Survey
* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project
* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project
2010 THE AUTHORS; BJU INTERNATIONAL Urological Oncology 2010 BJU INTERNATIONAL LYMPH NODE DISSECTION IN PROSTATE CANCER WITHROW ET AL. BJUI Therapeutic value of lymph node dissection at radical prostatectomy: a population-based case-cohort study BJU INTERNATIONAL Diana R. Withrow, Julie M. DeGroot, D. Robert Siemens*† and Patti A. Groome Division of Cancer Care and Epidemiology, Queen’s Cancer Research Institute, Kingston, Canada, and Departments of *Anatomy and Cell Biology and †Urology, Queen’s University, Kingston, Canada Accepted for publication 2 July 2010 Study Type – Prognostic (case series) Level of Evidence 4 OBJECTIVE • To examine the association between the number of lymph nodes removed in pelvic lymphadenectomy and the risk of prostate cancer death, particularly in low to intermediate risk prostate cancer patients. PATIENTS AND METHODS • Data on a subset of patients from a population-based case-cohort study was used to assess the effect of lymph node removal on prostate cancer-specific mortality. • The subset included in this report were those 281 patients from the parent study who were treated with prostatectomy and had a pelvic lymph node dissection and for whom we had a record of the number of nodes removed (the sub-cohort) and 41 patients fitting the same criteria who died of their prostate cancer within 10 years (the cases). • Study variables included number of lymph nodes removed, lymph node status, age, pretreatment PSA, T category, Gleason score and use of hormonal therapy. • We ran a Cox proportional hazards regression analysis that accounted for the study design and allowed us to consider these patient and disease characteristics as potential confounders of the association of interest. © What’s known on the subject? and What does the study add? Pelvic lymph node dissection (PLND) at radical prostatectomy is typically used to determine the loco-regional extent of cancer, and to help determine the risk of recurrence and need for adjuvant therapy. Recent improvements in imaging and risk stratification have potentially reduced the prognostic value of PLND, especially in low-risk patients. Lymph node dissections have been found to be of therapeutic value in other cancers sites, however there is contrasting evidence for prostate cancer. Beyond its prognostic value, a therapeutic benefit of pelvic lymph node removal at radical prostatectomy may exist in prostate cancer patients whose nodes are non-cancerous (node-negative). Specifically, in node-negative patients we estimated a 5% reduction in risk of prostate cancer mortality with each additional node removed that was marginally statistically significant (HR: 0.95, 95% CI: 0.89–1.02). In node-positive patients, it may be too late in the disease course for lymph node dissection to provide therapeutic benefit. • In a secondary analysis, the results were stratified by nodal status. RESULTS • The crude hazard ratio (HR), which is a measure of relative risk, was not statistically significantly associated with a reduction in the risk of prostate cancer mortality as the number of lymph nodes removed at PLND increased (HR: 0.97, 95% CI: 0.91–1.03). • None of the variables considered as potential confounders had an impact on the crude HR. Using two cut points to categorize the number of lymph nodes removed, one at 4 or more removed and the other at 10 or more removed resulted in HRs indicating a risk reduction of 25% in both cases, although these results were not statistically significant. • When we analyzed the association by pathological nodal status, we observed a possible increase in risk in the node-positive group (HR: 1.10, 95% CI: 0.86, 1.42), while those with negative lymph nodes may have benefited from increasing numbers removed (HR 0.95, 95% CI: 0.89,1.02). CONCLUSION • The results of this study indicate a possible therapeutic benefit of lymph node removal in node negative patients. Future research should focus on gaining a better understanding of the biologic mechanisms of a possible therapeutic benefit of PLND, particularly for those lower risk patients with histologically negative lymph nodes. KEYWORDS prostate cancer, lymph node dissection, therapeutic effect, population-based, casecohort study, prostate cancer mortality 2010 THE AUTHORS BJU INTERNATIONAL © 2 0 1 0 B J U I N T E R N A T I O N A L | 1 0 8 , 2 0 9 – 2 1 6 | doi:10.1111/j.1464-410X.2010.09805.x 209 W I T H R O W ET AL. INTRODUCTION The main role of a pelvic lymph node dissection (PLND) at the time of radical prostatectomy is to determine the loco-regional extent of cancer, and to help determine the risk of recurrence and any possible need for adjuvant therapy [1,2]. However, the relative benefit and the overall utility of PLND in prostate cancer management remains debated [3]. The stage migration afforded by widespread PSA testing has led to a corresponding decrease in the incidence of lymph node involvement over the last three decades [4]. Accordingly, recent guidelines from professional associations and advocacy organizations have suggested that PLND should be limited to higher risk patients in order to prevent possible complications in those with a lower risk of lymph node involvement [5,6]. Beyond prognostic value, lymph node dissections have been found to be of therapeutic value in other urologic cancers [7], but the therapeutic role of PLND in prostate cancer remains controversial. In node positive patients, a greater number of nodes removed could result in more accurate staging, leading to earlier adjuvant treatments and therefore better survival outcomes [8]. Several anatomical studies have suggested that lymph node metastases may often occur outside a standard PLND and therefore extending the dissection and increasing the number of lymph nodes taken could amplify the potential benefit [9,10]. Interestingly, the therapeutic benefit of extended PLND in prostate cancer treatment has also been demonstrated in lymph node negative patients [11] suggesting the need to differentiate between biologically and histologically positive nodes in regard to prostate cancer progression. The goal of this study was to assess the association between number of lymph nodes removed at PLND and prostate cancer-specific mortality, particularly in low to intermediate risk prostate cancer patients. In a secondary analysis, we examined if and how this relationship differed between patients who were found to have histologically positive nodes and those who were not. PATIENTS AND METHODS STUDY POPULATION We identified a subset of patients who met our inclusion criteria from a larger 210 population-based case-cohort study that was conducted for other purposes [12]. The target population from which this parent study population was drawn consisted of all prostate cancer patients diagnosed in Ontario, Canada between January 1st 1990 and December 31st 1998 who were treated with curative intent with either surgery or external beam radiotherapy. ‘Curative treatment intent’ was defined as a radical prostatectomy and/or lymph node dissection within 7 months of diagnosis or external beam radiotherapy within 9 months. These patients were identified using the Cancer Care and Epidemiology (CCE) database at the Queen’s University Cancer Research Institute, which contains surgical procedure information from hospital discharge data and radiotherapy information from cancer clinic data all linked to the Ontario Cancer Registry (OCR) [13]. Patients in the parent study population were randomly chosen from the target population based on Cancer Care Ontario (CCO) region to represent geographically-defined subpopulations from across Ontario. As per the case-cohort design, this random sample made up the study sub-cohort for the parent study (n = 1703). The cases in the parent study consisted of 591 patients in the target population who died of prostate cancer within 10 years [12]. PROSTATE CANCER MORTALITY Cases were identified using the OCR, which stores cause and date of death from the Ontario Registrar General death certifications. As per the case-cohort design, the oversampling of this group increased study power. Survival of the sub-cohort was also monitored using OCR data verified by chart review whenever possible. The vital status was known for all patients as of December 31st 1999 for a maximum of 10 years follow up post diagnosis. PLND EXPOSURE The number of lymph nodes removed was determined from the chart review, specifically from the surgical and pathological reports of lymph node dissections done at the time of prostatectomy. In our secondary analysis, patients were stratified based on nodal status. The number of positive nodes identified was drawn from the same surgical and pathological reports. COVARIATES The subset from the parent study included in the current report was those patients who were treated with prostatectomy and had a pelvic lymph node dissection with information available as to the number of nodes removed. DATA SOURCES The CCE database provided us with information about the patient’s vital status, date and cause of death. All other data used in this report are from an extensive provincewide review of treating hospital and cancer centre charts conducted by trained abstractors according to a standardized protocol. Information collected for the parent study included patient demographics; disease characteristics at diagnosis; treatment details; co-morbid illnesses and disease recurrence. We also reviewed secondary hospital, urologist and general practitioner records when hospital and regional cancer centre charts did not contain all the desired information. Age at diagnosis, pre-treatment PSA, T category, Gleason score and use of hormone therapy or orchiectomy following PLND were all considered as potential confounders. PSA, T category and Gleason score predict survival in prostate cancer patients undergoing prostatectomy. Patients with a clinical stage of T1a or T1b were combined and T3 or T4 were combined due to small numbers of patients in each of those groups. Use of androgen deprivation therapy (ADT) was considered a potential confounder as the number of lymph nodes removed may dictate the likelihood of ADT, and its use is presumed to affect survival [8]. We also tested for an association between year of diagnosis and number of lymph nodes removed to examine the possibility of era effects over the 10 years during which data were collected. STATISTICAL ANALYSES Characteristics of those sub-cohort members from the parent study who were included vs excluded from the study were compared to assess whether the exclusion criteria introduced any selection biases. The mean age in the two groups was compared using an independent t-test for means while the PSA, T category, Gleason score and use of ADT were compared using Pearson’s Chi Square tests. © BJU INTERNATIONAL © 2010 THE AUTHORS 2010 BJU INTERNATIONAL LYMPH NODE DISSECTION IN PROSTATE CANCER FIG. 1. Derivation of study population from the parent study. Patients who died of PCa prior to December 31st, 1999 Ncase = 591 Parent Study Population Ntotal = 2213 Patients treated for cure by radical prostatectomy w/in 7 months of diagnosis ncase = 82 ncohort = 672 ntotal = 729 ncase= 76 ncohort = 539 ntotal = 601 ncase = 74 ncohort = 524 ntotal = 584 Stratified Random Sample of All Men Diagnosed with PCa between 1990-1998 ncohort = 1703 Did not undergo intraoperative PLND ncase= 6 ncohort = 133 ntotal = 128 Pre-operative PLND ncase = 2 ncohort = 15 ntotal = 17 Missing # of LN removed ncase = 31 ncohort = 238 ntotal = 265 ncase = 43 ncohort = 286 ntotal = 319 Adjuvant RT ncase = 2 ncohort = 5 ntotal = 6 Study Population ncase = 41 ncohort = 281 ntotal = 313 Because of the study design, demographics of the cases and sub-cohort were compared using the Cox proportional hazards model, which was also used to assess the study hypothesis [14]. Associations are reported as hazard ratios, which are an estimate of relative risk. The variance estimates, which determine the width of the hazard ratio confidence intervals, were adjusted for the case-cohort sampling and the area-level stratification was accounted for using an approach and SAS macro developed by Langholz and Jiao [15]. Each covariate was modeled along with the number of lymph nodes to check whether the covariate changed the lymph node effect (i.e. confounded the lymph node effect). A covariate was deemed to be a confounder if it caused a change in the lymph node hazard ratio that was greater than 10%. many lymph nodes were taken (i.e. the impact of taking 3 nodes vs 2 is as strong as the impact of taking 11 nodes vs 10). We tested this assumption by also calculating categorical LN effects and plotting them along with the continuous effect’s projected impact line to see if the categorical effects lay close to that line. Our main analysis used the continuous variable: number of lymph nodes removed. This results in a hazard ratio which assumes that the impact of having one more lymph node removed is the same no matter how RESULTS © In our secondary analysis, hazard ratios for prostate cancer mortality in N1 and N0 patients were examined to investigate whether the result was consistent across these groups. Mean number of lymph nodes removed in N1 and N0 patients was also compared. All analyses were performed using SPSS (Version 16.0.0, SPSS Inc. Chicago, IL) and SAS software (Version 9.1, SAS Institute Inc., Cary, NC). The steps undertaken to identify our study population from within the parent study are outlined in Fig. 1. From the parent study of a region-stratified random sample of patients diagnosed in Ontario between 1990 and 1998 and treated with curative intent (n = 2213), only those patients who underwent radical prostatectomy were eligible for this study (n = 729). After inclusion and exclusion criteria were applied to this group, the study population consisted of 281 sub-cohort patients and 41 cases (9 patients belonging to both groups) for a total of 313 subjects. The average follow-up in the sub-cohort was 4.7 years (SD = 2.4 years). Table 1 presents the characteristics of those sub-cohort members who were included (n = 281) and excluded (n = 381) from this study. Mean age was similar in the two groups. More patients in the excluded group had unknown PSA and T category, although the differences were marginally statistically significant (P = 0.07, P = 0.12 respectively) and differences in Gleason were also marginally significant with 69% of the excluded group having a Gleason between 5 and 7 vs 58% in the included group (P = 0.15). Use of ADT was higher in the included patients (P = 0.02). The sub-cohort and cases included in this study are described in Table 2. Because of the case-cohort design, statistical comparisons were only possible using Cox regression. Those results are presented in Table 3. Table 3 presents a consistent picture of more severe disease in the cases: 55% of the sub-cohort had PSA ≤10 compared to 32% of the cases (although PSA was unknown for 22% of the case group), 73% in the sub-cohort had T2a disease or lower compared to 59% of the cases, 66% of the sub-cohort had a Gleason score <7 compared to 51% of cases and many more cases had ADT and N1 disease. Figure 2 shows the distribution of lymph nodes removed in the cases and sub-cohort. The mean number of nodes removed in cases was 6.3 (SD = 4.5) while the mean in the subcohort was 7.8 (SD = 6.0). The number of lymph nodes removed in all patients ranged from 0 to 35 (one patient had a PLND with zero nodes recorded) with a median of 6 and only 10 patients with more than 20 nodes removed. The number of nodes removed was not statistically significantly associated with year of diagnosis (P = 0.44). Table 3 presents the HRs and 95% confidence intervals (CI) from a series of separate Cox proportional hazards regression analyses that compared the risk of prostate cancer death 2010 THE AUTHORS BJU INTERNATIONAL © 2010 BJU INTERNATIONAL 2 11 W I T H R O W ET AL. among groups defined by the variable listed. Results are presented for the main study variable, number of lymph nodes removed and for all of the study covariates: age, PSA, clinical T category, Gleason score, and use of ADT. The lymph node HR indicates that for every extra lymph node removed, the risk of prostate cancer death was reduced by 3%. This result is not statistically significant and the 95% confidence interval indicates that the true effect may be as much as a 9% decrease in risk or as much as a 3% increase in risk. As shown in Table 2 and statistically tested here, the cases were more likely to have missing PSA. Use of ADT, and a Gleason score >7 were associated with an increased risk of prostate cancer death. When the covariates were each added separately to the Cox model that contained the lymph node effect, the lymph node hazard ratio remained the same at 0.97 or 0.96 depending on the covariate, indicating that none of these covariates was confounding our number of lymph nodes removed effect. We categorized the number of lymph nodes removed using two separate cutoffs to see if there was a threshold effect of lymph node removal on survival. When the split was at ≥4 nodes, those patients’ hazard ratio compared to the rest indicated a non-statistically significant protective effect (HR 0.73, 95% CI: 0.38, 1.40). The result when the cut was at ≥10 nodes was the same (HR 0.74, 95% CI: 0.33, 1.62). As shown in Table 4, patients with positive lymph nodes had, on average, 3.8 more nodes removed than those with negative results (P = 0.08). Stratification based on pathological nodal status indicated, as one might expect, an increase in risk with more lymph nodes removed in the positive group, although this result was also not statistically significant (HR: 1.10, 95% CI: 0.86, 1.42). Those with negative lymph nodes may have benefited from a 5% risk reduction for each extra node removed but the confidence interval indicates that we cannot rule out a 2% increase in risk (HR 0.95, 95% CI: 0.89,1.02). The HRs we calculated for lymph node removal assume an equal effect on prostate cancer survival for each extra lymph node removed. We tested this assumption by plotting the curve generated by our continuous effect against the HRs generated when categories of number of lymph nodes removed (in intervals of three) were created 212 TABLE 1 Characterisics of included and excluded subcohort members Age, mean ± SD Pre-treatment PSA (%) ≤4 ng/ml 4–10 ng/ml 10–20 ng/ml >20 ng/ml Unknown T category (%) T1a/T1b T1c T2a T2b T3a/T3b/T4 Unknown Gleason Score (%) 2–4 5–6 7 8–10 Unknown Hormone therapy or orchiectomy Included (n = 281) n Mean or percent 281 63.0 ± 6.0 years Excluded (n = 381) n Mean or percent 381 63.1 ± 6.2 years 30 124 76 32 19 10.7 44.1 27.1 11.4 6.8 28 182 91 34 46 7.4 47.8 23.9 8.9 12.1 0.07 23 85 96 67 6 4 8.2 30.2 34.2 23.8 2.1 1.4 17 132 130 78 3 21 4.5 34.6 34.1 20.5 0.8 5.5 0.12 70 116 47 26 22 14 24.9 41.3 16.7 9.3 7.8 5.0 77 172 91 19 22 7 20.2 45.1 23.9 5.0 5.8 1.8 0.15 P value 0.86 0.02 TABLE 2 Characteristics of the subcohort and prostate cancer mortality cases. Nine patients belonged to both the case and subcohort groups and are included in both groups Age, mean ± SD Pre-treatment PSA (%) ≤4 ng/ml 4–10 ng/ml 10–20 ng/ml >20 ng/ml Unknown T category (%) T1a/T1b T1c T2a T2b T3a/T3b/T4 Unknown Gleason Score 2–4 5–6 7 8–10 Unknown Hormone therapy or orchiectomy Pathological N status N0 N1 Subcohort (n = 281) n Mean or percent 281 63.0 ± 6.0 years Cases (n = 41) n Mean or percent 37 64.0 ± 6.2 years 30 124 76 32 19 10.7 44.1 27.1 11.4 6.8 1 10 11 6 9 2.4 29.3 26.8 19.5 22.0 23 85 96 67 6 4 8.2 30.2 34.2 23.8 2.1 1.4 2 10 10 13 1 1 4.9 26.8 26.8 34.2 4.9 2.4 70 116 47 26 22 14 24.9 41.3 16.7 9.3 7.8 5.0 6 13 7 8 3 7 14.6 36.6 17.1 24.4 7.3 17.1 273 8 97.2 2.8 31 6 82.9 17.1 © BJU INTERNATIONAL © 2010 THE AUTHORS 2010 BJU INTERNATIONAL LYMPH NODE DISSECTION IN PROSTATE CANCER FIG. 2. Distribution of number of LNs removed during intra-operative PLND (ncase = 41, ncohort = 281). Nine patients belonged to both the case and sub-cohort groups and therefore are included twice. 25.00 Percent, % 20.00 15.00 10.00 Cases 5.00 Sub cohort 21 ≥ 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 0.00 Number of LN removed in PLND Variable Lymph Nodes Age PSA ≤4 ng/ml 4–10 ng/ml 10–20 ng/ml >20 ng/ml Unknown T category T1a/T1b T1c T2a T2b T3a/T3b/T4 Unknown Gleason Score 2–4 5–6 7 8–10 Unknown Hormone Treatment Hazard ratio 0.97 1.03 95% Confidence interval 0.91–1.03 0.98–1.10 1.0 0.39 1.74 2.09 4.73 – 0.05–2.82 0.73–4.17 0.86–5.06 2.01–11.11 0.69 1.0 0.69 1.50 1.48 1.96 0.14–3.33 – 0.30–1.59 0.71–3.19 0.38–5.80 0.21–17.84 1.0 1.64 2.06 4.12 1.45 4.70 – 0.64–4.19 0.73–5.85 1.55–10.94 0.35–5.91 2.14–10.30 TABLE 3 Crude hazard ratios for prostate cancer specific mortality across patient and disease characteristics Mean # of nodes removed (n) 7.7 (273) 11.5 (8)* for means 5.9 6.1 SD *Mean difference between N0 and N1: P = 0.08. © Given that our study population was diagnosed and treated throughout the 1990’s, we were interested to see how the treatment of our study’s sub-cohort differed, if at all, from what would occur in Ontario today. We used the 2008 Cancer Care Ontario surgical and pathological guidelines for prostate cancer management, which apply to patients with potentially curable prostate cancer for whom radical prostatectomy is the preferred treatment option [16]. To mimic this target group, we identified all surgically-treated members of the original parent study subcohort who had known PSA, Gleason and T category values (n = 567) and stratified them based on CCO risk category. According to the CCO guidelines, PLND is optional for those patients who are considered low risk. As shown in Table 5, the low risk patients made up the largest proportion of our sub-cohort (46%) and PLND was performed in 75% of these patients. In high and medium risk patients, approximately 85% and 93% underwent PLND respectively. DISCUSSION A dash in the confidence interval column indicates the referent group. Confidence intervals in bold type are statistically significant. TABLE 4 Mean number of LNs removed in N1 and N0 members of the subcohort and hazard ratios for prostate cancer mortality stratified by N1 and N0 Pathological node status N0 N1 and a separate HR calculated for each category. The reference category was 0–3 nodes removed. Figure 3 shows that the continuous effect curve is a conservative estimate of the categorical results up to the removal of 15 nodes. Hazard ratio 0.95 1.10 Confidence interval 0.89–1.02 0.86–1.42 The importance of the lymphatic system in cancer treatment is not a new idea. In 1908, Moynihan stated ‘the surgery of malignant disease is not the surgery of organs; it is the anatomy of the lymphatic system,’ and went on to explain that regional lymph node removal was necessary to cure [17]. For over a century, the role of lymphadenectomy has been disputed and despite a recent surge in research examining the possible therapeutic value of PLND in prostate cancer treatment, the benefit of the procedure remains hotly debated and controversial. Studies in colon and bladder cancers, among others, have found that both disease specific and overall survival were improved as more nodes were recovered and that this benefit was present across node positive and node negative patients [18,19]. With respect to prostate cancer, Joslyn et al. found that removal of ≥10 lymph nodes significantly decreased the risk of cancer specific mortality in over 9000 patients over a 10-year period (HR = 0.81, 95%CI: 0.70–0.94) [11]. 2010 THE AUTHORS BJU INTERNATIONAL © 2010 BJU INTERNATIONAL 213 W I T H R O W ET AL. Although our findings were not statistically significant, our study suggests a possible therapeutic value of lymph node removal in node negative patients. This result has not been consistent across publications and one might argue that our finding of a HR of 0.95 (95% CI: 0.89,1.02) is simply a product of bias and/or chance. To support the validity of our findings, some study strengths should be highlighted. All of the information used in this analysis was initially collected for other research purposes [12], so the data abstractors were not aware of the hypothesis, exposure or outcome. In addition, while many studies use biochemical recurrence as an outcome, prostate cancer mortality is of greater clinical relevance [21] and it seems there is a gap in the literature regarding the effect of a more extended lymph node dissection on prostate cancerspecific mortality [22]. The case-cohort design allowed us to study this rare outcome much more efficiently (i.e. with a smaller overall sample size) than what would have been possible in a conventional cohort study. Some limitations to the dataset should also be pointed out. Retrospective studies do not have the same control of data quality and completeness as a prospective study. In particular, we are not aware of any research exploring the accuracy of the number of lymph nodes removed. Also, we had to exclude 265 patients because no lymph nodes were removed at surgery. When we compared the known characteristics of those included and those excluded from this study, the biggest differences between the two groups had to do with completeness of the charts. This may have introduced a selection bias, as those who had records including this information may have received a higher quality of treatment overall. Cause of death 214 FIG. 3. Relationship between hazard ratio and number of LNs removed. In the continuous series, LNs removed were treated as a continuous variable and HRs were generated based on the model. The categorical series found HRs for LNs grouped by intervals of three. 2 1.5 Hazard Ratio Unfortunately, those investigators were not able to consider disease characteristics at diagnosis or examine the role of ADT due to lack of available data in their data source, the National Cancer Institute’s Surveillance, Epidemiology and End Results (SEER) database. In other research, a greater number of nodes removed has also been significantly correlated with lower relapse rates (23% and 8% in a standard PLND and more extended PLND respectively) [9] and freedom from biochemical recurrence (43% vs 10% for the more limited lymph node dissection, P = 0.01) [20]. 1 Continuous 0.5 Categorical 0 −1 1 3 5 7 9 11 13 15 17 21 19 −0.5 −1 Number of Nodes Removed TABLE 5 2008 Cancer Care Ontario Guidelines applied to the subset of the parent study subcohort that had surgery with or without lymph node dissection and could be assigned to a risk category (n = 567) Risk category Low Medium High Definition PSA <10, Gleason ≤6, and T1 or T2 PSA 10–20, and/or Gleason 7 PSA >20, Gleason ≥8, or ≥ T3 data is often inaccurate; however one study has found that death certificates for prostate cancer are in excellent agreement with clinical records [23] and therefore we are confident in our assigned cause of death. Despite the increased efficiency of the casecohort design, we were not able to rule out chance variation in our results. This study had 80% statistical power to detect a 6% risk reduction (3% was observed) in the overall analysis and a 7% risk reduction (5% was observed) in the node negative group. It is integral to our analysis to understand the determinants of number of lymph nodes removed. Not surprisingly, the greatest predictor of lymph node yield is extent of lymphadenectomy [24]. This may be determined by surgeon preference but may also be influenced by the perceived extent of disease. Joslyn et al. found that patients with lymph node involvement had a significantly greater number of nodes removed compared with those with negative nodes [11]. In this data set, a mean of 11.5 and 7.7 nodes were PLND recommendation Optional Recommended Mandatory Sub cohort Total Received PLND n n (%) 261 196 (75.1) 217 184 (84.8) 89 83 (93.3) removed in pathologically N1 and N0 patients respectively. This trend could explain the increased risk associated with a greater number of nodes removed in node positive patients in this study because surgeons who recognized more advanced disease upon lymphadenectomy may have been more likely to perform a more extensive resection. Those lymph node positive patients who had more lymph nodes removed therefore may have been inherently at greater risk of prostate cancer mortality. It is somewhat surprising that a few patients (n = 10) had more than 20 lymph nodes removed; however, given that this dataset includes several high volume centres, it is possible that some surgeons did routinely perform a thorough, extended dissection resulting in a large number of nodes. In the United States and likely elsewhere, the number of prostate cancer patients receiving PLND has dropped significantly over the period of our study and has continued to drop since [25]. The rates presented in Table 5 indicate that many patients included in this © BJU INTERNATIONAL © 2010 THE AUTHORS 2010 BJU INTERNATIONAL LYMPH NODE DISSECTION IN PROSTATE CANCER study because they had a PLND would not have been so treated had they been diagnosed recently. Few studies have examined the impact of PLND in low-risk individuals [3] but current guidelines suggest that low risk patients are least likely to benefit from lymphadenectomy [16]. Notably, our study suggesting a therapeutic benefit may not be reproducible today because of this reduced use of lymphadenectomy among those most likely to be node negative. Sigurdson suggests three possible mechanisms of benefit that assume improvements in survival can be attributed to increased lymph node yield in node negative patients [26]. The first is that the surgeon who performs a more thorough PLND provides better overall cancer care. This explanation could be examined with further research that controls for surgeon variation. The second theory is that a greater lymph node yield is associated with a more thorough examination by the pathologist, whereby counting is done more carefully, leading to improved detection of metastases and a decreased prevalence of false negatives. This could lead physicians to provide more aggressive adjuvant treatment and presumably improve survival. If this theory were true, however, one would expect to see a greater benefit in node positive patients whereas our results suggest an opposite effect. Finally, Sigurdson suggests that a more extensive lymphadenectomy which yields more nodes may lead to a decreased risk of local and regional recurrence [26]. A more extensive lymphadenectomy removes not only more histologically positive nodes [9], but also has the potential to remove a greater proportion of biologically positive nodes: those that are histologically unidentified but capable of continued metastatic progression [22,27,28]. Our study suggests that in node positive patients, it may be too late in the disease for lymph node dissection to provide any therapeutic benefit. With contemporary prostate cancer management, the vast majority of patients undergoing surgical management will be of low risk for lymph node positivity (97.2% of our sub-cohort was N0) and yet our results suggest these patients may benefit from a more extensive dissection. In general, complication rates for PLND range from 2%-51% [29] although it is uncertain whether taking a greater number of lymph nodes is associated with higher rates of complications [29–33]. © While prospective studies or randomized controlled trials may provide us with a clearer picture of the therapeutic benefit of lymph node dissections, the recognized prognostic value of the procedure on the one hand and potential complications on the other limit the feasibility of these studies. This retrospective study does not confirm a therapeutic benefit of more extensive lymph node removal, but supports the idea that a better understanding of the anatomical and biological mechanisms is necessary [22]. If a greater number of biologically positive lymph nodes removed confers a survival benefit by removing micrometastases, biologically positive nodes must be present for any curative advantage to exist. We must therefore make attempts to identify biologically positive nodes and in so doing gain a better understanding of the ratio of biologically to histologically positive nodes and of how the removal of each benefits survival. 3 4 5 6 7 8 ACKNOWLEDGEMENTS We would like to thank Karleen Schulze and Zhi Song for their programming and statistical support during the study design and data collection periods. Dr Groome is the Canada Research Chair in Cancer Care Evaluation. This research was undertaken, in part, thanks to funding from the Canada Research Chairs Program. We wish to acknowledge the Ontario Cancer Registry and Cancer Care Ontario for providing us with the Ontario registry data used in this study. This research was supported by the Canadian Cancer Society Research Institute with funds from the Canadian Cancer Society. 9 10 11 12 CONFLICT OF INTEREST None declared. 13 REFERENCES 14 1 2 Arcangeli G, Micheli A, Verna L et al. Prognostic impact of transurethral resection on patients irradiated for localized prostate cancer. Radiother Oncol 1995; 35: 123–8 Saitoh H, Hida M, Shimbo T, Nakamura K, Yamagata J, Satoh T. Metastatic patterns of prostatic cancer. Correlation between sites and number of organs involved. Cancer 1984; 54: 3078–84 15 16 Briganti A, Blute M, Eastham J et al. Pelvic lymph node dissection in prostate cancer. Eur Urol 2009; 55: 1251–65 Wagner M, Sokoloff M, Daneshmand S. The role of pelvic lymphadenectomy for prostate cancer – therapeutic? J Urol 2008; 179: 408–13 Aus G, Abbou CC, Bolla M et al. EAU guidelines on prostate cancer. Eur Urol 2005; 48: 546–51 Heidenreich A, Aus G, Bolla M et al. EAU guidelines on prostate cancer. Eur Urol 2008; 53: 68–80 Leissner J, Hohenfellner R, Thuroff JW, Wolf HK. Lymphadenectomy in patients with transitional cell carcinoma of the urinary bladder; significance for staging and prognosis. BJU Int 2000; 85: 817–23 Messing EM, Manola J, Sarosdy M, Wilding G, Crawford ED, Trump D. Immediate hormonal therapy compared with observation after radical prostatectomy and pelvic lymphadenectomy in men with nodepositive prostate cancer. N Engl J Med 1999; 341: 1781–8 Heidenreich A, Ohlmann CH, Polyakov S. Anatomical extent of pelvic lymphadenectomy in patients undergoing radical prostatectomy. Eur Urol 2007; 52: 29–37 Gil-Vernet J. Prostate cancer: anatomical and surgical considerations. BJU Int 1996; 78: 161–8 Joslyn SA, Konety BR. Impact of extent of lymphadenectomy on survival after radical prostatectomy for prostate cancer. Urology 2006; 68: 121–5 Groome PA, Rohland SL, Brundage MD, Heaton JP, Mackillop WJ, Siemens DR. The relative risk of early death from comorbid illness among prostate cancer patients receiving curative treatment. J Urol 2005; 173: 53 Mackillop WJ, Zhang-Salomons J, Groome PA, Paszat L, Holowaty E. Socioeconomic status and cancer survival in Ontario. J Clin Oncol 1997; 15: 1680–9 Cox D. Regression models and life-tables (with discussion). J R Stat Soc 1972; 34: 187–220 Langholz B, Jiao J. Computational methods for case-cohort studies. Comput Stat Data Anal 2007; 51: 3737–48 Chin J, Srigley J, Mayhew LA et al. Guideline for optimization of surgical and pathological quality performance for radical prostatectomy in prostate cancer management: surgical and pathological 2010 THE AUTHORS BJU INTERNATIONAL © 2010 BJU INTERNATIONAL 215 W I T H R O W ET AL. 17 18 19 20 21 22 guidelines. Can Urol Assoc J 2010; 4: 13– 25 Moynihan B. The surgical treatment of cancer of the sigmoid flexure and rectum. Surg Gynecol Obstet 1908; 6: 463 Berger A, Sigurdson E, LeVoyer T et al. Colon cancer survival is associated with decreasing ratio of metastatic to examined lymph nodes. J Clin Oncol 2005; 23: 8706–12 Herr HW, Bochner BH, Dalbagni G, Donat SM, Reuter VE, Bajorin DF. Impact of the number of lymph nodes retrieved on outcome in patients with muscle invasive bladder cancer. J Urol 2002; 167: 1295–8 Allaf ME, Palapattu GS, Trock BJ, Carter HB, Walsh PC. Anatomical extent of lymph node dissection: impact on men with clinically localized prostate cancer. J Urol 2004; 172: 1840–4 Ward JF, Blute ML, Slezak J, Bergstralh EJ, Zincke H. The long-term clinical impact of biochemical recurrence of prostate cancer 5 or more years after radical prostatectomy. J Urol 2003; 170: 1872–6 Klein E, Kattan M, Stephenson A, Vickers A. How many lymphadenectomies does it take to cure one patient? Eur Urol 2008; 53: 13–5 216 23 Penson DF, Albertsen PC, Nelson PS, Barry M, Stanford JL. Determining cause of death in prostate cancer: are death certificates valid? J Natl Cancer Inst 2001; 93: 1822–3 24 Bochner BH, Cho D, Herr HW, Donat M, Kattan MW, Dalbagni G. Prospectively packaged lymph node dissections with radical cystectomy: evaluation of node count variability and node mapping. J Urol 2004; 172: 1286– 90 25 Kawakami J, Meng MV, Sadetsky N, Latini DM, Duchane J, Carroll PR. Changing patterns of pelvic lymphadenectomy for prostate cancer: results from CaPSURE. J Urol 2006; 176: 1382–6 26 Sigurdson E. Lymph node dissection: is it diagnostic or therapeutic? J Clin Oncol 2003; 21: 965–7 27 Pagliarulo V, Hawes D, Brands FH et al. Detection of occult lymph node metastases in locally advanced nodenegative prostate cancer. J Clin Oncol 2006; 24: 2735–42 28 Miyake H, Hara I, Kurahashi T, Inoue TA, Eto H, Fujisawa M. Quantitative detection of micrometastases in pelvic lymph nodes in patients with clinically localized prostate cancer by real-time reverse transcriptase-PCR. Clin Cancer Res 2007; 13: 1192–7 29 Briganti A, Chun FK, Salonia A et al. Complications and other surgical outcomes associated with extended pelvic lymphadenectomy in men with localized prostate cancer. Eur Urol 2006; 50: 1006– 13 30 Clark T, Parekh DJ, Cookson MS et al. Randomized prospective evaluation of extended versus limited lymph node dissection in patients with clinically localized prostate cancer. J Urol 2003; 169: 145–8 31 Bader P, Burkhard FC, Markwalder R, Studer UE. Disease progression and survival of patients with positive lymph nodes after radical prostatectomy. Is there a chance of cure? J Urol 2003; 169: 849– 54 32 Heidenreich A, Varga Z, Von Knobloch R. Extended pelvic lymphadenectomy in patients undergoing radical prostatectomy: high incidence of lymph node metastasis. J Urol 2002; 167: 1681– 6 33 Metcalf KS, Peel KR. Lymphocele. Ann R Coll Surg Engl 1993; 75: 387–92 Correspondence: Patti A. Groome, PhD, Division of Cancer Care and Epidemiology, Queen’s Cancer Research Institute, 10 Stuart Street, Level 2, Kingston, Ontario, Canada K7L 3N. e-mail: [email protected] © BJU INTERNATIONAL © 2010 THE AUTHORS 2010 BJU INTERNATIONAL