Download Therapeutic value of lymph node dissection at radical prostatectomy

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