Quantitative Histomorphometric Features of Prostate Cancer Predict Patients Who Biochemically Recur Following Prostatectomy

Abstract

Prostate cancer is the most commonly diagnosed cancer in men, accounting for 27% of the new male cancer diagnoses in 2022. If organ-confined, removal of the prostate through radical prostatectomy is considered curative; however, distant metastases may occur, resulting in a poor patient prognosis. This study sought to determine whether quantitative pathomic features of prostate cancer differ in patients who biochemically experience biological recurrence after surgery. Whole-mount prostate histology from 78 patients was analyzed for this study. In total, 614 slides were hematoxylin and eosin stained and digitized to produce whole slide images (WSI). Regions of differing Gleason patterns were digitally annotated by a genitourinary fellowship-trained pathologist, and high-resolution tiles were extracted from each annotated region of interest for further analysis. Individual glands within the prostate were identified using automated image processing algorithms, and histomorphometric features were calculated on a per-tile basis and across WSI and averaged by patients. Tiles were organized into cancer and benign tissues. Logistic regression models were fit to assess the predictive value of the calculated pathomic features across tile groups and WSI; additionally, models using clinical information were used for comparisons. Logistic regression classified each pathomic feature model at accuracies >80% with areas under the curve of 0.82, 0.76, 0.75, and 0.72 for all tiles, cancer only, noncancer only, and across WSI. This was comparable with standard clinical information, Gleason Grade Groups, and CAPRA score, which achieved similar accuracies but areas under the curve of 0.80, 0.77, and 0.70, respectively. This study demonstrates that the use of quantitative pathomic features calculated from digital histology of prostate cancer may provide clinicians with additional information beyond the traditional qualitative pathologist assessment. Further research is warranted to determine possible inclusion in treatment guidance.

Keywords: annotations; digital pathology; image processing; pathomic features; prostate cancer; whole slide images.

Figure 1.

Schematic representation of the annotation and tile extraction process, with second-order feature segmentations across a WSI. Top left: T2-weighted MR image used to model the prostate slicing jig. Custom prostate slicing jigs allow the prostate to be sliced to match the slice thicknesses of the MR image. Top right: whole-mount samples were stained, digitized, and annotated by a pathologist. Annotations were color-coded by class to extract representative tiles from each of the annotation classes: atrophy, HGPIN, G3, G4NC, G4CG, G5, and Seminal Vesicles (not pictured). Bottom: pathomic features are calculated across WSI and feature maps are overlaid on the original image.

Figure 2.

Pathomic feature segmentations. A representative Gleason 3 tile from the Huron microscope with pathomic feature maps. Calculated features include lumen roundness and area; cell fraction; epithelial roundness, area, and wall thickness. Calculated values are overlaid on the respective glands. Units of area maps are in mm², and thickness in mm. Roundness and cell fraction are unitless.

Figure 3.

Significant feature predictors of BCR with examples from cancer tiles. Top left: the patient who did not experience biochemical recurrence had regions of papillary to cribriform glands that had been previously associated with BCR. Bottom left: the patient who did experience BCR had regions of low-risk G3 cancer. Middle: feature maps overlaid on tiles.

Figure 4.

Top: ROC curves for each logistic regression model with AUCs and 95% CI. Pathomic feature models were generated for noncancer, cancer, all tiles, and WSI. Clinical feature models included a general model encompassing clinicopathological information, CAPRA score, and Grade Groups. Combined feature models included tile or WSI pathomic features with clinicopathological information. Bottom: Kaplan–Meier survival analyses were conducted to compare survival of pathomic features, cribriform presence, CAPRA score, and Gleason Grade Groups (P values calculated using log-rank test). In our patient cohort, higher grade cancers were not significantly more likely to recur. R1, low-risk; R2, high-risk; Cr1, cribriform glands absent; Cr2, cribriform glands present; C1, CAPRA 0–2; C2, CAPRA 3–5; C3, CAPRA 6–10; G1–5, Gleason Grade Group 1–5.