Automated Ki‐67 labeling index assessment in prostate cancer using artificial intelligence and multiplex fluorescence immunohistochemistry

Abstract

The Ki-67 labeling index (Ki-67 LI) is a strong prognostic marker in prostate cancer, although its analysis requires cumbersome manual quantification of Ki-67 immunostaining in 200-500 tumor cells. To enable automated Ki-67 LI assessment in routine clinical practice, a framework for automated Ki-67 LI quantification, which comprises three different artificial intelligence analysis steps and an algorithm for cell-distance analysis of multiplex fluorescence immunohistochemistry (mfIHC) staining, was developed and validated in a cohort of 12,475 prostate cancers. The prognostic impact of the Ki-67 LI was tested on a tissue microarray (TMA) containing one 0.6 mm sample per patient. A 'heterogeneity TMA' containing three to six samples from different tumor areas in each patient was used to model Ki-67 analysis of multiple different biopsies, and 30 prostate biopsies were analyzed to compare a 'classical' bright field-based Ki-67 analysis with the mfIHC-based framework. The Ki-67 LI provided strong and independent prognostic information in 11,845 analyzed prostate cancers (p < 0.001 each), and excellent agreement was found between the framework for automated Ki-67 LI assessment and the manual quantification in prostate biopsies from routine clinical practice (intraclass correlation coefficient: 0.94 [95% confidence interval: 0.87-0.97]). The analysis of the heterogeneity TMA revealed that the Ki-67 LI of the sample with the highest Gleason score (area under the curve [AUC]: 0.68) was as prognostic as the mean Ki-67 LI of all six foci (AUC: 0.71 [p = 0.24]). The combined analysis of the Ki-67 LI and Gleason score obtained on identical tissue spots showed that the Ki-67 LI added significant additional prognostic information in case of classical International Society of Urological Pathology grades (AUC: 0.82 [p = 0.002]) and quantitative Gleason score (AUC: 0.83 [p = 0.018]). The Ki-67 LI is a powerful prognostic parameter in prostate cancer that is now applicable in routine clinical practice. In the case of multiple cancer-positive biopsies, the sole automated analysis of the worst biopsy was sufficient.

BLEACH&STAIN 15-marker Multiplexed Imaging in 3,098 Human Carcinomas Reveals Six Major PD-L1–driven Immune Phenotypes with Distinct Spatial Orchestration

“…Image analysis was performed using the previously trained ( 22 ) deep-learning–based (U-Net) framework for cell detection, cell segmentation, intensity measurement of the used fluorophores (range 0–255, i.e., a continuous numerical value indicating the fluorescence signal strength), processing the intensity values, cell-to-cell distance, and cell-to-cell interaction analysis using Python Programming Language version 3.8 (RRID:SCR_008394; ref. 21 ), R version 3.6.1 (The R foundation, R Project for Statistical Computing; RRID:SCR_001905; ref.…”

“…Thresholding was used to annotate the first training set. This set was manually corrected upon necessity and then used to train “provisional” deep-learning systems to continuously extend the training set along with an improved accuracy of the deep-learning system ( 22 ). The final deep-learning systems (U-Net) for cell type identification has been trained on 800 to 2,500 tissue samples from more than 30 different carcinoma entities (Supplementary Figs.…”

“…S2 and S3) using the deep-learning frameworks Keras and Tensorflow (RRID:SCR_016345, in Python version 3.87) and the Visiopharm software package. The image analysis workflow of 17 different deep-learning systems and the spatial analysis algorithms have been described in detail earlier ( 22, 25 ).…”

Harnessing artificial intelligence for prostate cancer management

“…AI can effectively tackle various tasks beyond diagnosis and grading. For example, automatic measurement of cancer length and volume, 20 , 23 , 37 , 56 quantification of GP percentage, 43 , 57 , 58 recognition and quantification of perineural invasion, 21 , 59 quantification of immunohistochemistry (IHC) staining, 60 , 61 and detection and quantification of cribriform pattern. 34 , 62 , 63 AI models have also been developed to assess tumor purity of PCa using frozen H&E-stained slides.…”

Predictive value of systematic immune-inflammation index combined with Ki-67 index on prognosis of prostate cancer patients after laparoscopic radical prostatectomy

“…Ki-67 is a nuclear protein expressed in all phases (G1, S, G2 and M) of the cell cycle except the resting G0 phase, and its expression was linked with the proliferation of tumor cells [ 12 ]. Several studies have provided supporting evidence of prognostic role for Ki-67 index in PCa after prostatectomy [ 13 , 14 ]. Considering the potential prognostic significance of both SII and Ki-67 index individually, there is a growing interest in exploring the combined value of these markers in predicting outcomes of PCa patients following LRP.…”