**2. Clinical studies of our modified boron neutron capture therapy**

Since the 1950s, BNCT has been used to treat malignant gliomas, although the results have not been satisfactory. We modified the therapy in several ways to resolve problems previously existing, and applied this modified BNCT to malignant gliomas beginning in 2002 by using KURRI (Kawabata et al. 2003). First, we utilized an epithermal rather than a thermal beam to improve the distribution of thermal neutrons in deep sites. Second, we used both of the boron compounds that are currently available worldwide for BNCT: sodium borocaptate (BSH) and boronophenylalanine (BPA). These compounds reach different subpopulations of tumor cells and accumulate in them in a different fashion.

#### **2.1 Survival benefit from modified boron neutron capture therapy in combination with external beam fractionated X-ray treatment**

We evaluate the clinical results of a form of tumor selective particle radiation, BNCT in combination with external beam fractionated X-ray treatment (XRT) for newly-diagnosed GB patients. Between 2002 and 2006, we treated 21 patients of newly-diagnosed GB with BNCT utilizing sodium borocaptate and boronophenylalanine simultaneously. The first 10 were treated with only BNCT (protocol 1), and the last 11 were treated with BNCT followed by fractionated XRT of 20 to 30 Gy (protocol 2) to reduce the possibility of local tumor recurrence. No chemotherapy was applied until tumor progression was observed.

#### **2.1.1 Methods and the study design of our modified BNCT with XRT**

As mentioned previously, we modified the protocol of BNCT as follows: First, we started using epithermal neutrons instead of thermal neutrons to obtain good penetration for deepseated lesions. Second, we simultaneously used 2 different boron compounds (BSH and BPA) with different accumulation mechanisms to the tumor cells (Ono *et al.* 1999, Yokoyama *et al.* 2006). Third, we utilized a dose simulation work station, the Simulation Environments for Radiotherapy Applications (SERA; Idaho National Laboratory, Idaho Falls, ID) or the JAERI computational dosimetry system (JCDS). Fourth, 18F-labeled BPA positron emission tomography (18F-BPA-PET) (Fig.2) was applied for the estimation of the boron compound

been synthesized (Hiramatsu *et al.* 2011, Miyata *et al.* 2011, Wu *et al.* 2007, Yang *et al.* 2006), but only two of these currently are being used in clinically. The first, which has been used primarily in Japan, is sodium borocaptate or BSH, and the second is a dihydroxyboryl derivative of phenylalanine referred to as boronophenylalanine or BPA (Barth *et al.* 2005). The latter has been used in clinical trials in Japan, Europe and the United States, primarily for the treatment of high grade gliomas, and more recently for recurrent tumors of the head and neck region (Ariyoshi *et al.* 2007, Haginomori *et al.* 2009, Kimura *et al.* 2009). Following i.v. administration of either BPA or BSH by i.v. infusion, the tumor site is irradiated with

Recently, BNCT studies carried out by us at Osaka Medical College (OMC) and Kyoto University Research Reactor Institute (KURRI), in which BPA and BSH were administered in combination (Kawabata *et al.* 2003, Miyatake *et al.* 2005, Miyatake *et al.* 2009) for the patients with recurrent tumor after irradiation, or BNCT followed by an X-ray boost showed favorable responses in patients with newly-diagnosed glioblastoma (GB) and especially

Since the 1950s, BNCT has been used to treat malignant gliomas, although the results have not been satisfactory. We modified the therapy in several ways to resolve problems previously existing, and applied this modified BNCT to malignant gliomas beginning in 2002 by using KURRI (Kawabata et al. 2003). First, we utilized an epithermal rather than a thermal beam to improve the distribution of thermal neutrons in deep sites. Second, we used both of the boron compounds that are currently available worldwide for BNCT: sodium borocaptate (BSH) and boronophenylalanine (BPA). These compounds reach different subpopulations of tumor cells and accumulate in them in a different fashion.

**2.1 Survival benefit from modified boron neutron capture therapy in combination with** 

We evaluate the clinical results of a form of tumor selective particle radiation, BNCT in combination with external beam fractionated X-ray treatment (XRT) for newly-diagnosed GB patients. Between 2002 and 2006, we treated 21 patients of newly-diagnosed GB with BNCT utilizing sodium borocaptate and boronophenylalanine simultaneously. The first 10 were treated with only BNCT (protocol 1), and the last 11 were treated with BNCT followed by fractionated XRT of 20 to 30 Gy (protocol 2) to reduce the possibility of local tumor

As mentioned previously, we modified the protocol of BNCT as follows: First, we started using epithermal neutrons instead of thermal neutrons to obtain good penetration for deepseated lesions. Second, we simultaneously used 2 different boron compounds (BSH and BPA) with different accumulation mechanisms to the tumor cells (Ono *et al.* 1999, Yokoyama *et al.* 2006). Third, we utilized a dose simulation work station, the Simulation Environments for Radiotherapy Applications (SERA; Idaho National Laboratory, Idaho Falls, ID) or the JAERI computational dosimetry system (JCDS). Fourth, 18F-labeled BPA positron emission tomography (18F-BPA-PET) (Fig.2) was applied for the estimation of the boron compound

recurrence. No chemotherapy was applied until tumor progression was observed.

**2.1.1 Methods and the study design of our modified BNCT with XRT** 

**2. Clinical studies of our modified boron neutron capture therapy** 

neutrons, the source of which is a nuclear reactor.

those in high risk groups (Kawabata *et al.* 2009a).

**external beam fractionated X-ray treatment** 

accumulation prior to neutron irradiation (Imahori *et al.* 1998). Fifth, we filled the tumor removed cavity with air to obtain enough neutron flux, especially for the bottom of deepseated tumors (Sakurai *et al.* 2006). Sixth, we developed a central shielding method with a lithium plate at the center of the irradiation field to obtain uniform neutron distribution and increase the neutron flux relativey at the periphery in the radiation field (Ono 2006, Ono *et al.* 2000). With these modifications, even patients with deep-seated tumors can be treated by BNCT without craniotomy with a short hospital stay. In the present study, the revised protocol was used as a new protocol as follows.

Fig. 2. 18F labeled BPA positron emmision tomography (18F - BPA PET) has been applied for the estimation of the boron compound accumulation prior to BNCT. The tracer is fluororide labelled boron compound. This PET ensures the effectiveness of BNCT. 18F-BPA accumulates well and distributes precisely in the tumor lesion and the infiltrating tumor zone.

Twelve hours before the neutron irradiation, the patients were administered 100mg/kg of BSH intravenously for one hour. 700mg/kg of BPA was infused continuously to the patients for 6 hours before the irradiation, and they were positioned for neutron irradiation in the reactor (KURRI or JRR-4 (Japan Atomic Energy Agency Research Reactor 4)). Just after termination of continuous BPA infusion for 6hrs, neutrons were irradiated. We used the dose-planning workstation to calculate the radiation dose for tumors from the 18F-BPA-PET data and blood 10B concentrations obtained every 2 hours after BSH administration. We used an epithermal neutron beam. Following this, a 2 Gy daily fraction of XRT was applied, for a total of 20 to 30Gy. The total dose of XRT was decided based on the BNCT dose for the normal brain. In Protocol 1, we aimed to apply more than 30 gray-equivalent (Gy-Eq) for gross tumor volume (GTV) and less than 12 Gy-Eq for normal brain, as BNCT. In Protocol 2, we aimed to apply more than 40 Gy-Eq for GTV and less than 15 Gy-Eq for normal brain. No chemotherapy was applied for any of the patients until the tumor progression was confirmed histologically or by 18F-BPA-PET (Miyashita *et al.* 2008).

Survival time from histologically diagnosed as GB was compared with the survival time for the institutional historical controls who were treated by surgical removal followed by XRT and chemotherapy (mainly nitrosourea) from 1990 to 2006 in OMC. In this historical control

Clinical Study on Modified Boron Neutron Capture Therapy for Newly Diagnosed Glioblastoma 329

The RPA class distribution of 21 patients treated with BNCT at the initial diagnosis was as follows: Class III = 6 (29%); Class IV = 6 (29%); Class V = 8 (38%); Class VI = 1 (5%). The MSTs of the patients in classes III, IV, V, and VI were 23.5, 16.9, 13.2, and 9.8 months, respectively (Table 1). In historical control, the RPA class distribution was as follows: Class III = 3 (11%); Class IV = 14 (52%); Class V = 8 (30%); Class VI = 2 (7%). The distributions of each RPA class in BNCT group and institutional historical control group are a little bit different. We compare the survival of both groups in low risk RPA (class III and IV) and in high risk RPA (class V and IV) separately. The MST of BNCT group in low risk group was 18.5 months (n=12, 95% CI: 13.7-36.1) and that of historical control was 13.0 months (n=17, 95% CI: 8.6-18.0). There is statistical significance in log-rank test (p=0.028). The MST of BNCT group in high risk group was 12.2 months (n=9, 95% CI: 9.8-undetermined) and that of historical control was 7.4 months (n=10, 95% CI: 2.7-10.3). There is also statistical significance in log-rank test (p=0.0083). Therefore, it can be concluded that BNCT group shows the long survival in comparison with historical control not mainly by the difference of distribution of each RPA class in both groups. Our BNCT results for survival among the newly-diagnosed GB cases were favorable in comparison with those obtained from the

Fig. 4. Cumulative survival data for all newly diagnosed glioblastoma (protocol 1 and 2). External beam X-ray irradiation boost after boron neutron capture therapy (BNCT) (protocol 2, red line) was indicated for later 11 cases. This improve the median survival time as 23.5

months (vs 14.1 months for BNCT only (protocol 1, dotted red line)).

corresponding RTOG - RPA subclasses.

group, all cases were operated on to aim for the maximum tumor removal, the same as for the cases of the BNCT group. These Kaplan–Meier curves were calculated and the Log-rank test was used for statistical analysis. For the 21 patients who received BNCT, survival time was also compared with the corresponding recursive partitioning analysis (RPA) subclasses by the Radiation Therapy Oncology Group (RTOG) (Curran *et al.* 1993), as the international historical control, and European Organisation for Research and Treatment of Cancer - National Cancer Institute of Canada (EORTC-NCIC) trial (Stupp *et al.* 2009).

#### **2.1.2 Results from our past study**

Patients treated with BNCT (n=21) had a median ST of 15.6 months (95% confidence interval (CI): 12.2-23.9) after diagnosis (Fig. 3). Here the date of diagnosis is the initial debulking surgery date. This was significantly longer than the median survival time (MST) for the historical controls at our institute who were treated with surgical removal followed by XRT and chemotherapy (n=27, MST was 10.3 months (95% CI: 7.4-13.2), log-rank test p = 0.0035). The survival time from the date of diagnosis was calculated using the Kaplan-Meier method. The MST of the protocol 2 was 23.5 months (95% CI: 10.2 - undetermined) after diagnosis (n=11), and that of the protocol 1 patients (n=10) was 14.1 months (95% CI: 9.9- 18.5), although the difference was not statistically significant (Fig. 4).

Fig. 3. Cumulative survival data for all newly diagnosed glioblastoma (WHO grade 4, n=21). Blue line is our recent historical control treated by external beam X-ray irradiation. The median survival time (ST) of boron neutron capture therapy (BNCT) group (red line) is 15.6 months. There is statistical significance between both group in Log-rank test (p=0.0035).

group, all cases were operated on to aim for the maximum tumor removal, the same as for the cases of the BNCT group. These Kaplan–Meier curves were calculated and the Log-rank test was used for statistical analysis. For the 21 patients who received BNCT, survival time was also compared with the corresponding recursive partitioning analysis (RPA) subclasses by the Radiation Therapy Oncology Group (RTOG) (Curran *et al.* 1993), as the international historical control, and European Organisation for Research and Treatment of Cancer -

Patients treated with BNCT (n=21) had a median ST of 15.6 months (95% confidence interval (CI): 12.2-23.9) after diagnosis (Fig. 3). Here the date of diagnosis is the initial debulking surgery date. This was significantly longer than the median survival time (MST) for the historical controls at our institute who were treated with surgical removal followed by XRT and chemotherapy (n=27, MST was 10.3 months (95% CI: 7.4-13.2), log-rank test p = 0.0035). The survival time from the date of diagnosis was calculated using the Kaplan-Meier method. The MST of the protocol 2 was 23.5 months (95% CI: 10.2 - undetermined) after diagnosis (n=11), and that of the protocol 1 patients (n=10) was 14.1 months (95% CI: 9.9-

Fig. 3. Cumulative survival data for all newly diagnosed glioblastoma (WHO grade 4, n=21). Blue line is our recent historical control treated by external beam X-ray irradiation. The median survival time (ST) of boron neutron capture therapy (BNCT) group (red line) is 15.6 months. There is statistical significance between both group in Log-rank test

National Cancer Institute of Canada (EORTC-NCIC) trial (Stupp *et al.* 2009).

18.5), although the difference was not statistically significant (Fig. 4).

**2.1.2 Results from our past study** 

(p=0.0035).

The RPA class distribution of 21 patients treated with BNCT at the initial diagnosis was as follows: Class III = 6 (29%); Class IV = 6 (29%); Class V = 8 (38%); Class VI = 1 (5%). The MSTs of the patients in classes III, IV, V, and VI were 23.5, 16.9, 13.2, and 9.8 months, respectively (Table 1). In historical control, the RPA class distribution was as follows: Class III = 3 (11%); Class IV = 14 (52%); Class V = 8 (30%); Class VI = 2 (7%). The distributions of each RPA class in BNCT group and institutional historical control group are a little bit different. We compare the survival of both groups in low risk RPA (class III and IV) and in high risk RPA (class V and IV) separately. The MST of BNCT group in low risk group was 18.5 months (n=12, 95% CI: 13.7-36.1) and that of historical control was 13.0 months (n=17, 95% CI: 8.6-18.0). There is statistical significance in log-rank test (p=0.028). The MST of BNCT group in high risk group was 12.2 months (n=9, 95% CI: 9.8-undetermined) and that of historical control was 7.4 months (n=10, 95% CI: 2.7-10.3). There is also statistical significance in log-rank test (p=0.0083). Therefore, it can be concluded that BNCT group shows the long survival in comparison with historical control not mainly by the difference of distribution of each RPA class in both groups. Our BNCT results for survival among the newly-diagnosed GB cases were favorable in comparison with those obtained from the corresponding RTOG - RPA subclasses.

Fig. 4. Cumulative survival data for all newly diagnosed glioblastoma (protocol 1 and 2). External beam X-ray irradiation boost after boron neutron capture therapy (BNCT) (protocol 2, red line) was indicated for later 11 cases. This improve the median survival time as 23.5 months (vs 14.1 months for BNCT only (protocol 1, dotted red line)).

Clinical Study on Modified Boron Neutron Capture Therapy for Newly Diagnosed Glioblastoma 331

Fig. 5. Dose planning of neutron capture therapy for the tumor (upper) and for the normal brain (lower). (Gy-Eq: gray equivalent, star: air instillation to the tumor removed cavity)

Fig. 6. Serial MRI images (T1-weighted with Gd enhancement) of the patient treated by boron neutron capture therapy combined with fractionated external beam X-ray irradiation. (A: prior to surgical removal and BNCT, B: 12 months, C: 14 months and D: 5 years after BNCT).


\*Median survival time (months). RTOG: radiation therapy oncology group, EORTC: european organisation for research and treatment of cancer, NCIC: National Cancer Institute of Canada Clinical Trials, RPA: recursive partitioning analysis, XRT: X-ray radiation therapy, TMZ: Temozolomide, BNCT: boron neutron capture therapy.

Table 1. Comparison of survival data among RPA class in the RTOG database, EORTC-NCIC (XRT/TMZ) trial, and in our cases treated with BNCT.

#### **2.2 Representative case treated by modified boron neutron capture therapy combination with fractionated X-ray irradiation**

A 63-year old female had a left parietal tumor removed partially in a hospital in November 2004. The histopathological diagnosis was GB. She was introduced to our hospital for BNCT for the remaining lesion. While she waited for the BNCT, rapid growth of the tumor caused aggravations of right hemiparesis, total aphasia and consciousness disturbance, all of which caused us to remove the tumor again in our hospital prior to BNCT. The tumor was gross totally removed, and the patient received BNCT with air instillation in tumor removed cavity, followed by XRT. At the time of BNCT, the workstation calculates two different estimated doses, one is for the tumor and the other is for the normal brain (Fig. 5).

Dose calculation shows estimated gray-equivalent (Gy-Eq) isodose lines from BNCT. This is mainly based on the decay of the irradiated neutron, boron concentration and biological effectiveness of each tissue. The deepest part of the tumor was 7.5 cm from the scalp in this case. The irradiated minimum tumor dose by BNCT was improved by the air-instilation (star in the Fig. 5) methods from 18.9 Gy-Eq (without air) to 26.9 Gy-Eq (with air). The maximum irradiated point of the normal brain was 2 cm from the scalp and was 12.7 Gy-Eq and scalp surface was 9.0 Gy-Eq. An additional 20 Gy was applied by XRT (daily fraction of 2 Gy x 10 days) aimed for the deep part of the mass but included surface of the tumor and the normal brain.

No tumor recurrence on MRI and no neurological deficit were seen for 13 months after BNCT (Fig. 6). BNCT followed by fractionated XRT could control the tumor for 12 months (Fig. 6B). However, 14 months after BNCT, the patient suffered from right hemiparesis and MRI revealed a Gd-enhanced lesion from the surface area of the tumor removed cavity which was irradiated by BNCT with enough dose and also included into the boost XRT (Fig. 6C). Radiation necrosis was suspected as the pathological condition based on L/N ratio in BPA-PET, but we could not neglect the possibility of a recurrence of the GB. We performed a re-craniotomy to remove this lesion. The pathological diagnosis was radiation necrosis; no apparent tumor cells were found by pathologists. After the surgery, the right hemiparesis was improved, while no tumor progression was observed on MRI in the follow-up for more than 5 years (Fig. 6D). The patient received best medical treatment (anti-coagulants and vitamin E) for a year to prevent a further radiation necrosis.

EORTC-NCIC

III 17.9 14.8 18.7 13.4 23.5 IV 11.1 13.3 16.3 12.8 16.9 V 8.9 9.1 10.7 7.4 13.2 VI 4.6 Not reported 3.6 9.8 \*Median survival time (months). RTOG: radiation therapy oncology group, EORTC: european organisation for research and treatment of cancer, NCIC: National Cancer Institute of Canada Clinical Trials, RPA: recursive partitioning analysis, XRT: X-ray radiation therapy, TMZ: Temozolomide, BNCT: boron neutron capture therapy. Table 1. Comparison of survival data among RPA class in the RTOG database, EORTC-

XRT XRT TMZ/XRT XRT BNCT

(XRT/TMZ) BNCT

RTOG RPA class

the normal brain.

RTOG original

NCIC (XRT/TMZ) trial, and in our cases treated with BNCT.

vitamin E) for a year to prevent a further radiation necrosis.

**combination with fractionated X-ray irradiation** 

**2.2 Representative case treated by modified boron neutron capture therapy** 

estimated doses, one is for the tumor and the other is for the normal brain (Fig. 5).

A 63-year old female had a left parietal tumor removed partially in a hospital in November 2004. The histopathological diagnosis was GB. She was introduced to our hospital for BNCT for the remaining lesion. While she waited for the BNCT, rapid growth of the tumor caused aggravations of right hemiparesis, total aphasia and consciousness disturbance, all of which caused us to remove the tumor again in our hospital prior to BNCT. The tumor was gross totally removed, and the patient received BNCT with air instillation in tumor removed cavity, followed by XRT. At the time of BNCT, the workstation calculates two different

Dose calculation shows estimated gray-equivalent (Gy-Eq) isodose lines from BNCT. This is mainly based on the decay of the irradiated neutron, boron concentration and biological effectiveness of each tissue. The deepest part of the tumor was 7.5 cm from the scalp in this case. The irradiated minimum tumor dose by BNCT was improved by the air-instilation (star in the Fig. 5) methods from 18.9 Gy-Eq (without air) to 26.9 Gy-Eq (with air). The maximum irradiated point of the normal brain was 2 cm from the scalp and was 12.7 Gy-Eq and scalp surface was 9.0 Gy-Eq. An additional 20 Gy was applied by XRT (daily fraction of 2 Gy x 10 days) aimed for the deep part of the mass but included surface of the tumor and

No tumor recurrence on MRI and no neurological deficit were seen for 13 months after BNCT (Fig. 6). BNCT followed by fractionated XRT could control the tumor for 12 months (Fig. 6B). However, 14 months after BNCT, the patient suffered from right hemiparesis and MRI revealed a Gd-enhanced lesion from the surface area of the tumor removed cavity which was irradiated by BNCT with enough dose and also included into the boost XRT (Fig. 6C). Radiation necrosis was suspected as the pathological condition based on L/N ratio in BPA-PET, but we could not neglect the possibility of a recurrence of the GB. We performed a re-craniotomy to remove this lesion. The pathological diagnosis was radiation necrosis; no apparent tumor cells were found by pathologists. After the surgery, the right hemiparesis was improved, while no tumor progression was observed on MRI in the follow-up for more than 5 years (Fig. 6D). The patient received best medical treatment (anti-coagulants and

Fig. 5. Dose planning of neutron capture therapy for the tumor (upper) and for the normal brain (lower). (Gy-Eq: gray equivalent, star: air instillation to the tumor removed cavity)

Fig. 6. Serial MRI images (T1-weighted with Gd enhancement) of the patient treated by boron neutron capture therapy combined with fractionated external beam X-ray irradiation. (A: prior to surgical removal and BNCT, B: 12 months, C: 14 months and D: 5 years after BNCT).

Clinical Study on Modified Boron Neutron Capture Therapy for Newly Diagnosed Glioblastoma 333

chemotherapy with TMZ. Prescription dose by BNCT is regulated as not to be more than 13Gy-Eq for normal brain. Additional XRT is given with 3 gradient such as 8, 16, 24Gy from the surface of scalp to the bottom of tumor infiltrated zone (Fig. 7). Chemotherapy with TMZ is applied concomitantly during XRT treatments and adjuvant chemotherapy with the

Fig. 7. Illustrated image of the protocol combined with boron neutron capture therapy (BNCT) and 3 gradient fractionated X-ray irradiation (XRT). Gy: gray, fr: fraction.

18F-BPA-PET study is included for the diagnosis of these pathologies.

Based on our previous clinical study, the Hazard ratio of BNCT vs. XRT was simulated as 0.4, so the total estimated number of the patients who should be included in our new study become 45 totally. Primary end point is overall survival and these patients will be followed up for 2 years after the last patient treatment. The most important point in our protocol is diagnosis and treatment of radiation effects such as swelling, radiation induced edema, transient expansion of the tumor, pseudo- progression / response, and radiation necrosis.

same agent is repeated in outpatient clinic (Fig. 8) (Stupp *et al.* 2005).

#### **2.3 Phase II clinical study of boron neutron capture therapy combined with X-ray radiotherapy / Temozolomide in patients with newly diagnosed glioblastoma multiforme**

Results of our clinical study showed the efficacy of combination therapy with external beam XRT and BNCT. For our future study, we planned the Multi-centric Phase II clinical study named "*Boron Neutron Capture Therapy, Radiation Therapy, and Temozolomide in Treating Patients with Newly Diagnosed Glioblastoma Multiforme*" in Japan (OSAKA-TRIBRAIN0902, NCT00974987). The major differences of our protocol from the other past BNCT studies were simultaneous use of both BSH and BPA, and combination with XRT.
