The assertion that only maternal inheritance of TRIM28 variants predisposes to Wilms tumour arises from the observation that in one of the originally described series, 10/10 inherited (as opposed to de novo) variants were also present in the mother.(5) Such parent of origin effects in tumour predisposition syndromes have been described previously, notably for SDHD, SDHAF2 and MAX pathogenic variants, which are associated with PPGL when paternally inherited.(11, 12)
Perhaps the most intuitive molecular explanation for parent of origin effects is that the allele from one parent is inactive in the normal state due to an imprint established during gametogenesis. In that scenario, inheritance of a pathogenic loss of function variant from that parent would not cause additional aberration of gene function but inheritance from the other parent would lead to two non-functional alleles. However, to date no evidence has been reported that TRIM28 is an imprinted gene.(13, 14).
An alternative explanation of parent-of-origin effects associated with a non-imprinted cancer predisposition gene is that proposed for PPGL associated with pathogenic variants in SDHD and SDHAF2 which both map to chromosome 11 and predispose to PPGL when pathogenic variants are paternally inherited. In these cases, PPGL usually show loss of the maternally inherited chromosome 11 in the tumour tissue. This results in loss of the maternally expressed imprinted gene CDKN1C whilst the function of the paternally expressed growth factor IGF2 is unaffected (15). In contrast when a germline SDHD/SDHAF2 variant is maternally inherited, though somatic loss of the paternal chromosome 11 would be associated with biallelic SDHD inactivation, there would also loss of the functioning paternal IGF2 allele but the functional CDKN1C allele would be unaffected. Based on studies of PPGL associated with pathogenic variants in SDHD, it would be predicted that TRIM28 tumourigenesis would be favoured by somatic events that resulted in biallelic TRIM28 inactivation accompanied by loss of paternally expressed imprinted genes (and/or preservation of maternally expressed imprinted genes) mapping to chromosome 19. Penetrance from paternally inherited variants under this model would be due to mitotic recombination events involving this region. Whilst poor DNA quality precluded investigation of FFPE- tumour material in the present case, other groups have found that in Wilms tumours associated with constitutional or somatic TRIM28 pathogenic variants, loss of heterozygosity (LOH) can involve imprinted genes on chromosome 19, albeit without complete consistency in terms of included genes. Halliday et al reported LOH in a Wilms tumour from an individual with a constitutional TRIM28 frameshift variant where a distal q13.43 region of homozygosity included eight genes,(6) none of which are firmly considered as imprinted.(13). Armstrong et al performed copy number analysis in five Wilms tumours with somatic TRIM28 mutations, four of which showed copy number neutral LOH at 19q13.32 to 9q13.43(7) that includes TRIM28 along with a number of reported paternally expressed imprinted genes (ZIM2, PEG3, MIMT1, MIR371A).(13, 14) If the Halliday and Armstrong studies are taken together, no known imprinted genes are within the region of LOH in all tumours although some affected in multiple samples have greater potential relevance such as PEG3, a paternally expressed growth promoter that was hypothesised as of potential relevance by Mahamdallie et al.(5) It is a regulator of the tumour necrosis factor immune response and decreased expression has been associated with tumourigenesis.(16)
This report of Wilms tumour in the context of a paternally inherited TRIM28 pathogenic variant demonstrates that this mode of inheritance is possible despite pre-existing evidence of a parent of origin effect where penetrance would only result from maternal inheritance. The mechanistic basis of this observation is unclear but more frequent paired WGS analysis of Wilms tumours in clinical practice may beget the definition of common regions of LOH and mitotic recombination events. Whilst identification of a paternally inherited TRIM28 variant in a child appears to confer a lower risk, we suggest that clinicians still consider surveillance for Wilms tumour, particularly if it has been penetrant in a sibling. The question of surveillance is likely to arise with increasing frequency in the UK given that TRIM28 is one of only five cancer predisposition genes included in the Genomics England Newborn Genomes Programme.(17)