Homozygous CDKN2A/B deletions in low- and high-grade glioma: a meta-analysis of individual patient data and predictive values of p16 immunohistochemistry testing
Darius Noack, Johannes Wach, Alonso Barrantes‐Freer, Nils H. Nicolay, Erdem Güresir, Clemens Seidel
Abstract
CDKN2A/B deletions are prognostically relevant in low- and high-grade gliomas. Data on this is derived from heterogeneous series, an accurate estimation of survival risk from homozygous CDKN2A/B deletion is missing. Besides genetic testing, p16-immunohistochemistry (IHC) as a less cost intensive means for indirect detection of CDKN2A/B alterations is possible but not validated in larger datasets. The present meta-analysis aimed to (1) reconstruct individual patient data (IPD) and estimate overall survival (OS) stratified by CDKN2A/B status from all literature and to (2) determine accuracy of p16 testing for CDKNA2/B detection from published studies. For survival analysis according to CDKN2A/B status 460 records were screened, four articles with 714 participants were included. In IDH-wildtype (IDH-wt) gliomas, 57.07% harbored the deletion compared to 9.76% in IDH-mutant (IDH-mut) gliomas. Median OS of patients with IDH-wt gliomas and homozygous CDKN2A/B deletion was 13.0 months compared to 18.0 months with non-deleted CDKN2A/B (p = 0.014, Log-Rank). With homozygous deletion of CDKN2A/B the risk of death was increased by 1.5 (95%-CI 1.1-2.1). Median OS in patients with IDH-mut gliomas without CDKN2A/B deletion was 92.0 months compared to 40.0 months with CDKN2A/B deletion (p < 0.001, Log-Rank). CDKN2A/B deletions were associated with a significantly shorter OS (HR = 3.2; 95%-CI 2.2-5.5). For p16 IHC analysis, 10 eligible studies with 1087 examined samples were included. The cut-off for retention differed between the studies. In 588/662 p16 retained cases CDKN2A/B deletions was not detected, implying a negative predictive value (NPV) of p16 staining of 88.8%. Conversely, 279/425 p16 absent cases showed a CDKN2A/B deletion resulting in a positive predictive value (PPV) of 65.6%. Sensitivity of p16 staining for CDKN2A/B detection was 79.0%, specificity 80.1%. Highest diagnostic accuracy of p16 IHC was reached with a cut-off of > 5% and within IDH-mut glioma.