Orthogonal validation of anaplastic lymphoma kinase (ALK) immunohistochemistry with molecular analysis for ALK gene rearrangement is required to finetune staining protocols with the D5F3 clone and can impact external quality assessment.

Immunohistochemistry (IHC) for anaplastic lymphoma kinase (ALK) is used to screen tumours for ALK gene rearrangements, which generally result in ALK protein overexpression, depending on the fusion partner. Here, an interlaboratory validation of three IHC protocols using the D5F3 clone was performed. ALK IHC results were retrospectively compared with available molecular data on ALK fusions for 462 tumours, as part of an internal quality assessment. ALK IHC was read as 'negative', 'equivocal', or 'positive'. Two ALK IHC protocols showed increased staining intensity in the appendiceal myenteric plexus, but they also introduced diffuse, weak cytoplasmic staining in most non-ALK-rearranged epithelial and mesenchymal tumours, leading to equivocal or false-positive ALK IHC readout. One less sensitive protocol was more specific, as it showed good concordance with the molecular data, without extensive background staining: 33 of 34 ALK-rearranged tumours were positive (97% sensitivity), and 363 of 428 lesions without ALK fusions were negative (85% specificity). The negative and positive predictive values were 99% and 34%, respectively. Three of 58 equivocal tumours (5%) were ALK-rearranged, in contrast with 30 out of 40 positive tumours (75%). Despite its good concordance with ALK rearrangements, and thus good sensitivity, external quality assessment repeatedly considered this protocol as insufficiently sensitive. Our validation study shows that overly sensitive ALK IHC assays could result in false-positive tumours presenting with difficult-to-read focal or diffusely weak immunoreactivity, which could lead to potential overuse of confirmatory molecular tests. We therefore recommend carefully fine-tuning ALK IHC assays with the D5F3 clone by comparing diagnostic performance with molecular data.