Artificial intelligence modelling in differentiating core biopsies of fibroadenoma from phyllodes tumor

A Preliminary Investigation into Search and Matching for Tumor Discrimination in World Health Organization Breast Taxonomy Using Deep Networks

Breast cancer is one of the most common cancers affecting women worldwide. It includes a group of malignant neoplasms with a variety of biological, clinical, and histopathologic characteristics. There are more than 35 different histologic forms of breast lesions that can be classified and diagnosed histologically according to cell morphology, growth, and architecture patterns. Recently, deep learning, in the field of artificial intelligence, has drawn a lot of attention for the computerized representation of medical images. Searchable digital atlases can provide pathologists with patch-matching tools, allowing them to search among evidently diagnosed and treated archival cases, a technology that may be regarded as computational second opinion. In this study, we indexed and analyzed the World Health Organization breast taxonomy (Classification of Tumors fifth ed.) spanning 35 tumor types. We visualized all tumor types using deep features extracted from a state-of-the-art deep-learning model, pretrained on millions of diagnostic histopathology images from the Cancer Genome Atlas repository. Furthermore, we tested the concept of a digital "atlas" as a reference for search and matching with rare test cases. The patch similarity search within the World Health Organization breast taxonomy data reached >88% accuracy when validating through "majority vote" and >91% accuracy when validating using top n tumor types. These results show for the first time that complex relationships among common and rare breast lesions can be investigated using an indexed digital archive.

Precise diagnosis of breast phyllodes tumors using Raman spectroscopy: Biochemical fingerprint, tumor metabolism and possible mechanism

BACKGROUND

Breast fibroadenomas and phyllodes tumors are both fibroepithelial tumors with comparable histological characteristics. However, rapid and precise differential diagnosis is a tough point in clinical pathology. Given the tendency of phyllodes tumors to recur, the difficulty in differential diagnosis with fibroadenomas leads to the difficulty in optimal management for these patients.

METHOD

In this study, we used Raman spectroscopy to differentiate phyllodes tumors from breast fibroadenomas based on the biochemical and metabolic composition and develop a classification model. The model was validated by 5-fold cross-validation in the training set and tested in an independent test set. The potential metabolic differences between the two types of tumors observed in Raman spectroscopy were confirmed by targeted metabolomic analysis using liquid chromatography-tandem mass spectrometry (LC-MS/MS).

RESULTS

A total of 204 patients with formalin-fixed paraffin-embedded (FFPE) tissue samples, including 100 fibroadenomas and 104 phyllodes tumors were recruited from April 2014 to August 2021. All patients were randomly divided into the training cohort (n = 153) and the test cohort (n = 51). The Raman classification model could differentiate phyllodes tumor versus fibroadenoma with cross-validation accuracy, sensitivity, precision, and area under curve (AUC) of 85.58 % ± 1.77 %, 83.82 % ± 1.01 %, 87.65 % ± 4.22 %, and 93.18 % ± 1.98 %, respectively. When tested in the independent test set, it performed well with the test accuracy, sensitivity, specificity, and AUC of 83.50 %, 86.54 %, 80.39 %, and 90.71 %. Furthermore, the AUC was significantly higher for the Raman model than that for ultrasound (P = 0.0017) and frozen section diagnosis (P < 0.0001). When it came to much more difficult diagnosis between fibroadenoma and benign or small-size phyllodes tumor for pathological examination, the Raman model was capable of differentiating with AUC up to 97.45 % and 95.61 %, respectively. On the other hand, targeted metabolomic analysis, based on fresh-frozen tissue samples, confirmed the differential metabolites (including thymine, dihydrothymine, trans-4-hydroxy-l-proline, etc.) identified from Raman spectra between phyllodes tumor and fibroadenoma.

SIGNIFICANCE AND NOVELTY

In this study, we obtained the molecular information map of breast phyllodes tumors provided by Raman spectroscopy for the first time. We identified a novel Raman fingerprint signature with the potential to precisely characterize and distinguish phyllodes tumors from fibroadenoma as a quick and accurate diagnostic tool. Raman spectroscopy is expected to further guide the precise diagnosis and optimal treatment of breast fibroepithelial tumors in the future.

Refining the classification of breast phyllodes tumours

Phyllodes tumours of the breast are uncommon fibroepithelial neoplasms that pose recurrent classification challenges, in large part due to the multiple histological parameters of stromal hypercellularity and atypia, stromal mitotic count, stromal overgrowth and tumour borders, that are used for grading. While the World Health Organization (WHO) Classification of Breast Tumours provides recommendations on diagnostic features, defining criteria are not always applied in routine practice. Lack of concordance among pathologists in typing and grading further underscores the classification difficulties, especially in the borderline category. Although there has been significant molecular information on phyllodes tumours in recent years which has been diagnostically helpful, it has not been translated into daily clinical practice. In order to refine the classification of phyllodes tumours into one that is simple yet comprehensive, reproducible and prognostically precise, a multipronged approach is needed that leverages on global contributions of the International Fibroepithelial Consortium, support by the International Collaboration on Cancer Classification and Research (IC³ R) in amalgamating evidence translation, and guidance from the International Collaboration on Cancer Reporting (ICCR) for standardised reporting. It is hoped that the evidence generated can be used towards refining the classification of phyllodes tumours for the future.

Survey of recurrent diagnostic challenges in breast phyllodes tumours

BACKGROUND

Breast phyllodes tumours (PTs) are graded as benign, borderline, or malignant by analysis of multiple histological features. PT grading is often inconsistent, likely due to variation in the weighting of grading criteria by pathologists.

DESIGN

The hierarchy of use of diagnostic criteria was identified using a 20-question survey.

RESULTS

In all, 213 pathologists from 29 countries responded. 54% reported 10-50 PT cases per year. Criteria considered key to PT diagnosis were: increased stromal cellularity (84.3%), stromal overgrowth (76.6%), increased stromal mitoses (67.8%), stromal atypia (61.5%), stromal fronding (59.0%), periductal stromal condensation (58.0%), irregular tumour borders (46.3%), and/or lesional heterogeneity (33.7%). The importance of grading parameters were: mitotic activity (55.5%), stromal overgrowth (54.0%), stromal atypia (51.9%), increased stromal cellularity (41.7%), and nature of the tumour border (38.9%). 49% would diagnose malignant PT without a full array of adverse features. 89% used the term "cellular fibroepithelial lesion (FEL)" for difficult cases; 45% would diagnose an FEL with stromal fronding (but lacking other PT features) as fibroadenoma (FA), 35% FEL, and 17% PT. 59% deemed clinico-radiological findings diagnostically significant; 68% considered age (≥40 years) important in determining if an FEL was a FA or PT. In FELs from young patients, increased stromal cellularity (83%), fronding (52%), and mitoses (41%) were more common. 34% regarded differentiating cellular FA from PT as a specific challenge; 54% had issues assigning a borderline PT grade.

CONCLUSION

Criteria for grading PT lie on a spectrum, leading to interpretive variability. The survey highlights the criteria most used by pathologists, which do not completely align with WHO recommendations.

Discrimination between phyllodes tumor and fibro-adenoma: Does artificial intelligence-aided mammograms have an impact?

Background

The indulgence of artificial intelligence (AI) has been considered recently in the work up for the detection and diagnosis of breast cancer through algorithms that could supply diagnosis as the radiologist do. The algorithm learns from a supervised and continuous input of large and new data sets unlike the standard programming, which requires clear step-by-step instructions. The aim of this study is to assess the ability of AI scanned mammograms to aid the ultrasound in the discrimination between phyllodes tumors and fibro-adenomas.

Results

This is a retrospective analysis included 374 proven phyllodes tumors (PT) and fibro-adenomas (FA). Digital mammogram and breast ultrasound was performed for all the cases and each breast was given a “Breast Imaging Reporting and Data System” (BI-RADS) score. Included mammograms were scanned by AI with resultant a qualitative heatmap and a quantitative abnormality scoring of suspicion percentage.

The study included 164 PT (43.9%) and 210 FA (56.1%). BI-RADS category 2 was assigned in 40.1%, category 3 in 38.2%, category 4 in 18.5% and category 5 in 3.2% with median value of the AI abnormality scoring of 23%, 44%, 65% and 90% respectively. Sensitivity and specificity of the conventional imaging were 59.2% and 75.8% respectively. The AI abnormality scoring of 49.5% upgraded the sensitivity to 89.6% and specificity to 94.8% in the ability to discriminate PT from FA masses.

Conclusion

Artificial intelligence-aided mammograms could be used as method of distinction between PT from FA detected on sono-mammogram. The color hue and the quantification of the abnormality scoring percentage could be used as a one setting method for specification and so guide clinicians in their decision of conservative management or the choice of the surgical procedure.