Context.—Gleason grading of prostatic adenocarcinoma in core needle biopsies is important for predicting prognosis and selecting appropriate therapy. Previous studies have shown that Gleason scores assigned by general pathologists have a low correlation with those assigned by urologic pathologists, and that general pathologists tend to undergrade prostate carcinoma.

Objective.—To determine if the performance of general pathologists grading prostate needle biopsies has changed over time.

Design.—Four hundred sixteen prostate biopsies from men treated at a single community-based institution between 1987 and 2000 were reviewed by one urologic pathologist (A.A.R.). The correlation between the original Gleason score and the reviewer's score was determined over time.

Results.—Cases were divided into those performed and originally interpreted in the first half of the study (1987–1996) and those performed and originally interpreted in the second half (1996–2000). Overall concordance for exact Gleason score was 59% (244/416). The exact concordance of the Gleason score assigned by the original pathologist and the reviewer during the first half of the study was 51%, whereas in the second half of the study the concordance was significantly greater (66.3%, P = .002). However, when grouped into score categories of 6 or less, 7, and 8 or greater, there was no significant difference in the exact concordance between the first half of the study (78.3%) and the second half (78.4%). Fifty-five percent of the cases in which there was discordance were graded as 7 by the reference pathologist and 6 or less by the original pathologist. There was no correlation between concordance in Gleason score and the percentage of tissue involved by carcinoma.

Conclusion.—The concordance between general pathologists' Gleason grading and that of a reference pathologist in this study is much higher than that in previously reported studies. Although exact concordance has significantly improved over time, concordance by clinically significant groups has remained high throughout the study, is dominated by the difference between Gleason score 7 and 6 or less, and is unrelated to the size of the tumor focus.

Gleason grading of prostatic adenocarcinoma in core needle biopsies is important for predicting prognosis and selecting appropriate therapy. Several studies have highlighted that there may be limits to the correlation between the grade assigned to the biopsy and those found in the subsequent resection specimen.1–4 In addition, the interobserver agreement among pathologists for Gleason scoring varies considerably.5–7 In particular, previous studies have shown that urologic pathologists may agree on grouped Gleason scores in no more than 70% of cases,7 and general pathologists do significantly worse.6 One study with only 3 pathologists showed agreement on exact Gleason score among the 3 participants in only 10% of cases, and on grouped Gleason scores in only 44% of cases.5 In general, there is a consistent trend to undergrade prostate carcinoma.3,6 Such studies strongly suggest that central review is necessary for meaningful studies on Gleason grading of prostate cancer,5 and perhaps more importantly, call into question whether patients with prostate cancer diagnosed in the community can rely on Gleason scores assigned in that setting.

However, in the course of reviewing a relatively large number of cases for a longitudinal study of prostate cancer, we were impressed that the grading of prostate cancer by community-based pathologists was not only good, but improving. To determine if this were so, we correlated the Gleason scores of the original community-based pathologists with those of the central urologic pathologist over time.

Between 1987 and 2000, 416 men were treated for biopsy-proven adenocarcinoma of the prostate at a Harvard Associated Community Outreach Facility (St Anne's Hospital, Fall River, Mass). Biopsies were originally obtained and diagnosed at approximately 12 to 14 different institutions in the Boston area. This group represents essentially all men treated at this facility for whom adequate follow-up information could be obtained. All biopsies were subsequently reviewed by one urologic pathologist (A.A.R.). The Gleason scoring of this reference pathologist has previously been shown to correlate strongly with prognosis.8 The correlation between the original Gleason score assigned by the outside pathologist and the reviewer grade was determined over time.

Concordance between Gleason score groups was determined using a Pearson χ2 test. Correlation of the percentage of the tissue involved by carcinoma and Gleason scores was determined using the Mann-Whitney U test and a threshold of P < .05 for significance.

According to the reference pathologist, there were 6 (1%) Gleason score 4 cases, 19 (5%) Gleason score 5 cases, 162 (39%) Gleason score 6 cases, 179 (43%) Gleason score 7 cases, 20 (5%) Gleason score 8 cases, 27 (6%) Gleason score 9 cases, and 3 (1%) Gleason score 10 cases. Overall concordance between the original pathologist and the reference pathologist for exact Gleason score for the entire study was 59% (244/416).

Cases were divided into those performed and originally interpreted in the first half of the study (1987–1996, 187 cases) and those performed and originally interpreted in the second half (1996–2000, 229 cases). The exact concordance of the Gleason score assigned by the original pathologist and the reviewer during the first half of the study was 51%, whereas in the second half of the study correlation was significantly greater (66.3%, P = .002).

When grouped into scores of 6 or less, 7, and 8 or greater, we found no significant difference in the exact concordance for the first half of the study (78.3%) and the second half (78.4%). Of the 90 cases in which discordance was identified, 50 (55%) of the cases were graded as 7 by the reference pathologist and 6 or less by the original pathologist.

We found no significant difference in the percentage of tissue involved by tumor between the cases with either exact Gleason score or grouped Gleason score concordance and those in which the Gleason scores were different (P > .12 for all comparisons). In addition, the percentage of biopsies with 5% of the biopsy or less involved with tumor (ie, small tumor foci) were not significantly different in the first half (8.5%) and the second half (10%) of the study (P = .72).

There is no question that the assignment of Gleason score in prostate needle biopsies is important, affecting both prognosis and choice of therapy. Several articles have suggested that there may be significant disagreement in the assigning of Gleason scores in general,5,6 and in particular problems in the Gleason scores assigned by community-based pathologists.3,6 These articles suggested that community-based pathologists significantly undergrade prostate cancer when compared with urologic pathologists.3,6 If true, patients whose Gleason score is assigned by a community pathologist may not be getting as accurate data concerning their prognosis as they might if that score were assigned by a dedicated urologic pathologist.

However, our results clearly show that at least in the setting of this study, general pathologists are doing an excellent job in assigning Gleason scores. Since our data are essentially confined to the New England area, we are unable to determine whether these results are specific to this region or reflect the performance of pathologists nationally. Although the concordance for exact Gleason score has significantly improved over time, the concordance of clinically significant grouped Gleason scores was very high throughout the study, with an exact concordance for grouped Gleason scores of 78%. This percentage compares favorably with a concordance of only 70% for grouped Gleason scores assigned by dedicated urologic pathologists, a performance which has been deemed “acceptable.”7 While there are differences between these 2 studies (in particular, the cases selected for the urologic pathologists to review were selected in part to reflect difficult cases, while the current study included essentially all cases seen at an institution), they do represent the only data currently available on this subject. Although these differences make comparison difficult, at the very least they do suggest that Gleason scores assigned by general pathologists may be just as “acceptable” as those assigned by others.

However, although this performance is acceptable, as in prior studies,5 when community-based pathologists disagree with a reference urologic pathologist, they appear to continue to consistently undergrade carcinoma. Nearly 55% of the cases in which disagreements were found consisted of cases in which the reference pathologist assigned a score of 7, while the original pathologist had assigned a score of 6 or less. As others have noted,9 the distinction between a Gleason score of 6 and 7, although clinically significant, may be particularly difficult to assign consistently. We had wondered if the size of the tumor would be a factor in determining the level of concordance, but this was not the case. We thus conclude that this difference reflects an actual change in performance in grading observed over time. While these results are good, they still show that nearly 20% of these cases were not placed into the same prognostically relevant group by the 2 reviews. Since this is a sizable number and the difference appears to be clinically significant, efforts to improve the reproducibility of the assignment of these scores, such as continued educational efforts,6 remain worthwhile.

In conclusion, the concordance between general pathologists' Gleason grading and that of a reference pathologist in this study is much higher than that in previously reported studies. Although exact concordance has significantly improved over time, concordance in clinically significant groups has remained high throughout the study, is dominated by the difference between Gleason scores 7 and 6 or less, and is unrelated to the size of the tumor focus.

Fukagai
,
T.
,
T.
Namiki
,
H.
Namiki
,
R. G.
Carlile
,
M.
Shimada
, and
H.
Yoshida
.
Discrepancies between Gleason scores of needle biopsy and radical prostatectomy specimens.
Pathol Int
2001
.
51
:
364
370
.
Djavan
,
B.
,
K.
Kadesky
,
B.
Klopukh
,
M.
Marberger
, and
C. G.
Roehrborn
.
Gleason scores from prostate biopsies obtained with 18-gauge biopsy needles poorly predict Gleason scores of radical prostatectomy specimens.
Eur Urol
1998
.
33
:
261
270
.
Steinberg
,
D. M.
,
J.
Sauvageot
,
S.
Piantadosi
, and
J. I.
Epstein
.
Correlation of prostate needle biopsy and radical prostatectomy Gleason grade in academic and community settings.
Am J Surg Pathol
1997
.
21
:
566
576
.
King
,
C. R.
Patterns of prostate cancer biopsy grading: trends and clinical implications.
Int J Cancer
2000
.
90
:
305
311
.
McLean
,
M.
,
J.
Srigley
,
D.
Banerjee
,
P.
Warde
, and
Y.
Hao
.
Interobserver variation in prostate cancer Gleason scoring: are there implications for the design of clinical trials and treatment strategies?
Clin Oncol (R Coll Radiol)
1997
.
9
:
222
225
.
Allsbrook
,
W. C.
,
K. A.
Mangold
,
M. H.
Johnson
,
R. B.
Lane
,
C. G.
Lane
, and
J. I.
Epstein
.
Interobserver reproducibility of Gleason grading of prostatic carcinoma: general pathologist.
Hum Pathol
2001
.
32
:
81
88
.
Allsbrook
,
W. C.
,
K. A.
Mangold
, and
M. H.
Johnson
.
et al
.
Interobserver reproducibility of Gleason grading of prostatic carcinoma: urologic pathologists.
Hum Pathol
2001
.
32
:
74
80
.
Renshaw
,
A. A.
,
J. P.
Richie
,
K. R.
Loughlin
,
M.
Jiroutek
,
A.
Chung
, and
A. V.
D'Amico
.
The maximum diameter of prostatic carcinoma is a simple, inexpensive, and independent predictor of PSA failure in radical prostatectomies: results in a cohort of 434 patients.
Am J Clin Pathol
1999
.
111
:
641
644
.
Epstein
,
J. I.
,
C. R.
Pound
,
A. W.
Partin
, and
P. C.
Walsh
.
Progression following radical prostatectomy in men with Gleason score 7 tumor.
J Urol
1998
.
160
:
97
101
.

Author notes

Reprints: Andrew Renshaw, MD, Baptist Hospital of Miami, 8900 N Kendall Dr, Miami, FL 33176 ([email protected])