Time to Rethink Gleason Grading

Tumor grade is the strongest predictor of long-term prostate cancer outcomes and thus the most important factor in treatment decisions.¹ Prostate cancer grade has been traditionally reported using the Gleason score (GS), a system based on architectural tumor pattern first described in a 1967 study of 270 patients at the Minneapolis VA Hospital.² The Gleason score was defined as the sum of the two most common grade patterns (each pattern assigned a score from 1 to 5), yielding an overall GS ranging from 2 to 10.

Over time, several modifications have been made to the Gleason system to better inform clinical decision-making.³ Notably, Gleason pattern (GP) 1 and 2 are no longer used in practice, yielding a contemporary system in which Gleason scores range from 6 to 10.⁴ Another change was the recommendation to report the percentage of GP4 observed in Gleason score 7 tumors.⁵ Doing so draws an important distinction between patients with GS 3+4=7 cancers with, for example, <5% pattern 4, who may be candidates for active surveillance, and those with nearly 50% pattern 4 more likely to benefit from definitive treatment. This practice has been validated by subsequent data demonstrating a clear association between percentage GP4 and oncologic outcomes.^{6, 7} Finally, the increasingly complex Gleason score was translated to a more intuitive Grade Group (GG) classification ranging from 1 to 5.⁸

While the Gleason scoring system has evolved for the better, valid questions remain regarding the merit of a system based on proportions of tumor patterns rather than simply the volume of tumor known to drive clinical aggressiveness (i.e. GP4–5). In this issue of The Journal of Urology, Scuderi et al compared the prognostic value of percentage pattern 4 (i.e. percentage of overall tumor) with that of absolute volume of pattern 4 for adverse pathologic stage and biochemical recurrence (BCR) in patients undergoing radical prostatectomy (RP).

At the same time, the report draws attention to limitations of the current grading system. Take, for example, three patients diagnosed with prostate cancer on biopsy (Table 1). Patient 1 has a higher volume of aggressive tumor (i.e. pattern 4) than patients 2 and 3, yet by virtue of the current ratio-based system, patient 1 is assigned the lowest GS/GG. The implications of this system are apparent – the patient with the highest volume of aggressive cancer (patient 1) is categorized as favorable or unfavorable intermediate-risk and accordingly recommended for lower intensity management. Meanwhile, the patient with the lowest volume of aggressive cancer and overall cancer (patient 3) is categorized as high-risk and recommended for more intensive therapies.^{9, 10} These cases highlight the inherent flaw in the use of ratios to assign overall grade.

Table 1.

Gleason Score and Grade Group based on biopsy pathology findings

Patient	GP3	GP4	Total Tumor	% GP3	% GP4	GS	GG
1	8 mm	4 mm	12 mm	67%	33%	3+4=7	2
2	1 mm	2 mm	3 mm	33%	67%	4+3=7	3
3	0 mm	1 mm	1 mm	0%	100%	4+4=8	4

GP=Gleason Pattern, GS=Gleason Score, GG=Grade Group

Indeed, Scuderi and colleagues found that absolute volume of GP4 provides better discrimination for advanced pathologic stage and BCR than percentage of GP4 or overall GS/GG. Furthermore, they demonstrated that considering the volume of GP3 does not meaningfully improve discrimination for these outcomes. The authors should be commended for this thoughtful work, which raises other critical questions regarding current practice.

Should Gleason pattern 3 be called cancer?

There has been much discussion about whether Gleason score 6 should be labeled as cancer. A natural extension of this question is whether GP3 should be called cancer. More specifically, does consideration of GP3 add value to a cancer diagnosis in men with intermediate-risk disease? If the answer is no, as suggested by the current analysis, then labeling GP3 as cancer may not be justified. On the other hand, historical data suggest that the percentage of positive biopsy cores, including GS6, should be considered in risk-stratification.^{11, 12} These findings support the American Urological Association (AUA) guideline definition of favorable intermediate-risk disease, which requires that fewer than half of biopsy cores are positive for cancer.⁹ However, these studies largely included patients treated before the 2005 International Society of Urological Pathology (ISUP) modifications, and few of these patients underwent MRI-guided biopsy, which is known to improve detection of clinically-significant cancer.⁸

Should prostate cancer grading be based on the volume of pattern 4 (and 5)?

Multiple studies have demonstrated that the volume of GP4 is associated with an increased risk of adverse pathology in RP specimens and BCR.^{13, 14} Additionally, including GP3 volume seems to provide limited information and influences prostate-specific antigen (PSA) and PSA density similarly to benign tissue.¹⁵ It is only logical that tumor components known to lack metastatic potential (i.e. GP3) would have minimal impact on cancer outcomes and thus should not be a primary consideration in management decisions. The current analysis adds to the growing body of evidence that the volume of more aggressive tumor (i.e. GP4) should drive decision-making in this context.

What comes next?

Perhaps it is time to shift the conversation from how to label various Gleason grades to instead asking whether the Gleason system is our best approach for guiding contemporary cancer management. With advancements in artificial intelligence (AI), we can envision a future in which machine learning accurately predicts individualized outcomes from a synthesized analysis of pathology, PSA levels, MRI findings, and genomic data. Instead of trying to explain the significance of nuanced pathologic findings (e.g. 2mm vs. 4mm of GP4) to our patients, we could instead focus on discussing the likelihood of developing metastases under various management options based on a comprehensive assessment of their data. While this may seem like a distant dream to some, AI has already been shown to achieve high agreement with expert uropathologists, offering an objective component of grading to complement our pathology colleagues.¹⁶

While we await this future, Scuderi et al. evoke several important, actionable points. First, further research is needed to confirm the insignificance of GP3 volume, particularly in patients with high-volume GP3 disease. Similarly, as additional data validate a grading approach based on GP4–5 volume, the manner of reporting GP4–5 volume should be standardized. Finally, guideline definitions of intermediate-risk disease – if not prognostic risk groups altogether – should be re-evaluated to reflect this more contemporary understanding of grade. A more accurate and interpretable system awaits to benefit our patients.