Explore the vast range of titles available.

Improved markers of prognosis are needed to stratify patients with early-stage colorectal cancer to refine selection of adjuvant therapy. The aim of the present study was to develop a biomarker of patient outcome after primary colorectal cancer resection by directly analysing scanned conventional haematoxylin and eosin stained sections using deep learning. More than 12 000 000 image tiles from patients with a distinctly good or poor disease outcome from four cohorts were used to train a total of ten convolutional neural networks, purpose-built for classifying supersized heterogeneous images. A prognostic biomarker integrating the ten networks was determined using patients with a non-distinct outcome. The marker was tested on 920 patients with slides prepared in the UK, and then independently validated according to a predefined protocol in 1122 patients treated with single-agent capecitabine using slides prepared in Norway. All cohorts included only patients with resectable tumours, and a formalin-fixed, paraffin-embedded tumour tissue block available for analysis. The primary outcome was cancer-specific survival. 828 patients from four cohorts had a distinct outcome and were used as a training cohort to obtain clear ground truth. 1645 patients had a non-distinct outcome and were used for tuning. The biomarker provided a hazard ratio for poor versus good prognosis of 3·84 (95% CI 2·72–5·43; p<0·0001) in the primary analysis of the validation cohort, and 3·04 (2·07–4·47; p<0·0001) after adjusting for established prognostic markers significant in univariable analyses of the same cohort, which were pN stage, pT stage, lymphatic invasion, and venous vascular invasion. A clinically useful prognostic marker was developed using deep learning allied to digital scanning of conventional haematoxylin and eosin stained tumour tissue sections. The assay has been extensively evaluated in large, independent patient populations, correlates with and outperforms established molecular and morphological prognostic markers, and gives consistent results across tumour and nodal stage. The biomarker stratified stage II and III patients into sufficiently distinct prognostic groups that potentially could be used to guide selection of adjuvant treatment by avoiding therapy in very low risk groups and identifying patients who would benefit from more intensive treatment regimes. The Research Council of Norway.

The DoMore-v1-CRC marker was recently developed using deep learning and conventional haematoxylin and eosin-stained tissue sections, and was observed to outperform established molecular and morphological markers of patient outcome after primary colorectal cancer resection. The aim of the present study was to develop a clinical decision support system based on DoMore-v1-CRC and pathological staging markers to facilitate individualised selection of adjuvant treatment. We estimated cancer-specific survival in subgroups formed by pathological tumour stage (pT<4 or pT4), pathological nodal stage (pN0, pN1, or pN2), number of lymph nodes sampled (≤12 or >12) if not pN2, and DoMore-v1-CRC classification (good, uncertain, or poor prognosis) in 997 patients with stage II or III colorectal cancer considered to have no residual tumour (R0) from two community-based cohorts in Norway and the UK, and used these data to define three risk groups. An external cohort of 1075 patients with stage II or III R0 colorectal cancer from the QUASAR 2 trial was used for validation; these patients were treated with single-agent capecitabine. The proposed risk stratification system was evaluated using Cox regression analysis. We similarly evaluated a risk stratification system intended to reflect current guidelines and clinical practice. The primary outcome was cancer-specific survival. The new risk stratification system provided a hazard ratio of 10·71 (95% CI 6·39–17·93; p<0·0001) for high-risk versus low-risk patients and 3·06 (1·73–5·42; p=0·0001) for intermediate versus low risk in the primary analysis of the validation cohort. Estimated 3-year cancer-specific survival was 97·2% (95% CI 95·1–98·4; n=445 [41%]) for the low-risk group, 94·8% (91·7–96·7; n=339 [32%]) for the intermediate-risk group, and 77·6% (72·1–82·1; n=291 [27%]) for the high-risk group. The guideline-based risk grouping was observed to be less prognostic and informative (the low-risk group comprised only 142 [13%] of the 1075 patients). Integrating DoMore-v1-CRC and pathological staging markers provided a clinical decision support system that risk stratifies more accurately than its constituent elements, and identifies substantially more patients with stage II and III colorectal cancer with similarly good prognosis as the low-risk group in current guidelines. Avoiding adjuvant chemotherapy in these patients might be safe, and could reduce morbidity, mortality, and treatment costs. The Research Council of Norway.

Chromatin organisation affects gene expression and regional mutation frequencies and contributes to carcinogenesis. Aberrant organisation of DNA has been correlated with cancer prognosis in analyses of the chromatin component of tumour cell nuclei using image texture analysis. As yet, the methodology has not been sufficiently validated to permit its clinical application. We aimed to define and validate a novel prognostic biomarker for the automatic detection of heterogeneous chromatin organisation. Machine learning algorithms analysed the chromatin organisation in 461 000 images of tumour cell nuclei stained for DNA from 390 patients (discovery cohort) treated for stage I or II colorectal cancer at the Aker University Hospital (Oslo, Norway). The resulting marker of chromatin heterogeneity, termed Nucleotyping, was subsequently independently validated in six patient cohorts: 442 patients with stage I or II colorectal cancer in the Gloucester Colorectal Cancer Study (UK); 391 patients with stage II colorectal cancer in the QUASAR 2 trial; 246 patients with stage I ovarian carcinoma; 354 patients with uterine sarcoma; 307 patients with prostate carcinoma; and 791 patients with endometrial carcinoma. The primary outcome was cancer-specific survival. In all patient cohorts, patients with chromatin heterogeneous tumours had worse cancer-specific survival than patients with chromatin homogeneous tumours (univariable analysis hazard ratio [HR] 1·7, 95% CI 1·2–2·5, in the discovery cohort; 1·8, 1·0–3·0, in the Gloucester validation cohort; 2·2, 1·1–4·5, in the QUASAR 2 validation cohort; 3·1, 1·9–5·0, in the ovarian carcinoma cohort; 2·5, 1·8–3·4, in the uterine sarcoma cohort; 2·3, 1·2–4·6, in the prostate carcinoma cohort; and 4·3, 2·8–6·8, in the endometrial carcinoma cohort). After adjusting for established prognostic patient characteristics in multivariable analyses, Nucleotyping was prognostic in all cohorts except for the prostate carcinoma cohort (HR 1·7, 95% CI 1·1–2·5, in the discovery cohort; 1·9, 1·1–3·2, in the Gloucester validation cohort; 2·6, 1·2–5·6, in the QUASAR 2 cohort; 1·8, 1·1–3·0, for ovarian carcinoma; 1·6, 1·0–2·4, for uterine sarcoma; 1·43, 0·68–2·99, for prostate carcinoma; and 1·9, 1·1–3·1, for endometrial carcinoma). Chromatin heterogeneity was a significant predictor of cancer-specific survival in microsatellite unstable (HR 2·9, 95% CI 1·0–8·4) and microsatellite stable (1·8, 1·2–2·7) stage II colorectal cancer, but microsatellite instability was not a significant predictor of outcome in chromatin homogeneous (1·3, 0·7–2·4) or chromatin heterogeneous (0·8, 0·3–2·0) stage II colorectal cancer. The consistent prognostic prediction of Nucleotyping in different biological and technical circumstances suggests that the marker of chromatin heterogeneity can be reliably assessed in routine clinical practice and could be used to objectively assist decision making in a range of clinical settings. An immediate application would be to identify high-risk patients with stage II colorectal cancer who might have greater absolute benefit from adjuvant chemotherapy. Clinical trials are warranted to evaluate the survival benefit and cost-effectiveness of using Nucleotyping to guide treatment decisions in multiple clinical settings. The Research Council of Norway, the South-Eastern Norway Regional Health Authority, the National Institute for Health Research, and the Wellcome Trust.

The mitotic checkpoint protein BUB3, cyclin B1 (CCNB1) and pituitary tumor-transforming 1 (PTTG1) regulates cell division, and are sparsely studied in prostate cancer. Deregulation of these genes can lead to genomic instability, a characteristic of more aggressive tumors. We aimed to determine the expression levels of BUB3, CCNB1, and PTTG1 as potential prognostic markers of recurrence after radical prostatectomy. Protein levels were determined by immunohistochemistry on three formalin-fixed paraffin-embedded tissue sections from each of the 253 patients treated with radical prostatectomy. Immunohistochemistry scores were obtained by automated image analysis for CCNB1 and PTTG1. Recurrence, defined as locoregional recurrence, distant metastasis or death from prostate cancer, was used as endpoint for survival analysis. Tumors having both positive and negative tumor areas for cytoplasmic BUB3 (30%), CCNB1 (28%), or PTTG1 (35%) were considered heterogeneous. Patients with ≥1 positive tumor area had significantly increased risk of disease recurrence in univariable analysis compared with patients where all tumor areas were negative for cytoplasmic BUB3 (hazard ratio [HR] = 2.18, 95% confidence interval [CI] 1.41–3.36), CCNB1 (HR = 2.98, 95% CI 1.93–4.61) and PTTG1 (HR = 1.91, 95% CI 1.23–2.97). Combining the scores of cytoplasmic BUB3 and CCNB1 improved risk stratification when integrated with the Cancer of the Prostate Risk Assessment post-Surgical (CAPRA-S) score (difference in concordance index = 0.024, 95% CI 0.001–0.05). In analysis of multiple tumor areas, prognostic value was observed for cytoplasmic BUB3, CCNB1, and PTTG1.

Deep learning is expected to aid pathologists in tasks such as tumour segmentation. We developed a general tumour segmentation model for histopathological images and examined its performance in different cancer types. The model was developed using over 20,000 whole-slide images from over 4000 patients with colorectal, endometrial, lung, or prostate carcinoma. Performance was validated in pre-planned analyses on external cohorts with over 3000 patients across six cancer types. Exploratory analyses included over 1500 additional patients from The Cancer Genome Atlas. Average Dice coefficient was over 80% in all validation cohorts with en bloc resection specimens and in The Cancer Genome Atlas cohorts. No performance loss was observed when comparing the general model with single-cancer models specialised in cancer types from the development set. In conclusion, extensive and rigorous evaluations demonstrate that generic tumour segmentation by a single model is possible across cancer types, patient populations, sample preparations and slide scanners.

The number of publications on deep learning for cancer diagnostics is rapidly increasing, and systems are frequently claimed to perform comparable with or better than clinicians. However, few systems have yet demonstrated real-world medical utility. In this Perspective, we discuss reasons for the moderate progress and describe remedies designed to facilitate transition to the clinic. Recent, presumably influential, deep learning studies in cancer diagnostics, of which the vast majority used images as input to the system, are evaluated to reveal the status of the field. By manipulating real data, we then exemplify that much and varied training data facilitate the generalizability of neural networks and thus the ability to use them clinically. To reduce the risk of biased performance estimation of deep learning systems, we advocate evaluation in external cohorts and strongly advise that the planned analyses, including a predefined primary analysis, are described in a protocol preferentially stored in an online repository. Recommended protocol items should be established for the field, and we present our suggestions.The number of publications on deep learning for cancer diagnostics is rapidly increasing, but clinical translation is slow. This Perspective advocates performance estimation in external cohorts and strongly advises that a primary analysis is predefined in a standardized protocol preferentially stored in an online repository.

The current standard-of-care adjuvant treatment for patients with colorectal cancer (CRC) comprises a fluoropyrimidine (5-fluorouracil or capecitabine) as a single agent or in combination with oxaliplatin, for either 3 or 6 months. Selection of therapy depends on conventional histopathological staging procedures, which constitute a blunt tool for patient stratification. Given the relatively marginal survival benefits that patients can derive from adjuvant treatment, improving the safety of chemotherapy regimens and identifying patients most likely to benefit from them is an area of unmet need. Patient stratification should enable distinguishing those at low risk of recurrence and a high chance of cure by surgery from those at higher risk of recurrence who would derive greater absolute benefits from chemotherapy. To this end, genetic analyses have led to the discovery of germline determinants of toxicity from fluoropyrimidines, the identification of patients at high risk of life-threatening toxicity, and enabling dose modulation to improve safety. Thus far, results from analyses of resected tissue to identify mutational or transcriptomic signatures with value as prognostic biomarkers have been rather disappointing. In the past few years, the application of artificial intelligence-driven models to digital images of resected tissue has identified potentially useful algorithms that stratify patients into distinct prognostic groups. Similarly, liquid biopsy approaches involving measurements of circulating tumour DNA after surgery are additionally useful tools to identify patients at high and low risk of tumour recurrence. In this Perspective, we provide an overview of the current landscape of adjuvant therapy for patients with CRC and discuss how new technologies will enable better personalization of therapy in this setting.The current standard-of-care adjuvant treatment for patients with colorectal cancer is chemotherapy selected on the basis of conventional histopathological staging criteria; however, the clinical benefit from these regimens is limited. The authors of this Perspective discuss strategies to minimize toxicity and monitor efficacy of these regimens, and propose new tools for disease staging that could enable more personalized treatment decisions.

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.

Background Current risk stratification tools for prostate cancer patients under active surveillance (AS) may inadequately identify those needing treatment. We investigated DNA ploidy and PTEN as potential biomarkers to predict aggressive disease in AS patients. Methods We assessed DNA ploidy by image cytometry and PTEN protein expression by immunohistochemistry in 3197 tumour-containing tissue blocks from 558 patients followed in AS at a Norwegian local hospital. The primary endpoint was treatment, with treatment failure (biochemical recurrence or initiation of salvage therapy) as the secondary endpoint. Results The combined DNA ploidy and PTEN (DPP) status at diagnosis was associated with treatment-free survival in univariable- and multivariable analysis, with a HR for DPP-aberrant vs. DPP-normal tumours of 2.12 ( p  < 0.0001) and 1.94 ( p  < 0.0001), respectively. Integration of DNA ploidy and PTEN status with the Cancer of the Prostate Risk Assessment (CAPRA) score improved risk stratification (c-index difference = 0.025; p  = 0.0033). Among the treated patients, those with DPP-aberrant tumours exhibited a significantly higher likelihood of treatment failure (HR 2.01; p  = 0.027). Conclusions DNA ploidy and PTEN could serve as additional biomarkers to identify AS patients at increased risk of developing aggressive disease, enabling earlier intervention for nearly 50% of the patients that will eventually receive treatment with current protocol.