DICER1-sarcoma: an emerging entity

Abstract

In this issue of Modern Pathology, Kommoss et al. report on the clinicopathological and molecular features of DICER1-mutant and DICER1-wild type embryonal rhabdomyosarcomas (ERMS) in a series of genitourinary tumours. The authors studied 17 ERMS, including 9 uterine (four corpus and five cervix), one vaginal and seven urinary tract tumours. They showed pathogenic variants (PVs) of DICER1 in 9/9 uterine ERMS while all other ERMS were DICER1-wild type. The median age at diagnosis of patients with DICER1-mutant and DICER1-wild type ERMS was 36 and 5 years, respectively. Limited follow-up data suggested that DICER1mutant tumours might exhibit a less aggressive clinical course than DICER1-wild type. Cartilaginous elements were only observed in DICER1-mutant ERMS (6/9 DICER1-mutant ERMS) and in one case this was accompanied by foci of ossification. Unsupervised hierarchical clustering of array-based whole-genome DNA methylation data of their series together with an extended methylation data set including different rhabdomyosarcoma (RMS) subtypes from genitourinary and extra-genitourinary locations (n= 102), revealed a distinct cluster for DICER1-mutant ERMS. These tumours clearly segregated from the clusters of DICER1-wild type ERMS, alveolar RMS and MYOD1-mutant spindle cell and sclerosing RMS. Only one tumour, an ERMS in the maxilla of a 6-year-old boy, clustered with DICER1-mutant ERMS of the uterus. Subsequent sequencing analysis of this case identified two DICER1 PVs. The authors concluded that DICER1-mutant ERMS might represent a distinct subtype in the future classification of RMS. This paper adds to a series of publications in the last decade that have shown a strong association between both germline and somatic PVs in DICER1 in uterine ERMS, especially those arising in the cervix. As noted by the current authors, these tumours exhibit several characteristic features which can aid the pathologist in identifying a likely association with DICER1 PVs. Moreover, we and others have noted that many of these features, such as a subepithelial layer of malignant mesenchymal cells (cambium layer), areas of rhabdomyoblastic differentiation with positive staining with myogenin and myoD1, cellular/immature and occasionally malignant cartilage, foci of bone/osteoid and areas of anaplasia are shared by other sarcomas at different sites that possess either germline or somatic (or both) PVs in DICER1. Given the confusing proliferation of names for tumours that have similar appearances and a single genetic cause, we have argued for a simplified nomenclature of DICER1-related sarcomas, such that their tissue site and common molecular origins are recognised. For example, we suggest terms such as “primary peritoneal sarcoma, DICER1-mutant” and “primary cervical sarcoma, DICER1-mutant”. DICER1 is part of the microRNA biogenesis pathway and perturbations in microRNAs underlie many developmental and neoplastic disorders. DICER1 mutations have an unusual distribution in that in neoplasms, two hits are usually identified, but unlike most other genes that fit the “two-hit” hypothesis of tumour formation, the hits are very different in nature; one hit is usually a protein-truncating loss-of-function variant, whereas the other specifically involves one of very few “hotspot” sites located in the critical RNase IIIb domain of DICER1 protein. This can facilitate simple genetic testing in the molecular pathology laboratory because if a hotspot is not found on targeted sequencing, it is unlikely that the tumour will be DICER1-related. This can be helpful for example in trying to distinguish uterine ERMS from adenosarcoma, as the absence of a DICER1 PV makes a diagnosis of ERMS very unlikely, although the presence of a PV does not help distinguish between the two tumours. In other organ systems, this specificity has been exploited by commercial genomic assays such as Thyro-Seqv3, which is helpful in management of patients with a thyroid nodule. The assay includes only the hotspot regions of DICER1, but because of the association of hotspot variants with second hits elsewhere in the gene, the assay can identify persons who may have germline DICER1 PVs and hence DICER1 syndrome. Thus, genetic counselling should be offered to all persons found to have a hotspot PV in DICER1, unless a germline PV has already been excluded. How else might this emerging class of tumours be more easily identified by pathologists? The paper by Kommoss et al. suggests another intriguing possibility. The distinct cluster for DICER1mutant ERMS mentioned above likely reflects the common tissue of origin or differentiation state of these tumours. While DNA methylation arrays may not be widely available in pathology laboratories, a minimalist approach has recently been suggested in this journal and could be applied in clinical diagnostic pathology laboratories. To generalise this approach, it will first have to be demonstrated that DICER1-sarcomas do indeed form a discrete, identifiable entity. Recently, a novel RMS-like methylation class referred to as “SARC” has been identified, which is populated by central nervous system (CNS) sarcomas with DICER1 PVs. DICER1 may also be implicated in the related “SARC (malignant peripheral nerve sheath tumour (MPNST)-like)” methylation cluster, as a DICER1 PV was reported in one of five MPNSTs. While DICER1-related CNS sarcomas will be unlikely to share a tissue of origin with sarcomas arising in the urogenital tract, there may be other DICER1-dictated events that lead to clustering. Intriguingly, there is a notable excess of females with DICER1sarcomas and a possible link to the Müllerian system has been posited. The strongest clue that this clustering may be generalisable, however, comes from the one outlier in the study of Kommoss et al., mentioned above; this was a maxillary ERMS in a 6-year-old boy which was identified on the basis of the common clustering with truncating and hotspot DICER1 mutations only discovered subsequently. Therefore, further methylation studies to determine whether DICER1-mutant sarcomas do indeed form a distinct, identifiable subgroup of sarcomas are warranted. A recent review of the epigenomics of sarcoma suggests that many sarcomas can be viewed as epigenetic diseases with few, and sometime only one, mutation being required for their initiation. DICER1 was not mentioned in this review, but in many ways DICER1-mutant sarcomas fit the bill for being predominantly