Hypo-Segmented Heterogeneous Boron Distribution Application in Boron Neutron Capture Therapy

doi:10.21203/rs.3.rs-4892063/v1

Heterogeneous boron concentration distribution is the primary determinant that must be considered in boron neutron capture therapy (BNCT) dose calculations. However, taking into account the slip in BNCT dose calculation and assessment come from fusion of different image modalities and the partial volume effect of image voxel differences, it is necessary to reduce these errors when building a highly realistic computational model in order to provide high-confidence dose results. This article uses hypo-segmented heterogeneous boron distribution dose calculation to apply to brain tumor and head and neck cancer cases to evaluate dose cold spots in tumor and hot spots in organs at risk. Using hyper-segmented boron distribution dose results as a reference benchmark to compare with hypo-segmentation results, the mean tumor dose difference between the two is within 1.6%, and the D₈₀ difference is within 2.3%. Significant differences in maximum and minimum doses to the region of interest are expected between hyper- and hypo-segmentation methods, but the dose-volume histogram for both are close to each other. Then, post-BNCT follow-up images were compared with dose maps of hyper- and hypo-segmented boron distribution. The dose distribution echoes clinical response, indicating that the hypo-segmented boron distribution method has evaluation properties.

Physical sciences/Physics/Applied physics

Physical sciences/Physics/Nuclear physics

Boron-10 (¹⁰B) isotope is exercised to synthesize boron compounds that are stable, non-toxic, and targeted to specific tissue cells. The bulk ordinarily employed one at present is ¹⁰B-p-Boronophenylalanine (BPA)[1] which can be utilized for indications such as head and neck cancer[2], brain tumors[3], melanoma[4], osteosarcoma[5], and so on. Continuous BPA intravenous injection allows tissue cells to steadily accumulate 10B and undergo neutron irradiation. The leading dose of boron neutron capture therapy (BNCT) is boron dose, produced by ¹⁰B capturing neutrons and releasing alpha particle and lithium-7 isotope. The greater boron uptake ratio between lesion and normal tissue, the higher tumor control probability (TCP), and the lower normal tissue complication probability (NTCP). Therefore, it is a prominent course of action to achieve information on boron distribution to optimize a treatment plan for the irradiation goal.

Since the BPA distribution in the human body cannot be obtained through in vitro detection, BPA is labeled with fluorine-18 (¹⁸F) isotope and then positron emission tomography (PET) scan is operated to secure the relative boron ratio between tissues. Previous studies have validated that BPA and FBPA have the consistent pharmacokinetics[6], and FBPA PET has been used for BNCT oriented diagnostics for decades[7]. Routine protocol handles the lesion boron uptake to normal tissue ratio received approximately 60-minute after an injection of FBPA as the principle for enrollment in BNCT and dose calculation[8, 9]. The standard uptake value (SUV) is quantitative PET image data, which is defined as Eq. 1[10, 11]. Where I is radioactivity intensity measured in region of interest (ROI) in Bq/ml, A is decay-corrected amount of the PET tracer in Bq, and W is weight of the patient in kg. The SUV ratio of tumor to blood (TBR) and that of tumor to normal tissue (TNR) indicate the difference in uptake of boron and therapeutic benefit. Related studies have explored the dependence between tumor activity, malignancy and SUV of FBPA[12, 13]. TNR or TBR greater than 2.5 is applied as the screening criterion for BNCT, which is commonly used in clinical trials[14–16].

$$\:\text{S}\text{U}\text{V}=\frac{I}{A/W}$$

1

Nevertheless, the SUV of tumor delineated on PET images is distributed, which means that tumor cells have diverse capacities to accumulate ¹⁰B. There may be a distinction between the definition of tumor extent in functional images of PET and anatomical image of magnetic resonance imaging (MRI) and computed tomography (CT). Neuboron Medical Group provided an approach to quantify a heterogeneous boron concentration distribution by image fusion technique so as to accurately achieve the irradiation goal, including tumor prescribed dose and normal tissue limit dose[17]. The boron dose in the ROI is essentially resolved through the product of the spatial distribution of ¹⁰B and thermal neutrons. Consequently, regardless of whether homogeneous or heterogeneous boron distribution is utilized in BNCT dose calculation, comparable mean dose can be obtained, but the heterogeneous boron dose calculation method can more positively reflect the dose hot spots and cold spots in the ROI. That study used FBPA PETCT fusion images and hyper-segmentation boron concentrations to demonstrate the impact of heterogeneous boron on dose calculations.

This study follows the investigator-initiated trial (IIT) protocol with accelerator-based NeuPex system[18] at Xiamen Humanity Hospital (XHH), operating CT simulation (CT-sim), MRI, and FBPA PET/CT of the subject for multi-modality image fusion to create a treatment plan. Considering the error of image alignment and fusion and the significance of clinical response, working hypo-segmentation boron distribution within ROI and explore its prominence.

The subjects’ CT-sim, MRI, and PET scan were completed within a week to avoid disparities triggered by physiological and pathological transforms. The images provided by XHH were approved by the Institutional Review Board (IRB) under the National Key Research and Development Program (Grant No. 3502720201031). Employ image fusion technique to map anatomical information of MRI and functional information of PET to CT-sim image, and use it to design and produce treatment plans. Note that the PET image has been reconstructed with measured attenuation correction (MAC)[19], that is, the attenuation factor of ¹⁸F isotope during scanning has been taken into account. This study has recourse to the treatment planning system NeuMANTA[20] (Multifunctional Arithmetic for Neutron Transportation Analysis) to establish the computational model and exercises COMPASS[21] (COMpact PArticle Simulation System) as the dose calculation engine.

The TBR in this study is the SUV ratio of the target tissue to blood, and the target tissue can be a tumor or an organ at risk (OAR). For hyper-segmentation procedure, the TBR of voxels are classified to account for the heterogeneous boron distribution and minimize the orthogonal distance between the segmented TBR and the original TBR. The procedure is carried out according to Eq. 2, where TBR_upper and TBR_lower are the upper and lower limits of TBR of ROI, respectively. And a normalization factor k, as in Eq. 3, is manipulated to guarantee that the total ¹⁰B remains the constant after segmentation. Where N_i is the number of counts of the i^th group, and TBR(V) is that of V^th voxel in ROI. For hypo-segmentation course, limit the TBR range per segmentation, that is, limit TBR_upper and TBR_lower, to at least 0.5 to reduce the number of segmentation.

$$\:TBR=\frac{{TBR}_{upper}-{TBR}_{lower}}{50}\times\:\left(i-0.5\right)+{TBR}_{lower}i=1,\:2,\:3,\dots\:,\:50$$

2

$$\:k=\frac{{\int\:}_{V}^{}TBR\left(V\right)dV}{{\sum\:}_{i}^{}{N}_{i}{TBR}_{i}}$$

3

This article illustrates the application of boron concentration segmentation in glioblastoma multiforme (GBM) and buccal mucosal squamous cell carcinoma (SCC) case. The image acquisition of cases was completed within one week, and image information was listed in Table 1. The GBM case is a 49-year-old male who received the second BNCT treatment. The two treatments were separated by one month. The SCC case is 79-year-old female with a primary lesion located in the oral cavity. The FBPA/CTsim fusion images of GBM case and SCC case are shown in Fig. 1 and Fig. 2, respectively, and the gross tumor volume (GTV) was contour in red line. In addition, mucosa in the SCC case is defined as TBR greater than 2.0.

Table 1

Image information for GBM and SCC cases
Case	Image modality	Voxel size (mm³)	Manufacturer, model
GBM	MRI	0.65×0.65×2.50	Philips, Ingenia
	FBPA PET	3.65×3.65×3.27	GE, Discovery 690
	CTsim	0.69×0.69×2.50	Philips, Brilliance Big Bore
SCC	MRI	0.69×0.69×3.00	Philips, Ingenia
	FBPA PET	1.17×1.17×1.45	UIH, uMI Panorama 35S
	CTsim	1.07×1.07×3.00	Philips, Brilliance Big Bore

BNCT bioequivalent dose is sum of physical boron dose multiply by compound biological effectiveness (CBE), the product of physical neutron dose and relative biological effectiveness of neutron (RBE_n), and that of physical photon dose and RBE of photon (RBE_p), and the unit is Gy-Eq. Both cases were deliver BPA 200 mg/kg/hr for 2 hours before irradiation, and 100 mg/kg/hr during the irradiation. Utilize blood sample data of before and after irradiation to acquire average blood boron concentration during irradiation and employ to calculation dose distribution which the patient accepted. The boron concentration of all ROIs is conduct from TBR in FBPA PET/CTsim fusion image. The IIT accepted in these cases was conducted in accordance with the principles of the Declaration of Helsinki and the Guidelines for Good Clinical Practice. Approval was obtained from the Medical Ethics Committee of Xiamen Humanity Hospital (No. HAXM-MEC-20220901-005-02), and informed consent was obtained from all subjects. The trial was registered in the Chinese Clinical Trial Registration Center (registration number: ChiCTR2200066473) and the Medical Research Registration and archival information system of the National Safety and Health Security Information Platform.

The blood boron concentration of GBM cases during irradiation was 27.29 ppm. GTV TBR ranged from 0.9 to 3.7, with TBR greater than 2.0 and 2.5 accounting for 59% and 17% of the volume respectively. Brain TBR ranged from 0.3 to 2.4, with average of 0.95. The brain is with 0.04% of the volume having a TBR greater than 2.0 (V_TBR≥2.0) and 1.5% of the volume having a TBR greater than 1.5 (V_TBR≥1.5). Performed hyper- and hypo-segmented boron distribution dose calculations on GTV and brain respectively. The TBR-volume histogram is shown in Figure 3, where the segmented TBR is 0.5 for each segment. The dose-volume histogram (DVH) is shown in Figure 4.

The dose covered 80% volume (D₈₀) of GTV is 21.42 Gy-Eq (CBE_GTV=3.8, RBE_n=3.2, RBE_p=1.0) in hyper-segmented boron and 21.92 Gy-Eq in hypo-segmented boron, a difference of 2.3%. The minimum dose (D_min) of GTV in hyper-and hypo-segmentation are 8.43 Gy-Eq and 13.19 Gy-Eq, respectively, and the difference is 56.4%. However, the doses to GTV volumes of 0.6% and 0.3% are less than 14 Gy-Eq (V_D≤14Gy-Eq) in the hyper-and hypo-segmented boron calculations, respectively. The dose results of GTV and the difference between of two approaches are in Table 2. For brain, the maximum dose (D_max) and mean dose (D_mean) of hyper-segmented boron are 15.14 Gy-Eq (CBE_brain=1.3, RBE_n=3.2, RBE_p=1.0) and 2.46 Gy-Eq separately, while that of hypo-segmented boron are 14.46 Gy-Eq and 2.46 Gy-Eq, with differences of 4.5% and 0.0%, as shown in Table 3. Nevertheless, in both methods, the brain dose exceeding 14 Gy-Eq (V_D≥14Gy-Eq) is less than 0.4% volume. The dose map of hyper- and hypo-segmented boron distribution are shown in and Figure 5, and the dose cold spots are consistent with each other. The follow-up MRI of lesion changes are shown in Figure 6.

Table 2 GTV dose results of the GTV case in hyper- and hypo-segmented boron distribution

Boron distribution	D_max(Gy-Eq)	D_min(Gy-Eq)	D_mean(Gy-Eq)	D₈₀(Gy-Eq)	V_D≤14Gy-Eq
Hyper-segmented	43.52	8.43	26.52	21.42	0.6%
Hypo-segmented	41.08	13.19	26.95	21.92	0.3%
Difference	-5.6%	56.4%	1.6%	2.3%	-
*The maximum dose statistical error of both are less than 1.0%

Table 3 Brain dose results of the GTV case in hyper- and hypo-segmented boron distribution

Boron distribution	D_max(Gy-Eq)	D_mean(Gy-Eq)	V_D≤14Gy-Eq
Hyper-segmented	14.37	2.65	0.7%
Hypo-segmented	12.98	2.61	0.7%
Difference	-9.7%	-1.5%	-
*The maximum dose statistical error of both are less than 1.0%

For the SCC case, the blood boron concentration was 26.30 ppm during irradiation. The GTV TBR range is 0.9~8.1, with average of 3.52. The TBR-volume histogram is shown in Figure 7. TBR is greater than 2.0 for 78% of GTV volume and that of more than 2.5 for 65% of GTV volume. Mucosa TBR ranged from2.0~3.4. In hyper-segmentation approach, TBR range of GTV is divided more than 250 segments. In hypo-segmentation method, the TBR of GTV is divided into seven segments, and the TBR of mucosa is divided into three segments with TBR of1.0. The DVH is shown in Figure 8. Dose results of GTV in hyper- and hypo-segmented boron distribution is shown in Table 4. Mucosa D_max is 14.46 Gy-Eq (CBE_mucosa=2.5, RBE_n=3.2, RBE_p=1.0) in hyper-segmentation, volume of dose over 13 Gy-Eq (V_D≥13Gy-Eq) is less than 3.1%, and volume of dose over 6 Gy (V_D≥6Gy-Phy) is less than 2.3%. While, mucosa D_max is 15.14 Gy-Eq in hypo-segmentation, and V_D≥13Gy-Eq and V_D≥6Gy-Phyis less than 3.0% and 2.3% respectively. Dose results of mucosa in both methods is shown in Table 5. This patient developed mucositis grade 2 after receiving BNCT and recovered within a week. In lesion region, both heterogeneous boron methods have similar D_mean and D₈₀. The dose map is shown in Figure 9 and follow-up MRI of lesion changes is shown in Figure 10.

Table 4 GTV dose results of the SCC case in hyper- and hypo-segmented boron distribution

Boron distribution	D_max(Gy-Eq)	D_min(Gy-Eq)	D_mean(Gy-Eq)	D₈₀(Gy-Eq)	V_D≤14Gy-Eq
Hyper-segmented	60.79	3.60	22.45	13.45	22%
Hypo-segmented	64.82	4.01	22.57	13.42	22%
Difference	6.6%	11.4%	0.5%	-0.2%	-
*The maximum dose statistical error of both are less than 1.0%

Table 5 Mucosa dose results of the GTV case in hyper- and hypo-segmented boron distribution

Boron distribution	D_max(Gy-Eq)	V_D_≥_14Gy-Eq	V_D_≥_6Gy-Phy
Hyper-segmented	14.46	3.1%	2.3%
Hypo-segmented	15.14	3.0%	2.3%
Difference	4.7%	-	-
*The maximum dose statistical error of both are less than 1.0%

The BNCT dose of the ROI mainly comes from the boron dose, especially the ROI with high boron uptake. Traditional BNCT dose calculations assume that the amount of boron absorbed by normal tissue is the same as the blood boron concentration. However, some OARs may have specific amounts of boron uptake based on the patient's medical condition, based on the relative distribution of boron in the patient's body as predicted by the ¹⁸FBPA PET scan. Hyper-segmented boron distribution was utilized to simulate heterogeneous boron distribution in the ROI. However, describing boron distribution too finely in dose calculations and assessments may not be clinically meaningful. Since medical images exist macroscopically and have different modalities with different voxel sizes, fusion techniques may have errors. But it is certain that consideration of boron distribution is an important and necessary factor in BNCT dose assessment.

At this stage, it was found that dose distribution results calculated using hypo-segmented boron heterogeneous boron distribution dose are consistent with those of hyper-segmentation, and large deviations between D_max and D_min are expected. The reference[7] uses PET/CT images to calculate heterogeneous boron distribution, and the image fusion error is low, but it is still affected by partial volume effects; This study uses PET/CT-sim fusion in accordance with the BNCT process specifications implemented by XHH. Under different scanning time and different equipment, the image alignment fusion error may be larger. However, the DVH of hypo-segmented heterogeneous boron distribution is consistent with the results of hypo-segmentation, indicating that the use of this method to describe boron heterogeneity has the authenticity of the impact of macroscopic distribution on dose.

In the GBM case, it can be observed that the average boron uptake in brain tissue is similar to that in blood, i.e., TBR of brain is close to 1, but the distribution may not be uniform in brain tumor patients. Due to the impact of the epidemic, the patient did not receive FBPA PET scan in the first BNCT, so it was not possible to confirm whether the high uptake around the lesion was related to the treatment. This suggests the importance of the FBPA a PET scan. The patient was treated with the same irradiation method twice. The MRI image 90 days after the second BNCT could not measure the lesion. The tumor was evaluated as a complete response according to RANO (Response Assessment in Neuro-Oncology) criteria. This response was consistent with the dose calculation result of V_D≤14Gy-Eq ≤ 0.6%. The patient's side effect was mild cerebral edema after the first BNCT, which recovered after receiving a single dose of bevacizumab; The neurological symptoms worsened after the second treatment, and this response was poorly correlated with dose outcomes V_D≥14Gy-Eq ≤ 0.4% and D_mean of 2.46 Gy-Eq. However, considering that a patient receives BNCT twice, regardless of the tumor or OAR, the impact of the cumulative dose of the two treatments as well as the correction of the treatment interval should be taken into account.

the effect of the dose of the two treatments and the repair of the treatment interval should be considered. Therefore, consistency between treatment response and dose cannot be confirmed.

The SCC patient underwent MRI, FDG PET/CT scan and biopsy 90 days after receiving a single BNCT. The tumor was evaluated as complete response according to RECIST 1.1 (Response Evaluation Criteria in Solid Tumors). This result shows a significant difference from the heterogeneous boron distribution dose result V_D≤14Gy-Eq ≤ 22%. Considering that previous clinical trials used FBPA PET imaging to measure TBR for homogeneous boron distribution dose calculation, the reference[18] mentioned that the tumor range measured by FBPA PET image was smaller than the MRI-defined lesion range in most patients. It is obvious that the calculated dose reference provided by past experience has the potential to be overestimated. However, the dose-response relationship between common mucositis side effects in the BNCT of head-and-neck cancer is not clear due to differences in physical doses depending on how the anatomical location is defined or the TBR threshold is defined. Therefore, once again the importance of highly realistic dose calculations for therapeutic response is emphasized.

For the advancement of TCP, it is difficult for BNCT to proactively deliver the prescribed dose. Therefore, the dose cold spots generated in tumor areas with low BPA uptake may need to be enhanced through other radiotherapy modalities. That is to say, the low-dose area of the tumor is what needs to be paid attention to after BNCT. Tumor dose cold spots can be clearly defined through hypo-segmented boron dose calculation, and the dose distribution results are highly consistent with hyper-segmentation. For the improvement of NTCP, BNCT often focuses on brain when treating brain tumors, and mucosa when treating head and neck cancer. Brain tissue may have uneven boron distribution due to treatment or tumor effects. Mucosa has different BPA uptake than soft tissue. The application of segmentation boron concentration can improve the confidence of dose results, while the use of hypo-segmentation is sufficient to assess clinical response.

Competing interests

The authors declare no competing interests.

Author Contribution

Yi-Chiao Teng contributed to the general research objectives, synthesized study data and prepared the original manuscript;Jiang Chen and Wan-Bing Zhong created of models and implemented of the computer code and supporting algorithms; andYuan-Hao Liu contributed to reviewing and editing of the manuscript.

Acknowledgement

This research was made possible through the support of the National Natural Science Foundation of China under Grant No. 12261131621. Additional support was also provided by the National Key R&D program of China under Grant No. 2023YFE0197700, and the Scientific Research Fund of NIIT under Grant No. YK21-01-06.

Data Availability

The datasets used and/or analyzed during the current study are available from the corresponding author upon reasonable request.

Kankaanranta, L. et al. L-boronophenylalanine-mediated boron neutron capture therapy for malignant glioma progressing after external beam radiation therapy: a Phase I study. Int. J. Radiat. Oncol. Biol. Phys. 80 (2), 369–376 (2011).
Wang, L. W. et al. Boron Neutron Capture Therapy Followed by Image-Guided Intensity-Modulated Radiotherapy for Locally Recurrent Head and Neck Cancer: A Prospective Phase I/II Trial. Cancers. 15 (10), 2762 (2023).
Chen, Y. W. et al. Salvage Boron Neutron Capture Therapy for Malignant Brain Tumor Patients in Compliance with Emergency and Compassionate Use: Evaluation of 34 Cases in Taiwan. Biology (Basel), 10(4). (2021).
Hiratsuka, J. et al. Long-term outcome of cutaneous melanoma patients treated with boron neutron capture therapy (BNCT). J. Radiat. Res. 61 (6), 945–951 (2020).
Bortolussi, S. et al. Understanding the potentiality of accelerator based-boron neutron capture therapy for osteosarcoma: dosimetry assessment based on the reported clinical experience. Radiat. Oncol. 12 (1), 130 (2017).
Watanabe, T. et al. Comparison of the pharmacokinetics between L-BPA and L-FBPA using the same administration dose and protocol: a validation study for the theranostic approach using [(18)F]-L-FBPA positron emission tomography in boron neutron capture therapy. BMC Cancer. 16 (1), 859 (2016).
Ishiwata, K. 4-Borono-2-18F-fluoro-l-phenylalanine PET for boron neutron capture therapy-oriented diagnosis: overview of a quarter century of research. Ann. Nucl. Med. 33 (4), 223–236 (2019).
Lin, Y. C. et al. Similar T/N ratio between (18)F-FBPA diagnostic and BPA therapeutic dosages for boron neutron capture therapy in orthotropic tongue cancer model. Ann. Nucl. Med. 34 (1), 58–64 (2020).
Menichetti, L. et al. Positron emission tomography and [18F]BPA: a perspective application to assess tumour extraction of boron in BNCT. Appl. Radiat. Isot. 67 (Suppl), 7–8 (2009).
Kinahan, P. E. & Fletcher, J. W. Positron emission tomography-computed tomography standardized uptake values in clinical practice and assessing response to therapy. Semin Ultrasound CT MR. 31 (6), 496–505 (2010).
Soongsathitanon, S., Masa-Ah, P. & Tuntawiroon, M. A new Standard Uptake Values (SUV) calculation based on pixel intensity values. 6: pp. 26–33. (2012).
Kobayashi, K. et al. In vivo spatial correlation between 18F-BPA and 18F-FDG uptakes in head and neck cancer. Appl. Radiat. Isot. 115, 138–146 (2016).
Isohashi, K. et al. Exploration of the threshold SUV for diagnosis of malignancy using (18)F-FBPA PET/CT. Eur. J. Hybrid. Imaging. 6 (1), 35 (2022).
Lan, T. L. et al. Advances in Boron Neutron Capture Therapy (BNCT) for Recurrent Intracranial Meningioma. Int. J. Mol. Sci., 24(5). (2023).
Chen, Y. W. et al. Using precise boron neutron capture therapy as a salvage treatment for pediatric patients with recurrent brain tumors. Therapeutic Radiol. Oncol., 4. (2020).
Wang, L. W. et al. Clinical trials for treating recurrent head and neck cancer with boron neutron capture therapy using the Tsing-Hua Open Pool Reactor. Cancer Commun. 38 (1), 37 (2018).
Teng, Y. C. et al. Correcting for the heterogeneous boron distribution in a tumor for BNCT dose calculation. Sci. Rep. 13 (1), 15741 (2023).
Advances in Boron Neutron Capture Therapy. Vienna: INTERNATIONAL ATOMIC ENERGY AGENCY. (2023).
Shiri, I. et al. Direct attenuation correction of brain PET images using only emission data via a deep convolutional encoder-decoder (Deep-DAC). Eur. Radiol. 29 (12), 6867–6879 (2019).
Chen, J. et al. Development of Monte Carlo based treatment planning system for BNCT. Journal of Physics: Conference Series, 2313(1): p. 012012. (2022).
Zhong, W. B. et al. Introduction to the Monte Carlo dose engine COMPASS for BNCT. Sci. Rep. 13 (1), 11965 (2023).

No competing interests reported.

Hypo-Segmented Heterogeneous Boron Distribution Application in Boron Neutron Capture Therapy

Status:

Version 1

Abstract

Figures

Introduction

Methods

Results

Conclusion

Discussion

Declarations