Uterine leiomyoma with RAD51B::NUDT3 fusion: a report of 2 cases

Among uterine mesenchymal tumors, which represent a heterogeneous group of tumors arising from the smooth muscle and connective tissue of the uterus, uterine leiomyomas, also known as fibroids, are the most common and most important benign tumors with an estimated lifetime incidence of up to 70% [1].

Although current medical diagnosis of uterine mesenchymal tumors is based mainly on imaging and histological procedures, molecular tools are rapidly expanding and gaining relevance as a complement to conventional strategies in all clinical fields [2‐5]. In this sense, understanding the molecular aberrations can be of diagnostic value, as they can help distinguish between different types of uterine mesenchymal tumors with similar clinical and pathological features. In some cases, this knowledge can also be of therapeutic value. To date, some of the reported targetable aberrations include, for example, ALK, NTRK, ROS1, and other tyrosine kinase receptor rearrangements. The three main uterine leiomyoma molecular subtypes include (i) tumors with MED12 point mutations, (ii) tumors with biallelic loss of FH, and (iii) tumors with HMGA2 overexpression, commonly associated with chromosomal rearrangements (in HMGA1/HMGA2 or COL4A5/COL4A6), mainly resulting in the overexpression of these genes or reduced expression of CUX1 or CUL1 due to 7q deletions [6‐11]. Interestingly, while the latter genetic alteration represents the second most common category of usual-type leiomyoma, it is even more commonly found in cellular leiomyoma, where almost all cases (over 90%) show HMGA2 overexpression [12‐14]. In some of these tumors, rearrangement of HMGA2 is present [15]. The most common fusion partner of HMGA2 is RAD51B [16, 17]. However, the limited data suggests that fusions of RAD51B with other genes may also be present, and they seem to be mutually exclusive with other aberrations [18]. In our study, we described two cases of uterine leiomyoma with RAD51B::NUDT3 fusion, which occur in one case of usual-type and one case of cellular leiomyoma. Our cases represent the first tumors in which RAD51B::NUDT3 fusion has been found, but fusion of RAD51B with other genes than HMGA2 and NUDT3 can occur.

The study included two cases of uterine leiomyoma with RAD51B::NUDT3 fusion. One of them (case 1) was a routine diagnostic case from the Department of Pathology, First Medical Faculty and General University Hospital in Prague, in which the fusion was detected during the diagnostic work up. The second case (case 2) was uterine leiomyoma which was included in our previous study focused on genomic and transcriptomic profiling of uterine leiomyoma and uterine leiomyosarcoma [18].

The etiopathogenesis of uterine leiomyoma has been studied from several aspects including their molecular features. It has been shown that three main uterine leiomyoma molecular subtypes exist, including tumors with MED12 mutation, molecular aberrations leading to HMGA2 overexpression, and biallelic loss of FH [8, 10, 11, 20‐23]. These mutually exclusive aberrations seem to be a driver event and can be detected in approximately 80–90% of uterine leiomyomas. However, the frequency of molecular aberration occurring in uterine leiomyoma differs between leiomyoma subtypes. Most usual uterine leiomyomas (40–75%) are characterized by MED12 mutation, followed by 10–25% with HMGA2 overexpression [6, 20]. Specifically, in cellular leiomyoma, the most common is HMGA2 alteration (35% of cases), followed by chromosome 1p deletion (up to 25% of cases) and MED12 mutation (5–16% of cases) [24, 25]. In our previous study on cellular leiomyoma, deletion of chromosome 1p was mutually exclusive with other driver alterations [13]. However, the specific gene affected by this deletion is currently unknown [20]. FH alterations are mostly restricted to FH-deficient leiomyoma and leiomyoma with bizarre nuclei [22, 26, 27]. The incidence of FH alterations in usual and cellular leiomyoma is very rare, in the range 0–2.5% and 0–4%, respectively [10, 14].

Approximately 10% of uterine leiomyomas, however, does not belong to above-described categories. From these tumors, 38% in one study showed overexpression of HMGA1 [28]. Another study focused on 111 tumors, which were classified as negative for driver alteration based on Sanger sequencing and immunohistochemistry [17]. Forty-three of these tumors (39%) showed features typical for HMGA2-altered tumors including PLAG1 overexpression and 16 of them (14%) chromosomal rearrangements of HMGA2 (despite not having overexpression of HMGA2 by IHC), HMGA1, or PLAG1. HMGA1 and PLAG1 aberrations are not mutually exclusive with other alteration and can co-occur with MED12 mutation. Based on this, they have been suggested to be a secondary event related to tumor progression. Nevertheless, aberrations of both genes can occur also as an isolated finding and can be a driver event in uterine leiomyoma [17]. Other rare molecular driver aberrations occurring in uterine leiomyoma included somatic mutations in genes encoding six members of SRCAP histone-loading complex leading to H2A.Z loading defect [28]. It has been proved that patients with germinal mutation in the SRCAP members YEATS4 and ZNHIT1 predispose to uterine leiomyoma [28].

One recent study suggested the leading role of HMGA2 aberrations in uterine leiomyoma tumorigenesis, which is overexpressed even in leiomyomas with MED12 mutation [29]. However, the data in our previous study showed in three MED12 mutated cases neither HMGA2 overexpression on IHC level nor increased HMGA2 mRNA [13]. Some cases with HMGA2 overexpression are associated with HMGA2 translocations or aberrant splicing, but in most studies, simultaneous analysis of IHC expression and molecular aberrations was not performed and the exact incidence of cases showing HMGA2 rearrangement is not clear [30, 31]. The most common fusion partner of HMGA2 is RAD51B [16, 17, 32]. Other mechanisms potentially involved in HMGA2 overexpression are hypomethylation and regulation by the microRNA Let-7 family [31, 33, 34]. While it has been suggested that alteration of HMGA2 is considered to be the initial step leading to significant upregulation of PLAG1, the role of RAD51B should not be overlooked, as it is also important in uterine leiomyoma development [20]. In one study including 8 cases with HMGA2 rearrangement, 4 showed fusions with RAD51B, two with PTGER3, and two were rearranged without candidate partner gene [17]. Furthermore, other 4 cases in this study showed HMGA1 fusions, two with RAD51B, one showing complex rearrangement involving TRAF3IP2 and PRDM1, and one with PBX1. In our previous study on cellular leiomyoma, 33% (5/15) of tumors with HMGA2 overexpression showed HMGA2 rearrangement [13]. The fusion partners include C9orf92, PBX1, and RAD51B. In two cases, no fusion partner genes were found. In both cases, the rearrangement was within non-coding areas of chromosome 5. Another study comparing uterine leiomyoma and leiomyosarcoma found that a small percentage (3 out of 56) of leiomyoma cases showed a RAD51B fusion (with HMGA2, NCOR2, and NUDT3), and one of these cases with RAD51B::NUDT3 fusion is reported here in detail [18]. Interestingly, disruptions in NUDT3 have been shown to enhance cell migration in tumorigenic processes [35].

The expression of RAD51B and NUDT3 was detected on similar levels, which supports the hypothesis of balanced and reciprocal translocation event leading to these fusions.

While both cases showed 5-fold upregulation of RAD51B expression when compared to the matched healthy myometrium, there was no change in NUDT3 expression in case 1, with case 2 showing only 2-fold increased expression compared to the matched myometrium. Our findings support the previously published data suggesting that these fusions lead to the loss of physiological functions of RAD51B and NUDT3, resulting in a tumorigenic process. Concerning other RNA expression findings, transcriptional differences among leiomyomas harboring different genetic drivers have been described. Significant upregulation of PLAG1 was described in HMGA2 subtype of leiomyoma [20]. Overexpression of this gene can be also associated with upregulation of insulin-like growth factor-2 (IGF2) [36]. Moreover, overexpression of HMGA1 and/or HMGA2 is in leiomyomas common finding [20]. In our study, we have found upregulation or downregulation of several genes. In concordance with literary data, upregulation of HMGA1 and PLAG1 mRNA was detected in both cases. The HMGA2 mRNA expression in tumor and non-tumor tissue was below the level of reliable evaluation of expression pattern, which is in line with IHC negative results. Immunohistochemically detected PLAG1 protein expression showed weak positivity in case 1 and negativity in case 2. The IGF2 mRNA upregulation (≥ 10-fold) was observed only in usual leiomyoma. Furthermore, highly upregulated mRNA of cell surface marker CD24 was detected in our cases which correlates with previous findings of enriched CD24hi cells in leiomyoma. Another upregulated gene CAPN6 was detected also in both cases. Its upregulation has been previously described in uterine leiomyoma and was shown to be involved in proliferation and apoptosis while being mediated through the Rac1/PAK1 signaling pathway [37]. SNORA48 and SNORD10 (coding for small nucleolar RNAs) have not yet been described in leiomyomas, however were upregulated in both our cases. Some snoRNAs exhibit differential expression patterns in a variety of human cancers [38]. In one study, SFRP1 was significantly upregulated in leiomyomas relative to normal adjacent myometrium while other Wnt inhibitors such as APC, DKK1, and DKK3 were significantly downregulated [39]. We observed downregulation of DKK3 in both samples. Expression of SFRP1 was downregulated in cellular leiomyoma and upregulated in usual leiomyoma. The expression of APC and DKK1 was low and not reliable for evaluation of expression pattern. Furthermore, STK26 (previously known as MST4) downregulation was observed in both samples. This finding correlates with MST4 downregulation in leiomyomas relative to normal myometrium reported previously [40].

The knowledge about molecular features of uterine leiomyoma can be of practical value in differential diagnosis especially in tumors with some unusual morphological features, such as cellular leiomyomas. In some of these tumors, especially so-called highly cellular leiomyomas, the distinction from tumors with endometrial stromal differentiation, including low grade endometrial stromal sarcoma (LG-ESS), may be problematic. Most of these tumors can be distinguished based on combination of morphological and immunohistochemical features. However, rare tumors can have overlapping features between cellular leiomyoma and LG-ESS and in these tumors, molecular testing may be helpful. However, the knowledge of molecular aberrations occurring in endometrial stromal tumors is rapidly evolving and the spectrum of aberration is broadening. These aberrations do not occur in uterine leiomyoma. However, with increasing knowledge about molecular aberrations occurring in mesenchymal uterine tumors, new aberrations were described, which can occur in both cellular leiomyoma and endometrial stromal tumors. For example, tumors with KAT6B::KANSL1 and KAT6A::KANSL1 fusion resembling LG-ESS some of them with sex cord-like features have been described recently [41]. These tumors have potential to aggressive behavior, even though most of them were characterized by well-defined borders. However, the fusions detected in these tumors have been described in 1 case of uterine leiomyoma and 1 case of uterine leiomyosarcoma [42, 43].

In conclusion, our study showed that RAD51::NUDT3 fusion can occur in both usual and cellular leiomyoma. RAD51B may be a fusion partner of HMGA2 and HMGA1 but can occur in fusion with other genes including NUDT3 and seems to be a potential driver event in these tumors mutually exclusive with other driver aberrations defining molecular leiomyoma subtypes. Nevertheless, more data is needed to confirm the possibility of RAD51B altered uterine leiomyoma as a distinct molecular subtype. From practical point of view, we should add the RAD51B::NUDT3 fusion into the spectrum of fusions which can occur in leiomyocellular tumors, but has never been described in tumors of other histogenesis including inflammatory myofibroblastic tumor, endometrial stromal tumors, and tumors with kinase fusions such as NTRK, RET, and ROS1. Based on this, this fusion seems to be specific for tumors with leiomyocellular differentiation. This can be important for differential diagnosis between cellular leiomyoma and LG-ESS, but also for the differential diagnosis of tumors of other histogenesis, which can be in some cases with equivocal features complicated on morphological and immunohistochemical level only.