Evolution of germline TP53 variant classification in children with cancer

Abstract

Li-Fraumeni syndrome, caused by germline pathogenic variants in TP53, results in susceptibility to multiple cancers. Variants of uncertain significance (VUS) and reclassification of variants over time pose management concerns given improved survival with cancer surveillance for LFS patients. We describe the experience of TP53 variant reclassification at a pediatric cancer center. Methods: We reviewed medical records (2010–2019) of 756 patients seen in Texas Children’s Cancer Genetics Clinic. We noted initial TP53 classification and any reclassifications. We then classified TP53 variants following ClinGen TP53 variant curation expert panel recommendations using data from ClinVar, medical literature and IARC database. Results: Of 234 patients tested for TP53, 27 (11.5%) reports contained pathogenic/likely pathogenic (P/LP) variants and 7 (3)% contained VUS. By January 2022, 4 of 6 unique VUS and 2 of 16 unique P/LP variants changed interpretations in ClinVar. Reinterpretation of these 4 VUS in ClinVar matched clinical decision at the time of initial report. Applying TP53 VCEP specifications classified 3 VUS to P/LP/benign, and one pathogenic variant to likely benign. Conclusions: Planned review of variant significance is essential, especially for patients with high probability of LFS.

Introduction:

Li-Fraumeni syndrome (LFS) is caused by germline pathogenic variants of the TP53 tumor suppressor gene and results in susceptibility to a broad range of pediatric and adult onset cancers. The lifetime risk of cancer is very high, estimated at ≥70% for men and ≥90% for women [1, 2]. Five cancer types account for the majority of LFS tumors: adrenocortical carcinomas, breast cancer, central nervous system tumorss, osteosarcomas, and soft-tissue sarcomas. Hypodiploid acute lymphoblastic leukemia is suggestive of LFS, although hyperdiploid and diploid cases have been reported[3–5]. The diagnosis of LFS is established in a proband who meets classic LFS criteria and/or has a germline pathogenic variant in TP53 identified by molecular genetic testing [3]. Life-long, intensive surveillance for germline TP53 pathogenic variant carriers, termed the Toronto Protocol, aimed at early tumor detection, has been shown to improve survival [6, 7].

The majority (90%) of pathogenic variants include missense variants which most commonly occur in the DNA-binding domain between amino-acid residues 102–292. These can be further classified into variants that disrupt amino acids directly contacting DNA and variants that alter the conformation of the DNA-binding domain [8]. Another 10% of pathogenic changes are typical loss of function variants: nonsense or frameshift variants and deletions.

TP53 variants of uncertain significance (VUS) pose a difficult dilemma as a clinician may be uncertain whether to implement this intensive surveillance and downstream testing of family members. As more data is obtained, VUS can be reclassified as pathogenic or benign. The TP53 Variant Curation Expert Panel (VCEP) of the Clinical Genome Resource (ClinGen) has recently published germline TP53 specifications to the ACMG/AMP guidelines in order to improve consistency in variant interpretation and ultimately reduce the number of VUS or conflicting interpretations [9].

Prior studies of VUS in cancer susceptibility genes (primarily identified in adult populations) have demonstrated that the majority of reclassification is towards benign[10–12]. Less research has been done on reclassification of variants derived from pediatric cancer populations with a higher probability of LFS. In this analysis, we describe the experience with variant reclassification and evolution of TP53 results over a ten year period at the Texas Children’s Cancer Genetics Clinic. We focus on VUS and other reclassified variants, and consider features that suggest reclassification to pathogenic versus benign.

Methods:

After obtaining IRB approval, we reviewed medical records of pediatric patients seen in Texas Children’s cancer genetics clinic from Jan 2010 through Dec 2019 for individuals who underwent clinical TP53 testing. Data (including age, diagnosis, family history, genetic testing results, and relevant follow up) were collected on individuals with germline TP53 pathogenic (P), likely pathogenic (LP) or VUS variants. Our clinical interpretation at the time of initial report, which determined LFS screening and management, was also collected. Testing was performed at a variety of clinical laboratories with diagnostic methods changing over time although the majority of testing laboratories included both sequence analysis and deletion/duplication analysis.

We then searched the literature, ClinVar database [13] in January 2022, and TP53 International Agency for Research on Cancer (IARC) database [14] for changes in variant classification through subsequent years. Independently, we applied the recently published ClinGen TP53 VCEP specifications [9, 15] to update classifications and resolve variants with initial VUS classification or conflicting interpretations in ClinVar.

Results:

Of 756 patient records from 2010 to 2019, 234 individuals (30.9%) underwent TP53 gene analysis (Fig 1a Flow Chart). Indications for testing included personal or family history of cancers associated with LFS in 127 individuals (54%), and family history of LFS in 40 individuals (17%). TP53 testing was also available as a part of exome sequencing or gene panels for 41 individuals (18%) with cancers atypical of LFS, such as retinoblastoma, rhabdoid tumor, meningioma or paraganglioma, and for 26 individuals (11%) with non-cancerous indications, such as colon or thyroid polyps, bone marrow failure syndromes and overgrowth syndromes.

Figure 1.

A) Flow chart outlining results of the review of medical records as described in the methods. B) Lollipop plot of TP53 variants reported. The top diagram demonstrates pathogenic (P) and likely pathogenic (LP) variants, whereas the bottom diagram demonstrated variants of uncertain significance (VUS). Missense variants are represented in green, Nonsense variants are represnted in black, synonymous substitution affecting splicing is represented in pink. The proposed extent of the duplication described in the paper and the location of the intronic VUS are also demonstarated in the bottom diagram.

Twenty-seven individuals (11.5%) were found to have TP53 P or LP variants (16 unique variants). Familial testing was available for 22 out of the 27 individuals. Of these, three individuals had de novo changes (~14%), and the rest were inherited (~86%). Seven individuals (3%) were found to have VUS (6 unique variants). Familial testing was available for 4 out of the 7 individuals. Of these, two individuals had de novo changes, and two siblings had an inherited change. De novo status was known at the time of initial report, and did not change VUS interpretation. Results are outlined in Fig 1a. Supplementary Table 1 contains all nonbenign TP53 variants with indication for testing, clinical interpretation by a medical geneticist at initial report, number of family members (including the proband) tested positive for the specific variant, original lab interpretation, and current (2022) interpretation in ClinVar. Classifications by applying the new ClinGen TP53 VCEP specifications, including evidence codes applied are summarized in Supplementary Table 2.

The 16 P/LP unique variants included 12 missense variants, three frameshift variants, and one synonymous splicing variant. Eleven of the 12 P/LP missense variants (92%) were located in the DNA-binding domain (Figure 1b). Over this period, 14/16 (87%) variants have maintained P/LP interpretation in ClinVar. Surprisingly, changes in interpretation of pathogenicity were submitted to ClinVar for 2 variants (Table 1). The variant c.469G>T (p.Val157Phe) has conflicting interpretations of pathogenicity, with one VUS submission from a clinical lab in 2020, and two prior LP submissions. This variant was identified in a patient with astrocytoma and significant family history for LFS-related tumors, therefore with high prior probability for LFS. Applying the ClinGen TP53 VCEP specifications still resulted in pathogenic classification (Supplementary Table 2). The second variant, c.868C>T (p.Arg290Cys) was originally classified in 2013 as “mutation” by the lab, and was an unexpected finding from a panel test of a patient diagnosed with CLOVES syndrome who had no personal or family history of cancers. Current interpretation in ClinVar has changed to LB after review by the ClinGen TP53 expert panel. Applying the ClinGen TP53 VCEP specifications resulted in similar LB classification based on in-silico predictors (BP4) and functional assays (BP3). After the original lab interpretation of TP53 as “mutation”, the patient was followed with the intensive Toronto protocol for several years before reevaluation of his variant, at which time screening was discontinued. This individual continues to be cancer free with a follow up of 7 years.

Table 1:

TP53 variants of uncertain significant (VUS) and pathogenic variants, which changed classifications over time

TP53 Variant NM_000546.4	Indication	Clinical interpretation as LFS	no. family members with variant1	Year	Original lab report	ClinVar		Classification according to TP53 VCEP specifications
						Interpretation in 2022	Review status2
c.467G>A p.Arg156His	Juvenile polyposis syndrome	No	1	2014	VUS	Uncertain Significance (8 VUS, 1 LB)	***	VUS
c.469G>T p.Val157Phe	Astrocytoma FH of LFS	Yes	2	2015	LP	CI (2 LP, 1 VUS)	*	P
c.470T>C p.Val157Ala	Adrenocortical carcinoma	Yes	2	2012	VUS	LP	*	LP
c.710T>A p.Met237Lys	Choroid plexus ca AML	Yes	1	2019	VUS	LP	None	P
c.868C>T p.Arg290Cys	CLOVES syndrome	Yes	1	2013	Mutation3	LB	***	LB
c.1022T>A p.Phe341Tyr	Rhabdomyosarcoma	No	1	2012	VUS	Not reported	N/A	VUS
c.1101–49C>T N/A	Medulloblastoma	No	1	2013	VUS	B	*	B
Exon 2–6 duplication	Hyperploid B ALL MDS shortly after treatment	Yes	1	2013	VUS	LP	*	VUS

LFS= Li-Fraumeni syndrome, P= pathogenic, LP= likely pathogenic, VUS= variants of incertain significance, CI= conflicting interpretations of pathogenicity, LB= Likely benign, B= benign, MDS= myelodysplastic syndrome

^1.Number refers to both the proband and any positive relative.

^2.ClinVar review status:

*** reviewed by expert panel

** criteria provided, multiple submitters, no conflicts

* criteria provided, conflicting interpretations

* criteria provided, single submitter

^3.“Mutation”- exact wording of the lab report in 2013

Four of the six unique VUS were missense variants, one was an intronic variant not affecting splicing (SpliceAI prediction), and one was a large duplication (Table 1). Three missense variants were in the DNA binding domain, including two variants, c.470T>C (p.Val157Ala) and c.710T>A (p.Met237Lys), now reported as LP in ClinVar, and one variant, c.467G>A (p.Arg156His), which is still reported as VUS (Table 1, Figure 1b). The exon 2–6 duplication is now reported as LP in ClinVar. The intronic variant, c.1101–49C>T, is now reported as benign. The remaining missense variant, c.1022T>A (p.Phe341Tyr), has not been reported in ClinVar. Applying the ClinGen TP53 VCEP specifications classified 3/6 VUS: the two missense variants currently reported as LP in ClinVar were classified as LP and P; the intronic variant, c.1101–49C>T was classified as benign, similar to its current interpretation in ClinVar, following stand alone BA1 criteria and a variant mean allele frequency of 1.1% in gnomAD (Table 1, Suppementary Table 2). Interestingy, this variant was reported as VUS by the lab in 2013, before large population allele frequency databases, such as ExAC [16] became available. Importantly, all cases with reclassifications from VUS to LP and P were consistent with the clinical interpretation made at the time of testing. Therefore, no changes to medical management were required.

Discussion:

With the increasing use of genetic testing for hereditary-cancer risk genes, information regarding variant pathogenicity is constantly evolving. Timely reevaluation of TP53 variants is especially important as enhanced surveillance protocols significantly improves survival of LFS patients, but requires substantial use of medical resources and patient time [6, 7]. Additionaly, VUS classification causes significant anxiety to carriers and health care providers [17, 18]. Hence, TP53 VUS as well as results that seem inconsistent with the clinical history must be periodically reviewed to ensure that management and counseling offered to the patient and the family is appropriate. For example, ClinGen requires VCEPs to reassess LP and VUS classifications at least every 2 years and benign classifications when new large population datasets are released [19]. This type of new population evidence was responsible for the reclassifications from pathogenic to benign as seen in one variant in our report. Misclassifications of benign variants as pathogenic may also result from lack of ancestry-matched asymptomatic controls [20].

As mentioned previously, studies of VUS reclassifications in multiple cancer susceptibility genes in adult populations report the majority are reclassified towards benign [10–12]. However, that may not be the case for the TP53 gene. Bittar [21] and colleagues analyzed TP53 variants reported in 1844 adult patients from cancer risk evaluation programs in Brazil with a VUS prevalence of 0.65%. Four of 12 VUS in this study were reclassified using ACMG guidelines and the Sherloc framework resulting in 2 LP classification [22].

In our small series, TP53 VUS constituted up to 3% of nonbenign variants found, with three of six variants changing interpretation to likely pathogenic in ClinVar, and one variant changing interpretation to benign. The changes to LP match the clinical interpretation and surveillance decisions made at the time of the initial report. Thorough clinical evaluation based on tumor diagnoses and family history should guide interpretation in cases where classification remains unclear.

Importantly, ClinGen TP53 VCEP specifications applied to 43 pilot variants reduced VUS classifications from 28% to 12% [9], highlighting the importance of the ClinGen expert panel process. Applying the ClinGen TP53 VCEP specifications, including applying functional assay data and/or in-silico predictions from A-GVGD and BayesDel, to eight variants from this study led to 6 non-VUS reclassification.

In summary, our report highlights that ongoing review of variant significance is essential, especially for pediatric cancer patients with high prior probability for pathogenic variants based on their tumor diagnosis and family history.

Highlights.

• Reclassification of TP53 variants over time may significantly alter management

• Applying the ClinGen VCEP specifications to 8 variants led to reclassification in six

• Ongoing review of variants with high prior probability of pathogenicity from cancer patients is essential