Histological grading of breast carcinomas: A study of interobserver agreement

doi:10.1016/0046-8177(95)90010-1

Human Pathology

Volume 26, Issue 8, August 1995, Pages 873-879

https://doi.org/10.1016/0046-8177(95)90010-1 Get rights and content

Abstract

Interobserver variation in the histological grading of breast carcinoma was investigated using the hypothesis that optimal fixation, more precise grading guidelines, some experience, the use of training and test sets, and a comparison of results with an expert group might allow higher levels of agreement. For the training sets sections from 50 consecutive cases of breast carcinoma received at the Sir Charles Gairdner Hospital (SCGH) and fixed in both B5 and buffered formal saline (BFS) were graded by consensus of three pathologists at the SCGH and independently by consensus of two pathologists at the Nottingham City Hospital (NCH) using a modified Scarff-BloomRichardson histological grading system with guidelines as suggested by NCH pathologists. The section quality and degree of preservation of nuclear morphology were judged by NCH pathologists to be superior for B5-fixed material. Complete agreement in grade between SCGH and NCH results was achieved for 83.3% of B5-fixed cases and 73.5% of BFS-fixed cases (P = .05) with relative disagreement rates (RDRs) of 0.15 and 0.29 and kappa statistic values of 0.73 and 0.58, respectively. Approximately 80% complete agreement was achieved for tubule formation, nuclear score, and mitotic count, with RDRs ranging from 0.19 to 0.27 and kappa values from 0.46 to 0.69. There was a consistent bias in the SCGH results toward a higher tubule score in both B5- and BFS-fixed material because of a difference in interpretation of cribriform or complex gland patterns and a consistent bias in SCGH results toward a lower nuclear size/pleomorphism score for B5 and BFS material. For the test set sections from 50 further consecutive cases of breast cancer fixed in B5 were examined using similar criteria but taking into account the sources of error shown by the training set. Approximately 80% complete agreement was again achieved for grade components and grade (RDRs, 0.18 and 0.72). Systematic bias was reduced in the test set, but no other improvement was observed. Of the tumors designated as grade I by NCH, 87.5% were called grade I tumors by SCGH in the B5 training set, 84.6% in the B5 test set, and 66.6% in the BFS training set. The levels of agreement shown in both the training and test sets were satisfactory and represented a significant improvement over our previous study, suggesting that experience and precise grading guidelines are of value. The similar levels of agreement in training and test sets suggest that reasonable results can be achieved without direct training by expert groups. The value of optimal fixation may be the most important variable noted by the study.

References (30)

JM Brown et al.
Confirmation of a long-term prognostic index in breast cancer
The Breast
(1993)
JM Harvey et al.
Histological grading in breast cancer: Interobserver agreement and relation to other prognostic factors including ploidy
Pathology
(1992)
F Theissig et al.
Histological grading of breast cancer
JPA Baak et al.
The Multicentre Morphometric Mammary Carcinoma Project (MMMCP)
JC Fleege et al.
Analysis of measuring system parameters that influence reproducibility of morphometric assessments with a graphic tablet
Hum Pathol
(1988)
JPA Baak
Mitosis counting in tumors
Hum Pathol
(1990)
EE Chouinard et al.
Reproducibility of pathological grading in breast cancer
The Breast
(1994)
SG Hilsenbeck et al.
Improved methods of estimating mitotic activity in solid tumors
Hum Pathol
(1992)
JF Simpson et al.
Expression of mitoses per thousand cells and cell density in breast carcinoma: A proposal
Hum Pathol
(1992)
PJ Van Diest et al.
Reproducibility of mitosis counting in 2469 breast cancer specimens: Results from the Multicenter Morphometric Mammary Carcinoma Project
Hum Pathol
(1992)

G Contesso et al.

The importance of histologic grade in long-term prognosis of breast cancer: A study of 1010 patients, uniformly treated at the Institut Gustave-Roussy

J Clin Oncol

(1987)

BW Davis et al.

Prognostic significance of tumor grade in clinical trials of adjuvant therapy for breast cancer with axillary lymph node metastasis

Cancer

(1986)

CW Elston et al.

Pathological prognostic factors in breast cancer. I. The value of histological grade in breast cancer: Experience from a large study with long-term follow-up

Histopathology

(1991)

DE Henson et al.

Relationship among outcome, stage of disease and histologic grade for 22, 616 cases of breast cancer

MH Galea et al.

The Nottingham prognostic index in primary breast cancer

Breast Cancer Res Treat

(1992)

Cited by (372)

Deep learning supported mitoses counting on whole slide images: A pilot study for validating breast cancer grading in the clinical workflow
2023, Journal of Pathology Informatics
Breast cancer (BC) prognosis is largely influenced by histopathological grade, assessed according to the Nottingham modification of Bloom-Richardson (BR). Mitotic count (MC) is a component of histopathological grading but is prone to subjectivity. This study investigated whether mitoses counting in BC using digital whole slide images (WSI) compares better to light microscopy (LM) when assisted by artificial intelligence (AI), and to which extent differences in digital MC (AI assisted or not) result in BR grade variations.
Fifty BC patients with paired core biopsies and resections were randomly selected. Component scores for BR grade were extracted from pathology reports. MC was assessed using LM, WSI, and AI. Different modalities (LM-MC, WSI-MC, and AI-MC) were analyzed for correlation with scatterplots and linear regression, and for agreement in final BR with Cohen’s κ.
MC modalities strongly correlated in both biopsies and resections: LM-MC and WSI-MC (R² 0.85 and 0.83, respectively), LM-MC and AI-MC (R² 0.85 and 0.95), and WSI-MC and AI-MC (R² 0.77 and 0.83). Agreement in BR between modalities was high in both biopsies and resections: LM-MC and WSI-MC (κ 0.93 and 0.83, respectively), LM-MC and AI-MC (κ 0.89 and 0.83), and WSI-MC and AI-MC (κ 0.96 and 0.73).
This first validation study shows that WSI-MC may compare better to LM-MC when using AI. Agreement between BR grade based on the different mitoses counting modalities was high. These results suggest that mitoses counting on WSI can well be done, and validate the presented AI algorithm for pathologist supervised use in daily practice. Further research is required to advance our knowledge of AI-MC, but it appears at least non-inferior to LM-MC.
MultiNet: A deep neural network approach for detecting breast cancer through multi-scale feature fusion
2022, Journal of King Saud University - Computer and Information Sciences
Breast cancer diagnosis from biopsy tissue images conducted manually by pathologists is costly, time-consuming, and disagreements among specialists. Nowadays, the advancement of the Computer-Aided Diagnosis (CAD) system allows pathologists to identify breast cancer more reliably and quickly.For this reason, interest in CAD-based deep learning models has been increased significantly. In this study, we propose a “MultiNet” framework based on the transfer learning concept to classify different breast cancer types using two publicly available datasets that include 7909 and 400 microscopic breast images, respectively. The proposed “MultiNet” framework is designed to provide fast and accurate diagnostics for breast cancer with binary classification (benign and malignant) and multi-class classification (benign, in situ, invasive, and normal). In the proposed framework, features from microscopy images are extracted using three well-known pre-trained models, including DenseNet-201, NasNetMobile, and VGG16. The extracted features are then fed into the concatenate layer, making a robust hybrid model. The proposed framework yields an overall classification accuracy of 99% in classifying two classes. It also achieves 98% classification accuracy in classifying four classes. Such promising results will provide the opportunity to use “MultiNet” framework as a diagnostic model in clinics and health care.
ONEST (Observers Needed to Evaluate Subjective Tests) suggests four or more observers for a reliable assessment of the consistency of histological grading of invasive breast carcinoma: A reproducibility study with a retrospective view on previous studies
2022, Pathology Research and Practice
Histological grade is one of the most important prognosticators of breast cancer which is available for nearly all cases. It also makes part of several multivariable analysis derived combined prognostic profiles despite concerns about its reproducibility. The aims included a reproducibility study of grading in the light of a recently described statistical approach, ONEST (Observers Needed to Evaluate Subjective Tests) and review earlier reproducibility studies in the light of the ONEST analysis. Nine pathologists reviewed 50 core needle biopsies and 50 slides from different excision specimens and recorded the scores for gland (tubule) formation, nuclear pleomorphism and mitotic activity as well as histological grade. Overall percent agreement, Fleiss kappa and the intraclass correlation coefficient (ICC) were used for the analysis of reproducibility. ONEST data and curves were generated from 100 random permutations of the participants. ONEST suggested a minimum of 4 observers for the reliable evaluation of reproducibility for both the scored components and grade in either type of specimen. Our results suggested moderate or moderate to good reproducibility of grading (kappa values of 0.51 for excisions, and 0.54 for biopsies and ICCs of 0.70 and 0.69, respectively) with gland formation being the most and nuclear pleomorphism the worst consistently evaluated feature. In studies with sufficient participants (at least 4) and non-pairwise comparisons in the analysis, the reproducibility of histological grading is fair to moderate, whereas studies with fewer participants or pairwise kappa analysis suggest moderate to almost prefect agreement of the results. ONEST is a valuable complementation of reproducibility analyses.
Automated detection and grading of Invasive Ductal Carcinoma breast cancer using ensemble of deep learning models
2021, Computers in Biology and Medicine
Invasive ductal carcinoma (IDC) breast cancer is a significant health concern for women all around the world and early detection of the disease may increase the survival rate in patients. Therefore, Computer-Aided Diagnosis (CAD) based systems can assist pathologists to detect the disease early. In this study, we present an ensemble model to detect IDC using DenseNet-121 and DenseNet-169 followed by test time augmentation (TTA). The model achieved a balanced accuracy of 92.70% and an F1-score of 95.70% outperforming the current state-of-the-art. Comparative analysis against various pre-trained deep learning models and preprocessing methods have been carried out. Qualitative analysis has also been conducted on the test dataset. After the detection of IDC breast cancer, it is important to grade it for further treatment. In our study, we also propose an ensemble model for the grading of IDC using the pre-trained DenseNet-121, DenseNet-201, ResNet-101v2, and ResNet-50 architectures. The model is inferred from two validation cohorts. For the patch-level classification, the model yielded an overall accuracy of 69.31%, 75.07%, 61.85%, and 60.50% on one validation cohort and 62.44%, 79.14%, 76.62%, and 71.05% on the second validation cohort for 4×, 10×, 20×, and 40× magnified images respectively. The same architecture is further validated using a different IDC dataset where it achieved an overall accuracy of 90.07%. The performance of the models on the detection and grading of IDC shows that they can be useful to help pathologists detect and grade the disease.
Clinical impact of LncRNA XIST and LncRNA NEAT1 for diagnosis of high-risk group breast cancer patients
2021, Current Problems in Cancer
Citation Excerpt :
The remaining study participants (n = 22) were healthy volunteers recruited as a control group from usual medical checkups. Breast diseases’ patients were categorized according to their histopathological examination into 121 females with primary BC, who were histologically grouped according to their staging and grading,22,23 and 35 females with benign breast lesions. Hormone receptors in BC patients were examined using an immunohistochemical method as previously reported.24
Long noncoding RNAs (lncRNAs) are evolving as contributing biomarkers for many diseases. Among these lncRNAs, X inactive-specific transcript (XIST), and nuclear paraspeckle assembly transcript 1 (NEAT1) were studied as undesirable upregulated nucleic acid markers for unfavorable prognosis of cancer. The authors aimed to investigate their role as diagnostic markers for breast cancer (BC) patients with high-risk factors. Serum samples were obtained from BC patients (n = 121), patients with benign breast lesions (n = 35), and healthy volunteers (n = 22). Assessment of lncRNA XIST, and lncRNA NEAT1 expression was performed using real time PCR. Expression levels of the investigated lncRNAs were significantly higher in BC patients as compared to the other groups. Both lncRNAs were significantly correlated with BC laterality, lymph node involvement, and clinical stages. LncRNA NEAT1 reported a significant aberrant expression with pathological types, histological grading and, hormonal status. The sensitivity of lncRNA NEAT1 was superior for detection of BC with high risk-factors as compared to lncRNA XIST. In conclusion, the detection of lncRNAs in body fluids has demonstrated a significant importance for detecting BC patients with high-risk factors, and was related to hormonal receptors, thus may be used for determining the direction of treatment strategy.
Transfer Learning from Non-Medical Images to Medical Images Using Deep Learning Algorithms
2024, Frontiers in Health Informatics

View all citing articles on Scopus

¹: Dr Robbins is supported by the Prindiville Family Scholarship.

View full text

Original contributionHistological grading of breast carcinomas: A study of interobserver agreement

Abstract

The Breast

Pathology

Hum Pathol

Hum Pathol

The Breast

Hum Pathol

Hum Pathol

Hum Pathol

The importance of histologic grade in long-term prognosis of breast cancer: A study of 1010 patients, uniformly treated at the Institut Gustave-Roussy

J Clin Oncol

Prognostic significance of tumor grade in clinical trials of adjuvant therapy for breast cancer with axillary lymph node metastasis

Cancer

Pathological prognostic factors in breast cancer. I. The value of histological grade in breast cancer: Experience from a large study with long-term follow-up

Histopathology

Relationship among outcome, stage of disease and histologic grade for 22, 616 cases of breast cancer

The Nottingham prognostic index in primary breast cancer

Breast Cancer Res Treat

Original contribution
Histological grading of breast carcinomas: A study of interobserver agreement