Performance of longitudinal cancer screening in Li-Fraumeni syndrome

Li-Fraumeni syndrome (LFS), a cancer predisposition syndrome caused by germline [likely] pathogenic (P/LP) TP53 variants, confers elevated multi-organ cancer risks from infancy and high lifetime risk of multiple primary cancers. The heterogeneous cancer spectrum necessitates multi-modal lifelong cancer screening to facilitate early detection. The study included 145 individuals (62% female) with a germline P/LP TP53 variant who enrolled in the National Cancer Institute's IRB-approved longitudinal LFS study (NCT01443468) and received cancer screening at the NIH Clinical Center including annual whole-body MRI (WBMRI), brain MRI, breast MRI (females>20 years), mammography (females>40 years), bloodwork, triennial colonoscopy (>25 years), and abdominal ultrasound every 4 months (age 3-16 years). Median age was 34 years (range 3-68) at the first screening visit. In all, 772 annual screening visits were completed (median 5 visits per participant, range 1-10) and 3956 screening tests performed at a compliance of 92.8%. Screening tests led to 545 follow-up tests. WBMRI led to at least one follow-up test in 29.1% scans (n=225/772). Invasive biopsies comprised 8.3% (n=45/545) of all follow-ups, 48.9% (n=22/45) of which resulted in a cancer diagnosis. WBMRI led to 62% (n=28/45) of all biopsies, of which 42.8% (n=12/28) resulted in a cancer diagnosis. Of the 72 cancers/pre-cancers diagnosed during the study, 75% (n=54) were screen-detected (32/54, 59% from WBMRI). The 18 (25%) interval cancers were diagnosed at a median of 8 months (range 3-13) after last screening test. Prevalence of asymptomatic cancer was 7% at first visit, varying between 4% and 11% at subsequent visits. The screening protocol had a sensitivity of 71.7%, specificity 89%, false positive rate (FP) 11%, negative predictive value (NPV) 99.1%. Diagnostic properties of each screening test for screen-detectable cancers were analyzed using logistic regression with generalized estimating equations. WBMRI sensitivity was 94%, specificity 94.8% (between 91.6% for thoracic cancers and 97.6% for upper extremity), FP 27%, NPV 99.9%. Brain MRI had sensitivity of 85.7%, specificity 98%, FP 2%, NPV 99.9%. Breast MRI had sensitivity of 88.9%, specificity 83.6%, FP 16.3%, NPV 98.9%. Neither germline TP53 variant nor age significantly affected specificity. A previous cancer diagnosis in an anatomical region reduced specificity of WBMRI (OR=0.12, 95% CI 0.06-0.25, <0.001). A previous false positive finding in an anatomic region increased the probability of additional false positive findings in that region (p<0.001). This largest longitudinal LFS cancer screening study to date shows that multi-modal screening performs well with most reported cancers being screen-detected. FP rate is high but lower than previously reported. While further research on alternative screening strategies and cancer prevention is needed, screening is clinically crucial to early cancer detection in LFS.