The goal of our study was to determine if the type of TP53 somatic mutation (GOF or LOF, DNE- or DNE+, hotspot status, and CpG nucleotide position) varied in frequency between patients of different ancestry. Considering that the overall rate of somatic TP53 mutations in breast cancer differs by race (15–18), this is an important concern for study of TP53-mutant breast tumors and differences in outcomes and treatment response by race. We identified a modest difference between AA and NHW individuals, with NHWs slightly more likely to have GOF mutations. DNE- TP53 somatic mutations were associated with TNBC and ER.
Our finding that TP53 mutations without DNE activity were associated with TNBC (p = 0.03) and ER status (p = 0.005) is novel. From previous studies, ER- and TNBC have a higher frequency of TP53 mutations compared to ER + tumors (13). In this study, that only includes TP53-mutant tumors, we observed a higher proportion of ER- and TNBC tumors overall compared to unselected populations. This is consistent with previous studies that identified TP53 somatic mutations in 40–60% of all breast tumors versus ~ 85% of TNBC (1–3, 5). There has been some debate about the significance of mutant TP53 DNE versus GOF activity, as many common somatic mutations, including hotspot mutations, are both DNE + and GOF. Overall the importance of DNE + versus GOF is highly dependent on the specific mutation type, genetic and cellular context, loss of heterozygosity, and the phenotype evaluated (31). It is thus of great interest that the association with receptor status was only significant for DNE; there was not a significant association with receptor status or GOF/LOF. Thus mutant TP53 DNE may be a more important component of tumor subtype, though functional studies are needed to better understand this phenomenon.
Our cohort included somatic TP53 mutation data from TCGA, METABRIC, and IARC databases, studies identified for inclusion from literature, and 351 previously unpublished cases (Additional File 1). The frequency of hotspot mutations observed in our study (20%) was slightly lower than previous studies finding that 28% of TP53 mutations occurred at mutation hotspots (7). We observed that 36% of tumors from NHW individuals had GOF mutations compared to 29% in AA individuals (p = 0.04). This is opposite of what we expected to find as missense/GOF variants have been associated with poorer prognosis or worse outcomes in previous studies (9). We considered that this effect may be an artifact of more NHW patients sequenced with earlier technology, such as Sanger, which could bias the TP53 mutation detected to the exons more likely to have GOF mutations. However, there was no difference in use of Sanger vs NGS between these population groups, with 43.6% of NHW patients sequenced with Sanger, compared to 43.8% of AA patients. There also was no notable difference in the number of exons sequenced; 67.3% of NHW patients had at least exons 2–11 sequenced, compared to 68.6% of AA patients. Additionally, there was no difference in the percentage of unclassified variants between groups (7% in AA versus 9.9% in NHW for GOF/LOF, 14.3% in AA versus 17.3% in NHW for DNE). Thus, this difference does not appear to be due to technological differences in mutation detection or in mutation classification. Further studies of larger numbers of AA and NHW women are warranted to confirm this finding.
Participants with hotspot mutations were younger than those with non-hotspot mutations, with a mean age of 53.61 in hotspots versus 55.04 in non-hotspots, but this was not statistically significant (p = 0.065). Age did not correlate with DNE or GOF/LOF. This finding is somewhat unexpected. Susceptibility to hotspot mutations is likely due to properties of the genetic sequence being vulnerable to mutation, rather than purely selective growth advantage of tumor cells (7). A high proportion of hotspot mutations are CpG sites, a feature of mutation signature 1, which correlates with age, so it would seem more likely for somatic hotspot mutations at CpG sites to be associated with later age at diagnosis (8). However, a correlation for breast cancer has not yet been reported in the literature of which we are aware. (1). In Li-Fraumeni syndrome, germline TP53 missense mutations have been associated with an earlier age of breast cancer diagnosis compared to frameshift and nonsense mutations (32, 33). Furthermore, in one study, age of diagnosis was not associated with hotspot mutations in ovarian cancer (34). We did not have sufficient numbers of individuals with stage information to evaluate whether there was a correlation with age, stage and hotspot status. Data from this study and others suggest that age is not the most critical contributor to TP53 hotspot mutations at CpG sites, warranting additional study.
Strengths of this study include the large number of women included for study. Previous studies characterizing TP53 mutation types have not focused on race or ancestry and have relied on TCGA or IARC datasets alone. This study incorporated multiple sources including previously unpublished data. We limited the dataset to only include tumors with TP53 somatic mutations and only included participants with race or ancestry data. There are a number of limitations to this study. Many of the studies used self-reported race and ethnicity information, which may not reflect genetic ancestry, and may have been categorized differently by study depending on data collection method, such as distinguishing between NHW and Ashkenazi Jewish ethnicity. There may be differences in TP53 mutation types between ethnic groups within a racial group, such as between individuals of European ancestry from Greece and Finland. For countries that are predominantly one racial group and for which detailed racial information was not available from the study authors, we made an assumption that the individuals in that study were very likely to be of that racial group (e.g. Norway and European ancestry; China and Asian ancestry). Some older studies that used single-strand conformation polymorphism followed by Sanger to identify mutations may have missed some mutations. A small number of studies only performed analyses of exons 4 through 8 which could miss more LOF (splice, nonsense and frameshift) variants that occur in other exons compared to GOF or DNE-associated missense variants that predominantly map to these exons. Because of the mixed data sources, this study did not include large copy number losses or exon specific deletions as variants for study resulting in fewer TP53 LOF variants being included. As mutation signatures resulting in specific mutation types could have a basis in biological or environmental differences between races, we may have missed racial specific differences in mutations resulting in large DNA gains or losses. Finally, classifications of variants as GOF/ LOF and DNE were made based on studies in the literature. For some variants there was discordant information; for this study, we used the classifications from studies that were larger (tested more variants) or studies that included a larger number of different assays. It is possible that some of the rarer missense variants included were misclassified or may act differently in humans than in the system tested (e.g. yeast).