Introduction

Patients with Li-Fraumeni Syndrome (LFS), a dominantly inherited genetic condition, confers a significantly high risk of developing cancer. It is caused by deleterious germline variants in the tumor suppressor gene TP53 [1]. The prevalence ranges from 1 in 20,000 to 1 in 500 individuals in the general population [2].

LFS is characterized by the development of early-onset tumors, including premenopausal breast cancer, soft tissue sarcomas, osteosarcomas, adrenocortical carcinomas, and central nervous system (CNS) tumors. Additionally, lymphomas, lung cancer, gastrointestinal cancer, and melanoma have also been observed in individuals with LFS [1, 2]. Diagnosis of LFS is typically confirmed by detecting the presence of a pathogenic germline variant in TP53. Therefore, germline testing should be performed when LFS is clinically suspected by the classic or modified Chompret diagnostic criteria [3, 4].

Strategies for routine cancer surveillance are fundamental to monitoring and managing LFS patients. Guidelines have recommended various imaging scans, such as whole-body magnetic resonance imaging (WBMRI), brain MRI, breast MRI, and abdominal ultrasonography, which should be conducted shortly after diagnosis. However, the effectiveness of surveillance is uncertain, and it comes with challenges, such as costs, high rates of false positive results, overdiagnosis, the need for sedation during imaging procedures, and potential psychological impacts [5,6,7].

According to Toronto Protocol guidelines, regular screening of individuals with LFS can lead to early cancer diagnosis and improve overall survival rates. This protocol includes annual rapid WBMRI, which can detect malignant tumors of asymptomatic TP53 carriers. Most of the newly identified cancers were found early, allowing for curative treatment [8, 9]. Previous studies supported the role of WBMRI-based screening in the early detection of malignant tumors, reducing additional invasive procedures and minimizing the recalls for further evaluation. [10,11,12].

WBMRI has the potential to play a crucial role in the surveillance of this high-risk population. In 2017, Ballinger et al [12] conducted a meta-analysis to estimate the performance and frequency of cancer detection by WBMRI. Since then, diagnostic methods have become more accessible, and several additional studies have been published. Here, we performed a comprehensive systematic review and an updated meta-analysis to establish the frequency of asymptomatic cancer detection using WBMRI as part of an initial evaluation for TP53 mutation carriers.

Material and methods

Study eligibility

Included studies have met all the following criteria: (1) clinical trials, observational studies, or case series; (2) at least one publication on the searched databases; (3) WBMRI as an early detection method for tumor screening; (4) asymptomatic LFS patients; and (5) baseline detection rate of lesions and malignant lesions.

The exclusion criteria were: (1) letters, reviews, editorial comments, or case reports; (2) overlapping patient populations; (3) non-human studies; (4) combined analyzes with other syndromes; and (5) ongoing trials.

Search strategy

This study was designed in accordance with the Preferred Reporting Items for Systematic Reviews and Meta-analyses (PRISMA) reporting guidelines, and the prospective meta-analysis protocol was registered on PROSPERO (CRD42023441723) on July 12, 2023. Pubmed, Cochrane, and Embase databases were systematically searched on November 22, 2023. The following search strategy was used: (Li-Fraumeni OR “Li-Fraumeni syndrome” OR LFS OR TP53 OR “TP53 carriers” OR “germline TP53”) AND (WBMRI OR “whole-body magnetic resonance” OR “whole-body MRI” OR MRI) AND (“cancer screening” OR screening OR surveillance OR “risk management” OR detection OR lesions OR survival OR cancer OR cancers).

Data extraction

Two authors (M.I.D. and D.C.) independently extracted the data following predefined search criteria, data extraction, and quality assessment. Any discrepancies were resolved by discussing with a third author (P.R.). Furthermore, the references of included studies and systematic reviews were evaluated for additional studies. Data from studies of the Li-Fraumeni Exploration Research Consortium reported in the previous meta-analysis were extracted from the published review [13].

We collected baseline detection rates of lesions and cancers from each study. Clinical follow-up varied between studies; additional imaging and biopsy were the most frequently used diagnostic confirmatory methods. The standardized data extraction forms included the study characteristics (e.g., publication year, first author, journal), subject characteristics (e.g., the number of patients, patient age, type of cancer, TP53 variants), intervention details (e.g., WBMRI, brain MRI, biopsy, other imaging), and outcome measures (number of lesions detected, number of malignant lesions confirmed, prevalence, incidence).

Endpoints and sub-analysis

The outcomes of interest included the proportion of individuals with one or more suspicious lesions, the proportion of individuals with one or more new cancers, the proportion of lesions confirmed to be new cancer, the incidence rate of individuals with cancer per patient-round of WBMRI for baseline plus follow-up and for follow-up tests only (considered as one round of WBMRI screening equivalent to screening 100 patients). The leave-one-out sensitivity analysis was performed for the proportion of lesions found to be new cancers among suspicious lesions (see Fig. S2 Supplementary Material for the leave-one-out sensitivity analysis of Fig. 2B).

Risk of bias assessment

Two authors (L.R. and D.C.) evaluated the risk of bias for eligible studies using the ROBINS-I tool (v. 2016) for non-randomized studies of intervention [14]. Baseline cancer prevalence detected by WBMRI was assessed as a primary outcome, and studies were judged as presenting low, moderate, serious, or critical risk of bias. Disagreements were resolved through a consensus after discussing reasons for the discrepancy (see Figs. S3 and S4 Supplementary Material for the critical appraisal of individual studies and summary of the risk of bias).

Statistical analysis

We performed a proportional meta-analysis using the metaprop function to pool studies and estimate the frequency of our outcomes and their corresponding 95% confidence intervals (95% CI). Heterogeneity was evaluated with I2 statistics; I2 > 25% was considered significant for heterogeneity. Leave-one-out sensitivity analyses were performed to investigate outcomes with high heterogeneity. We performed all statistical analyses using R software version 4.3.1 with a random-effects model. All authors are responsible for the sincerity and transparency of all statistical analyses, mainly the author P.R.

Results

Studies selection and baseline characteristics

As detailed in Fig. 1, the initial search yielded 1687 results. After the removal of duplicates and exclusion by title and abstract, 82 studies were fully assessed. Most excluded studies did not include a population with LFS or did not use WBMRI for screening. Finally, eleven studies with 14 related publications were included. Of them, 10 were observational retrospective cohort studies (n = 651) and one was a case–control (n = 52) (Fig. 1) [9, 15,16,17,18,19,20,21,22,23,24].

Fig. 1
figure 1

Preferred Reporting Items for Systematic Reviews and Meta-Analysis (PRISMA) flow diagram of study screening and selection

Full size image

Our analysis included a total of 703 participants, 359 females (51%) and 344 males (49%) with a median age of 32 years (IQR 1–74). These individuals were part of 11 cohorts from 7 different countries. The median follow-up ranged between 2 and 11 years and studies that had a follow-up with an additional WBMRI ranged from 1 to 2 years after the first scan. The baseline characteristics of the included studies are described in Table 1. Other specified information such as the description of the intervention, personal and family history of cancer, multiple tumors, and eligibility criteria are described in Table 2.

Table 1 Baseline characteristics of included studies
Full size table
Table 2 Additional characteristics of included studies
Full size table

Most of the cancers found were CNS tumors (n = 16) followed by sarcomas (n = 12) and BC (n = 11) (Table 3).

Table 3 The spectrum of tumors in TP53 carriers
Full size table

Outcomes

We found an estimated detection rate of 31% (95% CI: 0.28, 0.34; I2 0%) of any suspicious lesions in asymptomatic TP53 carriers who underwent baseline WBMRI (Fig. 2). In total, 277 lesions that required clinical follow-up in 215 patients were identified. Of them, 46 lesions in 39 individuals had a cancer diagnosis confirmed. The estimated cancer diagnosis rate among suspicious lesions was 18% (95% CI: 0.13, 0.25; I2 45%; Fig. 2). WBMRI detected 41 of 46 cancers in the early-disease stage, and they were amenable to therapy with curative intent. The overall detection rate for new cancers was 6% (95% CI: 0.05, 0.08; I2 0%; Fig. 2).

Fig. 2
figure 2

Proportion of suspicious and malignant lesions in asymptomatic patients. A Proportion of individuals found to have one or more investigable lesions. B Proportion of lesions found to be new cancers among suspicious lesions. C Proportion of individuals with one or more new cancers among the total population. Life-Guard = Ruijs 2017 [18] SIGNIFY = Bancroft 2017 [20] SWEP53 = Omran 2022 [15]

Full size image

To estimate incidence, we considered one round of WBMRI screening equivalent to screening 100 patients. Our incidence rate was 0.02 patients per round based on baseline and follow-up WBMRI testing (95% CI: 0.01, 0.04; I2 0%; Fig. 2A). Additionally, we found a similar incidence rate of 0.02 patients per round based only on follow-up studies (95% CI: 0.01, 0.05; I2 0%; Fig. 2B) (see Fig. S1 (a) and S1 (b) on Supplementary material).

Quality assessment

Overall, the ten retrospective cohorts and one case–control study included in this analysis were considered to have a moderate risk of bias. They predominantly lacked to classify and clarify intervention methods (WBMRI protocols), thereby failing to meet the specified criteria for the third domain. The funnel plot analysis revealed no indication of a publication bias (see Fig. S5 Supplementary Material for the Funnel Plot for publication bias).

Sensitive analysis

Sensitivity analyses were conducted for the endpoint with high heterogeneity. Specifically, when examining the estimated cancer diagnosis rate among suspicious lesions (see Fig. 2), a leave-one-out analysis (Fig. S2 in the Supplementary Material) revealed that Tewattanarat and LiFe-Guard were responsible for increasing most of the heterogeneity [18, 21]. Yet, results remained consistent with overall analyses, even when each individual study was removed from the analysis.

Discussion

This systematic review with meta-analysis evaluates the utility of WBMRI in the early detection of malignancies in asymptomatic patients with LFS. Through the analysis of data from 703 patients, the study revealed a detection rate of 31% for suspicious lesions and a 6% detection rate for malignant tumors, indicating the potential of WBMRI in the early identification of cancerous conditions. Further analysis showed that 18% of lesions identified by WBMRI were malignant, highlighting the method’s effectiveness in distinguishing malignant from benign findings. The incidence of cancer per person-round of WBMRI was determined to be 2%, suggesting a measurable impact of WBMRI on the surveillance of LFS patients. These findings contribute to the limited body of evidence supporting the benefits of WBMRI for cancer screening in this high-risk population and underscore the importance of incorporating WBMRI into surveillance protocols to facilitate early detection and potentially improve clinical outcomes in asymptomatic LFS patients.

WBMRI is widely used in oncology, and it has been proposed as a surveillance strategy for LFS patients [9, 25]. Villani et al [9] proposed a tumor surveillance protocol, known as the Toronto protocol, for both adults and children affected by LFS including the use of WBMRI for cancer screening. All the tumors were diagnosed in asymptomatic patients, who were all alive by the end of the study. Conversely, the survival rate for patients who did not undergo surveillance during the same period was 23%. In another study of WBMRI conducted by Anupindi et al [25] abnormal findings were detected in 18% of the patients examined. Of which, a papillary thyroid carcinoma was confirmed to carry a TP53 pathogenic variant and WBMRI had 100% sensitivity and 94% specificity [25, 26].

Emerging studies have suggested improved clinical outcomes for TP53 mutation carriers with intensive screening. A pilot study (SIGNIFY) aimed to assess the incidence of malignancies diagnosed in asymptomatic TP53 mutation carriers using a non-contrast WBMRI against general population controls and found a prevalence of 13.6% on initial MRI and two cases of simultaneous primary cancers in two participants, arguing for the adoption of at least a baseline WBMRI scan in the screening of TP53 mutation carriers [20, 27,28,29].

Current screening guidelines for patients with LFS include a combination of clinical, radiologic, and pathologic evaluation, and most of them do not include screening with CT or PET-CT [7, 23]. Patients with LFS are especially susceptible to adverse effects from radiation exposure, with an elevated risk of secondary malignancy in the area of radiation when compared with the general population [17, 19, 30, 31]. For this reason, available data support WBMRI for surveillance, because WBMRI reduces radiation and maintains high sensitivity and specificity [19].

In a previous meta-analysis [13], from the Li-Fraumeni Exploration Research Consortium, 578 patients with deleterious germline TP53 mutations were included. This study presented a detection rate of 7% using baseline WBMRI, similar to this meta-analysis. A substantial proportion of tumors identified by surveillance were low-grade or premalignant lesions, in a total of 225.

This meta-analysis has some limitations. The surveillance protocols and follow-up scans were not homogenous across studies. The eligibility criteria regarding time since curative treatment for a previous cancer also differed in each study. To overcome some limitations, we considered all lesions that remained under investigation suspicious, and benign and incidental findings were not considered. In addition, we lack enough data to estimate the sensitivity and specificity, accuracy, and the number needed to screen within a surveillance protocol through WBMRI. Furthermore, longer follow-up periods will be important to determine whether occult cancers were missed and to assess for safety issues associated with WBMRI screening.

In conclusion, this Systematic Review and Meta-Analysis demonstrated that surveillance with WBMRI can be successful in diagnosing cancers in the early-disease stage, achieving an overall detection rate of 6% in patients with LFS, similar to the previous meta-analysis. This supports its use as a valuable diagnostic tool for early cancer detection in LFS asymptomatic patients, and improving patients’ outcomes through surveillance protocols. Further studies are needed with long-term follow-up.