Evaluation of two-dimensional total bone uptake (2D-TBU) and bone scan index (BSI) extracted from active bone metastatic burden on the bone scintigraphy in patients with radium-223 treatment

35 Objective: Radium-223 is a first-line alpha-emitting radionuclide treatment for patients 36 with metastatic castration-resistant prostate cancer (mCRPC) with bone metastases, and 37 the completion and prognosis of this treatment are intensively studied. Although the 38 spread-based bone scan index (BSI) and the novel index of intensity-based two-39 dimensional total bone uptake (2D-TBU) from anomalous hot spot counts of bone


Introduction
It is estimated that almost 1.4 million new cases of prostate cancer appear every year, making it the second most frequent cancer among men worldwide [1].Because prostate cancer causes bone metastasis at a high rate, its control and management are crucial [2,3].Radium-223 is a first-line alpha-emitting radionuclide treatment for patients with metastatic castration-resistant prostate cancer (mCRPC) with bone metastases, and the completion of 6 cycles of this treatment prolongs the median overall survival (OS) and prevents skeletal-related events [4].During radium-223 treatment, the condition of patients is comprehensively monitored using a combination of blood tests and diagnostic imaging.There are no established biomarkers for patient stratification and response assessment, and novel biomarkers are needed to optimize radium-223 treatment [5].
Recently, several studies presented results on predicting the completion of radium-223 treatment and patient selection, and reported on its association with blood indices [6][7][8].
Bone scanning via whole-body bone scintigraphy and single-photon emission computed tomography/computed tomography (SPECT/CT) is routinely performed in clinical practice to assess bone metastases.The bone scan index (BSI) extracted from bone scintigraphy represents the percentage of bone metastatic burden in the total skeletal mass [9,10], making it a useful tool for objective monitoring of bone metastases and predicting the prognosis in patients with prostate cancer [11,12].Furthermore, the threedimensional total bone uptake (3D-TBU) calculated from bone SPECT/CT images is considered an index of active bone metastatic burden that reflects the uptake intensity and can assess response to radium-223 treatment, unlike the spread-based BSI [13].
In contrast, bone SPECT/CT imaging has disadvantages for patients in clinical practice, such as radiation exposure, scan time extension, and examination cost [14,15].
To the best of our knowledge, no two-dimensional intensity-based bone metastasis index combining BSI and 3D-TBU has been evaluated in patients treated with radium-223 until now.Furthermore, studies on the relevance of image-based indicators for radium-223treated patients remain insufficient [5].In this paper, we evaluated the spread-based BSI and the novel intensity-based index of two-dimensional total bone uptake (2D-TBU) extracted from active bone metastatic burden on bone scintigraphy in patients with mCRPC treated with radium-223 to reveal new information on the spread and intensity of the bone metastatic burden, treatment response, treatment completion, and OS.

Patients
Twenty-seven Japanese patients with mCRPC treated with radium-223 (55 kBq/body weight, 4-week intervals, completed in 6 cycles) between July 28, 2016 and March 31, 2023 were retrospectively analyzed.All patients were scheduled for blood tests and bone scans of bone scintigraphy with SPECT/CT before, after 3 cycles, and after 6 cycles of radium-223 treatment, whenever possible.Our Institutional Ethics Committee approved this study (No. 2023-GB-114).
Bone scintigraphy was performed with a scan speed of 20 cm/min in a matrix size of 256 × 1,024.Bone SPECT was performed with 120 projections (10 s/projection) by dual detectors over 360°, a pixel size of 2.4 mm, in a matrix size of 256 × 256.CT was performed using an automatic exposure control of adaptive dose modulation (CARE Dose4D) under the conditions of 130 kV tube voltage and 140 ref.mAs.SPECT data were reconstructed using the quantitative ordered subset conjugate gradient maximization algorithm (xSPECT Quant; Siemens Healthineers, Erlangen, Germany) with one subset, 30 iterations, and a 6-mm Gaussian filter.CT data were reconstructed with a slice thickness of 3 mm using a medium sharp kernel (B50s) and a medium smooth kernel for attenuation correction (B31s).

Data analysis
The bone SPECT/CT images were analyzed using GI-Bone (Nihon Medi-Physics, Tokyo, Japan), and the intensity-based index of 3D-TBU was extracted based on the threshold of standardized uptake value (SUV) > 7 and Hounsfield units > 152 [13].
The bone scintigraphy images were analyzed using VSBONE BSI, version 2.1 (Nihon Medi-Physics, Tokyo, Japan).VSBONE BSI provides the BSI based on skeletal segmentation and detection of skeletal anomalous hot spots, using butterfly-type neural networks and a deep learning-based artificial intelligence approach [16].The BSI expresses the spread-based bone metastatic burden as a percentage of skeletal mass (rBSI).
Furthermore, the novel intensity-based index 2D-TBU was calculated based on counts in each skeletal anomalous hot spot (active intensity) using the following formula: where  is the number of anomalous hot spots extracted from VSBONE BSI and  and  are the mean pixel counts and the number of pixels in each anomalous hot spot, respectively.

Statistical analysis
All indices under study (i.e., 3D-TBU, 2D-TBU, BSI, Hgb, ALP, LDH, PSA, and age) were extracted as continuous variables and are expressed as median values and interquartile ranges (IQRs; 25-75 percentile).Each index was analyzed in terms of correlations, treatment response, and association with treatment completion, and survival analysis was performed.In this study, a numerical change of >30% after 3 and 6 cycle intervals of radium-223 treatment was classified as response to increase or decrease and otherwise as stable in treatment response [4].All statistical analyses were performed using EZR (Saitama Medical Center, Jichi Medical University, Saitama, Japan), a graphical user interface for R (The R Foundation for Statistical Computing, Vienna, Austria), and Statistics Toolkit (Chinese University of Hong Kong, Shatin, Hong Kong) [17].
Patients who completed radium-223 treatment and those who did not complete the treatment regimen were compared using the Mann-Whitney U test.Correlations were expressed using scatter plots and linear regression models.Spearman's rank correlation Patients who were alive on March 31, 2023 were censored for OS analysis.P-values < 0.05 were used to denote statistical significance.

Patients
Table 1 shows the study summary.We analyzed 27 patients, of whom 19 (70.4%) completed 6 cycles of radium-223 treatment and eight (29.6%) did not complete the treatment regimen.The reasons for discontinuing radium-223 treatment were as follows: hematological adverse events (n = 3), progressive disease (n = 3), visceral metastasis (n = 1), and treatment refusal (n =1). Figure 1 shows the analysis of VSBONE BSI.From 53 bone scans, 626 skeletal anomalous hot spots were detected.Table 2 shows the region of the hot spots analyzed.

Correlations
Figure 2 shows the scatter plots and linear regression models of the correlations between 2D-TBU and other indices.Significant positive linear correlations were confirmed between 2D-TBU and 3D-TBU, BSI, ALP, and PSA, and a negative linear correlation was found between 2D-TBU and Hgb.A significant positive linear correlation was confirmed between active intensity and rBSI, which are the basis for determining 2D-TBU and BSI.

Treatment response
Table 4 shows the response to radium-223 treatment.The treatment responses were classified according to 2D-TBU as follows: increased (42.1%), stable (26.3%), and decreased (31.6%).Similarly, the BSI values were classified as increased in 50.0% of the cases, stable in 21.1%, and decreased in 28.9%.Table 5 shows the degree of response agreement between 2D-TBU and other indices.The agreement for BSI was "almost," that for 3D-TBU was "substantial," and that for ALP was "fair."

Univariate analysis
Table 6 shows the univariate analysis and completion rate of radium-223 treatment.The univariate analysis revealed an association between radium-223 completion and median 2D-TBU and BSI levels.A high completion rate of radium-223 treatment was observed in low levels of age, 3D-TBU, 2D-TBU, BSI, ALP, LDH, and PSA and high levels of Hgb.A significant difference in the completion rate was observed in both median 2D-TBU (91.7% vs. 45.5%;p = 0.027) and BSI (91.7% vs. 45.5%;p = 0.027) levels.

Survival analysis
Figure 3 shows the Kaplan-Meier curves for OS according to radium-223 treatment completion.The median OS was 25.2 months (95% CI: 14.0-33.6months) in patients who completed the radium-223 treatment regimen and 7.5 months (95% CI: 3.3-14.2months) in those who did not complete the radium-223 treatment regimen.A significant difference in OS was observed between the two patient groups (p < 0.001).

Discussion
The completion of 6 cycles of radium-223 treatment prolongs the median OS and prevents skeletal-related events [4].Primary assessment and consideration of baseline status and disease characteristics are recommended prior to initiation of radium-223 treatment [18].Various studies have used blood markers and imaging indices for patient selection to determine who would most likely benefit from radium-223 treatment [6-8,19-PSA; high levels of Hgb were significant factors associated with treatment completion [6][7][8].Additionally, the active bone metastatic burden spread-based BSI is commonly used imaging index to evaluate patients treated with radium-223 [24][25][26].Kitajima et al. developed a novel nomogram, including BSI, to select patients with mCRPC to receive radium-223 treatment, and this nomogram was shown to be suitable for assisting therapeutic decision-making [27].Furthermore, Umeda et al. established a novel intensity-based bone scan index of 3D-TBU that is more accurate and sensitive than the BSI for patients receiving radium-223 treatment [13].Similarly, the novel bone metabolism volumetric index using bone SPECT/CT images could improve the low sensitivity of the BSI in patients with low disease grade [28].Skeletal anomalous hot spots contain information about the intensity of bone metastasis with accumulated counts.
Therefore, our research team hypothesized that the intensity-based active bone metastasis burden extracted from bone scintigraphy may be associated with radium-223 treatment, and consequently, they evaluated the novel 2D-TBU index in patients with mCRPC.
In correlation analysis, a strong linear correlation between 2D-TBU and BSI was found.Additionally, a strong linear correlation was also observed between rBSI and active intensity, which detailed the relationships between the spread and intensity of the bone metastatic burden.These results suggest that the spread of bone metastasis is accompanied by comprehensive enhanced active intensity of the bone metastatic burden.

Dittmann et al. and Oya et al. reported that a quantitative assessment of baseline 3D active
bone uptake value from bone SPECT/CT images provides prognostic information on the survival of patients treated with radium-223 [29,30].These results may also be applicable to our 2D-TBU.Part of the discrepancy between rBSI and activity intensity may be explained by discordance in spread and count retention over time [31].
Regarding treatment response, an "almost" agreement between 2D-TBU and BSI was confirmed, and all imaging indices showed a high rate of changes (i.e., increased or decreased) compared with blood markers.The numerical criteria for radium-223 treatment response have not been established [5]; however, imaging indices may be able to reflect the treatment response more sensitively.PSA showed a unilateral increasing trend; however, no significant PSA declines have been reported despite the OS benefit in radium-223 treatment [32][33][34].ALP is considered to have the highest sensitivity among the blood markers under study, as both positive and negative changes were confirmed after treatment.A response reduction of 30% for ALP was adopted as a secondary efficacy endpoint in ALSYMPCA clinical trials [4].The ALP reduction in this study may be explained by the radium-223 radiation effects on skeletal bone metastases with highly proliferating osteoblasts [5,34].In the univariate analysis, associations between the median BSI and 2D-TBU values at baseline and radium-223 treatment completion were confirmed.The results of this study are consistent with those of several previous studies, which confirmed associations of these indices with radium-223 treatment [24][25][26][27].In the survival analysis, patients who completed the radium-223 treatment regimen showed a significantly longer median OS.In this study, the median OS was 25.2 months (95% CI: completion group [6].Including our study, several studies support the importance of completing all 6 cycles of radium-223 treatment [6,7,18,19,23].
This study has several limitations.First, this study was conducted on a limited cohort at a single center.This study design can remove selection bias between institutions; however, the small sample size may have led to a large estimation width, a lower output of test significance, an increased influence of specific individuals on the results, and the inability to remove the effect of confounding factors.Second, image count intensity measured by bone scintigraphy and SPECT was affected by various factors.Acquisition system sensitivity and patient biological and physical characteristics can affect the count intensity-based image index of analysis results, and the standardization and optimization of images are required [35,36].Finally, the accuracy of VSBONE BSI for detecting bone metastasis has not been thoroughly verified.Although VSBONE BSI adopts deep learning-based segmentation, which has a higher segmentation accuracy than atlas-based segmentation, involuntary segmentation errors may occur [37].Additionally, all hot spots extracted by VSBON BSI have not been confirmed to be bone metastatic lesions [38].
Despite these limitations, our study provided important information about the 2D bone metastasis evaluation index extracted from bone scintigraphy in patients treated with radium-223.Further studies are required to maximize the benefits of the treatment for patients with mCRPC.
In conclusions, the spread-based BSI and novel intensity-based 2D-TBU from the active bone metastatic burden on bone scintigraphy were evaluated in patients treated with radium-223.BSI and 2D-TBU were highly correlated and had a strong agreement with response to radium-223 treatment.The BSI could contain information on the spread and comprehensive active intensity of the bone metastatic burden.Furthermore, both indices were found to be associated with treatment completion, with a superior prognosis in the completion group.This study supports the usefulness of preliminary assessment

Tables Captions
coefficient (r) was used to assess the correlations.Response agreement was determined based on Cohen's weighted kappa (k) as follows: almost agreement (k > 0.81); substantial agreement (k = 0.61-0.80);moderate agreement (k = 0.41-0.60);fair agreement (k = 0.21-0.40);slight agreement (k = 0-0.20);nonagreement (k < 0).Fisher's exact test was used to evaluate the distribution of radium-223 completion frequency with the median cutoff level.The association between radium-223 treatment completion and index was evaluated using a univariate logistic regression model.The Kaplan-Meier product limit estimator was used to assess the OS distribution, and the log-rank test was used to determine differences in OS between the two groups.Data were censored on March 31, 2023.Patients who were lost to follow-up were censored at the data of last contact.

Figure 1 :
Figure 1: Analysis of VSBONE BSI.Whole-body bone scintigraphy images and the analysis of VSBONE BSI.(a, b) Analysis of VSBONE BSI with skeletal segmentation in baseline bone scintigraphy, and (c, d) follow-up images of 3 cycles of radium-223 treatment intervals.Red indicates anomalous hot spots, and blue indicates normal hot spots.Evaluated index values are shown at the bottom of the figure, and changes in the values > 30% relative to previous bone scans are shown in bold letters (black: plus, red: minus).

Figure 2 :
Figure 2: Scatter plots and linear regression models.Scatter plots and linear regression models between 2D-TBU and another index: the relationship between 2D-TBU and (a) 3D-TBU and (b) BSI.The detailed relationship between active intensity calculating 2D-TBU and rBSI calculating BSI based on 626 anomalous hot spots.The relationship between 2D-TBU and (d) age, (e) Hgb, (f) ALP, (g) LDH, and (h) PSA.The black dashed line indicates the linear regression line, whereas r indicates Spearman's rank correlation coefficient.

Figure 3 :
Figure 3: Kaplan-Meier curves.The Kaplan-Meier curves for OS according to radium-223 treatment completion (n = 27).The black solid line indicates the survival of patients with radium-223 treatment completion, and the red solid line indicates the survival of patients without radium-223 treatment completion.

Table 2 :
Region of analyzed hot spots.