Risk factors of PSA progression and overall survival in patients with localized and locally advanced prostate cancer treated with primary androgen deprivation therapy

Background Primary androgen deprivation therapy (PADT) has played an important role in the treatment of prostate cancer. We sought to identify factors of PSA progression in our series of patients with localized and locally advanced prostate cancer treated with PADT. Methods Six-hundred forty-nine patients with localized and locally advanced prostate cancer who received PADT from 1998 to 2005 by Nara Uro-Oncology Research Group were enrolled. Age, T classification, stage, PSA level at diagnosis, Gleason score, laterality of cancer detected by biopsy and seminal vesicle involvement (SVI) were adopted as parameters of PSA progression. Cox’s proportional hazards model was used to determine the predictive factors for PSA progression. Results The median follow-up period and the median PSA level at diagnosis were 49 months and 15 ng/mL. The 5-year disease specific survival rate, overall survival rate and PSA progression-free survival (PFS) rate were 97.9 %, 91.9 % and 71.2 %, respectively. The univariate analysis showed that the PSA level at diagnosis, Gleason score, laterality of cancer detected by biopsy and SVI were independent predictive parameters of PSA-PFS. However, by multivariate analysis, only laterality of cancer detected by biopsy (unilateral vs. bilateral) was an independent predictive parameter of PSA-PFS (p = 0.034). The patients were classified into new risk groups base on three factors: PSA level at diagnosis, Gleason score, and laterality of cancer detected by biopsy. The PSA-PFS rates at 5-years in the low- (none or one factor), intermediate- (two factors) and high-risk (three factors) groups were 78.2 %, 62.5 % and 46.9 % (p < 0.001), respectively. Conclusion In localized or locally advanced prostate cancer patients who received PADT, laterality of cancer detected by biopsy was a significant predictor associated with a longer PSA-PFS. Our new risk grouping indicates the usefulness of PSA-PFS.

combination with castration was developed, and was shown to facilitate stronger androgen suppression.
Widespread screening for prostate-specific antigen (PSA) has led to a significant increase in the detection of early stage, clinically localized prostate cancer. Currently, treatment of localized prostate cancer remains controversial. In the US it is frowned upon to give ADT for localized disease. The CaPSURE data from the USA indicated that 44 % of patients underwent radical prostatectomy, 23 % received definitive radiotherapy and 20 % received primary androgen deprivation therapy (PADT) [2]. On the other hand, the Japan Prostate Cancer Study Group showed the corresponding figures were 39.5 %, 23.9 % and 28.0 %, respectively [3] and the figures from Nara Uro-Oncological Research Group (NUORG) were 40 %, 16 % and 38 %, respectively [4][5][6]. As background of the present study, several reasons why Japanese patients with localized and locally advanced prostate cancer hesitate to undergo radical prostatectomy and prefer to receive PADT are proposed. Firstly, all patients are completely covered by the public health insurance system in Japan [5]. Secondly, Japanese patients tolerate hormonal therapy well without severe side effects for a long time [7,8]. Thirdly, in those days radiotherapy was not widespread and doctors at hospitals where modalities for radiation therapy were not available usually chose PADT if the patients were unwilling to undergo radical prostatectomy [4,5]. Fourthly, in those days, 49.9 % of the patients with localized or locally advanced prostate cancer were considered as the D' Amico highrisk group. 5-year biochemical recurrence-free rate in the D' Amico high-risk group treated with prostatectomy estimated 46.3 % [9]. 51 % of the patients with localized or locally advanced prostate cancer received PADT [4].
Recently, ADT is used as the primary treatment for advanced prostate cancer, and the efficacy of PADT for localized or locally advanced prostate cancer has also been reported [10,11]. Mounting data on the efficacy and safety of ADT has brought about increased use of PADT in patients with localized or locally advanced prostate cancer in many countries, despite limited evidence to date on the impact on clinical outcomes [12][13][14].
We performed a retrospective study of the efficacy of PADT and identified risk factors for PSA progression in our series of patients with localized and locally advanced prostate cancer.

Methods
This study retrospectively evaluated 649 Japanese patients with localized and locally advanced prostate cancer who received PADT following diagnosis by the NUORG between January 1998 and December 2005. The diagnosis was based on prostate biopsy. Computed tomography, bone scans, magnetic resonance imaging and/or transrectal ultrasonography were used in all cases. These patients selected PADT for various reasons, including older age, patient's preference and comorbidity such as severe cardiovascular disease or other malignancies, although definitive therapy such as radical prostatectomy or irradiation is the standard treatment for patients with localized prostate cancer.
Follow-up data were retrieved from hospital medical records. Patients were followed every month for the first 3 months and every 3 months thereafter. PSA progression was defined as the first day when the PSA was increased for three consecutive times or when clear clinical radiological evidence of progressive disease was seen. PSA progression-free survival (PFS) rate was estimated by the Kaplan-Meier method and the log rank test was used to assess differences between groups: Age (≤75 vs. 76≤), T classification, stage (B vs. C), PSA level at diagnosis (<10 ng/mL vs. 10-20 ng/mL vs. 20 ng/ mL≤), Gleason score (6 vs. 7 vs. 8≤), laterality of cancer detected by biopsy (unilateral vs. bilateral) and seminal vesicle involvement (SVI; negative vs. positive). Based on the result of the log rank test, the Cox proportional hazards regression model was performed to analyze independent predictors of PSA progression.
We classified the patients into the modified D' Amico risk groups [15] and the Japan Cancer of the prostate Risk Assessment (J-CAPRA) risk groups [16]. The modified D' Amico risk grouping classifies patients into three risk groups based on PSA level at diagnosis and Gleason score: low-(PSA level at diagnosis ≤10 ng/mL and Gleason score ≤ 6; 112 patients), intermediate-(10 ng/mL < PSA level at diagnosis ≤20 ng/mL and/or Gleason score 7; 203 patients), and high-risk (PSA level at diagnosis >20 ng/ mL or 8 ≤ Gleason score; 334 patients). In J-CAPRA risk grouping, patients were assigned 1 point for Gleason score 7 and 2 points for Gleason score 8 to 10; 1 point for PSA level at diagnosis 20 to 100 ng/mL, 2 points for PSA 100 to 500 ng/mL, and 3 points for PSA higher than 500 ng/ mL; 1 point for stage T2c or T3a, 2 points for T3b, and 3 points for T4. Points for each variable are summed to yield a total score with a range of 0 to 12. The J-CAPRA score was also categorized to identify three groups at low-(0 to 2 points; 459 patients), intermediate-(3 to 7 points; 190 patients) and high-(8 to 12 points; 0 patient) risk of recurrence.
Statistical analysis was performed SPSS 11.0 J (SPSS Inc., Chicago, Illinois) and p < 0.05 was considered statistically significant. The product limit method of Kaplan-Meier was used to assess survival. The log-rank method was used to assess differences between groups. The Cox proportional hazards model was performed to analyze independent predictors of PSA-PFS. Only the variables that were found to be significant in the univariate analyses (p < 0.05) were entered into the multivariate analysis to determine the most significant factor for predicting disease outcome.
The Medical Ethics Committee of Nara Medical University approved this retrospective study.
We used a Cox's proportional hazards model to determine the predictive parameter of PSA progression. Based on the result of the log-rank test, age (≤75 vs. 76≤), stage (B vs. C), PSA level at diagnosis (<20 ng/mL vs. 20 ng/ mL≤), Gleason score (≤7 vs. 8≤), laterality of cancer detected by biopsy (unilateral vs. bilateral) and SVI (negative vs. positive) were adopted as clinicopathological parameters of PSA progression and T classification was excluded as a parameter in grouping the patients, because it was difficult to distinguish the cutoff point. PSA level at diagnosis, Gleason score, laterality of cancer detected by biopsy and SVI were the significant factors for a longer PSA-PFS. But, by multivariate analysis, only laterality of cancer detected by biopsy was an independent predictive parameter of PSA-PFS (Hazard ratio: 1.523, p = 0.034, 95 % confidence interval: 1.033-2.245) ( Table 3).
PSA-PFS rates at 5 years in low-, intermediate-and high-risk groups by the modified D' Amico risk grouping were 80.7 %, 78.5 % and 63.8 % (Fig. 4). A significant difference in PSA-PFS rate was observed between the intermediate-and high-risk groups (p = 0.003), but there was no difference between the low-and intermediaterisk groups (p = 0.493), as reported by Ueno et al. [14]. In the J-CAPRA risk grouping, PSA-PFS rates at 5 years in the low-and intermediate-risk groups were 78.3 % and 49.9 % (p < 0.001) (Fig. 5).
We classified the patients into our new risk groups based on three factors: PSA level at diagnosis (<20 ng/mL vs. 20 ng/mL≤), Gleason score (≤7 vs. 8≤) and laterality of cancer detected by biopsy (unilateral vs. bilateral). The low-(431 patients), intermediate-(153 patients) and highrisk (65 patients) groups included none or one, two and three factors, respectively. PSA-PFS rates at 5 years in the respective risk groups were 78.2 %, 62.5 % and 46.9 %, respectively, and a significant difference in the PSA-PFS rate was observed between groups by the log-rank test (p < 0.001) (Fig. 6). The Cox proportional hazards model showed the same result as the log-rank test (Table 4).

Discussion
Although PADT has been widely used for the treatment of prostate cancer at any early disease stage, there is not much information regarding the clinical outcomes associated with clinically localized and locally advanced prostate cancer treated by PADT. According to some reports, a survival advantage of CAB over castration monotherapy has been indicated [10,16,17]. Thus, the focus of the present study was placed on CAB rather than castration monotherapy as PADT to evaluate its efficacy in terms of long-term disease control of clinically localized and locally advanced prostate cancer.
At the present time, the younger patients with localized prostate cancer and locally advanced prostate cancer without complications have a tendency to select the radical treatment such as prostatectomy. But, in those days, the use of PADT was still common in patients with localized prostate cancer and locally advanced prostate cancer in Japan [4,6]. In many cases, the patients might select PADT by older age or some complications.
In this study, the PSA-PFS (71.2 % at 5 years) was similar to other previous reports [10,[16][17][18]. These results were worse than other treatment modalities such as prostatectomy and radiotherapy. If the PSA progression was defined as the day when the PSA at least 4 weeks later was 25 % increase over nadir with more than 2 ng/mL, the PSA-PFS might be better. The disease specific survival rate was very high (97.9 % at 5 years) even though 23.3 % of patients had stage C, suggesting a  [20].
On the other hand, there is growing evidence that ADT is associated with an increased risk of various comorbidities including ischemic heart disease, metabolic syndrome, glucose intolerance, and a decrease in bone mineral density [21][22][23][24]. As a result, patients who received PADT have worse overall survival compared with    conservative management [25,26]. In contrast, several reports have also shown no significant increase in cardiovascular mortality with ADT in men with prostate cancer [27][28][29]. Several parameters were isolated as predictors of PSA progression. Nadir PSA level and the percentage of positive biopsy cores remained as independent prognostic factors on multivariate analysis [18]. Younger patients (<70 years) and those with 6 ≤ Gleason score were at a higher risk of treatment failure [30]. Ueno et al. reported that PSA ≤20 ng/mL, Gleason score ≤7, and time to nadir PSA ≤6 months showed a good response to PADT  Fig. 4 Kaplan-Meier plots showing the incidence of PSA progression-free survival rate classified by D'Amico risk grouping [17]. In this study, there was no difference between Gleason score of 6 vs. 7 ≤ (p = 0.310), and we adopted a Gleason score ≤7 vs. 8 ≤ as a parameter of PSA progression in Cox's proportional hazards model. We found that PSA level at diagnosis, Gleason score, laterality of cancer detected by biopsy and SVI were significant factors for a longer PSA-PFS, except for age and stage by univariate analysis. Then, by multivariate analysis, only laterality of the cancer detected by biopsy was an independent predictive parameter of PSA-PFS. Firstly, we classified the patients using our four new risk groups: no, one, two and three factors. No significance difference was shown between the no factor  Our low-risk factor patients accounted for two-thirds of the T1c-T3b patients. For patients showing a good response to ADT, ADT showed an excellent effect in this study. This effect may be explained by the observation that resected specimens after neoadjuvant ADT were sometimes completely apoptotic. Kitagawa et al. analyzed the histological effects of ADT in specimens from patients treated with radical prostatectomy after neoadjuvant ADT [31]. They reported that histologically cured or nearly cured patients accounted for more than 40 % of the total number. In addition, the recurrence-free survival rate of patients with complete apoptosis was 100 %. These results supported our observation that some cases of localized prostate cancer could be cured by ADT alone. Schulman et al. also performed neoadjuvant ADT for 3 months before radical prostatectomy in patients with localized prostate cancer, and good histopathological effects [32].
In the modified D'Amico risk grouping, a significant difference in PSA-PFS rate was observed between the intermediate-and high-risk groups, but there was no difference between the low-and intermediate-risk groups. The J-CAPRA risk grouping included also metastatic cancer patients in addition to localized and locally advanced prostate cancer. Our new risk groups included only localized and locally advanced prostate cancer patients and a significant difference of PSA-PFS rate was observed between all groups. Our new risk grouping indicates the usefulness for localized and locally advanced prostate cancer patients treated with PADT.
There are several limitations to the current study. Firstly, there may be interobserver variation of the Gleason score between general pathologists and uropathologists. Secondly, the current study is retrospective and results should be interpreted accordingly.

Conclusions
Unilateral positive biopsy was a significant predictor associated with a longer PSA-PFS in localized or locally advanced prostate cancer patients who received PADT. Our new risk groups according to the three factors of PSA level at diagnosis, Gleason score and laterality of cancer detected by biopsy indicate the usefulness for PSA-PFS. The efficacy and toxicity of ADT for localized or locally advanced prostate cancer requires further study before it can be recommended as the primary treatment. In the future, a prospective randomized study or comparative study of QOL or medical cost compared with other treatments will be necessary to establish PADT as a recommended treatment for early prostate cancer. Our results provide potentially clinical useful predictive tools for physicians and patients contemplating PADT for localized or locally advanced prostate cancer as well as the outcomes necessary to design prospective studies of the treatment strategy.

Competing interests
The authors declare that they have no competing interests.
Authors' contributions AT contributed to analysis and interpretation of data and was involved in drafting the manuscript. TN contributed to conception and helped to draft the manuscript. MY, MM, SA and YC contributed to acquisition of data. EO and AH contributed to acquisition of data and helped to draft the manuscript. YH and KF conceived and supervised the study, helped to draft the manuscript and was involved in revising it critically for important intellectual content. All authors read and approved the final manuscript.