Low prevalence of germline TP53 and PALB2 mutations in unselected cohort of breast cancer patients from Brunei Darussalam

Study population

The study population includes unselected incident and prevalent breast cancer patients seen from 2017 till 2018 at The Brunei Cancer Centre (TBCC), Pantai Jerudong Specialist Centre (PJSC), the only cancer referral hospital in Negara Brunei Darussalam. Terminally-ill patients were excluded from the study. Among 218 patients who were seen during this period, 89 patients were approached where only 54 patients consented to participate in the study. All study participants provided written informed consent. Peripheral blood samples, demographic, and detailed family history data were collected from consenting patients. A retrospective review of the recruited study participants’ medical and histopathology reports was performed at TBCC after acquiring access permission from the relevant authorities.

Ethics statement and consent

This study was approved by the Medical and Health Research and Ethics Committee (MHREC) of Brunei Darussalam [MHREC/Edu/2016/4(3)]. Written informed consent for publication of patients’ details was obtained from the patients or through consent by proxy.

Next-generation sequencing (NGS)

Genomic DNA (gDNA) was extracted and purified from the collected peripheral blood sample using Wizard Genomic DNA purification kit (Promega) following manufacturer’s protocol. The DNA samples were then outsourced to Cancer Research Malaysia (CRM) for BRCA1, BRCA2, TP53, and PALB2 targeted panel sequencing.

The HBOC_4_v2 gene panel used was developed by CRM and the University of Melbourne and was used to screen for all coding exons ±2 bp intronic sequence of the BRCA1 (NM_007294.3), BRCA2 (NM_000059.3), PALB2 (NM_024675.3), and TP53 (NM_000546.4) genes. For each sample, 50 ng of gDNA were screened for germline mutations using Hi-PlexNGS platform. The targeted sequencing library was prepared using an amplicon-based approach and sequenced on a MiSeq (Illumina, USA), 2×300bp paired-end sequencing using MiSeq Reagent kit v2 300 cycles. On-target coverage parameter for sequencing was computed using Bedtools (2.17.0). Samples considered as successfully sequenced were those with >95% of amplicons with ≥10 read-pairs (≥20× read depth).

Raw sequencing reads were mapped to the human genome (hg19) using Bowtie2 (2.3.2). Variants were called using ROVER, annotated with ANNOVAR, and according to the Human Genome Variation Society (HGVS) nomenclature using Mutalyzer (2.0.26). Identified variants were classified and assigned to one of the following classes: (1) benign; (2) likely benign; (3) variants of uncertain significance; (4) likely pathogenic; or (5) pathogenic. Class 1 and Class 2 variants were not reported to end-user by CRM. Class 3 variants were reported as equivocal. Class 4 and Class 5 variants were reported as pathogenic mutations. All variants other than neutral polymorphisms were confirmed by Sanger sequencing. Bioinformatic analysis was performed by CRM and reports on the variant results excluding neutral polymorphisms (Class 1 and 2) were delivered upon completion.

Bioinformatics analysis

Although CRM have analysed and validated the results, the data were re-analysed as neutral polymorphisms (Class 1 and Class 2 variants) were not included in the provided reports. Integrative Genomics Viewer (IGV) v2.5.2 (Broad Institute) was used to view the binary files provided by CRM. The data were analysed using the public server at usegalaxy.eu.⁵⁶ Variants were called using varscan (Galaxy Version 2.4.2) and bcftools call (Galaxy Version 1.9+galaxy 1). The data generated by bcf tools call were filtered using VCF filter. All variants identified were annotated using the SNPeff Eff tool (Galaxy Version 4.3+T.galaxy1).

All variants identified were annotated according to the nomenclature used by HGVS recommendation guidelines, using the A of the ATG translation initiation codon as nucleotide +1. All identified missense variants were analysed in-silico using SIFT,⁵⁷ PolyPhen-2,⁵⁸ CADD,⁵⁹ FATHMM-MKL,⁶⁰ and DANN⁶¹ to predict the effect of amino acid substitution. Each prediction tools scored the variant as damaging or benign/neutral/tolerated. All variants identified in this study were also checked against the NCBI ClinVar, Varsome, and population frequency databases (gnomAD and 1000 genome).

Variant pathogenicity was explored and derived on the basis of the ClinVar database, population data, IARC TP53 database (R20, July 2019),⁶² UMD TP53 Mutation database (2017_R2), in-silico data, and functional data if available following the American College of Medical Genetics and Genomics and the Association for Molecular Pathology (ACMG/AMP) guidelines.⁶³ In this study, a variant was considered a pathogenic damaging mutation if it was a protein-truncating mutation caused by deleterious or frameshift mutation, or a missense mutation which has a confirmed association with the disease, or a missense variant which has been classified as likely pathogenic according to the ACMG standards and guidelines. We selectively Sanger sequenced only pathogenic mutations identified using NGS via our own ABI 3500 Genetic Analyzer.

Statistical analysis

Continuous data were presented as median and range. Chi-square or Fisher exact test was used to analyse the association between two independent variables in the population. The statistical analysis was performed using SPSS version 17.0 software for Windows. A p-value <0.05 was considered statistically significant.

Prevalence of TP53 and PALB2 germline mutations in unselected breast cancer patients

Although the population included in this study was sequenced for the four genes, this article only reports the prevalence and mutation spectrum of the TP53 and PALB2 genes. The prevalence and mutation spectrum of the BRCA1 and BRCA2 genes for these patients are presented elsewhere (unpublished reports, Siti Nur Idayu Matusin).

Of the 54 unselected breast cancer patients included in this study, two (3.7%) had germline BRCA2 deleterious mutations. A total of seven (7) variants were identified in the study population (Table 3). Two (28.6%) and five (71.4%) were TP53 and PALB2 variants respectively. No deleterious mutations were identified, and all the seven identified missense variants have been reported.

Evaluation of variants pathogenicity

Two (2) variants were classified as benign and five (5) as variants of uncertain significance (VUS) (Table 4). Among the VUS variants, four (4) had conflicting evidence for pathogenic and benign criteria, while one (1) did not have enough evidence for its pathogenicity. The one TP53 VUS variant was identified in one patient and the four PALB2 VUS variants were identified in a total of five patients. Thus, the prevalence of TP53 and PALB2 VUS carriers among Brunei breast cancer patients were 1.9% and 9.3%, respectively. All the identified VUS were found to be rare in the general population (<0.05% in gnomAD and/or 1000 Genome).

The two benign variants, TP53 c.215C>G and PALB2 c.1676A>G, were found in 75.9% and 37% of the study population, respectively. Both variants have been reported >10% in the global populations (rs1042522 and rs152451) suggesting the chances of these variants contributing to breast cancer susceptibility to be very low. Furthermore, the clinical significance of these variants has been depicted as benign in the NCBI ClinVar database, and all the in silico tools agreed that they had no impact on the TP53 and PALB2 protein, respectively.

Missense VUS variants

TP53 c.79C>T caused the change of the highly conserved proline at codon 27 to serine which was shown to increase TP53 protein binding affinity to MDM2 due to TP53-P27S being more α-helical compared to the wildtype TP53.^65,66 Functional yeast-based assays of this variant have shown that the TP53 protein was able to retain >80% of its transcriptional transactivation activity for the tested promoters, except for WAF (CDKN1A) and AIP where the residual TP53 transcriptional transactivation activity range at 41-50% and 61-70%,^67,68 respectively. Four of six in silico tools predicted a damaging effect of the variant on protein function (Table 4). In this study, the variant was found in a 59-year-old patient who exhibited a history of ≥2 familial colon cancer with poor prognosis. Germline TP53 mutations are associated with Li-Fraumeni syndrome and show poor prognosis in several types of cancers, including colon cancer.^15,16 The contribution of this variant in the occurrence of colon cancer in this patient’s family members is unknown. Collectively, the evidence provided with this variant was still insufficient to determine its clinical significance.

PALB2 c.149A>C results in a change of Lysine to Threonine at codon 50. All the in silico tools in this study predicted a damaging effect of the variant on protein function (Table 4). To our knowledge, no experimental functional assay demonstrating the impact of this variant on PALB2 protein function has been reported. In our study, the variant was identified in a 35-years old patient who did not have any family history of cancer. Together with the available population data (Table 4) on the variant occurrences in the general population, there is insufficient evidence to determine a definite conclusion on the variant’s significance.

PALB2 c.2289G>C results in a change of leucine at codon 763 to phenylalanine. Four of six in silico tools predict a damaging effect on protein function (Table 4). This variant has been reported in the literature in individuals affected with Hereditary Breast and Ovarian Cancer (HBOC) Syndrome, most of whom were of the East-Asian populations.^21,41 However, these studies do not provide clear conclusions on the association of the variant with HBOC syndrome i.e., the variant was classified as VUS. A recent study reported L763F as a moderate-risk somatic missense mutation.⁶⁹ A cellular homology directed DNA repair (HDR)-based assay showed no damaging effect of this variant on PALB2 protein function.⁷⁰ Based on the current available evidence, the criteria for benign and pathogenic are contradictory for this variant. In our study, this variant was identified in a 42-year-old patient with a family history of blood and colon cancer.

PALB2 c.2474G>C results in a change of arginine at codon 825 to threonine. Five of six insilico tools predict a benign effect of the variant on protein function (Table 4). To our knowledge, no experimental functional assay demonstrating the impact of this variant on PALB2 protein function has been reported. In our study, this variant is found to co-occur in an early-onset HBOC syndrome patient with a BRCA2 germline pathogenic variant c.5164_5165delAG (S1722Yfs*1725) providing supporting evidence for a benign role of this variant in the HBOC syndrome. However, the available population data (Table 4) on the variant occurrences in the general population are insufficient to allow a definite conclusion on the variant significance.

PALB2 c.3054G>C results in a change of glutamate at codon 1018 to aspartate. Five of six in silico tools predict a damaging effect of the variant on protein function (Table 4). To our knowledge, only one functional study has been conducted on this variant, and it was shown that it did not impair HDR, and also have no impact on the maintenance of IR-induced G2/M checkpoint.⁷¹ This variant has been reported in the literature in individuals affected with HBOC Syndrome, most of whom were of the East-Asian populations.^{21,26,32,33,37,41,42,45} However, these studies do not provide clear conclusions on the association of the variant with HBOC syndrome i.e., the variant was classified as either likely pathogenic based on prediction tools or VUS. The population data, specifically in the East-Asian population, suggested the variant is a benign polymorphism. However, the currently available evidence is still insufficient to fulfil the ACMG criteria for a definite benign or pathogenic classification (Table 4). In our study, this variant was identified in two unrelated patients: a 71-year-old breast cancer patient with a family history of neck and colorectal cancers, and a 45-years old breast cancer patient who did not have any family history of cancers.

TP53 P72R and response to chemotherapy

TP53 polymorphisms have been studied numerously in breast cancer.In one study of a cohort of 40 female patients no significant correlation was found between the studied polymorphisms and risk of premenopausal breast cancer.⁷² TP53 codon 72 is a hotspot of polymorphisms with rs1042522 (Pro72Arg) being the most studied SNP. The benign Pro72Arg (P72R) was located in the proline-rich region of the TP53, and the substitution directly affects the structure of the SH3-binding domain. TP53-R72 was shown to be more efficient in inducing apoptosis which influenced an increased longevity in humans, whereas TP53-P72 increased the induction of G1 arrest and was better at activating TP53-dependent DNA repair pathway.⁷³ The SNP has been shown to not have any significant association with breast cancer risk.⁷⁴ Functional analysis of P72R on TP53 protein function, including transcriptional activity, apoptosis, and cell proliferation assays performed both in human cells or in yeast cells showed no defective impact.⁷⁵ Interestingly, an in vivo study showed the potential mechanisms i.e., chronic inflammation, by which TP53 R72 variant contribute to enhanced breast cancer susceptibility.⁷⁶

As one of the common SNPs to be identified in different cancers, substantial studies have been conducted to measure the efficacy of drug response in the presence of this variant with other SNPs in different genes in different cancers. Breast cancer patients with P72P variant were shown to be less sensitive to anthracycline-based neoadjuvant treatment compared to heterozygous and homozygous P72R variants^77,78 and developed more toxicity.⁷⁸ The P/P genotype at codon 72 of TP53 was also reported to be associated with poor disease-free survival (DFS), particularly in patients who received adjuvant chemotherapy.⁷⁹

In the study, 41 of the study population were TP53-P72R carriers (Table 3). Applying the Hardy-Weinberg equilibrium, the genotype distributions of TP53 codon 72 variants in our study showed 15 (27.8%) were R/R, 26 (48.1%) were P/R, and 13 (24.1%) were P/P while the allelic frequencies were 0.519 (n = 56) for R and 0.418 (n = 52) for P. A total of 30 (55.6%) patients received neoadjuvant chemotherapy while the rest received adjuvant chemotherapy. Patients receiving neoadjuvant chemotherapy were assessed for their tumour pathological response based on histopathological criteria.⁸⁰ A complete response (Grade 3) was defined as no evidence of residual – and invasive if present – disease. A marked response (Grade 2) was defined as ≥50% reduction in tumour size with apparent or a few remaining cancer cells. A slight response (Grade 1) was defined as mild or moderate response to the treatment with marked changes in <50% of cancer cells. An absence of response (Grade 0) was defined as no change in cancer cells after treatment. All patients were assessed for adverse reaction to chemotherapy based on the Common Terminology Criteria for Adverse Events (CTCAE) v5.0 (National Cancer Institute).

Of the 30 patients who received neoadjuvant chemotherapy, 15 (50%) received anthracycline-based (with or without taxane) regimens, 5 (16.7%) received taxane-based regimens, 9 (30%) received taxane/platinum-based (with or without anthracycline) regimens, and 1 (3.3%) patient with missing record. Of the 29 patients, the tumour pathological response of five patients were not available as they had not completed their treatment cycles yet, giving a total of only 24 patients data to be assessed. Two (8.3%) patients achieved complete response, three (12.5%) patients showed no response to the treatment received, while three (12.5%) and sixteen (66.7%) showed slight and marked response, respectively. Due to the small number of patients, the response to neoadjuvant treatment and the TP53 codon 72 status were grouped into two groups to increase the reliability of the statistical analysis. There was no significant association between the TP53 codon 72 status of these patients with their tumour pathological response to neoadjuvant treatment (p = 0.277, Table 5).

Records for 2/54 patients with P72R variant on any development of adverse reaction to treatment were not available for analysis giving a total of only 52 patients records to be assessed. A total of 20 (38.5%) patients presented any grade of adverse reaction to chemotherapy, with one (5%) of them experiencing a severe Grade 3 reaction to treatment. There was no significant association between the TP53 codon 72 status with the development of adverse reactions to treatment in these patients (p = 0.510, Table 5).