ARID1A, BRG1, and INI1 deficiency in undifferentiated and dedifferentiated endometrial carcinoma: a clinicopathologic, immunohistochemical, and next-generation sequencing analysis of a case series from a single institution

Abstract

Undifferentiated/dedifferentiated endometrial carcinomas (UDEC and DDEC) are rare, aggressive uterine neoplasms, with no specific line of differentiation. A significant proportion of these cases feature mutations of SWI/SNF chromatin remodeling complex members, including ARID1A, SMARCA4, and SMARCB1 genes. To study these entities more comprehensively, we identified 10 UDECs and 10 DDECs from our pathology archives, obtained clinicopathologic findings and follow-up data, and performed immunohistochemical studies for ARID1A, BRG1 (SMARCA4), and INI1 (SMARCB1) proteins. In addition, we successfully conducted targeted next-generation sequencing for 23 samples, including 7 UDECs, and 7 undifferentiated and 9 well/moderately-differentiated components of DDECs. Cases consisted of 18 hysterectomies and 2 curettage/biopsy specimens. Patient age ranged from 47 to 77 years (median, 59 years), with a median tumor size of 8.0cm (range, 2.5-13.0cm). All cases demonstrated lymphovascular invasion and the majority (13/20) were FIGO stage III-IV. By immunohistochemistry, ARID1A loss was observed in 15 cases, BRG1 loss in 4, and all cases had intact INI1 expression. A trend for enrichment of the undifferentiated component of DDECs for ARID1A loss was seen, although not statistically significant. Sequencing revealed frequent pathogenic mutations in PTEN, PIK3CA, ARID1A, CTNNB1, and RNF43, a recurrent MAX pathogenic mutation, and MYC and 12p copy number gains. In DDECs, the undifferentiated component featured a higher tumor mutational burden compared to the well/moderately-differentiated component; however, the mutational landscape largely overlapped. Overall, our study provides deep insights into the mutational landscape of UDEC/DDEC, SWI/SNF chromatin remodeling complex member status, and their potential relationships with tumor features.