Design: Retrospective institutional cohort with paired IHC–NGS comparison
Population: 5,233 NSCLCs with NGS; 105 tested by both IHC and NGS
Primary outcome: Prevalence of MTAP deletion and IHC–NGS concordance
Effect: Deletion in 5.6%; concordance 97.3%

MTAP, on chromosome 9p21, is often deleted alongside CDKN2A, and its loss creates a metabolic weakness that targeted drugs are being developed to exploit. A US institutional review of 5,233 non-small cell lung cancers with a 447-gene NGS panel found homozygous MTAP deletion in 5.6%. It was more common in adenocarcinomas from never or light smokers and in tumours with EGFR mutations or ALK fusions.

In 105 cases tested by both methods, IHC and NGS agreed in 97.3% once sequencing from samples with 30% tumour or less was excluded and careful quality control applied. Of the two remaining discordant cases, one was a sequencing misinterpretation in a non-diploid genome and one showed complete protein loss with no detectable deletion. About 13% of tumours with loss stained heterogeneously, suggesting subclonal events.

The practical message: MTAP IHC is a reliable, cheap partner to NGS, and the better choice when the tumour fraction is low or the deletion may be subclonal. Interpreting heterogeneous staining needs care. MTAP status is not yet tied to an approved lung cancer therapy, so this is about building the assay before the drug arrives.

Validate MTAP immunohistochemistry in-house now, so it is ready if MTAP-directed therapies are approved.
Prefer IHC over NGS for MTAP status when tumour content is 30% or less.
Report heterogeneous MTAP staining explicitly; about one in eight losses was patchy.
Interpret NGS copy-number calls cautiously in tumours with complex, non-diploid genomes.
Expect MTAP loss more often in never-smoker adenocarcinomas and those with EGFR or ALK alterations.

Why it matters

Low-purity lung biopsies are common, and a cheap stain can rescue a biomarker result that sequencing would miss.

The statistics, in plain English

A 97.3% concordance in 105 cases means about three disagreements per hundred. The comparison excluded NGS results from low-purity tumours, which is exactly where IHC is said to perform better — so the concordance figure applies to good-quality samples.