European Urology

1. Introduction

Almost 40 yr ago, Donald F. Gleason pioneered his widely acknowledged grading system of prostate adenocarcinoma [1], and [2]. In the following years, the reliability of the Gleason system in predicting the outcome of patients with prostate cancer (pCA) undergoing surgery, radiotherapy (RT), or surveillance has been proven in several single- and multi-institution studies [3], [4], [5], and [6]. In its original version, the Gleason system considered only the most predominant and, if present, the second most predominant Gleason pattern. A potentially present third Gleason grade pattern (TGG), regardless of the tumour percentage or state of aggressiveness, did not contribute to the Gleason score. In 2005, the International Society of Urological Pathology (ISUP) decided that in prostate biopsies, a more aggressive TGG should replace a less aggressive secondary grade because of the assumption that this would reflect the aggressiveness of the whole tumour more accurately. Interestingly, in RP specimens, it was acknowledged that the presence of an aggressive TGG might be associated with an adverse prognosis. However, it was stated that, for example, renaming a Gleason 4 + 3 pCA with a tertiary pattern 5 into a Gleason 4 + 5 tumour would overestimate the aggressiveness of the cancer. Therefore, it was concluded that in RP specimens, a TGG should be mentioned but should not redefine the Gleason score [7], and [8].

Recently, different investigators defined the prevalence of a TGG and demonstrated a significant association with adverse histopathologic features and/or the risk of biochemical recurrence (BCR) after definite therapy for clinically localised pCA [7], [9], [10], [11], [12], [13], and [14]. Limitations, however, arise in consideration of these study designs, most of which were retrospective and often limited by low statistical power resulting from small study populations. Furthermore, the data were derived from time intervals during which pathologic reporting of a TGG might not have been mandatory. Consequently, the prevalence and definition of the reported TGG differed substantially in the above-mentioned studies, ranging from 13–50%, which might hamper the assessment of their clinical impact. This was underlined by a recently published meta-analysis concerning the influence of a TGG on RP specimens. It was suggested that the available data are still too limited and too heterogeneous to affect the proposal of the 2005 ISUP conference on Gleason grading [15].

To address this void, we hypothesized that accurate assessment of TGGs in contemporary RP specimens may unequivocally demonstrate their association with acknowledged adverse histopathologic features.

2. Methods

3. Results

Clinical and histopathologic characteristics of the study cohort are provided in Table 1. Overall, a TGG was present in 180 of the 800 (22.5%) patients. In 89 cases (49.4%) the TGG was <5% of the whole tumour volume, whereas in 91 cases (50.6%) the volume was ≥5%. In 140 cases (78%), the tertiary Gleason component was higher than the two most predominant Gleason patterns, whereas a lower TGG was reported in 38 cases (21%). Two specimens (1%) showed both a higher and a lower third and fourth Gleason pattern with the same amount of tumour volume. A higher TGG was more frequently observed in well- or moderately differentiated tumours. An overview of frequencies for the different TGGs at various Gleason scores is given in Table 2. Regarding pT classification, TGG prevalence significantly increased with advancing pathology. It was present in 106 of 585 patients with pT2 tumours (18%), 48 of 168 pT3a tumours (29%), and 27 of 46 pT3b tumours (59%), respectively (p < 0.001). With respect to lymph node status, the presence of a TGG was more frequently observed in patients with LNI (23% pN0, vs 60% pN1; p < 0.001). Evaluating RP Gleason score, a TGG was observed in 32% of patients with a Gleason score ≤6 (76/239), in 10% of Gleason 3 + 4 (39/402), and in 41% of Gleason ≥4 + 3 (65/159), respectively (Table 3). To account for the fact that seven pathologists examined the RP specimens, we checked the data for a potential interobserver bias. The frequency of a TGG differed between investigators but was not statistically significant (range: 14–29%; p = 0.129).

Table 2 Frequencies of different tertiary Gleason grade patterns according to different Gleason scores. Various Gleason scores were put together into the Gleason groups <7, 3 + 4, and ≥4 + 3

*Both higher and lower tertiary Gleason grade patterns were found within the prostatectomy specimen.

In univariate analysis for the entire cohort, the presence of a TGG pattern <5% was not statistically associated with adverse pathology. Conversely, a TGG pattern ≥5% was a significant risk factor for ECE (odds ratio [OR]: 6.44; p < 0.001, PA: 61%), SVI (OR: 8.21; p < 0.001, PA: 68%), PSM (OR: 3.85; p < 0.001, 58%), and LNI (OR: 7.05; p < 0.001, PA: 70%), respectively (Table 4). In multivariate analyses, a TGG pattern ≥5% showed an independent correlation to the presence of ECE (OR: 3.04; p < 0.001, gain of PA: 0.5%) and PSM (OR: 2.16; p = 0.007, gain of PA: 1.2%) but did not reach statistical significance for SVI after correcting for preoperative PSA, primary and secondary Gleason grade, ECE, SVI, and PSM (Table 5).

In subanalyses, we addressed the impact of a TGG on adverse pathology at Gleason score <7 (including Gleason score 2 + 3, 3 + 2, 3 + 3, and 2 + 4) and 3 + 3. Although a significant association of TGG on ECE and PSM was seen for the whole Gleason group <7, this effect vanished at Gleason score 3 + 3 only (Table 6), for which the amount of the TGG was <5% in all cases. In another subanalysis addressing the effect of a TGG 5 on ECE, SVI, PSM, and LNI in univariate analysis for the entire study cohort, a significant association to ECE, SVI, and PSM was also only seen if the amount of the TGG was at least 5% (Table 7).

4. Discussion

Almost 40 yr after its introduction, the Gleason system is still a commonly used instrument for histopathologic grading of pCA. The Gleason system accounts only for the two most predominant Gleason patterns. However, pCA is a heterogeneous disease, and the presence of multiple different tumour foci with particulate variable grades of aggressiveness within one prostate are frequently observed [20]. Among other studies that addressed the appearance of a TGG [21], Gleason himself reported in 1992 that more than half of RP specimens contain more than two patterns [22]. Nevertheless, a potential adverse influence of TGG on patient outcome was first reported only a few years ago by Pan et al [11]. In that study, patients with an RP Gleason score 5–6 and an additive worse differentiated tertiary grade had a significantly higher rate of BCR compared to their counterparts without a TGG [11]. In the following years, more studies uniformly confirmed the adverse impact of TGGs with respect to unfavourable histopathologic stage and/or a higher frequency of BCR after RP. Hattap et al [9] evaluated 228 patients with an RP Gleason score 7 (3 + 4 or 4 + 3) after prostatectomy for the presence of a tertiary Gleason pattern 5 containing at least 5% of the whole tumour volume. The presence of a TGG was an independent predictor for treatment failure. Van Oort and co-workers [13] showed that the presence of a TGG was likewise an independent predictor of PSA recurrence after RP. Interestingly, this applied to more and less aggressive tertiary components than the two predominant grades, indicating that a tertiary grade—regardless of differentiation—is an expression of multifocality and, accordingly, of worse prognosis. Mosse et al [10] evaluated the influence of TGG 5 on adverse histopathologic features in a series of 223 RP specimens with a tertiary component compared to 603 cases without a tertiary grade. In their study, a Gleason pattern 5 of >5% of the tumour volume was reported as the secondary component. They found a significant correlation between the presence of TGG and advanced pathological stage at Gleason score 3 + 4 and 4 + 3 but not for Gleason score 4 + 4. They concluded that the influence of a TGG is strongest at lower Gleason scores but diminishes for more aggressive tumours because of naturally adverse biological behaviour. Whittemore et al [14] evaluated 214 RP specimens with Gleason score 3 + 4 or 4 + 3 that were reexamined by two blinded pathologists for the presence of a tertiary Gleason 5 pattern. Patients with a TGG showed statistically significantly higher pathological tumour stage and a significantly decreased BCR-free survival.

Limitations, however, to most of the aforementioned studies are that data were retrospectively analysed and that the definition of TGG was not invariably consistent, which may impede the interpretation and reproducibility of the results. Additionally, the numbers of patients were relatively small in most of the studies, alleviating the statistical expressiveness, especially for subanalyses.

To the best of our knowledge, our study is the first one in which the prevalence of a TGG in RP specimens has been evaluated in a single-centre, contemporary, consecutive patient population. The prevalence of TGG was 22.5% in our study and is accordingly lower than in most of the above-mentioned studies, where frequencies ranging from 15% to >50% were reported [21], [22], and [23] This discrepancy might be partly explained by different study designs (eg, only the index tumour was further evaluated).

In our study, the prevalence of TGG increased significantly with higher tumour stage (p < 0.001), which is in accordance with previously reported results [13], and [24]. Recently, it was suggested that in multiracial populations, special subgroups may have unfavourable pCA characteristics at the time of RP [25]. Although we did not explicitly document race, we present data of a homogenous Caucasian referral group with a very low percentage of Asians and blacks, surely not exceeding 1% of the whole group, which corroborates the homogeneity of our study population.

In subanalyses, we stratified patients into TGGs <5% versus ≥5% of the whole tumour volume. Interestingly, in uni- and multivariate analysis, a significant association with adverse pathology was only seen for TGG volume ≥5%, while volume <5% showed no significant correlation. This finding was emphasized by further subanalyses. Although a TGG for the entire Gleason group <7 (including Gleason score 2 + 3, 3 + 2, 3 + 3, and 2 + 4) showed significant association to ECE and PSM, this effect vanished for Gleason 3 + 3 only, for which the volume of TGG was always <5%. This also applied for subanalyses on TGG 5 to adverse pathology for the entire cohort. Again, a significant correlation was only detected when the TGG volume exceeded 5%, suggesting that a certain threshold of TGG must be exceeded to exert a meaningful influence. These findings are in contrast with the results of Mosse et al [10], where a tertiary volume <5% showed significant association with higher tumour stage. However, in their study, only the index tumour was considered, so our results may not be entirely comparable. Recently, Sim et al analysed the influence of a TGG 5 at Gleason score 7 on pathologic stage and time to BCR after RP [24]. Although a significant association with BCR was noticed for the entire cohort, on subanalyses for Gleason 3 + 4 + 5 and 4 + 3 + 5 versus their respective counterparts without TGG, both groups showed a trend towards a higher risk of BCR without reaching statistical significance. However, in their series, the authors did not consider the volume of the tertiary Gleason pattern, which may explain the lower significance. Nevertheless, this finding nicely demonstrates that the phenomenon of a TGG is complex, because an adverse association at a certain Gleason score must not necessarily withstand subanalysis. This is eminently interesting for reproducing the findings of van Oort et al [13], who stated that any TGG, regardless of its pattern, is an indicator for treatment failure. In our series, a worse tertiary grade was reported overall in 78% of all TGGs. Further substratification into higher versus lower tertiary grade and tumour volume <5% versus ≥5% would have ended in very small figures with accordingly reduced statistical power and so was not conducted. Therefore, the assumption that a better-differentiated TGG has a negative impact on patient outcome must be confirmed in further studies before a definite conclusion can be drawn, because the study by van Oort et al is the only one of its kind so far that addresses this issue.

In contrast to the results of other studies in which a TGG was not significantly associated with LNI, in our series, a TGG ≥5% of the tumour volume showed in univariate analysis a significant association to positive lymph nodes (p < 0.001). However, our overall number of patients with positive lymph nodes was only 30 (3.8%), and we did not perform an extended lymphadenectomy in most of the cases, so this finding should be interpreted with caution. Accordingly, we decided not to include LNI in multivariable analysis because we also included outcome parameters in the multivariate analyses. Inclusion of LNI would have lowered the number of cases for every parameter and was therefore not conducted.

Limitations of our study are, first, a relatively small number of overall TGGs, which consequently lowers the statistical power, especially for subanalysis. Second, no follow-up data are available to address the prognostic significance of TGGs on biochemical failure or survival. Finally, histopathologic evaluation of the RP specimens was performed by several pathologists, and each specimen was evaluated only once; therefore, interobserver percentage of agreement for classifying the same RP specimen cannot be given.

5. Conclusions

A TGG in RP specimens is significantly associated with advanced pathology. Our data indicate that a certain amount of TGG must be exceeded to affect pathological stage. A potentially negative impact of a TGG in RP specimens on patients’ overall survival remains to be evaluated in further studies.