|Home | About | Journals | Submit | Contact Us | Français|
• To investigate pathological and short-term outcomes since the most recent Gleason system modifications by the International Society of Urological Pathology (ISUP) in an attempt to divide the current Gleason grading system into prognostically accurate Gleason grade groups.
• We queried the Johns Hopkins Radical Prostatectomy Database (1982–2011), approved by the institutional review board, for men undergoing radical prostatectomy (RP) without a tertiary pattern since 2004 and identified 7869 men.
• Multivariable models were created using preoperative and postoperative variables; prognostic grade group (Gleason grade ≤6; 3 + 4; 4 + 3; 8; 9–10) was among the strongest predictors of biochemical recurrence-free (BFS) survival.
• Significant differences were noted among the Gleason grade groups at biopsy; differences were noted in the race, PSA level, clinical stage, number of positive cores at biopsy and the maximum percentage of positive cores among the Gleason grade groups at RP.
• With a median (range) follow-up of 2 (1–7) years, 5-year BFS rates for men with Gleason grade ≤6, 3 + 4, 4 + 3, 8 and 9–10 tumours at biopsy were 94.6, 82.7, 65.1, 63.1 and 34.5%, respectively (P < 0.001 for trend); and 96.6, 88.1, 69.7, 63.7 and 34.5%, respectively (P < 0.001), based on RP pathology.
• The 2005 ISUP modifications to the Gleason grading system for prostate carcinoma accurately categorize patients by pathological findings and short-term biochemical outcomes but, while retaining the essence of the Gleason system, there is a need for a change in its reporting to more closely reflect tumour behaviour.
• We propose reporting Gleason grades, including prognostic grade groups which accurately reflect prognosis as follows: Gleason score ≤6 (prognostic grade group I); Gleason score 3+4=7 (prognostic grade group II); Gleason score 4+3=7 (prognostic grade group III); Gleason score 4+4=8 (prognostic grade group (IV); and Gleason score 9–10 (prognostic grade group (V).
The Gleason grading system for prostate cancer was developed in the 1960s and remains the strongest predictor of outcome for men diagnosed with prostate carcinoma . The current Gleason grading system has evolved from its original description, first modified by Gleason and Mellinger in 1974  and most recently in 2005 by the International Society of Urological Pathology (ISUP) . With the new modifications, a number of studies indicate that there is better correlation between biopsy and radical prostatectomy (RP) Gleason grade and stage, as well as with biochemical recurrence, than previously [4–8].
These newest modifications when applied to needle biopsies, while demonstrating improved correlation with RP grade, stage, margins, tumour volume and postoperative progression, are also more complicated than previous versions [4–9]. For instance, in the setting of high grade cancer on needle biopsy, if there is also low grade cancer on the same core occupying less than 5% of the cancer, the low grade cancer is not included in the Gleason score. The tertiary highest grade pattern is included in the biopsy score yet not the RP score. In several previous studies, we have shown that tertiary Gleason pattern has a risk of recurrence that lies in between the current grading system gradations; for example, Gleason score 6 with a tertiary higher grade component has a prognosis between Gleason score 6 and Gleason score 3 + 4 = 7 . The purpose of the present study was to investigate the pathological findings and short-term biochemical outcomes of tumours graded since the ISUP Consensus in an attempt to divide the current Gleason grading system into prognostically accurate Gleason grade groups.
The institutional-review-board-approved, Johns Hopkins Radical Prostatectomy Database (1982–2011) was queried for men undergoing RP since 2004, which was identified as the time of development and implementation of the most recent grading guidelines. A total of 7869 were identified, of whom 19 men were missing complete biopsy data and were excluded from the analysis of biopsy Gleason score. Pathological analysis of the RP specimen has been described previously [10,11]. Briefly, specimens were serially sectioned and submitted in entirety and reviewed by general surgical pathologists, with daily consultation of urological pathologists for issues relating to grade, stage and margins. In cases where there are three patterns on a biopsy core (typically Gleason patterns 3, 4 and 5), the Gleason score is derived by adding the most common pattern and the highest pattern as recommended . The following Gleason score groups were considered: ≤6; 3 + 4 = 7; 4 + 3 = 7; 8; and 9–10. In cases with multiple cores with differing Gleason scores, the highest score in a case was selected as the case's grade. In analysis of RP specimens, intermediate categories were created by adding the designation ‘T’ to each existing category when a higher tertiary pattern was present. Gleason score ≤6 could have a tertiary pattern 4 or 5, while T refers to tertiary pattern 5 only for the remaining Gleason score groups. The RP grade assigned was based on the tumour nodule with the highest grade, rather than an overall grade for all of the tumour in the prostate. In a minority of cases, pathological stage was unknown as a result of intraprostatic incision, which accounts for the discrepancy between the number of cases in Tables 1 and and22 compared with Table 3.
We used Kaplan–Meier analysis with a log-rank test to determine biochemical recurrence-free survival (BFS) among categories. Biochemical recurrence was defined as any postoperative PSA > 0.2 ng/mL. Univariate and multivariate regression models were created to determine predictors of BFS. Only significant variables in univariate analysis were included in the multivariate analysis. Concordance (c) indices were calculated for each model. For the multivariate analysis, the maximum percentage of positive cores (PPC) was assessed as >50 vs ≤50%.
Of the 5205 men with Gleason score ≤6 tumour on biopsy, there was only one with Gleason score 4 and only seven with Gleason score 5; the remaining 99.8% were Gleason score 6. Notably, significant differences in PSA, clinical stage, and number of positive cores were found among the Gleason grade groups at biopsy (Table 1). Patient and prostate cancer characteristics relating to RP findings are depicted in Table 2. The relationship between biopsy Gleason score and RP stage is shown in Table 3. With a median (range) follow-up of 2 (1–8) years, BFS improved incrementally with more well-differentiated tumours (Fig. 1, top). In multivariable analysis, Gleason score was among the strongest preoperative predictors of BFS (Table 4). The 2-year BFS rates for men with biopsy Gleason scores ≤6, 3 + 4, 4 + 3, 8 and 9–10 were 97.1, 90.6, 79.9, 70.9, and 51.5%, respectively (P < 0.001 for trend; P = 0.057 for Gleason score 4 + 3 vs 8 [Fig. 1]).
Of the 3548 men with Gleason score ≤6 tumours as their RP index tumour grade, there was one with Gleason score 4 and there were 44 with Gleason score 5; therefore, 98.7% of RP Gleason score ≤6 tumours were Gleason score 6. Notably, significant differences were found in race, PSA level, clinical stage, number of positive cores at biopsy and the maximum PPC among the Gleason grade groups at RP such that, in general, men with a higher Gleason score at RP were more often African–American, had higher PSA levels, more often had a palpable abnormality on DRE, had more positive cores at biopsy and a greater PPC (Table 2). The relationship of pathological stage to RP Gleason score is shown in Table 3. BFS rates are shown in Fig. 1 and are consistent with the comparison of pathological stage with a higher Gleason score corresponding to worse BFS. The 2-year BFS rates for RP Gleason scores ≤6, ≤6 + T, 3 + 4, 3 + 4 + T, 4 + 3, 4 + 3 + T, 8, 8 + T and 9–10 were 98.8, 97.2, 93.6, 90.3, 85.6, 73.3, 73.7, 60.5 and 58.5%, respectively (P < 0.001 for trend; Fig. 1, bottom). When postoperative variables were assessed in multivariable analysis, RP Gleason grade groups were also among the strongest predictors of BFS (Table 4). The c-index for the complete model was 0.8978. A similar model was created considering only primary Gleason score and ignoring tertiary Gleason patterns (not shown). The c-index for the model excluding tertiary patterns was 0.8970. Based on c-index calculations, considering tertiary Gleason pattern only nominally increases the predictive value of the model and the model excluding tertiary patterns can be considered as predictive as the complete model.
The Gleason scoring system has proved to be a robust and durable method for the grading of prostate carcinoma. Through several iterations its predictive ability has been improved and, as a consequence, it has become necessarily more complex. The complexity of the newest version of the Gleason grading system can lead to confusion among pathologists, urologists and patients alike. For instance, Gleason score 6 is typically recommended as the lowest grade to be assigned on biopsy material, however, the Gleason scale ranges from 2 to 10, such that patients are unduly concerned when told that they have Gleason score 6 cancer on biopsy, logically but incorrectly assuming that their tumour is in the mid range of aggressiveness. Another consequence of the modified grading system is that there is an expanded definition of Gleason pattern 4 to include a broader range of histological patterns. The original Gleason system restricted Gleason pattern 4 to cases with irregular cribriform architecture and fused glands. In the modified system, almost all cribriform patterns were considered Gleason pattern 4 along with poorly formed and fused glands. Subsequent to the initial publication of the modified system, data have been presented that all cribriform glands should be diagnosed as cribriform pattern 4 . Even in a highly selected set of images thought to be the best candidates for cribriform pattern 3, most experts interpret cribriform patterns as pattern 4 . Furthermore, 73% of the candidates of cribriform pattern 3 were associated with more definitive pattern 4 elsewhere on the needle biopsy specimen. There was poor reproducibility amongst experts as to cribriform pattern 3 vs pattern 4. In a more recent study, specifically addressing the prognosis of cribriform prostate cancer glands, both small and large cribriform glands were equally linked to progression after RP .
There are several prognostic consequences of the reclassification of many former Gleason score 6 tumours to Gleason score 7 in the modified system. Gleason score 6 tumours are currently more homogeneous and have a uniformly better prognosis. For example, we have shown that virtually no pure Gleason score 6 tumour is associated with progression after RP, whereas when using the original Gleason system this occasionally occurred . In the present study, 97.1 and 98.8% of patients with pure Gleason 6 cancer at biopsy and RP, respectively, were biochemically free of tumour 2 years after RP. Although other variables must be factored in, such as extraprostatic extension and margin status in the RP and serum PSA level and clinical stage on biopsy, in the modified Gleason system, Gleason score 6 has an excellent prognosis.
The difference between Gleason scores 3 + 4 and 4 + 3 has been well investigated for both post-RP series and studies of biopsies [15–22]. While many of these studies have divergent findings, the majority demonstrate worse pathological stage and biochemical recurrence rates for men with Gleason scores 4 + 3 compared with 3 + 4 [16,17,19–21]; however, each of these studies focused on specimens before the reclassification of the Gleason system in 2005 and, furthermore, those published since 2005 are skewed by the inclusion of older samples. In the past, any Gleason pattern 4 tumour was considered to be aggressive, and many authors considered Gleason score 7–10 tumours to be high grade. Using the modified Gleason system, the present study showed that a Gleason score 3 + 4 = 7 tumour has a very favourable prognosis with estimated 2-year BFS of 90.6 and 93.6% for biopsy and RP, respectively. These results warrant the separation of Gleason score 7 into two prognostic groups. Further evidence against combining all Gleason score 7 tumours into one category is that, in the present study, Gleason score 4 + 3 = 7 carcinomas behaved more similarly to tumours with Gleason score 8 than they did to those with Gleason score 3 + 4 = 7.
With Gleason score 8 tumours, the prognosis worsens considerably, yet not to the level of Gleason score 9–10. Within the literature, Gleason scores 8–10 are typically considered as one grade category. In large part, this results from many RP series having relatively few patients with Gleason score 9–10; for the purposes of having enough cases for meaningful statistical analysis they are combined with the more numerous cases of Gleason score 8. Although Gleason score 9–10 tumours are not rare, most of the men with this tumour grade have advanced clinical disease such that they are not candidates for RP. Given the almost ubiquitous lumping together of Gleason scores 8, 9 and10 in the literature, including within commonly used predictive tables and nomograms, it is not surprising that some urologists consider Gleason score 8 as being prognostically the same as Gleason score 9–10. As the present study clearly shows, however, Gleason score 9–10 tumours have almost twice the risk of progression compared with Gleason score 8. At biopsy and RP, 70.9 and 73.7%, respectively, of men with Gleason score 8 tumours are free of biochemical recurrence at 2 years. There is a precipitous drop in predicted cure for men with Gleason score 9 and10 cancer at biopsy and RP, where only 66.7 and 58.5%, respectively, are free of biochemical recurrence at 2 years. Unfortunately, for the reasons stated above, predictive tables and nomograms overestimate the cure rate for Gleason score 9–10 tumours, by combining these cases with the more common better prognosis Gleason score 8 lesions. There are also numerous studies in the radiation literature as to the worse prognosis for men with Gleason score 9–10 tumours as opposed to Gleason score 8 .
In the present study and in previous studies, the presence of a tertiary higher grade component was associated with an increased risk of biochemical recurrence, typically raising the risk of recurrence to an intermediate level between those of cancers without a tertiary component in the same Gleason score category and cancers in the next higher Gleason score category. For example, Gleason score 3 + 3 = 6 with tertiary pattern 4 has a prognosis that is in between Gleason score 3 + 3 = 6 without tertiary pattern 4 and Gleason score 3 + 4 = 7. Gleason score 3 + 4 = 7 with tertiary pattern 5 has a prognosis intermediate between Gleason score 3 + 4 = 7 without a tertiary pattern 5 and Gleason score 4 + 3 = 7. The one exception is RP Gleason score 4 + 3 = 7 with tertiary pattern 5, which has progression rates more similar to Gleason score 8 [24,25]. Other studies have found a different prognostic effect of tertiary higher grade patterns. For example, Turker et al.  demonstrated that Gleason score 6 tumours with tertiary higher grade patterns behaved in a similar way to Gleason score 3 + 4 = 7 tumours. They also reported that Gleason score 3 + 4 plus tertiary pattern 5, Gleason score 4 + 3 plus tertiary pattern 5, and Gleason score 8–10 all had similar prognoses .
The present study failed to show that adding the tertiary component enhanced the predictive value in multivariable analysis factoring in preoperative serum PSA levels and pathological stage, margins and Gleason grade. Whittemore et al.  also found that the prognostic significance of tertiary pattern 5 in Gleason score 7 tumours dropped out in multivariable analysis. In summary, tertiary Gleason patterns are typically associated with higher pathological stage and biochemical recurrence than cancers with the same Gleason score without tertiary patterns. The relative increase in risk of recurrence associated with tertiary patterns differs between studies. It is currently recommended that tertiary grade patterns are noted in the pathology report to accurately reflect the grade, although it is questionable whether tertiary grade patterns add prognostic information to routinely reported variables
It could be argued that the numbering system of the Gleason grading system should be abandoned and replaced with a new grading system of I–V, reflecting Gleason scores ≤6; 3 + 4 = 7; 4 + 3 = 7; 8; and 9–10, respectively. By doing so, we would avoid the current grading system that to all intents and purposes begins with 6. The Gleason system, however, is deeply entrenched in the history and literature of prostate carcinoma – a quick Medline search of ‘Gleason score’, ‘Gleason sum’ or ‘Gleason grade’ reveals 5957 articles. A reassignment of the Gleason numbers to a simple I–V grading system would create mass confusion for patient care and when comparing data with older studies using the conventional Gleason grading numbering, which goes from 2 to 10. The issue remains of how to best manage and discuss the grading of prostate carcinoma. On the one hand, we have demonstrated that an accurate grouping of Gleason scores can be accomplished with five prognostic grade groups, as opposed to the individual nine Gleason scores. On the other hand, oversimplification of Gleason grade classification, such as by combining Gleason scores 8–10 or categorizing patients into low-, intermediate- and high-risk categories based on Gleason scores <7, = 7 and >7 loses critical prognostic information . We recommend reporting whenever possible in analyses within the literature the five prognostic grade groups that we have demonstrated to be prognostically distinct. If it is not feasible to do so because of insufficient numbers of patients within each prognostic grade group, this limitation should be acknowledged.
There are several limitations to the present study, the first being the short-term follow-up with regard to BFS and lack of cancer-specific survival data. Since the modified Gleason system was only implemented in 2005, the data are necessarily constricted, but the fact that the data are internally consistent with incrementally worsening prognosis for each prognostic grade group, as would be expected, makes it unlikely that the results would change significantly with longer follow-up. The Kaplan–Meier analysis and curves take into account the length of follow-up, and weight greater prognostic information for those patients with longer follow-up, but studies with longer follow-up using the modified Gleason system are necessary to verify our results. Another potential weakness is that we did not have data on all cases for the number of cores sampled since the majority of cases were biopsied in outside institutions. Consequently, we report the number of positive cores as opposed to the fraction of positive cores (number of positive cores/total cores). In a previous study of >3000 needle biopsy specimens analysed at our institution, between 10 and 15 cores were sampled in 67.5% of the cases, and the remaining 32.5% were evenly distributed between <10 and >15 cores with only one patient with a sextant biopsy. Given the relatively narrow range of number of cores sampled, we have found in our data virtually the same statistical results whether the number or fraction of positive cores is used .
In summary, the 2005 ISUP modifications to the Gleason grading system for prostate carcinoma accurately categorize patients by pathological findings and short-term biochemical outcomes. While retaining the essence of the Gleason system, there is a need for a change in its reporting to more closely reflect tumour behaviour. In reporting grades on biopsy and RP, we propose recording the Gleason score, and including prognostic grade groups (Table 5). One would still report a case, for example, as ‘Gleason score 9’ or as ‘Gleason score 10’, (rather than as ‘Gleason score 9–10’) along with the prognostic grade group V. Patients will, for example, be reassured that when diagnosed with a Gleason score 6 that their prognostic grade group is I out of V, not Gleason score 6 out of 10. The same would apply for a Gleason score 3 + 4 = 7 tumour where both the prognostic grade group (II) in line with their tumour's relatively less aggressive behaviour. At the other end of the grade spectrum, men with Gleason score 9–10 tumours will more accurately be considered to have more aggressive tumours than those with Gleason score 8, and this can be factored into their management.
What's known on the subject? and What does the study add?
Conflict of Interest