The extent of pelvic lymph node dissection correlates with the biochemical recurrence rate in patients with intermediate- and high-risk prostate cancer

Authors


Dr Riccardo Schiavina, Department of Urology, University of Bologna, S. Orsola-Malpighi Hospital, Palagi 9, 40134, Bologna, Italy. e-mail: rschiavina@yahoo.it

Abstract

Study Type – Therapy (case series)

Level of Evidence 4

What’s known on the subject? and What does the study add?

Pelvic lymph-node dissection during radical prostatectomy for prostate cancer is certainly a fundamental staging procedure but its therapeutic role is yet under debate.

This retrospective study suggests that, in patients with intermediate- and high-risk of prostate cancer, the greater the number of lymph-nodes removed, the lower the risk of biochemical relapse, even in the presence of 1 or 2 lymph-node metastasis. However, the Will Rogers phenomenon must be considered due to the retrospective nature of the present study.

OBJECTIVE

  • • To assess the impact of pelvic lymph node dissection (PLND) and of the number of lymph nodes (LNs) retrieved during radical prostatectomy (RP) on biochemical relapse (BCR) in pNX/0/1 patients with prostate cancer according to the clinical risk of lymph node invasion (LNI).

PATIENTS AND METHODS

  • • We evaluated 872 pT2-4 NX/0/1 consecutive patients submitted to RP between October 1995 and June 2009, with the following inclusion criteria: (i) a follow-up period ≥12 months; (ii) the avoidance of neoadjuvant hormonal therapy or adjuvant hormonal and/or adjuvant radiotherapy; (iii) the availability of complete follow-up data; (iv) no pathological T0 disease; (v) complete data regarding the clinical stage and Gleason score (Gs), the preoperative prostate-specific antigen (PSA) level and the pathological stage.
  • • The patients were stratified as having low risk (cT1a-T2a and cGs ≤6 and PSA level < 10 ng/mL), intermediate risk (cT2b-T2c or cGs = 7 or PSA level = 10–19.9) or high risk of LNI (cT3 or cGs = 8–10 or PSA level ≥ 20).
  • • The 872 patients were divided into two LN groups according to the number of LNs retrieved: group 1 had no LN or one to nine LNs removed; group 2 had 10 or more LNs.
  • • The variables analysed were LN group, age, PSA level, clinical and pathological stage and Gs, surgical margin status, LN status and number of LN metastases; the primary endpoint was the BCR-free survival.

RESULTS

  • • The mean follow-up was 55.8 months.
  • • Of all the patients, 305 (35%) were pNx and 567 (65.0%) were pN0/1.
  • • Of the 567 patients submitted to PLND, the mean number of LNs obtained was 10.9, and 49 (8.6%) were pN1.
  • • In the 402 patients at low risk of LNI, LN group was not a significant predictor of BCR at univariate analysis, while in the 470 patients at intermediate and high risk of LNI, patients with ≥10 LNs removed had a significantly lower BCR-free survival at univariate and multivariate analysis.

CONCLUSION

  • • In our study population, a more extensive PLND positively affects the BCR-free survival regardless of the nodal status in intermediate- and high-risk prostate cancer.
Abbreviations
BCR

biochemical relapse

e-PLND

extended pelvic lymph node dissection

Gs

Gleason score

HR

hazard ratio

LN

lymph node

LNI

lymph node invasion

LNM

lymph node metastasis

PCa

prostate cancer

PLND

pelvic lymph node dissection

R0

negative surgical margin

R1

positive surgical margin

RP

radical prostatectomy.

INTRODUCTION

The role of pelvic lymph node dissection (PLND) in patients with prostate cancer (PCa) continues to be a controversial topic [1]. PLND during radical prostatectomy (RP) is still the most accurate staging procedure in PCa and extended PLND (e-PLND) significantly improves the detection of lymph node metastasis (LNM) [2–5]. It has been suggested that patients at high risk of lymph node invasion (LNI) could potentially benefit from e-PLND in terms of oncological outcome because of the removal of small metastases, but the therapeutic role of PLND has not been definitively shown [6–10]. The aim of the present study was to assess the impact of PLND and of the number of LNs retrieved on biochemical relapse (BCR) in pNX/0/1 patients according to the clinical risk of LNI.

PATIENTS AND METHODS

We reviewed the Institutional Prostate Cancer Database of the patients treated with RP at the Department of Urology of the University of Bologna between 25 October 1995 and 30 June 2009. The inclusion criteria were: (i) a follow-up period ≥12 months; (ii) the avoidance of neoadjuvant hormonal therapy or adjuvant hormonal and/or adjuvant radiotherapy; and (iii) the availability of complete follow-up data. The exclusion criteria were: (i) pathological T0 disease; (ii) incomplete data regarding the clinical stage and Gleason score (Gs), the preoperative PSA level and the pathological stage. The remaining 872 pT2-4 NX/0/1 patients were evaluated.

Until 2001, standard PLND (including obturator fossa and external iliac vein LNs) or e-PLND (including obturator fossa, external LNs, internal iliac LNs, both medially and laterally, and distal 2 cm common iliac LNs) were offered to most of the patients undergoing RP at our department, no matter the risk of LNI. In all cases, the extent of the dissection varied essentially according to the preference and the skill of the surgeon. In recent years, the Kattan nomogram and the Briganti nomogram [11–13] have frequently been utilized to assess the risk of LNI, to identify the low-risk patients who were generally spared PLND. The specimens of the LNs were sent to the pathologist in separate packages in all cases. The number of LNs examined served as a surrogate of the extent of the PLND.

The RP specimens were routinely processed by dedicated genitourinary pathologists. The axial step sections were obtained at 3 to 4 mm intervals from the apex to the base. A tumour present at the inked margin was defined as a positive surgical margin (R1). After routine separation of each LN with acetone, the total number of LNs was assessed and all the LNs were analysed microscopically. LNs ≤4 mm were completely embedded in paraffin with one haematoxylin and eosin section per paraffin block, while LNs >4 mm were cut into 2-mm-thick slices; when necessary, an immunohistochemical analysis was carried out. In all cases, the number of LNs retrieved and the number of LNMs were reported. All specimens were staged using the 2002 American Joint Committee on Cancer TNM staging system [14,15].

The follow-up protocol consisted of PSA measurement and DRE every 3 months during the first year. When negative, these tests were done every 6 months during year 2; thereafter, PSA level was measured annually. Imaging studies were assessed when deemed necessary. All data were retrospectively obtained by clinical visits or telephone calls and were entered in the database by one of the authors.

The 872 patients were stratified into three clinical risk groups as follows: patients at low risk of LNI, defined as cT1a/b/c-T2a and cGs ≤6 and PSA level < 10 ng/mL (n= 402 46.1%); patients at intermediate risk of LNI, defined as cT2b-T2c or cGs = 7 or PSA level = 10–19.9 (n= 347, 39.8%) and patients at high risk of LNI, defined as cT3 or cGs = 8–10 or PSA level ≥ 20 (n= 123, 14.1%) [16].

The patients were divided into two LN groups according to the number of LNs retrieved at RP: group 1 (n= 573, 65.7%) had no LNs removed (n= 305) or had one to nine LNs removed (n= 268); while group 2 (n= 299, 34.3%) had ≥ 10 LNs.

Clinical and pathological data included LN group, age, PSA level at diagnosis, clinical and pathological Gs (divided into Gs ≤ 6, Gs = 7, Gs = 8–10), clinical and pathological stage, surgical margin status (R0 vs R1), LN status and number of LN metastases.

The primary endpoint was the BCR defined as PSA level greater than 0.2 ng/mL, followed by any increase above this level or the start of any therapy due to an increase in PSA.

The study was performed in line with the Helsinki Declaration and national regulations. All patients provided informed consent for participation and anonymous publication of the data.

The means, standard deviations, medians, ranges and frequencies were used as descriptive statistics. Fisher’s exact test, the linear by linear association, Pearson’s chi-squared and the Mann–Whitney U-test were used to compare the two groups of patients. The BCR-free survival rates were estimated by the Kaplan–Meier method. The univariate and multivariate forward-stepwise Cox proportional hazards regression was applied to analyse the survivals, and the hazard ratio (HR) was evaluated together with the 95% CI. All analyses were conducted using the SPSS software (version 13.0; SPSS, Inc., Chicago, IL, USA) and two-tailed P < 0.05 was considered to indicate statistical significance.

RESULTS

The mean (range) follow-up was 55.8 ± 36.8 (12–159) months. Among the 872 pNX/0/1 patients, the BCR-free survival rates were 74.9 and 58.7% at 5 and 10 years, respectively. BCR was observed in 180 (20.6%) patients. Figures 1 and 2 show the BCR-free survival rates according to the LN status (P= 0.000) and to the clinical risk group (P= 0.000).

Figure 1.

BCR-free survival in the 872 pNX/0/1 patients according to LN status.

Figure 2.

BCR-free survival in the 872 pNX/0/1 patients according to clinical risk-group.

Table 1 lists the clinical and pathological features of the 872 pNX/0/1 patients; of the 567 (65.0%) pN0/1 patients submitted to PLND, the mean (median, range) number of LNs obtained was 10.9 ± 6.4 (11.0, 1–36); in those with one to nine LNs removed (n= 268, 47.3%), the mean (median) number of LNs retrieved was 5.7 ± 6.3 (5.0), while in patients with ≥10 LNs removed (n= 299, 52.7%), the mean (median) number was 15.7 ± 5.1 (14.0). Of 49 pN1 patients, the mean (median, range) number of positive LNs was 2.3 ± 1.7 (2, 1–9).

Table 1.  Clinical and pathological characteristics of the 872 pNX/0/1 patients
 Variable (n= 872) (mean ±sd, n or n[%])
Age, years (872 valid cases)65.8 ± 5.8
PSA level, ng/mL (872 valid cases)10.2 ± 8.2
Clinical stage 
 T1a8 (0.9)
 T1b16 (1.8)
 T1c318 (36.5)
 T2503 (57.7)
 T3a24 (2.8)
 T3b3 (0.3)
 No. of valida cases872
Clinical Gs 
 ≤6689 (79.0)
 7144 (16.5)
 8–1039 (4.5)
 No. of valid cases872
Pathological stage 
 T2505 (58.1)
 T3a262 (30.0)
 T3b105 (12.0)
 No. of valid cases872
Pathological Gs 
 ≤6  511 (62.2)
 7269 (32.7)
 8–1042 (5.1)
 No. of valid cases822
Margin status 
 R0631 (72.4)
 R1241 (27.6)
 No. of valid cases872
Risk group 
 Low risk402 (46.1)
 Intermediate risk347 (39.8)
 High risk123 (14.1)
 No. of valid cases872
LN status 
 pNX305 (35.0)
 pN0/1567 (65.0)
 pN0518 (91.4)
 pN149 (8.6)
 No. of valid cases872
LN group 
 0–9 LNs removed573 (65.7)
 ≥10 LNs removed299 (34.3)
 No. of valid cases872
Mean number of LNs removed: 
 pN0/110.9 ± 6.4
 No. of valid cases567
Mean number of positive LNs: 
 pN12.3 ± 1.7
 No. of valid cases49

Table 2 lists the clinical and pathological features pertaining to the number of LNs removed in the 402 pNX/0/1 patients at low risk of LNI. The two LN groups were similar, except for the pathological stage and the LN status: patients with more LNs removed had a higher pathological stage (P= 0.025) and were more frequently pN1 (P < 0.001). Figure 3 shows the Kaplan-Meier curve for BCR-free survival with patients stratified by LN group in the 402 patients at low risk of LNI. The LN group was not a significant predictor of BCR at univariate analysis (HR = 0.828, 95% CI = 0.409–1.674; P= 0.599). At multivariate analysis, only the pathological stage and the pathological Gs showed an independent significant relationship with the BCR (Table 3).

Table 2.  Clinical and pathological characteristics of the 402 NX/0/1 low-risk patients according to LN group
 Variable [mean ± SD, n or n (%)]P
Overall(n= 402)0–9 LNs(n= 315)≥10 LNs(n= 87)
  • *

    Mann–Whitney U-test.

  • †Linear by linear association.

  • ‡Fisher exact test.

Age, years (402 valid cases)65.6 ± 5.665.6 ± 5.765.6 ± 5.20.831*
PSA level, ng/mL (402 valid cases)6.1 ± 2.06.1 ± 2.06.3 ± 2.10.735*
Clinical stage   0.446
 T1a4 (1.0%)4 (1.3%)0 (0%)
 T1b 11 (2.7%)8 (2.5%)3 (3.4%)
 T1c196 (48.8%)156 (49.5%)40 (46.0%)
 T2a191 (47.5%)147 (46.7%)44 (50.6%)
 No. of valida cases40231587
Pathological stage   0.025
 T2290 (72.1%)234 (74.3%)56 (64.4%)
 T3a89 (22.2%)67 (21.3%)22 (25.3%)
 T3b23 (5.7%)14 (4.4%)9 (10.3%)
 No. of valid cases40231587
Pathological Gs   0.064
 ≤6312 (80.0%)251 (81.5%)61 (74.4%)
 773 (18.7%)55 (17.9%)18 (22.0%)
 8–105 (1.3%)2 (0.6%)3 (3.6%)
 No. of valid cases39030882
Margin status   0.496
 R0294 (73.1%)233 (74.0%)61 (70.1%)
 R1108 (26.9%)82 (26.0%)26 (29.9%)
 No. of valid cases40231587
LN status:   <0.001
 pNX224 (55.7%)224 (71.1%)0 (0%)
 pN0/1178 (44.3%)91 (28.9%)87 (100.0%)
 pN0174 (97.8%)91 (100%)83 (95.4%)
 pN14 (2.2%)0 (0%)4 (4.6%)
 No. of valid cases40231587
Figure 3.

BCR-free survival in the 402 pNX/0/1 patients at low-risk of LNI according to LN-group.

Table 3.  Univariate and multivariate analyses for clinical and pathological characteristics correlated with BCR-free survival in the 402 pNX/0/1 low-risk patients
 UnivariateMultivariate
HR (95% CI)PHR (95% CI)P
  • *

    HR estimates the increasing risk related to the increase of one category of the independent variable.

LN groups (2 vs. 1)0.828 (0.409–1.674)0.599
Age, years1.003 (0.949–1.060)0.923
PSA level, ng/mL1.009 (0.878–1.160)0.896
Clinical stage*2.042 (1.092–3.821)0.025
Clinical Gs*
Pathological stage*2.281 (1.518–3.427)<0.0011.684 (1.095–2.591)0.018
Pathological Gs*3.370 (2.052 –5.534)<0.0012.734 (1.618 –4.621)<0.001
Margin status (R1 vs R0)2.418 (1.332–4.389)0.004
LN status*1.100 (0.802–1.508)0.554
Number of positive LNs1.319 (1.067–1.630)0.010

Table 4 lists the clinical and pathological features pertaining to the number of LNs removed in the 470 pNX/0/1 patients at intermediate and high risk of LNI. The two LN-groups were similar, except for the clinical Gs and pathological Gs and the LN status; the patients with more LNs removed had a higher clinical Gs (P= 0.003), a higher pathological Gs (P= 0.009) and were more frequently pN1 (P= 0.000). Figure 4 shows the Kaplan–Meier curve for BCR-free survival with patients stratified by LN group in the 470 patients at intermediate and high risk of LNI (P= 0.021). At univariate analysis, the LN group was a significant predictor of BCR (HR = 0.668, 95% CI = 0.471–0.947, P= 0.023). At multivariate analysis, the LN group, the clinical Gs, the pathological stage, the LN status and the number of positive LNs showed an independent significant relationship with the BCR (Table 5). Figure 5 shows the Kaplan–Meier curve for BCR-free survival with patients stratified by LN group in the 33 pN1 patients with ≤ two LN metastases (P= 0.014); the LN group was a significant predictor of BCR in both univariate and multivariate analyses (Table 6).

Table 4.  Clinical and pathological characteristics of the 470 NX/0/1 intermediate- and high-risk patients according to LN group
 Variable [mean ± SD, n or n (%)]P
Overall(n= 470)0–9 LN(n= 258)10 or more LN(n= 212)
  • *

    Mann-Whitney U-test.

  • †Linear by linear association.

  • ‡Fisher exact test

Age, years (470 valid cases)65.9 ± 5.966.3 ± 6.365.4 ± 5.30.653*
PSA level, ng/mL (470 valid cases)13.7 ± 9.813.6 ± 9.813.7 ± 9.90.235*
Clinical stage   0.120
 T1a4 (0.9)2 (0.8)2 (0.9)
 T1b5 (1.1)3 (1.2)2 (0.9)
 T1c122 (26.0)71 (27.5)51 (24.1)
 T2312 (66.3)172 (66.6)140 (66.0)
 T3a24 (5.1)9 (3.5)15 (7.1)
 T3b3 (0.6)1 (0.4)2 (0.9)
 No. of valida cases470258212
Clinical Gs   0.003
 ≤6287 (61.1)174 (67.2) 113 (53.6)
 7144 (30.6)68 (26.2)76 (36.0)
 8–1039 (8.3)17 (6.6)22 (10.4)
 No. of valid cases470259  211
Pathological stage   0.094
 T2215 (45.7)124 (47.9)91 (43.1)
 T3a173 (36.8)97 (37.4)76 (36.0)
 T3b82 (17.5)38 (14.7)44 (20.9)
 No. of valid cases470259  211
Pathological Gs   0.009
 ≤6199 (46.1)124 (51.2)75 (39.5)
 7196 (45.3)102 (42.1)94 (49.5)
 8–1037 (8.6)16 (6.7)21 (11.0)
 No. of valid cases432242190
Margin status   0.355
 R0337 (71.7)181 (69.9)156 (73.9)
 R1133 (26.9)78 (30.1)55 (26.1)
 No. of valid cases470259   211
LN status:   <0.001
 pNX81 (17.2)81 (31.3)0 (0)
 pN0/1389 (82.8)178 (68.7)   211 (100)
 pN0344 (88.4)163 (91.5)  181 (85.7)
 pN145 (11.6)15 (9.2)30 (14.2)
 No. of valid cases470259   211
Number of positive LNs   0.215
 ≤231 (68.9)12 (80)19 (63.3)
 >314 (31.1)3 (20)  11 (36.6)
 No. of valid cases451530
Figure 4.

BCR-free survival in the 470 pNX/0/1 patients at intermediate and high risk of LNI according to LN group.

Table 5.  Univariate and multivariate analyses for clinical and pathological characteristics correlated with BCR-free survival in the 402 pNX/0/1 intermediate- and high-risk patients
 UnivariateMultivariate
HR (95% CI)PHR (95% CI)P
  • *

    HR estimates the increasing risk related to the increase of one category of the independent variable.

LN groups (2 vs. 1)0.668 (0.471–0.947)0.0230.498 (0.329–0.754)0.001
Age, years1.010 (0.980–1.041)0.522
PSA level, ng/mL1.015 (1.001–1.028)0.032
Clinical stage*1.578 (1.045–2.174)0.005
Clinical Gs*1.729 (1.381–2.166)<0.0011.631 (1.262–2.108)<0.001
Pathological stage*2.091 (1.677–2.607)<0.0011.686 (1.284 –2.214)<0.001
Pathological Gs*1.982 (1.533–2.562)<0.001
Margin status (R1 vs R0)1.445 (1.010–2.067)0.044
LN status*1.524 (1.242 –1.871)<0.0011.337 (1.027–1.741)0.031
Number of positive LNs1.845 (1.623–2.098)<0.0011.529 (1.296–1.805)<0.001
Figure 5.

BCR-free survival in the 33 pN1 patients with ≤ two LN metastasis according to LN group.

Table 6.  Univariate and multivariate analyses for clinical and pathological characteristics correlated with BCR-free survival in the 33 pN1 intermediate- and high-risk patients with ≤ two LNMs
 UnivariateMultivariate
HR (95% CI)PHR (95% CI)P
  • *

    HR estimates the increasing risk related to the increase of one category of the independent variable.

LN groups (2 vs 1)0.322 (0.131–0.792)0.0140.330 (0.120–0.903)0.031
Age, years1.056 (0.975–1.144)0.178
PSA level, ng/mL1.019 (1.987–1.052)0.257
Clinical stage*0.737 (0.335 –1.619)0.447
Clinical Gs*1.103 (0.650 –1.872)0.717
Pathological stage*1.230 (0.576–2.625)0.593
Pathological Gs*1.817 (0.811–4.071)0.147
Margin status (R1 vs R0)0.854 (0.347–2.101)0.731
Number of positive LNs (2 vs 1)0.863 (0.343–2.171)0.754

DISCUSSION

Pelvic lymph node dissection remains the gold standard procedure for LN staging in PCa, especially in intermediate- and high-risk patients [1,4,5], but a beneficial effect of PLND on cancer outcome has not been shown with certainty.

In a study on 5038 patients, Masterson et al.[7] examined the association between the number of LNs removed, the number of positive LNs and cancer progression in patients with clinically localized PCa, and found better BCR-free survival with more LNs removed in node-negative patients (P < 0.01). In a retrospective trial of 13 020 cases treated with RP using the Surveillance, Epidemiology and End Results (SEER) database, patients undergoing excision of at least four LNs (both node-positive and node-negative patients) or more than 10 nodes (only node-negative patients) had lower cancer-specific death rate at 10 years with respect to pNX [8]. Allaf et al.[17] found better BCR-free survival for patients submitted to e-PLND than to the limited form of PLND (43 vs 10%) among patients with metastatic involvement in less than 15% of the extracted nodes (P < 0.01). Moreover, many studies in node-positive patients showed that those with low nodal involvement and treated with RP plus e-PLND can have excellent survival rates even in the absence of further adjuvant treatments: these data ultimately support a possible therapeutic effect of PLND [18–22].

Among those who believe PLND does not have a therapeutic role, Murphy et al.[9] reported that the number of LNs removed at RP did not increase the chance of cure for 964 patients with pT2–4 node-negative patients. Similarly, in a study by DiMarco et al.[10] on 7036 RPs, the extent of PLND did not appear to affect PCa outcome in the node-negative cases [10]. Interestingly, in a study by Bhatta-Dhar et al.[23] on 806 low-risk patients, the omission of the limited PLND did not adversely affect the biochemical relapse rates at 6 years. We agree with these authors that limited PLND does not have a therapeutic role in patients at low risk of LNI, but these results cannot be transferred to the intermediate- and high-risk population.

In the present study we evaluated the impact of clinical and pathological characteristics as well as the impact of PLND and of its extent on BCR in pNX/0/1 patients according to the clinical risk of LNI. In the low-risk patients, the number of LNs retrieved did not show any relationship with BCR. On the other hands, in the groups at intermediate and high risk of LNI, PLND with 10 or more LNs retrieved independently predicted a lower risk of BCR at multivariate analysis.

It must be underlined that the intermediate- and high-risk patients with more nodes removed had a better BCR-free survival despite the fact that these patients had a significantly poorer prognosis in terms of higher pathological Gs and stage and were more frequently pN1. In addition, in the patients with limited LNI, patients with more LNs removed had a higher BCR-free survival. In our opinion, the strength of the present study is represented by the fact that, as the decision to perform the PLND and its extent varied according to the preference and skill of the surgeon, we were able to compare many patients at high risk of LN invasion (and who were not submitted to PLND or only to a limited form of PLND) with those at the same risk who underwent a more extended PLND; we believe this differing conduct among the surgeons offered an optimal study population for a retrospective analysis.

In line with other studies [7,8,17], the results of the present study showed a positive role of PLND on cancer outcome. The existence of micrometastatic disease in node-negative patients has been shown in many studies performed with molecular and immunohistochemical techniques [24–29]. A meticulous dissection in our patients might thus have reduced the risk of BCR by removing the small nodal micrometastases missed at histopathology.

Among the study limitations, the principal limit of the present study is its retrospective nature; furthermore, no histopathological revision of the LN specimens was performed to assess the presence of micrometastasis. We could not report the incidence of the complications related to PLND precisely, because of the retrospective design of the study. In addition, we utilized the number of LNs retrieved as a surrogate of the extent of PLND; finally, the relationship between cancer-specific survival and BCR is not precisely defined.

In conclusion, in the population of the present study, a more extensive PLND positively affects the BCR-free survival in patients with PCa at intermediate and high risk of LNI. Even if patients with more nodes removed had a significantly poorer prognosis in terms of pathological stage and Gs, they had a better BCR-free survival. It is reasonable to support a potential therapeutic effect of a more extensive PLND during RP thanks to the removal of the micrometastasis. The assessment of the influence of PLND on survival needs to be more fully examined in trials with longer follow-up, and prospective trials comparing e-PLND with limited or no PLND are awaited.

CONFLICT OF INTEREST

None declared.

Ancillary