Background: Bilateral macronodular adrenocortical disease (BMAD) is characterised by multiple bilateral macronodules causing hypercorticism. The clinical, biological and radiological presentation may vary from patient to patient. Despite this heterogeneity, the few microscopic descriptions have been consistent since the initial description. BMAD can be divided into three genetic groups: patients with alterations in ARMC5, KDM1A and those with no known genetic cause.  ARMC5 and KDM1A function as tumour suppressor genes. A germline event and a second somatic event lead to bi-allelic inactivation of the gene. For ARMC5, the somatic event may be heterogeneous from one nodule to another. The rare studies on KDM1A have only shown loss of heterozygosity by 1p deletion. There are few protein studies on BMAD, concentrating on immunohistochemical investigations, in particular on steroidogenesis enzymes.
Objective: to characterise BMAD heterogeneity on 3 levels: an exploration of microscopic heterogeneity with a search for correlations with patient genetics, an exploration by NGS of somatic genetic heterogeneity in BMAD and, an exploration by mass spectrometry of protein heterogeneity and its correlations with genetics and microscopy.
Results: The study involved 35 BMAD operated at Cochin hospital. Multiple factor analysis of microscopic characteristics distinguished 4 BMAD subtypes. Subtypes 1 and 2 have rounded fibrous septa in the macronodules. Subtype 1 contains 10-30% compact cells. This subtype correlates with an alteration of ARMC5. Subtype 2 contains 30-40% compact cells. This subtype was only visible when KDM1A was altered. Subtypes 3 and 4 do not contain fibrous septa. Subtype 3 is made up almost exclusively of clear cells. Subtype 4 contains 40-80% oncocytic cells. These subtypes include a majority of patients with no known genetic alteration. In cases of BMAD with ARMC5 or KDM1A alteration, NGS analysis showed that biallelic inactivation of the gene is necessary for nodule formation. It confirmed that pathogenic somatic events in ARMC5 are highly heterogeneous, whereas inactivation of KDM1A only occurs through loss of heterozygosity. Unsupervised clustering proteomics showed three groups of BMAD. The first group includes patients with ARMC5 inactivation who overexpress subunits of RNA polymerase II (POL II) and its partners. The second group includes patients with inactivation of KDM1A and those of subtype 4 who share a common signature and overexpress effectors of the mevalonate pathway involved in cholesterol biosynthesis. The final group includes subtypes 1 and 3, which overexpress certain superoxide anion generating proteins.
Conclusion: All the investigations enabled us to better characterise the heterogeneity of BMAD. The data revealed 4 microscopic groups and 3 proteomic groups which partially overlap with the genetic groups of BMAD already known. This work also clarifies the pathophysiology of the BMAD groups and highlights the importance of POL II and cholesterol synthesis enzymes. Between nodules, somatic ARMC5 events are heterogeneous, whereas KDM1A events are homogeneous. Other techniques such as spatial transcriptomics and the use of microdissection could help to characterise this intra-tumoral heterogeneity even further. The molecular mechanisms underlying BMAD without alterations in ARMC5 or KDM1A remain to be discovered.