The 2019 World Health Organization classification of tumours of the breast

The newly published World Health Organization (WHO) Classification of Tumours of the breast features significant changes compared to earlier editions. In this review, we outline the major changes in this important reference source for those diagnosing tumours, or engaged in cancer research, and describe the significant changes. For breast cancer, the overview acknowledges the treatment-relevant subtypes of invasive carcinoma (based on ER and HER2 status) and new data is added to support the differences in pathogenesis, treatment response and prognosis of these clinically relevant groupings. The WHO Classification of Tumours is increasingly evidence-based, with a clear update cycle, improved quality of illustrations, as well as content, led by an editorial board comprising pathologists, but increasingly incorporating input from other disciplines. The advent of the new website allows the use of whole slide images, and hyperlinks to evidence or external bodies that produce guidance on staging or reporting.


Introduction
The classification of breast tumours continues to evolve, with the integration of new knowledge from research rapidly being translated into clinical practice. Major changes are shown in Table 1. In this volume of the World Health Organization (WHO) classification of tumours series' fifth edition, which is an update of the fourth-edition breast tumours volume published in 2012, 1 the descriptions of breast tumours follow the familiar systematic approach of previous volumes, with the content now organised in sequence from benign epithelial proliferations and precursors, through benign neoplasms, to in-situ and invasive breast cancer, followed by mesenchymal and haematolymphoid neoplasms, tumours of the male breast, and genetic tumour syndromes.
A brief introduction prefaces the content pertaining to each major tumour group, to provide a general perspective and highlight key modifications. In the current volume, information on epidemiology, imaging, clinical features, grading, staging, molecular testing for hormone receptors and ERBB2 (HER2), post-therapy effects, core needle biopsy and fine-needle aspiration considerations, molecular pathology and genomics is now presented in the general overview that introduces the sections on invasive breast carcinoma (IBC), rather than in the first chapter as in the previous edition. Core biopsy diagnosis, an important preoperative tool, is addressed across multiple sections. The importance of molecular pathology in aiding diagnosis is recognised, with a specific subsection for each tumour type. Essential and desirable diagnostic criteria are also included, to reinforce key histopathological clues.

Breast carcinoma
Invasive breast cancers are still organised into chapters by their morphological subtypes, which remain clinically relevant. However, as the majority of cases are of no special type (NST), additional prognostic and predictive factors that aid significantly in treatment and outcome stratification are also focused on and reviewed in more depth in the invasive carcinoma overview section. The overview acknowledges the treatment-relevant subtypes of invasive carcinoma [based on oestrogen receptor (ER) and HER2 status], and new data are added to support the differences in pathogenesis, treatment response and prognosis of these clinically relevant groupings. Updates in defining and testing hormone receptor and HER2 status are presented, and there are updated sections on additional assays and parameters used in prediction and prognosis (including proliferation markers, androgen receptor, response to neoadjuvant therapy, gene expression assays, tumour-infiltrating lymphocytes (TILs), prognostic scoring systems, and programmed death-ligand 1 testing). The overview section of the molecular classification of breast cancers is also updated to include more recent data supporting classification schemes that have prognostic associations (including the intrinsic subtypes, integrative cluster subgroups, triple-negative subclassifications, and mutation-based profiling).
Standard prognostic indicators, such as tumour size, lymph node status, and Nottingham grade, continue to be highly relevant. An important change in this edition is the conversion of mitotic count from the traditional denominator of 10 high-power fields to a defined area expressed in mm 2 . This serves to standardise the true area over which mitoses are enumerated, because different microscopes have highpower fields of different sizes. This change will also be helpful for anyone reporting using digital systems. The score thresholds for mitotic counts based on the diameter of the high-power field and its corresponding area are shown in Table 2.
Updates to the 'Invasive breast carcinoma, NST' section include a revised definition of the mixed NSTspecial subtype (now expanded to include cases with 10-90% special subtype admixed with NST, with a recommendation to include parameters regarding both components). Classification of several patterns previously recognised as separate special rare subtypes have moved under the NST umbrella as 'special morphological patterns'. Carcinomas previously classified as the special subtype 'carcinoma with medullary features' (including medullary carcinoma, atypical medullary carcinoma, and invasive carcinoma NST with medullary features) have suffered from poor interobserver reproducibility and overlap in features with carcinomas that have basal-like molecular profiles and carcinomas associated with BRCA1 mutations. In addition, the increasing affirmation of the prognostic importance of TILs in high-grade breast cancers in explaining their good prognosis, including high-grade cancers not meeting strict medullary criteria, reduces the requirement for discrete separation of these tumours, which exist along a morphological continuum. Therefore, for clinical purposes, it is now proposed to consider carcinomas with a medullary pattern as representing one end of the spectrum of the TIL-rich IBC-NSTs rather than a distinct morphological subtype, and to use the term 'IBC-NST with medullary pattern'. In addition, oncocytic, lipid-rich, glycogen-rich clear cell, and sebaceous carcinomas, which are rarely encountered, are also now recognised as special patterns of NST along with carcinoma with osteoclast-like stromal giant cells, pleomorphic carcinoma, and choriocarcinomatous and melanotic patterns. Inflammatory, bilateral and non-synchronous breast carcinomas are also now recognised as distinct clinical presentations rather than special subtypes of breast cancer.
As well as classic lobular carcinoma in situ (LCIS), the pleomorphic and florid subtypes are now recognised. Pleomorphic LCIS shows marked nuclear atypia, and may include apocrine features, whereas in florid LCIS there is marked distension of terminal duct lobular units or ducts, often forming a mass-like appearance. It is now recognised that some invasive lobular carcinomas may be associated with extracellular mucin production.

Neuroendocrine tumours
Although neuroendocrine neoplasms (NENs) are allocated their own section, harmonised with those of other organ systems on the basis of a recent WHO workshop report, 2 it must be emphasised that true primary neuroendocrine tumours (NETs) of the breast remain uncommon and poorly defined. According to the proposed consensus terminology, well-differentiated NETs broadly correspond to grade 1 (carcinoid-like) and grade 2 (atypical carcinoid-like) tumours (regarded as carcinomas in the breast), whereas poorly differentiated neuroendocrine carcinomas (NECs) are typified by small-cell and large-cell carcinoma. Many breast tumours that show varying degrees of neuroendocrine differentiation belong to recognised entities, such as hypercellular mucinous carcinoma and solid papillary carcinoma of both in-situ and invasive types. Small-cell NEC does arise in the breast, often admixed with invasive carcinoma NST. Large-cell NEC has been added as an entity arising in the breast, albeit encountered very rarely. For well-differentiated NETs resembling carcinoid or atypical carcinoid tumour, it is prudent to exclude metastasis from another site. It is recommended that the classification of breast tumours showing neuroendocrine expression be based on the recognisable morphological tumour type, such as invasive carcinoma NST, mucinous carcinoma, or solid papillary carcinoma. 3 Because some degree of neuroendocrine expression is relatively common in invasive breast cancer NST, most breast cancers with neuroendocrine expression will ultimately be classified as invasive carcinoma NST with neuroendocrine differentiation. Only if neuroendocrine histological features and neuroendocrine marker expression are distinct or uniform enough to classify a cancer as one of the rare NETs or NECs of the breast should NEN terminology be used. NET or NEC of the breast are currently treated on the basis of standard breast cancer parameters (such as ER and HER2 status). The new WHO classification is not advocating routine evaluation for neuroendocrine markers in breast cancers.

Other tumour types and newly recognised entities
One important change in the classification of fibroepithelial tumours is the removal of well-differentiated liposarcoma as a histological criterion of malignancy in breast phyllodes tumours in the absence of additional supporting microscopic alterations. Evidence has emerged that these abnormal adipocyte populations residing within phyllodes tumours do not harbour the MDM2 aberrations that characterise welldifferentiated liposarcomas elsewhere. In light of the consensus opinion that this heterologous element does not have metastatic potential, it was agreed that A new entity included in this volume is mucinous cystadenocarcinoma, a unique invasive malignancy with a relatively good prognosis, featuring luminal mucin and cytomorphology resembling pancreatobiliary and ovarian mucinous cystadenocarcinoma. The entity 'tall cell carcinoma with reversed polarity' is introduced in the section about rare and salivary gland-type tumours, as there have been multiple reports of this entity, previously termed 'breast tumour resembling the tall cell variant of papillary thyroid carcinoma' as well as 'solid papillary carcinoma with reverse polarity', with these descriptions united by the consistent finding of IDH2 and PIK3CA mutations. 4 The new terminology of 'tall cell carcinoma with reversed polarity' incorporates portions of earlier terms used to describe this entity-'tall cell' and 'reversed polarity'. This revised term was a consensus agreement achieved during the WHO Editorial Board Meeting. It was also felt that having 'papillary thyroid carcinoma' in the terminology may be confusing and misleading. Periductal stromal tumour is now considered to be a variant of phyllodes tumour.
Mesenchymal tumours, haematolymphoid tumours and genetic tumour syndromes are covered in dedicated chapters in alignment with the approach being taken throughout this fifth edition of the series.

Conclusion
Tumour classification is a dynamic process, integrating multiple sources of information that have emerged since the previous WHO update. Digital pathology, which is becoming widely available, may enable the application of new artificial intelligence and computer learning tools to refine breast and other tumour classifications that will ultimately facilitate appropriate therapy and accurate prognostication.
Belgium, 14 University Medical Centre Utrecht, Utrecht, The Netherlands, and 15 International Agency for Research on Cancer, Lyon, France