Germline Sequencing Identifies Rare Variants in Finnish Subjects with Familial Germ Cell Tumors.

PURPOSE: Pediatric germ cell tumors are rare, representing about 3% of childhood malignancies in children less than 15 years of age, presenting in neonates or adolescents with a greater incidence noted in older adolescents. Aberrations in primordial germ cell proliferation/differentiation can lead to a variety of neoplasms, including teratomas, embryonal carcinoma, choriocarcinoma, and yolk sac tumors. PATIENTS AND METHODS: Three Finnish families with varying familial germ cell tumors were identified, and whole-genome sequencing was performed using an Illumina sequencing platform. In total, 22 unique subjects across the three families were sequenced. Family 1 proband (female) was affected by malignant ovarian teratoma, Family 2 proband (female) was affected by sacrococcygeal teratoma with yolk sac tumor in the setting of Cornelia de Lange syndrome, and Family 3 proband (male) was affected by malignant testicular teratoma. Rare variants were identified using an autosomal recessive or de novo model of inheritance.
RESULTS: For family 1 proband (female), an autosomal recessive or de novo model of inheritance identified variants of interest in the following genes: CD109, IKBKB, and CTNNA3, SUPT6H, MUC5AC, and FRG1. Family 2 proband (female) analysis identified gene variants of interest in the following genes: LONRF2, ANO7, HS6ST1, PRB2, and DNM2. Family 3 proband (male) analysis identified the following potential genes: CRIPAK, KRTAP5-7, and CACNA1B.
CONCLUSION: Leveraging deep pedigrees and next-generation sequencing, rare germline variants were identified that were enriched in three families from Finland with a history of familial germ cell tumors. The data presented support the importance of germline mutations when analyzing complex cancers with a low somatic mutation landscape.

Introduction

Pediatric germ cell tumors (GCT) represent about 3% of all childhood malignancies for children less than 15 years of age; this incidence increases to about 15% in adolescence and young adulthood.1 These tumors may arise from the gonads or extra-gonadal tissue. There are five subtypes of pediatric GCT: teratoma, germinoma, yolk sac tumor (also known as endodermal sinus tumor), embryonal carcinoma, and choriocarcinoma, as well as mixed malignant subtypes.1–3 Although infantile and pediatric GCT are biologically distinct from GCT of older adolescents and adults, treatment for all GCT is essentially the same and generally grounded in large doses of platinum-based chemotherapy combined with surgery. Thus, younger patients suffer more severe side effects from treatment.2,4

Predisposition to developing ovarian and testicular teratomas has been reported in families with multiple affected members.3 While the exact mechanisms underlying GCT development remain unclear, abnormalities in the KIT-ligand (KITLG) and KRAS pathway are thought to play a role.5,6 Amplifications in chromosome 12p and X chromosome have been implicated in nonseminomatous GCT. Yolk sac tumors also show gains in chromosomes 1q, 11q, 20q, as well as gains in chromosome 22; losses in chromosome 1p, 6q, and 16q have also been reported.7

There is evidence to support that methylation plays a role in GCT biology, and that methylation patterns identified in GCT can distinguish the five sub-types.8 Based on prevailing literature and our experience with families enriched with germ cell malignancies, germline variant that predisposes a progenitor germ cell to malignant transformation or makes the microenvironment more permissive to tumor growth is likely to be a strong candidate for unraveling the complex biology of these cancers.

Next-generation sequencing (NGS) has significantly advanced our understanding of the genomic landscape of familial cancers.2 Combining deep phenotyping with deep sequencing has the potential to identify key oncogenes involved with rare and aggressive cancers. The goal of this study was to apply these techniques to 3 independent families from Finland with an extensive history of neoplasia, but no known genomic lesions. Of interest, the people of Finland have a unique ancestry marked by a population bottleneck followed by geographic isolation on a peninsula.9 Consequently, the Finnish population has an enrichment in rare genetic variants. Our study leverages the distinctive genetics of this population.

Patients and Methods

Subject Enrollment

Three families from Finland were enrolled for this study. The study was approved by the ethical committee of Helsinki University Hospital, Helsinki, Finland. Written informed consent was obtained from all subjects and/or their parents, as well as the relatives from whom the samples were obtained for this study. A total of 22 unique subjects were sequenced across the 3 families, n=4, n=11, and n=7, respectively (). Family 1 proband had a malignant ovarian teratoma, diagnosed at 8 years of age, with a family history of ovarian teratoma (). Family 2 proband was diagnosed with a neonatal sacrococcygeal teratoma and a subsequent yolk sac tumor at 1.5 years of age, with a family history of malignancies (NOS) (). Family 3 had a proband with a malignant testicular teratoma, diagnosed at 1 year of age, with a family history of malignancies (NOS) (). Peripheral blood was collected from each proband and family members.

De-identified analysis of genomic data from the consented patients was conducted at Nemours (Dr. Crowgey) with IRB research determination (IRB# 1327922–1).

Whole-Genome Sequencing (WGS) and Genome Alignment

Whole-genome sequencing was conducted at Washington University School of Medicine by the McDonnell Genome Institute. Whole-genome sequencing was conducted using the NovaSeq 6000 high-throughput sequencing platform (Illumina, San Francisco, CA) at a depth of coverage of 30X. Samples were prepared using the TruSeq whole genome library preparation (Illumina). Fastq files were processed via fastqc for data quality (Babraham Institute) and trimmed based on adaptor sequences and quality (cutadapt). Trimmed reads were mapped to the human genome hg19 via bwa mem.10 Genome analysis toolkit (GATK) best practices, base quality score recalibration, indel realignment, and duplicate read removal were followed to generate a combined variant call file (VCF) using Haplotyper for all samples analyzed.11–13

Variant Annotation and Data Analysis

VCFs were annotated using VarSeq version 2.2.0 (Golden Helix®). Specifically, variants were annotated using: ExAC Variant Frequencies 0.3, gnomAD exomes and genome variant frequencies 2.1.1,14 dbSNP 151 (), Ensemble Genes 87, and functional predictions were determined via sift,15 polyphen2 HVAR,16 mutation taster,17 mutationAssessor,18 FATHMM,19 and FATHMM MKL coding. At least three of the functional predictions needed to predict the variant as damaging. Variants were filtered based on the following inheritance patterns when appropriate autosomal recessive, and de novo, using the following criteria: read depth >10, genotype qualities >20, MAF ExAC <0.01, MAF gnomAD <0.01, and effect of mutation (loss of function or missense for coding variants). The same logic was applied for non-coding variants located in introns and non-exonic (intergenic) locations.

A variant was considered as autosomal recessive for proband 1 and 2 if both the mother and father were carriers (0/1), unaffected siblings were either carriers or wildtype (0/1 or 0/0), and the proband was homozygous alternative (1/1). Additionally, for family 2 proband, the unaffected extended family members in the pedigree were used to further filter out variants if they were a homozygous alternative (1/1) genotype. Variants were considered as de novo for proband 1 and 2 if both the mother and father were homozygous wildtype (0/0), and the unaffected siblings were also homozygous wildtype (0/0). Additionally, for family 2 proband, the unaffected extended family members in the pedigree were used to further filter out variants if they were a carrier (0/1). For proband 3, unfortunately the mother’s sample was not available for sequencing. The same logic as above for probands 1 and 2 was applied, except that the mother’s parents’ genotypes were used in replacement.

Results

Family 1

Four subjects were sequenced in Family 1 pedigree: mother, father, proband (ovarian teratoma), and unaffected sibling (Figure 1). A history of ovarian teratoma was noted by the primary care physician (); however, germline samples from these extended family members were not available for sequencing. Using an autosomal recessive analysis, three variants of interest were identified in the following genes: CD109 (rs7741152), IKBKB (rs140485496), and CTNNA3 (rs192093851) (Table 1). All of the variants identified were rare within the other Finnish families analyzed with CD109 and IKBKB alleles only identified in Family 1, and 1 allele count outside of Family 1 for rs192093851 (CTNNA3). It was noted that all three variants had a <0.01 minor allele frequency (MAF) within the gnomAD database; however, they were >0.01 within the Finnish-specific genomes within gnomAD (Table 1). Inhibitor of NF-κB kinase subunit beta (IKBKB) is associated with immunodeficiency disorders (IMD15B and 15A) and is a serine kinase that plays a key role in the NF-κB signaling pathway. Gene variants and expression data have linked IKBKB with prostate and ovarian cancers.20 Of interest, IKBKB has been associated with KRAS, which is a known oncogene linked to GCTs.21

Figure 1

Family 1 pedigree. Four family members were sequenced via whole-genome sequencing: proband, mom, dada, and unaffected sister. The proband, black circle, was a malignant ovarian teratoma. Females are represented as circles; males are represented as squares. The number next to the pedigree represents the de-identified subject ID.

Table 1

Rare Autosomal Recessive and De Novo Variants Identified in Family 1 Proband

Inhertiance Patterm	Chr:Position	Ref/Alt	ID	Proband	Father	Mother	Sister	Gene Names	HGVS c. (Clinically Relevant)	HGVS p. (Clinically Relevant)	Effect	GnomAD	GnomAD European Finnish
				91/98	94/98	92/98	93/98
Autosomal recessive	6:74475675	G/A	rs7741152	1/1	0/1	0/1	0/1	CD109	NM_133493.4:c.1130G>A	NP_598000.2:p.Gly377Asp	missense_variant	0.0079	0.01932
	8:42178280	C/T	rs140485496	1/1	0/1	0/1	0/1	IKBKB	NM_001556.2:c.1606C>T	NP_001547.1:p.Arg536Trp	missense_variant	0.00373	0.02279
	10:68979466	C/A	rs192093851	1/1	0/1	0/1	0/0	CTNNA3	NM_013266.3:c.742G>T	NP_037398.2:p.Ala248Ser	missense_variant	0.00344	0.02396
De novo	1:6505837	AG/-		0/1	0/0	0/0	0/0	ESPN	NM_031475.2:c.1306_1307delAG	NP_113663.2:p.Ser436Leufs	frameshift_variant
	1:109268468	-/CCGTA	rs1342022852	0/1	0/0	0/0	0/0	FNDC7	NM_001144937.1:c.952_953insCCGTA	NP_001138409.1:p.Val318Alafs	frameshift_variant	0.0011	0
	1:109268470	T/A	rs1223692320	0/1	0/0	0/0	0/0	FNDC7	NM_001144937.1:c.955T>A	NP_001138409.1:p.Phe319Ile	missense_variant	0.00043	0
	1:109268473	GTG/-	rs1248759911	0/1	0/0	0/0	0/0	FNDC7	NM_001144937.1:c.958_960delGTG	NP_001138409.1:p.Val320del	inframe_deletion	0.00041	0
	1:109268476	-/AAAAAA	rs1225123862	0/1	0/0	0/0	0/0	FNDC7	NM_001144937.1:c.960_961insAAAAAA	NP_001138409.1:p.Lys321_Ser322insLysLys	inframe_insertion	0.00043	0
	4:190878556	-/A		0/1	0/0	0/0	0/0	FRG1	NM_004477.2:c.435_436insA	NP_004468.1:p.Met147Asnfs	frameshift_variant
	11:1213565	-/A		0/1	0/0	0/0	0/0	MUC5AC	NM_001304359.1:c.2680_2681insA	NP_001291288.1:p.Pro894Hisfs	frameshift_variant
	11:1213571	-/AACCCGGG	rs1300356008	0/1	0/0	0/0	0/0	MUC5AC	NM_001304359.1:c.2686_2687insAACCCGGG	NP_001291288.1:p.Thr896Lysfs	frameshift_variant	0.00158	0.003837
	12:109017014	T/G	rs1373341486	0/1	0/0	0/0	0/0	SELPLG	NM_001206609.1:c.1118A>C	NP_001193538.1:p.Tyr373Ser	missense_variant	0.00128	0
	17:27013754	G/A		0/1	0/0	0/0	0/0	SUPT6H	NM_003170.4:c.2647G>A	NP_003161.2:p.Glu883Lys	missense_variant
	18:34298343	ACCTTTCTGGGTTTG/-		0/1	0/0	0/0	0/0	FHOD3	NM_001281740.2:c.3082_3096delACCTTTCTGGGTTTG	NP_001268669.1:p.Thr1028_Leu1032del	inframe_deletion
	12:40897291	G/A	rs1276236479	0/1	0/0	0/0	0/0	MUC19	NM_173600.2:c.18907G>A	NP_775871.2:p.?	missense_variant	0.00581	0.003076
	14:22102317	G/T	rs200792644	0/1	0/0	0/0	0/0	OR10G2	NM_001005466.2:c.682C>A	NP_001005466.2:p.His228Asn	missense_variant	0.00646	0.0009194
	1:151317205	G/C	rs747555214	0/1	0/0	0/0	0/0	RFX5	NM_000449.3:c.352C>G	NP_000440.1:p.Arg118Gly	missense_variant	0	0

The de novo analysis identified 14 rare variants within proband one across 10 unique genes (Table 1). Nine of the variants had known RefSeq (rs#) identification numbers in the dbSNP database and MAF of >0.01 in the gnomAD genome database as well as within the Finnish-specific genomes within gnomAD (Table 1). Of interest, SUPT6H, a histone chaperone, has previously been shown to control estrogen-related transcription and linked to cancer.22 Additionally, MUC5AC has been associated with ovarian tumors23 and FRG1 expression levels have been shown to be aberrant in several cancers.

Family 2

Eleven subjects were sequenced in Family 2 pedigree: proband (sacrococcygeal teratoma – yolk sac tumor), unaffected brother with a benign sacrococcygeal cyst, unaffected sister, mother, father, two paternal aunts, paternal grandmother, 2 paternal great aunts, and a paternal cousin (; Figure 2). Of interest, the unaffected brother and nephew (unaffected sister) have a history of benign dermoid cysts (sacrococcygeal and ophthalmic area). Furthermore, the proband’s great grandfather (maternal side) had prostate cancer, and a great grandmother with stomach cancer, who has a daughter with leukemia and a granddaughter with cancer. An autosomal recessive analysis identified two rare variants in the following genes LONRF2 (rs116702638) and ANO7 (rs148609049) (Table 2). It was noted that all three variants had a <0.01 MAF within the gnomAD database; however, they were >0.01 within the Finnish-specific genomes within gnomAD (Table 1). Variant rs148609049 has previously been published as a germline variant that associates with significantly shorter survival rates in prostate cancer subjects.24 Aberrant methylation patterns for LON peptidase N-terminal domain and RING finger protein 2 (LONRF2) have been noted in rectal adenocarcinoma. Of interest, ANO7 is associated with aggressive prostate cancer.24

Figure 2

Family 2 pedigree. Eleven subjects were sequenced via whole-genome sequencing: proband, unaffected brother, unaffected sister, mother, father, two paternal aunts, paternal grandmother, 2 paternal great aunts, and a paternal cousin. The proband, black circle, was a sacrococcygeal teratoma – yolk sac tumor. Females are represented as circles; males are represented as squares. The number next to the pedigree represents the de-identified subject ID.

Table 2

Rare Autosomal Recessive and De Novo Variants Identified in Family 2 Proband

Inhertiance Pattern	Chr:Position	Ref/Alt	ID	Father	Mother	Proband	Sister	Brother	Aunt	Aunt	Grandma	Great aunt	Great aunt	Cousin	Gene Names	HGVS c. (Clinically Relevant)	HGVS p. (Clinically Relevant)	Effect	GnomAD	GnomAD European (Finnish)
				95/98	96/98	97/98	105/98	106/98	108/98	112/98	107/98	111/98	110/98	109/98
Autosomal recessive	2:100915330	G/T	rs116702638	0/1	0/1	1/1ʹ	0/1	0/1	0/1	0/1	0/1	0/1	0/1	0/0	LONRF2	NM_198461.3:c.1444C>A	NP_940863.3:p.His482Asn		0.008377	0.0201
	2:242128114	C/T	rs148609049	0/1	0/1	1/1ʹ	0/0	0/1	0/0	0/1	0/1	0/0	0/1	0/0	ANO7	NM_001001891.3:c.88C>T	NP_001001891.2:p.Arg30Ter		0.007657	0.03571
De novo	1:152278769	-/CC	rs1205280854	0/0	0/0	0/1	0/0	0/0	0/0	0/0	0/0	0/0	0/0	0/0	FLG	NM_002016.1:c.8593_8594insGG	NP_002007.1:p.Ala2865Glyfs	frameshift_variant	0.003439	0.02396
	6:43155555	CC/-	rs1235342930	0/0	0/0	0/1	0/0	0/0	0/0	0/1	0/0	0/0	0/0	0/0	CUL9	NM_015089.3:c.1686_1687delCC	NP_055904.1:p.Leu563Alafs	frameshift_variant	0.003898	0
	6:43155558	-/ACG	rs1217118305	0/0	0/0	0/1	0/0	0/0	0/0	0/1	0/0	0/0	0/0	0/0	CUL9	NM_015089.3:c.1688_1689insACG	NP_055904.1:p.Leu563_Leu564insArg	disruptive_inframe_insertion	0.006126	0
	1:152278771	TG/-	rs1407703398	0/0	0/0	0/1	0/0	0/0	0/0	0/0	0/0	0/0	0/0	0/0	FLG	NM_002016.1:c.8590_8591delCA	NP_002007.1:p.His2864Cysfs	frameshift_variant
	1:152278831	-/TTG	rs1448728360	0/0	0/0	0/1	0/0	0/0	0/0	0/0	0/0	0/0	0/0	0/0	FLG	NM_002016.1:c.8531_8532insCAA	NP_002007.1:p.Glu2844delinsAspLys	disruptive_inframe_insertion
	2:129025758	C/A	rs142919429	0/0	0/0	0/1	0/0	0/0	0/0	0/0	0/0	0/0	0/0	0/0	HS6ST1	NM_004807.2:c.1214G>T	NP_004798.3:p.Ser405Ile	missense_variant	0.001622	0.0008978
	3:195506483	TGTCGGTGACAGGAAGGGGGGTGGCGTGACCTGTGGATGCTGAGGAAC/-	rs1560301180	0/0	0/0	0/1	0/0	0/0	0/0	0/0	0/0	0/0	0/0	0/0	MUC4	NM_018406.6:c.11921_11968delGTTCCTCAGCATCCACAGGTCACGCCACCCCCCTTCCTGTCACCGACA	NP_060876.5:p.Arg3974_Thr3990delinsPro	disruptive_inframe_deletion
	3:195512233	-/GCCGAGGAAACGTTGGTGACAGGAAGACGGGTGGTGTCACCTGTGGAA	rs529636680	0/0	0/0	0/1	0/0	0/0	0/0	0/0	0/0	0/0	0/0	0/0	MUC4	NM_018406.6:c.6218_6219insTTCCACAGGTGACACCACCCGTCTTCCTGTCACCAACGTTTCCTCGGC	NP_060876.5:p.Gly2076_His2077insAspThrThrArgLeuProValThrAsnValSerSerAlaSerThrGly	disruptive_inframe_insertion	0.00004043	0
	6:43155547	C/T	rs1300018649	0/0	0/0	0/1	0/0	0/0	0/0	0/0	0/0	0/0	0/0	0/0	CUL9	NM_015089.3:c.1678C>T	NP_055904.1:p.Leu560Phe	missense_variant	0.004704	0
	6:43155560	-/GA	rs1264139231	0/0	0/0	0/1	0/0	0/0	0/0	0/0	0/0	0/0	0/0	0/0	CUL9	NM_015089.3:c.1690_1691insGA	NP_055904.1:p.Leu564Argfs	frameshift_variant
	6:43155560	T/C	rs1489405189	0/0	0/0	0/1	0/0	0/0	0/0	0/0	0/0	0/0	0/0	0/0	CUL9	NM_015089.3:c.1691T>C	NP_055904.1:p.Leu564Pro	missense_variant	0	0
	6:43155563	ACT/-	rs1189505569	0/0	0/0	0/1	0/0	0/0	0/0	0/0	0/0	0/0	0/0	0/0	CUL9	NM_015089.3:c.1694_1696delACT	NP_055904.1:p.Asn565_Ser566delinsThr	disruptive_inframe_deletion	0.004741	0.00003985
	11:1093483	GT/-	rs747244421	0/0	0/0	0/1	0/0	0/0	0/0	0/0	0/0	0/0	0/0	0/0	MUC2	NM_002457.4:c.5296_5297delGT	NP_002448.4:p.Val1766Thrfs	frameshift_variant
	12:11546859	-/CCTTGAGGCTGGTTGCCTCCTTGTGGGGGTGCTCCTTGTGGCTTTCCTGGAGGAGG		0/0	0/0	0/1	0/0	0/0	0/0	0/0	0/0	0/0	0/0	0/0	PRB2	NM_006248.3:c.153_154insCCTCCTCCAGGAAAGCCACAAGGAGCACCCCCACAAGGAGGCAACCAGCCTCAAGG	NP_006239.3:p.Ser52Profs	frameshift_variant
	12:40876971	C/A	rs769109166	0/0	0/0	0/1	0/0	0/0	0/0	0/0	0/0	0/0	0/0	0/0	MUC19	NM_173600.2:c.10217C>A	NP_775871.2:p.?	missense_variant	0.002807	0.0003193
	12:40876979	A/C	rs1273078187	0/0	0/0	0/1	0/0	0/0	0/0	0/0	0/0	0/0	0/0	0/0	MUC19	NM_173600.2:c.10225A>C	NP_775871.2:p.?	missense_variant	0.001104	0
	19:10908156	TCAGAGCTG/-	rs1295151630	0/0	0/0	0/1	0/0	0/0	0/0	0/0	0/0	0/0	0/0	0/0	DNM2	NM_001005360.2:c.1336-1006_1336-998delTCAGAGCTG		inframe_deletion	0.009196	0.000204
	19:41811728	-/CCCCCCCA		0/0	0/0	0/1	0/0	0/0	0/0	0/0	0/0	0/0	0/0	0/0	HNRNPUL1	NM_007040.5:c.2409_2410insCCCCCCCA	NP_008971.2:p.Thr804Profs	frameshift_variant
	19:52888076	-/ATGAGGTCAGGAGATCGAGACCATCCTGGCTAACAAGGTG		0/0	0/0	0/1	0/0	0/0	0/0	0/0	0/0	0/0	0/0	0/0	ZNF880	NM_001145434.1:c.1242_1243insATGAGGTCAGGAGATCGAGACCATCCTGGCTAACAAGGTG	NP_001138906.1:p.Lys415Metfs	frameshift_variant

The de novo analysis identified 19 rare variants within proband two across 10 unique genes (Table 2). Sixteen of the variants were found in dbSNP with MAF of >0.01 in the gnomAD genome database and within the Finnish-specific genomes within gnomAD (Table 2). Of interest HS6ST1, PRB2, and DNM2 have been previously linked to cancers.25,26

Family 3

Seven subjects were sequenced in Family 3 pedigree: proband (malignant testicular teratoma), un-affected siblings, dad, uncle (maternal), grandparents (maternal) (; Figure 3). Two InDels were identified in this family using an autosomal recessive pattern in the following genes (Table 3): CRIPAK and KRTAP5-7. CRIPAK encodes for a cysteine-rich PAK1 inhibitor protein that negatively regulates PAK1 expression. PAKs are a family of proteins that are involved in cytoskeletal dynamics, cell survival and proliferation and have been well associated with cancers.27 The de novo analysis identified 22 rare variants in 17 unique genes. Aberrant expression levels of CACNA1B have previously been associated with cancers.28 Of interest, TXNDC2 is solely expressed in testis and is predicted to regulate disulfide bonds, and TFE3 is a transcription factor linked to cancers and is ubiquitous in fetal and adult tissues.

Figure 3

Family 3 pedigree. Seven subjects were sequenced via whole-genome sequencing: proband, unaffected siblings, data, maternal uncle, and maternal grandparents. Females are represented as circles; males are represented as squares. The number next to the pedigree represents the de-identified subject ID.

Table 3

Autosomal Recessive and De Novo Variants Identified in Family 3 Proband

Inhertiance Pattern	Chr:Pos	Ref/Alt	ID	Proband	Sister	Father	Uncle	Grandmother	Sister	Grandfather	Gene Names	HGVS c. (Clinically Relevant)	HGVS p. (Clinically Relevant)	Effect	GnomAD	GnomAD European Finnish
				100/98	101/98	102/98	103/98	104/98	98/98	99/98
Autosomal recessive	4:1388626	ACGTGCCGATGCGGA	rs201732211	1/1’	0/0	0/0	0/0	0/0	0/0	0/0	CRIPAK			Missense	0.001105	0.002893
	11:71238676	CTGCTGCCAGTCCAGCTGCTGTAAGCCCTGCTGCTGCCAGTCCAGCTGCTGTAAGCCCTG	rs1393220620	1/1’	0/1	0/1	0/0	0/1	0/0	0/0	KRTAP5-7	NM_001012503.1:c.329_330insCTGCTGCCAGTCCAGCTGCTGTAAGCCCTGCTGCTGCCAGTCCAGCTGCTGTAAGCCCTG	NP_001012521.1:p.Cys132_Ser133insGlnSerSerCysCysLysProCysCysCysGlnSerSerCysCysLysProCysCysCys	Missense	0.0006583	0
De novo	1:152284478	A/G	rs199888588	0/1	0/0	0/0	0/0	0/0	0/0	0/0	FLG	NM_002016.1:c.2884T>C	NP_002007.1:p.Trp962Arg	missense_variant	0.02648	0.000609
	1:240371009	T/C	rs200975594	0/1	0/0	0/0	0/0	0/0	0/0	0/0	FMN2	NM_001305424.1:c.2909T>C	NP_001292353.1:p.Leu970Pro	missense_variant	0.002757	0.0000499
	3:195507970	G/A	rs201000518	0/1	0/0	0/0	0/0	0/0	0/0	0/0	MUC4	NM_018406.6:c.10481C>T	NP_060876.5:p.Pro3494Leu	missense_variant	0.0004797	0.000198
	3:195507971	G/A	rs767776470	0/1	0/0	0/0	0/0	0/0	0/0	0/0	MUC4	NM_018406.6:c.10480C>T	NP_060876.5:p.Pro3494Ser	missense_variant	0.0004074	0
	3:195507973	A/C	rs200368431	0/1	0/0	0/0	0/0	0/0	0/0	0/0	MUC4	NM_018406.6:c.10478T>G	NP_060876.5:p.Ile3493Ser	missense_variant	0.001908	0.0001092
	4:9245723	A/T	rs1411951281	0/1	0/0	0/0	0/0	0/0	0/0	0/0	USP17L17	NM_001256857.1:c.119A>T	NP_001243786.1:p.Lys40Met	missense_variant	0.001287	0.0008681
	5:140222138	-/CCAACTGATCTGATATATTGTATAGTTTAATAGCTTCTCTAGTCATCTTAAACAGGGTTGG	rs782205437	0/1	0/0	0/0	0/0	0/0	0/0	0/0	PCDHA1	NM_018900.3:c.2394+53868_2394+53869insCCAACTGATCTGATATATTGTATAGTTTAATAGCTTCTCTAGTCATCTTAAACAGGGTTGG	NP_061734.1:p.Ser411Thrfs	frameshift_variant	0.00042	0.0002905
	7:97619354	A/C	rs1390112627	0/1	0/0	0/0	0/0	0/0	0/0	0/0	OCM2	NM_006188.3:c.61+2T>G		splice_donor_variant	0.00007551	0
	8:12286291	C/T	rs1211097648	0/1	0/0	0/0	0/0	0/0	0/0	0/0	FAM86B2	NM_001137610.1:c.593G>A	NP_001131082.1:p.Arg198Gln	missense_variant	0.002466	0.0007554
	9:140773613	-/ACGACACGGAGCCCTATTTCATCGGGATCTTT	rs370237172	0/1	0/0	0/0	0/0	0/0	0/0	0/0	CACNA1B	NM_000718.3:c.390+1_390+2insACGACACGGAGCCCTATTTCATCGGGATCTTT		splice_donor_variant	0.005113	0.003842
	11:1212902	C/G	rs200292517	0/1	0/0	0/0	0/0	0/0	0/0	0/0	MUC5AC	NM_001304359.1:c.2018C>G	NP_001291288.1:p.Thr673Ser	missense_variant	0.002578	0.00328
	11:1213367	G/A	rs748424415	0/1	0/0	0/0	0/0	0/0	0/0	0/0	MUC5AC	NM_001304359.1:c.2483G>A	NP_001291288.1:p.Arg828His	missense_variant	0.0006029	0.00007648
	11:1266007	T/C	rs774768277	0/1	0/0	0/0	0/0	0/0	0/0	0/0	MUC5B	NM_002458.2:c.7897T>C	NP_002449.2:p.Trp2633Arg	missense_variant	0.00009861	0.00004068
	11:1266815	G/A	rs200874900	0/1	0/0	0/0	0/0	0/0	0/0	0/0	MUC5B	NM_002458.2:c.8705G>A	NP_002449.2:p.Arg2902His	missense_variant	0.0006634	0.0002522
	11:1269763	C/A	rs200531133	0/1	0/0	0/0	0/0	0/0	0/0	0/0	MUC5B	NM_002458.2:c.11653C>A	NP_002449.2:p.Arg3885Ser	missense_variant	0.002626	0.008074
	11:1605904	-/CCCCCCACAAGAACCGCAGCCCCCC	rs1316818204	0/1	0/0	0/0	0/0	0/0	0/0	0/0	KRTAP5-1	NM_001005922.1:c.576_577insGGGGGGCTGCGGTTCTTGTGGGGGG	NP_001005922.1:p.Ser193Glyfs	frameshift_variant	0.00003881	0
	11:64032525	C/G		0/1	0/0	0/0	0/0	0/0	0/0	0/0	PLCB3	NM_000932.2:c.2755C>G	NP_000923.1:p.Arg919Gly	missense_variant
	12:40882472	-/AGAGACAATTGGACTATCAGCTGGAGTGAT	rs1159825088	0/1	0/0	0/0	0/0	0/0	0/0	0/0	MUC19	NM_173600.2:c.15716_15717insAGAGACAATTGGACTATCAGCTGGAGTGAT	NP_775871.2:p.?	disruptive_inframe_insertion	0	0
	14:19378000	A/G	rs761048370	0/1	0/0	0/0	0/0	0/0	0/0	0/0	OR11H12	NM_001013354.1:c.407A>G	NP_001013372.1:p.Asp136Gly	missense_variant	0.00398	0.0002808
	18:9887384	T/C	rs1442865736	0/1	0/0	0/0	0/0	0/0	0/0	0/0	TXNDC2	NM_001098529.1:c.908T>C	NP_001091999.1:p.Leu303Pro	missense_variant	0.001197	0.0001793
	18:76754481	-/CCCCCCCCCCCCCCCCG		0/1	0/0	0/0	0/0	0/0	0/0	0/0	SALL3	NM_171999.3:c.2489_2490insCCCCCCCCCCCCCCCCG	NP_741996.2:p.Ser832Profs	frameshift_variant
	X:48887808	T/C	rs1286561719	0/1	0/0	0/0	0/0	0/0	0/0	0/0	TFE3	NM_006521.5:c.1589A>G	NP_006512.2:p.Glu530Gly	missense_variant

Non-Coding Variant Analysis

Rare variants (gnomAD MAF <0.01) in non-coding regions of the genome were analyzed, introns or intergenic regions. Heterozygous variants were only considered for a proband if the parents and unaffected siblings were homozygous wildtype (0/0), and homozygous variants were only considered for a proband if the parents and unaffected siblings were homozygous wildtype or heterozygous (0/1). A variant was considered as known if it was listed in dbSNP 151. The range of intronic variants detected per proband was 2526–3439 (left side, Figure 4), with the majority of variants per proband were known heterozygous. The range of intergenic variants detected was 8320–10826 (right side, Figure 4), with the majority of the variants also being known heterozygous.

Figure 4

Non-coding variants identified in the probands. Rare non-coding variants were identified in introns (left bars) and non-exonic regions (intergenic, right bars) per each proband. A variant was considered to be known if it was listed in dbSNP 151. Known homozygous variants are in blue, novel homozygous variants are in orange, known heterozygous variants are in grey, and novel heterozygous variants are in yellow.

Discussion

GCT are primarily sporadic cancers in early childhood or adolescence with a variety of subtypes. Thus, the mechanisms involved in GCT are poorly understood. The identification of families with relatively little admixture and multiple GCT cases offers an opportunity to identify novel genes that may drive or modulate tumor onset and subtype. In this study, we performed WGS on three such non-consanguinous families from Finland, where the proband in each family had a different GCT subtype. The NGS analysis revealed different genes with an autosomal recessive or de novo inheritance pattern in each family.

Family 1, proband with malignant ovarian teratoma, analysis yielded three potential autosomal recessive variants of interest in the following genes: CD109, IKBKB, and CTNNA3. Although variants had a MAF <0.01 in the gnomAD genome database, they did have higher MAF in the Finnish-specific genome data. We hypothesize that germline variants might cause a predisposition, but these higher allele frequencies in the Finnish population indicate that a second hit/additional variant in combination would be necessary. CD109 is highly associated with cancers29 and aberrant expression patterns are associated with squamous cell carcinoma.30 IKBKB is a potent activator of the NF-kB pathway, which plays a major regulatory role during developmental transitions and the ability of KRAS to transform tissues with aberrant NF-κB signaling is well established.31 Additionally, dysregulation of microRNAs that target IKBKB has been associated with ovarian tumors.32 The de novo variant analysis identified rare variants within the gnomAD Finnish data, with the most interesting gene candidates being SUPT6H, a histone chaperone, MUC5AC has been associated with ovarian tumors,23 and FRG1.

An autosomal recessive analysis for family 2 yielded two potential variants of interest in the following genes: LONRF2, and ANO7. ANO7 encodes for the protein anoctamin-7, which has calcium-dependent phospholipid scramblase activity. The variant detected is a loss of function variant resulting in a stop gain at amino acid 30 in the translated protein and has previously been published as a germline variant that associates with significantly shorter survival rates in prostate cancer subjects.24 The de novo analysis identified rare variants within the gnomAD Finnish data and included HS6ST1, PRB2, and DNM2, which all have been previously linked to cancers.25,26

Family 3 had variants of interest identified in CRIPAK and KRTAP5-7. CRIPAK is a novel interacting partner of PAK1.33 PAK1, also referred to as p21-activated kinase 1, is a complex gene involved in many cellular signaling pathways including mitosis. Recent studies have described PAK1 has a relevant oncogene in ovarian carcinoma34 and breast carcinoma,35 phenotypes recorded within this family history. However, the pedigree and variant results for Family 3’s proband suggest a potential sporadic case of a neoplasm. A de novo analysis did identify rare variants in TXNDC2, which is solely expressed in testis and is predicted to regulate disulfide bonds, and TFE3, a transcription factor linked to pediatric cancers that is ubiquitously expressed in fetal and adult tissues.

A current limitation of this study is the lack of functional studies, which are required to further validate the biological significance of the variants identified. It is challenging to functionally validate these types of complex variants associated with a complex phenotype, as models are difficult to establish. Unfortunately, tumor testing, which looks for abnormalities in cancer cells, was not conducted at the time of diagnosis for the probands. These data could have provided additional insight for interpreting the germline variants identified. Furthermore, the non-coding variant analysis identified numerous variants that are difficult to interpret without additional functional testing.