Articles

Risk of prostate cancer diagnosis and mortality in men with

a benign initial transrectal ultrasound-guided biopsy set:
a population-based study
Nina Klemann, M Andreas Røder, J Thomas Helgstrand, Klaus Brasso, Birgitte G Toft, Ben Vainer, Peter Iversen

Summary
Background The risk of missing prostate cancer in the transrectal ultrasound-guided systematic biopsies of the Lancet Oncol 2017
prostate in men with suspected prostate cancer is a key problem in urological oncology. Repeat biopsy or MRI-guided Published Online
biopsies have been suggested to increase sensitivity for diagnosis of prostate cancer, but the risk of disease-specific January 13, 2017
http://dx.doi.org/10.1016/
mortality in men who present with raised prostate-specific antigen (PSA) concentration and a benign initial biopsy
S1470-2045(17)30025-6
result remains unknown. We investigated the risk of overall and prostate cancer-specific mortality in men with a
See Online/Comment
benign initial biopsy set. http://dx.doi.org/10.1016/
S1470-2045(17)30024-4
Methods Data were extracted from the Danish Prostate Cancer Registry—a population-based registry including all Copenhagen Prostate Cancer
men undergoing histopathological assessment of prostate tissue. All men who were referred for transrectal Center (N Klemann MD,
ultrasound-guided biopsy for assessment of suspected prostate cancer between Jan 1, 1995, and Dec 31, 2011, in M A Røder PhD,
J T Helgstrand MD,
Denmark were eligible for inclusion. Follow-up data were obtained on April 28, 2015. The primary endpoint was the Prof K Brasso PhD,
cumulative incidence of prostate cancer-specific mortality, analysed in a competing risk setting, with death from other Prof P Iversen MD) and
causes as the competing event. Department of Pathology
(B G Toft MD,
Prof B Vainer DMSc),
Findings Between Jan 1, 1995, and Dec 31, 2011, 64 430 men were referred for transrectal ultrasound-guided biopsy, of Rigshospitalet, Copenhagen
whom 63 454 were eligible for inclusion. Median follow-up was 5·9 years (IQR 3·8–8·5) and the total follow-up time, University Hospital,
from the enrolment of the first patient on Jan 1, 1995, until the extraction of causes of death on April 28, 2015, was Copenhagen, Denmark
20 years. 10 407 (30%) of 35 159 men with malignant initial biopsy sets died from prostate cancer, compared with Correspondence to:
541 (2%) of 27 181 men with benign initial biopsy sets. Estimated overall 20-year mortality was 76·1% (95% CI Dr Nina Klemann, Copenhagen
Prostate Cancer Center,
73·0–79·2). In all men referred for transrectal ultrasound-guided biopsy, the cumulative incidence of prostate cancer- Rigshospitalet, Copenhagen
specific mortality after 20 years was 25·6% (24·7–26·5) versus 50·5% (47·5–53·5) for mortality from other causes. In University Hospital,
men with benign initial biopsy sets, the cumulative incidence of prostate cancer-specific mortality was 5·2% (3·9–6·5) 2200 Copenhagen N, Denmark
versus 59·9% (55·2–64·6) for mortality from other causes. In men with PSA concentrations 10 μg/L or lower and ninakhp@gmail.com

benign initial biopsy sets (2779 men), the cumulative incidence of prostate cancer-specific mortality was 0·7% (0·2–1·3).
Cumulative incidence of prostate cancer specific mortality in men with benign initial biopsy sets was 3·6% (95% CI
0·1–7·2) for men with a PSA higher than 10 ng/mL but 20 ng/mL or less (855 men) and 17·6% (12·7–22·4) and for
men with a PSA higher than 20 ng/mL (454 men).

Interpretation The first systematic transrectal ultrasound-guided biopsy set holds important prognostic information.
The 20-year risk of prostate cancer-specific mortality in men with benign initial results is low. Our findings question
whether men with low PSA concentration and a benign initial biopsy set should undergo further diagnostic
assessment in view of the high risk of mortality from other causes.

Funding Capital Region of Denmark’s Fund for Health Research, Danish Cancer Society, Danish Association for
Cancer Research, and Krista and Viggo Petersen’s Foundation.

Introduction MRI-guided biopsies can increase sensitivity and

Since the 1980s, transrectal ultrasound-guided systematic
biopsies have been the standard of care in the diagnostic
work-up of men judged to be at risk of prostate cancer,1
but the prognostic implications of a benign first biopsy set
are not well known. The detection rate of prostate cancer
from the initial transrectal ultrasound-guided biopsy
depends on several factors, such as prostate volume,
number of biopsy cores, and prostate-specific antigen
(PSA) concentration,2–5 but prostate cancer can be missed
because of sampling error or anterior location of the
lesion.6,7 Recent data suggest that MRI and subsequent
specificity of diagnosis, thus avoiding unnecessary
biopsies.8–10 However, although MRI-guided biopsy is able
to detect prostate cancer that is initially missed by
transrectal ultrasound-guided biopsy, the consequence of
a benign initial result on later prostate cancer-specific
mortality is not well described.
In 2010, the European Randomized Study of Screening
for Prostate Cancer (ERSPC) reported that prostate cancer-
specific mortality in 3056 men presenting with a benign
biopsy set on the initial transrectal ultrasound-guided
biopsy was 0·03% after 11 years of follow-up.11 However,

www.thelancet.com/oncology Published online January 13, 2017 http://dx.doi.org/10.1016/ S1470-2045(17)30025-6 1

 Articles
Research in context
Evidence before this study
We searched PubMed with the terms “transrectal-guided
biopsy”, “prostate”, “prostatic cancer”, “benign”, “negative”,
“prognosis”, “PSA”, and “mortality” for articles published in
English between Jan 1, 1990, and May 15, 2016. This search
retrieved one relevant trial assessing the risk of prostate
cancer-specific death following a benign initial biopsy set. In
2010, the European Randomized Study of Screening for
Prostate Cancer (ERSPC) showed that the risk of prostate
cancer-specific mortality in men randomly assigned to
prostate-specific antigen (PSA) screening was 0·03% in those
who presented with a benign initial biopsy set.
Added value of this study
Findings from our population-based study support the notion
that the initial transrectal ultrasound-guided biopsy set holds
important prognostic information. Men with a benign initial
biopsy set have a low risk of prostate cancer-specific mortality
and a high risk of mortality from other causes. These data should
be interpreted with recognition that men in Denmark in the
the prognostic role of the first transrectal ultrasound-
guided biopsy has not previously been described at a
population-based level. We aimed to assess this role in an
analysis of all men referred for transrectal ultrasound-
guided biopsy in Denmark between 1995 and 2011.
Methods
Study design and participants
In this population-based analysis, we included all men
undergoing transrectal ultrasound-guided biopsy between
Jan 1, 1995, and Dec 31, 2011, in Denmark. Patient data
were extracted from the Danish Prostate Cancer Registry
(DaPCaR)—a population-based registry including all men
undergoing histopathological assessment of prostate
tissue. Methods for DaPCaR have been described in detail
elsewhere;12 the database includes integration of data from
several national Danish registries. All initial biopsy sets
registered in DaPCaR were used in the analysis. DaPCaR
has been approved by the Danish Data Protection Agency
and the Danish Health Authority. No ethics approval was
required for the study.
Procedures
For the purpose of this study, all initial transrectal
ultrasound-guided biopsy sets, repeat biopsy sets, age at
biopsy, PSA concentrations, biopsy Gleason scores, and
causes of death (prostate cancer vs other causes) were
extracted. Patients were stratified by PSA concentration
(≤10 μg/L vs >10 to ≤20 μg/L vs >20 μg/L) at referral.
Gleason scores were divided into five groups (≤6, 7, 3 + 4,
4 + 3, or ≥8). Patients categorised as having a Gleason
score of 7 included patients for whom only the score was
available, and all patients for whom the Gleason score
2 www.thelancet.com/oncology Published online January 13, 2017 http://dx.doi.org/10.1016/S1470-2045(17)30025-6
studied period are likely to have been underassessed by biopsy
and underscreened compared with a contemporary setting due to
the low uptake of PSA testing in the Danish general population
during a large part of the studied period and the use of 6-core
biopsy until early 2000s which is not standard of care today. Our
data also show that the combination of a benign initial biopsy
result and PSA concentration at biopsy could further substratify
patients into risk groups in terms of disease-specific mortality.
Implications of all the available evidence
Overtreatment and overdiagnosis remain major problems in
prostate cancer after the introduction of PSA testing. The need
for repeat biopsy in men with benign initial biopsy results is
questionable, and restricting these additional investigations to
men with high PSA concentrations and other risk factors might
lower the risk of overdiagnosis and the number of unnecessary
biopsies. The optimum follow-up strategies in men who
present with benign initial biopsy results is still unclear, and
warrants further investigations, including use of imaging
modalities and biomarkers.
added up to 7—eg, 2 + 5. For causes of death, we combined
information about vital status from the Central Person’s
Registry (updated daily) with causes of death from the
national Cause of Death Register. If men were registered
as dead, but did not have a death certificate as of
April 28, 2015, or if the cause of death was unspecified, a
manual review of medical records was done by the first
author (NK) and coauthor JTH. Dates of deaths were
obtained from the national Cause of Death Register.12
Prostate cancer was judged to be the cause of death if a
patient had received androgen deprivation therapy,
had evidence of metastatic disease, had received
chemotherapy or palliative radiotherapy for prostate
cancer, or had been referred for palliative care for prostate
cancer. Metastatic disease was deemed to be present when
positive bone scans or CT scans had been reported in the
patient’s chart. Docetaxel and cabazitaxel (either or both)
were considered prostate cancer-specific chemotherapy.
Radiotherapy was classed as prostate cancer-specific when
aimed at the prostate or bony metastases. If a patient had
been referred for palliative care, manual review was done
to ensure that this referral was due to prostate cancer and
not because of other malignancies.
Biopsy sets consisted of six cores until the early 2000s
when guidelines were changed to a 10–12-core systematic
biopsy scheme. PSA concentrations were obtained from
laboratories across Denmark and were included in the
analysis if taken a maximum of 2 years before and
3 months after the biopsy procedure. Biopsies containing
adenocarcinoma, possible adenocarcinoma, dysplasia,
or high-grade prostatic intraepithelial neoplasia, and
neuroendocrine or small cell differentiation were judged
to be malignant, and only biopsy sets containing

exclusively normal tissue were deemed benign. The
histopathological assessment followed the guidelines
from the International Society of Urological Pathology
(ISUP) during the entire investigation period. Histological
reassessment of specimens before the 2005 revision was
not done. Follow-up data were obtained on April 28, 2015.
Patients were excluded if follow-up data were insufficient
(eg, information about vital status was unavailable or
there were insufficient biopsy results).
Outcomes
The primary endpoint was the cumulative incidence of
prostate cancer-specific death and the secondary endpoint
was the cumulative risk of prostate cancer diagnosis
following a benign initial biopsy set.
Statistical analysis
We did not do a power calculation for this study. All men
referred for prostate biopsy were included. For survival
analysis, a competing risk setting was used with death
from other causes as the competing event. We used the
competing risk assessment as an alternative to Kaplan-
Meier analysis because it accounts for competing events.
Statistically, men “compete” for events—ie, if most men
have one event, the risk of the competing event will
decrease accordingly. The Kaplan-Meier analysis is
dichotomous, which means that events cannot compete,
and the risk of an event can be overestimated because of
censoring of competing events. In large cohorts, Kaplan-
Meier analysis can be imprecise.13 For the risk of cancer in
men who underwent repeat biopsy or transurethral
resection of the prostate, the competing risk model
accounts for the fact that a man can die before repeat
biopsy or resection, but also, if he is alive, it also accounts
for the risk of harbouring a different Gleason score—ie,
the model allows for several competing outcomes, which
better reflects the true risk of a specific Gleason score in
men who undergo repeat biopsy or transurethral resection.
Patients who were alive were right-censored on
April 28, 2015. The cumulative incidence of mortality
(prostate cancer-specific and other causes) was estimated
at 20 years, apart from analyses including PSA concen-
trations, in which 15-year estimates were used because
PSA concentrations were missing for a large part of the
early inclusion period.
Univariate cause-specific hazard for prostate cancer
mortality was calculated through the use of competing
risk with PSA logarithmically transformed on a base 2
scale. Estimates are presented with 95% CI and should be
interpreted as the hazard ratio (HR) for prostate cancer-
specific mortality per two-times increase in PSA
concentration.
Two negative predictive values (NPVs) were calculated:
the proportion of patients with benign initial biopsy sets
who did not die from prostate cancer, and the proportion
who were not subsequently diagnosed with prostate
cancer. Thus, the calculation of NPV does not account for
www.thelancet.com/oncology Published online January 13, 2017 http://dx.doi.org/10.1016/S1470-2045(17)30025-6
Articles
time-to-event. Statistical analyses were done with SPSS
version 22.0 and R version 3.3.1. p values less than 0·05
were deemed significant.
Role of the funding source
The funders of the study had no role in study design,
data collection, data analysis, data interpretation, or
writing of the report. All authors had access to the raw
data. The corresponding author had full access to all the
data in the study and had final responsibility for the
decision to submit for publication.
Results
Between Jan 1, 1995, and Dec 31, 2011, 64 430 men
were referred for transrectal ultrasound-guided biopsy
(figure 1). 976 men were excluded because of insufficient
follow-up data, typically non-Danish residents; therefore,
63 454 patients were included in the analysis (entire
cohort). For the analyses according to initial biopsy result
(malignant or benign), 1114 (2%) of these 63 454 patients
were excluded because of diagnoses other than normal
prostate tissue or prostate cancer from the biopsy (eg,
bladder cancer). 35 159 (55%) men were diagnosed with
prostate cancer on the initial biopsy and 27 181 (43%) had
a benign initial biopsy set.
Table 1 shows patient characteristics (age and PSA
concentration) at the time of biopsy. The total follow-up
time, from the enrolment of the first patient on Jan 1, 1995,
until the extraction of causes of death on April 28, 2015,
was 20 years. Median follow-up time when calculated
for the entire cohort was 5·9 years (IQR 3·8–8·5). In
the entire cohort, 1636 men were followed up for more
than 15 years. 1261 men with benign initial transrectal
64 430 patients referred for transrectal
ultrasound-guided biopsy
976 excluded because of
insufficient follow-up data
(eg, non-Danish residents)
63 454 included in analysis
1114 excluded (pathology
reports with diagnosis not
related to the prostate)
35 159 malignant first biopsy 27 181 benign first bopsy
As of April 28, 2015: As of April 28, 2015:
17 225 (49%) alive 20 486 (75%) alive
10 407 (30%) died from prostate 541 (2%) died from prostate
cancer cancer
7527 (21%) died from other causes 6154 (23%) died from other causes
Figure 1: Patient selection and outcomes
3
 Articles

ultrasound-guided biopsy sets and 359 men with
malignant first biopsy sets were followed up for more
than 15 years (totals 1620 because additional 16 patients
were excluded due to pathology reports). During follow-
up, 24 629 men in the entire cohort died, including
10 948 (17%) who died from prostate cancer. Of 35 159 men
with malignant initial biopsy sets, 10 407 (30%) died from
prostate cancer, compared with 541 (2%) of 27 181 men
with benign initial biopsy sets. Estimated overall 20-year
mortality was 76·1% (95% CI 73·0–79·2). In patients with
malignant initial biopsy sets, estimated overall mortality
after 20 years was 85·7% (95% CI 82·4–89·0), compared
with 65·1% (60·3–69·8) in patients with benign initial
biopsy sets.
PSA concentrations at diagnosis were available for
13 895 (22%) of 63 454 men in the entire cohort. In patients
with benign initial biopsy sets, PSA concentrations were
available for 5933 (22%) of 27 181 individuals. When
Median age,
years (IQR)
All men referred for biopsy
sampling (n=63 454)
Benign initial biopsy set (n=27 181)
See Online for appendix
Malignant initial biopsy set
(n=35 159)
Men undergoing repeat biopsy or
TURP (n=8526)
Men with benign initial biopsy sets,
diagnosed with prostate cancer on
repeat biopsy or TURP (n=2845)
PSA=prostate-specific antigen. TURP=transurethral resection of the prostate.
Table 1: Patient characteristics at the time of biopsy or TURP
including only PSA measurements taken a maximum of
2 years before and 3 months after the biopsy procedure,
data were available for 10 121 (16%) of 63 454 men in the
entire cohort, and for 4132 (15%) of 27 181 men with
benign initial biopsy sets. The median time interval
between the PSA test and biopsy was –21 days
(IQR –44 to –7). 1233 (12%) of 10 121 PSA measurements
within the defined time interval were taken after the
biopsy procedure but only up to 3 months after and
616 (6%) of 10 121 were taken within the first month after
the procedure. Unfortunately, PSA measurements were
unavailable for more than 70% of patients because
laboratories did not store data electronically, were closed,
or data had been discarded. Many PSA measurements
were missing for the early inclusion period (1995–2000),
which is why we were only able to calculate an estimated
15-year cumulative incidence of mortality for men with
available PSA data. Median PSA concentration at first
biopsy differed significantly between patients who had
benign (7·7 μg/L [IQR 5·5–12·0]) and malignant
Median PSA (17·3 μg/L [8·3–58·6]) biopsy sets (p<0·0001).
concentration,
μg/L (IQR) In all men referred for transrectal ultrasound-guided
biopsy in Denmark, the estimated cumulative incidence
68·0 (63·0–75·0) 11·0 (6·6–28·0)
of prostate cancer-specific mortality after 20 years was
67·0 (62·0–73·0) 7·7 (5·5–12·0)
25·6% (95% CI 24·7–26·5) versus 50·5% (47·5–53·5)
70·0 (64·0–77·0) 17·3 (8·3–58·6)
for mortality from other causes (figure 2, appendix p 1).
Stratified by PSA concentration at the time of referral,
67·0 (62·0–72·0) 8·6 (6·2–14·2) men with PSA of 10 μg/L or lower had an estimated
cumulative incidence of prostate cancer-specific mortality
66·0 (61·0–72·0) 10·0 (6·6–21·0) after 20 years of 6·6% (95% CI 4·5–8·8), compared with
28·1% (24·6–31·6) for mortality from other causes
(appendix p 1). In univariate analysis, increasing PSA
concentration was significantly associated with the risk
of prostate cancer-specific mortality (HR 1·57 [95% CI
1·55–1·60], p<0·0001).

100 Mortality from other causes

Prostate cancer-specific mortality
Cumulative incidence of death (%)

75

50 II

25

0
0 5 10 15 20
Time from date of biopsy (years)
Number at risk 63 454 55 999 46 157 31 619 18 834 9910 5161 2488 979 235 24
Number censored* 0 0 4052 14 237 24 127 31 125 34 917 37 074 38 367 39 011 39 202

Figure 2: Risk of prostate cancer-specific mortality versus mortality from other causes in all men referred for transrectal ultrasound-guided biopsy
*Cumulative number of men censored.

4 www.thelancet.com/oncology Published online January 13, 2017 http://dx.doi.org/10.1016/S1470-2045(17)30025-6

 Articles

100 Mortality from other causes

Prostate cancer-specific mortality
Cumulative incidence of death (%)

75

50

25

0
0 5 10 15 20
Time from date of biopsy (years)
Number at risk 35 159 28 867 22 099 13 838 7309 3417 1528 603 206 33 1
Number censored* 0 0 1995 7174 11 825 14 626 16 016 16 720 17 042 17 194 17 225

Figure 3: Risk of prostate cancer-specific mortality versus mortality from other causes in patients with a malignant initial biopsy set
*Cumulative number of men censored.

100 Mortality from other causes

Prostate cancer-specific mortality
Cumulative incidence of death (%)

75

50

25

0
0 5 10 15 20
Time from date of biopsy (years)
Number at risk 27 181 25 683 22 875 16 996 11 114 6376 3564 1849 768 202 23
Number censored* 0 0 1555 6215 11 151 15 142 17 475 18 880 19 815 20 301 20 465

Figure 4: Risk of prostate cancer-specific mortality versus mortality from other causes in patients with a benign initial biopsy set
*Cumulative number of men censored.

In men with malignant initial transrectal ultrasound- cumulative incidence of mortality from other causes
guided biopsy sets, the estimated cumulative incidences after 15 years was 32·3% (95% CI 25·1–40·0) in men
of prostate cancer-specific mortality and mortality from with PSA concentrations of 10 μg/L or lower at referral,
other causes were similar at 43·6% (95% CI 42·1–45·1) 36·2% (30·2–42·3) in men with PSA of 10–20 μg/L, and
and 42·1% (38·9–45·3) after 20 years, respectively 32·1% (29·3–34·9) in men with PSA higher than
(figure 3, appendix p 2). Men with malignant first biopsy 20 μg/L (appendix p 2). In univariate analysis, increasing
sets and PSA concentrations of 10 μg/L or lower at the PSA concentration was significantly associated with the
time of referral had a cumulative incidence of prostate risk of prostate cancer-specific mortality in men with
cancer-specific death of 16·9% (95% CI 10·3–23·5) after malignant initial biopsy sets (HR 1·42 [95% CI
15 years, increasing to 23·4% (17·7–29·0) in patients 1·40–1·45], p<0·0001).
with PSA of >10 ≤20 μg/L, and 55·7% (52·9–58·4) in In men with benign initial biopsy sets, the estimated
those with PSA higher than 20 μg/L (appendix p 2). The 20-year cumulative incidence of prostate cancer-specific

www.thelancet.com/oncology Published online January 13, 2017 http://dx.doi.org/10.1016/S1470-2045(17)30025-6 5

 Articles

100 Mortality from other causes

Prostate cancer-specific mortality
Cumulative incidence of death (%)

75

50

25

0
0 5 10 15 20
Time from date of biopsy (years)
Number at risk 2779 2710 2360 1650 1040 518 237 84 21 4 0
Number censored* 0 0 194 669 1169 1558 1780 1919 2013 2067 2067

Figure 5: Risk of prostate cancer-specific mortality versus mortality from other causes in patients with a benign initial biopsy set and prostate-specific
antigen concentration of 10 μg/L or lower
*Cumulative number of men censored.

mortality was 5·2% (95% CI 3·9–6·5) compared with Nearly half of patients diagnosed on repeat biopsy were
59·9% (55·2–64·6) for mortality from other causes diagnosed with histological low-risk disease (table 2). Of
(figure 4, appendix p 3). In men with benign initial biopsy the remaining 18 655 patients for whom no reassessment
sets, the cumulative incidence of prostate cancer-specific was done, 191 (1%) patients had prostate cancer confirmed
death after 15 years was 0·7% (95% CI 0·2–1·3) in as the cause of death.
men with PSA concentrations of 10 μg/L or lower In an exploratory post-hoc analysis, patients with benign
(figure 5) compared with 3·6% (0·1–7·2) in men with PSA initial biopsy sets, the risk of later detection of prostate
of 10–20 μg/L, and 17·6% (12·7–22·4) in men with PSA of cancer, irrespective of Gleason score, was 11·1% (95% CI
higher than 20 μg/L. The cumulative incidence of 10·6–11·6) after 20 years (appendix p 4). When stratified by
mortality from other causes was 26·1% (95% CI PSA concentration at the time of referral, the risk of future
22·0–30·2) in men with PSA concentrations of 10 μg/L or cancer detection after 15 years was 7·6% (95% CI 6·5–8·7)
lower (figure 5) compared with 39·8% (31·1–48·5) in men in patients with PSA concentrations of 10 μg/L or lower,
with PSA of 10–20 μg/L, and 56·2% (45·5–66·9) in those increasing to 12·1% (9·5–14·7) in patients with PSA of
with PSA higher than 20 μg/L (appendix p 3). In univariate 10–20 μg/L, and 25·2% (20·4–29·9) in patients with PSA
analysis, increasing PSA concentration was significantly higher than 20 μg/L (appendix p 4). The overall 20-year
associated with the risk of prostate cancer-specific risk of future detection of potentially clinically significant
mortality in men with benign initial biopsy sets (HR 2·20 prostate cancer based on Gleason scores in men with
[95% CI 2·07–2·34], p<0·0001). initial benign biopsies was 3·4% (95% CI 3·1–3·6) for
The estimated 20-year cumulative incidence of Gleason score 7 and was 2·3% (2·1–2·6; appendix p 5) for
mortality from other causes was lower in men with Gleason score 8 or higher. When stratified by PSA
malignant initial biopsy sets (42·1% [95% CI 38·9–45·3]) concentration, the risk of detecting prostate cancer with a
than in men with benign initial biopsy sets (59·9% Gleason score of 7 after 15 years was 2·3% (95% CI
[55·2–64·6]). 1·6–2·9) in patients with PSA concentrations of 10 μg/L
Of the 27 181 patients who presented with a benign initial or lower, compared with 2·8% (1·7–3·9) in patients with
transrectal ultrasound-guided biopsy, 8526 underwent PSA of 10–20 μg/L, and 7·5% (4·9–10·0) in patients
reassessment with a second biopsy or transurethral with PSA of more than 20 μg/L (appendix p 5). For patients
resection of the prostate (TURP) within a median of with PSA concentrations of 10 μg/L or lower, the
320 days (IQR 118·8–973·0). Median age at the subsequent 15-year risk of detecting prostate cancer with a Gleason
procedure was 67·0 years (IQR 62·0–72·0; table 1). score of 8 or higher was 0·9% (95% CI 0·5–1·3), increasing
2845 (10%) men were diagnosed with prostate cancer or more than ten times to 10·4% (6·6–14·1) in patients with
possible adenocarcinoma; 2408 of these patients were PSA of more than 20 μg/L. In patients with PSA of
diagnosed on repeat biopsy and 437 were diagnosed on 10–20 μg/L, the 15-year risk of detecting prostate cancer
transurethral resection. Table 2 shows the histopathology with a Gleason score of 8 or higher was 2·6% (95% CI
for cancer diagnoses following benign initial biopsy. 1·4–3·7; appendix p 5).

6 www.thelancet.com/oncology Published online January 13, 2017 http://dx.doi.org/10.1016/S1470-2045(17)30025-6


The NPV, calculated as the proportion of patients with
benign initial biopsies who did not die from prostate
cancer, was 98% (26 640 of 27 181 patients). When
calculated as the proportion of patients who did not later
become diagnosed with prostate cancer, the NPV was
90% (24 336 of 27 181).
Discussion
In this population-based cohort with long-term follow-
up, our findings show that the initial transrectal ultra-
sound-guided biopsy set holds important prognostic
information, and men with benign initial biopsy results
are at low risk of prostate cancer-specific mortality. Our
results should be interpreted with recognition that data
were derived from a cohort that was potentially
underscreened and underassessed by biopsy compared
with a contemporary setting, due to a low uptake of
PSA testing in the general population in Denmark and
the use of 6-core biopsy set for a large part of the studied
period. Furthermore, our data show that men with
benign initial biopsy results have a significant risk of
mortality from other causes compared with the risk of
prostate cancer-specific mortality. Our findings are in
accordance with observations from the ERSPC trial.11
Of 3056 men assigned to PSA screening who had a
benign initial transrectal ultrasound-guided biopsy
result, only 287 were diagnosed with prostate cancer
and seven eventually died from the disease after 11 years
of follow-up.11 Estimated progression-free survival and
prostate cancer-specific survival in patients with benign
initial biopsies were almost 80% and 97% after 14 years,
respectively. Importantly, as in the DaPCaR data, the
ERSPC trial included both sextant and extended
10–12-core biopsies, and no MRI-guided biopsies were
done. These findings could serve to substantially reduce
overtreatment and overdiagnosis of men screened for
prostate cancer by restricting the need for repeat
biopsies to men with high PSA concentrations and
other risk factors, such as family history.
To the best of our knowledge, our study is the largest
and most comprehensive analysis of men with benign
transrectal ultrasound-guided biopsy results. Although
median follow-up was 5·9 years, estimated overall
mortality was 76·1% after 20 years for the entire cohort,
which supports an acceptable degree of data maturity.
DaPCaR is based on integration of several national
registries with high validity, including manual validation
of diagnoses and causes of death.12 The Nordic countries
excel by the many comprehensive, up-to-date registries,
largely due to the fact that every citizen is given a Central
Person’s Registry identification number, allowing for the
combination of data stored in different registries. For
death certificates, the delay in the entry of data into the
Danish Death Certificate Registry is short. Since 2009,
death certificates have been assessed electronically and
reported directly at the time of death to the registry and
then validated electronically with the internationally
www.thelancet.com/oncology Published online January 13, 2017 http://dx.doi.org/10.1016/S1470-2045(17)30025-6
Articles
Repeat biopsy Transurethral
(n=2408) resection of the
prostate (n=437)
Unspecified adenocarcinoma 179 (7%) 21 (5%)
Gleason score
≤6 921 (38%) 237 (54%)
3+4 215 (9%) 38 (9%)
4+3 74 (3%) 19 (4%)
7 366 (15%) 61 (14%)
≥8 405 (17%) 61 (14%)
Possible adenocarcinoma 248 (10%) 0
Data are n (%).
Table 2: Histopathological subtypes of prostate cancer detected on
repeat biopsy or transurethral resection of the prostate after benign
initial biopsy
recognised computer software from the Iris Institute. For For more on Iris see
DaPCaR, we manually reviewed around 65 000 pathology https://www.dimdi.de/static/en/
klassi/irisinstitute/about-iris/
reports and thousands of patient charts to ensure high- index.htm
quality data. Unfortunately, this tedious validation process
took an extra year after the integration of the other
registries into the DaPCaR on April 28, 2015; we then
spent a year reviewing pathology reports (for Gleason
Scores).
Admittedly, our data do show that some men with
initial negative biopsy results ultimately die from prostate
cancer, but the incidence of prostate cancer-specific death
in this group of men is low. Our data do not procure
information about which biopsy strategy should be used
to detect prostate cancer in a population per se, but add
to the longstanding discussion about whether a
systematic transrectal ultrasound-guided biopsy can
diagnose patients at risk of prostate cancer-related
mortality. Our results suggest that the NPV of a benign
initial biopsy is very high, and even at the current
duration of follow-up, the magnitude of the dataset and
number of events suggest that the NPV is unlikely to
change substantially with longer follow-up.
In the recent decade, whether or not transrectal
ultrasound-guided biopsy is sufficient to identify clinically
significant prostate cancer has been called into question,
and MRI with targeted biopsies has been introduced to
further enhance the diagnostic accuracy. Recently, the
PROMIS case-control study compared MRI-guided biopsy
versus transrectal ultrasound-guided biopsy with a
template mapping biopsy scheme as a reference test in
576 men.14 Results of the study showed that MRI-guided
biopsies have a higher sensitivity for diagnosis of clinically
significant prostate cancer (defined as the presence of
Gleason score ≥4 + 3 or maximum cancer core length
≥6 mm) compared with transrectal ultrasound-guided
biopsies. In all recent MRI publications, the term clinically
significant prostate cancer refers to a histopathological
definition. Originally, clinically significant prostate cancer
referred to the Epstein criteria, defined as tumours larger
7
 Articles
than 0·2 cm³ in volume, a Gleason pattern 4 or 5, or PSA
concentration higher than 0·15 μg/L.15,16 The criteria were
defined by correlating pathological features to biochemical
recurrence, tumour-free survival, and prostate cancer-
specific death following radical prostatectomy in patients
with non-palpable prostate cancer.17–19 The MRI definition
of clinically significant prostate cancer most often refers to
the University College London criteria (ie, the presence of
Gleason grade ≥4, prostate cancer involvement of ≥4 mm
of the biopsy core, or a Prostate Imaging Reporting and
Data System [PI-RADS] score >3).7,20 Furthermore, other
studies have shown that MRI-guided biopsies significantly
increase detection of prostate cancer compared with
transrectal ultrasound-guided biopsies in men with
previously benign biopsies (14–39% vs 25%,
respectively).8–10 Of these detected cancers, 25–48% are
judged clinically significant.21–23 Repeat biopsy with
transrectal ultrasound-guided biopsy has shown a
detection rate of 10–34%, with 16·5–29·8% of cancers
perceived as clinically significant,11,24,25 which is similar to
the data presented here. Currently, no MRI study has
followed up patients until death and the prognostic value
of clinically significant prostate cancer detected by MRI-
guided biopsy is unknown. Our data indicate that there
could be a discordant association between the
histopathological definition of clinically significant
prostate cancer and the actual risk of dying from the
disease. This discrepancy is probably fuelled by lead-time
bias and stage migration because of increasing uptake of
PSA testing in most countries, including Denmark.
Nonetheless, MRI, other imaging modalities, or
biomarkers could potentially help to reduce the number of
men who undergo unnecessary biopsies, especially if
these instruments can select men who can avoid any
biopsies.
Although diagnosis might be missed, the primary
endpoint in our study was risk of missing lethal prostate
cancer and the prognostic role of the initial transrectal
ultrasound-guided biopsy. It must be recognised that
some men with benign initial biopsy results eventually
develop prostate cancer and subsequently die from the
disease. Although a pertinent problem in our study was
missing PSA data, we still showed that PSA concentration
at first biopsy adds substantially to the prognostic value
of the biopsy information. However, because PSA
measurements were not available for all patients, this
added prognostication should be interpreted with
caution. The optimum follow-up strategy for men with
benign initial transrectal ultrasound-guided biopsy
results has yet to be determined, and further research
may show if monitoring beyond PSA concentration and
repeat biopsies (eg, MRI-guided) can pinpoint patients
at the highest risk.
The observed difference in overall mortality between
patients with initial positive or initial negative biopsy
results might well be explained by the fact that patients
with negative biopsies indeed had a low risk of the disease
8 www.thelancet.com/oncology Published online January 13, 2017 http://dx.doi.org/10.1016/S1470-2045(17)30025-6
that they were originally suspected of having. However,
these patients might represent a select cohort of PSA-
screened men, and thus be affected by healthy volunteer
and lead-time bias. Other explanations could be the
differences in the cohort sizes and the fact that the risk of
having an initial negative biopsy increased over time,
reflected by the increased use of PSA testing in the Danish
general population during the course of this study.
Recently, results of the PROTECT trial, which
randomly assigned 1643 PSA-screened men to definitive
therapy, including included radical prostatectomy and
external-beam radiotherapy, or observation, showed that
prostate cancer-specific survival was 98·8% after
10 years even in men undergoing active monitoring.26
Our study does not have the stringent follow-up
schedule of such a randomised trial, and the risk of
prostate cancer-specific mortality in men with benign
initial biopsy sets was higher than that reported in the
PROTECT trial. However, it should be emphasised that
no specific national protocol for follow-up after a
negative biopsy has ever existed in Denmark, and
18 655 men never underwent a repeat biopsy.
Limitations to our study include the unavailability of
information about clinical stage (TNM) at the time of
diagnosis, and incomplete PSA data, since many PSA
measurements had not been stored or were not available
in an electronic format. Our data include no information
about treatment or follow-up schedules. In view of the
results of the PROTECT,26 SPCG-4,27 and PIVOT28 trials,
the effects of treatment aiming to cure prostate cancer in
men who were subsequently diagnosed with prostate
cancer on the risk of disease-specific mortality are
suspected to be low. We recognise that the Gleason
grading system changed during the studied period, and
no re-evaluation of the histopathological assessment was
done. Recent data suggest that the latest ISUP 2005
reclassification of the Gleason grading system has
introduced a grade migration,29 mainly because of an
upgrading of Gleason grade 3 to 4. The possible effect of
change in the ISUP 2005 on risk of mortality in DaPCaR
data is currently being explored in another study by our
group.
In conclusion, in this Danish population-based setting,
the first transrectal ultrasound-guided biopsy set offers
strong prognostic information, and men with benign
initial biopsy results are at low risk of prostate cancer-
specific mortality but at high risk of mortality from other
causes. PSA concentration at initial biopsy could add
further prognostic value. Our data could be used to
inform men about the need for further biopsy assessment
when the first transrectal ultrasound-guided biopsy
result is negative.
Contributors
NK, MAR, KB, and PI participated in study design. NK, MAR, KB, BV,
and PI contributed to writing of the report. MAR, KB, BGT, BV, and PI
provided study supervision. All authors contributed to data collection,
data analysis, and data interpretation.

Declaration of interests
We declare no competing interests.
Acknowledgments
NK has received financial support from the Capital Region of Denmark’s
Fund for Health Research, the Danish Cancer Society, the Danish
Association for Cancer Research, and Krista and Viggo Petersen’s
Foundation. We would like to acknowledge our data manager,
Günther Momsen, for his tremendous efforts following the complex
conversion and integration of several large Danish registries into the
Danish Prostate Cancer Registry (DaPCaR). Without his support,
DaPCaR would never have been possible. We also thank the
Departments of Urology and Pathology at Rigshospitalet, Copenhagen,
Denmark, for outstanding support.
References
1 Mottet N, Bellmunt J, van den Bergh RCN, et al. Guidelines on
prostate cancer. 2015. https://uroweb.org/wp-content/uploads/09-
Prostate-Cancer_LR.pdf (accessed Dec 12, 2016).
2 Hugosson J, Carlsson S. Overdetection in screening for prostate
cancer. Curr Opin Urol 2014; 24: 256–63.
3 Park S, Miyake H, Hara I, Eto H. Predictors of prostate cancer on
repeat transrectal ultrasound-guided systematic prostate biopsy.
Int J Urol 2003; 10: 68–71.
4 Aus G, Ahlgren G, Hugosson J, Pedersen KV, Rensfeldt K,
Söderberg R. Diagnosis of prostate cancer: optimal number of prostate
biopsies related to serum prostate-specific antigen and findings on
digital rectal examination. Scand J Urol Nephrol 1997; 31: 541–44.
5 Djavan B, Margreiter M. Biopsy standards for detection of prostate
cancer. World J Urol 2007; 25: 11–17.
6 Numao N, Kawakami S, Sakura M, et al. Characteristics and clinical
significance of prostate cancers missed by initial transrectal 12-core
biopsy. BJU Int 2012; 109: 665–71.
7 Abd-Alazeez M, Ahmed HU, Arya M, et al. The accuracy of
multiparametric MRI in men with negative biopsy and elevated
PSA level—can it rule out clinically significant prostate cancer?
Urol Oncol 2014; 32: 45.e17–22.
8 Arsov C, Rabenalt R, Blondin D, et al. Prospective randomized trial
comparing magnetic resonance imaging (MRI)-guided in-bore
biopsy to MRI-ultrasound fusion and transrectal ultrasound-guided
prostate biopsy in patients with prior negative biopsies. Eur Urol
2015; 68: 713–20.
9 Anastasiadis AG, Lichy MP, Nagele U, et al. MRI-guided biopsy of
the prostate increases diagnostic performance in men with elevated
or increasing PSA levels after previous negative TRUS biopsies.
Eur Urol 2006; 50: 738–48.
10 Beyersdorff D, Taupitz M, Winkelmann B, et al. Patients with a
history of elevated prostate-specific antigen levels and negative
transrectal US-guided quadrant or sextant biopsy results: value of
MR imaging. Radiology 2002; 224: 701–06.
11 Schröder FH, van den Bergh RCN, Wolters T, et al. Eleven-year
outcome of patients with prostate cancers diagnosed during
screening after initial negative sextant biopsies. Eur Urol 2010;
57: 256–66.
12 Helgstrand JT, Klemann N, Røder MA, et al. Danish Prostate Cancer
Registry (DaPCaR)—methodology and early results from a novel
national database. Clin Epidemiol 2016; 8: 351–60.
www.thelancet.com/oncology Published online January 13, 2017 http://dx.doi.org/10.1016/S1470-2045(17)30025-6
Articles
13 de Glas NA, Kiderlen M, Vandenbroucke JP, et al. Performing survival
analyses in the presence of competing risks: a clinical example in
older breast cancer patients. J Natl Cancer Inst 2015; 108: djv366.
14 Ahmed HU, Bosaily AES, Brown LC, et al. The PROMIS study:
a paired-cohort, blinded confirmatory study evaluating the accuracy of
multi-parametric MRI and TRUS biopsy in men with an elevated
PSA. 2016 ASCO Annual Meeting; Chicago, IL, USA; June 3–7, 2016.
Abstract 5000.
15 Bastian PJ, Mangold LA, Epstein JI, Partin AW. Characteristics of
insignificant clinical T1c prostate tumors. A contemporary analysis.
Cancer 2004; 101: 2001–05.
16 Epstein JI, Walsh PC, Carmichael M, Brendler CB. Pathologic and
clinical findings to predict tumor extent of nonpalpable (stage T1c)
prostate cancer. JAMA 1994; 271: 368–74.
17 Han M, Snow PB, Epstein J, et al. A neural network predicts
progression for men with gleason score 3+4 versus 4+3 tumors after
radical prostatectomy. Urology 2000; 56: 994–99.
18 Epstein JI, Pizov G, Walsh PC. Correlation of pathologic findings
with progression after radical retropubic prostatectomy. Cancer 1993;
71: 3582–93.
19 Bastian PJ, Gonzalgo ML, Aronson WJ, et al. Clinical and pathologic
outcome after radical prostatectomy for prostate cancer patients with
a preoperative Gleason sum of 8 to 10. Cancer 2006; 107: 1265–72.
20 Moore CM, Kasivisvanathan V, Eggener S, et al. Standards of
reporting for MRI-targeted biopsy studies (START) of the prostate:
recommendations from an International Working Group. Eur Urol
2013; 64: 544–52.
21 Vickers AJ, Cronin AM, Aus G, et al. A panel of kallikrein markers
can reduce unnecessary biopsy for prostate cancer: data from the
European Randomized Study of Prostate Cancer Screening in
Göteborg, Sweden. BMC Med 2008; 6: 19.
22 Abraham N, Mendhiratta N, Taneja S. Patterns of repeat prostate
biopsy in contemporary clinical practice. J Urol 2015; 193: 1178–84.
23 Moore CM, Robertson N, Arsanious N, et al. Image-guided prostate
biopsy using magnetic resonance imaging-derived targets:
a systematic review. Eur Urol 2013; 63: 125–40.
24 Tan N, Lane BR, Li J, Moussa AS, Soriano M, Jones JS.
Prostate cancers diagnosed at repeat biopsy are smaller and less
likely to be high grade. J Urol 2008; 180: 1325–29.
25 Djavan B, Mazal P, Zlotta A, et al. Pathological features of prostate
cancer detected on initial and repeat prostate biopsy: results of the
prospective European Prostate Cancer Detection study. Prostate
2001; 47: 111–17.
26 Hamdy FC, Donovan JL, Lane JA, et al. 10-year outcomes after
monitoring, surgery, or radiotherapy for localized prostate cancer.
N Engl J Med 2016; 375: 1415–24.
27 Bill-Axelson A, Holmberg L, Garmo H, et al. Radical prostatectomy
or watchful waiting in early prostate cancer. N Engl J Med 2014;
370: 932–42.
28 Wilt TJ, Brawer MK, Jones KM, et al. Radical prostatectomy versus
observation for localized prostate cancer. N Engl J Med 2012;
367: 203–13.
29 Berg KD, Thomsen FB, Nerstrøm C, et al. The impact of the 2005
International Society of Urological Pathology consensus guidelines
on Gleason grading—a matched-pair analysis. BJU Int 2016;
117: 883–89.
9