Sarcomas and Cancer Predisposition Syndromes

An ESUN Article by Abha Gupta, MD and David Malkin, MD

Introduction

Genetic susceptibility to cancer has been the subject of considerable interest in defining the etiology and natural history of cancer, and also for providing guidance and intervention for prevention or early detection strategies for affected families. Practicing oncologists continue to place high importance on enquiring about a family history of cancer and to explore this history not only at the time of diagnosis, but also throughout the patient’s active and long-term follow-up care. In this way, any evolving cancers in the patient or their family members can be documented. The purpose of this review is to describe common genetic or heritable conditions associated with sarcomas, in order to remind families and health care providers to consider each patient in such context.

In its broadest terms, cancer is initiated by a genetic mutation in a particular cell which then is transmitted to each daughter cell forming the cancer. A somatic gene mutation is one which is acquired in a post-meiotic cell division and is restricted to the cancer cell, whereas a germline mutation is one which is found in all cells of the host organism. When clusters of cancers occur in families in a reproducible pattern, these families are considered to define a familial cancer predisposition syndrome. While all cancer predisposition syndromes are associated with early age of onset of tumors when compared with their sporadic counterparts, some appear to confer an increased risk of predominantly adult-onset cancers (e.g. familial breast-ovarian cancer or hereditary non-polyposis colorectal cancer), whereas others predispose to cancers primarily of childhood (e.g. hereditary retinoblastoma or Beckwith-Wiedemann syndrome) or early-onset cancers of both children and adults (e.g. von Hippel-Lindau disease, multiple endocrine neoplasia, or Li-Fraumeni syndrome).

In adolescents and young adults between the ages of 15-29 years of age, the incidence of bone and soft-tissue sarcomas (including Kaposi’s sarcoma) represents 11% of all cancers.¹ Soft-tissue sarcomas (STS) represent about 8% of all malignant tumors of children, with rhabdomyosarcoma (RMS) accounting for about half of the cases.² The most common bone tumors include osteosarcoma and Ewing sarcoma. Non-random somatic molecular or chromosomal alterations are commonly observed in a wide variety of sarcomas, such as c-kit mutations in gastrointestinal stromal tumors, PAX3/7-FKHR translocations in alveolar rhabdomyosarcoma or EWS-FLI translocations in Ewing sarcoma. In addition, while most sarcoma patients do not have a striking family history of cancer, sarcomas are common manifestations of cancer predisposition syndromes, which are in turn associated with well-defined heritable or germline genetic abnormalities. The majority of sarcomas occur sporadically, however, a significant minority of children with either soft-tissue or bone sarcomas are identified as having a genetic predisposition to malignancy.³ The frequency likely represents an underestimation, because only recently have oncologists recognized the power of family history and attention to genetic manifestations of cancer causation. As new genes and novel constellations of tumor clusters are discovered, more genotype:phenotype correlations in sarcoma patients will likely be found.

Most cancer predisposition syndromes associated with the development of sarcoma manifest in childhood. In some instances, such as Li-Fraumeni syndrome, malignancy is the defining phenotype of the syndrome. In other situations, the increased risk of malignancy is one of many features of the syndrome, which may also be characterized by other congenital anomalies. In this review, we describe phenotypic characteristics and heritable genetic changes associated with cancer predisposition syndromes in which mutant gene carriers are at an increased risk to develop sarcoma.

Heritable Conditions Defined by an Increased Risk of Cancer

1. Retinoblastoma

One of the most striking genetic associations for sarcomas is that of mutation in the retinoblastoma (RB) gene, RB1, and osteosarcoma. The RB1 tumor suppressor gene was the first inherited cancer susceptibility gene identified in humans.⁴ In a longitudinal survey of 693 cases of bilateral RB, 15% developed second primary tumors, most of which were osteosarcoma (OS).⁵ One third of these tumors occurred outside the radiation field for the RB. Even among 113 patients who did not receive radiation from this cohort, there were 10 cases of second neoplasms (2 OS, 1 soft-tissue sarcoma) for an excess risk of 3.1%.⁵ The cumulative incidence of secondary tumors occurring within a previously radiated field in heritable RB patients has been reported to be greater than 50%, and is dependent on dose of radiation.⁶ In addition to the risk for OS, a more recent study highlights that survivors of hereditary RB are also at an increased risk of developing soft-tissue sarcomas within the radiation field, especially leiomyosarcoma.⁷

Why Osteosarcoma in Retinoblastoma Patients?

The tissue-specific development of cancers, especially OS in RB patients, suggests that the RB1 gene may be involved in the pathogenesis of both RB and OS, and that there is a specific function of RB1 protein in bone development. The landmark study of Hansen, et al. was the first to demonstrate that loss of constitutional heterozygosity (LOH) for the RB1 locus on chromosome 13 was present in OS from patients with and without heritable RB.⁸ RB1 plays a role in the inhibition of proteins important for proliferation of osteoblasts and stimulates proteins important for their terminal differentiation (Figure 1).⁹ Outside the context of RB, loss of heterozygosity of the RB1 gene is the most frequent genetic change in OS, and is observed in almost half of sporadic tumours. In multivariate analysis, some reports suggest that LOH of RB1 may be associated with an unfavorable outcome in OS.¹⁰

2. TP53 and Li-Fraumeni syndrome

In 1969, Drs. Frederick Li and Joseph Fraumeni, Jr. identified four families with an increased susceptibility to cancer in which at least two cases of rhabdomyosarcoma (RMS) occurred in infancy.¹¹ A twenty-year prospective evaluation of these families, together with 20 others, led to a more comprehensive definition of what is now termed Li-Fraumeni syndrome (LFS).¹² LFS is an autosomal dominant disorder characterized by diagnosis of a bone or soft tissue sarcoma at an early age (< 45 years) in an individual whose family exhibits a diverse array of other cancers including breast cancers, brain tumors and adrenocortical carcinomas as multiple primary tumors in multiple members in one ancestral line. The probability of developing cancer approaches 40% by age 20, and >90% by age 70; greater than 50% of these patients will develop any type of second tumor. In 1990, germline mutations in the TP53 tumor suppressor gene were identified as the molecular event responsible for cancer predisposition in the majority of LFS families.¹³ This initial observation has been substantiated in numerous subsequent studies suggesting that 60-85% of classic LFS families and approximately 10% of LFS-like families (who are characterized by a less restrictive cancer phenotype than classic LFS) harbor detectable germline TP53 mutations, most commonly missense mutations.^14-17 Mutant TP53 gene carriers have an extremely high lifetime risk of cancer with 75% of males and 100% of females developing cancer by age 60. Various other genetic and epigenetic factors are under investigation to determine their role in modifying the genetic penetrance of the underlying TP53 alteration, which may explain the variability in age of onset and specific tumor type within and between different LFS families.

The prevalence of germline TP53 mutations in children with sarcoma has been examined in several studies. Among 151 families of children with childhood-onset sarcoma, five were identified as having classic LFS pedigrees.¹⁸ Another study screened 235 children with OS from 33 institutions, and found that approximately 3% carry constitutional germline TP53 mutations.¹⁹ In children with RMS, the prevalence of germline TP53 mutations is greater in children less than 3 years of age at diagnosis.²⁰ Most TP53 sequencing studies that defined population-specific frequencies predate the use of high throughput automated sequencing technology and failed to span the entire coding (and intervening non-coding) regions of the gene, therefore likely underestimating the true mutation frequency. Even so, based on these results, it has been proposed that, regardless of family cancer history, children with very early onset RMS or OS should be considered candidates for TP53 testing. This will serve to identify risk of secondary cancer development as well as potentially define the risk for as yet unaffected family members. There are very few reports describing the prevalence of germline TP53 mutations in adults with sarcoma (outside the context of an LFS family history), and although they likely occur in a small percentage of patients (perhaps 1-2%), further comprehensive studies are required to better define this population frequency.²¹

Acquired TP53 alterations are detectable in 25-42% of OS²² and in 20% of non-osteosarcoma bone and soft tissue sarcomas.²³ While TP53 mutations usually encode for altered protein structure, as a result of missense base alterations in the gene, mutations in OS and other bone sarcomas more commonly lead to protein truncation and absent protein expression.²⁴ The mechanisms for the unique nature of TP53 dysfunction in bone tumors is not well understood, and interestingly, TP53 mutation status does not appear to correlate to outcome in patients with OS.²²

In addition to the occurrence of sarcomas in the setting of LFS, there appears to be another relationship between breast cancer and sarcoma, which may or may not be TP53-dependent. For example, an excess of breast cancer has been observed in the mothers of children and adolescents with sarcomas.^25,26 When ascertained from the adult perspective, a similar pattern is observed. Bennett, et al. examined 402 women with breast cancer and reported a significant increase in soft tissue and bone sarcoma in their first degree relatives.²⁷ Similarly, another study observed an increased risk of early-onset OS among children of mothers with breast cancer or melanoma.²⁸ However, in one series, only 1 of 21 families with breast cancer and sarcoma were found to have TP53 germline mutations.²⁹ Thus other genes, including BRCA2 (associated with early onset breast-ovarian cancer) may also explain familial aggregations of breast cancer and sarcoma, and require further investigation.³⁰

3. Desmoid Tumors and Familial Adenomatous Polyposis

Familial adenomatous polyposis (FAP) is an autosomal dominantly inherited colon cancer predisposition syndrome. It is caused by germline mutations in the adenomatous polyposis coli (APC) tumor suppressor gene.^31-34 The prevalence of FAP is estimated at 1 in 5000–10,000, and although many APC mutations have been identified, only 50% of patients with the FAP phenotype have mutations in APC.^35,36

One of the most common extra-colonic manifestations of FAP is aggressive fibromatosis (AF). AF, also termed desmoid tumor, is a benign, locally invasive soft-tissue lesion composed of a monoclonal proliferation of spindle (fibroblast-like) cells.³⁷ Ten–twenty percent of FAP patients develop desmoid tumors, a frequency approximately 850 times that of the general population.³⁸ Although not truly malignant, desmoid tumours are a major cause of morbidity and mortality in FAP patients.³⁹ The occurrence of desmoids in FAP are more common after prophylactic colectomy, and although they arise following occurrence of biallelic APC mutations, they are independent of the specific type of germline APC mutation.⁴⁰ Thus, the exact risk for the development of desmoids in FAP is not defined, and likely other genetic factors play a role in determining the risk of desmoid tumor formation.

Inactivating mutations of the APC gene lead to stabilization of the ß-catenin protein which subsequently disrupts transcriptional regulation of a complex set of genes that contribute to mesenchyme cell differentiation/proliferation⁴¹ and ultimately lead to promotion of these tumors.⁴² Interestingly, sporadic cases of aggressive fibromatosis also harbour mutations in APC or more commonly, in the ß-catenin gene.⁴³

Syndromes Associated with Cancer Predisposition

RecQ family DNA helicases are enzymes that unwind DNA and are thus important to maintain overall genomic integrity.⁴⁴ Constitutional mutations in the genes encoding the RecQ enzymes BLM, WRN, or RTS give rise to the hereditary disorders Bloom Syndrome (BS), Werner Syndrome (WS) and Rothmund-Thompson Syndrome (RTS), respectively, all of which are associated with a predisposition to cancer. The spectrum of malignancies associated with each of the RecQ deficiency syndromes is variable, but the non-epithelial cancers in WS and RTS are dominated by sarcomas, especially OS.⁴⁵

1. Genomic Instability: Rothmund-Thomspon Syndrome

RTS is a rare autosomal recessive disorder characterized by a variety of features, including poikiloderma in early childhood (upon which the diagnosis is often made), skeletal dysplasias, small stature, sparse head and facial hair, juvenile cataracts, gastrointestinal disturbances and an increased predisposition to osteosarcoma.⁴⁶ Seven percent (17/260) of RTS patients reported in the English language literature have developed OS, but a more contemporary case-series of 41 patients reported 13 (32%) with OS.⁴⁷ One landmark study described the risk of OS in RTS based on mutations in the RECQL4 gene.⁴⁸ In this study, 23/33 patients with RTS examined had truncating mutations in RECQL4, including all 11 patients with OS. OS did not develop in patients who lacked truncating mutations. The clinical features and response to therapy of OS in RTS patients is similar to that in patients who have sporadic OS.⁴⁹ The frequency and role of RECQL4 mutations in sporadic OS is not known, although it is the focus of intense investigation.

2. The Ras Family

Ras genes encode small guanosine triphosphatases (GTPases) that include H-RAS, N-RAS and K-RAS.⁵⁰ The RAS proteins are switches that connect receptor cell signaling to downstream nuclear pathways which generally mediate cell proliferation and differentiation. Activating ras mutations that encode ‘oncogenic ras’ are found in >30% of malignancies, predominantly epithelial cancers and melanoma, and have now been implicated in developmental disorders (see Table below).

Cancer Predisposition Syndromes Associated with Childhood

	Syndrome	Gene	Sarcoma	Other Cancers	Non-Cancer Features
1.	Retinoblastoma	RB1	OS	Breast
2.	Li-Fraumeni	TP53	OS, RMS Non-RMS Soft Tissue Sarcoma	Breast, Brain, ACC Leukemia, CPC, Multiple others
3.	Familial Adenomatous Polyposis	APC	Desmoids	Colon, Hepatoblastoma	Osteoid osteomas
4.	Rothmund-Thompson	RECQLA	OS	Skin	Poilioderma, juvenile cataracts, short stature, sparse hair, skeletal dysplasias
5.	Neurofibromatosis	NF1	MPNST	Optic gliomas, Neurofibromas	Learning disabilities, Lisch nodules, axillary freckling, café au lait spots
6.	Costello	HRAS	RMS	Cutaneous papillomas, Neuroblastoma, Bladder carcinoma	Dysmorphic craniofacial features, cardiac defects, failure to thrive, musculoskeletal anomalies, developmental delay
7.	Beckwith-Wiedemann	CDKN1C/ NSD1 p57Kip	RMS	Hepatoblastoma, Wilms	Macrosomia, macroglossia abdominal wall defect, hemihyperplasia, ear anomalies, visceromegaly, renal abnormalities, neonatal hypoglycemia
8.	HLRCC	FH	LMS, Leiomyomas	Renal Cell Carcinoma

a) Neurofibromatosis Type 1

Von Recklinghausen's neurofibromatosis (NF type 1), one of the most common genetic disorders, with an incidence of 1 in 3000, is caused by mutations in the NF1 tumor suppressor gene, neurofibromin, a modulator of the ras pathway.⁵¹ The disease is characterized by learning disabilities, multiple café-au-lait spots, axillary or inguinal freckling, neurofibromas, Lisch nodules (iris hamartomas), and distinctive bony lesions.⁵² The best described association between NF1 and cancer is the increased risk of central nervous system tumors (i.e., optic nerve pathway pilocytic astrocytomas)⁵¹ and STS, particularly malignant peripheral nerve sheath tumors (MPNST).⁵³

MPNST is a spindle cell sarcoma that arises in proximity to peripheral nerves, or shows nerve sheath differentiation. Although they are one of the most common non-rhabdomyosarcoma soft tissue sarcomas in children, >80% of these tumors occur in adults.^54-56 The proportion of all MPNST that arise specifically in NF1 patients ranges from 17-67%,^57-59 with a lifetime risk of MPNST in NFI estimated to be between 8-13%.⁶⁰

Neurofibromatosis type 1 itself has been shown to be a poor prognostic factor in the overall survival from MPNST.^57,61 However, another series demonstrated that although NF1 patients tended to present with larger tumors, there was no impact of the underlying condition on overall survival.⁶² Among the non-neurogenic sarcomas, RMS and gastrointestinal stromal tumours (GIST) are worthy of mention. RMS is encountered in NF1 patients at a greater frequency than in the general population. A recent review of cases from the Intergroup Rhabdomyosarcoma Study Group (IRSG) reported five cases of RMS patients with NF1 out of the 1,025 cases enrolled in IRS-IV.⁶³ To date, no correlations between specific genetic mutation subtypes in the NF1 gene and the particular spectrum of tumors that present have been identified.

GISTs are becoming increasingly recognized in association with NF1. Although activating mutations in KIT and PDGRF are the responsible oncogenic force in most sporadic GISTs, the NF1-related GISTs do not have KIT or PDGFRA mutations,⁶⁴ and instead, are driven by alternate mechanisms such as somatic inactivation of the wild-type NF1 allele in the tumor.⁶⁵ Similarly, GISTs are part of Carney’s triad, a syndrome of unknown etiology comprising pulmonary chondroma and paraganglioma. GISTs in patients with Carney’s triad also have wild type KIT and PDGRFA.⁶⁶

b) Costello Syndrome

Costello Syndrome (CS) is a rare syndrome which is characterized by multiple congenital anomalies including dysmorphic craniofacial features, cardiac defects, ectodermal and musculoskeletal anomalies, failure to thrive and developmental delay.⁶⁷ Recently, activating germline mutations in H-RAS have been identified in the majority of patients with CS,⁶⁸ and in all CS with malignancy.⁶⁹ Although not historically considered a cancer syndrome per se, over half of CS patients will develop benign cutaneous papillomas and up to 20% will develop malignancies with the most common being embryonal RMS.⁷⁰ Since the first report of rhabdomyosarcoma in a patient with CS in 1998,⁷¹ a total of 100 cases of CS have been identified with 10 cases of RMS.⁷² As with many of the other CPS, the diagnosis of cancer often precedes the diagnosis of CS. The frequency of RMS in CS has been thought to be significant enough to warrant clinical surveillance screening similar to that in Beckwith-Wiedemann syndrome.⁷² However, because germline H-RAS mutations are rarely seen in sporadic RMS, it is not likely that routine genetic testing in these patients would be fruitful. Furthermore, other molecular events presumably determine the precise tumor phenotype in CS and play a role in tumorigenesis in this syndrome.⁷³

3. Beckwith-Wiedemann Syndrome

Beckwith-Wiedemann Syndrome (BWS) was originally defined by the presence of macrosomia, macroglossia and abdominal wall defects (omphalocele, umbilical hernia, diastasis recti). However, the penetrance of clinical features of BWS is variable and the diagnosis can be established if at least three diagnostic findings are present of those included above in addition to hemihyperplasia, embryonal tumors, adrenocortical cytomegaly, ear anomalies, visceromegaly, renal abnormalities, neonatal hypoglycemia, or a positive family history.⁷⁴ The molecular basis for BWS is complex and involves autosomal dominant inheritance, uniparental disomy, and genetic imprinting of the chromosome 11p15 locus.⁷⁵ The overall risk for tumor development in children with BWS is approximately 7.5% (Wiedemann, 1983), but there are many factors which influence this. Children with BWS have an increased risk for embryonal tumors within the first 5–8 years of age (Wiedemann, 1983). Tumors most frequently found in BWS are Wilms tumor (WT), adrenocortical carcinoma (ACC), rhabdomyosarcoma (RMS), and hepatoblastoma (HB). This specific profile of histologies likely reflects a common genetic pathway in the development of BWS.⁷⁶

Children with BWS who develop embryonal tumors such as rhabdomyosarcoma and hepatoblastoma are more likely to have epigenetic changes in domain 2 of 11p15,⁷⁷ whereas Wilms tumor is more strongly associated with epigenetic alterations in domain 1 of the locus or uniparental disomy.⁷⁸ Genomic imprinting is a form of inheritance where the ‘imprinted’ genes are expressed uniquely from one allele, representing each respective parent. The mis-regulation of imprinted gene expression is termed loss of imprinting (LOI) and has been found in the IGF2 gene in both embryonal and alveolar RMS in 11/14 cases.^79,80 RMS has often been cited as a tumor of risk in BWS, although only 7 cases of RMS in BWS have been reported in the literature.⁸¹ The genetic basis of BWS is tightly linked to that of embryonal type RMS (ERMS), in that loss of heterozygosity of 11p15 is commonly found (72%) in sporadic ERMS, and this locus harbours a constellation of targets for BWS candidate genes.⁸²

Absolute risk of tumor development ranges from 3-43% in BWS patients depending on the specific genes involved in loss of imprinting.⁸³ Due to the overall increased risk of malignancy, especially for Wilms’ tumor and hepatoblastoma, it is now considered standard practice to perform regular screening of children with abdominal ultrasound and serum alpha-fetoprotein every 3 months until their ninth birthday.^84,85

4. Uterine Leiomyomatosis and Renal Papillary Cell Carcinoma

Novel hereditary cancer syndromes are still being identified. Uterine leiomyomas (fibroids) are benign clonal tumors arising from the smooth muscle of the uterine wall. They are extremely common, occurring in perhaps more than 50% of women, and are a source of significant morbidity.⁸⁶ Due to this significant burden, investigators sought to identify its molecular basis and identified autosomal dominant inheritance for uterine leiomyomatosis and renal papillary cell carcinoma (HLRCC) localized to chromosome 1.⁸⁷ The gene is now known to be the tumor suppressor, and it encodes fumarate hydratase (FH), an enzyme in the tricarboxylic acid cycle.⁸⁸ Cancer cells upregulate glucose metabolism, the Warburg effect, which is actually the basis for the widespread application of positron emission tomography in which a glucose analog tracer (2-18fluoro-2-deoxy-D-glucose) is used to differentiate between normal and tumor tissue. FH represents one mitochondrial enzyme of the TCA which mediates glucose metabolism of the cell. The role of mitochondrial enzymes in tumorigenesis is thought to be related to the effect on homeostasis of hypoxia-inducible factor, HIF, and subsequently, apoptosis.⁸⁹

HLRCC is considered a cancer predisposition syndrome due to the increased standardized incidence ratio of 6.1 for RCC and 71 for uterine leiomyosarcoma, the malignant counterpart for benign fibroids.⁹⁰ However, the prevalence of FH mutations in sporadic LMS is rare,⁹¹, even in patients with younger age of onset (<45 years) of the tumor.⁹² The prevalence of germline FH mutations in sporadic RCC patients has not been explored.

Screening in Cancer Predisposition Syndromes

The topic of screening for the identification of cancer predisposition is one shrouded in an enormity of literature and debate, the scope of which is beyond this review. Genetic testing can offer information for cancer risk for the patient, but also for thus-far healthy family members, both children and adults. With respect to sarcomas however, specific issues must be considered:

1. Sarcomas can occur in any anatomic location, so unlike for breast, ovarian or colon cancer, there is no role for risk-reductive surgery.
2. Sarcomas secondary to genetic predisposition can manifest in childhood; thus, testing of patients or their family members would often occur at an age when parents act as surrogate decision makers.

The American Society of Clinical Oncology agrees that the ‘scope of parental authority encompasses the right to decide for or against testing’ children, but parents should be made aware of the strong arguments against testing before making their decisions.⁹³ Nevertheless, because the cancer risk in some sarcoma-associated CPS is particularly high in children (i.e. BWS and LFS), genetic testing and associated clinical surveillance for early cancer detection are considered appropriate. The topic of screening for cancer predisposition syndromes specifically in children has been reviewed, and the approach to testing still depends on the particular ethical considerations of specific institutions and cancer genetics teams.⁹⁴

Prenatal testing for LFS has been reported to lower psychosocial stress, and introduces the option of early screening programs; current studies are in progress to evaluate whether such programs alter clinical outcome.⁹⁵ Pre-implantation genetic testing for cancer predisposition has been reported for a number of conditions, and continues to raise important ethical and legal concerns.⁹⁶ Based on experience with early detection surveillance in other CPS, including BWS, multiple endocrine neoplasia and von Hipppel Lindau disease, it is possible that as more effective screening programs are developed, early detection of smaller tumors will be demonstrated to reduce long-term morbidity and mortality for these high-risk patients. The use of FDG-PET imaging for detection of tumors in LFS has been reported, although its utility as a broad screening tool bears further evaluation.⁹⁷

As the discussions of the appropriate role of genetic testing for cancer predisposition continue, it should be offered in the context of pre-and post-test counseling provided by trained experts. Families should be made aware of all the benefits and risks to having genetic testing performed and extra care should be taken when testing of children is involved.

As the field of cancer genetics continues to evolve, it is also incumbent on practicing oncologists to continue to be cognizant of family cancer history and to explore this history with patients not only at the time of diagnosis, but also throughout their active and long-term care to ensure that evolving cancers in other family members are documented. Frequently updated pedigrees may make it possible to recognize patient and family members with potential risk status and consider that the apparent ‘sporadic’ occurrence of sarcoma might actually represent a more complex cancer predisposition syndrome family. Studies of these rare families have improved our understanding of the fundamental genetic basis of human cancer. Although to date, there is no direct relevance of learning about cancer predisposition syndromes to the development of new targeted therapies, novel familial clusters, as well as new ‘cancer genes’, are sure to be discovered with continued vigilant observation of this intriguing set of patients.