Pediatric Cancer Predisposition and Surveillance: An Overview, and a Tribute

The prevalence of childhood cancer attributable to genetic predisposition was generally considered very low. However, recent reports suggest that at least 10%of pediatric cancer patients harbor a germline mutation in a cancer predisposition gene. Although some of these children will have a family history suggestive of a cancer predisposition syndrome, many others will not. Evidence from recent pediatric studies suggests that surveillance and early detection of cancer in individuals carrying a germline cancer predisposing mutation may result in improved outcomes. However, there is a lack of consistency in the design of cancer surveillance regimens across centers both nationally and internationally. To standardize approaches, the Pediatric Cancer Working Group of the American Association for Cancer Research (AACR) convened aworkshop, during which consensus screening recommendations for children with the most common cancer predisposition syndromes were developed. In general, we considered a 5% or greater chance of developing a childhood cancer tobe a reasonable threshold to recommend screening. Conditions for which the cancer risk was between 1% to 5% were addressed individually. In a series of manuscripts accompanying this article, we provide recommendations for surveillance, focusing on when to initiate and/or discontinue specific screening measures, which modalities to use, and the frequency of screening. Points of controversy are also reviewed. We present the outcome of our deliberations on consensus screening recommendations for specific disorders in 18 position articles as Open Access publications, which are freely available on an AACR-managed website. Clin Cancer Res; 23(11); e1–e5. 2017 AACR. See all articles in the online-only CCR Pediatric Oncology Series. Tribute to Alfred G. Knudson Jr Dr. AlfredG. Knudson is considered bymany to be the father of modern cancer genetics and heritable predisposition. Al died on July 10, 2016, as this workshop was still in its planning stages. It was the unanimous consensus of the workshop participants to dedicate our efforts to Al Knudson and his pioneering work in this field. In his initial statistical analysis of hereditary and nonhereditary cancer, he hypothesized that "retinoblastoma is a cancer caused by two mutational events. In the dominantly inherited form, one mutation is inherited via the germinal cells and the second occurs in somatic cells. In the nonhereditary form, both mutations occur in somatic cells" (1). This "two-hit theory" of the genetic origin of retinoblastoma was extended to other pediatric cancers, such as neuroblastoma andWilms tumor (2, 3), aswell as many cancers occurring in adults. Although this model may not explain the genetic etiology of all heritable cancers, it has been a guiding principle for cancer susceptibility and pathogenesis around the world for 45 years. In addition to this and many other contributions to our understanding of the genetic basis of cancer, Al was also an "intellectual pollinator." He took great delight in going to meetings and visiting scientists around the world, and then sharing his insights with the other scientists whom he encountered. He was a humblemanwhodidnot seekpersonal credit or acknowledgment for his contributions, but rather, he delighted in the successes of others. Nevertheless, Al certainly won his share of accolades, including the Charles S. Mott Prize from the General Motors Foundation in 1988, election to theNational Academy of Sciences in 1992, the David A. Karnofsky Memorial Award and Lecture from the American Society of Clinical Oncology in 1997, the LaskerClinicalMedical ResearchAward in1998, theKyotoPrize in 2004, and the Award for LifetimeAchievement inCancer Research from the American Association for Cancer Research (AACR) in 2005. However, his contributions to science and inspiration to many cannot be measured by awards or words. He was a gentle and inspiring giant in the field, and he will be deeply missed. Pediatric Cancer Predisposition: Introduction Cancer in children is generally considered a rare and sporadic event, and only a small percentage of cases were previously thought to result from a genetic predisposition. However, a number of features suggest that at least a subset of pediatric cancers result from a genetic predisposition. These include (i) family history of the same or related cancers; (ii) bilateral, Children's Hospital of Philadelphia, Philadelphia, Pennsylvania. St. Jude Children's Research Hospital, Memphis, Tennessee. Baylor College of Medicine and Texas Children's Hospital, Houston, Texas. Intermountain Primary Children's Hospital and Huntsman Cancer Institute/University of Utah, Salt Lake City, Utah. The Hospital for Sick Children, University of Toronto, Toronto, Ontario, Canada. Corresponding Author: Garrett M. Brodeur, Children's Hospital of Philadelphia, University of Pennsylvania, 3501 Civic Center Boulevard, CTRB Room 3018, Philadelphia, PA 19104-4302. Phone: 215-590-2817; Fax: 215-590-3770; E-mail: [email protected]. doi: 10.1158/1078-0432.CCR-17-0702 2017 American Association for Cancer Research. CCR PEDIATRIC ONCOLOGY SERIES www.aacrjournals.org e1 on December 30, 2017. © 2017 American Association for Cancer Research. clincancerres.aacrjournals.org Downloaded from multifocal, or multiple cancers; (iii) earlier age at diagnosis than sporadic tumors of the same type; (iv) physicalfindings suggestive of a predisposition syndrome; and (v) occurrence of specific tumor types that frequently occur in the context of genetic predisposition (4). A child presenting with any of these five features should be referred for evaluation to a cancer predisposition program (5). Examples of physical findings associated with specific cancer predisposition syndromes include caf e-au-lait macules in neurofibromatosis type I, macroglossia in Beckwith–Wiedemann syndrome (BWS), or macrocephaly in PTEN hamartoma tumor syndrome; however, these findings also occur in individuals without cancer predisposition. Examples of specific pediatric cancer diagnoses that are frequently seen in the setting of underlying cancer susceptibility include pleuropulmonary blastoma in DICER1 syndrome, malignant rhabdoid tumors in rhabdoid tumor syndrome, and adrenocortical carcinoma in Li-Fraumeni syndrome (LFS). Recent reports using genome-scale germline sequencing of pediatric cancer cohorts not selected for genetic risk (6–8) suggest at least 10% of pediatric cancer patients harbor a germline mutation in known cancer predisposition genes. This is probably an underestimate, because there are also patients who fulfill clinical criteria for a cancer predisposition syndrome but lack identifiable germline mutation in the known gene(s) currently associated with those conditions. In addition, there are pediatric patients with cancer whose families exhibit a higher frequency of cancer than expected yet do not fit any patterns for a known predisposition syndrome. Furthermore, some individuals are predisposed to develop cancer on the basis of epigenetic changes (e.g., BWS or hemihypertrophy) that would not be detected by traditional DNA sequencing. Cancer genomes are increasingly scrutinized to identify variants that are actionable for targeted therapy, but changes or patterns of somatic mutation can sometimes be identified that may represent an unanticipated germline predisposition. In addition, pediatric oncologists are increasingly sending patient samples for paired tumor/normal sequencing, with a primary goal to identify potentially targetable genetic lesions in the tumor. This analysis may also inadvertently lead to the identification of pathogenic variants that reflect germline mutations in cancer susceptibility genes, a possibility forwhichpatients and their families donot always receive pretest genetic counseling. It is likely that this type of systematic analysis of tumors (and paired germline samples) will become routine in pediatric oncology clinical care in the next few years, maybe even one day becoming the standard of care at diagnosis and even more likely at relapse. For example, the NCI/Children's Oncology Group Pediatric MATCH trial will use such a study design for its sequencing. Overall, we expect that the proportion of cancer patients identified as carriers of cancer susceptibility mutations will increase as both targeted genetic evaluations and genomic analyses of cancer cells are implemented. Individuals with cancer predisposition syndromes carry a significantly increased risk of developing one or more cancers. Therefore, focused surveillance on the types of cancer(s) to which the individual is most predisposed, and during the period of greatest risk, should substantially improve their outcome through early detection. Indeed, many centers are starting to follow proposed early tumor surveillance protocols that have been published for a few disorders such as LFS and BWS/hemihypertrophy (9–11). However, even for these disorders, surveillance has not been performed consistently across different centers, and there are still many inherited cancer predisposition disorders for which either no protocols or multiple published protocols exist. This practice variability makes it difficult to compare studies from different centers or groups and creates uncertainty for clinicians caring for patients with these individually rare disorders. An AACR-sponsored workshop was held in Boston, Massachusetts, from October 6 to 8, 2016, to develop consensus recommendations for cancer surveillance of children and adolescents with heritable cancer predisposition. Sixty-five professionals from 11 countries were present, including 51 physician directors or codirectors of cancer predisposition programs (pediatric oncologists or medical geneticists), seven genetic counselors, three radiologists, three directors of adult cancer predisposition programs, and one pediatric endocrinologist. The main goal of this meeting and the postworkshop activities was to review the existing data and practices, and to establish international consensus recommendations for cancer surveillance for the most common cancer predisposition syndromes. These pediatric cancer syndromeswere organized intonine groups basedon specific themes. Attendees were organized into these groups based on prior experience and expertise. Major Cancer Predisposition Syndromes This AACR workshop focused on the 50 most common syndromes that predispose individuals to the development of cancer in the first 20 years of life. These syndromes were then divided into ninemajor groupsbasedon themajor cancer typeswithwhich they are associated: (i) LFS, (ii) neurofibromatoses, (iii) overgrowth syndromes and Wilms tumor, (iv) neural tumors, (v) gastrointestinal cancer predisposition, (vi) neuroendocrine syndromes, (vii) leukemia predisposition, (viii) DNA instability syndromes, and (ix) miscellaneous syndromes. The disorders and associated genes for each of these categories are summarized in Table 1. Cancer Surveillance Considerations Prior to the workshop, each group reviewed the published literature to determine the current state of screening recommendations fromexperts,professionalorganizations,orgroupsthatcare forsuch patients. Attendees also provided information from their own centersaswellaspersonalexperiencewithregardtocancer incidence and surveillance protocols. The groups frequently consulted additional experts on specific disorders for their input both before and after the workshop. In general, the recommended surveillance protocols are designed for asymptomatic individuals who are genetically predisposed to develop cancer, although some individuals are identified only after the development of their first cancer. The first question addressed was whether or not it was appropriate to undertake cancer surveillance for children or adolescents with a given cancer predisposition syndrome. For this decision, the consensus of the group was that surveillance was recommended when there is a 5% or greater risk of developing cancer during the first 20 years of life and when effective screening modalities existed. Surveillance was not recommended or consideredworthwhile if the risk for an individualmalignancy during the first 20 years of life was less than 1%. Conditions in which the cancer risk during childhood fell between 1% and 5% were discussed on an individual basis. We supported surveillance if the screening modalities were relatively cost-effective and noninvasive, and/or if the outcome was so poor for clinically detected CCR PEDIATRIC ONCOLOGY SERIES Clin Cancer Res; 23(11) June 1, 2017 Clinical Cancer Research e2 on December 30, 2017. © 2017 American Association for Cancer Research. clincancerres.aacrjournals.org Downloaded from tumors that any possibility for early detection might enhance survival. If screening was indicated, then the remaining questions for surveillance included the following: 1. What procedure(s) should be done? 2. How often should screening be performed? 3. At what age should screening start, and if/when should it stop? 4. Should the screening procedures change over time, such as a change in frequency or type of screening with age to account for changes in cancer risk? Wealso askedwhether any evidence exists that the recommended surveillance leads to improved clinical outcome, but formost cases, this will have to await the implementation of the recommendations proposed in the position articles for the respective disorders. Evidence that Surveillance Improves Outcome It is generally accepted that early identification of tumors when smaller and less likely to be metastatic will improve clinical outcome. This is the underlying principle that supports cancer surveillance approaches for adults (e.g., screening for colon, breast, prostate, or other cancers, especially in predisposed individuals). However, relatively few studies of cancer surveillance in children and adolescents have been published. In a recent study, a clinical surveillance protocol was implemented using frequent biochemical and imaging studies for asymptomatic individuals with LFS (9, 10). Forty subjects underwent surveillance, whereas 49 did not, and the 5-year overall survival (OS) was 89% in the surveillance group versus 60% in the nonsurveillance group. These data suggest that surveillance was associatedwith improved OS from tumors detected in patients with germline TP53 mutations even after 11 years of follow-up (9, 10). Another group undertook a cost–benefit analysis of children with BWS undergoing screening for Wilms tumor and hepatoblastoma (11). Using a conservative model, they determined that screening for these tumors with abdominal ultrasound was predicted to be cost-effective. Although the focus of the studywas not on outcome, screening also resulted in an improved survival (11). A review of characteristics and outcome of children with BWS and Wilms tumor treated on National Wilms Tumor Study Group protocols showed a trend toward smaller tumors over time that was not seen in non-BWS patients with Wilms tumor, suggesting that existing screening protocols led to earlier detection (12). Together, these data on cancer surveillance of LFS and BWS suggest that screening enables the detection of smaller tumors, allowing for less-intensive therapy, less organ toxicity, and better outcomes. Consistent Approach to Pediatric Cancer Patients and Their Families The primary goal of this CCR Pediatric Oncology Series is to develop consensus recommendations for the management of children at significant hereditary risk for cancer. However, a parallel issue discussed in more detail elsewhere in this series (see ref. 5) includes the need for a consistent and thoughtful approach to genetic testing for these families. Training in pediatric hematology-oncology does not emphasize the importance and methods with which to obtain and record a complete family cancer history. Similarly, gaps exist in training around the appropriate methods for germline genetic testing and the issues that should be addressed when obtaining consent for tumor testing. Genomic testing of tumor tissue can reveal mutations in genes that may also be present in the germline. Mutations present in the tumor at an allele frequency near 50% can be acquired, or they may reflect a germline mutation (13). A number of studies have demonstrated that parents of children with cancer are concerned about a hereditary contribution to the cancer diagnosis in their child, with possible implications for other family members (14). Table 1. Major subgroups of pediatric cancer susceptibility disorders reviewed Predisposition group Specific disorders reviewed