Retrospective validation of bone risk stratification criteria for men with de novo metastatic hormone-naive prostate cancer in China

Introduction

Improvements in public awareness of prostate cancer (PCa) and the popularisation of prostate-specific antigen (PSA) examinations have resulted in early detection of increasing numbers of PCa cases. However, many cases still show de novo metastatic PCa (mPCa) at the time of diagnosis. According to the American Cancer Association, approximately 34,500 patients will die due to PCa (mainly metastatic stage) in the United States in 2022 (Siegel et al., 2022). mPCa is incurable and the main treatment modality is androgen deprivation therapy (ADT) or ADT-based therepy (Klotz, 2008). However, the prognosis of patients treated with ADT varies considerably, indicating significant individual differences in the status of metastatic hormone-sensitive prostate cancer (mHSPC).

The most common metastatic site of PCa is the bone. The prognosis of bone metastases in different sites and the number of bone metastases varies greatly (Suzuki et al., 2021). Soloway et al. (1988) proposed an extent of disease (EOD) grading system for patients with mHSPC treated with ADT, which was based on the number of bone metastases detected on bone scans; in this system, the patients were classified into five groups according to tumor grade: Grade 0 = normal imaging findings or imaging appearance of benign bone disease; Grade 1 < 6 bone metastases, with each metastasis involving less than 50% of the vertebral body (bone lesions more than 50% of the vertebral body were counted as two lesions); Grade 2 = 6–20 bone metastases; Grade 3 = lesion size >20 but less than a “super scan”; and Grade 4 = “super scan” or involvement of more than 75% of the ribs and vertebrae. Some researchers proposed the extremity bone metastases model (EBM) based on the sites of metastasis (Crawford et al., 1989; Eisenberger et al., 1998). Gravis et al. (2016) proposed high- and low-volume disease (H/LVD) models in the CHAARTED trial, taking into account the site and number of bone metastases. HVD was defined by the presence of visceral metastases and/or four bone lesions with at least one lesion outside the vertebral column and/or pelvis; otherwise, patients were defined as having LVD. Fizazi et al. (2017) proposed the concepts of high-and low-risk disease (H/LRD) in the LATITUDE trial based on the site and number of metastases and the Gleason score. Patients with at least two of the following three risk factors associated with poor prognosis were defined as having HRD: a Gleason score of 8 or higher, at least three bone metastases, and visceral metastases. Other patients were defined as having LRD. However, the significance of these risk stratification models in patients with mHSPC receiving ADT-based therepy has not been fully assessed due to the differences in the study populations enrolled in trials.

ADT-based therapy is more effective in Asian patients with mHSPC (Fukagai et al., 2006; Suzuki et al., 2020; Bernard et al., 2017; Cooperberg et al., 2016). The number of patients diagnosed with PCa in China is increasing annually (Liu et al., 2019; Akamatsu et al., 2019), with many patients showing bone metastases at the time of diagnosis. Therefore, we aimed to validate and compare the effectiveness of the four bone-related risk stratification criteria for patients with mHSPC treated with ADT-based therapy, in order to provide clinicians with more accurate prognostic tools to help clinical decision-making.

Methods

Results

AUC analysis

An area under the curve (AUC) analysis of the four risk stratification systems was performed based on the time to CRPC and time to OS, and the findings of these analyses are illustrated in Figs. 1 and 2. Fig. 1 shows the CRPC-based comparison of the four risk stratification models (EOD, EBM, H/LVD, and H/LVD) on the basis of their CRPC AUCs. Compared to the three other models, EOD showed higher 1-, 2-, and 3-year CRPC AUC values. Fig. 2 provides a OS-based comparison. Similar to the results of CRPC, EOD also showed a higher 1-, 2-, and 3-year OS AUC value.

DCA analysis

The DCA curves for the four models were plotted according to the time to CRPC and OS (Fig. 3). As shown in Fig. 3A, the net benefit of the EOD model on OS was better than those of the other risk stratification models. As for CRPC, the net benefit of the EOD model was second only to that of the H/LRD model when the threshold was <0.5; when the threshold was >0.5, the EOD model outperformed the other models (Fig. 3B).

Discussion

In this study, we validated the four bone risk stratification criteria for men with de novo metastatic hormone-naive prostate cancer in China using AUC analysis and DCA. The results showed that the EOD model is more suitable than the H/LVD, H/LRD, and EBM models for Chinese patients with mHSPC treated with ADT-based therapy. The EOD model showed better performance in differentiation and a higher net benefit, and was an independent prognostic factor for CRPC and OS.

Racial differences are important consideration in the discussion of the reasons for the differences in prognosis between global and Chinese patients. Asian patients are more sensitive to hormone therapy in comparison with patients of other racial groups (Fukagai et al., 2006; Suzuki et al., 2020; Bernard et al., 2017; Cooperberg et al., 2016). Previous genomic analyses of PCa in China have revealed that the genomic alterations in Chinese patients are significantly different from those in Western patients (Li et al., 2020). ADT/ADT-based treatment is a “popular” treatment choice for metastatic prostate cancer. However, the prognosis of those patients varies considerably. In the LATITUDE trial, the addition of abiraterone acetate plus prednisone to ADT in patients with mHSPC resulted in a significant improvement in OS and progression-free survival in comparison with placebo and prednisone plus ADT (Fizazi et al., 2017). The CHAARTED trial showed that docetaxel systemic chemotherapy combined with ADT prolonged the OS of patients with HVD in comparison with ADT alone (Gravis et al., 2016). Totally, four bone risk stratification criteria are popularly used in clinical practice, i.e., H/LVD, H/LRD, EBM and EOD criteria. Studies are warranted to assess the clinical utility of the existing four risk stratification models for bone metastasis in patients with mHSPC. To the best of our knowledge, this was the first study to thoroughly validate and compare the prognostic significance of several bone risk stratification models based on different perspectives in Chinese patients.

Our study included 140 Chinese patients with mHSPC. The patients were followed up for relevant information on CRPC and OS. Comparative analysis of the AUC values of the H/LVD, H/LRD, EBM, and EOD models showed that the EOD model was significantly better in differentiation than the other models. Comparison of the DCA curves associated with OS for patients with mHSPC demonstrated that the EOD model had the highest net benefit. The DCA showed that H/LRD model had the highest net benefit when the threshold was <0.5, while the EOD model was second only to the H/LRD model. When the threshold was >0.5, the EOD model had the highest net benefit.

Bone metastatic burden is one of the important prognostic markers for patients with mHSPC (Akamatsu et al., 2019; Miyoshi et al., 2015). Several studies reported that the four models mentioned above were the independent risk factor affecting TTCRPC and OS (Narita et al., 2019; Kawahara et al., 2020). Similarly, our study comfirmed that EOD was an independent prognostic factor for TTCRPC and OS. However, of note, H/LVD, H/LRD and EBM were not the independent prognostic factors for OS in univariable and multivariable analyses. The HVD and patients with EBM group showed a shorter TTCRPC than the LVD and the patients without EBM group, respectively (Supplementary Material S1). One possible explanation is that these bone prognosis models may not be fully applicable to Chinese patients with mHSPC.

The biology of PCa is being studied worldwide and has potential clinical applications. However, the use of genetic markers as therapeutic targets is currently limited in clinical practice. Among clinical factors, the number and site of bone metastases, presence of visceral metastases, PSA level, testosterone level, and other treatments are considered important factors affecting the prognosis of patients with mHSPC (Tian et al., 2020; Hori et al., 2011). Similarly, in the univariable analysis, we found that Hb, ALP, Fib and EOD were statistically significant for OS (Table 2). In the multivariable analysis, EOD (≥2 v.s. <2; HR 3.353, 95% CI [1.361–8.259], P = 0.009) was still the independent factor influencing the OS. Collectively, EOD was a strong predictor for prognosis.

This study had several limitations. First, the retrospective nature of the study was an inherent limitation. Second, this study included data from a single centre and had a relatively small sample size. Therefore, multi-centre, large-sample data are required to confirm the findings of this study and to validate the accuracy of the model. Larger studies with longer follow-up periods are warranted to validate and compare the prognostic impact of different prognostic models. In addition, there is a need to further incorporate some of the newly developed drugs for ADT-based therepy, or to conduct a more detailed delineation of the suitability of the individual models for different drugs.

Conclusion

The EOD model yields reasonable risk factors for use in Chinese mHSPC patients, providing further evidence supporting clinical decision-making.

Supplemental Information