A neoplasm is a population of abnormal cells which persist and proliferate excessively.

Almost all cell types and tissues give rise to neoplasia: epithelium, connective tissue, bone marrow, lymphoid, neural tissue &c, but nine out of 10 cancers are invasive carcinomas, that is, malignant neoplasms of epithelial origin.

In this article, we talk about the importance of understanding where cancer comes from as we seek earlier detection and appropriate treatment.

Many steps to cancer

Neoplasms arise by way of repeated cycles of clonal expansion which establish populations of cells with accumulated mutations and epigenetic alterations permitting and promoting increasingly abnormal cell behaviour until the threshold of malignancy is passed.

Above figure: Standard model of neoplasia. The tree represents a multicellular organism. An abnormal cell acquires a survival advantage that it transmits to its descendants. This cycle repeats; eventually, a clone emerges with invasive and metastatic potential (in this case, clone 5). Intermediate stages may be identifiable as intraepithelial neoplasia (dysplasia).

Neoplastic but not yet malignant cell populations are often identifiable in histological sections by their abnormal appearance, individually and collectively. The ability to identify such lesions is important, as they are potentially treatable, and the increased risk of cancer associated with them may justify enhanced screening or surveillance. Classic examples of treatable pre-malignant lesions include cervical intraepithelial neoplasia and dysplasia in Barrett’s oesophagus.

Let’s mind our language

The terminology of these precursor lesions is complex and confusing, with a jumble of names and qualifiers for equivalent phenomena at different sites, including atypical hyperplasia, dysplasia, carcinoma in situ, intraepithelial neoplasia and intraepithelial lesion.

Dysplasia was classically stratified on a four-category scale:

1. Mild dysplasia
2. Moderate dysplasia
3. Severe dysplasia
4. Carcinoma in situ.

From this terminology, a reasonable person might infer a qualitative difference between dysplasia in its various degrees and carcinoma in situ, but this would be a mistake. There is no difference between the likely future behaviour of ‘severe dysplasia’ and ‘carcinoma in situ’ at any site.

(Another undesirable aspect of the term ‘carcinoma in situ’ is that it links a lesion by definition not invasive, and therefore not in the most widely accepted sense a malignant neoplasm, with a lesion which is unquestionably malignant; because the term ‘carcinoma’ most often means an established invasive malignancy. More logical terminology abolishes this artificial distinction).

Imperfect inter-observer agreement and the need to specify action thresholds, also reinforce the case for simpler terminology and fewer categories.

Making sense of precursor lesions 

Above figure: normal cervical squamous epithelium, contrasted with CIN 2. In the normal epithelium (left), cells provided by asymmetric division of stem cells in the basal layer enter the suprabasal proliferation zone, which is about one quarter of the thickness of the epithelium. As they migrate towards the surface of the epithelium they leave the cell cycle and differentiate. In CIN 2 (right), the proliferation zone is expanded to occupy at least three-quarters of the epithelium. There is a mitosis near the surface (ringed), where it should not be. A narrow zone of partial differentiation remains.

In the uterine cervix, cervical intraepithelial neoplasia (CIN) was introduced as a three-category scale:

1. CIN1 (mild dysplasia)
2. CIN2 (moderate dysplasia)
3. CIN3 (severe dysplasia and carcinoma in situ)

and CIN has been further simplified further to LSIL (CIN1) and HSIL (CIN2+CIN3): low and high grade ‘squamous intraepithelial lesion’. A comparable pragmatic approach is taken in the GI tract where dysplasia (glandular and squamous) is classified as low grade or high grade.

The table below summarises nomenclature at some representative sites:

Low grade/high-grade thresholds are site-specific; moderate dysplasia is low grade in the GI tract, but high grade in the cervix.

Comparable simplification of terminology has not occurred for breast and remains confusing. This reflects its complex biology and a conservative approach to terminology by workers in the field. This is one of the few tissues in which, for better or worse, carcinoma in situ is retained as a diagnosis.

Two patterns of atypical hyperplasia (that is, dysplasia) and carcinoma in situ are recognised in the breast: ‘ductal’ and ‘lobular’. Despite these names both of them characteristically involve both the glandular lobules as well as larger ducts. The ‘atypical hyperplasias’ resemble but do not fully meet the criteria for the diagnosis of carcinoma in situ.

Accurate diagnosis of the atypical hyperplasias and carcinoma in situ in breast is all about architecture and context. It is extremely important not to miss a diagnosis of invasive carcinoma, with its potential for metastasis to distant sites via lymphatics or via the blood. Very often this is straightforward but can be challenging.

In summary: Let’s agree on what we’re talking about!

Above figure: In the upper two panels, we see neoplastic cells of ductal carcinoma in situ (DCIS) replacing normal epithelium lining a duct (left) and (right) colonising glandular acini of breast tissue in the same patient. The very different architecture of these two areas is retained from normal structures now occupied by neoplastic cells.

The panel below is an invasive carcinoma. The tissue architecture retained in DCIS (although not necessarily easy to recognise) is lost completely, with irregular, angulated glandular acini embedded in an abnormal stroma of plump myofibroblasts and myxoid extracellular matrix. This stromal desmoplastic reaction is an important clue to invasive malignancy.

Nomenclature wars have been fought between ‘lumpers’ and ‘splitters’. Application of Occam’s razor (‘entities are not to be multiplied more than necessary’) is frequently of use in the pruning of unnecessary complexity. Roger Sessions expressed it admirably: everything should be as simple as it can be but not simpler! (a remark usually attributed to Einstein).

The message to workers dealing with histological precursors of cancer is to be sure you understand and agree on the terminology you and your collaborators will use for the tissues you are interested in, and the criteria defining your categories. Don’t overcomplicate things.

This is an area in which agreement between human observers is notoriously imperfect. It remains to be seen how well image analysis and deep learning will improve our accuracy.