Li-Fraumeni Syndrome in the Cancer Genomics Era

A syndrome is a medical condition characterized by 1 or more symptoms or signs that occur together in a person more frequently than expected by chance. Historically, all medical syndromes were defined by the phenotypes present. With the increasing use of a “genome-first” approach (1), however, the ground is shifting as to what defines a syndrome. Li-Fraumeni syndrome (LFS) was narrowly defined clinically by Li and Fraumeni (2), and then broader but still restricted clinical criteria were established and later updated by a group led by the late Thierry Frébourg (3). These approaches limited the scope of the syndrome because only those who met criteria were offered the labor-intensive genetic tests that could be performed at the time. With falling sequencing costs, most multigene panel tests for the common adult and pediatric cancers will include TP53; thus, many more people with germline pathogenic variants (GPVs) are being identified. Therefore, increasingly large numbers of such people do not fulfil any established criteria for LFS. If a syndrome is primarily defined by genotype, it will lead to a change in the definition of what constitutes that syndrome (Figure 1). Changes in the way Li-Fraumeni syndrome (LFS) has been defined. Shown above and below the central (blue) image are periods of time defining the genetic and diagnostic eras of LFS. More recent eras are to the right of the figure. The genetic era is divided into pre-Sanger (ie, before genetic testing was available); post-Sanger (when sequencing was generally gene by gene); and a next-generation sequencing (NGS) era, when multigene panel testing, including copy number analysis, became widely available for testing of germline DNA. Most recently, paired NGS of normal and tumor DNA has allowed for a more accurate attribution of causality to germline pathogenic variants (GPVs) (cancer genomics era). The diagnostic era is divided into the earlier clinical phase and the more recent genetic phase. Shown in a dark blue circle are the core tumors of LFS (2,3), with an outer gray ring of possible LFS cases that did not meet clinical criteria. Identifying GPVs in TP53 in core tumors (cGPVTP53, brown text) confirmed the clinical diagnosis. As NGS made testing easier and less expensive, noncore tumors were identified in people with GPVs in TP53 (ncGPVTP53, pink text), but their biological significance could not be determined. Therefore, during this era, the range of tumors possibly falling within the LFS spectrum increased (blue shading, becoming fainter with time). Adding second hits (ncGPVTP53+LOH) has helped narrow the spectrum, shown by the blue shading leading to a lighter blue ring outside the dark blue circle of core tumors, while leaving other tumors (gray ring) outside the spectrum. The work of Ceyhan-Birsoy and others (4,12) shows that tumor studies are not required to confirm the status of core tumors. *Li-Fraumeni syndrome as defined by the classical (2) and Chompret (3) criteria. Changes in the way Li-Fraumeni syndrome (LFS) has been defined. Shown above and below the central (blue) image are periods of time defining the genetic and diagnostic eras of LFS. More recent eras are to the right of the figure. The genetic era is divided into pre-Sanger (ie, before genetic testing was available); post-Sanger (when sequencing was generally gene by gene); and a next-generation sequencing (NGS) era, when multigene panel testing, including copy number analysis, became widely available for testing of germline DNA. Most recently, paired NGS of normal and tumor DNA has allowed for a more accurate attribution of causality to germline pathogenic variants (GPVs) (cancer genomics era). The diagnostic era is divided into the earlier clinical phase and the more recent genetic phase. Shown in a dark blue circle are the core tumors of LFS (2,3), with an outer gray ring of possible LFS cases that did not meet clinical criteria. Identifying GPVs in TP53 in core tumors (cGPVTP53, brown text) confirmed the clinical diagnosis. As NGS made testing easier and less expensive, noncore tumors were identified in people with GPVs in TP53 (ncGPVTP53, pink text), but their biological significance could not be determined. Therefore, during this era, the range of tumors possibly falling within the LFS spectrum increased (blue shading, becoming fainter with time). Adding second hits (ncGPVTP53+LOH) has helped narrow the spectrum, shown by the blue shading leading to a lighter blue ring outside the dark blue circle of core tumors, while leaving other tumors (gray ring) outside the spectrum. The work of Ceyhan-Birsoy and others (4,12) shows that tumor studies are not required to confirm the status of core tumors. *Li-Fraumeni syndrome as defined by the classical (2) and Chompret (3) criteria. In this issue of the Journal, Ceyhan-Birsoy et al. (4) use the Memorial Sloan Kettering Integrated Mutation Profiling of Actionable Cancer Targets resource to interrogate normal and tumor DNA for TP53 pathogenic variants in 17 922 people with cancer and conclude that National Comprehensive Cancer Network criteria for TP53 germline testing will miss people with LFS. Importantly, the authors show that interrogating paired normal-tumor DNA samples of affected individuals can provide insights into the tumor spectrum truly associated with TP53 GPVs as opposed to identifying people in whom the variants are unlikely to be causal for that cancer. By sequencing a large number of people with mainly adult-onset cancers, Ceyhan-Birsoy et al. (4) identified a far broader range of tumors arising in people with TP53 GPVs than would be expected from existing clinical criteria. This phenomenon has been previously noted (5,6) and is therefore not unexpected. The additional insights in the current paper (4) come from the paired DNA sequencing of the tumor at the same time as the normal tissue. The authors matched their identified TP53 GPVs with somatic second hits (loss of heterozygosity [LOH] and intragenic disrupting variants) affecting the wild-type allele to determine whether biallelic inactivation of TP53 had occurred. As expected, pathogenic variants in TP53 were rare: 50 of 17 922 people had a pathogenic variant (0.28%), and 16 of the 50 (32.0%) were not germline in origin. People fulfilling National Comprehensive Cancer Network LFS criteria were more likely to have a second hit in their tumor compared with those who did not meet criteria (P = .03). Five noncore spectrum tumors had LOH (2 lung cancers and 1 each of colorectal, duodenal, and high-grade serous ovarian carcinoma). Four of these cancers have a high frequency of somatic TP53 alterations (http://www.tumorportal.org/view?geneSymbol=TP53). Although some missense TP53 variants appear to act in a dominant-negative manner, the authors (4) did not notice any difference in the LOH frequency in loss-of-function variants compared with dominant-negative variants. Thus, although it may not completely apply to TP53 (7), the 2-hit model has served as a critical framework for understanding the function of tumor suppressor genes. For autosomal dominantly inherited disorders, the second hit is a somatic event; therefore, one could argue that the biological significance of any GPV can be fully appreciated only in the context of the tumor occurring in the person in whom the GPV was found. We have created a simple quantitative framework that approximates germline risk using cancer genomics, named the Etiologic Index (EI) (8), that is applied to tumor suppressor genes to help define the limits of inherited cancer syndromes. It is reasonably accepted that breast, ovarian, pancreatic, and prostate cancer are established “syndromic” cancers in heterozygotes for BRCA1 GPVs, whereas the status of other cancers is uncertain. In our analysis (8), the EI for these established cancers was 6.7 for BRCA1 and 6.2 for BRCA2 (reflecting that >84.2% of these tumors had biallelic inactivation of BRCA1/2). For all nonestablished cancers, the values were 1.6 and 1.7, respectively (meaning that 38.2% of these cancers had a second [somatic] hit in BRCA1/2 combined). When applied to the data of Ceyhan-Birsoy et al. (4), where LOH of the wild-type allele in TP53 GPV heterozygotes was observed in 24 of 25 (96)% of core LFS spectrum–type tumors vs 5 of 11 (45%) of the other tumors, we can say that the EI for LFS-spectrum tumors was 25/1, or 25.0, whereas for the nonspectrum tumors it was 11/6 = 1.83. Note that the proportion of second hits and hence EI in nonspectrum TP53 cancers is similar to the nonestablished numbers for BRCA1 and BRCA2. Consistent with this low to medium risk estimated by EI, 5 of 29 (17.2%) of cancers that are likely to be caused by TP53 GPVs are not part of the core spectrum that has defined the syndrome (4). Second hits can help define a syndrome, but they may not by themselves be precise enough. In the case of BRCA1/2, mutational signatures are associated with homologous recombination repair deficiency (9), and for TP53, copy number signatures could be detectable (10). Therefore, tumor sequencing signatures open the door to define which cancer is caused directly by TP53 GPVs. More generally, cancer susceptibility syndrome definitions could be expanded from the person’s phenotype to the include the person’s tumor genotype (Figure 1). The increasingly recognized contribution of noncore cancers to the totality of tumors observed in those with TP53 GPVs has led to a proposed renaming of LFS to “heritable TP53-related cancer (hTP53rc) syndrome” (11), which, although accurate, lacks the simplicity of the existing eponymous designation. Perhaps we need to accept that the syndrome, whatever it is called, is broader than previously thought and, given the unpredictable nature of the phenotypic manifestation of a GPV in TP53, new approaches, such as liquid biopsies, will be required to screen unaffected people with a GPV in TP53. In affected people, testing the associated tumor, as shown by Ceyhan-Birsoy et al. (4), will be of special value not only for determining the provenance of the variant but also for attributing clinical significance to its presence. None. Role of the funder: Not applicable. Disclosure: In 2019, Dr Foulkes held funds on behalf of a hospital clinical cancer genetic testing laboratory that were provided by Astra Zeneca and used to improve variant classification of BRCA-related genes. Dr Polak has no disclosures. Author contributions: Writing—original draft: WDF, PP; writing–review and editing: WDF, PP. Acknowledgements: We thank Thibaut Matis, MD, for his contribution to the figure. All data analyzed for this editorial were derived from Ceyhan-Birsoy et al. (4).

Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.