Recommendations for Performance Evaluation of Machine Learning in Pathology: A Concept Paper From the College of American Pathologists.

CONTEXT.­: Machine learning applications in the pathology clinical domain are emerging rapidly. As decision support systems continue to mature, laboratories will increasingly need guidance to evaluate their performance in clinical practice. Currently there are no formal guidelines to assist pathology laboratories in verification and/or validation of such systems. These recommendations are being proposed for the evaluation of machine learning systems in the clinical practice of pathology. OBJECTIVE.­: To propose recommendations for performance evaluation of in vitro diagnostic tests on patient samples that incorporate machine learning as part of the preanalytical, analytical, or postanalytical phases of the laboratory workflow. Topics described include considerations for machine learning model evaluation including risk assessment, predeployment requirements, data sourcing and curation, verification and validation, change control management, human-computer interaction, practitioner training, and competency evaluation. DATA SOURCES.­: An expert panel performed a review of the literature, Clinical and Laboratory Standards Institute guidance, and laboratory and government regulatory frameworks. CONCLUSIONS.­: Review of the literature and existing documents enabled the development of proposed recommendations. This white paper pertains to performance evaluation of machine learning systems intended to be implemented for clinical patient testing. Further studies with real-world clinical data are encouraged to support these proposed recommendations. Performance evaluation of machine learning models is critical to verification and/or validation of in vitro diagnostic tests using machine learning intended for clinical practice.

Gene expression profiling of clear cell papillary renal cell carcinoma: comparison with clear cell renal cell carcinoma and papillary renal cell carcinoma.

Clear cell papillary renal cell carcinoma is a distinct variant of renal cell carcinoma that shares some overlapping histological and immunohistochemical features of clear cell renal cell carcinoma and papillary renal cell carcinoma. Although the clear cell papillary renal cell carcinoma immunohistochemical profile is well described, clear cell papillary renal cell carcinoma mRNA expression has not been well characterized. We investigated the clear cell papillary renal cell carcinoma gene expression profile using previously identified candidate genes. We selected 17 clear cell papillary renal cell carcinoma, 15 clear cell renal cell carcinoma, and 13 papillary renal cell carcinoma cases for molecular analysis following histological review. cDNA from formalin-fixed paraffin-embedded tissue was prepared. Quantitative real-time PCR targeting alpha-methylacyl coenzyme-A racemase (AMACR), BMP and activin membrane-bound inhibitor homolog (BAMBI), carbonic anhydrase IX (CA9), ceruloplasmin (CP), nicotinamide N-methyltransferase (NNMT), schwannomin-interacting protein 1 (SCHIP1), solute carrier family 34 (sodium phosphate) member 2 (SLC34A2), and vimentin (VIM) was performed. Gene expression data were normalized relative to 28S ribosomal RNA. Clear cell papillary renal cell carcinoma expressed all eight genes at variable levels. Compared with papillary renal cell carcinoma, clear cell papillary renal cell carcinoma expressed more CA9, CP, NNMT, and VIM, less AMACR, BAMBI, and SLC34A2, and similar levels of SCHIP1. Compared with clear cell renal cell carcinoma, clear cell papillary renal cell carcinoma expressed slightly less NNMT, but similar levels of the other seven genes. Although clear cell papillary renal cell carcinoma exhibits a unique molecular signature, it expresses several genes at comparable levels to clear cell renal cell carcinoma relative to papillary renal cell carcinoma. Understanding the molecular pathogenesis of clear cell papillary renal cell carcinoma will have a key role in future sub-classifications of this unique tumor.

Suppression of Tak1 promotes prostate tumorigenesis.

More than 30% of primary prostate cancers contain a consensus deletion of an approximately 800 kb locus on chromosome 6q15.1. The MAP3K7 gene, which encodes TGF-ß activated kinase-1 (Tak1), is a putative prostate tumor suppressor gene within this region whose precise function remains obscure. In this study, we investigated the role of Tak1 in human and murine prostate cancers. In 50 well-characterized human cancer specimens, we found that Tak1 expression was progressively lost with increasing Gleason grade, both within each cancer and across all cancers. In murine prostate stem cells and Tak1-deficient prostatic epithelial cells, Tak1 loss increased proliferation, migration, and invasion. When prostate stem cells attenuated for Tak1 were engrafted with fetal urogenital mesenchyme, the histopathology of the grafts reflected the natural history of prostate cancer leading from prostatic intraepithelial neoplasia to invasive carcinoma. In the grafts containing Tak1-suppressed prostate stem cells, p38 and c-jun-NH(2)-kinase activity was attenuated and proliferation was increased. Together, our findings functionally validate the proposed tumor suppressor role of Tak1 in prostate cancer.