The Principles of Clinical Cytogenetics - Extra Materials - Springer

Genomic Imprinting and Uniparental Disomy 521
Wilms Tumor/Rhabdomyosarcoma
In a number of embryonal tumors, loss of heterozygosity (LOH) of a specific parental allele has
been observed. In all cases studied, the maternal allele is preferentially lost. This suggests that duplication
of some paternal alleles results in enhanced cell proliferation, whereas duplication of certain
maternal alleles can inhibit cell proliferation.
In Wilms tumor and rhabdomyosarcoma, LOH involves chromosome 11 (129–131). LOH does
not involve markers for 11p13, the proposed Wilms tumor locus, but only markers on 11p15.5
(130). Known imprinted genes in the 11p15.5 region include H19, IGF2, and CDKNIC (p57KIP2 )
(see above). The expression of CDKNIC is reduced in Wilms tumor (118). In addition, by using
several overlapping subchromosomal transferable fragments from 11p15 distinct from H19 and
IGF2, Koi et al. (132) were able to obtain in vitro growth arrest of rhabdomyosarcoma cells. These
observations suggest that CDKNIC, which is normally active on the maternal allele only, might be
a candidate for a tumor suppressor gene. Loss of the active CDKNIC allele on the maternal chromosome
results in tumor development. In addition to LOH, another possible mechanism, loss of
imprinting LOI (see the subsection on BWS), has been proposed. Ogawa et al. (133) reported
biallelic IGF2 RNA synthesis in four of 30 Wilms tumors they studied. Thus, “relaxation” of IGF2
gene imprinting on the maternal allele has occurred, resulting in its expression. This would be
equivalent to having two copies of an active IGF2 gene, as would occur with a paternal duplication
or with paternal UPD. A similar biallelic expression of IGF2 was reported in 30% of breast cancer
patients studied (134). Disruption of the imprinting mechanism (i.e., LOI), might therefore also
play a role in tumorigenesis. A third possible mechanism has also been proposed in a proportion of
Wilms tumor patients. In some patients, LOI was observed in both the Wilms tumor tissue and the
normal adjacent kidney tissue, but IGF2 expression was significantly higher in tumor tissue. The
overexpression in tumor tissue was accompanied by activation of all four IGF2 promoters (135).
These studies indicate that although genomic imprinting plays an important role in tumorigenesis,
a single mechanism does not account for all cases.
Retinoblastoma/Osteosarcoma
In retinoblastoma and osteosarcoma, loss of both functional copies of the retinoblastoma gene
(RB) on chromosome 13 at band q14 has been observed (136). In familial cases, a mutation in
one of the alleles is present in the germline. De novo mutations in the germline occur preferentially
in the paternal chromosome (137,138), consistent with the general observation that new
germline mutations arise predominantly during spermatogenesis. In sporadic, nonfamilial tumors,
loss of function of both alleles occurs somatically. In sporadic osteosarcomas, the initial mutation
occurs preferentially on the paternal chromosome 13 (139), suggesting that genomic imprinting
might be involved. Data are less clear in sporadic retinoblastomas. No predilection in the parental
origin of the somatic allele loss was noted in some studies (138,140), but a preferential loss of the
maternal allele, which implies a preferential initial somatic mutation on the paternal allele, was
reported in one study (141). Thus, the role of genomic imprinting in retinoblastoma is unclear at
this time.
Neuroblastoma
In neuroblastoma, deletions of chromosome 1p and amplification of the MYCN gene on chromosome
2p are frequently seen (142). Preferential amplification of the paternal MYCN allele in
neuroblastoma tumor tissues has been reported (143). In tumors with MYCN amplification, loss of
parental 1p alleles was found to be random (143,144). In tumors without MYCN amplification, loss
of 1p was previously reported to be preferentially maternal (16 of 17 cases) (144), but random in a
more recent study (145). Thus, there is no clear evidence that the putative tumor suppressor gene
mapped at 1p36.2-36.3 is imprinted, and the role of imprinting in neuroblastoma is not clear at the
present time.