Cancer incidence, patterns, and genotype-phenotype associations in individuals with pathogenic or likely pathogenic germline TP53 variants: an observational cohort study

Abstract

Background: Li-Fraumeni syndrome, caused primarily by pathogenic or likely pathogenic germline TP53 variants, is a rare, variably penetrant, cancer predisposition syndrome with very high risks of cancer starting in childhood, including the risk of multiple primary malignancies over an individual's lifespan. We aimed to characterise and quantify cancer incidence, patterns, and genotype-phenotype associations in individuals with pathogenic or likely pathogenic germline TP53 variants.

Methods: This observational cohort study was done in 480 carriers of pathogenic or likely pathogenic germline TP53 variants enrolled in the National Cancer Institute's referral-based longitudinal Li-Fraumeni syndrome study between Aug 1, 2011, and March 24, 2020. Data on personal and family history of cancer were obtained through study questionnaires and validated by medical records. Variants were categorised on the basis of both loss-of-function (LOF) and dominant-negative effect (DNE) properties. Cancer incidence associated with Li-Fraumeni syndrome was compared with that of the general population using the Surveillance, Epidemiology, and End Results (SEER) 1975-2017 registry. Cancer incidence was evaluated with family-clustered Cox regression models and competing risk methods. This study is registered with ClinicalTrials.gov, NCT01443468.

Findings: Individuals with Li-Fraumeni syndrome had a nearly 24 times higher incidence of any cancer than the general population (standardised incidence ratio 23·9; 95% CI 21·9-26·0), with the highest comparative incidence from childhood to 30 years of age. The overall cancer incidence remained 10·3 (95% CI 7·9-13·2) times higher than that of the general population after age 50 years. In women, when considering breast cancer as a competing risk, the probability of a first diagnosis of a non-breast cancer malignancy was substantially lower than that of any first cancer (24·4% [95% CI 19·6-30·5] vs 50·4% [43·5-56·5] by age 33·7 years). Overall, DNE_LOF and notDNE_LOF variants were associated with earlier age at first and second cancer compared with notDNE_notLOF and DNE_notLOF variants. The time interval from first to second cancer was shorter among carriers whose first cancer diagnoses were later in life. Multiple cancers were diagnosed within a short timeframe in some individuals, regardless of the order of cancer occurrence.

Interpretation: This study adds granularity to the understanding of cancer incidence and patterns in individuals with pathogenic or likely pathogenic germline TP53 variants. Integration of age range-specific cancer incidence estimates, cancer-free survival by functional variant group, the potential impact of risk-reducing mastectomy on female cancer incidence, and data on subsequent malignancies will be important for the development of strategies to optimise cancer screening and management for these individuals.

Funding: Intramural Research Program, Division of Cancer Epidemiology and Genetics, National Institutes of Health.

Conflict of interest statement

Declaration of interests KCA, JNH, MNF, PLM, and SAS are unpaid members of the ClinGen TP53 Variant Curation Expert Panel. MNF is a co-developer of CancerGene Connect and a member of the National Accreditation Program for Breast Centers Board, representing the National Society of Genetic Counselors. All other authors declare no competing interests.

Copyright © 2021 Elsevier Ltd. All rights reserved.

Figures

Figure 1.

Standardized Incidence Ratios (SIR) and 95% Confidence Interval (CI) for the main cancer types included in the analysis. Synchronous cancers were counted independently. Dashed line at 1 represents the expected SEER-associated SIR. Results for any cancer (plus breast in situ), leukemia, lymphoblastic leukemia, MDS/AML, lymphoma, leiomyosarcoma, liposarcoma, and rhabdomyosarcoma are provided in Appendix p 8.

Figure 2.

Probability of diagnosis of a first cancer by age, stratified by sex (A, B) and functional variant group (C, D). A, C: Any first cancer. B, D: Any non-sex-specific or non-breast first cancer, considering sex-specific (breast, gynecological, prostate) or breast cancers only as a competing risk. Synchronous cancers of the same category were counted as single events; synchronous cancers of different categories were counted independently. DNE_LOF class was considered as a baseline. P-values calculated using family-clustered Cox-proportional-hazard (A, C) and proportional causespecific hazard (B, D) models. Abbreviations: LOF, loss-of-function; DNE, dominant-negative effect.

Figure 3.

A, B: Probability of diagnosis of a second cancer, by time after first cancer diagnosis (A) and age at second cancer (B), accounting for death prior to second cancer as a competing risk. Stratification by age at first cancer diagnosis. Individuals with a diagnosis at age 0–17 years were considered as a baseline. P-values calculated using family-clustered proportional cause-specific hazard models. C: Timeline plot showing time intervals between primary cancers and last follow-up or death, restricted to the 128 individuals with at least two primary cancers and known age at diagnosis for all cancers. Synchronous cancers were counted as single events. Blue lines represent time intervals between primary cancers; orange lines represent time interval between the last diagnosed primary cancer and either last follow-up or death.