Whole genome sequencing and single-cell transcriptomics identify KMT2D inactivation as a potential new driver for pituitary tumors: a case report

Abstract

The pituitary gland is a main component of the endocrine system and a master controller of hormone production and secretion. Unlike neoplastic formation in other organs, Pituitary Neuroendocrine Tumors (PitNETs) are frequent in the population (16%) and, for unknown reasons, almost never metastatic. So far, few genes have been identified as drivers for PitNETs, such as GNAS in somatotroph tumors and USP8 in corticotroph tumors. Using whole genome sequencing, we uncover a potential novel driver, the histone methyltransferase KMT2D, in a patient in his 50s suffering from a mixed somato-lactotroph tumor. Coverage ratio between germline and tumor revealed extensive chromosomal alterations. Single-cell RNA sequencing of the tumor shows up-regulation of known tumorigenic pathways compared to a healthy reference, as well as a different immune infiltration profile compared to other PitNETs, more closely resembling the profile of carcinomas than adenomas. Genome-wide DNA methylation analysis identified 796 differentially methylated regions, including notable hypomethylation in the promoter of SPON2, an immune-related gene. Our results show that tumors considered quiet and non-aggressive can share drivers, features, and epigenetic alterations with metastatic forms of cancer, raising questions about the biological mechanisms controlling their homeostasis.

Fig. 1. Immunostaining markers for somatolactrotroph PitNETs.

a Hematoxylin and eosin staining. b POU1F1 staining. c PRL staining. d GH staining.

Fig. 2. Evidence of a somatic variant in KMT2D.

a IGV output of Patient 16’s tumor WGS coverage showing the somatic mutation. b IGV output of Patient 16’s germline WGS coverage. c log2 Matched germline and tumor coverage ratio.

Fig. 3. Single cell clustering analysis.

a Initial clustering with 13 clusters. b Annotated UMAP based on canonical markers and automated annotation pipeline. c Copykat heatmap representing chromosomal alterations. Blue for deletion and orange for amplification. Y axis bar represent predicted aneuploid and diploid cells. d Quantification of KMT2D-expressing cells across PitNET samples with paired scRNA-seq and WGS data, compared to healthy control tissue.

Fig. 4. In depth analyses of patient 16's tumor compared to other PitNET samples or healthy references.

a Differential gene expression analysis results showing HALLMARKS up- and downregulated between patient 16 and reference. b Tumor and immune cells proportion in PitNET samples where scRNA-seq is available. c Immune cells composition in the PitNETs predicted by Azimuth.

Fig. 5. Boxplot representing beta values of the top 1000 most variable probes in 16 PitNETs where methylation data and WGS were available.

POU1F1 lineage includes somato- and lactotroph tumors. TBX19 lineage produces corticotroph tumors and NR5A1 gonadotroph tumors (Non-functioning).