Design: Systematic review and meta-analysis
Population: 35 studies, 1,146 histological and 35 cytological salivary tumour samples
Primary outcome: Pooled SOX10 positivity by tumour type
Effect: Pleomorphic adenoma 95.8%; acinic cell 97.4%; Warthin's 2.3%; salivary duct carcinoma 3.7%

This meta-analysis pooled 35 studies of SOX10 immunohistochemistry in primary salivary gland tumours: 1,146 histological and 35 cytological samples.

Pooled positivity was high in pleomorphic adenoma (96%), polymorphous adenocarcinoma (97%), canalicular adenoma (99%), basal cell adenoma (91%), adenoid cystic carcinoma (97%), acinic cell carcinoma (97%), intercalated duct-type intraductal carcinoma (91%) and microsecretory carcinoma (96%), and somewhat lower in myoepithelioma (75%). It was low in Warthin's tumour (2%), oncocytoma (8%), salivary duct carcinoma (4%) and apocrine intraductal carcinoma (10%). Polyclonal antibodies and the EP268, N-20 and A-2 clones performed alike; the BC34 clone stained fewer positive tumours.

The practical uses are a positive SOX10 supporting acinic cell over secretory-pattern mimics, and a negative result favouring salivary duct carcinoma or oncocytic lesions. Because the clone matters, a negative stain with BC34 should be read with caution, and the clone belongs in the report.

Use SOX10 positivity to support acinic cell, adenoid cystic or pleomorphic adenoma in a salivary differential.
Expect SOX10 negativity in Warthin's tumour, oncocytoma and salivary duct carcinoma.
Check which antibody clone your laboratory uses; BC34 stained fewer positive tumours.
State the clone in reports where a SOX10 result drives the diagnosis.
Interpret myoepithelioma with care; only three-quarters were positive.

Why it matters

A negative stain may reflect the reagent rather than the tumour, which changes how it should be read.

The statistics, in plain English

Pooled positivity is the average proportion of each tumour type that stained. Some intervals run above 100% because of how small proportions were pooled; read those as 'nearly all'. Wide intervals, as for myoepithelioma, reflect few or inconsistent studies.