SEARCH

SEARCH BY CITATION

Keywords:

  • prostate cancer;
  • ISUP Gleason score;
  • needle biopsy;
  • prostatectomy;
  • biochemical failure

Abstract

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. MATERIALS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. ACKNOWLEDGEMENTS
  8. CONFLICT OF INTEREST
  9. REFERENCES

OBJECTIVE

To determine whether the 2005 International Society of Urologic Pathology (ISUP) Gleason Grading Consensus is clinically more useful than the conventional Gleason score (CGS), we compared the CGS and ISUP GS (IGS) of prostate needle biopsy (NB) and radical prostatectomy (RP) specimens, and evaluated the prognostic value of the ISUP GS.

PATIENTS AND METHODS

Of 250 patients undergoing RP, 103 with clinical stage T1–2 N0M0 were enrolled. Pathological tumour grades of NB and RP specimens were classified according to CGS by experienced pathologists in the central pathology department of our hospital, and retrospectively according to IGS by one uropathologist at the central pathology department of another hospital. All patients had RP with no neoadjuvant or adjuvant therapy. We analysed associations of CGS and IGS with biochemical recurrence-free survival (BRFS) after RP.

RESULTS

The concordance rates between NB and RP specimens by CGS and IGS were 64.1% and 69.9%. Under-grading and over-grading rates by CGS and IGS were 28.2% and 7.8% for NB, and 27.2% and 2.9% for RP, respectively. There was a significant difference in the over-grading rate between CGS and IGS (P = 0.026). When CGS and IGS of NB and RP specimens were compared, the concordance rates were similar, at 67% and 69.9%. The IGS was higher, by 15.6% in NB and by 20.4% in RP specimens, than CGS. Patients were divided into three groups based on IGS of NB specimens (≤6, 7 and ≥8). These groups differed significantly in BRFS after RP (P = 0.022); CGS showed no such association.

CONCLUSIONS

The IGS of NB specimens were significantly associated with BRFS after RP. The ISUP system is thus clinically useful for determining the most appropriate treatments for patients with early-stage prostate cancer.


Abbreviations
ISUP

International Society of Urologic Pathology

RP

radical prostatectomy

(C)(I)GS

(conventional) (ISUP) Gleason score

NB

needle biopsy

BRFS

biochemical recurrence-free survival.

INTRODUCTION

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. MATERIALS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. ACKNOWLEDGEMENTS
  8. CONFLICT OF INTEREST
  9. REFERENCES

Prostate cancer is one of the most common neoplasms in men in the USA and other Western countries. The current widespread use of serum PSA levels and prostate biopsy has increased the detection rate of prostate cancer, resulting in a considerable shift toward earlier stages [1]. Furthermore, it is very important to diagnose prostate cancer at an early stage, when curative treatment is most likely. More patients are now choosing radical prostatectomy (RP), external beam radiotherapy, brachytherapy and active surveillance. In planning these treatments, the primary prognostic factors are the Gleason score (GS) of needle biopsy (NB) specimens, serum PSA level and the DRE findings [2,3].

Among the many factors influencing whether curative treatment or active surveillance is the best option, the GS of NB specimens is more important for patients with early-stage prostate cancer. However, it has been shown that the GS from NB specimens has an inherent sampling error and differs from the GS of RP specimens [4]. The most frequent discordance has been under-grading of the NB GS, although rates of under-grading have gradually decreased since the early 1990s [5]. Some reports have indicated that extended prostate biopsy significantly improves the GS concordance between NB and RP specimens [6,7]. Nevertheless, clinicians should be aware of the GS discrepancy between NB and RP specimens when determining the most appropriate treatments for their patients, especially those with early-stage prostate cancer.

The 2005 International Society of Urologic Pathology (ISUP) Gleason Grading Consensus was proposed to achieve consensus in specific areas of Gleason grading [8]. In particular, the ISUP Gleason grading for NB and RP specimens, the representative points of which are described in below, was altered. There have been no reports evaluating ISUP Gleason grading which include PSA (biochemical) recurrence-free survival (BRFS) after RP. In the present retrospective study, we compared conventional GS (CGS) and ISUP GS (IGS) of NB and RP specimens. Furthermore, we analysed the relationships of CGS and IGS with BRSF after RP, and thereby evaluated the clinical usefulness of IGS specifically in NB specimens.

MATERIALS AND METHODS

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. MATERIALS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. ACKNOWLEDGEMENTS
  8. CONFLICT OF INTEREST
  9. REFERENCES

In all, 250 patients with prostate cancer diagnosed as clinical stage T1–2 N0M0 received RP at Yokohama City University Hospital from January 1996 to December 2006. Serum PSA levels were determined with a third-generation PSA assay kit. PSA failure after RP was considered to be present when PSA was increased by >0.2 ng/mL. Before RP, all patients had TRUS-guided NB transperineally with 8–12 cores, to confirm the diagnosis of prostate cancer. Clinical stage was determined using MRI, CT and bone scintigraphy.

Of these 250 patients, 103 who did not have neoadjuvant or adjuvant therapies, e.g. hormonal or radiotherapy, were enrolled in the study. Pathological tumour grades of the specimens obtained from NB and RP were classified according to CGS by several pathologists in the central pathology department of Yokohama City Hospital. Later, these NB and RP specimens were classified according to IGS by one uropathologist at the central pathology department of Yokohama City University Medical Center.

In the present study, we used the ISUP Consensus Gleason Grading System [8]. The differences between the ISUP and CGS are shown in Table 1. One pathologist determined the pathology of the dominant tumour nodules, based on rules 7 and 8 in Table 1.

Table 1.  Differences between conventional and 2005 ISUP Gleason scoring
CGS2005 IGS
A diagnosis of GS <4 possible on NBGS of NB specimens <4 rarely if ever made.
A partial cribriform pattern, large cribriform, is diagnosed as Gleason pattern 3Most cribriform patterns would be diagnosed as Gleason pattern 4 while specimens with only rare cribriform lesions would satisfy the diagnostic criteria for cribriform pattern 3.
The same GS are used for NB and RP specimensDifferent GS used for NB and RP specimens.
High-grade tumour of small quantity (<5%) on NB should be excluded based on GS (5% threshold rule)High-grade tumour of any quantity on NB should be included within the GS.
Tumours on NB should be graded by listing the primary and secondary patterns, i.e. excluding tertiary patternThe GS of RP specimens should be assigned based on the primary and secondary patternsFor the tertiary pattern on NB specimens, both the primary pattern and the highest grade should be recorded.For RP specimens, the pathologist should assign the GS based on the primary and secondary patterns + a comment on the tertiary pattern
Separate or overall scoring to assess all grades of NB specimens are usedWhen NB specimens show different grades in separate cores, individual GS should be assigned to these cores (separate scoring).
The grade of the largest portion should be assigned even if the second largest portion is of higher gradeWhen RP specimens show different grades in separate tumour nodules, a separate GS should be assigned to each of the dominant tumour nodules

The results were assessed statistically using the Mann–Whitney U-test and chi-square test. BRFS was determined by the Kaplan-Meier method, and the significance of differences determined by the log-rank test; in all tests P < 0.05 was considered to indicate a statistically significant difference.

RESULTS

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. MATERIALS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. ACKNOWLEDGEMENTS
  8. CONFLICT OF INTEREST
  9. REFERENCES

The characteristics of the 103 who had not received adjuvant therapies before or after RP are shown in Table 2. Using the risk classification of D’Amico et al.[9], the 103 patients were divided into three groups, i.e. 34 low-risk (33%), 37 intermediate-risk (36%) and 32 high-risk (31%).

Table 2.  The characteristics of the 103 patients
VariablesMedian (95% CI)
Observation period, years 2.1 (1.92–2.40)
Age, years67 (66–67)
Serum PSA level at RP, ng/mL 9.2 (10.7–14.1)
Clinical stage
 cT151 (49.5)
 cT251 (49.5)
 Unknown 1 (1.0)
% positive biopsy cores25.0 (22.0–28.6)
GS NB by CGS
 ≤648 (46.6)
 737 (35.9)
 8–1018 (17.5)
Risk classification by CGS
 Low34 (33.0)
 Intermediate37 (35.9)
 High32 (31.1)

Correlations between the CGS and IGS of NB and RP specimens are shown in Table 3. The CGS for the NB specimen was concordant with that of the RP specimen in 66 cases (64%). The under-grading rate was 28% (29 cases) and the over-grading rate was 8% (eight cases) for CGS. IGS yielded an under-grading rate of 27% (28 cases), similar to the CGS results. The concordance rate for IGS was slightly higher (70%) than that for CGS, while the over-grading rate was significantly lower with IGS (3%, three cases) than with CGS (8%, eight cases; P = 0.026).

Table 3.  Correlation of CGS and IGS results between NB and RP specimens
NBRP GS, n
≤678–10
CGS
NB GS
 ≤62620 2
 7 327 7
 8–10 0 513
IGS
NB GS ≤62318 1
 7 329 9
 8–10 0 020

Next, we compared the CGS and IGS of NB and RP specimens (Table 4). Although the concordance rates for NB and RP specimens by CGS and IGS were similar, at 67% and 70%, the IGS were higher, by 16% in NB and by 20% in RP specimens, than CGS. However, there were no significant differences between CGS and IGS (P = 0.469 for NB and P = 0.329 for RP, Mann–Whitney U-test). Over-grading rates for NB and RP specimens were 13% and 10% by CGS and IGS, respectively.

Table 4.  Comparison of CGS and IGS of NB and RP specimens
CGSIGS, nTotal
≤678–10
  1. Mann–Whitney U-test, P = *0.469, †0.329.

NB*
CGS
 ≤63414048
 7822737
 8–10051318
 Total424120103
RP
CGS
 ≤61811029
 78341052
 8–10022022
 Total264730103

Table 5 shows the results grouped by the risk categories of D’Amico et al.[9]; 10 (29%) of the 34 categorized as low-risk, based on IGS, were intermediate-risk according to CGS results. Five (14%) and four (11%) of the 37 intermediate-risk, based on CGS, patients were low- and high- risk, respectively, by IGS. However, there were no statistically significant differences in risk classification between CGS and IGS (P = 0.678, Mann–Whitney U-test).

Table 5.  Risk classifications with CGS and IGS
RiskclassificationIGS, nTotal
LowIntermediateHigh
  1. Mann–Whitney U-test, P = 0.678.

CGS
 Low2410034
 Intermediate528437
 High042832
 Total294232103

Clinical staging showed that T1c and T2 each accounted for half the patients; we compared the components of RP pathological staging by CGS and IGS of NB specimens, divided into two groups, GS ≤6 and ≥7. As shown in Table 6, pathological staging by CGS showed predominantly pT2 and pT3a lesions. Among the patients with CGS ≤6 for NB specimens, 37% (pT2) had organ-confined lesions, while 26% of those with CGS ≥7 had organ-confined lesions. However, by contrast with the CGS results, IGS showed an increase in GS ≥7 cases (33%) among those with pT2 tumours, and similar results were obtained for pT3 cases, i.e. 20% in pT3a and 7% in pT3b. However, none of the differences in pathological staging between CGS and IGS were statistically significant (P = 0.198, chi-square test).

Table 6.  Comparison of NB GS and pathological staging of RP specimens
NB GSPathological stageTotalP chi-square
pT2pTapTb
CGS    0.198
 ≤63710148 
 ≥72619651 
IGS
 ≤630 9039 
 ≥73320760 

As shown in Fig. 1A, BRFS in patients with NB specimens classified by CGS did not differ significantly among the three GS groups (≤6, 7 and ≥8). However, when IGS was applied to re-evaluate NB specimens, there was a significant difference in BRFS among these three groups (P = 0.022; Fig. 1B). Likewise, the BRFS in RP specimens, classified by CGS or IGS, were analysed. There were no significant differences in BRFS among the three groups when RP specimens were classified by CGS. However, as when IGS was used to classify NB specimens, the BRFS differed significantly among the three groups (P = 0.045; data not shown).

image

Figure 1. Associations of BRFS according to GS from NB specimens graded by CGS and IGS. (a) In three groups, i.e. CGS ≤6, 7 and ≥8, BFRS was evaluated after RP. (b) In three groups, i.e. IGS ≤6, 7 and ≥8, BFRS was evaluated after RP.

Download figure to PowerPoint

DISCUSSION

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. MATERIALS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. ACKNOWLEDGEMENTS
  8. CONFLICT OF INTEREST
  9. REFERENCES

The present study showed that nearly a quarter of NB specimens were under-graded in comparison with RP specimens, according to both CGS and IGS evaluations. The GS concordance rates between NB and RP specimens in this study were 64% and 70% by CGS and IGS, respectively, rates similar to those of previous reports (30–74%) [10]. Although we anticipated increased GS up-grading with IGS in both NB and RP specimens, the up-grading rates were only 16% and 20%, respectively. Interestingly, the GS classification of NB by IGS reflected BRFS after RP, although that by CGS did not. Thus, IGS is more reliable for treatment decision-making, especially for early-stage prostate cancer managed with RP, brachytherapy, external beam radiotherapy, hormonal therapy or active surveillance [11].

It was previously reported that the NB pathological findings of focal prostate cancers are not a reliable surrogate indicator, and thus should not influence treatment decisions. One explanation is that prostate cancer is a multifocal disease with satellite tumours [12]. Horinger et al.[13] reported that 52 RP specimens had multifocality, showing altogether 196 foci of prostate cancer (the mean number of cancer foci per specimen was 3.76). Other earlier reports also showed that most prostate cancers with high or low tumour volumes were multifocal [14,15]. In these reports, more intriguing results were documented, i.e. very small prostate cancer foci had high GS [13,15]. This might explain why the CGS of NB fails to predict the CGS of RP specimens. CGS does not indicate tertiary cancer lesions or small cancer foci in NB specimens with relatively high Gleason grades, while IGS indicates the primary pattern and the highest grade even in cancer lesions which have small volumes in NB specimens. The present observation that IGS of NB specimens showed significant differences in BRFS after RP suggests the importance of documenting the highest GS in NB.

Khoddami et al.[16] reported that primary Gleason pattern 4 had predictive value in patients with Gleason score 7 tumours after RP, meaning that a higher Gleason pattern is important for the prognosis of localized tumours after RP or radiotherapy. From the current results we conclude that the influence of IGS on clinical outcome reflects the grading of NBs based even on minimal findings of high-grade tumour cells. We speculate that Gleason himself might not have recognized the importance of small areas of a third pattern, especially those with the highest grade pattern [17]. However, unlike in the Gleason era, many thin cores from different sites in the prostate can currently be obtained by sextant needle biopsies in patients with early-stage disease, e.g. T1c tumours. Recently, to determine optimum treatments for localized prostate cancer, use of the Partin table, which predicts tumour extension by combining PSA level, clinical stage and GS, has become widespread [18]. Hence, the Partin table might be influenced by IGS, and accumulation of more pathological data on RP specimens will be needed to obtain a new ‘Partin table’. Although urologists should prospectively consider differences in NB pathology between CGS or IGS, in practice it would be better to use IGS to obtain of the newest and most reliable nomogram for predicting the outcome of this disease. It is well understood that taking more biopsy cores improves the accuracy of the GS obtained by biopsy in predicting the final GS at RP [9,19]. Furthermore, Antunes et al.[20] reported that the percentage of positive biopsy cores reflected the biochemical outcome after RP. In addition, the present results show pathological grading by IGS to reconfirm the accuracy of the biopsy GS before applying treatments for localized prostate cancer.

We showed that 27% more NB than RP specimens were under-graded by IGS, while 28% of NB specimens were under-graded by CGS. The under-grading rate with IGS in our data is consistent with those of earlier reports using CGS [10,21,22]. Previously, GS under-grading of NB specimens was the most prevalent error in comparison with GS grading of RP specimens. Even when the same pathologist assessed both samples, a third of them were still under-graded [22]. Although in the present study the same pathologist evaluated both NB and RP specimens by IGS, the under-grading rate was no less than that by CGS (Table 2). The study of King [23] documented that biases in pathological interpretation and sampling effects contribute to grading discordance. A bias problem in pathological interpretation might be resolved by the widespread use of IGS, especially introducing an assessment of NB specimens.

Fukagai et al.[24] indicated that grading errors were most frequent in well-differentiated carcinoma, and that most involved under-grading with poorer NB-RP correlations. They suggested that pathologists tend to under-grade biopsies of moderately differentiated carcinoma as well-differentiated carcinoma. Furthermore, a recent study, examining trends in GS over a period of 15 years, showed that tumours reported as GS ≤6 in NB specimens are prone to be under-graded, while those reported as GS 8–10 are prone to be over-graded [5]. In the light of the change in the GS reporting trend between 1992 and 2006, a progressive increase in Gleason grade 3 and a decrease in Gleason grade 2 were confirmed [25], showing a tendency for an increase in the incidence of moderately differentiated tumours and a decrease in that of well-differentiated tumours. Based on those analyses, the reduction in GS discrepancy between NB and RP specimens would be expected from the redefinition of the Gleason grading system for NB specimens [26,27]. In this respect, the IGS system provides a new definition, described above. The Consensus Conference on Gleason Grading of Prostatic Carcinoma recommended that the GS be the sum of the most common and the highest grade patterns, and that cribriform patterns be diagnosed as Gleason pattern 4 with only rare cribriform lesions satisfying the diagnostic criteria for cribriform pattern 3 [8]. According to the IGS system, under-grading of NB, i.e. GS ≤6, would decrease markedly, while over-grading, i.e. GS >6, would increase.

In conclusion, although the GS under-grading rates in NB and RP specimens did not differ between the CGS and IGS systems, the GS of NB specimens determined by IGS were associated with a significant difference in BRFS after RP, with GS scores similar to those determined using RP specimens. Given that an extended NB scheme and/or the development of a more suitable ultrasound device is expected, the IGS system would enhance the accuracy of histological assessments of prostate cancer. This will allow physicians caring for patients with prostate cancer to optimize therapies such as RP, brachytherapy, external beam radiotherapy and active surveillance.

ACKNOWLEDGEMENTS

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. MATERIALS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. ACKNOWLEDGEMENTS
  8. CONFLICT OF INTEREST
  9. REFERENCES

This study was supported by a Grant-in-Aid for Scientific Research (C) from the Ministry of Education, Culture, Science and Technology, a grant for the 2007 Suzuki Foundation, Yamaguchi Endocrine Disease Foundation, the Third-term Cancer Control Strategy Program from the Ministry of Health, Labor and Welfare, and a grant for 2008 Strategic Research Projects (No. K20003 and K20010) of Yokohama City University, Japan.

REFERENCES

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. MATERIALS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. ACKNOWLEDGEMENTS
  8. CONFLICT OF INTEREST
  9. REFERENCES