Unusual phenotypes in patients with a pathogenic germline variant in DICER1

DICER1 is a member of ribonuclease III (RNaseIII) family and responsible for microRNA processing. Thus, DICER1 modulates gene expression. Germline pathogenic variants in DICER1 cause a tumour predisposition syndrome (OMIM 601200), which is characterized by occurrence of pleuropulmonary blastoma, Sertoli-Leydig cell tumour, cystic nephroma, multinodular goiter, embryonic rhabdomyosarcoma of the cervix uteri and other tumour types [1].

Mosaic somatic missense DICER1 variants in the RNase IIIb domain are linked to GLOW syndrome, an acronym from the reported core features of global developmental delay, lung cysts, overgrowth, and Wilms’ tumour (OMIM 618272) [2]. The discussion of whether GLOW syndrome is a separate entity is ongoing: meanwhile some patients with pathogenic germline variants in DICER1 are reported matching the phenotype of GLOW syndrome and not all mosaic variants in the RNAase IIIb domain lead to the typical GLOW syndrome phenotype but instead are in line with the variable expressivity and reduced penetrance of DICER1-associated features [3].

Conventional karyotyping as well as chromosomal whole-genome microarray from peripheral blood lymphocytes revealed normal results.

Whole exome sequencing (WES) in patients 1, 2 and 4 was carried out using a probe-based capture method to enrich the target regions (IDT Coraville, IA, USA). Alignment to the reference genome (hg19 or hg38), variant calling and analysis was performed using an in-house pipeline based on SeqMule and Kggseq. In patient 3, WES was performed on the proband and her mother as previously reported with modification in capture kit, Roche NimbleGen’s SeqCap EZ Human Exome Library, v3.0 with 64 Mb of exonic sequence targeted (Roche NimbleGen, Inc., Madison, WI) and sequencer NovaSeq 6000 (Illumina, San Diego, CA). Variants were called using three callers, HaplotypeCaller (version 3.8-1-0-gf15c1c3ef), UnifiedGenotyper (version 3.8-1-0-gf15c1c3ef) and FreeBayes (version v0.9.14-24-gc292036). Variants passed the GATK hard filter (QD < 2.0, FS > 60.0, MQ < 40.0, MQRankSum < − 12.5, ReadPosRankSum < − 8.0, SOR > 3.0 for SNV and QD < 2.0, FS > 200.0, ReadPosRankSum < − 20.0, SOR > 10.0 for INDEL), ABHet is between 0.2 and 0.8, and called by at least two of three callers.

In case 2, the heterozygous DICER1 missense variant c.4031C>T; p.(Ser1344Leu) was identified in DNA of blood lymphocytes. This variant is located in the RNase IIIa domain (Fig. 2) and has previously been reported as a somatic mutation in various cancers, including Wilms tumour and as somatic hotspot mutation in uterine cancer [5, 6]. No other obvious pathogenic variants were detected by whole-exome sequencing (WES). Unfortunately, genetic testing could not be performed on patient’s parents nor on her two healthy siblings.

In case 3, the heterozygous nonsense variant c.3073G>T; p.(Glu1025*) in DICER1 was identified in DNA of blood lymphocytes. Exome sequencing the patient did not reveal any other homozygous, compound heterozygous, X-linked or dominant pathogenic or likely pathogenic sequence variants in a gene known to be associated with a human phenotype (by OMIM listing), other than the known pathogenic DICER1 variant.

In case 4 molecular genetics revealed a heterozygous variant in DICER1 (Exon 21; c.3234_3237dupTGGC; p.(Val1080Trpfs*12)). This frameshift variant has not been described previously. The DICER1 variant was inherited from the 50-year-old father without a history of tumours. Interestingly, heterozygosity for a frameshift variant c.5915_5916del, p.(Cys1972Tyrfs*11) (Chr6(GRCh38):g.157206936_157206937del) in ARID1B was detected in addition. This variant has not yet been described in the genome variant databases (gnomAD, ClinVar) and is suspected to be pathogenic. The variant was not detected in blood from the parents, suggesting a de novo origin. ARID1B mutations are associated with autosomal dominant inherited Coffin-Siris syndrome type 1 (OMIM 135900), which may in part explain some of the additional clinical features of the patient.

Most germline DICER1 pathogenic variants are loss-of-function variants (LOF) [1]. LOF variants as in our patients 1, 3 and 4 are not likely to cause skewed miRNA processing, however, one cannot exclude the possibility that developmental defects might arise from DICER 1 haploinsufficiency. Developmental delay and a syndromic phenotype combined with classical DICER1-associated tumours have been rarely reported and are described in association with a 14q32 deletion encompassing the DICER1 locus [7, 8] (Table 1). One female patient with 14q32 deletion showed developmental delay, particularly in the field of speech in combination with mild facial dysmorphism (thick eyebrows, wide nasal base and bulbous nose) [9] as our patient 4 with a frameshift variant in DICER1 and in ARID1B. This similarity of both patients raises the possibility that developmental delay and facial dysmorphism in both patients may be associated with DICER1. However, the situation is complicated by the fact that the patient also has the characteristics of Coffin-Siris syndrome and most additional phenotypic features like tracheomalacia and arachnoidal cyst may be well explained by the latter diagnosis [10]. Our patient 4 shows that, nevertheless, it cannot be excluded that in individual cases two rare conditions may occur simultaneously and that a comprehensive molecular genetic diagnosis is necessary in the presence of additional symptomatology.

Developmental delay is also a feature of GLOW syndrome (mosaic DICER1 variant in the RNase IIIb domain) [2]. There are several similarities between the phenotypes of patient 2 and the two previously published patients with GLOW syndrome such as macrocephaly and macrosomia at birth, Wilms tumour, hydrocephalus, hypertelorism, lung cysts and developmental delay [2]. Unfortunately, segregation analysis in the family of patient 2 could not be undertaken and thus mosaicism in our patient cannot be ruled out, but allele distribution argues for a germline heterozygous DICER1 variant. The detected missense variant c.4031C>T; p.(Ser1344Leu) within the RNase IIIa domain has been described so far only as somatic mutation in patients with Wilms tumour and other cancers, but not as germline variant, leaving a rest of uncertainty on the pathogenicity of this variant [11]. Interestingly, mutations within the RNase IIIa domain have been shown to phenocopy mutations in the RNase IIIb domain presumably due to the constrained proximity of the RNase IIIa and RNase IIIb as shown by structural and evolutionary coupling analyses [6]. This constrained proximity could also explain why the presumed germline variant in patient 2 leads to a phenotype similar to the one described in the two patients with GLOW syndrome associated with mosaic DICER1 mutation in the RNase IIIb domain, suggesting a genotype–phenotype relationship with missense mutations in RNase III. Whether a missense variant as in patient 2 is sufficient to lead to tumor development or a second somatic hit in DICER1 is needed, as presumed in patients with LOF variants, remains speculative.

Pierre-Robin sequence and other skeletal abnormalities as described in patient 1 may also be a rare phenotypic presentation of a DICER1 variant. Pierre-Robin sequence is a craniofacial anomaly which includes mandibular hypoplasia, glossoptosis and often cleft palate. There are several genetic causes leading to this phenotype [12]. Interestingly, DICER has been shown to play a role in nucleolar function, and heterozygous pathogenic variants in genes involved in nucleolar homeostasis have been identified to cause various craniofacial disorders [13, 14]. One case of Pierre-Robin sequence associated with DICER1 variant was previously described [15]. In mice, a conditional Dicer1 deletion leads to late embryonic lethality and severe craniofacial dysmorphism, including a secondary cleft palate [16].

Abnormalities in optic nerves as in patient 4 were previously described in patients with DICER1 germline variants [17]. In a mouse model, conditional Dicer1 deletion in the retina led to developmental disorder of retinal cells [18]. Although ocular involvement has to be discussed in this patient in context of the identified ARID1B variant, in Coffin-Siris syndrome ocular findings are more likely to manifest as strabismus, nystagmus, cataract, hypophoria, astigmatism, hypermetropia and anisomyopia [19], with optic nerve hypoplasia being described only occasionally [20].

In summary, germline pathogenic DICER1 variants may not only be associated with the occurrence of certain tumour types, but might also rarely include developmental features, like Pierre-Robin sequence, developmental delay, facial dysmorphisms, and ocular abnormalities. Actually, it is rather surprising that DICER1 pathogenic germline variants do not lead to an even more severe clinical phenotype, since DICER1 is an absolutely central molecule in RNA interference: Dicer catalyses the first step of RNA interference and initiates the formation of the RNA-induced silencing complex (RISC), where argonaute endonuclease, is able to degrade mRNA whose sequence is complementary to the resulting siRNA. Due to this fundamental mechanism, effects in all kinds of cells are conceivable. Nevertheless, we acknowledge that the role of concomitant pathogenic variants in other genes cannot be ruled out, or may indeed modify the phenotype such as in patient 4. The coincidence of developmental phenotypes and pathogeneic DICER1 variant merits further evaluation.