Gleason Grading

Gleason grades range from 1 to 5, representing a spectrum from well-differentiated to poorly differentiated tumor architecture:

• Grade 1: Well-formed, uniform glands tightly packed
• Grade 2: Slightly less uniform glands with more stroma between them
• Grade 3: Infiltrative glands with marked variation in size and shape
• Grade 4: Fused, cribriform, or poorly-formed glands
• Grade 5: Solid sheets, cords, or single cells with no glandular differentiation

Grades 1 and 2 are rarely diagnosed in contemporary practice due to improved diagnostic criteria.

The final Gleason Score is the sum of the two most prevalent architectural patterns observed in the biopsy specimen:

• Primary Grade: The pattern occupying the largest tumor area
• Secondary Grade: The second most prevalent pattern
• Tertiary Grade: A minor high-grade component (Grade 4 or 5) noted separately when present

For example, a tumor with predominantly Grade 3 and a smaller Grade 4 component receives a Gleason Score of 3+4=7. The order matters: 3+4 carries a better prognosis than 4+3, despite both summing to 7.

In 2014, the ISUP Grade Group system was introduced to simplify risk stratification and reduce overtreatment of Gleason 6 disease:

• Grade Group 1: Gleason Score ≤6 (low risk)
• Grade Group 2: Gleason Score 3+4=7 (intermediate favorable)
• Grade Group 3: Gleason Score 4+3=7 (intermediate unfavorable)
• Grade Group 4: Gleason Score 8 (high risk)
• Grade Group 5: Gleason Score 9-10 (very high risk)

This system directly addresses the psychological impact of a "6 out of 10" score by clarifying that Gleason 6 is the lowest clinically significant grade.

Deep learning models now automate Gleason grading from whole-slide images (WSIs) using multiple instance learning:

• Patch-level classification: A convolutional neural network or vision transformer classifies individual tissue regions into Gleason patterns
• Slide-level aggregation: Predictions are combined to determine the primary and secondary patterns across the entire specimen
• Inter-observer agreement: AI systems achieve Cohen's Kappa >0.85 with expert genitourinary pathologists, exceeding the inter-pathologist agreement of ~0.70

Automated systems reduce turnaround time and provide quantitative, reproducible grading for clinical trials and active surveillance protocols.

The cribriform architecture—characterized by solid nests of tumor cells punctuated by round, punched-out glandular lumina—is a Gleason Grade 4 pattern with independent prognostic significance:

• Associated with higher rates of biochemical recurrence after radical prostatectomy
• Linked to PTEN loss and aggressive genomic features
• Presence of cribriform morphology in Gleason 7 disease portends outcomes closer to Gleason 8

Pathologists now specifically report cribriform pattern presence, as it influences decisions between active surveillance and definitive treatment.

Intraductal carcinoma of the prostate (IDC-P) is a distinct entity where malignant cells expand pre-existing ducts, often adjacent to high-grade invasive cancer:

• Not assigned a Gleason grade but reported separately
• Strongly associated with Gleason Score 8-10 disease and germline BRCA2 mutations
• Independently predicts early biochemical failure and metastasis

Misclassifying IDC-P as Gleason 4 cribriform pattern can lead to understaging. Computational models are being trained to distinguish these morphologically similar but biologically distinct entities.

The five fundamental architectural grades that form the basis of the Gleason system, classified by the degree of glandular differentiation:

• Pattern 1: Circumscribed, uniform, closely packed glands
• Pattern 2: More loosely aggregated, less uniform glands with stromal separation
• Pattern 3: Infiltrative, angular glands of variable size with definitive stromal invasion
• Pattern 4: Fused, cribriform, or poorly-formed glands with ragged infiltration
• Pattern 5: Solid sheets, cords, or single cells with central comedonecrosis

Patterns 1 and 2 are rarely diagnosed in contemporary practice, with most biopsies showing patterns 3-5.

The Gleason Score is the sum of the two most prevalent architectural patterns observed in the specimen:

• Primary Grade: The pattern occupying the largest tumor volume
• Secondary Grade: The second most prevalent pattern (must constitute at least 5% of the tumor)
• Tertiary Grade: A higher-grade pattern occupying <5% is noted separately but not included in the score

Scores range from 6 (3+3) to 10 (5+5). A score of 3+4=7 is less aggressive than 4+3=7 because the primary pattern drives prognosis.

The International Society of Urological Pathology (ISUP) introduced a simplified 5-tier grading system in 2014 to address confusion around Gleason Score 7:

• Grade Group 1: Gleason Score ≤6 (3+3)
• Grade Group 2: Gleason Score 7 (3+4)
• Grade Group 3: Gleason Score 7 (4+3)
• Grade Group 4: Gleason Score 8 (4+4, 3+5, 5+3)
• Grade Group 5: Gleason Score 9-10 (4+5, 5+4, 5+5)

This system is now standard in pathology reports alongside the traditional Gleason Score.

Gleason grading is subject to significant inter-observer variability, particularly in distinguishing:

• Pattern 3 vs. tangential sectioning of Pattern 4 glands
• Cribriform Pattern 4 vs. intraductal carcinoma
• Borderline cases between Grade Groups 2 and 3

Reported Cohen's Kappa values range from 0.43-0.68 for general pathologists, improving to 0.70-0.85 for genitourinary specialists. This variability drives the need for AI-assisted grading systems that provide consistent, reproducible scoring.

Gleason grading directly informs risk stratification and treatment selection:

• Low Risk: Grade Group 1, PSA <10, stage ≤T2a → Active surveillance
• Intermediate Risk: Grade Group 2-3, PSA 10-20, stage T2b-T2c → Surgery or radiation
• High Risk: Grade Group 4-5, PSA >20, stage ≥T3a → Multimodal therapy

The Gleason Score remains the single strongest pathological predictor of biochemical recurrence, metastasis-free survival, and cancer-specific mortality in localized prostate cancer.

Deep learning models for automated Gleason grading typically employ Multiple Instance Learning (MIL) on gigapixel whole-slide images:

• Patch-level classification: CNNs or Vision Transformers classify individual tissue regions into Gleason patterns
• Slide-level aggregation: Attention-based pooling combines patch predictions into a final Grade Group
• Performance: State-of-the-art models achieve quadratic weighted Kappa >0.85 against genitourinary pathologist consensus

FDA-cleared systems now provide real-time decision support, reducing inter-observer variability and flagging suspicious regions for pathologist review.