The subtype and grade on a cutaneous squamous cell carcinoma report

**The subtype.** The UK dataset uses a modified WHO classification and recognises that a squamous cell carcinoma may come from surface epidermis or from hair follicle epithelium, either of which can be in situ or invasive and low or high risk (RCPath G124). If none of the named subtype features are present, the tumour is reported as **no special type**, also called classic, and that is what most are. The subtypes national guidelines and NICE treat as clinically high risk are **acantholytic** (cells falling apart from one another, leaving pseudoglandular spaces), **desmoplastic** (defined as having more than 30 percent desmoplastic stroma) and **spindle cell** or sarcomatoid. The dataset notes a real disagreement about spindle cell tumours: the Armed Forces Institute of Pathology accepts spindle cell carcinoma arising after radiotherapy as high risk but regards spindle cell tumours on sun-exposed skin as not carrying the same aggressive potential. **Adenosquamous** carcinoma is high risk to WHO. Invasive squamous cell carcinoma with adjacent Bowen's disease is usually treated as high risk, although there is increasing debate about whether that should be restricted to areas not exposed to ultraviolet light. **Basaloid** squamous cell carcinoma is uncommon in skin, must be told from basal cell carcinoma with immunohistochemistry (BerEP4 and EMA), may arise in pre-existing basaloid Bowen's disease, and is poorly differentiated by definition.

**The grade.** Both UICC and AJCC equate grades 1 to 4 with well, moderately, poorly and undifferentiated, and accept that 3 and 4 can be combined. The UK dataset considered the original Broders method, which scores the percentage of well-differentiated tumour present, and after consultation adopted a three-grade system instead. Well differentiated tumours show easily recognisable and often abundant keratinisation, obvious squamous cells with visible intercellular bridges, minimal pleomorphism and mostly basal mitoses. Moderately differentiated tumours are more disorganised, with more nuclear and cytoplasmic pleomorphism, more and abnormal mitoses, and keratin limited to keratin pearls, horn cysts and scattered single keratinised cells. In poorly differentiated tumours it may be hard to establish the nature of the lesion at all unless intercellular bridges or small foci of keratinisation are found; rarely the tumour is completely anaplastic and keratin immunohistochemistry is needed.

The grading rule matters and is not obvious. AJCC gives no guidance on what percentage of a differentiated component sets the grade, so the UK dataset follows the widely used approach of classifying a tumour by **its most poorly differentiated region, irrespective of the percentage present**. A tumour that is 95 percent well differentiated with one poorly differentiated focus is a poorly differentiated tumour on a British report. Recording the percentages of each component is optional. Loss of differentiation correlates with clinical risk, which is the whole reason the grade is there.