Genotype–phenotype associations within the Li-Fraumeni spectrum: a report from the German Registry

Li-Fraumeni syndrome (LFS; OMIM151623) is a cancer predisposition syndrome caused by pathogenic variants (PVs) in the TP53 tumor suppressor gene and represents one of the best characterized genetic causes of cancer in children and adults [1‐4]. The use of modern DNA-sequencing methods has revealed TP53 germline PVs in individuals who do not meet established clinical LFS criteria, leading to a Li-Fraumeni spectrum classification [5]. We analyzed factors influencing the cancer risk across this spectrum. The overall aim of such studies is to improve risk prediction to inform cancer surveillance.

Founded in 2017, the German Cancer Predisposition Syndrome Registry collects information on genotypes, personal medical details, family histories, and surveillance, as well as a range of biospecimens. The cutoff date for study inclusion for the present analysis was July 31, 2021. Patients with a germline TP53 PV (pathogenic or likely pathogenic) or with a somatic mosaic TP53 PV were included. All variants were curated according to TP53 specific guidelines [6]. Classic LFS criteria [2], Chompret criteria [4] as well as the Li-Fraumeni spectrum classification [5] were assessed. To search for genotype–phenotype correlations we used functional data from Kato [7], Giacomelli [8], Kotler [9] as well as estimated dominant negative effects based on studies by Monti [10] and Dearth [11]. We tabulated the 94 LFS families and applied the Fisher's exact test to analyze whether the phenotypes (1) LFS versus attenuated LFS and (2) occurrence of childhood cancer other than adrenocortical carcinoma (ACC) alone versus cancer free childhood except ACC were associated with specific genotypic/functional TP53 PV subgroups. A P value of < 0.01 was considered statistically significant. Ethics review and informed consent were obtained.

An overview of all variants, functional data categories, and associated phenotypes including personal and family histories are provided in Additional file 1. The cohort comprises 141 individuals from 94 families; 43 (30.5%) individuals were children or adolescents < 18 years, whereas 98 (69.5%) individuals were adults. There were 98 female and 43 male patients (male-to-female ratio: 0.44). This uneven gender distribution may be due to females being tested more frequently in the context of a breast cancer diagnosis. Four cases with somatic mosaicism were reported. TP53 PVs as well as statistically significant genotype–phenotype correlations are depicted in Fig. 1.

According to the Li-Fraumeni spectrum classification [5], the cohort included 79 individuals with LFS, 33 LFS carriers as well as 14 individuals with attenuated LFS and 15 attenuated LFS carriers. No consistent signs of anticipation were observed. In the entire cohort, 33 families (35.1%) did not meet any of the established LFS testing criteria. Thirty-four LFS patients (30.4%) had multiple (between two and five) malignancies, whereas six patients with attenuated LFS (20.7%) had a history of multiple (between two and four) malignancies. Overall, 134 neoplasms occurred in 79 LFS patients, whereas 26 malignancies occurred in 14 individuals with attenuated LFS (Fig. 2). In patients with LFS, breast cancer ≤ 30 years, osteosarcoma, rhabdomyosarcoma, non-rhabdomyosarcoma soft tissue sarcoma, ACC, and central nervous system tumors were diagnosed in 73 of 134 (55%) patients. In individuals with attenuated LFS, more than half of the tumors diagnosed were breast cancers > 30 years. The proportion of miscellaneous neoplasms not known to be strongly associated with TP53 germline PVs was 34.6% in patients with attenuated LFS compared to 17.9% in patients with LFS. Altogether, 65 breast cancers occurred in the entire cohort, 26 of which were HER2 + , 24 were HER2-, and for 15 tumors histological details were not available.

Kato partially functional variants were statistically significantly associated with a cancer-free childhood, apart from childhood ACC (10 out of 53 families without childhood cancer except ACC versus 0 out of 41 families with childhood cancer except ACC alone, P value < 0.01). Typical LFS childhood cancers (i.e., rhabdomyosarcoma, osteosarcoma, choroid plexus carcinoma, medulloblastoma, other brain tumors, and leukemia)—excluding ACC—occurred exclusively in individuals with NULL variants or non-functional missense variants. In general, childhood cancer occurred in more than half of the families with NULL (58.8%) or non-functional missense (52%) variants, whereas in families with partially functional variants ACC was observed as the only childhood cancer, affecting 30% of these families. We observed a statistically significant association between NULL variants and LFS, while this variant type was rare among patients with attenuated LFS: 32 out of 73 families with LFS carried NULL variants, whereas NULL variants were present in two out of 21 families with attenuated LFS (P value < 0.01). We did not observe additional statistically significant associations when analyzing the other functional variant subgroups. Case ascertainment, differences in overall survival, family size, and/or family clustering may have introduced a potential bias and represent a limitation of our study.

Despite this limitation, these data suggest that future more detailed genotype–phenotype correlations may allow for accurate cancer risk prediction (time to first malignancy and second cancer risk) and personalized cancer surveillance. Large, international collaboration is required to reach the statistical power to make such risk predictions. Our findings are in agreement with previously published results assessing the correlation between TP53 genotypes and various other cancer phenotypes in LFS [12, 13]. The observation that a substantial proportion of patients is missed using established LFS testing criteria suggests that the criteria require modification.