Boron neutron capture therapy for glioblastoma ... - Springer Link

Journal of Neuro-Oncology 33: 141–152, 1997.  1997 Kluwer Academic Publishers. Printed in the Netherlands.

Boron neutron capture therapy for glioblastoma multiforme using p-boronophenylalanine and epithermal neutrons: Trial design and early clinical results Jeffrey A. Coderre,1 Eric H. Elowitz,2 Manjeet Chadha,3 Richard Bergland,2 Jacek Capala,1 Darrel D. Joel,1 Hungyuan B. Liu,1 Daniel N. Slatkin1 and Arjun D. Chanana1 1 Medical Department Brookhaven National Laboratory, Upton, NY, 11973, U.S.A.; Departments of 2 Neurosurgery and 3 Radiation Oncology, Beth Israel Medical Center, 16th Street at First Avenue, New York, NY, 10003, USA

Key words: boron neutron capture therapy, glioblastoma multiforme, boronophenylalanine, Phase I/II trial, epithermal neutrons Summary A Phase I/II clinical trial of boron neutron capture therapy (BNCT) for glioblastoma multiforme is underway using the amino acid analog p-boronophenylalanine (BPA) and the epithermal neutron beam at the Brookhaven Medical Research Reactor. Biodistribution studies were carried out in 18 patients at the time of craniotomy using an i.v. infusion of BPA, solubilized as a fructose complex (BPA-F). There were no toxic effects related to the BPA-F administration at doses of 130, 170, 210, or 250 mg BPA/kg body weight. The tumor/ blood, brain/blood and scalp/blood boron concentration ratios were approximately 3.5:1, 1:1 and 1.5:1, respectively. Ten patients have received BNCT following 2-hr infusions of 250 mg BPA/kg body weight. The average boron concentration in the blood during the irradiation was 13.0 ± 1.5 µg 10B/g. The prescribed maximum dose to normal brain (1 cm3 volume) was 10.5 photon-equivalent Gy (Gy-Eq). Estimated maximum and minimum doses (mean ± sd, n = 10) to the tumor volume were 52.6 ± 4.9 Gy-Eq (range: 64.4–47.6) and 25.2 ± 4.2 Gy-Eq (range: 32.3–20.0), respectively). The estimated minimum dose to the target volume (tumor + 2 cm margin) was 12.3 ± 2.7 Gy-Eq (range: 16.2–7.8). There were no adverse effects on normal brain. The scalp showed mild erythema, followed by epilation in the 8 cm diameter field. Four patients developed recurrent tumor, apparently in the lower dose (deeper) regions of the target volume, at post-BNCT intervals of 7, 5, 3.5 and 3 months, respectively. The remaining patients have had less than 4 months of post-BNCT follow-up. BNCT, at this starting dose level, appears safe. Plans are underway to begin the dose escalation phase of this protocol.

Introduction Boron neutron capture therapy (BNCT) is a binary treatment modality that can selectively irradiate tumor tissue (for recent reviews see [1, 2]). BNCT uses drugs containing a stable isotope of boron, 10 B, that are capable of preferentially accumulating in the tumor, which is then irradiated with low energy (thermal or epithermal) neutrons. The interaction of 10B

with a thermalized neutron (neutron capture) causes the 10B nucleus to split, releasing an alpha (4He) and a lithium (7Li) particle. These products of the 10 B(n, α)7Li reaction have a combined path length in tissue of approximately 14 µm, or about the diameter of one or two cells, thus restricting most of the resulting ionizing energy to 10B-loaded cells. If the particle track passes through the cell nucleus, the probability is high that the resulting damage will

Please indicate author’s corrections in blue, setting errors in red 121181 NEON ART.NO 731 AS (601) ORD.NO 235601.Z

142 inactivate any subsequent mitosis. The key to effective BNCT is the preferential accumulation of 10B in the tumor relative to the surrounding normal tissues. The amino acid analog p-boronophenylalanine (BPA) has been under investigation at Brookhaven National Laboratory (BNL), and elsewhere, as a boron delivery agent for BNCT of glioblastoma multiforme. Preclinical studies in the 9L rat gliosarcoma brain tumor model at BNL have shown that BPA, delivered orally, is effective in BNCT of the intracerebral 9L rat gliosarcoma [3]. Histologic examination of the brains of long-term BNCT survivors showed scar tissue replacing the tumor with no serious damage to the contiguous normal brain [4]. Intraperitoneal injection of the more soluble BPAfructose complex (BPA-F) produced much higher tumor boron concentrations in the rat intracerebral 9L gliosarcoma than were previously possible [5]. Higher boron concentrations in tumor allowed demonstrably curative radiation doses to be delivered to the tumor while maintaining the dose to the normal brain vascular endothelium below the threshold of tolerance. This resulted in long-term survival of over 90% of BNCT-treated rats [5] without any noticeable neurologic deficit. The relative biological effectiveness (RBE) of the various high-linear energy transfer (LET) components of the total BNCT dose has been measured in both tumor and normal tissues [6, 7], indicating a therapeutic ratio (tumor dose/normal tissue dose) for BNCT of the rat 9L gliosarcoma in excess of 5:1 [5, 8]. An epithermal neutron irradiation facility, with greater tissue-penetration characteristics than those of the thermal neutron beams used in the USA in the 1950s and early 1960s or of the thermal neutrons beams currently in use in Japan, has been designed and installed at the Brookhaven Medical Research Reactor (BMRR) [9]. The response of the normal dog brain to irradiation with epithermal neutrons in the presence of either BPA or BSH [10] is consistent with the biological effectiveness factors determined in the rat [6, 7]. Based on these preclinical results, studies were initiated in patients with glioblastoma multiforme. A series of 17 patients undergoing surgical removal of tumor (glioblastoma or melanoma) received

BPA orally as the free amino acid. Favorable tumor/ blood boron concentration ratios were obtained but the absolute amount of boron in the tumor would have been insufficient for BNCT [11]. More recently a Phase I/II clinical trial of BNCT using intravenously administered BPA-F was begun. Preliminary results have been reported regarding the biodistribution of i.v. BPA-F [12, 13] and BNCT of the first glioblastoma patient treated with BPA-F and epithermal neutrons [13, 14]. This report will serve to provide more BPA-F biodistribution results and to provide clinical observations on the patients treated with BNCT to date.

Methods BPA-F BPA was solubilized for iv infusion at neutral pH by formation of a complex with fructose [15]. Injection solutions of the BPA-fructose complex (BPA-F) were prepared at a concentration of 30 mg BPA/ml (0.14 M) by a modification of published procedures [15, 16]. Briefly, BPA (10B-enriched, L-isomer) was combined with a 10% molar excess of fructose in water (65% of the total volume needed to make a 0.14 M solution). The pH was adjusted to between 9.5 and 10.0 with NaOH for 2–3 minutes and then readjusted to 7.4 with HCl. The volume was then adjusted with water to yield a 0.14 M solution. The solution was passed through a 5000 molecular weight-cutoff filter (Sartorius EASY FLO, Sartorius AG, Go¨ttingen, Germany) to remove endotoxins, and through a 0.22 µm-pore size sterilization filter (Nalge Company, Rochester, NY) into empty, sterile infusion bags. A fresh solution of BPA-F was prepared for each individual patient study and was used within 48 hours. All BPA-F injection solutions were prepared and tested for sterility and pyrogenicity at BNL prior to use. Aliquots of the BPA-F injection solution were incubated in thioglycolate medium and tryptic soy broth at 27 °C or 35 °C and also applied to blood agar plates incubated under aerobic or anaerobic conditions. All cultures were monitored for 14 days. The solution was released for use if all cultures were negative after 24 hours. A


143 Limulus Amebocyte Lysate test kit (Pyrogent, BioWhittaker, Inc., Walkersville, MD) was used to verify that the endotoxin content of the BPA-F solution is below acceptable limits (< 5 EU/kg body weight).

Radiobiological considerations BNCT produces a mixture of high- and low-linear energy transfer (LET) radiations each with different tissue-penetration characteristics and different efficacies in biological systems. In addition to the high-LET products of the 10B(n, α)7Li reaction, the interaction of the neutron beam with the nuclei of elements in tissue produces high-LET dose components. The capture of thermal neutrons by nitrogen in tissue, the 14N(n, p)14C reaction, releases a proton with an energy of 590 keV. Contaminating fast neutrons (those with kinetic energies > 10 keV) in the epithermal neutron beam, through collisions with hydrogen nuclei in tissue, produce recoil protons with similar average energy through the 1H(n, n′)p reaction. In experimental BNCT radiobiology, the uniformly distributed effects of the nitrogen capture proton and the fast neutron recoil proton are generally measured as the combined proton dose. To express the total BNCT dose in a common unit, and to compare BNCT doses with the effects of conventional photon irradiation, each of the high-LET dose components (physical dose in Gy) is multiplied by an experimentally determined factor to correct for different degrees of biological effectiveness. The total effective, photon-equivalent BNCT dose is then expressed as the sum of the RBE-corrected components with a unit named Gy-Eq (Gray-Equivalent). The short ranges of the two high-LET products of the 10B(n, α)7Li reaction make the microdistribution of the boron relative to target cell nuclei of particular importance. Thus, there is a boron localization factor to be considered in determining the biological effectiveness of the 10 B(n, α)7Li reaction. The dependence of the biological effect on variations in the microdistribution of different boron compounds, or the same boron compound in different tissues, makes the term RBE inappropriate in describing the biological effective-

ness of the 10B(n, α)7Li reaction in BNCT. The term ‘CBE factor’ has, therefore been introduced and is defined as the product of the true, geometry-independent RBE for these high-LET particles multiplied by a boron localization factor, which, will be different for each boron compound or for a given compound in different tissues. Biological effectiveness factors for all the BNCT dose components have been experimentally determined, mostly in animal model systems, and serve as a guide in the estimation of photon-equivalent dose in human BNCT trials. Table 1 lists the biological effectiveness factors (RBE of the beam protons, CBE factor for the 10B(n, α)7Li reaction) that have been used in the clinical BNCT dosimetry described in this report.

Definition of brain regions The tumor volume is defined as the contrast-enhanced volume evident in CT or MRI images. The target volume is defined as the tumor volume plus a 2 cm-wide margin around the tumor volume: a zone likely to contain infiltrating glioblastoma tumor cells. The thermal neutron fluence from the epithermal beam reaches a maximum at approximately 2.5 cm deep from the scalp surface (1 cm depth in the brain). The absolute peak dose is defined as the maximum Gy-Eq dose to a 1 cm3 volume (a single ‘voxel’ in the treatment planning software) at the depth of the maximum thermal neutron fluence.

Treatment planning The BNCT treatment planning software was developed at the Idaho National Engineering Laboratory [17]. BNCT treatment planning is quite different from that for conventional radiotherapy [17, 18]. In BNCT, to achieve optimal beam orientation, it is necessary to maneuver the patient in the path of a fixed beam. During the postoperative treatment planning head scan, tatooed triangulation points are identified by fiducial markers for CT or MRI scans. The spatial distribution information for tumor and normal tissues establish a baseline coor-


144 dinate system which is used by the BNCT treatment planning software to reconstruct a 3-D head geometry and identify the beam entry point coordinates as well as preferred orientation of the head relative to the beam axis [17]. 10B concentrations in tumor and normal tissues are entered into the treatment planning program, along with RBE and CBE factors derived from in vivo animal experiments. The total BNCT radiation dose fields are displayed as photon-equivalent (Gy-Eq) isodose contours superimposed on the CT or MRI images. The absolute peak dose is asssumed to represent the 100% point (1 cm3 volume) in an isodose contour plot. Dose/volume histograms for the brain and the tumor volume are calculated so that the beam location and treatment time can be determined with consideration given to both tumor dose and sparing of normal tissues.

BNCT clinical trial: study design The objectives of the Phase I/II clinical trial are: 1) To determine a safe starting dose for BNCT using epithermal neutrons. 2) To evaluate adverse effects of BNCT at this starting dose, if any. 3) To evaluate the therapeutic effectiveness of this starting BNCT dose in patients with glioblastoma multiforme.

The clinical trial consists of two parts: 1) a BPA biodistribution study at the time of craniotomy to confirm preferential accumulation of BPA in the tumor, and 2) BNCT irradiation 3–4 weeks after surgery. Informed consent was obtained prior to the BPA biodistribution study and again at the time of BNCT. Major eligibility criteria included: 1) Radiologically documented supratentorial lesion for which the most likely clinical diagnosis is glioblastoma multiforme. The tumor must be unilateral and unifocal. The deepest part of the contrast-enhanced tumor margin on radiographic images should be ≤ 6.0 cm from the skin surface. Diagnosis of glioblastoma must be confirmed histopathologically after surgery to proceed to BNCT. 2) No prior radiation therapy to the brain, no prior immunotherapy and/or systemic or intrathecal chemotherapy. 3) Age greater than 18 years. 4) No history of phenylketonuria. 5) No pacemaker, no metal prostheses, or metal implant of any kind in the head or neck region. (Dental work was excepted after calculations revealed negligible added dose from the induced radioactivity.) 6) Karnofsky Performance Status (KPS) [19] must be ≥ 70. The starting dose to the normal brain was chosen with primary consideration given to safety and de-

Table 1. Biological effectiveness factors used in calculating photon equivalent doses during BNCT at the Brookhaven Medical Research Reactor Dose component

Biological effectiveness factor: (CBE or RBE)

10

B(n, α)7Li reaction (BPA-fructose)

tumor (9L rat gliosarcoma)a = 3.8 CNS (rat spinal cord)b = 1.35 human skin (moist desquamation)c ≈ 2.5

beam protons [14N(n,p)14C and 1H(n,n′)p]

tumor (9L rat gliosarcoma)a = 3.2 dog braind = 3.3 dog skine = 3.0

a

Determined from the results of in vivo/in vitro clonogenic assays of the rat 9L gliosarcoma [6]. Determined from dose-response studies of the irradiated rat spinal cord with an endpoint of limb paralysis within 7 months [7]. c Determined from thermal neutron-based BNCT of cutaneous melanoma in patients [20]. d Determined in normal dogs in BMRR epithermal neutron beam using MRI to detect damage to the blood brain barrier [10]. e Determined in dog skin using the BMRR epithermal neutron beam [27]. b


145 tection of possible adverse effects. The administered dose of BPA-F was 250 mg BPA/kg delivered as a 2 hour i.v. infusion. The start of the irradiation was approximately one hour after the end of the BPA-F infusion. The duration of the irradiation was based on the calculated average boron concentration in the blood derived from the measured blood boron concentration at the beginning of the BNCT irradiation, and at the irradiation midpoint. The irradiation time was adjusted to deliver the prescribed normal brain endothelium dose of 10.5 GyEq to the peak-dose volume. BNCT was delivered in a single fraction for about 40–50 min (with a pause at the approximate mid-point for blood sampling) using a single field. The peak and average radiation doses to the normal brain were determined from the measured boron concentration in the blood during the irradiation. The following conditions were used in this initial protocol: 1) The prescribed absolute peak dose (to a 1 cm3 volume) will be 10.5 Gy-Eq. 2) The peak dose rate will be kept below 27 cGyEq/min so as not to exceed that used by others for photon irradiation of the whole brain. 3) The average dose to the whole brain volume will not exceed 7.5 Gy-Eq. 4) The maximum dose to the scalp will not exceed 20 Gy-Eq. 5) The total radiation dose to the deepest part of the contrast-enhancing region of tumor should be ≥ 20 Gy-Eq. Patient follow-up includes regular physical, clinical and neurological examinations as well as brain scans at 2 days post-BNCT, at 3, 5, and 12 weeks and quarterly thereafter. Steroid dependency and quality of life are also assessed.

ies: 130 mg BPA/kg (n = 5); 170 mg BPA/kg (n = 6); 210 mg BPA/kg (n = 3); 250 mg BPA/kg (n = 3). One patient received only 100 mg BPA/kg for a biodistribution study, but this was inadvertent. All patients received 250 mg BPA/kg for BNCT. All of the BPA-F infusions were 2 hours in duration, except for two of the biodistribution studies carried out with 170 mg BPA/kg, for which each infusion lasted for 1 hour.

BPA biodistribution in blood The boron concentration in the blood increases in a quasi-linear fashion during the course of the i.v. infusion. As soon as the infusion is stopped, the boron concentration in blood drops abruptly. Figure 1 shows the boron concentration in blood as a function of time for 11 patients who received 250 mg BPA/kg. The data from three patients who received 250 mg/kg at the time of biodistribution and from eight patients who received 250 mg BPA/kg prior to BNCT are combined in Figure 1. Data from 2 additional infusions at the time of BNCT were not included in the graph because of problems (with either the infusion or in drawing blood samples)

Results To date, 18 patients have been entered into the iv BPA-F phase of the BPA biodistribution studies, 8 at the first debulking operation and 10 at the second debulking surgery. Ten of these patients went on to receive BNCT (9 after 2 craniotomies, 1 after the first craniotomy). The administered dose of BPA has been escalated during the biodistribution stud-

Figure 1. Boron concentrations in blood versus time during and after a 2-hour infusion (0–2 hrs) of BPA-F that delivered 250 mg BPA/kg body weight. Points represent the mean ± sd from 11 patients.


146 that did not affect the BNCT procedure but resulted in blood boron measurements at time points that did not match the rest of the patients. The maximum boron concentration obtained in the blood at the end of a 250 mg BPA/kg infusion was 22.1 ± 3.4 µg 10B/g (mean ± sd). The maximum boron concentrations obtained in the blood samples from two patients at the end of a 1-hour infusion that delivered 170 mg BPA/kg were 25.0 and 21.3 µg 10B/g, respectively. The maximum boron concentration obtained in the blood at the end of a 2 hour infusion is approximately a linear function of the administered dose: BPA-F infusions of 130 mg BPA/kg (n = 5), 170 mg BPA/kg (n = 4) and 210 mg BPA/kg (n = 3) produced maximum blood boron concentrations of 13.1 ± 1.9, 14.2 ± 2.2, and 17.3 ± 3.7 µg 10B/g, respectively [12, 13]. The data in Figure 1 were fit with a single exponential function of the form y = A(1-e-at) during the infusion (0–2 hrs) and with a double exponential function of the form y = [B(e-bt) + (C(e-ct)] after the end of the infusion (t > 2 hrs). The clearance of boron from the blood, after the end of the infusion, follows biexponential kinetics related to redistribution and renal excretion of unmetabolized BPA. The initial clearance phase had a decay half-time of 0.54 hours; the second phase, a halftime of 6.2 hours (Figure 1).

BPA biodistribution in tumor Over 120 individual tumor samples from 17 patients (one patient’s tumor was later identified as a ganglioglioma and was not included) in this report have been analyzed for gross boron content by direct current plasma atomic emission spectroscopy. Boron concentrations measured in tumor samples varied among patients and even within multiple samples from an individual patient [12, 13]. Tumor samples were taken 0.3 to 2.5 hours after the end of the infusion and their average boron concentration (normalized arithmetically to an infusion dose of 250 mg BPA/kg) ranged from 16 to 62 µg 10B/g. Histologic examination of tumor sections adjacent to the samples used for boron analysis showed considerable variation in the proportion of necrotic tissue. An analysis that provides an average boron concen-

Figure 2. (A) Scalp/blood boron concentration ratios as a function of time after the end of the infusion in 16 patients who received 2-hour infusions (open circles) and 2 patients who received 1-hour infusions of 170 mg BPA/kg (filled circles). (B) Brain/blood boron concentration ratios as a function of time after the end of the BPA-F infusion in 11 patients who received 2-hour infusions of BPA-F (open circles) and from 1 patient who received a 1-hour infusion of 170 mg BPA/kg (filled circle). In both Figure 2A and 2B, all 2-hour infusion data (µg 10B/g) obtained from patients who received doses lower than 250 mg BPA/kg have been normalized to an injected dose of 250 mg BPA/kg.

tration value for a heterogeneous sample underestimates the amount of boron in viable, highly cellular areas of a glioblastoma. Quantification of the cellularity in histologic sections adjacent to the samples of tumor used for


147 gross boron analysis recently showed a linear relationship between cellularity and boron concentration (Coderre et al., manuscript submitted for publication). The analysis indicates that, in human glioblastoma free from microscopic necrosis, the boron concentration at the time of tumor sampling (or BNCT) is about 3–4 times higher than the 10B concentration present in the blood.

BPA biodistribution in scalp and normal brain Small (< 100 mg) full-thickness biopsies of scalp are obtained during the debulking surgery. Figure 2A shows the boron concentration in scalp samples plotted as the ratio to the boron concentration in the blood as a function of the time after the end of the infusion. Data obtained with 2-hour infusions at 130, 170 and 210 mg BPA/kg have been extrapolated to 250 mg BPA/kg. The fitted curve shown in Figure 1 was used for the blood value in this analysis (open circles, Figure 2A). The scalp/blood boron concentration ratios range from 1 to 2; for treatment planning, an average scalp/blood boron concentration ratio of 1.5:1 was used. These results are consistent with data from Japan using 170 mg BPA/kg for thermal neutron-mediated BNCT of malignant melanoma [20], which show skin/blood boron concentration ratios in the 1.3:1 to 1.5:1 range. The concentration of boron in the normal brain has been usually equal to or less than that in the blood. Figure 2B shows data for samples of normal brain tissue required to be removed during the debulking operations in 11 patients. Data obtained at 130, 170 and 210 mg BPA/kg were extrapolated to 250 mg BPA/ kg for inclusion in Figure 2B. Data from two patients who received a 1-hour infusion of 170 mg BPA/kg are included (without normalization) as solid circles in Figure 2A (scalp) and in Figure 2B (brain).

Dosimetry for brain The thermal neutron fluence from the epithermal beam reaches a maximum at a depth of approximately 1 cm in the brain, or about 2.5 cm deep to the

Figure 3. Iso-thermal neutron fluence contours in a lucite cube phantom. The collimator (8 cm diameter aperture) is the same as is used in the patient irradiations. The contours shown are relative values expressed in percent.

scalp surface. Figure 3 shows the thermal neutron isofluence contours in the cubic lucite phantom used for epithermal beam dosimetry. The collimator used for patient irradiations in the epithermal beam is 16 cm in diameter facing the reactor port, tapering to 8 cm in diameter at the patient’s scalp. The following calculations of doses to brain and tumor are based on one case example, presented here for illustrative purposes. The desired (prescribed) average 10B concentration in the blood during the irradiation was 15 µg 10B/g. To reach the prescribed dose of 10.5 Gy-Eq to the peak dose volume (1 cm3 volume at a depth of 2.5 cm from the scalp surface) with 15 µg 10B/g in the blood required a peak thermal neutron fluence of 3.3 × 1012 n/cm2. This corresponds to a total irradiation time of 88 MW-min (or 44 min at 2 MW BMRR power). The BNCT dose components in the absolute peak physical dose (1 cm3 voxel) to the normal brain totaled 7.9 Gy, which was comprised of 3.47 Gy from the 10B(n, α)7Li reaction (assuming 15 µg 10B/g in the blood), 3.66 Gy gamma, 0.33 Gy fast neutrons, and 0.40 Gy from the 14N(n, p)14C reaction. Using the RBE and CBE factor values listed in Table 1, the peak physical dose of 7.9 Gy corresponds to the 10.5 Gy-Eq prescribed maximum brain dose. For the 10 patients who have received BNCT under this protocol, the average measured 10B concentration in the blood during the irradiation was 13.0 ± 1.5 µg 10B/g (range: 11.2–15.4). As an example, Figure 4A shows a mag-


148

Figure 5. Dose-volume relationships for the tumor, the target volume (tumor volume + 2 cm margin), and the normal brain assuming 15 µg 10B/g in the blood and 45 µg 10B/g in the tumor. The values in parentheses represent the volume of each region.

Figure 4. (A) MRI image one day prior to craniotomy from one of the patients treated under the BNCT Phase I/II protocol. (B) Schematic illustration of the treatment plan prepared for this patient based on an MRI image obtained one week prior to BNCT. The isodose contours for brain and tumor are expressed in GyEq units and assumed 15 µg 10B/g in the blood and 45 µg 10B/g in the tumor.

protocol. Figure 4B shows the isodose contours (Gy-Eq) for this case example assuming 15 µg 10B/g in blood and 45 µg 10B/g in tumor. The average dose to the entire brain volume was 2.8 Gy-Eq (ipsilateral hemisphere, 4.4 Gy-Eq; contralateral hemisphere, 1.2 Gy-Eq). Figure 5 shows the dose-volume histogram for normal brain corresponding to the example shown in Figure 4B. The peak dose to normal brain was brought to 10.5 Gy-Eq in all patients. Accordingly, for all patients, the dose-volume histograms for normal brain were nearly identical to the example shown in Figure 5. Examination of the normal brain isodose contours on MRI images at different levels in the brain allowed an estimation of the maximum dose to any part of other vital structures. For this case example, these doses (expressed as a percentage of the absolute peak dose, 10.5 Gy-Eq) were: ipsilateral basal ganglia, 65%; hypothalamus, 50%; cerebral midline, 25%; optic chiasm, ≤ 20%; retina, ≤ 10%.

Dosimetry for tumor netic resonance image taken one day prior to craniotomy from one of the patients treated under this

The estimation of tumor dose depends on the value


149 assumed for the tumor/blood boron concentration ratio. Correlation of the measured boron concentrations with the degree of cellularity in histological sections taken from the same tumor samples indicate that non-necrotic, viable glioblastoma accumulates about 3–4 times more BPA than is present in the blood (J.A. Coderre et al., manuscript submitted for publication). For calculation of the dose to tumor, a tumor/blood 10B concentration ratio of 3:1 was assumed, which may be conservative. The estimated dose to tumor is also dependent on the size, the shape and, particularly, the depth of individual tumors. For the example shown in Figure 4B, assuming 45 µg 10B/g in tumor and 88 MW-min irradiation, the maximum dose to tumor was 47.2 GyEq. The minimum dose to the tumor volume was ≈ 50% of the maximum dose, 24.2 Gy-Eq. The significant therapeutic ratio is apparent and is due to the higher 10B concentrations present in the tumor than in the normal tissues and to the higher CBE factor for tumor than for normal brain (see Table 1). Figure 5 shows the dose-volume histogram for the tumor volume and the target volume (tumor plus a 2-cm margin). The minimum dose to the target volume for this case example is 10.0 Gy-Eq. This is an estimate of the minimum dose to individual tumor cells or small clusters of tumor cells infiltrating the normal brain, under the assumption that these cells accumulate 10B to the same extent as did the main tumor mass. For the 10 patients treated under this protocol, the average doses (mean ± sd) to tumor and target volume are as follows: Maximum tumor dose: 52.6 ± 4.9 Gy-Eq (range: 64.4–47.6). Minimum tumor dose: 25.2 ± 4.2 Gy-Eq (range: 32.3–20.0). Minimum target volume dose: 12.3 ± 2.7 Gy-Eq (range: 16.2–7.8).

Safety The BNCT procedure outlined in this protocol appears to be safe. There have been no adverse effects from any of the two-hour BPA-F infusions at doses up to and including 250 mg BPA/kg, nor were there any adverse effects noted in the two cases studied

using 170 mg BPA/kg infused over 1 hour. There have been no unexpected acute neurological effects noted to date following BNCT. Four patients have been followed for more than 4 months with no neurological effects observed; several more patients have now been observed for approximately three months post-BNCT. In a number of patients, the zone of maximal thermal neutron fluence occurred in the tumor, or at least in a region of ‘normal brain’ already traversed by the neurosurgeon. Thus, in some cases, the highest dose delivered to normal brain unaffected by tumor or surgery may have been only 80–90% of the prescribed 10.5 Gy-Eq. The estimated radiation dose to the scalp was 12– 13 Gy-Eq. The scalp showed mild erythema in a few patients at approximately two weeks post BNCT. Epilation was observed in all cases within the 8 cm diameter treatment field. There was no impairment of healing of the scalp incision in the two patients who have undergone a repeat craniotomy for glioblastoma recurrence 9 and 6 months post-BNCT.

Effectiveness The BNCT clinical trial is only recently underway and statistically valid clinical results regarding the effectiveness of this therapy for glioblastoma are not yet available. Of the 10 patients treated, one had spinal cord metastases (that became clinically evident only after entry into this protocol) and died of diffuse carcinomatosis of the craniospinal axis two months after BNCT. A second patient died of uncontrolled local recurrence 11 months after BNCT, of which the first 7 months were without serious symptoms of disease. Of the remaining 8 patients, three showed evidence of recurrent disease at postBNCT intervals of 5, 3.5 and 3 months, respectively. Two of these have undergone a repeat craniotomy. In the other, the evidence is radiographic only. Retrospective analysis of the estimated dose to tumor at the site of recurrence is underway. In two of these patients the preliminary estimates of the tumor doses at the site of recurrence are 8.8–13.3 Gy-Eq and 7.5–11.9 Gy-Eq, respectively. A trend is emerging such that the time from BNCT to evidence of tumor recurrence may correlate with the minimum dose to


150 the target volume. Further observations will be required to show whether this trend is statistically significant.

Discussion The treatment of glioblastoma multiforme has been confounded by the nature of the tumor. These high grade tumors are rapidly progressive, relatively radiation resistant, display variable cell metabolism, and persistently infiltrate into surrounding normal brain tissue. The majority of patients die of local recurrence; the patient who died with significant spinal cord metastases was exceptional. Numerous clinical trials have combined surgery with radiation therapy and various systemic therapies in the treatment of glioblastoma. While brachytherapy may improve survival in selected patients, these approaches have, in general, failed to significantly alter the clinical course of this disease [21]. Focused radiotherapy is capable of destroying glioblastoma in situ. Both fast neutron therapy and stereotactic radiosurgery have been reported to sterilize tumor regions at the center of a treatment volume but normal tissue complications and local recurrence have limited the effectiveness of both of these modalities. We have analyzed the radiation doses from other kinds of radiotherapy that resulted in local tumor control to plan dosimetry of BNCT for glioblastoma multiforme. Stereotactic radiosurgery given as a 15–35 Gy (mean 19.7 Gy) boost to the tumor volume following 54–60 Gy of whole brain fractionated therapy provided no improvement in overall survival [22]. The minimum tumor dose ranged from 10–20 Gy (mean 11.7 Gy) in that study. Of 31 patients reported, there were no recurrences within the central treatment volume, 19 recurred within 2 cm of the treatment volume, and 4 recurred at distances > 2 cm from the treatment volume. Fast neutron therapy has been reported to be more effective than fractionated photon therapy in destroying tumor within the treatment volume, but local recurrence and normal tissue toxicity have limited the effectiveness [23, 24]. For example, Catterall and co-workers [24] have described a pros-

pective, controlled trial of fast neutrons (13.0 or 15.6 neutron Gy in 12 fractions over 4 weeks) compared with photons (50.0 or 55.0 Gy over 5–6 weeks). Neutron therapy did not improve survival, but did result in a greater antitumor effect. Patients in the neutron group usually died of lethal brain necrosis with no residual disease, whereas patients in the photon group usually died of locally recurrent disease [24]. Accelerated neutron therapy (16–18 neutron Gy in 2, 3, or 6 weeks) has also been reported to produce local control of tumor, but fatal post-irradiation gliosis [25]. Laramore has reviewed the side-effects in the normal brain of fast neutron therapy (and of other high-LET radiotherapies) for the treatment of gliomas and other brain tumors [26]. The results of a number of studies were similar: local control is possible but brain damage allows no improvement in quality of life or in overall survival. The selective delivery of radiation dose to the tumor during BNCT is due primarily to the biodistribution of the boron compound, not to the incident beam of neutrons. This provides a potential advantage of BNCT over other forms of radiation therapy. BNCT, in theory, can involve whole-brain irradiation and still yield a substantial therapeutic ratio. We have demonstrated high tumor/normal tissue boron concentration ratios in surgical samples from glioblastoma patients. If individual tumor cells, or small clusters of tumor cells infiltrating the normal brain, preferentially accumulate the boron delivery agent to the same degree as has been shown for surgical tumor samples, they will receive significantly higher dose than the immediately adjacent normal brain. Comparison of tumor doses reported for stereotactic radiosurgery and fast neutron therapy that are associated with local tumor control or peripheral recurrence with the BNCT doses described in this report has already provided some initial insight into the interpretation of BNCT results and in the planning of future BNCT protocols. Stereotactic radiosurgery delivering a 15–35 Gy (19.7 Gy mean) boost after 54–60 Gy of conventionally fractionated photon therapy was locally effective in the central portion of the treatment volume. Doses of 13– 15.6 neutron Gy, or ≈ 39–51 photon-equivalent Gy assuming a neutron RBE of between 3 and 3.3 [26],


151 produced the same effect. The maximum tumor doses in the BNCT clinical trial, 52.6 ± 4.9 Gy-Eq (range: 64.4–47.6), are in the same range as those locally effective doses. The minimum tumor dose of 10–20 Gy (mean 11.7 Gy) in the stereotactic radiosurgery report was associated with tumor recurrence within 2 cm of the tumor margin in the majority of patients [22]. The minimum tumor dose in the BNCT trial reported here was 25.2 ± 4.2 Gy-Eq (range: 32.3–20.0). Perhaps the most important estimate of the minimum dose to tumor cells is the minimum target volume dose, the calculated dose to tumor cells invading the brain 2 cm deep to the main tumor mass. The use of a single field in this initial series of patients resulted in a considerable dose inhomogeneity as a function of depth (Figures 4B, 5). The minimum target volume dose in the 10 BNCT patients reported here was only 12.3 ± 2.7 Gy-Eq (range: 16.2–7.8). The initial analysis of the dose in the area of radiographically documented tumor recurrence for two BNCT patients was in the range of 7.5–13.3 Gy-Eq. The results of this initial trial, therefore suggest that these doses were insufficient for complete tumor control, and are in concordance with the stereotactic radiosurgery literature. The BNCT doses used in the protocol described here are the first level in a dose escalation protocol. The results indicate that the doses delivered to the scalp and the normal brain are safe and, perhaps, conservative. Plans are underway to increase the BNCT dose to the tumor by: 1) changing the beam collimation to increase the neutron penetration characteristics; 2) increasing the peak dose delivered to the normal brain; 3) using two treatment fields instead of one for deeper tumors; and 4) after further BPA-F biodistribution studies, increasing the administered dose of BPA-F.

Acknowledgements P. Micca, M. Nawrocky, M. Makar, A. Lomonte and C. Fisher provided excellent technical support in the preparation and testing of BPA-F and in boron analysis. G. Morris and J. Hopewell provided constructive criticism and stimulating discussions during this study. We are particularly grateful to the

staff of the neurosurgical operating room and the pathology laboratory at Beth Israel Medical Center for assistance during the biodistribution studies. The staff of the BNL Safety and Environmental Protection and Reactor Divisions contributed expertly and skillfully to all phases of preclinical and clinical BNCT studies. The Pharmaceutical Resources Branch of the National Cancer Institute provided the BPA for most of these studies. This work was supported by the Office of Health and Environmental Research of the U.S. Department of Energy under Contract DE-AC02-76CH00016.

References 1. Slatkin DN: A history of boron neutron capture therapy of brain tumours – Postulation of a brain radiation dose tolerance limit. Brain 114: 1609–1629, 1991 2. Barth RF, Soloway AH, Fairchild RG, Brugger RM: Boron neutron capture therapy for cancer. Realities and prospects. Cancer 70: 2995–3007, 1992 3. Coderre JA, Joel DD, Micca PL, Nawrocky MM, Slatkin DN: Control of intracerebral gliosarcomas in rats by boron neutron capture therapy with p-boronophenylalanine. Radiat Res 129: 290–296, 1992 4. Coderre J, Rubin P, Freedman A, Hansen J, Wooding TS, Joel D, Gash D: Selective ablation of rat brain tumors by boron neutron capture therapy. Int J Radiat Oncol Biol Phys 28: 1067–1077, 1994 5. Coderre JA, Button TM, Micca PL, Fisher CD, Nawrocky MM, Liu HB: Neutron capture therapy of the 9L rat gliosarcoma using the p-boronophenylalanine-fructose complex. Int J Radiat Oncol Biol Phys 30: 643–652, 1994 6. Coderre JA, Makar MS, Micca PL, Nawrocky MM, Liu HB, Joel DD, Slatkin DN, Amols HI: Derivations of relative biological effectiveness for the high-LET radiations produced during boron neutron capture irradiations of the 9L rat gliosarcoma in vitro and in vivo. Int J Radiat Oncol Biol Phys 271129, 1993 7. Morris GM, Coderre JA, Hopewell JW, Micca PL, Nawrocky MM, Liu HB, Bywaters A: Response of the central nervous system to boron neutron capture irradiation: Evaluation using rat spinal cord model. Radiother Oncol 32: 249–255, 1994 8. Coderre JA, Morris GM, Micca PL, Nawrocky MM, Fisher CD, Bywaters A, Hopewell JW: The therapeutic ratio in BNCT: Assessment using the rat 9L gliosarcoma brain tumor and spinal cord models. In: Mishima Y (ed) Neutron Capture Therapy for Cancer. Plenum Press, New York, 1996, pp 757–762 9. Liu HB, Brugger RM, Greenberg DD, Rorer DC, Hu JP, Hauptman HM: Enhancement of the epithermal neutron


152

10.

11.

12.

13.

14.

15.

16.

17.

18.

beam used for boron neutron capture therapy. Int J Radiat Oncol Biol Phys 28: 1149–1156, 1994 Husikamp R, Gavin PR, Coderre JA, Phillip KHI, Wheeler FJ: Brain tolerance in dogs to boron neutron capture therapy with borocaptate sodium (BSH) or boronophenylalanine (BPA). In: Mishima Y (ed) Neutron Capture Therapy for Cancer. Plenum Press, New York, 1996, pp 591–598 Coderre JA: A Phase 1 biodistribution study of p-boronophenylalanine. In: Moss R, Gabel D (eds) Boron Neutron Capture Therapy: Towards Clinical Trials of Glioma With BNCT. Plenum Press, New York, 1992, pp 111–121 Bergland R, Elowitz E, Coderre JA, Joel D, Chadha M: A Phase I trial of intravenous boronophenylalanine-fructose complex in patients with glioblastoma multiforme. In: Mishima Y (ed) Neutron Capture Therapy for Cancer. Plenum Press, New York, 1996, pp 739–746 Coderre JA: Boron neutron capture therapy. In: Leibel S, Phillips T (eds) Textbook of Radiation Oncology. W.B. Saunders Company, Philadelphia, in press Coderre JA, Bergland R, Chadha M, Chanana A, Elowitz E, Joel DD, Liu HB, Slatkin DN, Wielopolski L: BNCT of glioblastoma multiforme using BPA-fructose and epithermal neutrons. In: Mishima Y (ed) Neutron Capture Therapy for Cancer. Plenum Press, New York, 1996, pp 553–561 Yoshino K, Suzuki A, Mori Y, Kanihana H, Honda C, Mishima Y, Kobayashi T, Kanda K: Improvement of solubility of p-boronophenylalanine by complex formation with monosaccharides. Strahlenther Oncol 165: 127–129, 1989 LaHann TR, Lu DR, Daniell G, Kraft SL, Gavin PR, Bauer WF: Bioavailability of intravenous formulations of p-boronophenylalanine in dog and rat. In: Soloway AH, Barth RF, Carpenter DE (eds) Advances in Neutron Capture Therapy. Plenum Press, New York, 1993, pp 585–589 Nigg DW: Methods for radiation dose distribution analysis and treatment planning in boron neutron capture therapy. Int J Radiat Oncol Biol Phys 28: 1121–1134, 1994 Wallace SA, Mathur JN, Allen BJ: Treatment planning figures of merit in thermal and epithermal boron neutron capture therapy of brain tumors. Phys Med Biol 39: 897–906, 1994

19. Karnofsky DA, Burchenal JH: The clinical evaluation of chemotherapeutic agents in cancer. In: McLeod CM (ed) Evaluation of Chemotherapeutic Agents. Columbia University Press, New York, 1949, pp 191–205 20. Fukuda H, Hiratsuka J, Honda C, Kobayashi T, Yoshino K, Karashima H, Takahashi J, Abe Y, Kanda K, Ichihashi M, Mishima Y: Boron neutron capture therapy of malignant melanoma using 10B-paraboronophenylalanine with special reference to evaluation of radiation dose and damage to the skin. Radiat Res 138: 435–442, 1994 21. Leibel SA, Scott CB, Loeffler JS: Contemporary approaches to the treatment of malignant gliomas with radiation therapy. Semin Oncol 21: 198–219, 1994 22. Masciopinto JE, Levin AB, Mehata MP, Rhode BS: Stereotactic radiosurgery for glioblastoma: a final report of 31 patients. J Neurosurg 82: 530–535, 1995 23. Laramore GE, Griffin TW, Gerdes AJ, Parker RG: Fast neutron and mixed beam (neutron/photon) teletherapy for grades III and IV astrocytomas. Cancer 42: 96–103, 1978 24. Catterall M, Bloom HJG, Ash DV, Walsh L, Richardson A, Uttley D, Gowing NFC, Lewis P, Chaucer B: Fast neutrons compared with megavoltage x-rays in the treatment of patients with supratentorial glioblastoma: A controlled pilot study. Int J Radiat Oncol Biol Phys 6: 261–266, 1980 25. Saroja KR, Mansell J, Hendrickson FR, Cohen L, Lennox A: Failure of accelerated neutron therapy to control high grade astrocytomas. Int J Radiat Oncol Biol Phys 17: 1295– 1297, 1989 26. Laramore GE: Injury to the central nervous system after high LET radiation. In: Gutin PH, Leibel SA, Sheline GE (eds) Radiation Injury to the Nervous System. Raven Press, Ltd., New York, 1991, pp 341–360 27. Gavin PR, Wheeler FJ, Huiskamp R, Siefert A, Kraft S, DeHaan C: Large animal studies of normal tissue tolerance using an epithermal neutron beam and borocaptate sodium. In: Gabel D, Moss R (eds) Towards Clinical Trials of Glioma. Plenum Press, New York, 1992, pp 197–209 Address for offprints: J.A. Coderre, Medical Department, Brookhaven National Laboratory, Upton, NY 11973-5000, USA
