Genomics in Cushing's Disease: The Dawn of a New Era.

Symptomatic pituitary tumors represent 5% to 10% of all brain tumors and are mostly benign. Up to 30% of macroadenomas grow invasively (mostly defined clinically by imaging), 15% may behave clinically “aggressive” in terms of treatment resistance, and 0.2% become malignant with metastases. In 2017, the World Health Organization (WHO) Classification of Endocrine Tumors published a new classification of pituitary tumors (1). The focus of the new proposal was 2-fold. Firstly, it establishes an unambiguous classification of tumors of anterior pituitary origin. This was achieved through the morpho-functional classification into 7 tumor subtypes arising from 3 pituitary lineages. Lineage tracing is achieved by immunohistochemical staining for pituitary transcription factors. Thus, pituitary tumors are classified into lactotroph, somatotroph, and thyrotroph (PIT1 lineage); corticotroph (TPIT lineage) or gonadotroph (SF1 lineage); null cell (immunonegative for PIT1, TPIT and SF1); and plurihormonal tumors. Immunohistochemical (IHC) staining for pituitary transcription factors should now be part of everyday routine histopathology assessment. Secondly, it should describe pituitary tumors with aggressive biologic behavior. This includes increased risk of tumor recurrence after resection, resistance to medical therapy and radiation, and invasive growth. The old WHO 2004 classification category of “atypical” adenoma was abandoned, since it was too unspecific. Instead, the current classification lists risk factors for aggressive behavior, such as p53 expression, a Ki67 index >3%, or any of the following subtypes: sparsely granulated somatotroph adenoma, silent corticotroph adenoma, Crook cell adenoma, null-cell adenoma, prolactinoma in males, and silent plurihormonal PIT1 adenomas.

Whereas the morpho-functional classification is mostly undisputed, the WHO classification stirred up an ongoing discussion about the usefulness of the proposed risk factors. Can histology alone identify those tumors at high risk of recurrence? Which is the impact of hormonal characteristics and imaging modalities? Histopathology expertise is nowadays an essential part of the multidisciplinary pituitary tumor board, but it is only 1 of the many puzzle pieces that should be considered in the risk assessment and treatment of pituitary tumor patients. Along the same lines, composite classification systems such as the French 5-tiered classification system take clinical, imaging, and histopathology factors into account, including tumor size and invasiveness, as determined by magnetic resonance imaging (MRI), immunohistochemical subtype, and proliferative markers, including Ki-67, mitotic count, and p53 positivity (2). It has been validated in several independent cohorts and allows the identification of tumors at higher risk of aggressive behavior, which can translate into closer surveillance strategies.

The tumorigenesis of Cushing’s disease has been obscure until 2015, when our group and others identified recurrent somatic heterozygous activating driver mutations in the ubiquitin-specific protease USP8 gene (3, 4). These mutations characterize a subgroup of patients who are mostly female, of younger age, and have microadenomas. The prevalence of mutations ranges between 20% and 60%. In subsequent whole exome sequencing studies, additional mutations in the glucocorticoid receptor NR3C1, the BRAF oncogene, the deubiquitinase USP48, and TP53 were identified, although at much lower rates (5, 6). Moreover, Neou et al have recently demonstrated that USP8 mutant and USP8 wild-type corticotroph tumors cluster into 2 distinct groups with quite different transcriptomic profiles (7).

The present manuscript by Uzilov et al reports a comprehensive study in 22 corticotroph tumors. The primary hypothesis driving the study is the classical assumption that the tumor genome is responsible for aggressive biological behavior. The authors selected corticotroph tumors at high risk for aggressive growth: more than half of the tumors were macroadenomas (including 6 silent corticotroph tumors), 2 were Crooke’s cell tumors, and 1 was a corticotroph carcinoma. Obviously, this deviates from the random corticotroph tumor series, which typically consists of more than 90% microadenomas. Remarkably, the authors were able to perform a detailed, integrative analysis of multiple genomic biomarkers on formalin-fixed paraffin-embedded tissue sections. This includes whole exome sequencing in search of somatic molecular mutations, somatic copy number variations, tumor mutational burden, microsatellite instability, allelic fractions and clonality, tumor purity, and mutational signatures.

Based on these data, the authors suggest that corticotroph tumors can be categorized into 2 major subtypes. Subtype 1 is characterized by key USP8 driver mutations. These tumors demonstrate low levels of somatic copy number variations, have a low level of microsatellite instability, and are typically microadenomas with nonaggressive biologic behavior. Subtype 2, which includes corticotroph macroadenomas, silent corticotroph tumors, and Crook cell adenomas, is characterized by a high level of somatic copy number variations (sCNV) indicating chromosome instability and often by TP53 mutations. sCNV are associated with cancer prognosis and response to treatment in other tumors. The Uzilov study suggests a close association between chromosome instability and aggressive corticotroph tumor behavior. If confirmed by others, these genomic markers could help to identify high-risk corticotroph tumors that may need closer monitoring and more intense treatment. This small but comprehensive study is an important step forward towards the dawn of the genomic area of Cushing’s disease.