Case selection
This retrospective study included 113 consecutively collected cases of TNBC who underwent surgery at Seoul St. Mary's Hospital between 2018 and 2022. This study was approved by the Institutional Research Ethics Board of the Catholic University of Korea with a waiver of patient’s informed consent (IRB No. KC22SISI0716). Cases who underwent neoadjuvant therapy or had a tumor size of less than 1.5 cm in its greatest dimension were excluded.
Confirmation of TNBC cases involved IHC for ER (clone SP1; cat. no. 790-4324; Roche/Ventana, ready to use), PR (clone 1E2; cat. No. 790-4296; Roche/Ventana, ready to use), and HER2 (clone 4B5; cat. No. 790-4493; Roche/Ventana, ready to use). Additional HER2 silver in situ hybridization (SISH), when necessary, was conducted using the VENTANA HER2 Dual ISH DNA Probe Cocktail. Classification of histological types for cases was based on the 5th edition of the WHO classification of breast tumors34. The histological type, histological grade, and tumor stage, as per the 8th edition of the American Joint Committee on Cancer (AJCC) TNM classification, were retrieved from their pathology reports.
Immunohistochemistry
Tissue acquisition to fixation time was minimized to within 2 hours. Samples were then fixed in 10% neutral buffered formalin (NBF) for 6 – 72 hours and sliced at 5 mm intervals. Sections of 4-µm-thick were obtained from paraffin-embedded blocks, deparaffinized in xylene, and rehydrated through a graded series of ethanol. P53 IHC was performed for 113 TNBC cases with DO7 antibody (cat. No. 800-2912; Roche Diagnostics, IN, USA; ready to use) and a Leica Bond III system (Leica Biosystems, Inc., Buffalo Grove, IL, USA) following the manufacturer’s protocol.
One of four experienced pathologists reviewed p53 IHC whole slides while blinded to tumor's TP53 mutation status. The p53 IHC staining pattern was manually assessed based on nuclear expression in tumor cells. The proportion of expression across all tumor cells was manually quantified in 5% increments for each case, ranging from 0 to 100%. Three other pathologists reviewed assessments. Cases with discordant measures were discussed until a consensus was reached. To define a cutoff for p53 IHC overexpression, a p53 expression rate-dependent receiver operating characteristic (ROC) curve was analyzed using TP53 mutation status as the endpoint to optimize the Youden index.
Whole exome sequencing
Whole exome sequencing was conducted for formalin-fixed, paraffin-embedded tissue blocks of a representative normal and tumor slide for the 113 TNBC cases. Paired-end sequences were generated with a NovaSeq6000 (Illumina, San Diego, CA, USA) platform following the manufacturer's protocols. The sequencing quality was checked using FastQC. Sequence reads were aligned to human genome assembly hg19 (GRCh37) using Burrows-Wheeler Aligner (BWA MEM, http://bio-bwa.sourceforge.net/), specifically the BWA-MEM algorithm35.
Paired normal and tumor sequence reads were aligned and processed together to identify somatic TP53 mutations. MuTect2 was used to detect somatic mutations. Mutations were filtered using the filtration tool of the Genome Analysis Toolkit (GATK). Functional annotation of filtered mutations was performed with SnpEff. Interpretation was performed with dbSNP and SNPs from the 1000 Genome Project. Further annotation with databases including ESP6500, ClinVar, dbNSFP, and American College of Medical Genetics and Genomics (ACMG) information was performed using an in-house program and SnpEff. VCF files of somatic mutations were processed to MAF files using the vcf2maf program. The ‘maftools’ R package was used to filter out synonymous mutations.
Copy number and loss of heterozygosity analysis
Copy number analysis of TP53 was performed using GISTIC 2.0 (Genomic Identification of Significant Targets in Cancer, RRID:SCR_000151) applied to segmentation files obtained from Sequenza analysis36,37. The analysis was undertaken to pinpoint significant loci of somatic CNVs. The analysis was configured with a focal length cutoff of 0.5 to focus solely on focal alterations. A confidence level of 0.9 was set to ensure high reliability of findings, along with a q-value threshold of 0.05 to distinguish between significant and non-significant results. Additionally, the X chromosome was excluded to eliminate variations related to sex chromosomes. The algorithm calculated G-scores for each region, taking into account the amplitude of copy number changes and their frequency across the sample set. Subsequently, it categorized these changes into homozygous deletions, heterozygous deletions, low-level gains, and amplifications.
To characterize the LOH at the TP53 locus, the GenomicRanges package in R was employed to process segmented CNV data from whole-exome sequencing. We annotated segmented data for traceability and determined LOH status for each segment by assessing the absence of heterozygosity. This binary information was then integrated into a dataset. Specifically, segments overlapping with the TP53 gene on chromosome 17, indicating LOH at the TP53 locus, were identified. These data were compiled into a TP53-specific LOH dataset for further analysis. If LOH occurred without a change in copy number in TP53, it was designated as copy neutral LOH.
Database for functional classification of TP53 mutation
Information on functional properties of TP53 mutations was compiled in the NCI TP53 database (R20 version; https://tp53.isb-cgc.org)38. Using this database, we systematically explored and analyzed functional properties of mutant p53. In this dataset, data were extracted from publications that reported functional assessment of mutant p53 proteins in human or yeast cells covering various aspects, such as transcriptional activities on p53-response elements, DNE on activities of wild-type p53, the ability to transactivate promoters not induced by wild-type p53, and the capability to promote cell growth and confer tumorigenicity. This assessment was conducted either by transfection and overexpression of mutant proteins or by evaluating endogenous mutants.
Based on functional outcomes in the experimental database, TP53 mutations in our cases were categorized into either LOF, DNE, or GOF. Specifically, we considered LOF when the mutant protein resulted in the loss of functional properties of wild-type p53. DNE was considered when mutant proteins inhibited the wild-type protein in transactivation or cell growth assays. GOF encompassed functional properties exhibited by the mutant protein but not by the wild-type counterpart. In the current study, a mutant p53 was classified as having GOF when it met at least one of the following categories: 1) tumorigenic property (in nude mice) in transfected cells; 2) interference with p73 activity; 3) transactivation of genes repressed by wild-type p53; 4) resistance to a cytotoxic drug; 5) increase growth rate; 6) cooperation with an oncogene for the transformation of rat embryonic fibroblast or mouse embryonic fibroblast cells; and 7) alteration of mutant p53 stability and activity by impairing regulators, including HSC70 and MDM221,39.
Statistical analysis
The Youden Index was utilized to determine the optimal cutoff value for p53 overexpression, defined as sensitivity + specificity − 1. The performance of the p53 expression cutoff point in predicting TP53 mutation was assessed using AUC, sensitivity, specificity, PPV, negative predictive value, and accuracy. The relationship between p53 IHC pattern and TP53 mutation type was assessed using the Chi-square test or Fisher's exact test, as appropriate. Results were considered significant if the p-value was equal to or less than 0.05. All statistical analyses were conducted using R version 4.3.1 (R Foundation for Statistical Computing, Vienna, Austria).