Diagnostic accuracy of methylthioadenosine phosphorylase immunohistochemistry for CDKN2A homozygous deletion and for distinguishing mesothelioma from reactive mesothelial proliferations: A systematic review and bivariate meta‐analysis

Abstract Distinguishing mesothelioma from reactive mesothelial proliferations is a recurring problem in serous effusion and small‐biopsy practice, where invasion cannot be assessed and diagnosis rests on cell block immunohistochemistry. Loss of methylthioadenosine phosphorylase (MTAP) immunostaining is an inexpensive surrogate for CDKN2A homozygous deletion; however, its accuracy used alone has not been quantified. This is a systematic review of diagnostic test accuracy studies (PRISMA‐DTA, QUADAS‐2, and Open Science Framework). Four databases were searched to June 18, 2026. Eligible studies reported MTAP immunohistochemistry alone with an extractable 2 × 2 table against CDKN2A homozygous deletion (primary reference) or the integrated mesothelioma‐versus‐reactive diagnosis (secondary). Bivariate/hierarchical summary receiver operating characteristic (HSROC) meta‐analysis used mada, with a prespecified univariate summary where specificity approached 100%. Nineteen studies contributed 21 two‐by‐two tables, five of nine secondary arms being effusion cytology cell blocks. Against CDKN2A homozygous deletion (seven studies), pooled sensitivity was 75.9% (95% CI, 56.6%–88.4%) and specificity was 79.4% (95% CI, 65.8%–88.6%); the area under the HSROC curve (AUC) was 0.845. For mesothelioma versus reactive proliferations (nine pleural arms), pooled sensitivity was 52.0% (95% CI, 44.4%–59.5%), with specificity at or near 100% (bivariate sensitivity, 51.7%; bivariate specificity, 95.5%; AUC, 0.758). Peritoneal data were too sparse to pool (one cohort; sensitivity, 33.3%). Sensitivity did not differ by specimen format (histology, 51.6%; effusion cytology, 52.9%). MTAP immunohistochemistry shows useful but imperfect concordance with CDKN2A homozygous deletion, and high specificity for the mesothelioma‐versus‐reactive distinction. On a cell block or small biopsy, a confident complete‐loss call is essentially specific for malignancy; molecular confirmation by fluorescence in situ hybridization, copy number analysis, or a next‐generation sequencing panel should be retained where MTAP is equivocal or uninterpretable, and for peritoneal disease.