TP53 mutations as potential prognostic markers for specific cancers: analysis of data from The Cancer Genome Atlas and the International Agency for Research on Cancer TP53 Database

Mutations in the tumor suppressor gene TP53 are associated with a variety of cancers. Therefore, it is important to know the occurrence and prognostic effects of TP53 mutations in certain cancers. Over 29,000 cases from the April 2016 release of the International Agency for Research on Cancer (IARC) TP53 Database were analyzed to determine the distribution of germline and somatic mutations in the TP53 gene. Subsequently, 7,893 cancer cases were compiled in cBioPortal for Cancer Genomics from the 33 most recent The Cancer Genome Atlas (TCGA) studies to determine the prevalence of TP53 mutations in cancers and their effects on survival and disease-free survival times. The data were analyzed, and it was found that the majority of TP53 mutations were missense and the major mutational hotspots were located at codons 248, 273, 175, and 245 in exons 4–8 for somatic mutations with the addition of codon 337 and other mutations in exons 9–10 for germline mutations. Out of 33 TGCA studies, the effects of TP53 mutations were statistically significant in nine cancers (lung adenocarcinoma, hepatocellular carcinoma, head and neck squamous cell carcinoma, acute myeloid leukemia, clear cell renal cell carcinoma (RCC), papillary RCC, chromophobe RCC, uterine endometrial carcinoma, and thymoma) for survival time and in five cancers (pancreatic adenocarcinoma, hepatocellular carcinoma, chromophobe RCC, acute myeloid leukemia, and thymoma) for disease-free survival time. It was also found that the most common p53 mutation in hepatocellular carcinomas (R249S) was a much better indicator for poor prognosis than TP53 mutations as a whole. In addition, in cases of ovarian serous cystadenocarcinoma, the co-occurrence of TP53 and BRCA mutations resulted in longer survival and disease-free survival times than the presence of neither TP53 nor BRCA mutations. TP53 mutations are potential prognostic markers that can be used to further improve the accuracy of predicting survival and disease-free survival times of cancer patients.


Introduction
The tumor suppressor gene TP53 encodes for the p53 protein, which serves a role in DNA repair, cell cycle arrest, apoptosis, and other pathways that prevent the development of cancers. The TP53 gene and its corresponding protein are frequently inactivated or partially disabled by mutations that lead to increased risks of developing cancer. Somatic TP53 mutations are very frequent in most human cancers, occurring in 5 to 80% of them, depending on the cancer type and stage [1].

Domains of p53
The TP53 gene contains 19,200 nucleotides in its 11 exons and 10 introns. Of the 11 exons, exon one, the first half of exon two, and the majority of exon 11 are non-coding exons [2].
The two transactivation domains are encoded in exons 2-4, the DNA-binding domain in exons 4-8, the tetramerization domain in exons 9-10, and the basic domain in exons [10][11]. Each domain of this 393 amino acid protein has its own distinct function.
Transactivation domain one (AD1) and transactivation domain two (AD2) are comprised of amino acids 1-92. AD1 is required for cell cycle arrest but is dispensable for apoptosis [3].
AD2 is responsible for apoptosis and is aided by the proline-rich domain which is now part of AD2 [4,5]. The DNA-binding domain (DBD), comprised of amino acids 102-292, is essential for the role of p53 as a sequence-specific transcription factor. The DBD of p53 binds to a specific DNA sequence to activate transcription, mediate apoptosis, and conduct cell cycle arrest to suppress the growth of tumor cells [6]. The DBD is followed by the tetramerization domain (TD), comprised of amino acids 326-356, which aids the DBD in binding to DNA and other proteins by increasing the strength of interactions between p53 and other structures [3] (Fig 1).

Li-Fraumeni and Li-Fraumeni-like syndromes
TP53 is the only gene so far identified in which mutations are definitively associated with Li-Fraumeni (LFS) and Li-Fraumeni-like (LFL) syndromes, which predispose patients to certain types of cancers. Over 50% of families with LFS have an inherited mutation in the TP53 gene [7]. The most common germline mutations in tumors are located at amino acids 248, 337, 273, and 175 [2]. It is known that these mutations can inactivate or disrupt the function of the p53 protein and increase risks of early onset cancer [8].

Objective of study
Past research has determined the most common TP53 mutations, examined the development of p53 alterations, and identified cancers that TP53 mutations were prevalent in.
This study attempted to establish TP53 mutations as potential prognostic markers for specific cancers by investigating if the presence of TP53 mutations in certain cancers was beneficial or detrimental to survival time and disease-free survival time. The distribution of TP53 mutations at specific exons/introns and codons was examined, and the potential of using specific mutations as cancer prognostic markers was evaluated.  Fig 4A).

Germline mutations
The distribution of germline mutations is different from the distribution of somatic 14]. In this study, almost every sample of the two aforementioned cancers had TP53 mutations, but for other types of cancers such as uveal melanomas, which had a sample size of 80, there were no occurrences of TP53 mutations (Table 1).

Overall survival
Of the 33 cancers analyzed, the difference in survival time of 10 cancers was found to be statistically significant when comparing samples with and without TP53 mutations ( Table 1).
The cancers were separated into four groups. The first group consisted of cancers in which TP53 mutations were beneficial. The second group included cancers in which cases without TP53 mutations survived up to two times longer than cases with TP53 mutations. mutations may require more frequent screening and treatment than cases without TP53 mutations due to shortened periods between relapses, but overall survival time would be the same (Fig 6B).
In liver hepatocellular carcinoma, the median disease-free survival time was 11.79 months for cases with TP53 mutations and 25.3 months for cases without TP53 mutations.
Overall, those without TP53 mutations lived disease-free for 2.15 times longer. According to the Kaplan-Meier estimates, 20.5% of cases with mutations remained disease-free, while only 10.9% of cases without mutations remained disease-free at the end of the study (Fig 6C).
In acute myeloid leukemia, the median disease-free survival time was 10.3 months for cases with TP53 mutations, and 17.3 months for cases without TP53 mutations. Those without mutations were disease-free for 1.68 times longer than those with mutations ( Fig 6D). mutations (including R249S) and 60.84 months for cases without TP53 mutations. Since the differences in survival time among the three groups was statistically significant, this specific mutation is an indicator for poor prognosis in liver hepatocellular carcinoma (Fig 7).

Discussion
This study used TP53 mutations instead of immunohistochemistry (IHC) data even though both were available. Since the p53 protein sometimes accumulates when there is no mutation and sometimes does not accumulate when there is a mutation present, the presence of false-positives and false-negatives making IHC less reliable than sequencing for TP53 mutations [15]. Older research uses IHC, but due to the availability of faster and cheaper modern sequencing methods, more genetic data on TP53 has become available for analysis.
Some mutations affected the survival time or time before relapse, while others had no effect. In 26 of the 33 cancers studied, there was data on survival time. With these data, TP53 mutations can be predictive markers as to how long patients will survive, and when the cancer may relapse. For example, with ovarian serous cystadenocarcinomas, one can predict that if a patient does not have a TP53 mutation, then she will have a 45% chance of surviving to month 31, while another patient with the same disease will have a 70% chance of surviving to month 31 if she has a TP53 mutation (Fig 5A). This adds accuracy to preexisting survival estimates that do not consider TP53 mutations as a factor. The same applies to disease-free survival time. In ovarian cystadenocarcinomas, while 55% of patients without TP53 mutations are expected to relapse by month 11, only 30% of patients with TP53 mutations are expected to relapse in the same amount of time (Fig 6A).
Past studies have determined the function of the p53 protein and studied the development TP53 mutations. Since the establishment of the IARC TP53 Database in 1991, many researchers have analyzed and used its data to identify the different TP53 polymorphisms that exist in human populations [16]. Researchers have also identified the causes of TP53 mutations, including geographic differences and known carcinogens [11]. In addition, the effects of TP53 mutations on the function and structure of the p53 protein were studied, which led to the identification of the functional domains of p53 in 2006 [3,17].
Using TP53 mutation data from the newest IARC Database, which is the April 2016, R18 release, and data on cBioPortal, which compiles all of the data from the most recent TCGA cancer studies, this study is the first to represent the newest data available. Applying previous knowledge of p53 functional domains and their locations, we were able to identify the distribution of mutations at exons/introns to determine that the DBD, which is responsible for binding p53 to DNA to regulate cell growth, was affected the most by TP53 mutations. By separating mutations into somatic and germline mutations, we found that germline mutations were common at introns and in the TD, AD1, and AD2, which was rare among somatic mutations.
Using survival and genomic data in multiple TCGA studies on cancers, this study compared Kaplan-Meier estimates of cases with and without TP53 mutations to predict survival time and disease-free survival time. Unlike preexisting studies, we examined the prognostic effect of TP53 mutations in a whole variety of cancers, instead of one specific cancer, and found that survival was significantly affected in 10 different kinds of cancers. This study is also one of the only studies that looked at the significance of TP53 mutations on the time before relapse. We also discovered that TP53 mutations in ovarian serous cystadenocarcinomas, which mostly occurred in the DBD, increased the survival and the disease-free survival time of patients, contrary to a study published in 2007 that said "TP53 mutations within the [DBD] have been repeatedly associated with shorter survival" [18]. The distribution of TP53 mutations was also used to identify the most prevalent mutations in certain cancers. The mutations in liver hepatocellular carcinoma were analyzed and the most common one was R249S. Survival data for cases with this specific mutation showed that it negatively affected the prognosis of patients significantly. Due to a previous lack of data, specific TP53 mutations were not considered to be utilized as prognostic markers until now.
Nevertheless, for some of the cancers, more data is necessary. For seven of the cancers, survival data was not available, and for twelve of the cancers, disease-free survival time data was not available either. Also, for some of the cancers, only small amounts of data were available since only a small percentage of cases had TP53 mutations. Lastly, some Kaplan-Meier estimates could be misleading because of censoring that results in horizontal and vertical lines which indicate loss of data [19]. More data can lead to better accuracy for cancers like clear cell RCC, where there were only nine cases of TP53 mutations and four of them were censored.

Data was downloaded from the International Agency for Research on Cancer (IARC)
TP53 database on all 28,000+ somatic and 800+ germline mutations [2]. Exhibit Both Common and Distinct Properties" [3].
Genomic data from various cancer studies can be found in cBioPortal for Cancer Genomics, available online at cbioportal.org [9,10]. The data sets of the most recent The Cancer  [20]. However, when combined with other factors, the accuracy of survival and relapse estimates can be further improved. The most common p53 mutation (R249S) in liver hepatocellular carcinomas was found to be an indicator for poor prognosis.
In future studies, when more data becomes available, by identifying and analyzing the most common TP53 mutations and the cancers they are prevalent in, it will be possible to predict the invasiveness or aggressiveness of cancers based upon whether they contain TP53 mutations or not, and in cases with TP53 mutations, based upon which specific TP53 mutations are present [21]. This information can be useful in determining the progression of cancers, which is crucial when deciding the type of treatment to be used. Also, the effects of specific mutations in cancers can be investigated to a deeper extent on the molecular level to improve accuracy of TP53 mutations as prognostic markers. This study demonstrated that TP53 mutations have statistically significant effects on survival and disease-free survival time, and thus can be successfully used as prognostic markers.
Specific germline TP53 mutations and the cancers that they are associated with can be identified to predict the cancers certain individuals and their children are predisposed to, which can help implement screening procedures that allow for earlier detection and treatment.
Individuals with a certain germline TP53 mutation can be screened for specific cancers that they are at risk of developing, thus increasing the likelihood of finding and successfully treating premalignant tumors and growths through preventative care. However, more data on germline TP53 mutations is required to establish reliable criteria for screening. As of now, there are still less than one thousand samples in the IARC TP53 Database, the largest database in the field.
Nevertheless, with increasing data availability, future studies may yield further insights.