Implications of all the available evidence It is possible to utilise deep learning to develop biomarkers for automatic prediction of patient outcome directly from conventional histopathology images. In colorectal cancer, the marker was found to be a clinically useful prognostic marker in analysis of a large series of patients who received consistent, modern cancer treatment.

Background:The high degree of genomic diversity in cancer represents a challenge for identifying objective prognostic markers. We aimed to examine the extent of tumour heterogeneity and its effect on the evaluation of a selected prognostic marker using prostate cancer as a model.Methods:We assessed Gleason Score (GS), DNA ploidy status and phosphatase and tensin homologue (PTEN) expression in radical prostatectomy specimens (RP) from 304 patients followed for a median of 10 years (interquartile range 6–12). GS was assessed for every tumour-containing block and DNA ploidy for a median of four samples for each RP. In a subgroup of 40 patients we assessed DNA ploidy and PTEN status in every tumour-containing block. In 102 patients assigned to active surveillance (AS), GS and DNA ploidy were studied in needle biopsies.Results:Extensive heterogeneity was observed for GS (89% of the patients) and DNA ploidy (40% of the patients) in the cohort, and DNA ploidy (60% of the patients) and PTEN expression (75% of the patients) in the subgroup. DNA ploidy was a significant prognostic marker when heterogeneity was taken into consideration. In the AS cohort we found heterogeneity in GS (24%) and in DNA ploidy (25%) specimens.Conclusions:Multi-sample analysis should be performed to support clinical treatment decisions.

Molecular classification of colorectal cancer (CRC) is currently based on microsatellite instability (MSI), KRAS or BRAF mutation and, occasionally, chromosomal instability (CIN). Whilst useful, these categories may not fully represent the underlying molecular subgroups. We screened 906 stage II/III CRCs from the VICTOR clinical trial for somatic mutations. Multivariate analyses (logistic regression, clustering, Bayesian networks) identified the primary molecular associations. Positive associations occurred between: CIN and TP53 mutation; MSI and BRAF mutation; and KRAS and PIK3CA mutations. Negative associations occurred between: MSI and CIN; MSI and NRAS mutation; and KRAS mutation, and each of NRAS, TP53 and BRAF mutations. Some complex relationships were elucidated: KRAS and TP53 mutations had both a direct negative association and a weaker, confounding, positive association via TP53–CIN–MSI–BRAF–KRAS. Our results suggested a new molecular classification of CRCs: (1) MSI+ and/or BRAF-mutant; (2) CIN+ and/or TP53– mutant, with wild-type KRAS and PIK3CA; (3) KRAS- and/or PIK3CA-mutant, CIN+, TP53-wild-type; (4) KRAS– and/or PIK3CA-mutant, CIN–, TP53-wild-type; (5) NRAS-mutant; (6) no mutations; (7) others. As expected, group 1 cancers were mostly proximal and poorly differentiated, usually occurring in women. Unexpectedly, two different types of CIN+ CRC were found: group 2 cancers were usually distal and occurred in men, whereas group 3 showed neither of these associations but were of higher stage. CIN+ cancers have conventionally been associated with all three of these variables, because they have been tested en masse. Our classification also showed potentially improved prognostic capabilities, with group 3, and possibly group 1, independently predicting disease-free survival. Copyright © 2012 Pathological Society of Great Britain and Ireland. Published by John Wiley & Sons, Ltd.