted to show the large increase when compared against C, Ne and He ion beams of low LET. Doses of 2 Gy produced higher RBE (Table 2), and RBE_2Gy ranged from 1.02 to 4.20 (Table 2). The ratio of RBE_2Gy to RBE₁₀ was 1.02 to 2.13. RBE plays an important role in the treatment of patients, because it directly affects the resulting biological damage from radiation, and the common RBE used is at 10% survival. RBE_2Gy is of interest because the clinically fractionated doses given were mostly between 1.8 and 2 Gy. Fertil and Malaise have reported that 2 Gy irradiation could provide valuable information and is a good indicator for intrinsic tumour cell radiosensitivity to photon radiation [49].

U87’s RBE₁₀ values ranged from 1.23 to 2.14 (Table 2), and increased with LET up to 85 keV/μm. U87 α and RBE₁₀ values showed a trend of progressively increasing with ion mass and LET. The increase in α was prominent and consistent, but β values of U87 did not increase with LET and heavier ion mass (Figs 4 and 5). U87 α/β values ranged from 2.31 to 304 (Table 2). The RBE_2Gy ranged from 1.39 to 5.59, and the ratio of RBE_2Gy to RBE₁₀ ranged from 1.13 to 2.64 (Table 2).

LN18 RBE₁₀ values ranged from 0.76 to 2.11 (Table 2). There was a trend of LN18 α and RBE₁₀ values increasing with LET and ion mass (Fig. 4). The β values did not show any consistent trend (Fig. 5). α/β values above 20 are seen for LET ≥85 keV/μm. The RBE_2Gy of LN18 ranged from 0.79 to 5.37, and the ratio of RBE_2Gy to RBE₁₀ ranged from 0.97 to 2.53. Only with LN18 was there a lower than unity ratio of RBE_{2 Gy} to RBE₁₀ (Table 2).

The results demonstrate that treatment of GBM could be ion specific with respect to effective cell killing. For example, for cell line LN18, ¹²C ions at 85 keV/μm were observed to be the most effective ion for cell inactivation. In contrast, ²⁰Ne ions were more effective for T98 inactivation. The most effective LET for U87 and LN18 was that of the ¹²C ion at 85 keV/μm. Conversely, the ¹²C ions at 100 keV/μm did not increase cell inactivation, measured in terms of D₁₀ survival; indeed, the reverse was true. ¹²C ions at LET of 100 keV/μm were more effective for the cell line T98G, and ²⁰Ne ions at LET of 85 keV/μm were also effective for T98G (Table 2). The LET of the ⁴He ions at 2.3 and 13 keV/μm was considered to be too low for any significant inactivation of GBM cells. However, due to limited beam time availability, ⁴He experiments were not conducted for U87.

The difference between the mean each of the GBM α-values has statistical significance (P < 0.01), except for that for ²⁰Ne ions (LET 85 keV/μm) and ¹²C ions (LET 100 keV/μm); and between the mean RBE₁₀, except for ¹²C ions (LET 100 keV/μm).

DISCUSSION

The charged ion irradiations to the three GBM cell lines resulted in a range of biological effectiveness. The responses of these GBM cell types to ⁴He, ¹²C and ²⁰Ne as a function of LET and dose also differed from one GBM type to the other. The results for RBE₁₀ versus RBE_2Gy, RBE₁₀ as a function of LET, α and β versus LET; comparison of previously reported data using GBM tumour cells, and the GBM genotype characteristics are discussed below.

RBE₁₀ and RBE_{2 Gy}

The RBE values of ions are relevant for treatment planning as they are used to calculate the Gy Equivalent dose (GyE). However, this is dependent on many factors, leading to further uncertainties. The classic RBE is conservatively taken from D₁₀, but the RBE at D_2Gy may need to be considered because the doses given to patients are conventionally 1.8–2 Gy per fraction. From Table 2, RBE_{2 Gy} for U87 peaks at 100 keV/µm, and RBE₁₀ for U87 peaks at 85 keV/μm. For LN18, the RBE₁₀ and RBE_{2 Gy} both peak at 85 keV/μm. The U87 and LN18 RBE₁₀ were not in agreement with reported RBE₁₀ peak values at 100–200 keV/μm [40, 41, 50]. These differences in RBE values could be related to the repair capacity of the cells [51]. There was a trend for RBE₁₀ and RBE_{2 Gy} to increase with LET, and for T98G the RBE₁₀ reported here may not reflect the true peak measured RBE because the maximum LET used was 100 keV/μm [52, 53].

However, our results do concur with previous reports in that different ions with different energies but the same LET had different RBE₁₀, as shown by the results for ²⁰Ne and ¹²C-ions with LET of 85 keV/μm (Table 2 and Fig. 3) [50]. Our results were also in agreement with Tobias et al. [54] in that RBE₁₀ was dependent on cell type, because each individual GBM cell line studied had a different genetic alteration. The intrinsic and genetic alterations of the GBM cells may be the parameters that determine the radiosensitivity. In general, there is a trend for RBE₁₀ to increase with the Z values of ions. RBE is dependent on many factors and could be selected from D₁₀, D₃₇ or D₁ [17]; D_{2 Gy} may also need to be considered, because the clinical doses given to patients are typically 1.8–2 Gy per fraction. Moreover, RBE_{2 Gy} is important, because some strains of GBM exhibit low dose HRS [20, 21]. Using RBE₁₀ alone may miss the intrinsic and inherent subtle nature of some strains of this tumour. It has been reported that 2 Gy irradiations provide the most valuable information and could be a good indicator for intrinsic tumour cell radiosensitivity to photon radiation [49]. The RBE was higher with D_{2 Gy}, and the D_{2 Gy/}D₁₀ ratio could demonstrate (infer) the radiobiological effectiveness of the respective ions and LET at 2 Gy; and may indicate the potential ion of choice. From both RBE₁₀ and RBE_{2 Gy}, there is an inference that, for LET values of 85 keV/μm, ²⁰Ne ions (85 keV/μm) are appropriate for T98G; and ¹²C ions (85 keV/μm) are more effective for both U87 and LN18 (Fig. 3A and B; Table 2). In addition to RBE_2Gy and its ratio to RBE₁₀; the ratio of plateau to Bragg peak LET may be significant; for example, with T98G, ¹²C (LET 13.3 keV/μm) had a RBE₁₀ of 0.88, and for ¹²C (LET 85 keV/μm) the RBE₁₀ was 1.69. The plateau to Bragg peak ratio was therefore 1.92. For U87, the ratio was 1.74 (2.14/1.23), and for LN18 2.22 (2.11/0.95). This suggests that the RBE along the entire depth–dose curve should be taken into consideration.

RBE as a function of LET

The trend of RBE increasing with LET was observed in the T98G cell line, but not for U87 or LN18. U87 and LN18 RBE₁₀ peaked at ¹²C (LET 85 keV/μm), and an increase in LET to 100 keV/μm resulted in lower RBE (Fig. 4). The results show that higher LET did not equate to higher RBE for these GBM cell lines. The true peak RBE for T98G was not ascertained, because the maximum LET studied here was 100 keV/μm. Since the LET range studied was limited, it would not be appropriate to plot the RBE as a function of Z*²/v² (where Z* is the effective charge and v is the relative velocity of the ion) instead of LET [55]. Further work using an appropriate extended range of LET may be warranted.

α, β and LET

The α value characterizes the linear component of cell kill in the linear quadratic equation. The results show there was a trend of α increasing for LET up to 100 keV/μm (Fig. 4). This is in agreement with Hall and Giaccia, that the optimal LET for producing a biological effect is 100 keV/μm [17]. At this density of ionization, the average separation between ionizing events approximately coincides with the 2 nm diameter width of a DNA double helix and has the highest probability of causing a double-strand break by a single charged particle. There are also ion-kill and delta ray (equivalent to low-LET) effects, which lead to variation in radiosensitivity with LET [55]. There was a trend for T98G to be more radioresistant than U87 and LN18 with low LET. However, this trend was not observed with high LET, where the type of ion used was the dominant factor in determining radiosensitivity. The β values did not show any trend (Fig. 5).

Comparison with previously reported results

Comparing our results with others, the T98G α and β values for X-rays reported by Suzuki et al. (0.064/0.046) differed to some extent from those in this report (0.049/0.109). Also, for the ¹²C ion (LET 13.3 keV/μm), our results (0.201/0.053) differed from Suzuki et al.’s (0.127/0.065), as did those for the ¹²C ion (LET 77 keV/μm versus 85 keV/μm) (0.432/0.0895 versus 0.643/0.075). This could be due to the different dose rates: for X-rays 0.85 Gy/min vs 1.0 Gy/min, and for ¹²C ions 1.2 Gy/min vs 3.0 Gy/min [27]. U87 and LN18 have increased radiosensitivity to ¹²C ions compared with ²⁰Ne ions at the same LET of 85 keV/μm, which concurs with the results of Suzuki et al., who reported that ¹²C ions are more effective at the same LET [42] for mutation frequency/yield. In addition, it has been suggested that beams of different ion species could cause different types of DNA damage [56]; different ion species with different structures of energy deposition may cause qualitatively different damages, even with the same LET [42], and a difference in the track structure of the core and penumbra exist for different kinds of ions with similar LET [57]. Our results were also in agreement with those of Short et al., that T98G and U87 exhibited HRS to X-ray irradiation [20, 21]. Short et al. state that HRS usually occurs at doses of <1 Gy with X-rays; our slightly different results could be due to our dose rate of 1 Gy/min (as compared with that of Short et al., 0.2–0.4 Gy/min). Furthermore, the methods and medium employed were different. Low-dose HRS is common in radioresistant glioma and is more marked in more radioresistant cell lines [21]. There is an inference that treating GBM with high fractionated doses for both low- and high-LET radiation may not be beneficial for all GBM cell lines.

Comparing ¹²C ion irradiation of U87 by Combs et al. [29], despite the different setting, materials and methods (X-ray dose rates of 2.3 versus 1 Gy/min and ¹²C ions with a LET of 103 versus 100 keV/μm), irradiation method (active raster-scanning techniques versus passively delivered ¹²C ions) and media used, the results varied only slightly in α values: X-rays 0.176 versus 0.157, ¹²C ions RBE_α 4.77 versus 7.0 and ¹²C ion RBE₁₀ 2.38 versus 2.12.

GBM genotypes

GBM has the hallmark of extensive genomic instability [2] and is known to be radioresistant to conventional X-ray therapy. However, T98G is one of the strains of GBM that demonstrate HRS to doses of <1 Gy but radioresistance to high doses [20]. It has been suggested that inducible repair processes are important components of radioresistance apparent in these cells at high doses [20]. The genotypes of GBM are directly related to radioresistance. p53 and PTEN mutations are two known genetic alterations of GBM [4, 58]. Mutated p53 is known to be involved in radioresistance to photons and charged ion radiations. Carbon ion irradiations can potentially overcome this resistance due to their high LET [25]. PTEN (on chromosome 10) and p53 are both tumour suppressor genes that have been linked to radioresistance [5, 12–15]. Therefore, GBM cell lines with wild-type p53 and PTEN will be more radiosensitive than the GBM cell lines with mutated p53 and PTEN. Moreover, the severity of the mutations also plays an important part (heterozygous or homozygous). From the genotypes of the GBM studied, namely, T98G (p53 homozygous mutated and PTEN mutated), U87 (p53 wild-type homozygous and PTEN mutated) and LN18 (p53 heterozygous mutated and wild-type PTEN), the expectation would be that T98G would be the most radioresistant, U87 would be of intermediate resistance and LN18 would be the least radioresistant.

U87 has homozygous wild-type p53 and PTEN mutation, whereas LN18 has p53 heterozygous mutation and wild-type PTEN, and they both have p53 wild type (one or two alleles). Wild-type p53 tumour suppressor is a key mediator of an ATM-dependent DNA damage response cascade following cellular exposure to ionizing radiation. Mutated p53 can give rise to alteration in G1 and G2 cell-cycle checkpoint control, cell death, DNA repair, and genetic stability. As T98G has homozygous mutated p53, they will be expected to be more radioresistant as compared with U87 and LN18.

Although, both U87 and T98G have mutated PTEN, they were not compared, because it has been found that the PTEN locus revealed a p53 binding element directly upstream of the PTEN gene in the regulation of PTEN transcription by p53 [15], and that PTEN was required for p53-mediated apoptosis and demonstrates a functional interaction between these two genes [15]. Since T98G has both mutated p53 and PTEN, the expectation is that they will be more radioresistant.

Our U87 and LN18 results have demonstrated that wild-type p53 cell lines (homozygous or heterozygous) are not radioresistant to ¹²C ions with high LET (85–100 keV/μm). In contrast, T98G, with mutated p53 (homozygous), exhibits generally lower α values, an indication of resistance (Table 2).

Another observable classification of GBM cell lines can be based on ion specificity, showing that U87 and LN18 cell lines are more sensitive to ¹²C than to ²⁰Ne (Fig. 1B and Fig. 3A). Conversely, T98G is more radiosensitive to ²⁰Ne ions (Fig. 1C and Fig. 3B). LN18 is radioresistant to ⁴He ions at high doses, but is radiosensitive to low doses, and the reverse is true for T98G. The survival curves (Fig. 2A and B) are not equal in size: B is smaller than A.

Due to the genetic differences between these tumours, it has been suggested that individual predictive assays (and tailored treatments) may prove beneficial for defining growth parameters and determining inherent radiosensitivity suited to high-LET therapy [31]. Probably, with charged ion radiation, eligibility criteria may be essential for the benefit of patients [59], as has been demonstrated for tumours in patients with 1p and 19q co-deletion, which are sensitive to chemotherapy with procarbazine, lomustine and vincristine. The status of chromosomes 1p and 19q are now used as eligibility criteria, rather than histologic assessment [4]. With this precedent, it may be necessary for a paradigm shift in stratifying genetic alteration in the design of clinical trials for GBM patients using charged particle radiotherapy.

There was a statistically significant difference (P < 0.01) between the mean RBE₁₀ (except for LET 100 keV/μm) and α value [except for ²⁰Ne ions (LET 85 keV/μm) and ¹²C ions (LET 100 keV/μm)]. There is evidence that personalized charged ion radiotherapy is indicated for GBM patients [5, 10, 31]. Moreover, there is ion specificity for GBMs, as it was shown for U87 and LN18 cell lines that ¹²C was more effective at inactivation than ²⁰Ne, whereas the T98G cell line was more radiosensitive to ²⁰Ne ions. It was demonstrated that individual GBM cell types responded differently to the same LET and same charged ion radiation. The genotypes of GBM are likely to be directly associated with radiosensitivity and radioresistance, because there was a statistically significant difference between the determined α values and RBE₁₀ parameters. The influence of the choice of ion, energy and treatment dose on the benefits for GBM patients may have to be considered. It may be necessary to stratify GBM genetic alteration in the design of future clinical trials for GBM patients using charged ion radiotherapy, rather than only using the histopathological GBM types.

In conclusion, the individual GBM cell types respond differently to the same LET and same charged ion radiation. The statistically significant difference (P < 0.01) between the mean RBE₁₀ (except for LET 100 keV/μm) and the α value [except for ²⁰Ne ions (LET 85 keV/μm) and ¹²C ions (LET 100 keV/μm)] infer that the same charged ion therapy may not be effective for all GBM types. In addition, the genotypes of GBM may also be contributory factors to effective treatment of GBM.

Future work on GBM response to ¹²C and ²⁰Ne ions using a wider range of LET for each ion species (with its unique Z value) is necessary in order to determine the maximum value of RBE, because this study could not determine the true maximum value of RBE. Of interest is also the effect of hypoxic conditions, with a LET of 85 keV/μm possibly providing a better understanding of the hypoxic tumour response to charged ion radiation. There is also a need to investigate the capacity of GBM cells to repair after ion irradiation.

CONFLICT OF INTEREST

The authors declare that there is no conflict of interest.

FUNDING

This work was supported in part by the International Open Laboratory NIRS and by NIRS/HIMAC; and was fully funded by the PARTNER of the European Commission, FP7 People (Marie Curie) Programme, under Grant Agreement No 215840, 2008–2012.