Design: Retrospective diagnostic immunohistochemistry study with sequencing
Population: 16 collecting duct carcinomas and 35 upper tract urothelial carcinomas
Primary outcome: NF2/Merlin protein loss by immunohistochemistry
Effect: Loss in 10/16 (63%) collecting duct vs 0/35 (0%) urothelial carcinomas

Collecting duct carcinoma and upper tract urothelial carcinoma can look alike, especially on small biopsies, yet they are treated differently. Because NF2 alterations recur in collecting duct carcinoma and are uncommon in urothelial carcinoma, this study tested NF2/Merlin immunohistochemistry on 16 collecting duct carcinomas and 35 upper tract urothelial carcinomas. The study was published in June.

Loss of NF2/Merlin staining was seen in 10 of 16 collecting duct carcinomas (63%) and in none of 35 urothelial carcinomas. All six collecting duct carcinomas with an NF2 mutation showed loss; of seven without a detectable alteration, two still showed loss.

The marker is therefore specific in this differential but not sensitive: retained staining does not exclude collecting duct carcinoma. NF2/Merlin loss also occurs in other renal cell carcinoma subtypes, so it only helps once the differential has been narrowed.

The case numbers are small, as expected for a rare tumour, and the findings need confirming in other laboratories.

Add NF2/Merlin to the panel for an infiltrative renal carcinoma with tubules when collecting duct carcinoma and urothelial carcinoma are both possible.
Read loss as supporting collecting duct carcinoma once urothelial markers are negative.
Do not use retained staining to exclude collecting duct carcinoma: a third of cases kept it.
Remember NF2/Merlin loss is not specific among renal cell carcinoma subtypes.

Why it matters

The distinction changes treatment, and on a small biopsy one objective stain can tip a difficult call.

The statistics, in plain English

Zero of 35 urothelial carcinomas showed loss, which suggests high specificity, but with 35 cases the true false-positive rate could still be a few per cent. Sensitivity of 63% from 16 cases is an imprecise estimate.