MON-085 - DNA Methylation Profiling Identifies Clinically Distinct Subgroups of Corticotroph and Lactotroph Pituitary Tumors

Pituitary tumors account for approximately 15-20% of all primary brain tumors and demonstrate a wide range of clinical behavior from benign to treatment-refractory and/or metastatic, and can be silent or functional. Emerging data suggest that epigenetic regulation contributes to this clinical variability, which is most pronounced in corticotroph and lactotroph lineages. We performed methylation profiling on the Illumina InfiniumMethylation EPIC 850k BeadChip platform and next-generation sequencing (NGS) using the MSK-IMPACT platform on a discovery cohort of 47 patients enrolled on a research protocol. The cohort comprised of 17 lactotroph and 30 corticotroph tumors; 18 benign and 29 aggressive tumors. Unsupervised hierarchical clustering separated corticotroph tumors into 3 subgroups (cluster 1, 2, and 3) and lactotroph tumors into 2 subgroups (cluster 4 and 5). Within the overall cohort, methylation clustering discriminated aggressive from benign disease (p < 0.001). Within the corticotroph lineage, clusters 1 (n=8) and 2 (n=10) demonstrated hypermethylation relative to cluster 3 (n=11) and were enriched with aggressive tumors (p=0.001), tumors with recurrent chromosomal loss-of-heterozygosity (LOH) on NGS using the FACETS algorithm (p=0.034), and silent clinical behavior (p < 0.001). Cluster 3 was primarily comprised of benign, functional tumors that achieved surgical remission (p=0.019). Within the lactotroph lineage, cluster 5 (n=4) was comprised of only aggressive tumors and included the two SF3B1-mutant tumors in our cohort, whereas cluster 4 (n=12) had a more heterogenous pattern. Tumors from a validation cohort (n=39), which included tumors from a previously reported external cohort, were assigned a methylation cluster by k=5 nearest-neighbor majority vote in batch-corrected high dimensional methylation space. The validation cohort mapped to the methylation clusters by lineage, clinical and genomic features, including an enrichment of silent tumors in clusters 1 and 2 compared to cluster 3 (p < 0.001), and universal mapping of USP8-mutant tumors to cluster 3 (p=0.005). Previously identified transcription clusters from the external validation cohort established differences in expression between clusters 1 and 2 compared to cluster 3 (p=0.016), potentially due to greater epigenic silencing. When analyzing the combined discovery and validation cohorts, aggressiveness differed significantly among the five clusters (p < 0.001), with clusters 2 and 5 exclusively including aggressive tumors (100%), compared to 47% in cluster 1, 17% in cluster 3, and 38% in cluster 4. Among lactotrophs, cluster 5 had tumors with higher Ki67, compared to cluster 4 (p=0.001). In this study, we show that DNA methylation segregates pituitary tumors primarily by lineage but also distinct subgroups with clinically relevant phenotypic differences.

*Unless otherwise noted, all abstracts presented at ENDO must not be released to the press or the public until the date and time of presentation. For oral presentations, the abstracts are embargoed until the session begins. The Endocrine Society reserves the right to lift the embargo on specific abstracts that are selected for promotion prior to or during ENDO.*