Two cases of mosaic germline SVA insertions in SMARCB1: implications for rhabdoid tumour predisposition diagnosis Authors: Laura Raiti, Kyoko E. Yuki, Lise Cinq-Mars, Edward J. Higginbotham, Vilma Navickiene, Yisu Li, Zhuozhi Wang, Bhooma Thiruv, Tristan Charlinski, Bailey Gallinger, Rosemarie E. Venier, Annie Huang, Avram Denburg, Kathy Chun, Anita Villani Published: 2026 ABSTRACT Rhabdoid tumour predisposition syndrome (RTPS) is a highly penetrant cancer predisposition syndrome caused by germline variants in SMARCB1 or less frequently in SMARCA4. Genetic testing for this syndrome involves sequence and deletion/duplication analysis of these two genes. Standard clinical testing is limited in detecting structural variants. Here we describe two patients who tested negative on standard clinical germline panel testing for RTPS but were each found to have a mosaic germline insertion of an SVA (SINE-VNTR-Alu) element in the SMARCB1 gene by more advanced comprehensive genomic analysis. These two cases demonstrate the importance of structural variants and broader genomic sequencing for individuals suspected of having an underlying germline cancer predisposition syndrome, such as RTPS. Introduction Genetic predisposition to childhood onset malignancies has been reported to occur in at least 10-15% of children with cancer1–5. As such, germline genetic testing for hereditary cancer predisposition syndromes (CPS) is an important aspect of assessment of children with malignancies, especially those associated with a highly penetrant CPS. Rhabdoid tumour predisposition syndrome (RTPS) is an autosomal dominant, highly penetrant CPS characterized by a spectrum of tumours that includes central nervous system (CNS) atypical teratoid rhabdoid tumour (ATRT), extracranial malignant rhabdoid tumour (MRT) of the kidney, and extrarenal MRT involving the liver, heart and soft tissues of the head and neck, thorax, retroperitoneum and pelvis (split 65%, 9% and 26%, respectively)6. RTPS is caused by germline loss-of-function pathogenic/likely pathogenic variants in SMARCB1 (RTPS1) or SMARCA4 (RTPS2)7,8. Single nucleotide variants (SNVs) or frameshift mutations leading to a premature stop codon are the most common germline aberrations; however, partial or complete gene copy number changes can also occur9. Approximately 35% of patients with rhabdoid tumours have an underlying pathogenic or likely pathogenic variant in SMARCB1, diagnostic of RTPS19,10. Clinical genetic testing for RTPS involves sequence and deletion/duplication analysis of these two genes. Germline genetic testing for RTPS and other CPS is typically performed using next-generation sequencing (NGS), which enables sequence analysis and exon-level copy number variant (CNV) detection at an average read depth of ~30x. Complementary methods include Sanger sequencing of exons and their intron/exon boundaries, in addition to multiplex ligation-dependent probe amplification (MLPA) for copy number analysis11. Standard clinical testing methods have limitations, particularly when structural variants or deep intronic variants are present. For example, a deep intronic hotspot variant located in intron 1 of SMARCB1 has been previously reported in two patients manifesting ATRTs, diagnostic of RTPS112. Thus, there is likely to be under-recognition of germline predisposition in patients with rhabdoid tumours, which creates missed opportunities for personalized clinical management, including institution of surveillance and cascade testing. It is important to consider the limitations of standard clinical testing and consider other mechanisms of SMARCB1 or SMARCA4 deficiency in patients where there is a high suspicion of RTPS, such as those with recurrent RTPS-associated tumours. Here we describe two cases in which mosaic germline structural variants in SMARCB1 were found through advanced comprehensive genomic analysis following negative standard clinical germline testing. Results Case One: an infant with a malignant rhabdoid tumour of the kidney and synchronous ATRT A 6-month-old term female presented to hospital with reduced oral intake and weight loss over the preceding 4 weeks. An abdominal mass was detected on examination with initial sonographic imaging revealing a left-sided renal mass. She underwent upfront left total nephrectomy and was subsequently diagnosed with abdominal stage III malignant rhabdoid tumour of the kidney with positive lymph nodes and evidence of pre-operative intra-abdominal rupture. Histology revealed loss of INI1/SMARCB1 nuclear staining in tumour cells. The patient developed seizure activity during the inpatient admission, and completion of her staging investigations revealed a suspected synchronous ATRT, with neuroimaging demonstrating a solitary well-circumscribed ovoid CNS lesion at the lateral border of the right caudate head with restricted diffusion. Given the small size of the lesion (6 mm) and deep location, it was not amenable to neurosurgical biopsy. Cancer family history was only significant for a maternal great-grandmother with breast cancer. Clinical genetic testing was arranged. DNA from peripheral blood was sent for a standard NGS panel for SMARCB1 and SMARCA4, which did not identify any pathogenic or likely pathogenic sequence or copy number variant. The patient was managed with three cycles of induction chemotherapy: alternating cycles of Ifosfamide, Carboplatin and Etoposide (ICE) and Vincristine, Doxorubicin and Cyclophosphamide (VDC) and intrathecal (IT) topotecan. Imaging assessment following two cycles of chemotherapy demonstrated no evidence of locoregional recurrence, and she was clinically well. Her CNS lesion was also evaluated with no significant interval change in size, no new lesions, and interval reduction in pachymeningeal enhancement. She went on to receive two of three planned cycles of high dose chemotherapy with carboplatin and thiotepa and autologous stem cell transplantation. Unfortunately, prior to her third planned cycle of high dose chemotherapy, she had widespread recurrence and disease progression of her extracranial MRT, evidenced by a new heterogeneous large soft tissue mass centered in the left renal bed, left retro/para-aortic lymphadenopathy, left retroperitoneal soft tissue mass, multiple peritoneal implants with large ascites and intrathoracic nodal metastases. No further disease-directed therapy was administered, and she died at 14 months of age. Advanced molecular analysis The patient was enrolled in an institutional precision oncology program (SickKids Cancer Sequencing Program, KiCS)13. The KiCS platform is a custom pan-cancer DNA panel that provides ~1000x coverage for 864 cancer-associated genes. Paired tumour:normal DNA sequencing detected somatic copy neutral loss of heterozygosity (cnLOH) of whole chromosome 22, which includes the SMARCB1 gene (Fig. 1a).Fig. 1Case 1 copy number analysis results.a-c Copy number output, generated by VIA (version 7). a Tumour DNA copy number analysis with the NGS panel sequencing showed somatic cnLOH of chromosome 22 (red box), where SMARCB1 resides. No germline SMARCB1 copy number variant was identified. Closer scrutiny of the copy number profile of the SMARCB1 gene (chr22:24,129,153-24,180,196) revealed several reads in exon 1 of SMARCB1 that do not map to SMARCB1 in either the b tumour or c germline sample. The number of reads where the mate maps to chromosome 20 was greater in the tumour (VAF ~ 30-50%) than in the germline (VAF ~ 5-6%) due to the cnLOH on chromosome 22 in the tumour. d IGV screenshot (generated from IGV version 2.11.2) of short read WGS of the SMARCB1 exon 1 region in the tumour showing read-depth evidence of the 17-bp target site duplication (TSD) that supports the NGS panel data. The gray bars indicate the mapped reads while the magenta-coloured reads map to chromosome 20 and the SVA_D element. Case 1 copy number analysis results. a-c Copy number output, generated by VIA (version 7). a Tumour DNA copy number analysis with the NGS panel sequencing showed somatic cnLOH of chromosome 22 (red box), where SMARCB1 resides. No germline SMARCB1 copy number variant was identified. Closer scrutiny of the copy number profile of the SMARCB1 gene (chr22:24,129,153-24,180,196) revealed several reads in exon 1 of SMARCB1 that do not map to SMARCB1 in either the b tumour or c germline sample. The number of reads where the mate maps to chromosome 20 was greater in the tumour (VAF ~ 30-50%) than in the germline (VAF ~ 5-6%) due to the cnLOH on chromosome 22 in the tumour. d IGV screenshot (generated from IGV version 2.11.2) of short read WGS of the SMARCB1 exon 1 region in the tumour showing read-depth evidence of the 17-bp target site duplication (TSD) that supports the NGS panel data. The gray bars indicate the mapped reads while the magenta-coloured reads map to chromosome 20 and the SVA_D element. A closer investigation of SMARCB1 during copy number analysis of the tumour revealed a high number of unmapped reads in exon 1 (Fig. 1b). The same pattern was evident in the germline sample at a much lower level (Fig. 1c), suggesting germline mosaicism. Manual inspection of the bam files revealed an insertion of uncertain origin at nucleotide position 76 in exon 1 of the SMARCB1 gene along with a duplication of nucleotides 60 to 76. Short read whole genome sequencing (WGS) of the tumour was suggestive of an insertion of an SVA_D transposable element with a polyA tail of unknown length, as well as exon 2 of the DEFB124 gene and intron 4 of the REM1 gene originating from chromosome 20, flanked by the 17-bp duplicated SMARCB1 exon 1 sequence. The NGS panel results suggested this insertion was mosaic in the germline at a variant allele frequency (VAF) of ~5-7%. Sanger sequencing with primers specific to exon 2 of DEFB124 and intron 1 of SMARCB1 confirmed the presence of this insertion in the patient’s germline and tumour samples. Long read WGS sequencing of the patient’s tumour confirmed the insertion to be a 2.618 kb SVA_D transposable element from chromosome 20 involving exon 2 of DEFB124 and intron 4 of REM1, with a 69-bp polyA tail (Fig. 2). This insertion is predicted to disrupt transcription of SMARCB1, resulting in loss of function.Fig. 2Genomic structure of the SVA_D insertion for Case 1.Long read sequencing confirmed the complex rearrangement found by short read sequencing, which included a 2.618 kb SVA_D transposable element from chromosome 20 involving the DEFB124 and REM1 genes, followed by a 69-bp polyA tail. The insertion at nucleotide 76 is flanked by a 17-bp duplication of SMARCB1 exon 1 sequence (nucleotides 60 to 76). TSD – target site duplication. Genomic structure of the SVA_D insertion for Case 1. Long read sequencing confirmed the complex rearrangement found by short read sequencing, which included a 2.618 kb SVA_D transposable element from chromosome 20 involving the DEFB124 and REM1 genes, followed by a 69-bp polyA tail. The insertion at nucleotide 76 is flanked by a 17-bp duplication of SMARCB1 exon 1 sequence (nucleotides 60 to 76). TSD – target site duplication. Case Two: a 15-year-old female with a history of ATRT and recurrent schwannomas This patient presented at 2 years and 9 months of age with a short history of vomiting, headache, neck pain and intermittent lethargy. Neuroimaging revealed a large heterogeneous solid and cystic fourth ventricular mass inseparable from the tectum with moderate obstructive hydrocephalus. There was no evidence of metastatic disease by neuroimaging or by CSF cytology. She underwent a gross total resection with histopathological confirmation of a diagnosis of ATRT. Immunohistochemistry (IHC) analysis of her tumour demonstrated that the cells were non-reactive for INI1/SMARCB1. She underwent three cycles of induction chemotherapy with Cisplatin, Cyclophosphamide, Vincristine, Etoposide, and triple IT chemotherapy. She was then consolidated with three cycles of high dose chemotherapy (Carboplatin and Thiotepa) with autologous stem cell transplantation as a radiation sparing approach. She was in a radiological complete response following induction and consolidation therapy. The patient went on to receive 12 cycles of maintenance therapy, including IT topotecan and oral tamoxifen/cis-retinoic acid, over 12 months and remains in remission of her CNS ATRT for greater than 10 years. At age 11, she developed a right foot schwannoma, which was resected, and IHC revealed a mosaic staining pattern for INI1/SMARCB1. She was given a clinical diagnosis of RTPS1 and followed with imaging surveillance using CNS and WBMRI and abdominal ultrasound. At age 13, she had a surveillance-detected right distal femoral osteochondroma, which was resected and did not show INI1 loss. Further surveillance detected two additional schwannomas in her retroperitoneum and her right tibial nerve at age 13, both of which were resected and showed a mosaic staining pattern for INI1 by IHC. At age 15, surveillance imaging detected a fourth schwannoma demonstrating a small mass abutting the right lateral plantar nerve. As this lesion is asymptomatic at present, she has not undergone a resection but continues to be monitored with imaging surveillance. Genetic testing at the time of her ATRT diagnosis, including MLPA and Sanger sequencing of SMARCB1, did not identify a pathogenic or likely pathogenic sequence variant or copy number alteration. Following her first schwannoma, repeat testing by MLPA, single nucleotide polymorphism (SNP) microarray and Sanger sequencing of SMARCB1 detected no pathogenic sequence or copy number variant. Germline NGS was subsequently ordered for NF2, LZTR1 and SMARCA4, with MLPA for CNV analysis of LZTR1 and NF2, which did not yield any significant findings. Trio whole exome sequencing (WES) was later conducted, with re-analysis after one year, which did not identify a pathogenic sequence variant to explain her phenotype. Family history pertaining to cancer is significant for multiple adult-onset cancers, but none within the expected RTPS spectrum. There were no young onset malignancies in the family identified (Fig. 3).Fig. 3Family pedigree for Case 2.The proband shows an extensive family history of cancer on the maternal side. Age of diagnosis of the cancer is indicated in brackets. Family pedigree for Case 2. The proband shows an extensive family history of cancer on the maternal side. Age of diagnosis of the cancer is indicated in brackets. Advanced molecular analysis The patient was enrolled in the KiCS study, and paired tumour:normal DNA panel sequencing of the patient’s right foot schwannoma revealed only loss of whole chromosome 22, which includes SMARCB1, while sequencing of the retroperitoneal schwannoma detected no aberrations involving SMARCB1. No copy number change or SNV affecting SMARCB1 was detected in the germline. Short read WGS of both tumours and the germline provided no further insights at that time. RNASeq of the patient’s two tumours detected what initially appeared to be a translocation between OR6J1 on chromosome 14 and SMARCB1; germline RNASeq detected the same rearrangement but at a lower level. Manual inspection of the short read WGS bam files of the two schwannomas and the germline was suggestive of a mosaic insertion of an SVA_D transposon from intron 1 of OR6J1 into intron 7 of SMARCB1. In the germline, three of the 19 informative reads showed the insertion, indicating low level mosaicism. Long read WGS revealed an insertion of a 2.978 kb SVA_D transposable element originating from chromosome 14 into intron 7 of the SMARCB1 gene (Fig. 4a). The insertion occurred at genomic coordinate chr22:23828661 and was flanked by a 15-bp duplication of intron 7 sequence (chr22:23,828,650-23,828,665).Fig. 4Characterization of the SVA_D insertion at the DNA and RNA level for Case 2.a Insertion of a 2.974 kb SVA_D element from chromosome 14 occurred within intron 7 of SMARCB1 at genomic position chr22:23828661. The insertion is flanked by a 15-bp duplication of sequence from intron 7 (chr22:23,828,650-23,828,665). Sequencing of the reverse transcribed SMARCB1 cDNA showed a 515-bp segment of the SVA_D element creating a cryptic exon between exons 7 and 8. b Examination of the sequence surrounding the 515-bp cDNA insertion revealed canonical splice sites flanking the cryptic exon, as well as a putative branch site upstream of the splice acceptor site. TSD – target site duplication. Characterization of the SVA_D insertion at the DNA and RNA level for Case 2. a Insertion of a 2.974 kb SVA_D element from chromosome 14 occurred within intron 7 of SMARCB1 at genomic position chr22:23828661. The insertion is flanked by a 15-bp duplication of sequence from intron 7 (chr22:23,828,650-23,828,665). Sequencing of the reverse transcribed SMARCB1 cDNA showed a 515-bp segment of the SVA_D element creating a cryptic exon between exons 7 and 8. b Examination of the sequence surrounding the 515-bp cDNA insertion revealed canonical splice sites flanking the cryptic exon, as well as a putative branch site upstream of the splice acceptor site. TSD – target site duplication. Sequencing of the reverse transcribed cDNA showed that a 515-bp segment of the SVA_D had been transcribed between exons 7 and 8 of SMARCB1 (Fig. 4b). Closer examination of the SVA_D genomic sequence surrounding the transcribed cDNA segment revealed canonical splice sites immediately flanking the insertion. Further, a putative branch site (“GCCTCAC”) and string of pyrimidines were found just upstream of the 515-bp segment, suggesting that a cryptic exon had been created between exons 7 and 8 as a result of this insertion. Figure 5 illustrates the junction sites of the cDNA and protein sequence of the (a) normal and (b) SVA_D insertion SMARCB1 transcripts. The serine 329 residue of the normal sequence is replaced by arginine in the putative protein product of the cryptic SVA_D exon insertion, followed by a novel protein sequence with multiple stop codons. The first stop signal occurs fourteen codons after the arginine 329 residue where the protein product is expected to truncate.Fig. 5cDNA and protein sequence of the normal and SVA_D insertion SMARCB1 transcripts of Case 2.a The junction of the normal exon 7 and exon 8 cDNA sequence is indicated by a short vertical line (|). The serine (S) at residue 329 is encoded partially by the end of exon 7 and the beginning of exon 8. b With the insertion of the cryptic SVA_D exon, the S329 residue is changed to an arginine (R), followed by unknown protein sequence. The first stop codon (*) is encountered fourteen codons after the R329 residue, where the protein product is expected to truncate. cDNA and protein sequence of the normal and SVA_D insertion SMARCB1 transcripts of Case 2. a The junction of the normal exon 7 and exon 8 cDNA sequence is indicated by a short vertical line (|). The serine (S) at residue 329 is encoded partially by the end of exon 7 and the beginning of exon 8. b With the insertion of the cryptic SVA_D exon, the S329 residue is changed to an arginine (R), followed by unknown protein sequence. The first stop codon (*) is encountered fourteen codons after the R329 residue, where the protein product is expected to truncate. Discussion These case reports illustrate the importance of advanced molecular testing to identify a diagnosis of an underlying germline CPS and highlight the value of considering mosaicism and complex structural rearrangements in the spectrum of molecular mechanisms that can lead to a diagnosis of a CPS. We present two cases of germline mosaic insertion of an SVA element into the SMARCB1 gene. SVAs are the most recently evolved class of retrotransposons, a type of transposable element, that have been associated with disease14. The mechanisms by which SVAs cause disease include changes to normal splicing patterns, large genomic deletions and exonic insertions which subsequently lead to loss-of-function mutations14. SVAs are divided into subtypes or subfamilies (A-F and F1) according to the SINE (short interspersed nuclear element) region, with the expansion of SVA_A to SVA_D occurring before the divergence of humans, chimpanzees and gorilla, whereas SVA_E, SVA_F and SVA_F1 are restricted to humans15,16. To our knowledge, this is the first time two SVA_D insertions have been reported in the mosaic state leading to RTPS1 and RTPS-associated tumours. While both cases have an insertion of an SVA_D element, the insertion for Case 1 occurs in the first exon of SMARCB1 (Fig. 2), disrupting the reading frame and presumably leading to nonsense mediated decay, as it occurs so early in the transcript. On the other hand, the insertion for Case 2 occurs within intron 7 of SMARCB1 and appears to create a cryptic exon that is inserted between exons 7 and 8 of SMARCB1 (Fig. 4). RNA sequencing studies suggest that the SMARCB1 insertion allele may escape nonsense-mediated decay and remain transcribed to a certain level. The aberrant protein product produced by the SVA_D insertion allele is expected to result in a truncated protein after exon 7, followed by a stretch of 13 unique amino acids. Insertion of a transposable element into SMARCB1 is rare but has been reported in germline samples of patients with rhabdoid tumours17,18. In two siblings diagnosed neonatally with ATRT, Sabatella and colleagues utilized optical genome mapping (OGM) and long read sequencing to identify a germline SVA_E insertion within intron 2 of SMARCB1 that caused aberrant splicing, leading to the loss of a functional allele17. The second hit in the tumour was cnLOH of a large region on chromosome 22 that included the SMARCB1 gene. This insertion was found to be mosaic in the mother. In another report, Thomas et al. used targeted NGS, further verified by Sanger sequencing, to detect an insertion of a AluY transposable element into exon 2 of SMARCB1 in the tumour of a child with ATRT18. The AluY insertion led to a frameshift truncating mutation, occurring in the context of a large region of cnLOH on chromosome 22 involving SMARCB118. The insertion was not observed in the germline; hence, both hits to SMARCB1 occurred only in the tumour. SVAs have also been associated with other CPS, including early onset breast cancer in the setting of hereditary breast and ovarian cancer syndrome with SVA insertions in BRCA1 and BRCA219–21. Walsh et al. reported a family with early onset breast cancer who was found to have an SVA intronic insertion in BRCA1, leading to the formation of a pseudoexon or exonisation of the SVA, and the introduction of a premature stop codon19. Deuitch and colleagues reported an SVA insertion in BRCA2, ultimately disrupting its DNA-binding domain20. With the use of long-read nanopore targeted sequencing, Ohori et al. identified an intronic SVA insertion in BRCA1 of a proband with bilateral triple-negative breast cancer and their mother who also had triple-negative breast cancer21. SVAs have also been implicated in Lynch syndrome, with exonic SVA insertions in MSH2 leading to a premature stop codon or aberrant splicing22,23, and an SVA insertion into exon 5 of MSH6 leading to aberrant splicing23. Finally, similar to the mechanism described in Case 2 and the case involving BRCA119 above, an intronic SVA insertion in PMS2 has been reported, which activates a cryptic splice acceptor site leading to exonisation of part of the SVA and a frameshift with a premature stop codon24. Germline mosaicism of SMARCB1 mutations have been reported in ~6-21% of patients with rhabdoid tumours25–28. Several groups have published case reports of siblings with malignancy and a diagnosis of RTPS in the setting of no pathogenic variants or deletions identified in the parents, but recurrence in subsequent pregnancies due to gonadal mosaicism9,10,25,29,30. In a small cohort of 16 patients with malignant rhabdoid tumour, eight had germline abnormalities, of which three were identified to be mosaic26. Larger cohorts reported by Thomson et al. and Fleischmann et al. detailed cases of germline mosaic variants in 6-21% of their cohorts, with VAFs between 0.9% and 33%, and 2.7% to 9.5%, respectively27,28. Importantly, those with mosaic germline variants were noted not to have an association with earlier onset disease or worse survival outcomes but described to be at risk of developing subsequent RTPS-related tumours later in life27. While germline mosaicism of SMARCB1 mutations in RTPS is relatively frequent, it was important to rule out other possibilities that could account for our germline findings. Tumour contamination of the germline seems highly unlikely, as the tumour and germline samples were all collected and extracted at different time points for both Case 1 and Case 2. The initial negative standard NGS panel for SMARCB1 and SMARCA4 for Case 1 was performed in another laboratory on a different peripheral blood sample. Subsequent to our mosaic germline finding, that laboratory also found evidence of the insertion at ~5% upon manual inspection. For Case 2, the paired tumour:normal KiCS panel analysis and short read WGS were performed on the same peripheral blood sample, while RNASeq and cDNA confirmation were performed on two other peripheral blood draws a year and a half apart. Furthermore, a fourth blood draw was used for the long read WGS. In addition, we also looked for somatic-only variants in both cases and found no evidence of them in the corresponding germline, providing further evidence of the absence of tumour contamination in the germline. Contamination with circulating tumour DNA (ctDNA) represents a second alternative explanation. However, the size of ctDNA is <200 bp, and since both of our SVA insertions are much greater than 2 kb, this possibility was also ruled out. While insertion of transposable elements is a well-established mechanism of disease, it was important to establish that the SVA insertions discovered in our two patients were not common or rare variants in the population. Examination of gnomAD SV data in and around SMARCB1 exon 1 and intron 7 showed the absence of SVA elements in these regions, suggesting that the insertions found in our two patients are indeed private and likely the pathogenic mechanism underlying their disease. The clinical significance of a diagnosis of RTPS is considerable. We have recently broadened our understanding of the risk of RTPS-associated tumors, acknowledging there is a risk for late onset tumours beyond the age of 531. As such, molecular confirmation of a diagnosis of RTPS1, as in Case 2, informs ongoing clinical and radiological surveillance in accordance with local and international consensus guidelines11,31,32. A germline diagnosis of RTPS1, or a germline post-zygotic mosaic aberration, as is the case for our two patients, also has implications for the patient’s reproductive planning and can help inform risk for other family members. Cascade testing for the variant can be initiated for at-risk family members to determine their personal cancer risks and surveillance recommendations. As both patients we describe were found to have mosaic germline SVA insertions suggesting post-zygotic occurrence, we expect the recurrence risk for siblings and future pregnancies to be low. These cases expand the body of evidence to support routine comprehensive genomic analysis for the identification of underlying constitutional predisposition in patients with rhabdoid tumours. While the cases presented here showed high clinical suspicion for underlying RTPS, it is likely that a much higher proportion of patients with rhabdoid tumours are germline-driven, requiring advanced molecular methods for diagnosis. SVAs and complex structural rearrangements represent a variant class that remains underreported and may be particularly difficult to detect in the mosaic state; accordingly, these may represent an underappreciated mechanism of pediatric cancer predisposition beyond rhabdoid malignancies. These findings significantly impact clinical care, enabling personalized management for the patient and family. Methods Ethics Patients (or their substitute decision makers) were consented to the SickKids’ Cancer Sequencing (KiCS) Program, approved by the institutional Research Ethics Board (REB 1000051699)13. Sequencing performed through the KiCS program was carried out in accordance with standard procedures, as previously described and outlined below13. Informed consent was also obtained to share their anonymized data for scientific publications in case report form. CARE guidelines: Consensus-based Clinical Case Reporting Guideline Development checklist33 has been considered while writing these case reports and can be found in the Supplementary information. DNA extraction Genomic DNA was extracted using the QIA Symphony automated extraction system (Qiagen, Germany) following the manufacturer’s protocol. Short read DNA sequencing For both cases, paired tumour:normal DNA analysis using an 864-gene panel was performed by NGS to a depth of ~1,000x through the SickKids Cancer Sequencing (KiCS) Program as previously described13. Short read WGS sequencing was performed on DNA extracted from fresh frozen tumour tissue for Case 1 and from tumour and peripheral blood for Case 2. Briefly, DNA was fragmented to 350 bp fragments on the Covaris Sample Prep Station E220. Library preparation was performed from 1 µg of fragmented DNA using the Illumina TruSeq DNA PCR-Free library preparation on an Agilent Bravo robot. Sequencing was performed on Illumina NovaSeq 6000 instruments to an average depth of coverage greater than 30x. Sequencing reads were mapped to the human reference genome build GRCh38 (hg38) using the Illumina DRAGEN DNA pipeline version 4.2. Analysis of small sequence variants, copy number variants, and structural variants was performed using the Illumina DRAGEN DNA pipeline version 4.2. Paired tumour:normal WGS DNA analysis for Case 2 was performed on an Illumina HiSeq X by the Centre for Applied Genomics (TCAG) (Toronto, Canada) as previously described13. Copy number analysis Copy number analysis was performed as previously described13. The NxClinical analysis program has been rebranded as VIA, and version 7.0 was used. Long read DNA sequencing Long read sequencing was performed on DNA extracted from fresh frozen tumour tissue for Case 1 and from peripheral blood for Case 2. DNA samples were quantified using the Qubit dsDNA High Sensitivity Assay. Agilent Genomic DNA ScreenTape was used to ascertain whether most fragments had high molecular weight ( > 30 kb). Subsequently, the Covaris g-TUBE was used to shear 3 µg of DNA to fragments with sizes ranging from 15-20 kb. The fragment sizes were checked again using Agilent Genomic DNA ScreenTape. Sheared DNA was then used for library preparation using the SMRTbell Prep Kit 3.0 (Pacific Biosciences, Menlo Park, CA) following the manufacturer’s recommended protocol. In brief, DNA was end-repaired, ligated with SMRTbell barcoded adapters, and size-selected using the PippinHT System (Sage Science, Beverly, MA) to remove fragments below 10 kb. Subsequently, libraries were validated on the Agilent Genomic DNA ScreenTape, checking for size and quantified by the Qubit dsDNA High Sensitivity Assay. Finally, validated libraries were set up for sequencing on the PacBio Revio, loading one sample per Revio SMRT Cell for a movie time of 24 h. The samples were analyzed with a comprehensive workflow that included both a) realignment of HiFi reads to a reference human genome to detect small variants and large structural rearrangements and b) de novo assembly. The workflow used pbmm2 to align reads to the GRCh38 reference genome and DeepVariant v1.5, a deep learning-based variant caller, to detect SNV and Indels. Structural variants were called with pbsv and copy number variants with HifiCNV. HiPhase was used to phase the small variants and the structural variants. A haplotagged bam file was generated for downstream analyses and visualizing phased variants in IGV or other browsers. The workflow used pb-cp-tools to identify and compute the modification probabilities of CpG/5mC sites. TRGT was used to genotype tandem repeats. De novo assembly of HiFi reads was performed using hifiasm, a haplotype-resolves de novo assembler. The small variants were annotated using an Annovar-based annotation pipeline and the structural variants were annotated using custom scripts. RNA sequencing RNA analysis was performed by whole transcriptome with polyA enrichment through the SickKids Cancer Sequencing (KiCS) Program as previously described13. Long-range and nested PCR Long-range PCR of the SMARCB1 gene [NM_003073.3] was performed using the Qiagen UltraRun LongRange PCR Kit (Qiagen, Germany) with gene-specific primers (10 pmol/μl) under the following conditions: initial denaturation at 93 °C for 3 min; 32 cycles of denaturation at 93 °C for 30 s, annealing at 62 °C for 15 s, extension at 68 °C for 8.5 min (0.5 min/kb); final extension at 68 °C for 10 min. This was followed by nested PCR amplification using nested primers (10 pmol/μl) under the following conditions: initial denaturation at 94 °C for 2 min; 2 cycles of denaturation at 94 °C, annealing at 70 °C, extension at 72 °C, each for 30 s; 2 cycles of denaturation at 94 °C, annealing at 67 °C, extension at 72 °C, each for 30 s; 33 cycles of denaturation at 94 °C, annealing at 65 °C, extension at 72 °C, each for 30 s; final extension at 72 °C for 5 min. The resulting PCR products were analyzed by agarose gel electrophoresis and subjected to Sanger sequencing. Sequencing data were generated using an Applied Biosystems 3730 DNA Analyzer (Thermo Fisher Scientific, USA). RT-PCR Total RNA was extracted using the PAXgene™ Blood RNA Kit (Qiagen, Germany) according to the manufacturer’s instructions. First-strand complementary DNA (cDNA) was synthesized using the SuperScript™ III First-Strand Synthesis System (Invitrogen, USA). Presence of the insertion for Case 2 was confirmed by targeted reverse transcription PCR (RT-PCR) under the following conditions: initial denaturation at 95 °C for 10 min; 35 cycles of denaturation at 95 °C for 15 s, annealing at 55 °C for 30 s, extension at 72 °C for 30 s; final extension at 72 °C for 10 min, followed by agarose gel electrophoresis and Sanger sequencing of the PCR amplicons. Sequencing data were generated using an Applied Biosystems 3730 DNA Analyzer (Thermo Fisher Scientific, USA). Supplementary information CARE-checklist-English-2013_mosaicgermlineSVAinsertionsinSMARCB1_FINAL CARE-checklist-English-2013_mosaicgermlineSVAinsertionsinSMARCB1_FINAL