Journal of Cancer Research and Clinical Oncology

, Volume 140, Issue 8, pp 1413–1419

Outcomes and predictive factors of prostate cancer patients with extremely high prostate-specific antigen level

  • Kouji Izumi
  • Wen-Jye Lin
  • Hiroshi Miyamoto
  • Chiung-Kuei Huang
  • Aerken Maolake
  • Yasuhide Kitagawa
  • Yoshifumi Kadono
  • Hiroyuki Konaka
  • Atsushi Mizokami
  • Mikio Namiki
Original Article – Clinical Oncology

DOI: 10.1007/s00432-014-1681-8

Cite this article as:
Izumi, K., Lin, WJ., Miyamoto, H. et al. J Cancer Res Clin Oncol (2014) 140: 1413. doi:10.1007/s00432-014-1681-8

Abstract

Purpose

Prostate-specific antigen (PSA) is a useful biomarker of prostate cancer (PCa). High-risk localized PCa is defined using T stage, Gleason score (GS), and PSA. However, PSA level defining high-risk PCa is at most 20 ng/mL. In PCa patients with high PSA, it is unclear whether PSA itself can be a prognostic factor.

Methods

Of 642 patients who were diagnosed as PCa, 90 patients with PSA > 100 ng/mL were retrospectively analyzed. Patients were divided into three groups according to PSA level: very high (>1,000 ng/mL), moderately high (200–1,000 ng/mL), and slightly high (100–200 ng/mL).

Results

There were no significant differences in overall survival or PCa-specific survival (PCaSS) among the three groups. Regardless of PSA level, high M stage and GS significantly reduced PCaSS. When the risk classification was made using M stage and GS (high risk = M1 and GS ≥ 9, low risk = M0 and GS < 9, and intermediate risk = others), PCaSS was significantly different among high-, intermediate-, and low-risk groups with 5-year survival rates of 58.2, 80.6, and 100 %, respectively. Although there were no differences in treatment performed during the castration-resistant stage, patients undergoing alternative anti-androgen and zoledronic acid treatment had better PCaSS after being castration-resistant.

Conclusions

As PSA could not be a prognostic factor in PCa patients with high PSA > 100 ng/mL, the novel risk classification using M stage and GS may help clinicians to predict PCaSS and to plan follow-up schedules after diagnosis.

Keywords

Prostate-specific antigen Prostate cancer TNM classification Gleason score Castration-resistant prostate cancer 

Introduction

Prostate-specific antigen (PSA) is useful as a biomarker of disease progression and prognosis in prostate cancer (PCa) patients in addition to TNM classification and Gleason score (GS). High-risk localized PCa is defined using T stage, Gleason score, and PSA; however, PSA level used to define high-risk PCa is at most 20 ng/mL (D’Amico et al. 1998). Nomograms such as the Partin table also focus on PSA < 10 ng/mL for localized PCa (Naito et al. 2008; Partin et al. 2001). Meanwhile, little attention has been paid to patients with high PSA > 100 ng/mL regardless of localized or metastatic disease. PSA screening systems in Western countries, such as the USA and European countries, are well established because of the high prevalence and high mortality rate of PCa (Siegel et al. 2013; Ferlay et al. 2010; Andriole et al. 2009; Schroder et al. 2009). Therefore, PCa patients with high PSA > 100 ng/mL may rarely be found. However, around 25 % of PCa patients were found with PSA > 100 ng/mL in Japan (Cooperberg et al. 2009), and probably, more PCa patients would be found with high PSA in other Asian and developed countries. For clinicians, PSA is often the earliest information of PCa patients among PCa risk factors because GS is found with PCa diagnosis on biopsy specimens and TNM classification is determined with radiological examinations after diagnosis of PCa. In such PCa patients with high PSA, clinicians tend to predict poorer prognosis associated with higher PSA; however, it is still unclear whether PSA itself is actually a prognostic factor. As shown in Fig. 1, the PCa patient with a PSA level of 4182 ng/mL at diagnosis lived much longer than the patient with a PSA level of 108 ng/mL at diagnosis who died of PCa, in contrast to our initial expectations. In addition, it is unclear which types of treatment after being castration-resistant affect patient prognosis in men with PSA > 100 ng/mL at diagnosis. In this study, we examined whether PSA can predict patient prognosis in PCa patients with high PSA > 100 ng/mL and clarified factors and treatments that affect the survival of such PCa patients.
Fig. 1

Cases with PSA > 100 ng/mL. PSA at diagnosis in Case 1 and Case 2 was more than 4,000 and around 100 ng/mL, respectively. Generally, prognosis of Case 1 with higher T and N stage than Case 2 seemed to be worse than that of Case 2. However, survival time of Case 1 was much longer than that of Case 2. CAB combined androgen blockade, ZOL zoledronic acid, AA anti-androgen, EMP estramustine phosphate

Materials and methods

Of 642 patients who were diagnosed as having PCa at Kanazawa University Hospital between January 2000 and December 2010, 90 patients with PSA > 100 ng/mL were retrospectively analyzed with regard to their backgrounds, treatments, and survival. The patients were divided into three groups based on their PSA levels, i.e., very high (VH, >1,000 ng/mL), moderately high (MH, 200–1,000 ng/mL), and slightly high (SH, 100–200 ng/mL).

Overall survival (OS), PCa-specific survival (PCaSS), and castration-resistant PCa (CRPC)-free survival (CFS) were compared among these groups. Three patients for whom the date of being CRPC was unclear were removed from the analysis of CFS. In addition, other risk factors were also examined to determine their prognostic significance. In addition, to clarify the better treatment strategy for patients with high PSA, PCaSS after being CRPC was analyzed according to the treatment strategy. CRPC was defined as the status after three continuous elevations of PSA during androgen-deprivation therapy (ADT), including medical or surgical castration with or without anti-androgen treatment.

Statistical analyses were performed using the commercially available software Prism (GraphPad, San Diego, CA). Comparisons between three groups were performed by one-way ANOVA. Comparisons of tendencies among different groups were performed by χ2 test for trends. Comparisons of survival described as Kaplan–Meier curves were performed by log-rank test or log-rank test for trends. In all analyses, p < 0.05 was taken to indicate statistical significance.

Results

OS and PCaSS in overall patients

All patients underwent combined androgen blockade consisting of luteinizing hormone-releasing hormone analog and anti-androgen as primary treatment except one patient who was treated with surgical castration alone. OS and PCaSS in all patients are shown in Fig. 2, and relatively long survival was estimated; 5-year survival rates of OS and PCaSS were 72.2 and 79.1 %, respectively, and 10-year survival rates were 51.5 and 67.7 %, respectively.
Fig. 2

OS and PCaSS in overall patients. Five-year survival rates of OS and PCaSS were 72.2 and 79.1 %, respectively, and 10-year survival rates were 51.5 and 67.7 %, respectively

Patient background, OS, PCaSS, and CFS in each group according to PSA value

Patient backgrounds are shown in Table 1. The highest PSA value at diagnosis was 16,702 ng/mL. Although there were no differences in age or follow-up time among groups, T, N, and M stages and GS tended to be worse as PSA increased. However, there were no significant differences in OS, PCaSS, and CFS among VH, MH, and SH groups (Fig. 3).
Table 1

Patient backgrounds

 

VH

MH

SH

p

n

21

37

32

 

Median age (years)

73 (60–86)

75 (52–92)

74.5 (50–95)

0.9909

Median PSA (ng/mL)

3,349 (1,018–16,702)

281 (203–979)

138 (102–197)

<0.0001

T

 1, 2

2

8

5

0.0315

 3

5

15

19

 

 4

14

14

8

 

N

 0

4

20

16

0.0264

 1

17

17

16

 

M

 0

1

11

24

<0.0001

 1b

15

21

7

 

 1c

5

5

1

 

GS

 6, 7

2

9

15

0.0184

 8

5

12

3

 

 9, 10

14

16

14

 

 NA

1

0

0

 

Median FU, days

988 (10–4,165)

1,424 (29–4,218)

1,390 (7–4,612)

0.3409

CRPC dev

 Yes

11

22

13

0.2932

 No

10

15

19

 

Values in parentheses indicate range

VH very high, MH moderately high, SH slightly high, GS Gleason score, NA not available, FU follow-up, CRPC Dev castration-resistant prostate cancer development

Fig. 3

OS, PCaSS, and CFS in VH, MH, and SH. Based on PSA at diagnosis, patients were divided into very high (VH, >1,000 ng/mL, n = 21), moderately high (MH, 200–1,000 ng/mL, n = 37), and slightly high (SH, 100–200 ng/mL, n = 32) groups. However, there were no significant differences in OS, PCaSS, or CFS among VH, MH, and SH groups (p = 0.2381, p = 0.7742, and p = 0.1909, respectively)

OS and PCaSS according to T, N, and M stages and GS

As it was demonstrated that PSA could not be used to predict patient survival when PSA value was higher than 100 ng/mL, other factors were examined to determine whether they could be used as prognostic factors in this patient cohort. As shown in Fig. 4, there were no differences in OS or PCaSS among each T stage or N stage. On the other hand, there were significant differences in both OS and PCaSS among M0, M1b, and M1c stages (p = 0.0075 and p = 0.0048, respectively). GS could not predict OS, but there were significant differences in PCaSS among each GS group (i.e., GS ≤ 7, GS = 8, and GS ≥ 9, p = 0.0166).
Fig. 4

OS and PCaSS according to T, N, and M stage and GS. T, N, and M stage and GS were examined to determine whether they can be used as prognostic factors in patients with PSA > 100 ng/mL. There were no differences in OS or PCaSS among each T and N stage (p = 0.0535 and p = 0.2884 with T stage, respectively, and p = 0.2381 and p = 0.1530 with N stage, respectively). On the other hand, there were significant differences in both OS and PCaSS among each M stage (p = 0.0075 and p = 0.0048, respectively). GS could not predict OS (p = 0.1273). However, there were significant differences in PCaSS among each GS group (p = 0.0166)

Novel risk classification of PCa with PSA > 100 ng/mL

Based on the observations that high M stage and GS worsened PCaSS, a novel risk classification combining M stage and GS was developed to better predict PCaSS. High risk was defined as M1 and GS ≥ 9, low risk was defined as M0 and GS < 9, and intermediate risk was defined as neither high nor low risk. With the exception of one patient for whom the GS was not available, 89 patients were divided into these risk groups; the numbers of patients in the high-, intermediate-, and low-risk groups were 32, 36, and 21, respectively. As shown in Fig. 5, PCaSS was clearly distinguished according to risk, and this risk classification was a better predictor of PCaSS than using M stage or GS alone (p = 0.0039). Five-year survival rates in the high-, intermediate-, and low-risk groups were 58.2, 80.6, and 100 %, respectively, and 10-year survival rates in these groups were 29.1, 80.6, and 92.3 %, respectively.
Fig. 5

Novel risk classification of PCa with PSA > 100 ng/mL. High risk was defined as M1 and GS ≥ 9, low risk was defined as M0 and GS < 9, and intermediate risk was defined as neither high nor low risk. The numbers of patients in the high-, intermediate-, and low-risk groups were 32, 36, and 21, respectively. PCaSS was clearly distinguished according to risk, and this risk classification showed better prediction of PCaSS than using M stage or GS alone (p = 0.0039). Five-year survival rates in high-, intermediate-, and low-risk groups were 58.2, 80.6, and 100 %, respectively, and 10-year survival rates were 29.1, 80.6, and 92.3 %, respectively

Treatments performed and PCaSS after being CRPC

As there was no significant difference in CFS among VH, MH, and SH, treatments performed during CRPC stage were examined to determine whether they affect PCaSS after being CRPC. There were no significant differences in treatments performed among VH, MH, and SH groups (Table 2). As shown in Fig. 6, although docetaxel (DTX), estramustine phosphate (EMP), and ethinylestradiol (EE) did not improve PCaSS, patients who underwent alternative anti-androgen (AA) and zoledronic acid (ZOL) therapy had significantly better PCaSS after being CRPC than those who did not (p = 0.0088 and p = 0.0137). ZOL is usually administered not when PCa has become castration-resistant but when bone metastasis is found. Therefore, PCaSS from diagnosis was also analyzed; however, ZOL showed no improvement.
Table 2

Treatment performed in CRPC patients

 

VH

MH

SH

p

DTX

 Yes

3

7

2

0.5613

 No

8

15

11

 

ZOL

 Yes

7

15

4

0.0841

 No

4

7

9

 

EMP

 Yes

7

16

7

0.522

 No

4

6

6

 

EE

 Yes

3

9

2

0.2749

 No

8

13

11

 

AA

 Yes

10

19

11

0.8953

 No

1

3

2

 

Other abbreviations are the same as in Table 1

DTX docetaxel, ZOL zoledronate, EMP estramustine phosphate, EE ethinylestradiol, AA alternative anti-androgen

Fig. 6

Treatments performed and PCaSS after being CRPC. Although docetaxel (DTX), estramustine phosphate (EMP), and ethinylestradiol (EE) did not improve PCaSS (p = 0.9936, p = 0.3699, and p = 0.4685, respectively), patients who underwent alternative anti-androgen (AA) and zoledronic acid (ZOL) therapy had significantly better PCaSS after being CRPC than those who did not (p = 0.0088 and p = 0.0137). However, ZOL did not improve PCaSS from the time of diagnosis (p = 0.1092)

Discussion

As the number of PCa patient with PSA > 100 ng/mL at diagnosis may not be high in many Western countries (Cooperberg et al. 2009), there have been very few studies of the prognosis in patients with such high PSA. However, the number of patients with PSA > 100 mg/mL in other countries should not be neglected. In South Korea, PSA at diagnosis was more than 100 ng/mL in 18.3 % of PCa patients, and this is similar to the rate of 25 % for Japanese patients with primary ADT (Cooperberg et al. 2009; Jang and Kim 2012). Many patients with PSA > 100 ng/mL were found clinically; however, there have been no assessments of the relationship between PSA and prognosis in such cases. For example, as shown in Fig. 1, it is unclear whether there are any differences in background or prognosis between patients with PSA = 100 ng/mL and PSA = 4,000 ng/mL. Compared to previous reports of progression-free survival in patients classified according to clinical stage, CFS in the overall patient cohort in this study was similar to stage III patients and much better than stage IV patients, and OS in the overall patient cohort was between stage III and stage IV patients (Hinotsu et al. 2007). These results indicate that even if PSA at diagnosis is higher than 100 ng/mL, patients still have better survival than stage IV patients. Interestingly, when patients were divided into three groups according to PSA level, there were no differences in OS, PCaSS, or CFS among groups, indicating that PSA is no longer a prognostic factor at levels beyond 100 ng/mL. Although PSA level is important for prognosis in patients with PSA < 100 ng/mL, especially between 4 and 20 ng/mL, prognosis may reach a plateau in those with PSA > 100 ng/mL. This result is consistent with those of a previous study that showed that the hazard ratio of progression in patients with 100–500 ng/mL and > 500 ng/mL was similar, compared to patients with PSA < 20 ng/mL as a reference (Kitagawa et al. 2013). In contrast, both M stage and GS were found to be prognostic factors in PCaSS in this study. In the VH group, each T, N, and M stage and GS were significantly higher than those in the SH group. However, prognosis was not significantly different among VH, MH, and SH. This discrepancy may have been due to the variance in patient background. That is, patients with high M stage did not necessarily have high GS in the VH group. The novel risk classification based on the data in this study for patients with PSA > 100 ng/mL clearly distinguished PCaSS among risk groups and solved the problem that PSA is not useful as a prognostic marker. This risk classification was superior to each T, N, and M stages as well as GS for predicting PCaSS. Previously, the risk classification for patients undergoing primary ADT was developed as Japan Cancer of the Prostate Risk Assessment (J-CAPRA) by Cancer of the Prostate Strategic Urologic Research Endeavor (CaPSURE) and the Japan Study Group of Prostate Cancer (J-CaP) using TNM classification, GS, and PSA at diagnosis. Patients were scored from 0 to 12 by J-CAPRA score, and progression-free survival rates in each scored patient group were significantly different (Cooperberg et al. 2009). J-CAPRA score is very useful for predicting the outcome and is applicable to patients with both localized and advanced disease. However, scoring may be complicated, especially in T stage and GS. The risk classification in this study may be applicable as simplified for the specific PCa population with PSA > 100 ng/mL with a simple scoring system.

As the high PSA level may affect treatment outcome, the effects of treatment during CRPC on PCaSS were also investigated in this study. DTX was first shown to improve OS of CRPC patients with phase III study, and EMP, EE, AA, and ZOL were reported to contribute to better prognosis (Tannock et al. 2004; Petrylak et al. 2004: Kojima et al. 2004; Ueno et al. 2013; Matsumoto et al. 2013; Izumi et al. 2010). DTX, EMP, and EE did not improve PCaSS after CRPC. In Japan, DTX was approved for PCa in 2008 and the patient number was relatively low, which may have influenced the outcome. Meanwhile, AA improved PCaSS after CRPC. PSA is downstream of androgen/AR signaling, and a high level of PSA may indicate highly activated androgen/AR signaling. AA using flutamide was reported to decrease the PSA level in 88 % of CRPC patients via the reduction of adrenal androgens (Narimoto et al. 2010). It is possible that this mechanism functioned well in such high PSA patients. However, this may be biased because only seven patients did not undergo AA due possibly to their poor condition and were moved to DTX treatment immediately. ZOL is administered for patient with bone metastases (M1b) for the reduction of skeletal-related events (Saad et al. 2004). Therefore, the patients given ZOL may be regarded as having more advanced disease. Nevertheless, ZOL improved PCaSS after CRPC. A previous study showed that ZOL improved progression-free survival in patients with GS ≥ 8 and suggested that it may have an effect in some advanced patients (Ueno et al. 2013). However, ZOL did not improve PCaSS from the time of diagnosis, and the possibility of bias should also be taken into consideration.

This study had a number of limitations. The small sample size may have prevented determination of the precise statistical significance of differences between groups. Larger prospective studies with longer follow-up periods and data from other ethnic backgrounds are needed to confirm our findings. Moreover, as a variety of anti-neoplastic treatments, including DTX, AA, EMP, EE, and others not described here for CRPC, were performed, some outcomes may not have been impacted only by M stage and GS or by AA and ZOL.

This is the first report showing the reduced meaning of PSA at diagnosis when the value is more than 100 ng/mL and the concise and useful risk classification of patients with PSA > 100 ng/mL. In addition, AA and ZOL may bring about better outcomes in CRPC patients with PSA > 100 ng/mL. These findings may help clinicians to predict PCaSS and to plan appropriate follow-up schedules after diagnosis.

Conflict of interest

None declared.

Copyright information

© Springer-Verlag Berlin Heidelberg 2014

Authors and Affiliations

  • Kouji Izumi
    • 1
  • Wen-Jye Lin
    • 2
  • Hiroshi Miyamoto
    • 3
  • Chiung-Kuei Huang
    • 4
  • Aerken Maolake
    • 1
  • Yasuhide Kitagawa
    • 1
  • Yoshifumi Kadono
    • 1
  • Hiroyuki Konaka
    • 1
  • Atsushi Mizokami
    • 1
  • Mikio Namiki
    • 1
  1. 1.Department of Integrative Cancer Therapy and Urology, Graduate School of Medical ScienceKanazawa UniversityKanazawaJapan
  2. 2.Immunology Research CenterNational Health Research InstitutesZhunan, Miaoli CountyTaiwan
  3. 3.Pathology and Urology, James Buchanan Brady Urological InstituteJohns Hopkins University School of MedicineBaltimoreUSA
  4. 4.Department of MedicineThe Warren Alpert Medical School of Brown UniversityProvidenceUSA