Pathology (June 2010) 42(4), pp.

339–343

PROSTATE

Gleason scoring: a comparison of classical and modified (International

Society of Urological Pathology) criteria using nadir PSA as a clinical end
point
BRETT DELAHUNT*, DAVID S. LAMB*, JOHN R. SRIGLEY*{, JUDY D. MURRAY*,
CHANTELLE WILCOXx, HEMAMALI SAMARATUNGAk, CHRISTOPHER ATKINSON{,
NIGEL A. SPRY{, DAVID JOSEPH{ AND JAMES W. DENHAMx
*Department of Pathology and Molecular Medicine, Wellington School of Medicine and Heath Sciences,
University of Otago, Wellington, {Oncology Services, Christchurch Hospital, Christchurch, New Zealand;
{Department of Pathology and Molecular Medicine, McMaster University, ON, Canada; xNewcastle Prostate
Cancer Centre and University of Newcastle, New South Wales, kAquesta Pathology, Queensland, and
{Department of Radiation Oncology, Sir Charles Gairdner Hospital, Perth, Western Australia, Australia

Summary 19742 has gained almost universal acceptance amongst

Aim: To compare the distribution and predictive perfor-
mance of Gleason grade and scores derived using classical
and modified (International Society of Urological Pathology)
criteria.
Methods: Classical and modified Gleason grades and
scores were assigned to cases of prostate carcinoma
accessioned by the Trans-Tasman Radiation Oncology
Group RADAR trial. Separate scores were derived for each
grading system based on the percentage of each Gleason
grade per case (area-based score) and the score of the
highest scoring core. The predictive performance of each of
the four Gleason scores assigned to each case was
evaluated using nadir prostate specific antigen (nPSA) as a
clinical end point.
Results: Modified Gleason scoring resulted in an upward
shift of scores, primarily resulting from the reclassification of
classical pattern 3 to modified pattern 4. On re-grading
classical Gleason score 7 cores, there was a 64% decrease
in the number of cores with 5 25% Gleason pattern 4 tumour,
while the number of cores with 75–100% Gleason pattern
4 tumour increased by 96%. All four scoring models
performed reasonably well as predictors of nPSA; however,
on comparison of the prognostic gradients of the grade
groupings, classical Gleason scoring outperformed modified
Gleason scoring.
Conclusion: The overlap of the predictive performance of
Gleason pattern 3 with Gleason pattern 4, suggests that review of
the defining features of modified pattern 4 may improve the
prognostic prediction of modified Gleason scoring.
Key words: Prostate, adenocarcinoma, Gleason grade, International
Society of Urological Pathology, prostate specific antigen.
INTRODUCTION
The grading scheme for prostatic adenocarcinoma
proposed by Gleason in 19661 and further validated in
anatomical pathologists and urologists.3,4 Despite the
widespread usage of the grading system it has been
acknowledged that the grading criteria, as proposed by
Gleason, might not be applicable to the current clinical
situation. In particular, it has been noted that some of the
constituent categories of the Gleason grading system do not
accord with subsequent clinical behaviour.4 This was
reinforced by the conclusions of a conference held in
Stockholm, Sweden, in 2004 where it was recommended
that the criteria of the Gleason grading system should be
reviewed.5
In 2005 the International Society of Urological
Pathology (ISUP) convened a consensus conference with
the objective of re-assessing the defining criteria of each
of the five patterns of the Gleason grading system.6 At
this meeting the criteria of each of the Gleason grades
were reviewed and recommendations were made relating
to grading criteria and the method of calculation of the
Gleason score for both thin core biopsies and radical
prostatectomy specimens.
Over the ensuing five years since the ISUP consensus
conference, there have been few studies that have investi-
gated the clinical utility of the amended grading criteria. In
these studies the concordance of Gleason patterns,7–9 the
correlation between the Gleason scores of thin core biopsies
and matched radical prostatectomy specimens,7 and the
effect on risk assessment criteria7,8 comparing both classical
and modified grading, were evaluated. Additionally, mod-
ified grading criteria have been correlated with serum
prostate specific antigen (PSA) levels,10 tumour volume on
biopsy10 and pathological stage.11 Investigation of the
prognostic significance of modified grading is limited to
one study which utilised time to biochemical failure as a
surrogate end point.8
The present study utilised data from the multinational
Randomized Androgen Deprivation and Radiotherapy
(RADAR) trial, co-ordinated by the Trans-Tasman Radia-
tion Oncology Group,12 to compare the distribution of

Print ISSN 0031-3025/Online ISSN 1465-3931 # 2010 Royal College of Pathologists of Australasia
DOI: 10.3109/00313021003787924
340 DELAHUNT et al.
Gleason patterns and scores assigned using classical and
modified grading criteria. We have recently shown that post-
radiation PSA nadir (nPSA) can be used to determine the
prognostic value of co-variables (such as the Gleason
scoring system) in men treated with radiation, with or
without neo-adjuvant androgen deprivation, if used in
multi-variable models that are adjusted for pre-treatment
(initial) PSA (iPSA), radiation dose delivered and duration
of androgen deprivation (ADT).13 In this present study we
have investigated the prognostic performance of both
classical and modified Gleason scoring systems, using nPSA
as a clinical end point.13
MATERIALS AND METHODS
Patients with locally advanced prostate cancer and without radiological
evidence of distant metastases were recruited to the RADAR trial between
2003 and 2007. The trial has a 2 6 2 randomisation design with patients
receiving 6 or 18 months androgen deprivation and 18 months of
zoledronic acid or no zoledronic acid, in addition to radical external beam
radiotherapy. A comprehensive pathology review was an integral part of
the trial. This was undertaken in order to confirm a diagnosis of prostate
carcinoma, and for the standardisation of Gleason grading and scoring to
be used as a major co-variable in the main trial analyses to be undertaken in
2012.
Sections of the thin core biopsies from patients enrolled in the RADAR
trial were forwarded from reporting laboratories for review, with all cases
being coded at the source laboratory. Matching clinical data were also
retrieved, with patient age and tumour stage, treatment protocol, PSA
levels at the time of diagnosis, and nPSA data being available for each
patient.
Sections were initially reviewed in order to confirm the presence of
malignancy. Gleason grades and a Gleason score was then assigned to each
core containing carcinoma according to the criteria of Gleason (1974).2
Each core was examined by light microscopy utilising a Zeiss Integra-
tionsplatte II eyepiece grid (Zeiss, Germany) with five cross-hatched lines
forming a grid of 25 squares as previously described.14 Sections were
examined at 2006 magnification and the Gleason pattern of tumour
present under intersecting lines was recorded. This was continued for the
complete area of tumour in the core and the percentage of each pattern of
tumour present within the core was recorded. Two final Gleason scores
were then derived for each case. The first of these was based on the area of
each pattern present within all cores sampled (area-based score). The
second Gleason score was taken as the score of the individual core having
the highest score in each case (highest scoring core score).
This process was then repeated for all cores in each case, with re-
grading of tumours based upon the recommendations of the ISUP
consensus conference.6 As for assignment of classical Gleason scores,
two final modified Gleason scores (area-based score and highest scoring
core score) were derived for each case. In both instances the
recommendation of the ISUP consensus conference, that if a tertiary
component of higher grade tumour was present, then the final score
would be based upon the primary pattern and the higher of the
remaining patterns, was followed. The distribution of classical and
modified Gleason grades and scores were then compared. Finally the
relationship between classical Gleason scores, modified Gleason scores,
and nPSA was examined.
The prognostic value of the classical and modified Gleason systems
was investigated using nPSA as a clinical end point. To make these
comparisons we composed multiple logistic regression models with
nPSA at a cut point of 50.1/0.1 ng/mL as the dependent variable.
Gleason/modified Gleason scores were the independent variables of
interest, defined as a categorical variable (6/7/8). All models were
adjusted for the independent variables iPSA, ADT duration and
radiation dose treated as categorical variables: [iPSA (510/107
520/20–550/50 ng/mL), ADT duration (6 months/18 months),
radiation dose (66/70/74 Gy / and high dose-rate brachytherapy boost)].
Pathology (2010), 42(4), June
Comparison of the models was achieved by determining and comparing
the area under the receiver operating characteristics curve (AUC) for
the predictive probabilities of each model.15 In the event of ties,
models exhibiting superior prognostic gradients in their histological
grade groupings were considered to reflect superior grading methodol-
ogies. Stata version 9 statistical software was used in these analyses
(Stata, USA).
RESULTS
Data from 590 men with locally advanced prostatic
adenocarcinoma with minimum clinical follow-up of 2.5
years from the completion of radiotherapy were evaluated
in this report. Patients ranged in age from 48 to 85 years,
with a mean age of 68 years. Serum PSA at diagnosis
and post-radiation nPSA values ranged from 1.59 to
135.0 ng/mL and from 0.003 to 66.6 ng/mL, with medians
of 14.2 ng/mL and 0.1 ng/mL, respectively.
From the 590 cases a total of 5567 core biopsies were
undertaken, with the number of cores per case ranging
from 1 to 25 (median 9 cores per case). On review
adenocarcinoma was detected in 3231 cores, with the
number of positive cores ranging from 1 to 20 per case
(median 5 cores per case).
The distribution of Gleason patterns for the cancer-
positive cores assigned according to classical Gleason and
ISUP modified scoring are summarised in Fig. 1. The major
shift in scores resulted from re-grading of areas of tumour
from classical Gleason pattern 3 to modified Gleason
pattern 4. This shift in cases showing Gleason pattern 4 is
reflected in the detailed analysis of the percentage of
Gleason pattern 4 tumour in cases assigned a Gleason score
of 3þ 4 ¼ 7 or 4þ 3 ¼ 7 utilising classical criteria (Fig. 2).
On re-grading, the number of cores with 5 25%, 25–49%
and 50–74% pattern 4 tumour decreased by 64%, 32% and
12%, respectively, while the number of cores with 75–100%
Gleason pattern 4 tumour increased by 96%. The Gleason
scores for all cases assigned according to classical or
modified criteria, with final scores calculated by both the
area-based and highest scoring core methods, are shown in
Fig. 3. The inclusion of tertiary patterns of higher grade
than the secondary pattern in the final score resulted in
upgrading of 44 cases upon modified grading, while the
disregarding of any lower grade comprising 5 5% of the
tumour focus influenced the final tumour score in only
three cases.
Comparison of classical and modified Gleason score
with nPSA shows that all of the models performed
reasonably well, with AUCs ranging between 0.744 and
0.757 (Table 1). Slightly higher AUCs were seen in the
two classical Gleason score models but the difference
between these and the two modified Gleason scores was
not significant. When the prognostic gradients of the
grade groupings were compared, the classical Gleason
scoring system outperformed the modified Gleason
scoring system, regardless of whether the scores were
derived using the area of involvement or on the highest
scoring core methods. In models (not shown) re-grouping
of the scores produced by the modified Gleason scoring
system (58/8/48) failed to improve the prognostic
gradient. The area based scoring methods exhibited
slightly better prognostic gradients than the highest
scoring core method for both the classical and modified
Gleason scoring systems.
 ISUP GLEASON GRADING 341

Fig. 1 Distribution of Gleason patterns in 3231 cores containing carcinoma classified according to classical and modified grading criteria.

Fig. 2 Percentage of Gleason pattern 4 tumour in cores with Gleason score 7 carcinoma assigned using classical and modified Gleason grading criteria.

DISCUSSION our series. In the majority of cases the observed upward

In our series of tumours we have shown modified
Gleason scores to be higher than those based upon classical
Gleason scoring criteria. This upshift in scoring is due, in
part, to the inclusion of higher tertiary grades in the final
Gleason scores, although this had only limited impact.
Similarly, the recommendation that low volumes (5 5%) of
lower grade tumour should be ignored for scoring
purposes6 contributed little to the upgrading of cases in
shift in scoring resulted from the re-grading of often
substantial proportions of tumour from classical Gleason
pattern 3 to modified Gleason pattern 4. This is further
illustrated by our observation that the percentage of
pattern 4 carcinoma in Gleason score 7 tumours markedly
increased when tumours were re-graded using modified
Gleason grading criteria. Further upgrading of cases was
seen when the final modified Gleason score was based upon
342 DELAHUNT et al. Pathology (2010), 42(4), June

Fig. 3 Gleason scores of 590 cases of prostate carcinoma comparing classical and modified Gleason criteria with final scores based on grade percentage (area-
based score) and highest scoring core based score.

Table 1 Multiple logistic regression models comparing the classical and modified Gleason scoring system for scores derived either by the area involved or
highest scoring core methods

Classical Gleason score Modified Gleason score

Area based score Parameter OR 95% Cl p Parameter OR 95% Cl p

Gleason 6 1.00 – – Gleason6 1.00 – –

Gleason 7 1.91 1.09–336 0.024 Gleason 7 1.59 0.69–3.66 0.271
Gleason 8þ 2.97 1.56–5.78 0.001 Gleason 8þ 2.47 1.04–5.82 0.039
AUC: 0.750 AUC: 0.744

Highest scoring core

based score Parameter OR 95% Cl p Parameter OR 95% Cl p

Gleason 6 1.00 – – Gleason 6 1.00 – –

Gleason 7 1.71 0.91–3.20 0.097 Gleason 7 1.37 0.55–3.47 1.37
Gleason 8þ 2.88 1.57–5.30 0.001 Gleason 8þ 2.44 1.04–5.73 0.041
AUC: 0.757 AUC: 0.745

Co-variables not shown in the models are initial (pre-treatment) PSA, ADT duration and radiation dose administered (all treated as categorical variables as
defined in the Methods).
95%CI, 95% confidence intervals; AUC, area under the receiver operating characteristics curve; OR, odds ratio.

the highest graded core, rather than area of individual
modified Gleason patterns within all the cores in each case.
While it has been shown that highest grade based scores
correlate best with scores seen in matched radical prosta-
tectomy specimens,16 it is apparent from our data that this
scoring methodology has a considerable effect on the final
tumour score in individual cases, which in turn will
influence treatment options.
Our findings of an upward shift in tumour scores
following modified Gleason grading are in accord with
the results of previous studies. In 2008 Helpap and Egevad
showed re-grading to result in a significant shift in Gleason
scores, with Gleason score 6 tumours in their series
decreasing from 48.4% to 22.0% of cases, while Gleason
score 7 tumours increased from 25.5% to 67.9% of cases.7
Somewhat similar results were demonstrated by Billis et al.8
who found Gleason primary patterns to increase by 1 or 2
units, respectively, in 16.8% and 0.6% of cases in their
series upon re-grading, while Gleason scores declined by 1
unit in 4.07% and increased by 1 or 2 units in 24.4% and
3.5% of cases. The decline in Gleason scoring in a small
percentage of cases is surprising. This appears to have been
controlled by the secondary pattern which was decreased
by 1 unit in 14.5% of cases, however, the details of this
were not fully explained. Furthermore, in that study, it was
not possible to differentiate the impact of modified grading
criteria from that of basing the final Gleason score on the
core showing the highest tumour grade.

In the most recent study to investigate the effects of
modified Gleason grading, 34.5% of cases were up-graded
with no cases down-graded.9 In this series, re-grading
resulted in increased proportions of Gleason score 8, 9 and
10 tumours, while the proportion of Gleason score 5, 6 and
7 (which included both 3þ 4 and 4þ 3 tumours) decreased.
In 91.8% of upgraded cases the highest scoring core rule
influenced this up-grade, while the inclusion of the tertiary
grade was influential in only 7.7% of the cases.
Serum PSA levels and cancer extent in thin core biopsies
have been significantly correlated with modified Gleason
grading, although a similar correlation was observed when
tumours were graded using classical criteria.10 Modified
scoring has also been shown to significantly predict
pathological stage for clinical T1c, T2a, T2b and T2c
tumours.11
Prognostic assessment of modified Gleason grading is
limited to one study, with time to biochemical (PSA) failure
being taken as the end point.8 In this study there was no
association between classical Gleason score and interval to
biochemical failure, while for modified grading, this was
weakly significant. There was considerable overlap between
the curves for Gleason scores 5–6, and Gleason score 7
tumour groups, while the Gleason score 8–10 cohort was
associated with a higher biochemical failure rate for the
first 60 months of a maximum 94 month follow-up period.
Assessment of prognostic factors for prostate adenocar-
cinoma is hampered by the long natural history of the
disease.4 While it is acknowledged that the tumour is
associated with a substantial morbidity and mortality,
survival intervals of 10 years or more are commonplace
even when relapse has occurred. Importantly, however, it
has been noted that mortality rises three-fold as late as 15
years following diagnosis.17 As a consequence, prospective
studies investigating prognostic factors for prostate adeno-
carcinoma are fraught with logistical difficulties, while
retrospective studies are often confounded by the rapid
evolution of treatment modalities.
In view of these problems the importance of identifying
outcome surrogates for prostate carcinoma is well recog-
nised and several candidate end points have been pro-
posed.18 The importance of post-radiation nPSA as a
predictor of biochemical failure, local tumour progression,
distant biochemical failure and prostate cancer specific
survival was highlighted using data from the TROG 96.01
randomised trial of short duration ADT and radiation in
men with locally advanced prostate cancer.13 Because post-
radiation nPSA was found to be dependent on the iPSA
and the treatment regimen used, it cannot be used as a
surrogate end point; however, it can be used to assess the
prognostic value of other variables in models appropriately
adjusted for iPSA and treatment regimen.
In the present report, post-radiation nPSA data have
indicated that the classical Gleason scoring system provides
better prognostic discrimination than the modified Gleason
system. It has also shown that the Gleason scores derived
using the area-based method provide better prognostic
discrimination than those derived using the highest core
score. Importantly there was no significant difference
between modified Gleason scores 6 and 7 when analysed
ISUP GLEASON GRADING 343
separately. This suggests that current patterns included in
the modified Gleason grade 4 category do not identify
survival (nPSA) data different to that provided by Gleason
grade 3. This would further suggest that it is timely to
review category 4 criteria with a view to re-categorising
those features which have survival characteristics similar to
Gleason pattern 3.
Address for correspondence: Professor B. Delahunt, Department of
Pathology and Molecular Medicine, University of Otago – Wellington,
PO Box 7343, Newtown, Wellington 6242, New Zealand. E-mail:
bd@wnmeds.ac.nz
References
1. Gleason DF. Classification of prostate carcinoma. Cancer Chemother
Rep 1966; 50: 125–8.
2. Gleason DF, Mellinger GT and the Veterans Administration
Cooperative Urological Research Group. Prediction of prognosis for
prostatic adenocarcinoma by combined histological grading and
clinical staging. J Urol 1974; 11: 58–64.
3. Lamb DS, Slaney D, Smart R, et al. Prostate cancer: the new evidence
base for diagnosis and treatment. Pathology 2007; 39: 537–544.
4. Delahunt B, Srigley JR, Lamb DS. Gleason grading: consensus and
controversy. Pathology 2009; 41: 613–4.
5. Amin M, Boccon-Gibod L, Egevad L, et al. Prognostic and predictive
factors and reporting of prostatic carcinoma in prostatic needle biopsy
specimens. Scand J Urol Nephrol 2005; 39: 20–33.
6. Epstein JI, Alsbrook WC, Amin MB, et al. The 2005 International
Society of Urological Pathology (ISUP) Consensus Conference on
Gleason Grading of Prostatic Carcinoma. Am J Surg Pathol 2005; 29:
1228–42.
7. Helpap B, Egevad L. The significance of modified Gleason grading of
prostatic carcinoma in biopsy and radical prostatectomy specimens.
Virchows Arch 2006; 449: 622–7.
8. Billis A, Guimaraes MS, Freitas LL, et al. The impact of the 2005
International Society of Urological Pathology consensus conference on
standard Gleason grading of prostate carcinoma in needle biopsies.
J Urol 2008; 180: 548–53.
9. Kuroiwa K, Uchino H, Yokomizo A, Naito S. Impact of reporting
rules of biopsy Gleason score for prostate cancer. J Clin Pathol 2009;
62: 260–3.
10. Helpap B, Egevad L. Correlation of modified Gleason grading of prostate
cancer with age, serum prostate specific antigen and tumor extent in
needle biopsy specimens. Anal Quant Cytol Histol 2008; 30: 133–8.
11. Naito S, Kuroiwa K, Kinukawa N, et al. Validation of Partin tables
and development of a preoperative nomogram for Japanese patients
with clinically localized prostate cancer using 2005 International
Society of Urological Pathology consensus on Gleason grading: data
from the Clinicopathological Research Group for Localized Prostate
Cancer. J Urol 2008; 180: 904–10.
12. Haworth A, Kearvell R, Greer PB, et al. Assuring high quality
treatment delivery in clinical trials – Results from the Trans-Tasman
Radiation Oncology Group (TROG) study 03.04 ‘RADAR’ set-up
accuracy study. Radiotherapy Oncol 2009; 90: 299–306.
13. Lamb DS, Denham JW, Joseph D, et al. A comparison of
the prognostic value of early PSA based variables following
external beam radiotherapy, with or without preceding androgen
deprivation: analysis of data from the TROG 96.01 randomized trial.
Int J Radiation Oncology Biol Phys 2010; Mar 27: (Epub ahead of
print).
14. Delahunt B, Nacey JN. Renal cell carcinoma. Histological indicators
of prognosis. Pathology 1987; 19: 258–63.
15. Harrell FE, Lee KL, Mark DB. Multivariable prognostic models:
issues in developing models, evaluating assumptions and adequacy,
and measuring and reducing errors. Stat Med 1996; 15: 361–87.
16. Kunz GM Jr, Epstein JI. Should each core with prostate cancer be
assigned a separate Gleason score? Hum Pathol 2003; 34: 911–4.
17. Crawford DE. Understanding the epidemiology, natural history, and
key pathways involved in prostate cancer. Urology 2009; 73 (Suppl 5A):
4–10.
18. Denham JW, Steigler A, Wilcox C, et al. Time to biochemical failure
and prostate-specific antigen doubling time as surrogates for prostate
cancer-specific mortality: evidence from the TROG 96.01 randomised
controlled trial. Lancet Oncol 2008; 9: 1058–68.