Full text of DOI:10.1016/S0022-5347(05)67716-6

Evaluation Only. Created with Aspose.PDF. Copyright 2002-2021 Aspose Pty Ltd.
0022-5347/00/1634-1171/0
®
THE JOURNAL OF UROLOGY
Vol. 163, 1171–1177, April 2000
Copyright © 2000 by AMERICAN UROLOGICAL ASSOCIATION, INC.®
Printed in U.S.A.
IMPACT OF POSITIVE SURGICAL MARGINS ON PROSTATE CANCER
RECURRENCE AND THE USE OF SECONDARY CANCER TREATMENT:
DATA FROM THE CAPSURE DATABASE
GARY D. GROSSFELD, JAMES J. CHANG, JEANETTE M. BROERING, DAVE P. MILLER,
JENNY YU, SCOTT C. FLANDERS, JAMES M. HENNING, DAVID M. STIER
AND PETER R. CARROLL
From the Department of Urology, University of California San Francisco Urology Outcomes Research Group and University of California
San Francisco-Mount Zion Cancer Center, University of California San Francisco and Lewin-TAG, Inc., San Francisco, California, and
TAP Holdings, Inc., Deerfield, Illinois
ABSTRACT
Purpose: We determined the impact of positive surgical margins on prostate specific antigen
(PSA) recurrence and secondary treatment in patients who underwent radical prostatectomy as
definitive local treatment for prostate cancer.
Materials and Methods: We reviewed the pathology reports of 1,383 patients in the CaPSURE*
database, a longitudinal disease registry of men with prostate cancer, who underwent radical
prostatectomy as definitive local treatment. Pathological stage, Gleason score, and the number
and location of any positive surgical margins were determined in each patient. PSA recurrence
was defined as PSA 0.2 ng./ml. or greater on 2 consecutive occasions after radical prostatectomy.
Secondary cancer treatment consisted of radiation or androgen deprivation after radical prosta-
tectomy. Adjuvant and nonadjuvant secondary treatment was given within and more than 6
months after radical prostatectomy, respectively. Kaplan-Meier event rates of PSA recurrence
and secondary treatment were calculated for patients with positive and negative surgical mar-
gins. We performed multivariate Cox proportional hazards analysis to adjust for clinical differ-
ences in groups.
Results: Patients with positive surgical margins were significantly more likely to undergo
secondary adjuvant or nonadjuvant cancer treatment and have PSA recurrence than those with
negative margins. After adjusting for patient age, ethnicity, PSA at diagnosis, pathological stage
and Gleason score, surgical margin status was an important independent predictor of PSA
recurrence and secondary treatment (p ϭ 0.06 and 0.0011, respectively). The number of positive
margins and positive margin location had little impact on the outcomes measured.
Conclusions: These data indicate that surgical margin status is an independent predictor of
PSA recurrence and secondary cancer treatment in patients who underwent radical prostatec-
tomy as definitive local therapy for prostate cancer.
KEY WORDS: prostatic neoplasms, prostatectomy, prostate-specific antigen, prognosis
The primary objectives of radical prostatectomy include com- tracapsular disease extension and prostatectomy Gleason score,
plete excision of the local tumor burden and cure of the disease.
Cancer at the inked surgical resection margin of the radical
prostatectomy specimen suggests that local resection may have
been incomplete. Despite advances in surgical technique, im-
proved patient selection for definitive local therapy and migra-
tion to lower disease stage at presentation, positive surgical
margins are relatively common, ranging from 14% to 46% in
radical prostatectomy series.1
Margin status is 1 of several well-defined preoperative and
pathological predictors of outcome in patients who undergo
radical prostatectomy. Factors, such as preoperative serum
prostate specific antigen (PSA), clinical T stage and biopsy
Gleason score, may be used to predict pathological tumor stage2
as well as the likelihood of biochemical failure after surgery.3, 4
Pathological characteristics of the radical prostatectomy speci-
men, such as lymph node status, seminal vesicle invasion, ex-
are also useful in this regard. While these characteristics are
inherent properties of the local tumor that cannot be altered by
the treating physician, in some cases surgical margin status
may be influenced by the operative technique.5, 6
To date most studies of the impact of margin status on
outcome after radical prostatectomy have demonstrated that
a positive margin has an adverse impact on outcome.6–15
However, this finding is not universal.3 Even in studies in
which margin status was predictive of outcome there was
uncertainty regarding the relative contribution of margin
status when considered with other important predictors,
such as pathological T stage, preoperative PSA and prosta-
tectomy Gleason score. Moreover, although margin status
may significantly influence outcome after radical prostatec-
tomy, to our knowledge the impact of such a finding on
resource use and secondary cancer treatment has not been
studied previously. We used data from the CaPSURE data-
base, a longitudinal disease registry of patients with biopsy
proved prostate cancer, to evaluate the impact of margin
status on prostate cancer recurrence and resource use in
patients who underwent radical prostatectomy as definitive
local therapy for prostate cancer.
Accepted for publication October 29, 1999.
Supported by an educational grant from TAP Holdings, Inc.
* Technology Assessment Group, San Francisco, California.
Editor’s Note: This article is the fourth of 5 published in this
issue for which category 1 CME credits can be earned. In-
structions for obtaining credits are given with the questions
on pages 1294 and 1295.
1171

Evaluation Only. Created with Aspose.PDF. Copyright 2002-2021 Aspose Pty Ltd.
1172
IMPACT OF POSITIVE SURGICAL MARGINS ON PROSTATE CANCER RECURRENCE AND TREATMENT
MATERIALS AND METHODS
CaPSURE project description. CaPSURE is a longitudinal
observational database of patients with prostate cancer who
were recruited through a network of urologists at 29 commu-
nity and academic urology practice sites distributed region-
ally throughout the United States. At each practice site pa-
tients with biopsy proved adenocarcinoma of the prostate are
invited to enroll in the study in a consecutive fashion as they
present for outpatient care. Patients may be enrolled in the
database despite considerable time elapsed since the tumor
diagnosis. Median time since the diagnosis is 2 years.
At study enrollment the urologist completes an extensive
retrospective prostate cancer history based on the existing
medical record. Additional data are recorded prospectively at
each office visit subsequent to the baseline visit, including
new procedures, treatments and diagnostic tests. Hospital
discharge summaries are obtained for all hospitalizations
reported, and all International Classification of Diseases-9
diagnoses and procedures are abstracted. Patients are fol-
lowed until death or study withdrawal. Institutional review
board approval for the study is obtained at each clinical site
and patient informed consent for participation is required for
data collection. Additional details of the project methodology
have been reported previously.16
Subjects. Between June 1995 and January 1998 we invited
4,631 patients to participate in our study, of whom 4,459
(96%) agreed. Of the 4,459 men 1,999 who underwent radical
prostatectomy for definitive local treatment of prostate can-
cer were eligible for study. However, 53, 417 and 146 patients
were excluded from analysis because radical prostatectomy
was performed before 1989, the pathology report of the rad-
ical prostatectomy specimen was not available and they re-
ceived neoadjuvant androgen deprivation preoperatively, re-
spectively. Thus, 1,383 patients treated initially with radical
prostatectomy were included in our current analysis (fig. 1).
Outcomes measured. The primary end points of our analy-
sis were secondary cancer treatment, including radiotherapy,
medical hormonal therapy or orchiectomy, at any time after
radical prostatectomy and postoperative biochemical disease
recurrence, defined as serum PSA 0.2 ng./ml. or greater on 2
consecutive occasions at least 6 weeks after surgery. We
attempted to distinguish secondary treatment in the context
of planned adjuvant therapy from that initiated in response
to physician assessment of treatment failure or disease pro-
gression. Although study physicians recorded all treatment
received and the dates of such treatment, they were not
asked to record the clinical rationale for treatment. Thus, we
did not specifically determine whether treatment was given
as part of an adjuvant therapy program or administered in
response to the discovery of disease recurrence or progres-
sion. Therefore, we hypothesized that planned adjuvant ther-
apy was initiated within 6 months of radical prostatectomy,
while nonadjuvant treatment was instituted more than 6
months after definitive local surgery. The likelihood of sec-
ondary treatment was analyzed independently for adjuvant
and nonadjuvant secondary therapy.
Many cases were enrolled in the database despite a con-
siderable lapse in time since the diagnosis and treatment.
Because PSA values were not routinely recorded between
diagnosis and enrollment in the database, serial PSA data
were not complete for all cases in the database. Thus, our
analysis using PSA recurrence as the end point for treatment
failure included only patients who underwent radical pros-
tatectomy within 6 months of database enrollment and for
whom serial PSA data were complete. Patients with complete
PSA data tended to be the cohort with the most recent sur-
gery. They were somewhat younger with lower serum PSA at
diagnosis, reflecting changing patterns of diagnosis and
treatment during the study period.
FIG. 1. CaPSURE enrollment and eligibility of subjects who un-
derwent radical prostatectomy as of January 1998.
clinical characteristics were analyzed for the study group
overall. Pathological results of the radical prostatectomy
specimen were determined based on a comprehensive review
of the surgical pathology report of the radical prostatectomy
specimen. Data elements obtained from the pathology re-
ports included lymph node status, primary and secondary
Gleason score of the radical prostatectomy specimen, tumor
location and extracapsular disease extension or seminal ves-
icle invasion. Using this information a final pathological T
stage was assigned to each patient using the 1992 American
Joint Committee on Cancer TNM staging system.17 We also
determined surgical margin status from each pathology re-
port. A positive margin was defined as tumor extending to
the inked surgical margin at any location of the radical
prostatectomy specimen, and we recorded the number and
location of each positive margin.
Each pathology report was read independently by 2 urolo-
gists after multiple training sessions during which the scor-
ing of each data element was discussed. Inter-reader reliabil-
ity was determined for each data element and an intraclass
correlation coefficient test statistic of 0.75 or greater was
considered as excellent reproducibility.18, 19 For variables
that did not achieve that level of agreement final agreement
among urologists was achieved after a consensus reading by
another urologist, especially for margin status, extracapsular
disease extension and final pathological tumor stage. In rare
cases of a discrepancy between the initial 2 readings a third
independent urologist reviewed all pathology reports and
Data analysis. Baseline demographic and preoperative settled any discrepancies between the 2 primary readers.

Evaluation Only. Created with Aspose.PDF. Copyright 2002-2021 Aspose Pty Ltd.
IMPACT OF POSITIVE SURGICAL MARGINS ON PROSTATE CANCER RECURRENCE AND TREATMENT
1173
For each demographic characteristic we performed the chi- cluding adjuvant therapy in 23 and 64, and nonadjuvant
square or t test to determine statistically significant differ- therapy in 69 and 82 with negative and positive surgical
ences in patients with positive and negative surgical mar-
gins. Similar analysis was done to compare the 2 groups with
respect to PSA at diagnosis, pathological T stage and pros-
tatectomy Gleason score. For patients who had no secondary
treatment recorded, the time at risk for secondary treatment
equaled the time elapsed between radical prostatectomy and
September 1998, or the earliest date that the patient was lost
to followup, withdrew from the study or died. For those with
secondary treatment recorded, the time at risk for secondary
treatment equaled the time elapsed between radical prosta-
tectomy and secondary treatment. The time at risk for PSA
recurrence was calculated similarly using the date of the
initial PSA value of 0.2 ng./ml. or greater as the recurrence
date.
The rate of secondary treatment and PSA recurrence in each
group was determined using the Kaplan-Meier method and
compared using the log rank test. Secondary treatment and
PSA recurrence were also compared based on the number of
positive margins and positive margin location. To control for
baseline differences in the positive and negative margin groups
we constructed a multivariate Cox proportional hazards model
to adjust for differences in clinically relevant variables, includ-
ing serum PSA at diagnosis, pathological T stage, adjuvant
treatment and prostatectomy Gleason score with ␣ ϭ 0.05 con-
sidered statistically significant. All data were analyzed using a
commercially available computer software package.
margins, respectively. Mean and median followup for pa-
tients without adjuvant treatments was 3.19 and 2.73 years,
respectively. Table 2 and figure 2 show the risk of undergoing
secondary cancer treatment after radical prostatectomy
based on margin status. Patients with positive surgical mar-
gins were significantly more likely to receive adjuvant or
nonadjuvant secondary treatment than those with negative
margins (p ϭ 0.0001).
To adjust for significant differences in pathological, clinical
and demographic factors in patients with positive and nega-
tive margins, respectively, we performed Cox proportional
hazards analysis. In this model the rate of nonadjuvant sec-
ondary treatment was adjusted for patient age, ethnicity,
serum PSA at diagnosis, pathological T stage, prostatectomy
Gleason score and adjuvant treatment (table 3). After adjust-
ing for all factors we continued to find a significant difference
in the rate of secondary treatment in those with positive and
negative surgical margins. Patients with positive margins
were approximately 1.9-fold more likely to undergo second-
ary cancer treatment after radical prostatectomy than those
with negative margins (hazards ratio 1.88, 95% confidence
interval [CI] 1.29 to 2.74, p ϭ 0.0011, table 3).
PSA recurrence after surgery. The risk of PSA recurrence
after radical prostatectomy was determined only in 380 pa-
tients who underwent radical prostatectomy within 6 months
of enrollment in the database. Mean and median followup in
this group was 1.43 and 1.29 years, respectively. Serial PSA
data were complete for this subset of patients. Table 2 and
figure 2 show the risk of PSA recurrence after radical pros-
tatectomy based on margin status. Patients with positive
surgical margins were significantly more likely to have PSA
recurrence than those with negative surgical margins (p ϭ
0.0001). A similar result was obtained after excluding pa-
tients who underwent secondary adjuvant treatment within
6 months of surgery (p ϭ 0.0001).
RESULTS
Overall no tumor was identified at the inked surgical mar-
gins in 829 patients (60%), tumor was present at the inked
surgical margin in 465 (margin positive, 34%) and the pathol-
ogy report was indeterminate with respect to margin status in
89 (6%). Table 1 shows demographic, clinical and pathological
characteristics of the study population according to margin
status. There was no significant difference in the 2 groups in
regard to patient age at diagnosis. However, patients in the 2
groups differed significantly with respect to ethnicity, patholog-
ical T stage, serum PSA at diagnosis and Gleason score of the
radical prostatectomy specimen (p ϭ 0.007, 0.001, 0.001 and
0.001, respectively, table 1). Generally a higher percent of pa-
tients with positive surgical margins had extracapsular disease
extension compared with those who had negative margins. Sim-
ilarly patients with positive surgical margins tended to have
higher serum PSA at diagnosis and a higher prostatectomy
Gleason score.
To adjust for significant differences in pathological, clinical
and demographic factors in patients with positive and nega-
tive surgical margins we performed Cox proportional hazards
analysis. In this model the risk of PSA recurrence was ad-
justed for patient age, ethnicity, serum PSA at diagnosis,
pathological T stage, prostatectomy Gleason score and adju-
vant treatment. After adjusting for all factors we noted that
patients with positive surgical margins were 2.6-fold more
likely to have PSA recurrence than those with negative mar-
gins (95% CI 0.96 to 7.06, p ϭ 0.06, table 3).
Secondary cancer treatment. Overall 261 patients received
secondary cancer treatment after radical prostatectomy, in- men with positive margins 262 (56%) had 1 positive margin,
Impact of 1 versus multiple positive margins. Of the 465
TABLE 1. Clinical and demographic characteristics based on margin status
Characteristic
Mean age (range)
Race:
No. Pts.
Neg. Margins
No. Pts.
Pos. Margins
p Value
63.3 (39.6–79.6)
63.2 (40.7–78.1)
0.80
0.007
White
713
80
32
86.4%
9.7%
3.9%
381
71
12
82.1%
15.3%
2.6%
Black
Other
PSA at diagnosis (ng./ml.):
0.001
0.001
0.001
0.0–4.0
125
478
131
54
15.9%
60.7%
16.6%
6.9%
34
225
129
62
7.6%
50.0%
28.7%
13.8%
4.1–10.0
10.1–20.0
Greater than 20
Disease stage:*
PT2a/b
330
378
62
40.4%
46.3%
7.6%
97
163
128
77
20.9%
35.0%
27.5%
16.6%
PT2c
PT3a/b
PT3c/4a
47
5.8%
Prostatectomy Gleason score:
2–6
7
8–10
Overall
482
233
56
62.5%
30.2%
7.3%
206
153
74
47.6%
35.3%
17.1%
827
463
* Positive lymph nodes in 43 patients.

Evaluation Only. Created with Aspose.PDF. Copyright 2002-2021 Aspose Pty Ltd.
IMPACT OF POSITIVE SURGICAL MARGINS ON PROSTATE CANCER RECURRENCE AND TREATMENT
1174
TABLE 2. Estimated likelihood of secondary treatment and
PSA recurrence 3 years postoperatively based on margin status
(p ϭ 0.0001)
Neg. Margins
Pos. Margins
End Point
No.
Pts.
% Likelihood
No.
Pts.
% Likelihood
Ϯ SE
Ϯ SE
Secondary treatment:
Adjuvant (estimated
at 6 mos. postop.)
Nonadjuvant
829
2.9 Ϯ 0.6
465
14.3 Ϯ 1.7
806
240
7.2 Ϯ 1.1
8.0 Ϯ 3.6
401
130
20.9 Ϯ 2.5
22.0 Ϯ 4.5
PSA recurrence
while 179 (38%) had 2 or more at radical prostatectomy.
Compared with patients with negative margins those with
positive margins were significantly more likely to receive
adjuvant or nonadjuvant secondary treatment postopera-
tively (p ϭ 0.0001, table 4). Similarly patients with multiple
positive surgical margins were significantly more likely to
receive adjuvant or nonadjuvant secondary treatment than
those with negative surgical margins (p ϭ 0.0001, table 4).
We also compared patients according to single versus multi-
ple positive margins. While men with multiple positive mar-
gins were significantly more likely to receive adjuvant sec-
ondary treatment than those with 1 positive surgical margin
(p ϭ 0.004), we observed no difference with respect to non-
adjuvant treatment in these 2 groups (table 4).
Patients with 1 or multiple positive margins were signifi-
cantly more likely to have PSA recurrence after surgery than
those with negative margins (p ϭ 0.0004 versus 0.0001, table
4). However, there was no significant difference in the like-
lihood of PSA recurrence when comparing patients with 1
versus multiple positive margins (p ϭ 0.36, table 4).
Impact of margin location. Recent studies suggest that
positive margin location may provide important information
with respect to prognosis in patients who underwent radical
prostatectomy. For example, when all other margins are
negative, tumor at the prostatic apical margin only may
confer no additional risk of disease recurrence.15, 20 In con-
trast, a positive margin at the bladder neck or prostatic base
may portend a worse outcome than a positive margin else-
where.7 To test these hypotheses patients in our current
study were separated into 3 groups based on a positive mar-
gin location at the apex only, base only and all others, includ-
ing those with a positive margin at any other site or a com-
bination of the apex and/or base with any other site. Those in
the apex and base only groups were then subdivided accord-
ing to 1 versus 2 or more positive margins at the designated
location (table 5). In most of these analyses positive margin
site had no significant impact on secondary cancer treatment
or PSA recurrence after radical prostatectomy. However, pa-
tients with 1 or more positive margins at the prostatic base
were significantly more likely to undergo secondary adjuvant
treatment than those with positive margins elsewhere (p ϭ
0.01).
DISCUSSION
The results of our study demonstrate that a positive sur-
gical margin at radical prostatectomy increases the risk of
disease recurrence and significantly affects additional cancer
treatment after surgery. This increased risk persists even
when adjusting for other important predictors of disease
recurrence, including pathological T stage, serum PSA at
diagnosis and prostatectomy Gleason score. We noted a sim-
FIG. 2. Kaplan-Meier estimates of likelihood after radical prosta-
tectomy based on surgical margin status. A, secondary adjuvant
treatment. B, secondary nonadjuvant treatment. C, PSA recurrence.
ilar rate of disease recurrence and secondary treatment in a significant factor influencing disease recurrence or second-
patients with 1 or multiple positive margins, suggesting that ary treatment.
any positive margin rather than the number of positive mar-
The association of margin status with outcome in patients
gins has the most important effect on outcome postopera- who underwent surgery for clinically localized prostate can-
tively. Moreover, in most cases positive margin site was not cer has been discussed previously. While most studies indi-