Dissecting TSC2-mutated renal and hepatic angiomyolipomas in an individual with ARID1B-associated intellectual disability

ARID1B haploinsufficiency (OMIM #135900, *614556) represents a frequent cause for neurodevelopmental disorders (NDD) [1] comprising nonspecific intellectual disability [2] or Coffin-Siris syndrome [3, 4]. The AT-rich interactive domain-containing protein 1B encoded by the ARID1B gene is a subunit of the SWI/SNF (SWItch/Sucrose Non-Fermentable) chromatin remodeling complex. This complex and many of its components play an important role in a broad spectrum of neoplasms, acting mainly as tumor suppressors [5]. ARID1B therefore belongs to a growing number of genes in which germline mutations cause NDDs while somatic mutations are involved in cancer [6]. Though there is no clinical evidence that individuals with NDDs due to germline mutations in most of these genes and particularly in ARID1B bear an increased risk for tumors [6, 7], this has been repeatedly proposed and discussed [3, 8].

We report on a young woman previously diagnosed with intellectual disability due to a de novo mutation in ARID1B, in whom a huge renal and multiple hepatic angiomyolipomas (AMLs) were diagnosed. Investigating their molecular cause revealed a post-zygotic stop-variant in TSC2 in addition to the previously known germline variant in ARID1B.

Analysing exome data from blood and renal and hepatic AMLs showed the known variant in ARID1B, as expected, in about 50% of the reads, respectively (Fig. 2a and Additional file 1: Figure S1). No further pathogenic variant in ARID1B, nor loss of heterozygosity (LOH) was observed in tumor tissues (Additional file 1: Figures S2 and S3). Analysis of TSC1 and TSC2 revealed the pathogenic variant c.3099C > G, p.(Tyr1033*) [17] in TSC2 (NM_000548.4) in 67.3% (270/401) of the reads in DNA from the renal and in 41.2% (49/119) of reads from the hepatic AML, while it was not detected in any of the 179 reads from blood (Fig. 2a and Additional file 1: Figure S1). Panel sequencing on DNA from unaffected renal and liver tissues and from fibroblasts did not detect the TSC2 variant in any of the 31 to 508 reads, respectively (Fig. 2a and Additional file 1: Figure S4), but in one of 1210 reads in DNA from tubular cells. However, given its location at the beginning of the read and as it occurred in only one read in an overlapping read pair, it most likely represents an artefact (Fig. 2a and Additional file 1: Figure S5).

Searching exome data for additional variants in renal or hepatic AML and absent in blood revealed a variant in SIPA1L3 (c.187_192dup, p.(Ala63_Thr64dup) (NM_015073.2); COSM5959440), which was additionally excluded by Sanger sequencing in DNA from fibroblasts and healthy renal and hepatic tissues. Dysfunction of SIPA1L3 has been discussed as a possible contributor to the phenotype of some malignancies as it led to abnormalities of epithelial cell morphogenesis, cytoskeletal structure and adhesion in a colorectal adenocarcinoma cell line and as somatic SIPA1L3 alterations were found in various, mainly epithelial cancer types [18, 19]. However, it has not yet been reported as a tumor driver ( [20] and Cosmic Cancer Gene Census (CGC)), and it was not found to be mutated in a study searching for additional somatic mutations in TSC-related lesions [21]. Therefore, we consider this variant most likely a passenger variant, though we cannot exclude a contributing effect in the tumor pathogenesis in our patient. Copy number analysis from exome data indicated a possible deletion of TSC2 in a subpopulation of tumor cells. However, analysis of heterozygous variant positions at this locus did not show a significant deviation from 0.5, therefore not supporting validity of this finding (Additional file 1: Figures S2 and S3 and Additional file 2). These analyses are challenged by variability in exome data (e.g. coverage uniformity) and compromised tumor purity. No other large regions of LOH or CNVs were detected in the kidney or liver AMLs, indicating a relatively stable constellation in both tumors (Additional file 1: Figures S2 and S3; Additional file 2).

Tuberous Sclerosis Complex (TSC; OMIM #191100, *605284, #613254, *191092) is a multisystemic disorder generally caused by germline variants in TSC1 or TSC2. Multiple renal and extra-renal angiomyolipomas (AMLs) represent one of the characteristic manifestations of TSC. Renal AMLs are commonly observed in 80% of patients, while hepatic AMLs are less frequent and occur in ca. 13% of affected individuals [22]. When renal and hepatic AMLs were detected in the herewith reported individual with ARID1B associated NDD and without other frequent and typical signs of TSC, we considered various possibilities: a) The AMLs might be causally related to the germline ARID1B mutation. b) The AMLs represent a mild and atypical manifestation of TSC in the sentence of multiple diagnoses as reported recently in a large study that revealed aberrations in two or more disease loci in 4.9% of 7374 individuals with various phenotypes tested by clinical exome sequencing [23]. Or c) Another cause, un-related to the known ARID1B variant and not necessarily deducible from the AML phenotype might be responsible for the unusual presentation.

When performing exome sequencing on DNA from the renal and hepatic AMLs, we found no indication for a contribution of ARID1B or other commonly cancer-related genes such as TP53 to their genesis. While exome sequencing on blood did not detect a variant in the TSC genes TSC1 or TSC2, we detected the same pathogenic TSC2 variant in both renal and hepatic tumor tissues. This and an additional variant in SIPA1L3 also exclusively found in both AMLs underline their common clonal origin and open two possibilities: a) the multiple hepatic AMLs being metastases from the huge renal AML as rarely reported particularly for epithelioid AMLs [24], or b) the AMLs resulting from a low-grade mosaicism for the TSC2 variant. The first possibility seems extremely unlikely based on the discordant histopathological features of the AMLs. The huge renal tumor was predominantly composed of epithelioid cells entrapping only a very minor fatty component. In contrast, the resected hepatic AML was almost devoid of such epithelioid cells and was composed almost exclusively of fat cells closely mimicking a lipoma. Further supporting the second possibility, mosaicism has been frequently observed for TSC2 variants, including low-grade mosaicism with allele frequencies < 1% or variants only detected in skin tumor biopsies but not in blood or saliva [25].

Of note, we did not reliably observe the TSC2 variant in any of the sequencing reads from unaffected kidney, unaffected liver, fibroblasts or renal tubular cells, which might be expected in case of mosaicism. Given that the lesional cell of these TSC-related mixed neoplasms, the so-called “perivascular epithelioid cell” (PEC) has no known physiological counterpart, renal and hepatic AMLs might have their origin in a common, yet unknown mesenchymal precursor cell that expanded clonally during tumor development [26]. This has previously been discussed for the co-occurrence of sporadic pulmonary lymphangioleiomatosis and renal AMLs with somatic TSC2 variants [27] and for TSC-associated AMLs [28]. Presence of the TSC2 variant in only very few tumor precursor cells might therefore hinder detection in healthy tissues even by next-generation-sequencing (overview in Fig. 2b).

In contrast to many other tumors, AMLs are genetically relatively stable, and bi-allelic loss of TSC2 has been postulated as a sufficient and the main driver for AML development [28]. Given the absence of a second TSC2 variant, TSC2-LOH or any other evident tumor driver apart from the unclear variant in SIPA1L3 in the renal and hepatic AMLs of the herewith reported individual, a contribution of the germline ARID1B variant to the tumorigenesis cannot be excluded.

In any case, identifying the TSC2 variant in the herewith reported individual has important consequences for her medical care. It now prompts TSC specific surveillance including regular echocardiograms, high-resolution computed tomography of the lung and brain MRIs in addition to regularly checks of the remaining kidney and the liver. As mTOR inhibitors have been proven to be an efficient therapy in TSC [29], this treatment option will be evaluated and discussed based on her future disease course and manifestations.

Next to the value of combining genetic and pathologic findings, this unique case demonstrates and emphasizes several generally important aspects of genetic and medical care such as a) considering multiple (independent) diagnoses in a single individual, b) considering mosaicism or post-zygotic variants and taking the efforts to detect them, as c) a proper diagnosis might have important consequences on treatment and surveillance. These efforts can be challenged by potential mosaicism in multi-tumor syndromes where the mutation-carrying cell of origin is still unknown and might represent a very minor yet unidentifiable cell population leading to false-negative results.