Dataset Information

Histotype classification of ovarian carcinoma: A comparison of approaches.

ABSTRACT: OBJECTIVE:Major changes in the classification of ovarian carcinoma histotypes occurred over the last two decades, resulting in the current 2014 World Health Organization (WHO) diagnostic criteria that recognize five principal histotypes: high-grade serous, low-grade serous, endometrioid, clear cell, and mucinous carcinoma. We assessed the impact of these guidelines and use of immunohistochemical (IHC) markers on classification of ovarian carcinomas in existing population-based studies. METHODS:We evaluated histotype classification for 2361 ovarian carcinomas diagnosed between 1999 and 2009 from two case-control studies using three approaches: 1. pre-2014 WHO ("historic") histotype; 2. Standardized review of pathology slides using the 2014 WHO criteria alone; and 3. An integrated IHC assessment along with the 2014 WHO criteria. We used Kappa statistics to assess agreement between approaches, and Kaplan-Meier survival curves and Cox proportional hazards models to evaluate mortality. RESULTS:Compared to the standardized pathologic review histotype, agreement across approaches was high (kappa?=?0.892 for historic, and 0.849 for IHC integrated histotype), but the IHC integrated histotype identified more low-grade serous carcinomas and a subset of endometrioid carcinomas that were assigned as high-grade serous (n?=?25). No substantial differences in histotype-specific mortality were observed across approaches. CONCLUSIONS:Our findings suggest that histotype assignment is fairly consistent regardless of classification approach, but that progressive improvements in classification accuracy for some less common histotypes are achieved with pathologic review using the 2014 WHO criteria and with IHC integration. We additionally recommend a classification scheme to fit historic data into the 2014 WHO categories to answer histotype-specific research questions.

SUBMITTER: Peres LC

PROVIDER: S-EPMC6292681 | biostudies-literature | 2018 Oct

REPOSITORIES: biostudies-literature

ACCESS DATA

Publications

Histotype classification of ovarian carcinoma: A comparison of approaches.

Peres Lauren C LC Cushing-Haugen Kara L KL Anglesio Michael M Wicklund Kristine K Bentley Rex R Berchuck Andrew A Kelemen Linda E LE Nazeran Tayyebeh M TM Gilks C Blake CB Harris Holly R HR Huntsman David G DG Schildkraut Joellen M JM Rossing Mary Anne MA Köbel Martin M Doherty Jennifer A JA

Gynecologic oncology 20180816 1

<h4>Objective</h4>Major changes in the classification of ovarian carcinoma histotypes occurred over the last two decades, resulting in the current 2014 World Health Organization (WHO) diagnostic criteria that recognize five principal histotypes: high-grade serous, low-grade serous, endometrioid, clear cell, and mucinous carcinoma. We assessed the impact of these guidelines and use of immunohistochemical (IHC) markers on classification of ovarian carcinomas in existing population-based studies.<h ...[more]

PMID: 30121132

Similar Datasets

Project description:Background:The understanding of ovarian cancer pathogenesis has recently shifted to recognize distinct changes in how ovarian cancer histotypes are defined. Using the 2014 World Health Organization (WHO) diagnostic guidelines, we classified ovarian cancer histotypes in Surveillance, Epidemiology, and End Results (SEER) cancer registry data and examined survival patterns by histotype and disease stage. Methods:We extracted data on 28 118 incident epithelial ovarian cancer cases diagnosed in 2004-2014 from SEER and defined histotype using the 2014 WHO guidelines (high-grade serous, low-grade serous, endometrioid, clear cell, mucinous, carcinosarcoma, and malignant Brenner tumors). By histotype and disease stage, we estimated Kaplan-Meier survival curves and calculated age-adjusted overall and cause-specific survival estimates. Cox proportional hazards regression models were used to estimate histotype-specific hazard ratios (HRs) and 95% confidence intervals (CIs) by disease stage while adjusting for age at diagnosis, region, race/ethnicity, and receipt of surgery. Results:Within two years after diagnosis, localized/regional-stage carcinosarcoma and distant-stage mucinous, clear cell, and carcinosarcoma had a higher risk of mortality compared with high-grade serous, with the most pronounced association for localized/regional carcinosarcoma (>1-2-year time period: HR = 3.81, 95% CI = 2.74 to 5.30) and distant-stage mucinous (0-1-year time period: HR = 3.87, 95% CI = 3.45 to 4.34). In the time period more than four to 10 years after diagnosis, hazard ratios for all histotypes relative to high-grade serous, irrespective of disease stage, were less than 1.00. Cumulatively, both localized/regional and distant-stage low-grade serous and endometrioid carcinomas had the most favorable outcomes. Conclusions:Our large study, which is representative of the United States population and incorporates the most current knowledge of ovarian cancer pathogenesis, highlights the need to recognize ovarian cancer as a set of distinct diseases and not a single entity. Only then will we be able to effectively target the unique features of each histotype to reduce ovarian cancer mortality.

Project description:Alzheimer's disease (AD) is a progressive neurodegenerative disorder, accounting for nearly 60% of all dementia cases. The occurrence of the disease has been increasing rapidly in recent years. Presently about 46.8 million individuals suffer from AD worldwide. The current absence of effective treatment to reverse or stop AD progression highlights the importance of disease prevention and early diagnosis. Brain structural Magnetic Resonance Imaging (MRI) has been widely used for AD detection as it can display morphometric differences and cerebral structural changes. In this study, we built three machine learning-based MRI data classifiers to predict AD and infer the brain regions that contribute to disease development and progression. We then systematically compared the three distinct classifiers, which were constructed based on Support Vector Machine (SVM), 3D Very Deep Convolutional Network (VGGNet) and 3D Deep Residual Network (ResNet), respectively. To improve the performance of the deep learning classifiers, we applied a transfer learning strategy. The weights of a pre-trained model were transferred and adopted as the initial weights of our models. Transferring the learned features significantly reduced training time and increased network efficiency. The classification accuracy for AD subjects from elderly control subjects was 90%, 95%, and 95% for the SVM, VGGNet and ResNet classifiers, respectively. Gradient-weighted Class Activation Mapping (Grad-CAM) was employed to show discriminative regions that contributed most to the AD classification by utilizing the learned spatial information of the 3D-VGGNet and 3D-ResNet models. The resulted maps consistently highlighted several disease-associated brain regions, particularly the cerebellum which is a relatively neglected brain region in the present AD study. Overall, our comparisons suggested that the ResNet model provided the best classification performance as well as more accurate localization of disease-associated regions in the brain compared to the other two approaches.

Project description:There is evidence that some ovarian tumours evoke an immune response, which can be assessed by tumour infiltrating lymphocytes (TILs). To facilitate adoption of TILs as a clinical biomarker, a standardised method for their H&E visual evaluation has been validated in breast cancer.We sought to investigate the prognostic significance of TILs in a study of 953 invasive epithelial ovarian cancer tumour samples, both primary and metastatic, from 707 patients from the prospective population-based SEARCH study. TILs were analysed using a standardised method based on H&E staining producing a percentage score for stromal and intratumoral compartments. We used Cox regression to estimate hazard ratios of the association between TILs and survival.The extent of stromal and intra-tumoral TILs were correlated in the primary tumours (n = 679, Spearman's rank correlation = 0.60, P < 0.001) with a similar correlation in secondary tumours (n = 224, Spearman's rank correlation = 0.62, P < 0.001). There was a weak correlation between stromal TIL levels in primary and secondary tumour samples (Spearman's rank correlation = 0.29, P < 0.001) and intra-tumoral TIL levels in primary and secondary tumour samples (Spearman's rank correlation = 0.19, P = 0.0094). The extent of stromal TILs differed between histotypes (Pearson chi2 (12d.f.) 54.1, P < 0.0001) with higher levels of stromal infiltration in the high-grade serous and endometriod cases. A significant association was observed for higher intratumoral TIL levels and a favourable prognosis (HR 0.74 95% CI 0.55-1.00 p = 0.047).This study is the largest collection of epithelial ovarian tumour samples evaluated for TILs. We have shown that stromal and intratumoral TIL levels are correlated and that their levels correlate with clinical variables such as tumour histological subtype. We have also shown that increased levels of both intratumoral and stromal TILs are associated with a better prognosis; however, this is only statistically significant for intratumoral TILs. This study suggests that a clinically useful immune prognostic indicator in epithelial ovarian cancer could be developed using this technique.

Dataset Information

Histotype classification of ovarian carcinoma: A comparison of approaches.

Publications

Histotype classification of ovarian carcinoma: A comparison of approaches.

Similar Datasets

OmicsDI is part of the ELIXIR infrastructure