Update on Surveillance Guidelines in Emerging Wilms Tumor Predisposition Syndromes

In the previously published AACR guidelines (1), a single document described the recommended monitoring for patients with overgrowth syndromes and other syndromes associated with Wilms tumor development. Since that publication, a number of new germline pathogenic/likely pathogenic variants (PV) associated with Wilms tumors have been described. For the most part, these PVs were identified through next-generation sequencing cohort studies in patients and families with Wilms tumor. The phenotype of patients with these newly described germline PVs is either subtle or consists only of increased Wilms tumor risk (2, 3). Therefore, in contrast to the companion guideline in this AACR Pediatric Cancer Working Group series (1, 4), these patients will frequently come to an oncologist’s or geneticist’s attention after they or a family member has already developed a Wilms tumor and has had germline genetic testing due to a family history, specific features of their tumor, or as part of routine care for patients with Wilms tumor. The most commonly affected genes in this group (TRIM28, REST, and DIS3L2) may be as prevalent in patients with Wilms tumor as WT1 PVs, and any clinician looking after these children should be aware of them (3). We will also discuss specific situations in which there is no clear predisposing germline PV, but there are other factors that may increase a patient’s risk of developing Wilms tumors: families with multiple affected members but no identified germline PV, children with bilateral or multifocal tumors and their family members, and children with mosaic 11p15 (epi) genetic alterations without phenotypic features of Beckwith–Wiedemann syndrome (BWS). Knowledge continues to evolve rapidly in Wilms tumor predisposition and associated phenotypes, and recommendations made in these guidelines will require ongoing revision over time. However, these recommendations consider the best available current data. As in the companion article, these recommendations are not necessarily harmonized with recommendations from the International Society of Pediatric Oncology (SIOP) because of differences in practice between North America, Europe, and elsewhere (5). In particular, surveillance will often be offered in North America, Australia/New Zealand, and elsewhere when the risk of developing a tumor is greater than 1%. This is a lower threshold than 5%, which is commonly used in Europe (6). Surveillance should be done with renal ultrasonography every 3 months—see the companion article for further discussion of radiologic considerations.

In contrast with BWS and other well-described syndromes, the age range wherein there is significant risk for Wilms tumor is still largely undefined for these syndromes. In the companion article focused largely on BWS and WT1-associated syndromes, surveillance is recommended until 7 years of age as this is the range in which 95% of patients will have developed tumors (4). Using a similar rubric, we will recommend surveillance until 8 years of age as this age range would encompass 95% of tumors for patients with TRIM28 PVs, which have been among the most well described (see “TRIM28” for further discussion). However, clinicians should be aware that the age window for tumor risk in these patients is still being defined and there may be a long tail of risk with some PVs.

Germline PVs in TRIM28, a gene encoding a transcription factor with a role in establishing and maintaining epigenetic transcriptional repression, have been described in patients with Wilms tumors in several reports since the last AACR consensus guidelines (7–9). A large cohort study in the United Kingdom found germline PVs in this gene in 2% of “sporadic” Wilms tumor cases and 8% of families with multiple affected members (3). In all cases, the germline PV is heterozygous, and where examined, complete loss of expression has been found in the tumor through loss of heterozygosity or epigenetic silencing of the second allele (10).

Wilms tumors associated with PVs in TRIM28 are associated with specific tumor histology—86% are epithelial predominant and 45% of all epithelial-predominant tumors harbor TRIM28 PVs (10, 11). Epithelial-predominant tumors, without other poor prognostic features, have a better prognosis than other Wilms tumors and are being investigated as a group that can be treated with reduced-intensity therapy in current clinical trials (NCT04005820). Current evidence suggests that the prognosis in these tumors is not altered by constitutional TRIM28 PVs.

It is unclear from current evidence that whether patients with Wilms tumors and TRIM28 PVs are at increased risk for second tumors after completion of therapy. However, extrapolating from other better characterized syndromes (BWS and WT1-associated syndromes), it is prudent to assume that this risk is higher than that in the general population, and continued surveillance, as recommended for other unaffected carriers, is recommended after completion of treatment. The absolute risk for individuals with TRIM28 PVs who do not have Wilms tumors is still being defined but is likely higher than that in the general population given the presence of families with multiple affected members. Five families have been well described, and in these families, the majority of individuals with PVs developed Wilms tumor. However, this may reflect a selection bias, and so both the penetrance and the gene-specific risk have not yet been defined. Thus, surveillance by renal ultrasound every 3 months is recommended in this population. Based on reported cases, 86% have developed Wilms tumor before 7 years of age and 95% before 8 years of age, with two patients who developed Wilms tumor in their 11th year of life. At this time, we recommend surveillance until 8 years of age (10).

As there is a strong association between epithelial-predominant histology and TRIM28 PVs, we recommend that all patients with this tumor histology have genetic counseling and TRIM28 analysis. However, where available, it is advisable that this gene be tested as part of a panel that includes other genes that predispose to Wilms tumors. We also recommend cascade testing for all first-degree family members of affected probands. Although there is evidence that this variant is predominantly, but not exclusively (12), maternally inherited, these data are still evolving, and so we recommend that all first-degree family members have genetic testing regardless of the proband’s or relative’s gender (9).

Although homozygous PVs in DIS3L2 (coding for an exoribonuclease) are known to cause Perlman syndrome—a congenital overgrowth and Wilms tumor predisposition syndrome—recent evidence demonstrates that heterozygotes are also at an increased risk for Wilms tumors compared with the general population. Hol and colleagues (2) found that 4 of 56 patients with Wilms tumor in the Netherlands investigated with a whole-exome sequencing panel carried a heterozygous germline PV in the DIS3L2 gene (7%). The pattern of Wilms tumor and other phenotypes related to this variant are still under investigation, but emerging data from a multicenter retrospective cohort suggest that the tumors are rarely bilateral and do not have a specific histology (13). In this cohort, 82% of probands had exon 9 deletions. As studies of PVs in this gene have described its frequency in a population that has already developed Wilms tumor and only a small number of families with multiple affected members have been reported (in conference abstracts only), the penetrance is currently undefined. However, given its low frequency in the general population (allele frequency ∼0.05 in gnomAD), most members of the panel felt that the risk is high enough to support surveillance. Most of the tumors described in this population occurred before 7 years of age, although one occurred at 8.3 years (Sophie Van Peer and colleagues—personal communication). These data suggest that extending surveillance until 8 years of age is warranted until further data are reported.

In 2015, REST, a zinc-finger transcription factor with roles in transcriptional and epigenetic regulation during embryonic development, (14) was established as a Wilms tumor predisposition gene (3). Mahamdallie and colleagues identified inactivating REST PVs within four families with familial Wilms tumor. There were nine affected individuals in the families and the PV tracked with Wilms tumor. They also identified a REST PV in 9 of 519 (∼2%) individuals with Wilms tumor and no family history of this cancer (15). Later in 2019, Mahamdallie and colleagues (3) expanded their cohort of familial and nonfamilial Wilms tumor cases, finding additional PVs in REST, and concluded that PVs in REST account for about 2% of unselected patients with Wilms tumor.

Since then, Wilms tumor cohort and case–control studies, as well as a small case series, have identified an additional eight probands, three of whom had familial Wilms tumor, with germline loss-of-function PVs in REST (2, 16–19). Combined with the earlier described patients, the ages at diagnosis ranged from 6 months to 6 years. The majority had unilateral Wilms tumors, with two bilateral Wilms tumors reported, and nephroblastomatosis was identified in at least two patients. Furthermore, in all familial cases, there were multiple unaffected obligate or confirmed carriers of the REST PV, indicating reduced penetrance for Wilms tumors. Given that REST PVs are found in a significant number of individuals who have Wilms tumor with or without a family history of Wilms tumor, we recommend testing for variants in REST as part of panel testing in any patient with Wilms tumor. If a PV is found, we recommend cascade testing and surveillance by renal ultrasound for PV carriers every 3 months until 8 years of age.

Other nonmalignant phenotypes may be associated with PV in REST. The two families described by Salomatina and colleagues (19) also had gingival fibromatosis (GF) tracking with Wilms tumor occurrence and REST PVs. PVs in REST have been described in families with autosomal dominant progressive sensorineural hearing loss (SNHL) with onset in the second decade, GF, and Jones syndrome, a combination of both conditions (20, 21). There seems to be a genotype/phenotype correlation such that most of PVs in REST associated with Wilms tumor are loss-of-function variants occurring in exons 2 and 3. The PV most associated with SNHL is a deep intronic variant affecting splicing (NC_000004.12:g.56927594C>G), and GF is most often associated with frameshift or nonsense variants within exon 5, possibly leading to a dominant negative phenotype, although these genotypes and phenotypes are not mutually exclusive. It is possible that the previously described patients with Wilms tumor had undetected GF or SNHL that would have presented later, after the development of Wilms tumor. We recommend that clinicians be aware of the risk of SNHL or GF in individuals with REST PVs, but there are no sufficient data at this time for specific monitoring recommendations.

In 2014, loss-of-function PVs in CTR9—a key component of the PAF1 complex involved in transcription regulation in embryonic organogenesis—were associated with Wilms tumor development in 3 of 35 families with familial Wilms tumor (22). A fourth family was reported in 2018 (23). Within these families, 9 of 14 individuals who were known or obligate carriers of a CTR9 PV developed Wilms tumor from ages 7 to 39 months, at least one of whom had bilateral disease. All were confirmed or presumed to have inherited the PV from their father, and three of four PVs led to the same in-frame deletion of exon 9, resulting in the deletion of 78 amino acids of the protein. Of note, in the study above, no PVs in CTR9 were detected by sequencing 587 individuals with nonfamilial Wilms tumor (22). We recommend testing for variants in CTR9 as part of panel testing in any patient with Wilms tumor. If a PV is found, we recommend cascade testing and surveillance by renal ultrasound for PV carriers every 3 months until 8 years of age.

As noted above, several recent large sequencing studies have sought to identify novel genes that may predispose patients to the development of Wilms tumors. Within these studies, three additional genes of interest were identified with less frequency than the genes discussed above. These include FBXW7, NYNRIN, and KDM3B (Res Sq rs.3.rs-2675436; refs. 3, 18, 24).

Five patients have been reported with Wilms tumor and PVs in FBXW7, which codes for an F-box protein involved in ubiquitination and normally acts as a tumor suppressor (25). Four of these cases were reported in a cohort study that included 890 patients, suggesting that the incidence of germline FBXW7 variants in patients with Wilms tumor is <1% (3). The fifth patient was identified by Stoltze and colleagues (18) in prospective whole-genome sequencing of patients with Wilms tumor. A separate patient was diagnosed with a Wilms tumor in the setting of both germline FBXW7 and WT1 variants, and in this case, the increased risk was primarily attributed to the better understood WT1-associated predisposition (25). Of these patients, two died from relapsed or refractory Wilms tumor, one died of an osteosarcoma diagnosed in adulthood, and one was being treated for relapsed disease at the time of publication (3, 18). These adverse outcomes are notable as Wilms tumor typically has an overall favorable prognosis. Other renal tumors reported in patients harboring germline PVs in FBXW7 include a rhabdoid tumor of the kidney (3) and renal cell carcinoma (26). Somatic variants in FBXW7 have been identified in multiple cancer types in both adults and children as potential driver mutations (25, 27). Additional nontumor-related clinical manifestations have been shown to be associated with FBXW7 germline variants. Stephenson and colleagues (28) recently reported on FBXW7 neurodevelopmental syndrome that is characterized by varying degrees of developmental delay, intellectual disability, hypotonia, and gastrointestinal issues. Interestingly, in this patient cohort (35 patients; ages 2–44 years), there were no reports of Wilms tumor. It is also notable that of all the cases of Wilms tumor with FBXW7 PVs cited above, only one patient was also noted to have hypotonia, and there were no reports of developmental or intellectual delays. Although none of the variants reported in the FBXW7 neurodevelopmental syndrome were found in patients with Wilms tumor, pathogenic truncating and missense variants across the gene were found in both diseases. As of now, there have not been any families reported in which multiple members have malignancy and FBXW7 variants. Therefore, estimating the true incidence of Wilms tumor in patients that harbor FBXW7 variants is difficult because indications for sequencing this gene are only now emerging clinically. In the absence of clear evidence, we do not make any recommendations about surveillance or cascade testing for patients harboring PVs in this gene at this time. However, institutional and family preferences should be taken into account for each case.

Biallelic truncating PVs in NYNRIN have been reported in multiple patients with Wilms tumor, seen in an autosomal recessive inheritance pattern. Unilateral Wilms tumor has been reported in three patients (all at 2 years of age; ref. 3), and bilateral Wilms tumor has been reported in four patients (ages not reported; Research Square rs.3.rs-2675436-v1). No Wilms tumor was reported in parents harboring a heterozygous NYNRIN variant. There are no reported clinical characteristics that have been associated with biallelic variants in NYNRIN other than Wilms tumor. One patient with unilateral Wilms tumor was reported to also have epilepsy, hypothyroidism, and intellectual disability (3), but there are not enough patients to determine if these characteristics are related to the underlying genetic alterations. Much like the challenge with FBXW7 germline variants, it will be necessary to study additional families that harbor NYNRIN germline PVs to be able to estimate the true risk for Wilms tumor. Similar to FBXW7, we do not make any specific surveillance or cascade testing recommendations for patients diagnosed with homozygous NYNRIN germline variants prior to a Wilms tumor diagnosis.

The pattern of clinical characteristics in individuals with germline PVs in KDM3B is slowly emerging. These individuals all have varying degrees of intellectual disability and short stature. Additionally, many also have characteristic facial features including a wide mouth, pointed chin, long ears and a low nasal columella, feeding difficulties in infancy, skin pigmentation changes, and joint hypermobility (29). The KDM3B protein family functions as histone demethylation proteins, which regulate gene expression (30). Abnormal functioning may lead to the promotion of tumorigenesis via malfunction of tumor-suppressor genes (30). Somatic KDM3B variants have been noted in multiple tumor types (31–33). Given this understanding of gene function, there is some concern that a germline variant may lead to increased risk for tumor development. There have been few patients identified with germline KDM3B variants and pediatric malignancies including bilateral Wilms tumor (diagnosed at 4 years of age), hepatoblastoma (10 years; ref. 15), acute myeloid leukemia (13 years), and Hodgkin lymphoma (14 years; ref. 24). Given the apparent low risk for any specific tumor, no tumor surveillance is recommended at this time for patients with germline KDM3B variants.

Approximately one third of individuals with Wilms tumor have a relative with Wilms tumor (3, 34). In an exome sequencing study that aimed to identify new Wilms tumor predisposition genes, the study cohort included 91 individuals with familial Wilms tumor from 49 families (in which 2 or more individuals had Wilms tumors because of an unknown genetic or epigenetic cause; ref. 3). Sixty-five families were investigated over time by these authors and reported together. Causative constitutional variants were found in 35% of the families, the most common of which were in the REST (8%), TRIM28 (8%), and WT1 (6%) genes. For the two thirds of families without a genetic cause identified, there remains a high likelihood of an underlying genetic predisposition. Notably, this exome study prioritized identification of truncating mutations in tumor-suppressor genes, so other types of coding variation, as well as epigenetic, noncoding, and mosaic variants, all remain as potential causes of the familial Wilms phenotype. To date, the genes associated with the Wilms tumor predisposition locus mapped to 17q12-21 (“FWT1-linked families”) and with other regions identified in prior genome-wide association studies (2p24, 11q14, 5q14, 22q12, and Xp22) have yet to be further characterized (35, 36). Although the penetrance and age distribution of Wilms tumor is difficult to define in familial cases with negative genetic testing, we recommend surveillance until 8 years of age for individuals with two first- or second-degree relatives with a history of Wilms tumor. Clinicians should be aware that penetrance of Wilms tumor risk for many known PVs is still undefined and may be unique to each family. Therefore, decisions about surveillance in these cases should take patient, family, and institutional preferences into account.

Bilateral Wilms tumors have classically been considered to be an archetype of the two-hit hypothesis. Similar to bilateral retinoblastomas, bilateral Wilms tumors occur earlier in life than unilateral disease and were in fact described in Knudson’s follow-up article. However, unlike bilateral retinoblastoma in which the majority of patients have constitutional RB1 PVs (37), <50% of children with bilateral Wilms tumors have identified predisposition syndromes (38, 39). As with children with unilateral Wilms tumors, the most common single predisposition is BWS, reviewed in a companion article (4). Disorders involving WT1 PVs are also more common in children with bilateral Wilms tumors. Data on more recently identified Wilms tumor predispositions are still emerging; however, there is no clear association between more recently described genes (REST, TRIM28, DIS3L2, and CTR9) and bilateral disease (Table 1).

Although most bilateral Wilms tumors are not associated with the previously described predisposition syndromes, emerging evidence suggests that the underlying mechanisms for this condition put these children at risk for secondary tumors. Coorens and colleagues (40) used multiple sampling and deep sequencing to demonstrate that nonneoplastic kidneys in children with bilateral disease often demonstrate clonal nephrogenesis associated with focal increased DNA methylation at the H19 imprinting locus on chromosome 11p15.5. These findings suggest that these kidneys have a “field defect” that puts them at risk for subsequent tumors even after treatment for the initial tumor. This finding is clinically supported by evidence that children with bilateral disease develop second Wilms tumors at a higher rate than those with unilateral disease (41). Therefore, it is likely that these molecular events occur in a mosaic fashion and predispose children to develop multiple Wilms tumors even in the absence of syndromic features or constitutional alterations in known predisposition genes.

There is evidence that suggests that females with bilateral disease are more at risk for subsequent Wilms tumors than males (41). Two additional studies have described a low level of somatic mosaicism for increased DNA methylation at the H19 locus in children with bilateral Wilms tumors, with a higher preponderance of this finding in females. However, there is no current evidence that these mosaic alterations are heritable.

Universal genetic testing and genetic counseling should be offered to all children with bilateral Wilms tumors. This testing should include methylation and copy-number analysis of chromosome 11p15.5, sequencing and copy-number analysis of WT1, and other common predisposition genes—especially REST, TRIM28, DIS3L2, and CTR9. The data presented above support ongoing tumor surveillance (abdominal ultrasounds every 3 months until at least 8 years of age) for any child who has been treated for a bilateral Wilms tumor regardless of whether the child has syndromic features or harbors a variant in a known predisposition gene. At this time, the data do not support the surveillance for relatives of children with bilateral Wilms tumors in the absence of a PV in a known predisposition gene.

Several studies have indicated that children with Wilms tumors may harbor pathogenic alterations at chromosome 11p15.5 without other features of BWS. In 2008, Scott and colleagues (42) measured DNA methylation in blood using methylation-specific multiplex ligation-dependent probe amplification (MS-MLPA) in all children enrolled in clinical trials for Wilms tumors in the United Kingdom. In this setting, 13/437 patients (3%) had 11p15.5 alterations without any reported syndromic features of BWS. This study did not report how many of these patients had bilateral disease, and it did not report long-term outcomes.

More recently, Murphy and colleagues (43) have reported a low level of increased DNA methylation measured by DNA methylation array at the H19 imprinting control region in blood from children with Wilms tumors enrolled in Children’s Oncology Group clinical trials. This increase in DNA methylation was statistically significant but was below the threshold that would be reported in clinical tests (44). Patients with bilateral tumors and females were much more likely to demonstrate this finding. In the studies by both Murphy and Scott, syndromic features were reported independently by each participating site, and patients were not necessarily examined by a clinician with experience in recognizing BWS (42, 43).

Fiala and colleagues (45) similarly demonstrated low levels of increased DNA methylation at the H19 imprinted locus in blood from a small group of patients with Wilms tumor treated at a single center (n = 24). In agreement with Murphy and colleagues (43), these patients were more likely to be female and more likely to have bilateral disease, although some with unilateral disease were reported. All patients in this study were examined by a clinician with expertise in diagnosing BWS, and 62.5% were reported to not have syndromic features.

Finally, Hol and colleagues (2) reported on the clinical and molecular findings of all children diagnosed with Wilms tumor in the Netherlands from 2015 until 2020. All children were examined by a clinician with expertise in diagnosing BWS, and all had molecular testing with a standard clinical MS-MLPA assay. Contrary to prior reports that only 2% of these patients have BWS, this study demonstrated that 16% had at least one clinical feature of BWS in addition to Wilms tumor. Also, 40% of these patients had pathogenic alterations at 11p15 in blood, whereas another 40% had the finding in kidney tissue only. An additional 3.5% of children without BWS features had pathogenic alterations at 11p15, in line with the frequency described by Scott and colleagues (42). The authors did not report whether the upper limit of normal for DNA methylation at the H19 imprinting control region was different in this study than numbers commonly used in North American clinical laboratories.

Less well represented in the literature is the extent to which somatic mosaicism affecting primarily renal tissue predisposes to Wilms tumor. As described above, Coorens and colleagues (40) demonstrated that clonal nephrogenesis is associated with Wilms tumor and renal-specific gain of methylation at the H19 imprinting control region, and other groups have similarly reported renal H19 gain of methylation in children with Wilms tumor and without concordant findings in blood (Research Square rs.3.rs-2675436-v1; ref. 46).

Although several studies have converged to demonstrate that a larger proportion of patients with Wilms tumors than previously thought have molecular and clinical features consistent with BWS, it is unclear whether these children have an increased risk for metachronous Wilms tumors as none of these studies reported long-term outcomes. It is likely that children with clinical features of BWS—especially those with isolated lateralized overgrowth—are in fact on the BWS spectrum (47). Thus, they are likely at risk of metachronous tumors and should undergo the same surveillance as other children with BWS (see accompanying article). However, further data are needed to determine if there is a subset of patients with mosaic pathogenic alterations at chromosome 11p15 without clinical features who require surveillance. Given the centrality of a clinical BWS diagnosis to the decision for surveillance, we recommend that every child with Wilms tumor should be examined by a clinician with expertise in recognizing the BWS clinical spectrum. At this time, we do not recommend surveillance for children without clinical features of BWS. We also advocate for additional research delineating both careful clinical phenotyping of patients with Wilms tumor to evaluate for subtle features of BWS and the other syndromes discussed here.

A.S. Doria reports grants from Terry Fox Foundation, PSI Foundation, Society of Pediatric Radiology, Radiological Society of North America, CARRA Foundation, and Novo Nordisk Access to Insight Grant outside the submitted work. J.R. Hansford reports other support from Bayer Pharmaceuticals, Alexion Pharmaceuticals, and Servier Pharmaceuticals outside the submitted work. K.W. Pajtler reports grants from Federal Ministry of Education and Research (01GM2205A) and German Childhood Cancer Foundation (DKS2021.02) during the conduct of the study. J.M. Kalish reports grants from Alex’s Lemonade Stand Foundation, Damon Runyon Cancer Research Foundation, and Rally Foundation for Childhood Cancer Research and other support from Lorenzo “Turtle” Sartini Jr Endowed Chair in Beckwith-Wiedemann Syndrome Research and the Victoria Fertitta Fund through the Lorenzo “Turtle” Sartini Jr Endowed Chair in Beckwith-Wiedemann Syndrome Research during the conduct of the study, as well as grants from St. Baldrick’s Foundation outside the submitted work. No disclosures were reported by the other authors.