Abstract
Objective: To identify key risk factors and construct a predictive model for the progression of high bone tumor burden prostate cancer (HBTB-PCa) to castration-resistant prostate cancer (CRPC). Methods: This retrospective study included 367 HBTB-PCa patients treated between January 2018 and May 2021, with 286 cases progressed to CRPC (progression group) and 81 cases did not (non-progression group). Patients were randomly divided into training (n=257) and validation (n=110) sets at a 7:3 ratio. Logistic regression was used to identify independent risk factors, and a Nomogram was built to predict progression risk. Model performance was evaluated using receiver operating characteristic (ROC) curves, calibration plots, and decision curve analysis (DCA). Results: Compared with the non-progression group, patients in the progression group had significantly higher rates of perineural invasion (P=0.011), Gleason score ≥8 (P=0.002), and T4 stage (P=0.012). Laboratory markers including ALP (P<0.001) and LDH (P<0.001) were also elevated in the progression group. Multivariate analysis identified perineural invasion (P=0.032), Gleason score (P=0.002), initial PSA (P=0.025), ALP (P=0.011), LDH (P<0.001), and ALB (P=0.019) as independent predictors of progression to CRPC. The Nomogram demonstrated strong discrimination power (AUC=0.845 in the training set; AUC=0.746 in external validation), with LDH being the most influential predictor. DCA indicated a net clinical benefit up to 77.82%. Conclusions: Perineural invasion, Gleason score ≥8, and elevated ALP and LDH are closely associated with progression from HBTB-PCa to CRPC. The constructed Nomogram (internal AUC=0.845; external AUC=0.746) offers a practical tool for individualized risk assessment and guiding treatment planning in clinical settings.
Keywords: High tumor burden, bone metastasis, prostate cancer, castration-resistant prostate cancer, prediction model
Introduction
Prostate cancer (PCa) is one of the most common malignancies of the male genitourinary system [1,2]. According to GLOBOCAN 2020, Pca ranks second in incidence and fifth in mortality among male cancers worldwide [3]. The incidence and mortality of PCa in China, though lower than the global average, are rising rapidly due to population aging and the increasing use of PSA screening [4]. In 2020, there were approximately 115,000 new PCa cases and 51,000 PCa-related deaths in China, posing a threat to people’s health [3].
PCa is often asymptomatic in its early stages, resulting in low rates of early diagnosis, and the lack of widespread screening programs further contributes to a high proportion of patients being diagnosed at intermediate or advanced stages [5-7]. Many newly diagnosed patients present with metastatic hormone-sensitive PCa, often accompanied by dysuria and skeletal complications, with a five-year survival rate of only 28.2% [8]. In addition, up to 70% of advanced PCa patients develop bone metastases, as the bones microenvironment are conducive for tumor cell proliferation, leading to complications such as bone pain and pathological fractures [9]. High bone tumor burden (HBTB) Pca refers to a clinically severe state characterized by bone metastasis, typically with more than four bone metastatic sites, with at least one lesion outside the axial skeleton (i.e., beyond the spine, pelvis, or ribs), or accompanied by visceral metastasis such as those in lung or liver [10,11]. These patients experience higher mortality and require more complex treatment strategies.
Androgen deprivation therapy (ADT) remains the cornerstone of treatment for advanced PCa. Although most patients initially respond to ADT, progression to castration-resistant prostate cancer (CRPC) is common, with highly variable time to progression [12]. Research has shown that patients with high tumor burden exhibit more rapid progression to CRPC and shorter overall survival compared to those low tumor-burden [13]. Clinically, bone metastases are typically assessed using whole-body bone scintigraphy via single-photon emission computed tomography (SPECT), with contrast-enhanced MRI employed when needed. While these modalities provide anatomic information, they fail to account for other prognostic variables, such as Gleason score or clinical stage [14]. Therefore, there is still a lack of an accurate predictive tools for determining the timing of progression from HBTB-PCa to CRPC.
Nomograms offer a graphical prediction tool that integrates multiple predictors to estimate the probability of clinical events. It provides an individual risk assessment, turning complex data models into intuitive graphs that are easy for doctors and patients to understand. In this study, we developed a nomogram model based on the risk factors identified through Logistic regression to predict the risk of progression from HBTB-PCa to CRPC, facilitating personalized treatment planning and improving patient outcomes.
Methods and materials
Patient source
This retrospective study included HBTB-PCa patients treated at the Second Affiliated Hospital of Guangzhou Medical University between January 2018 and May 2021. The study was approved by the Medical Ethics Committee of the Second Affiliated Hospital of Guangzhou Medical University). Given the retrospective nature of this study, informed consent was waived by the Ethics Committee.
Eligibility and exclusion criteria
Inclusion criteria: 1) Histologically confirmed diagnosis of PCa via transrectal ultrasound-guided prostate biopsy [6]; 2) Diagnosis of HBTB-PCa based on chest, abdomen, and pelvic CT scanning, SPECT whole body bone imaging, or whole body PET-CT imaging; 3) Availability of complete baseline data and pre-treatment hematological findings.
Exclusion criteria: 1) Patients who were not initially diagnosed or treated with endocrine therapy at our institution; 2) Those had a history of other malignancies; 3) Those had recently received radiotherapy or chemotherapy; 4) Those with no definite diagnosis of HBTB-PCa (Figure 1).
Figure 1.
Sample screening flow chart.
Definition of CRPC
According to the Chinese Guidelines for the Diagnosis and Treatment of Urology and Andrology Diseases (2022 edition) [15], disease progression in PCa patients receiving androgen deprivation therapy (ADT) is defined as either prostate-specific antigen (PSA) progression or radiographic progression despite achieving castration-level serum testosterone (<1.7 nmol/L or <50 ng/dL).
PSA progression is defined by three consecutive tests (at least 1-week intervals) showing elevated PSA levels, with the last test showing an increase by more than 50% from the baseline value and the final PSA value exceeding 2 ng/mL. This persistent elevation indicates disease activity despite appropriate testosterone suppression.
Radiographic progression is defined by the appearance of new lesions on imaging, including at least two new bone lesions on bone scintigraphy, or new soft tissue lesions evaluated using RECIST criteria.
Patient grouping
According to the eligibility and exclusion criteria, a total of 367 cases were included, including 286 patients in the progression group and 81 patients in the non-progression group. Then the patients were assigned to a training group (n=257) and a validation group (n=110) at a 7:3 ratio.
Data collection
Patient data were obtained from electronic medical records and follow-up records. Baseline data (recorded at first admission) included age, body mass index (BMI), ethnicity, marital status, education level, family history of cancer, intraductal carcinoma, perineural invasion (PNI), comorbidities (hypertension and diabetes), T stage, N stage, and Gleason score. Laboratory parameters included PSA, serum testosterone, hemoglobin (HGB), alkaline phosphatase (ALP), lactate dehydrogenase (LDH), and albumin (ALB).
Statistical analysis
Data analyses were conducted using SPSS 26.0. The normality of continuous variables were analyzed using the Kolmogorov-Smirnov test. Normally distributed data were expressed as mean ± standard deviation and compared using independent sample t-tests. Non-normally distributed data were presented as interquartile range (IQR) and analyzed using the Mann-Whitney U test. For the comparison of categorical variables, we used chi-square (χ2) tests. When the expected frequencies were greater than or equal to 5, the standard chi-square test was applied. For cases where the expected frequencies were less than 5, we used the Yates’ continuity correction to ensure the accuracy of the test results. A two-tailed P-value <0.05 was considered statistically significant.
Logistic regression was used to identify factors associated with progression from HBTB-Pca to CRPC. The predictive performance of significant factors was evaluated using the receiver operating characteristic (ROC) curves. Model calibration and clinical utility were assessed using calibration curves, the Hosmer-Lemeshow (H-L) goodness-of-fit test, and decision curve analysis (DCA). The Nomogram was developed using the rms package in R software.
Results
Comparison of baseline data with laboratory indicators between the two groups
No significant differences were observed in terms of age, BMI, ethnicity, marital status, education level, family history of cancer, intraductal carcinoma, hypertension, diabetes, or N staging between the two groups (P>0.05, Table 1). However, the progression group showed higher proportions of PNI (P=0.011), Gleason score ≥8 (P=0.002), and T4 stage (P=0.012) compared to the non-progression group.
Table 1.
Comparison of baseline data between the two groups
| Variables | Progression group (n=81) | Non-progression group (n=286) | χ2 | P |
|---|---|---|---|---|
| Age (years old) | 0.959 | 0.327 | ||
| ≥65 years old | 34 | 103 | ||
| <65 | 47 | 183 | ||
| BMI | 1.903 | 0.168 | ||
| ≥25 kg/m2 | 20 | 51 | ||
| <25 kg/m2 | 61 | 235 | ||
| Ethnicity | 2.954 | 0.086 | ||
| Han | 68 | 214 | ||
| Others | 13 | 72 | ||
| Marital status | 0.363 | 0.547 | ||
| Married | 75 | 272 | ||
| Divorced | 6 | 14 | ||
| Education level | 0.396 | 0.529 | ||
| ≥ High school | 47 | 177 | ||
| < High school | 34 | 109 | ||
| Family history of cancer | 0.165 | 0.685 | ||
| With | 4 | 20 | ||
| Without | 77 | 266 | ||
| Intraductal carcinoma | 0.485 | 0.486 | ||
| With | 6 | 31 | ||
| Without | 75 | 255 | ||
| Perineural invasion | 6.454 | 0.011 | ||
| With | 41 | 189 | ||
| Without | 40 | 97 | ||
| Hypertension | 1.479 | 0.224 | ||
| With | 36 | 149 | ||
| Without | 45 | 137 | ||
| Diabetes | 1.322 | 0.25 | ||
| With | 28 | 80 | ||
| Without | 53 | 206 | ||
| T-staging | 6.379 | 0.012 | ||
| T4 | 42 | 192 | ||
| <T4 | 39 | 94 | ||
| N-staging | 0.565 | 0.452 | ||
| N1 | 49 | 186 | ||
| N0 | 32 | 100 | ||
| Gleason score | 10.017 | 0.002 | ||
| ≥8 | 36 | 183 | ||
| <8 | 45 | 103 |
Note: BMI: Body Mass Index, PI: Perineural Invasion, PSA: Prostate-Specific Antigen, HGB: Hemoglobin, ALP: Alkaline Phosphatase, LDH: Lactate Dehydrogenase, ALB: Albumin.
Analysis of laboratory indicators showed no significant differences in initial PSA, testosterone, HGB, or ALB between the two groups (P>0.05). In contrast, ALP (P<0.001) and LDH (P<0.001) levels were significantly elevated in the progression group (Table 2).
Table 2.
Comparison of laboratory parameters between the two groups
| Indicators | Progression group (n=81) | Non-progression group (n=286) | T/Z | P |
|---|---|---|---|---|
| Initial PSA (ng/mL) | 167.49±18.28 | 172.12±24.58 | 1.855 | 0.065 |
| Testosterone (nmol/L) | 17.03±7.92 | 18.11±7.50 | 1.092 | 0.277 |
| HGB (g/L) | 119.79±18.33 | 120.72±15.61 | 0.416 | 0.678 |
| ALP (U/L) | 155.37±28.02 | 142.90±28.72 | 3.469 | <0.001 |
| LDH (U/L) | 219.08±26.39 | 196.86±25.88 | 6.79 | <0.001 |
| ALB (g/L) | 38.00 [33.00, 42.00] | 40.00 [34.00, 45.00] | 1.861 | 0.063 |
Note: PSA: Prostate-Specific Antigen, HGB: Hemoglobin, ALP: Alkaline Phosphatase, LDH: Lactate Dehydrogenase, ALB: Albumin.
Predictive value of laboratory indicators for progression to CRPC
The predictive value of laboratory indicators was assessed using ROC curves. The results showed that all laboratory indicators demonstrated AUC values greater than 0.5, indicating varying degrees of predictive capability. LDH exhibited the highest predictive performance (AUC=0.721; Figure 2E; Table 3), followed by ALP (AUC=0.637; Figure 2D; Table 3). In contrast, initial PSA (AUC=0.563, Figure 2A), testosterone (AUC=0.545, Figure 2B), HGB (AUC=0.512, Figure 2C), and ALB (AUC=0.568, Figure 2F) showed limited predictive value (Table 3).
Figure 2.
ROC curves for predicting patient progression using laboratory indicators. A. ROC curve of initial PSA for predicting patient progression. B. ROC curve of testosterone for predicting patients’ progression. C. ROC curve of HGB for predicting patient progression. D. ROC curve of ALP for predicting patient progression. E. ROC curve of LDH for predicting patients’ progression. F. ROC curve of ALB for predicting patients’ progression. Note: Receiver Operating Characteristic (ROC), Prostate-Specific Antigen (PSA), Hemoglobin (HGB), Alkaline Phosphatase (ALP), Lactate Dehydrogenase (LDH), Albumin (ALB).
Table 3.
ROC curve parameters for laboratory parameters in predicting disease progression
| Marker | AUC | 95% CI | Specificity | Sensitivity | Youden index | Cutoff |
|---|---|---|---|---|---|---|
| Initial PSA | 0.563 | 0.498-0.563 | 88.89% | 30.07% | 18.96% | 186.980 |
| Testosterone | 0.545 | 0.473-0.545 | 64.20% | 50.00% | 14.20% | 18.835 |
| HGB (g/L) | 0.512 | 0.436-0.512 | 29.63% | 78.67% | 8.30% | 108.500 |
| ALP (U/L) | 0.637 | 0.569-0.637 | 72.84% | 55.24% | 28.08% | 153.650 |
| LDH (U/L) | 0.721 | 0.661-0.721 | 81.48% | 55.94% | 37.43% | 215.435 |
| ALB (g/L) | 0.568 | 0.498-0.568 | 71.60% | 41.96% | 13.56% | 41.500 |
Note: AUC: Area Under the Curve, PSA: Prostate-Specific Antigen, HGB: Hemoglobin, ALP: Alkaline Phosphatase, LDH: Lactate Dehydrogenase, ALB: Albumin.
Grouping for model development and validation
Patients were randomly assigned to training (n=257) and validation (n=110) cohorts at a 7:3 ratio. No statistically significant differences were found in baseline characteristics or laboratory parameters between the two groups (P>0.05, Table 4), confirming the comparability of the two cohorts.
Table 4.
Comparison of baseline data between validation and training groups
| Variables | Training group (n=257) | Validation group (n=110) | χ2 | P |
|---|---|---|---|---|
| Age (years old) | 0.063 | 0.802 | ||
| ≥65 years old | 97 | 40 | ||
| <65 | 160 | 70 | ||
| BMI | 0.615 | 0.433 | ||
| ≥25 kg/m2 | 47 | 24 | ||
| <25 kg/m2 | 210 | 86 | ||
| Ethnicity | 0.159 | 0.690 | ||
| Han | 196 | 86 | ||
| Others | 61 | 24 | ||
| Marital status | 0.000 | 0.998 | ||
| Married | 243 | 104 | ||
| Divorced | 14 | 6 | ||
| Education level | 0.189 | 0.664 | ||
| ≥High school | 155 | 69 | ||
| <High school | 102 | 41 | ||
| Family history of cancer | 1.022 | 0.312 | ||
| With | 19 | 5 | ||
| Without | 238 | 105 | ||
| Intraductal carcinoma | 1.213 | 0.271 | ||
| With | 23 | 14 | ||
| Without | 234 | 96 | ||
| Perineural invasion | 0.000 | 0.988 | ||
| With | 161 | 69 | ||
| Without | 96 | 41 | ||
| Hypertension | 0.312 | 0.577 | ||
| With | 132 | 53 | ||
| Without | 125 | 57 | ||
| Diabetes | 1.194 | 0.275 | ||
| With | 80 | 28 | ||
| Without | 177 | 82 | ||
| T-staging | 0.461 | 0.497 | ||
| T4 | 161 | 73 | ||
| <T4 | 96 | 37 | ||
| N-staging | 0.138 | 0.710 | ||
| N1 | 163 | 72 | ||
| N0 | 94 | 38 | ||
| Gleason score | 0.715 | 0.398 | ||
| ≥8 | 157 | 62 | ||
| <8 | 100 | 48 | ||
| Initial PSA | 1.386 | 0.239 | ||
| ≥186.980 | 195 | 77 | ||
| <186.980 | 62 | 33 | ||
| Testosterone | 0.109 | 0.741 | ||
| ≥18.835 | 138 | 57 | ||
| <18.835 | 119 | 53 | ||
| HGB (g/L) | 0.169 | 0.681 | ||
| ≥108.500 | 58 | 27 | ||
| <108.500 | 199 | 83 | ||
| ALP (U/L) | 0.483 | 0.487 | ||
| ≥153.650 | 134 | 53 | ||
| <153.650 | 123 | 57 | ||
| LDH (U/L) | 0.016 | 0.901 | ||
| ≥215.435 | 135 | 57 | ||
| <215.435 | 122 | 53 | ||
| ALB (g/L) | 0.063 | 0.802 | ||
| ≥41.500 | 160 | 64 | ||
| <41.500 | 97 | 46 |
Note: BMI: Body Mass Index, PI: Perineural Invasion, PSA: Prostate-Specific Antigen, HGB: Hemoglobin (HGB), ALP: Alkaline Phosphatase, LDH: Lactate Dehydrogenase, ALB: Albumin.
Identification of predictive factors for the progression from HBTB-PCa to CRPC
Logistic regression analysis was performed to identify factors associated with progression from HBTB-PCa to CRPC. Univariate analysis revealed that PNI (P=0.008), T stage (P=0.039), Gleason score (P<0.001), initial PSA (P=0.021), ALP (P<0.001), LDH (P<0.001), and ALB (P<0.001) were significantly associated with disease progression of HBTB-PCa to CRPC (Table 5). Multivariate analysis subsequently identified PNI (P=0.032), Gleason score (P=0.002), initial PSA (P=0.025), ALP (P=0.011), LDH (P<0.001), and ALB (P=0.019) as independent risk factors for HBTB-PCa progression to CRPC (Table 6).
Table 5.
Univariate analysis of risk factors for the progression of high bone tumor burden prostate cancer to castration-resistant prostate cancer
| Variable | β | SE | P | OR | 95% CI | |
|---|---|---|---|---|---|---|
|
| ||||||
| Lower | Upper | |||||
| Age (years old) | -0.235 | 0.306 | 0.442 | 0.790 | 0.435 | 1.449 |
| BMI | -0.366 | 0.368 | 0.319 | 0.693 | 0.343 | 1.464 |
| Ethnicity | -0.792 | 0.414 | 0.055 | 0.453 | 0.188 | 0.972 |
| Marital status | 0.360 | 0.612 | 0.556 | 1.434 | 0.381 | 4.478 |
| Education level | 0.222 | 0.304 | 0.466 | 1.248 | 0.684 | 2.261 |
| Family history of cancer | 0.448 | 0.648 | 0.489 | 1.565 | 0.498 | 6.907 |
| Intraductal carcinoma | 0.330 | 0.572 | 0.564 | 1.391 | 0.497 | 4.953 |
| Perineural invasion | 0.814 | 0.305 | 0.008 | 2.256 | 1.242 | 4.122 |
| Hypertension | 0.387 | 0.302 | 0.200 | 1.473 | 0.816 | 2.681 |
| Diabetes | -0.529 | 0.312 | 0.090 | 0.589 | 0.321 | 1.094 |
| T-staging | 0.628 | 0.304 | 0.039 | 1.874 | 1.031 | 3.410 |
| N-staging | 0.015 | 0.311 | 0.962 | 1.015 | 0.545 | 1.856 |
| Gleason score | 1.293 | 0.314 | 0.000 | 3.643 | 1.987 | 6.827 |
| Initial PSA | -0.997 | 0.434 | 0.021 | 0.369 | 0.146 | 0.816 |
| Testosterone | -0.404 | 0.307 | 0.187 | 0.667 | 0.362 | 1.210 |
| HGB | -0.265 | 0.347 | 0.444 | 0.767 | 0.394 | 1.547 |
| ALP | -1.303 | 0.339 | 0.000 | 0.272 | 0.136 | 0.517 |
| LDH | -1.935 | 0.390 | 0.000 | 0.144 | 0.063 | 0.297 |
| ALB | -0.666 | 0.333 | 0.046 | 0.514 | 0.261 | 0.970 |
Note: BMI: Body Mass Index, PI: Perineural Invasion, PSA: Prostate-Specific Antigen, HGB: Hemoglobin, ALP: Alkaline Phosphatase, LDH: Lactate Dehydrogenase, ALB: Albumin.
Table 6.
Multivariate analysis identifying independent risk factors for the progression of high bone tumor burden prostate cancer to castration-resistant prostate cancer
| Variable | β | SE | P | OR | 95% CI | |
|---|---|---|---|---|---|---|
|
| ||||||
| Lower | Upper | |||||
| Perineural invasion | 0.767 | 0.358 | 0.032 | 2.154 | 1.071 | 4.394 |
| T-staging | 0.561 | 0.363 | 0.122 | 1.753 | 0.861 | 3.592 |
| Gleason score | 1.133 | 0.357 | 0.002 | 3.105 | 1.556 | 6.351 |
| Initial PSA | -1.092 | 0.487 | 0.025 | 0.336 | 0.120 | 0.830 |
| ALP | -0.977 | 0.383 | 0.011 | 0.376 | 0.173 | 0.785 |
| LDH | -1.911 | 0.430 | 0.000 | 0.148 | 0.060 | 0.329 |
| ALB | -0.930 | 0.397 | 0.019 | 0.395 | 0.176 | 0.841 |
Note: BMI: Body Mass Index, PI: Perineural Invasion, PSA: Prostate-Specific Antigen, HGB: Hemoglobin, ALP: Alkaline Phosphatase, LDH: Lactate Dehydrogenase, ALB: Albumin.
Nomogram model construction and validation
A nomogram was developed based on the six independent predictors identified: PNI, Gleason score, initial PSA, ALP, LDH, and ALB. LDH demonstrated the most significant effect on the progression risk, followed by the other five variables, each contributing significantly to the model (Figure 3A).
Figure 3.
Construction of a Nomogram model for the prediction of progression from HBTB-PCa to CRPC and its internal validation. A. Construction of the Nomogram model. B. Clinical benefit of the Nomogram model in predicting the progression of HBTB-PCa to CRPC. C. Consistency between the predicted probability and the observed probability. D. ROC curve for analyzing the discrimination ability of the Nomogram. Note: Castration-Resistant Prostate Cancer (CRPC), High Bone Tumor Burden (HBTB), Prostate Cancer (PCa), Receiver Operating Characteristic (ROC), Alkaline Phosphatase (ALP), Lactate Dehydrogenase (LDH), Albumin (ALB).
ROC curve, calibration curve, and DCA were used to evaluate the discrimination, calibration, and clinical practical value of the model, respectively. In the training set, the DCA curve revealed a high net clinical benefit, with the threshold probability curve (red line) consistently above the “treat-all” and “treat-none” lines, indicating strong clinical utility, with the highest yield reaching 77.82% (Figure 3B). The calibration curve generated using Bootstrap (iterations =500) showed close alignment between the predicted and observed probabilities in the training set, with the calibration line overlapping the diagonal reference line, indicating good model calibration; the concordance index (C-index) was 0.845 (95% CI: 0.787-0.903), and the goodness-of-fit test yielded a P-value of 0.552, indicating adequate model fit (Figure 3C). Furthermore, ROC curve analysis demonstrated strong discriminatory performance, with an AUC of 0.845 for predicting progression from HBTB-PCa to CRPC (Figure 3D).
The predictive model was then validated using the validation cohort. The DCA curve also demonstrated a high net benefit in the validation set (Figure 4A). Similarly, the calibration curve showed close alignment between the predicted and observed probabilities in the validation set. The C index was 0.746 (95% CI: 0.628-0.865), and the goodness-of-fit test yielded a P-value of 0.401, indicating good model fit (Figure 4B). Additionally, ROC curve showed an AUC of 0.746 in the validation set, supporting the model’s discriminatory ability in predicting the progression of HBTB-PCa to CRPC though it is slightly lower than the AUC obtained in the training set (Figure 4C).
Figure 4.
External validation of the Nomogram model. A. Clinical benefit of the Nomogram model in predicting the progression of HBTB-PCa to CRPC in validation cohort. B. Consistency between the predicted probability and the observed probability in validation cohort. C. ROC curve for analyzing the discrimination ability of the Nomogram in validation cohort. Note: Castration-Resistant Prostate Cancer (CRPC), High Bone Tumor Burden (HBTB), Prostate Cancer (PCa), Receiver Operating Characteristic (ROC).
Discussion
Prostate cancer (PCa) is one of the most common malignancies among older men, ranking high in both incidence and mortality rates globally [16,17]. While early-stage or localized PCa can often be effectively controlled with radical surgery or radiotherapy, advanced PCa remains prevalent in China, especially in patients with HBTB. These patients typically experience poor prognosis and significant psychological stress [10]. Given these challenges, identifying reliable predictors for progression to CRPC is critical in guiding timely interventions, optimizing clinical decision-making, and improving patient quality of life.
In this study, Logistic regression analysis identified six key risk factors independent associated with progression from HBTB-PCa to CRPC: PNI, Gleason score, initial PSA, ALP, LDH, and ALB. These variables reflect distinct aspects of tumor biology, disease burden, and systemic response. Among them, PNI is particularly noteworthy, as it illustrates how tumor cells can utilize nerve sheaths to invade surrounding tissues. This phenomenon is closely related to alterations in tumor microenvironment and neurotrophic signaling pathways that facilitate cancer cell migration [18,19]. Van et al. [20] reported that PNI significantly increases the risk of aggressive disease and is associated with higher chances of positive surgical margins, influencing postoperative outcomes. A comprehensive review by Niu et al. [21] similarly highlighted that PNI is frequently observed in multiple malignancies, including PCa, and correlates with adverse clinicopathological parameters. Further, Reeves et al. [22] identified PNI as an independent predictor for biochemical recurrence in patients undergoing radical prostatectomy.
Another essential predictor is the Gleason score, which reflects the level of histological differentiation and aggressiveness of PCa. A high Gleason score (≥8) signifies a more heterogeneous and poorly differentiated tumor population, which is often associated with reduced sensitivity to conventional therapies and an increased risk of progression to CRPC [23,24]. Notably, PCa with a Gleason score of 9-10 has been associated with significantly worse outcomes under androgen deprivation therapy compared to those with a Gleason score below 8 [25,26]. In line with this, Ham et al. [27] highlighted that a Gleason score of 9-10 was associated with higher all-cause and PCa-specific mortality than a Gleason score of 8, reinforcing the importance of thorough risk stratification in clinical management.
In terms of laboratory indicators, initial PSA, ALP, LDH, and ALB were identified as significant predictors of progression in HBTB-PCa patients. Elevated PSA generally suggests an increased tumor burden and more aggressive disease behavior [28]. Increased ALP and LDH levels are commonly associated with active bone involvement and enhanced tumor metabolic activity, suggesting both skeletal metastases and systemic disease progression [29,30]. Schlack et al. [29] noted that elevated ALP levels during enzalutamide therapy were associated with worse prognosis in osseous metastatic CRPC patients, while Poteska et al. [30] observed that dynamic changes in ALP and LDH levels may correlate with clinical benefit, disease-free survival, and overall survival, even in the absence of a PSA response. Additionally, hypoalbuminemia may reflect poor nutritional status and a systemic pro-inflammatory state, both of which are correlated with more advanced disease states and worse outcomes.
Building on these observations, we integrated the six independent risk factors (PNI, Gleason score, initial PSA, ALP, LDH, and ALB) to construct a Nomogram model for predicting HBTB-PCa progression to CRPC. The model demonstrated strong predictive performance, with an AUC of 0.845 in the training set and an AUC of 0.746 in the validation, underscoring its discrimination and robustness. Among the predictors, LDH contributed the most to the model, potentially reflecting an association with hypoxic tumor microenvironments and heightened metabolic activity. Moreover, the Nomogram incorporated other biological and clinical factors into an individualized risk profile, which may facilitate in making more personalized treatment strategies, such as early intensification of ADT, prompt initiation of radiotherapy, or the incorporation of novel hormonal agents.
While the Nomogram demonstrated favorable discrimination and clinical utility, slight deviations at higher probability thresholds were observed in the calibration analysis. These discrepancies could stem from a relatively modest sample size during external validation or the heterogeneity of patient characteristics. To improve generalizability, further validation in larger, multicenter cohorts with broader demographic and clinical diversity is warranted. Moreover, future studies may consider incorporating emerging biomarkers-such as genomic alterations, molecular imaging parameters-to further enhance the model’s predictive accuracy. Incorporating such advanced diagnostic tools could also contribute to a deeper understanding of the mechanisms underlying disease progression.
Limitations
Despite providing important insights, this study has several limitations. First, its retrospective, single-center design may introduce selection and information biases, thereby limiting causal inferences and generalizability. Second, the relatively short follow-up period for some patients restricted the evaluation of long-term outcomes and survivorship trends. Third, although internal and external validations were performed, the overall sample size remains modest for constructing a comprehensive prediction tool; larger cohorts are needed to improve the model’s reliability and refine cut-off points. Fourth, certain potentially relevant variables-such as genetic alterations (e.g., BRCA mutations) and advanced imaging findings-were not included in the analysis but may have enriched our understanding of disease progression.
Future directions
To address these concerns, future studies should adopt prospective, multicenter designs with standardized protocols. Increasing the sample size and diversity of patient populations will improve the robustness and external validity of the nomogram. Furthermore, incorporating novel biomarkers and leveraging next-generation sequencing data may provide deeper insights into the molecular pathways driving CRPC progression. Incorporating these factors with traditional clinical and laboratory variables could yield an even more powerful, personalized tool. Meanwhile, extended follow-up periods would shed light on the Nomogram’s utility for long-term prognostic assessments and support more dynamic decision-making regarding therapeutic escalation or de-escalation.
Conclusion
In summary, this study highlights several key risk factors-PNI, Gleason score ≥8, initial PSA, elevated ALP and LDH, and decreased ALB-that are independently associated with HBTB-PCa progression to CRPC. By integrating these variables into a Nomogram (internal AUC=0.845; external AUC=0.746), we offer a practical and user-friendly clinical tool for individualized risk assessment and treatment planning. This model has the potential to facilitate early identification of high-risk patients and guide personalized therapeutic strategies, ultimately improving clinical outcomes in HBTB-PCa patients.
Acknowledgements
The Grants from the National Natural Science Foundation of China (82160577, 82073762); Key Program for Regional Joint Funds of Natural Science Foundation of Guangdong Province (No. 2020B1515120094, No. 2021B1515120053); The Program for Science and Technology Plan of Guangzhou (No. 2023A03J0405; 202102010071); and The Guangdong Medical Science and Technology Research Fund (No. A2023205).
Disclosure of conflict of interest
None.
References
- 1.Pinsky PF, Parnes H. Screening for prostate cancer. N Engl J Med. 2023;388:1405–1414. doi: 10.1056/NEJMcp2209151. [DOI] [PubMed] [Google Scholar]
- 2.Waiting on Prostate Cancer. N Engl J Med. 2023;388:e54. doi: 10.1056/NEJMp2303613. [DOI] [PubMed] [Google Scholar]
- 3.Sung H, Ferlay J, Siegel RL, Laversanne M, Soerjomataram I, Jemal A, Bray F. Global cancer statistics 2020: GLOBOCAN estimates of incidence and mortality worldwide for 36 cancers in 185 countries. CA Cancer J Clin. 2021;71:209–249. doi: 10.3322/caac.21660. [DOI] [PubMed] [Google Scholar]
- 4.Chen W, Zheng R, Baade PD, Zhang S, Zeng H, Bray F, Jemal A, Yu XQ, He J. Cancer statistics in China, 2015. CA Cancer J Clin. 2016;66:115–132. doi: 10.3322/caac.21338. [DOI] [PubMed] [Google Scholar]
- 5.Wilt TJ, Jones KM, Barry MJ, Andriole GL, Culkin D, Wheeler T, Aronson WJ, Brawer MK. Follow-up of prostatectomy versus observation for early prostate cancer. N Engl J Med. 2017;377:132–142. doi: 10.1056/NEJMoa1615869. [DOI] [PubMed] [Google Scholar]
- 6.Ahdoot M, Wilbur AR, Reese SE, Lebastchi AH, Mehralivand S, Gomella PT, Bloom J, Gurram S, Siddiqui M, Pinsky P, Parnes H, Linehan WM, Merino M, Choyke PL, Shih JH, Turkbey B, Wood BJ, Pinto PA. MRI-targeted, systematic, and combined biopsy for prostate cancer diagnosis. N Engl J Med. 2020;382:917–928. doi: 10.1056/NEJMoa1910038. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7.Wang X, Zhao J, Marostica E, Yuan W, Jin J, Zhang J, Li R, Tang H, Wang K, Li Y, Wang F, Peng Y, Zhu J, Zhang J, Jackson CR, Zhang J, Dillon D, Lin NU, Sholl L, Denize T, Meredith D, Ligon KL, Signoretti S, Ogino S, Golden JA, Nasrallah MP, Han X, Yang S, Yu KH. A pathology foundation model for cancer diagnosis and prognosis prediction. Nature. 2024;634:970–978. doi: 10.1038/s41586-024-07894-z. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8.Wala J, Nguyen P, Pomerantz M. Early treatment intensification in metastatic hormone-sensitive prostate cancer. J. Clin. Oncol. 2023;41:3584–3590. doi: 10.1200/JCO.23.00723. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9.Siegel RL, Miller KD, Jemal A. Cancer statistics, 2019. CA Cancer J Clin. 2019;69:7–34. doi: 10.3322/caac.21551. [DOI] [PubMed] [Google Scholar]
- 10.Ye X, Huang X, Fu X, Zhang X, Lin R, Zhang W, Zhang J, Lu Y. Myeloid-like tumor hybrid cells in bone marrow promote progression of prostate cancer bone metastasis. J Hematol Oncol. 2023;16:46. doi: 10.1186/s13045-023-01442-4. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11.Li S, Kang Y, Zeng Y. Targeting tumor and bone microenvironment: novel therapeutic opportunities for castration-resistant prostate cancer patients with bone metastasis. Biochim Biophys Acta Rev Cancer. 2024;1879:189033. doi: 10.1016/j.bbcan.2023.189033. [DOI] [PubMed] [Google Scholar]
- 12.Hoffman KE, Penson DF, Zhao Z, Huang LC, Conwill R, Laviana AA, Joyce DD, Luckenbaugh AN, Goodman M, Hamilton AS, Wu XC, Paddock LE, Stroup A, Cooperberg MR, Hashibe M, O’Neil BB, Kaplan SH, Greenfield S, Koyama T, Barocas DA. Patient-reported outcomes through 5 years for active surveillance, surgery, brachytherapy, or external beam radiation with or without androgen deprivation therapy for localized prostate cancer. JAMA. 2020;323:149–163. doi: 10.1001/jama.2019.20675. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13.Das M. Androgen deprivation therapy for prostate cancer. Lancet Oncol. 2017;18:e567. doi: 10.1016/S1470-2045(17)30670-8. [DOI] [PubMed] [Google Scholar]
- 14.Cook GJ, Azad G, Padhani AR. Bone imaging in prostate cancer: the evolving roles of nuclear medicine and radiology. Clin Transl Imaging. 2016;4:439–447. doi: 10.1007/s40336-016-0196-5. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15.National Cancer Center; Prostate Cancer Expert Committee of National Cancer Quality Control Center. Quality control index for standardized diagnosis and treatment of prostate cancer in China (2022 edition) Zhonghua Zhong Liu Za Zhi. 2022;44:1011–1016. doi: 10.3760/cma.j.cn112152-20220803-00530. [DOI] [PubMed] [Google Scholar]
- 16.Nadamuni M, D’Amico AV, Donovan JL, Hamdy FC. Decision making in prostate cancer. N Engl J Med. 2023;389:1335–1338. doi: 10.1056/NEJMclde2307619. [DOI] [PubMed] [Google Scholar]
- 17.Ni X, Luu M, Ma W, Zhang T, Wei Y, Freedland SJ, Ye D, Daskivich TJ, Zhu Y. Impact of Asian race on prognosis in de novo metastatic prostate cancer. J Natl Compr Canc Netw. 2023;21:733–741. e3. doi: 10.6004/jnccn.2023.7027. [DOI] [PubMed] [Google Scholar]
- 18.Chen Y, Chen Z, Mo J, Pang M, Chen Z, Feng F, Xie P, Yang B. Identification of HCG18 and MCM3AP-AS1 that associate with bone metastasis, poor prognosis and increased abundance of M2 macrophage infiltration in prostate cancer. Technol Cancer Res Treat. 2021;20:1533033821990064. doi: 10.1177/1533033821990064. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19.Ding P, Guo H, Sun C, Yang P, Kim NH, Tian Y, Liu Y, Liu P, Li Y, Zhao Q. Combined systemic immune-inflammatory index (SII) and prognostic nutritional index (PNI) predicts chemotherapy response and prognosis in locally advanced gastric cancer patients receiving neoadjuvant chemotherapy with PD-1 antibody sintilimab and XELOX: a prospective study. BMC Gastroenterol. 2022;22:121. doi: 10.1186/s12876-022-02199-9. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20.van der Slot MA, Remmers S, Kweldam CF, den Bakker MA, Nieboer D, Busstra MB, Gan M, Klaver S, Rietbergen JBW, van Leenders GJLH Anser Prostate Cancer Network. Biopsy prostate cancer perineural invasion and tumour load are associated with positive posterolateral margins at radical prostatectomy: implications for planning of nerve-sparing surgery. Histopathology. 2023;83:348–356. doi: 10.1111/his.14934. [DOI] [PubMed] [Google Scholar]
- 21.Niu Y, Förster S, Muders M. The role of perineural invasion in prostate cancer and its prognostic significance. Cancers (Basel) 2022;14:4065. doi: 10.3390/cancers14174065. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22.Reeves FA, Battye S, Roth H, Peters JS, Hovens C, Costello AJ, Corcoran NM. Prostatic nerve subtypes independently predict biochemical recurrence in prostate cancer. J Clin Neurosci. 2019;63:213–219. doi: 10.1016/j.jocn.2019.01.052. [DOI] [PubMed] [Google Scholar]
- 23.Peng Z, Wang Y, Wu X, Yang S, Du X, Xu X, Hu C, Liu W, Zhu Y, Dong B, Pan J, Bao Q, Qian K, Dong L, Xue W. Identifying high gleason score prostate cancer by prostate fluid metabolic fingerprint-based multi-modal recognition. Small Methods. 2024;8:e2301684. doi: 10.1002/smtd.202301684. [DOI] [PubMed] [Google Scholar]
- 24.Bulten W, Kartasalo K, Chen PC, Ström P, Pinckaers H, Nagpal K, Cai Y, Steiner DF, van Boven H, Vink R, Hulsbergen-van de Kaa C, van der Laak J, Amin MB, Evans AJ, van der Kwast T, Allan R, Humphrey PA, Grönberg H, Samaratunga H, Delahunt B, Tsuzuki T, Häkkinen T, Egevad L, Demkin M, Dane S, Tan F, Valkonen M, Corrado GS, Peng L, Mermel CH, Ruusuvuori P, Litjens G, Eklund M PANDA challenge consortium. Artificial intelligence for diagnosis and Gleason grading of prostate cancer: the PANDA challenge. Nat Med. 2022;28:154–163. doi: 10.1038/s41591-021-01620-2. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 25.Yang DD, Mahal BA, Muralidhar V, Martin NE, Orio PF, Mouw KW, King MT, Choueiri TK, Trinh QD, Hoffman KE, Spratt DE, Feng FY, Nguyen PL. Androgen deprivation therapy and overall survival for gleason 8 versus gleason 9-10 prostate cancer. Eur Urol. 2019;75:35–41. doi: 10.1016/j.eururo.2018.08.033. [DOI] [PubMed] [Google Scholar]
- 26.Qin XP, Lu QJ, Yang CH, Wang J, Chen JF, Liu K, Chen X, Zhou J, Pan YH, Li YH, Ren SC, Liu JM, Liu WP, Qian HJ, Yi XL, Lai CY, Qu LJ, Gao X, Xu YS, Chen Z, Zhuo YM. CRMP4 CpG hypermethylation predicts upgrading to gleason score ≥8 in prostate cancer. Front Oncol. 2022;12:840950. doi: 10.3389/fonc.2022.840950. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 27.Ham WS, Chalfin HJ, Feng Z, Trock BJ, Epstein JI, Cheung C, Humphreys E, Partin AW, Han M. New prostate cancer grading system predicts long-term survival following surgery for gleason score 8-10 prostate cancer. Eur Urol. 2017;71:907–912. doi: 10.1016/j.eururo.2016.11.006. [DOI] [PubMed] [Google Scholar]
- 28.Akan S, Güner ND, Ediz C, Şahin A, Verit A. Is PSA still the best predictor for biochemical recurrence after radical prostatectomy in high-risk prostate cancer? J Invest Surg. 2022;35:1733–1738. doi: 10.1080/08941939.2022.2101165. [DOI] [PubMed] [Google Scholar]
- 29.Schlack K, Krabbe LM, Rahbar K, Isenberg K, Semjonow A, Schrader AJ, Boegemann M. ALP bouncing and LDH normalization in bone metastatic castration-resistant prostate cancer patients under therapy with Enzalutamide: an exploratory analysis. Transl Androl Urol. 2021;10:3986–3999. doi: 10.21037/tau-20-1117. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 30.Poteska R, Rahbar K, Semjonow A, Schrader AJ, Boegemann M, Schlack K. The prognostic potential of alkaline phosphatase and lactic acid dehydrogenase in bmCRPC patients without significant PSA response under enzalutamide. BMC Cancer. 2022;22:375. doi: 10.1186/s12885-022-09483-7. [DOI] [PMC free article] [PubMed] [Google Scholar]