Study Type – Therapy (case series)
Level of Evidence 4
Study Type – Therapy (case series)
To examine the long-term rates of biochemical recurrence (BCR)-free survival, cancer-specific mortality (CSM)-free survival, and overall survival (OS) in patients with prostate cancer treated with open radical prostatectomy (RP) in the prostate-specific antigen (PSA) era.
PATIENTS AND METHODS
The study comprised 436 patients who were treated with RP between 1992 and 1997 at our institution. None received adjuvant/salvage therapy in the absence of BCR. The BCR-free, CSM-free and OS rates
were defined using the Kaplan-Meier method. Multivariable Cox-regression models were used to test the effect of age, preoperative PSA level, neoadjuvant hormonal therapy, pT stage, lymph node status, RP Gleason sum and surgical margin status on BCR.
The median follow-up of censored patients was 122, 128, and 132 months for, respectively, BCR-free, CSM-free and OS estimates. The 10-year event-free survival rates for the same endpoints were 60%, 94% and 86%, respectively. Preoperative PSA level, RP Gleason sum, pT stage, lymph node status, and surgical margin status were independent predictors of BCR (all adjusted P < 0.05).
This study is the first to evaluate the long-term cancer control outcomes after RP from a European country in the PSA era. Our data indicate that RP provides excellent long-term survival rates in patients with clinically localized prostate cancer. Although ≈40% of patients have BCR after 10 years of follow-up, the CSM rate after 10 years is as low as 6%.
seminal vesicle invasion
Radical prostatectomy (RP) represents the most frequently chosen therapeutic option for treating newly diagnosed localized prostate cancer . Excellent long-term survival rates after RP have repeatedly been reported by several investigators [2–8]. However, virtually all studies evaluating long-term cancer control outcomes after RP stem from the USA. Moreover, several of these studies included patients who were treated before the implementation of PSA assays into routine clinical practice. Interestingly, in recent years several authors reported improved progression-free survival rates after RP in men treated in the PSA era [9,10]. However, due to the lack of long-term cancer control reports after RP from outside the USA, it is questionable whether long-term data derived from the USA and Europe are entirely comparable. This might especially be true in the light of reports suggesting that stage and grade migrations in patients diagnosed with localized prostate cancer and treated with RP affected the USA and Europe to a different extent .
To address this omission we evaluated the long-term cancer control outcomes of patients who were treated with RP for prostate cancer at one European institution in the PSA era, between 1992 and 1997. Specifically, we analysed long-
term biochemical recurrence (BCR)-free survival rates, cancer-specific mortality (CSM)-free survival rates and overall survival (OS) rates after RP.
PATIENTS AND METHODS
From January 1992 to December 1997, overall 673 patients diagnosed with clinically localized prostate cancer had open retropubic RP at our institution. Of these, 237 were excluded due to missing clinical variables (55), or because they were lost to follow-up (182); this resulted in 436 assessable patients. All RPs were performed by staff urologists. Most patients (417, 95.6%) had a standard lymph node dissection, comprising the obturator fossa, regardless of their respective risk profile for having lymph node metastasis. The whole RP specimen was processed using serial step-sections at 3 mm intervals according to the Stanford protocol . Tumours were graded according to the Gleason system , including the entire prostate with every tumour focus. For comparisons, pathological staging was assigned using the 2002 TNM system . Margins of resection were considered to be positive if one or more cancer cells extended to the inked surface. Patients with positive surgical margins but no extraprostatic extension (EPE) were considered to have organ-confined (OC) disease.
The patient and/or the referring urologist were contacted to record the postoperative medical history, including PSA levels, occurrence of BCR, particular chosen therapeutic option in the case of BCR, death, and the cause of death. BCR was defined as a PSA level of >0.1 ng/mL and increasing after RP. None of the patients had any adjuvant/salvage treatment after RP and before BCR. In the absence of BCR, PSA values were measured at least quarterly for the first year, followed by biannual measurements for the second year and annually from the third year onwards.
The BCR-free, CSM-free and OS rates were defined using the Kaplan-Meier method. Patients who did not develop the event of interest were censored at their respective last day of follow-up. Unknown causes of death in patients with known BCR were considered as CSM (five men). Subsequently, Kaplan-Meier curves were used to graphically illustrate the effect of pT stage (pT2 vs pT3a vs pT3b vs pT4), pN stage (pN0 vs pN1 vs pNx), RP Gleason sum (≤6 vs 3 + 4 vs 4 + 3 vs ≥8), and of the surgical margin status (positive vs negative) on BCR-free survival and CSM-free survival, respectively. Statistically significant differences between groups were assessed using the log-rank test.
Finally, multivariable Cox-regression models were used to identify independent predictors of BCR. Tested predictor variables consisted of pT stage, pN stage, RP Gleason sum, the surgical margin status, age at RP, the preoperative PSA level, and the presence or absence of neoadjuvant hormonal treatment. As the preoperative PSA level before the start of neoadjuvant therapy was unknown in some patients exposed to neoadjuvant therapy, all multivariable models were repeated after the exclusion of these men. All tests were two-sided with a significance level set at 0.05.
The descriptive characteristics of the study population are shown in Table 1. The median follow-up for censored patients was 122, 128 and 132 months for, respectively, BCR-free, CSM-free and OS analyses. The median (range) patient age at RP was 62 (44–75) years and the preoperative PSA level was 9.1 (0.12–120.0) ng/mL. Most patients did not receive neoadjuvant hormonal treatment (87.4%). Overall, 182 patients (41.7%) had OC disease at RP, 132 (30.3%) had EPE (pT3a), 103 (23.6%) had seminal vesicle invasion (SVI, pT3b), and 19 (4.4%) had pT4 disease. Most patients had no evidence of lymph node invasion (89.9%), but 25 (5.7%) had and 19 (4.4%) did not undergo lymphadenectomy (pNx). The Gleason sum of the RP specimens were as follows: 139 patients (31.9%) had a Gleason sum of ≤6, 202 (46.3%) of 3 + 4, 75 (17.2%) of 4 + 3 and 20 (4.6%) of ≥8.
|Variable||Mean (median, range) or n (%)|
|Age, years||61.5 (62, 44–75)|
|Preoperative PSA, ng/mL||13.5 (9.1, 0.12–120.0)|
|RP Gleason score|
|3 + 4||202 (46.3)|
|4 + 3||75 (17.2)|
|Overall +ve surgical margins||81 (18.6)|
|Follow-up (censored patients), months|
|BCR||115 (122, 1–191)|
|CSM||120 (128, 1–191)|
|OS||124 (132, 1–191)|
|Salvage therapy after BCR|
Figure 1 shows the Kaplan-Meier plots for BCR-free, CSM-free and OS. During the follow-up, 173 patients (39.7%) had a BCR; 72 (16.5%) died, of which CSM was recorded in 31 (7.1%) overall. The 5- and 10-year BCR-free survival rates for all patients were 68.4% and 59.9%, respectively. The CSM-free survival and OS estimates for the same time points were, respectively, 98.0% and 93.5% vs 95.0% and 86.0%.
Figure 2 shows the Kaplan-Meier plots of pT stage, pN stage, RP Gleason sum and surgical margin status on BCR-free survival. The 10-year BCR-free survival estimates according to pT stage were 87% for pT2 vs 53% for pT3a vs 28% for pT3b vs 6% for pT4 disease (log-rank pair-wise comparison over strata, all P < 0.001). Comparisons according to lymph node status showed a 10-year BCR-free survival of 78% for patients who did not have a lymphadenectomy vs 63% for pN0 patients vs 8% for patients with evidence of lymph node metastases. The difference between pN1 vs pN0 and pNx patients was statistically significant (all P < 0.001). Conversely, the difference between pNx and pN0 patients was not statistically significant (P = 0.2). Patients with a RP Gleason sum of ≤6 had a 10-year BCR-free survival of 90%, vs 58% for Gleason sum 3 + 4, 21% for Gleason sum 4 + 3 and 11% for Gleason sum ≥8. All these differences were statistically significant (all P < 0.001), except for the difference between Gleason sum 4 + 3 vs ≥8 (P = 0.5). Finally, patients with negative surgical margins had a 10-year BCR-free survival rate of 68% vs 24% for patients with positive surgical margins (P < 0.001).
Table 2 shows the results of the univariable and multivariable Cox-regression analyses predicting BCR. In addition to the univariable results from the Kaplan-Meier curves, PSA level was a statistically significant predictor of BCR (hazard ratio 1.03; P < 0.001). Moreover, PSA level, pT stage, pN stage, RP Gleason sum and the surgical margin status were independent predictors of BCR (all adjusted P = 0.04). No significant changes were recorded when patients were excluded in whom the PSA level before initiation of neoadjuvant hormonal therapy was unknown.
|Variable||HR (95% CI), P|
|Age||0.99 (0.97–1.01), 0.4||0.98 (0.96–1.01), 0.2|
|PSA level||1.03 (1.02–1.04), <0.001||1.01 (1.003–1.02), 0.008|
|Neoadjuvant hormonal treatment|
|Yes vs no||1.46 (0.97–2.20), 0.07||1.02 (0.66–1.57), 0.9|
|pT3a vs pT2||4.19 (2.63–6.69), <0.001||2.25 (1.32–3.82), 0.003|
|pT3b vs pT2||9.15 (5.77–14.5), <0.001||3.97 (2.30–6.84), <0.001|
|pT4 vs pT2||25.04 (13.34–47.01), <0.001||8.45 (3.99–17.92), <0.001|
|pN1 vs pN0||5.44 (3.45–8.59), <0.001||1.89 (1.13–3.16), 0.02|
|pNx vs pN0||0.50 (0.19–1.36), 0.2||1.43 (0.49–4.18), 0.5|
|RP Gleason sum||<0.001||<0.001|
|3 + 4 vs ≤6||4.38 (2.56–7.48), <0.001||2.88 (1.64–5.03), <0.001|
|4 + 3 vs ≤6||13.59 (7.76–23.79), <0.001||6.36 (3.51–11.55), <0.001|
|≥8 vs ≤6||17.87 (8.96–35.65), <0.001||4.65 (2.18–9.90), <0.001|
|Surgical margin status|
|Positive vs negative||3.50 (2.55–4.80), <0.001||1.88 (1.35–2.62), <0.001|
Figure 3 shows the Kaplan-Meier plots of pT stage, pN stage, RP Gleason sum and surgical margin status on CSM-free survival. The 10-year CSM-free survival rate for patients with pT2 disease was 98%, vs 96% for pT3a, 85% for pT3b and 72% for pT4. The differences in CSM-free survival between pT2, pT3a and pT3b were statistically significant (all P = 0.048), but the difference between pT3b and pT4 disease was not (P = 0.1). Patients with no evidence of lymph node metastases had a 10-year CSM-free survival rate of 95%, vs 68% for lymph node positive patients (P < 0.001). Stratification according to the RP Gleason sum gave 10-year CSM-free survival rates of, respectively, 100%, 94%, 83% and 80% for patients with a RP Gleason sum of ≤6 vs 3 + 4 vs 4 + 3 vs ≥8. The differences in the 10-year CSM rates between patients with Gleason sum ≤6 and all other strata was statistically significant, as was the difference between Gleason sum 3 + 4 and 4 + 3 and ≥8 (all P = 0.02). Conversely, the difference between Gleason sum 4 + 3 and ≥8 was not statistically significant (P = 1.0). Finally, the 10-year CSM-free survival rates according to surgical margin status were 99% for patients with negative and 81% for positive surgical margins (P < 0.001).
Of the 173 patients who had a BCR, 126 (72.8%) received salvage therapy. However, due to the diversity of different salvage therapy schemes (e.g. radiotherapy with/without hormonal treatment vs continuous hormonal therapy alone vs intermittent hormonal therapy alone), the effect of salvage therapy vs no salvage therapy on CSM-free survival was not assessed. Moreover, we did not use a multivariable analysis predicting CSM, as in the current analysis the effect of salvage treatment could not be sufficiently evaluated, as there were too few events (31), which significantly lowers the reliability of such an analysis because of the limited statistical power.
Within the last few decades the diagnosis and management of prostate cancer has changed significantly. With the widespread implementation of PSA testing into clinical practice, men are currently commonly diagnosed with earlier stage disease and when younger. Moreover, RP was substantially modified during the last 25 years [15,16] and is now considered a procedure providing high cancer control rates with low morbidity . Finally, results from a prospective randomized trial showed improved oncological outcomes in men treated with RP for localized prostate cancer and a long life-expectancy (age at randomization ≤65 years) compared to watchful waiting . Consequently, RP now represents the most frequently chosen treatment for localized prostate cancer in men with a life-expectancy of >10 years .
Although several authors reported the long-term cancer control outcomes after RP from the USA [2–8], long-term data from Europe in the PSA era are lacking. This prompted us to examine long-term cancer control rates after RP from a European tertiary-care referral centre. To ensure a long-term follow-up, we focused on patients who had RP ≥10 years ago.
Our overall BCR-free survival analyses showed a 10-year BCR-free survival rate of 60%. This is slightly lower than reported in studies from the USA, where reported 10-year BCR-free survival rates were 68–74% (Table 3) [2,6–9]. However, importantly, up to 58% of the present patients had evidence of EPE, which is substantially higher than in all North American studies. Moreover, most North American studies also included patients from before the PSA era, where tumour recurrences (local or metastatic) were due to the lack of a tumour marker to indicate early recurrence, usually detected later. Finally, none of the present patients received adjuvant or salvage treatment in the absence of BCR. Conversely, high-risk patients (e.g. patients with positive surgical margins and/or lymph node involvement) frequently received adjuvant treatment in the absence of BCR or were excluded from survival analyses in most North American studies [6–8]. These differences between the present and previous studies could explain our overall lower 10-year BCR-free survival rate. Moreover, we used a relatively low PSA threshold (0.1 ng/mL) to define BCR. Conversely, most North American studies used PSA levels of 0.2–0.4 ng/mL as the threshold indicating BCR. As some patients received salvage treatment at very low PSA levels (e.g. 0.1 ng/mL), unfortunately we could not assess the BCR rates using higher PSA thresholds, e.g. 0.2 ng/mL, which might further explain the more favourable recurrence-free survival rates reported in previous studies. However, when BCR-free survival rates were examined after stratification for stage and grade, there were excellent long-term BCR-free survival rates for patients with OC disease (87%) and for patients with low-grade disease (90%). These values are highly comparable to those reported from other series (Table 3). Notably, 28% of patients with SVI did not have a BCR after 10 years. This finding indicates that about a quarter of such patients are cured by RP alone and do not require additional treatment.
|Study||No. of patients||Year of RP||Definition of progression||Median follow-up, months||Survival rates, %|
|Present||436||1992–97||PSA >0.1 ng/mL||122 (BCR-free survival)||Overall 10-year 60||Overall 10-year 94|
|OC 87||OC 98|
|EPE 53||EPE 96|
|128 (CSM-free survival)||SVI 28||SVI 85|
|LNI 8||LNI 68|
|||3478||1983–2003||PSA >0.2 ng/mL or local recurrence or distant metastases||65||Overall 10-year 68||10-year 97 (nfs)|
|EPE 53–62 (depending on the surgical margin status)|
|||2404||1982–99||PSA >0.2 ng/mL or local recurrence or distant metastases||75.6 (mean)||10-year 74 (nfs)||10-year 96 (nfs)|
|||1000||1983–98||PSA >0.4 ng/mL or local recurrence or distant metastases||53.2||Overall 10-year 75||Overall 10-year 97.6|
|OC disease 92.2||OC 99.8|
|EPE 71.4||EPE 97.1|
|SVI 37.4||SVI 94.6|
|LNI 7.4||LNI 90.0|
|||752||1954–94||PSA >0.1 ng/mL or local recurrence or distant metastases||136.8||10-year: 71.2 (nfs)||10-year 95.5 (nfs)|
Our multivariable analysis identified preoperative PSA level, advanced pT stage, presence of lymph node metastases, high Gleason sum, and positive surgical margins as independent predictors of BCR. Interestingly, the absence of neoadjuvant hormonal therapy was not an independent predictor of BCR. This finding corroborates the results and conclusions from previous studies, that exposure to neoadjuvant androgen deprivation before RP does not improve cancer control and should therefore not be used [19,20].
We also examined the long-term CSM-free survival estimates. Despite the relative high proportion of patients with EPE or high-grade disease the overall 10-year CSM-free survival rate was 94%. This rate is an accordance with the findings from North American studies, where 10-year CSM-free survival rates were 90–97% (Table 3) [2,4,6–8]. As expected, the 10-year CSM-free survival rates were highest for patients with OC disease and men with low-grade tumours. Specifically, only 2% of patients with OC disease died from cancer-related causes within 10 years after RP, and there were no cancer-related deaths in patients with well differentiated disease (Gleason sum ≤6), which resulted in a CSM-free survival rate of 100% at 10 years. Possibly more interesting are the relatively high CSM-free survival rates of patients with unfavourable disease. Men with pT4 disease, highly aggressive cancers (Gleason score ≥8), or lymph node metastases, had 10-year CSM-free survival rates of 72%, 80% and 68%, respectively. This finding illustrates that 10 years after RP, more than a third of patients with lymph node metastases do not die from their disease. However, virtually all patients with such unfavourable RP findings received additional treatment in case of BCR. Consequently, these relatively high survival estimates are not due to RP alone, but are the combined effect of a multimodal approach. Notably, of the 31 men overall who died from cancer-related causes, most (28, 93%) developed BCR within 5 years after RP and only three (10%) had BCR at >5 years after RP. This finding indicates that only 10% of patients developing BCR at >5 years after RP will die from cancer-related causes within 10 years after RP.
It is tempting to compare our results to long-term results of alternative treatment schemes for prostate cancer, e.g. radiotherapy or active surveillance. However, because our study was retrospective and there was no comparison group, such comparisons are difficult to make and should be interpreted with caution. In any event, this study was not conducted to indicate the superiority of RP over other therapies but to generate authentic long-term data after RP, which can be used for patient counselling.
Our study has other limitations; e.g. it was from one centre, included relatively few patients and about a third of patients who were treated with RP during the study interval could not be included in the current analysis for different reasons (missing clinical variables or lost to follow-up). Moreover, the effect of salvage therapy on CSM-free survival could not be systematically assessed, due to the variety of chosen salvage treatments. Finally, contacting the treating physician and/or the patient for gathering follow-up information might introduce information bias. However, we found that in the vast majority of cases the patients were extremely well informed about their medical follow-up after RP and consequently we do not believe that information bias significantly affected our results.
Despite these limitations, our report is, to the best of our knowledge, the first from a European country to evaluate long-term cancer control outcomes after RP in the PSA era. Moreover, the median follow-up of our study substantially exceeds that of all previous reports from the PSA era.
In conclusion, our results indicate that RP provides excellent long-term cancer control rates in patients with clinically localized prostate cancer. At 10 years after RP, there are almost no cancer-related deaths in patients with OC or well-differentiated disease. Moreover, more than two-thirds of patients with highly unfavourable RP findings, such as lymph node metastases or pT4 disease, do not die from their disease within 10 years after RP.
This project was supported by the German Federal Ministry of Education and Science in the framework of the program for medical genome research (FKZ:01GS08189).
CONFLICT OF INTEREST