Association of TP53 Arg72Pro (rs1042522) Polymorphism with Pancreatic Cancer Risk in a Patient Cohort

Simple Summary

Pancreatic cancer is one of the most deadly forms of cancer because it is often diagnosed too late, when treatments are less effective. Currently, there is no practical way to screen individuals who are at risk but have no symptoms. In this study, we describe a specific genetic variation in a gene called TP53, which is known to play a role in many types of cancer. We found that this variation appears to be more common in people with pancreatic cancer than in the general population. This suggests that individuals carrying this variation may have a higher risk of developing the disease. These findings may help open a new line of research aimed at identifying high-risk populations.

Abstract

Pancreatic cancer is expected to become the second leading cause of death by 2030 in Western countries. There is a need to pinpoint high-risk populations since extensive screening would be economically impractical. Methods: This study, conducted on liquid biopsies of patients affected by pancreatic ductal adenocarcinoma (PDAC), sequenced, by NGS, the main genes involved in pancreatic carcinogenesis. Results: The study was discontinued due to a low recruitment rate. NGS analysis, conducted on included patients, revealed the TP53 variant rs1042522 in 30 out of 35 patients, with a cytosine (C) replaced by a guanine (G), hence inserting an Arginine in the final protein instead of a Proline. The presence of the rs1042522 variant confers an odds ratio of 6.11 for PaC and an OR of 20 for homozygosity G/G when comparing our cohort of PaC patients to a healthy population from the 1000GenomeProject. Conclusion: These findings could identify a very-high-risk population deserving of being screened for PDAC, even though a wider validation of rs1042522 as a risk factor is needed. Impact: These preliminary data may open the way for identification of a population more prone to developing pancreatic cancer.

1. Background

Pancreatic cancer is the fourth leading cancer-related cause of death in Western countries and is estimated to become the second by 2030 [1]. It has a high mortality rate since only a small percentage of patients are diagnosed in a resectable, hence curable, stage. Approximately 80% of patients are not amenable to resection at diagnosis due to metastatic disease or local infiltration, and another 41.4% of patients that are candidates for surgery have unresectable tumors found during exploratory laparoscopy or laparotomy [2]. Pancreatic ductal adenocarcinoma (PDAC) typically harbors mutations in a few highly recurrent genes—KRAS, TP53, CDKN2A, and SMAD4—commonly termed “mountain genes”. These mutations impact essential cellular processes such as proliferation, DNA repair, and signaling pathways. Their sequential alteration reflects the genetic progression model of pancreatic carcinogenesis, emphasizing their central role in disease development and highlighting their potential as targets for early detection strategies [3]. Liquid biopsy provides a non-invasive signature of the tumor; it is based on circulating genetic material coming from cellular turnover and thus especially from the tumor [4]. This study, conducted on liquid biopsies of patients affected by pancreatic ductal adenocarcinoma (PDAC), has sequenced the main genes involved in pancreatic carcinogenesis: KRAS, TP53, SMAD4, and CDKN2A.

2. Methods

The Institutional Review Board of the University of Sapienza approved the protocol (Prot. 243 SA_2018, 12 December 2018), where every participating center had to subscribe to the IRB’s guidelines. As per the Helsinki Declaration, the protocol was registered at clinicaltrials.gov with the identification number NCT03524677. All consecutive naive patients with non-metastatic PaC who were eligible for upfront pancreatic resection were enrolled on a multicenter basis from 2017 to 2020. As this was a pilot study, a sample size of 50 patients was deemed small enough (considering, as stated, no additional risk for the patient) but large enough to guarantee that the final contingency table was not sparse. The enrollment took place in 3 institutions: University Hospital Sant’Andrea (Rome), the promoter, University Hospital A. Gemelli (Rome), and Community Hospital G. Salvini (Garbagnate Milanese, MI). After signed informed consent was received, blood samples in EDTA were preoperatively collected and centrifuged twice within 1 h at 2300 rpm for 10 min. Plasma that was obtained was preserved and stored at −80 °C. A protected anonymous multiparametric database was filled with relevant information about the patients and disease. Circulating free DNA (cfDNA) was extracted from 1 to 3 mL of plasma samples using the Plasma/Serum cf-DNA/cf-RNA Advanced Fractionation Kit (Norgen Biotek, Cat. 68300), following the manufacturer’s protocol. Briefly, plasma samples collected in EDTA tubes were centrifuged and supernatants were treated with Binding Buffer A, Proteinase K, and Lysis Buffer A. DNA was isolated using proprietary spin columns, washed, and eluted in 25–50 µL of Elution Buffer F. The concentration and the degree of fragmentation of extracted DNA were assessed by RT-PCR (DNA Myriapod^® NGS Cancer panel DNA quantification strips DNA AMP LAB, Diatech Pharmacogenetics). It was then sequenced with a Myriapod^® NGS 56G Onco panel (Diatech Pharmacogenetics) on the MiniSeqTM Illumina platform (Illumina Inc., San Diego, CA, USA) for the analysis of hot-spot mutations of 56 genes. Gene panels were amplified using Multiplex-PCR using 25 ng of total DNA. After purification and indexing, amplified fragments were selected by enrichment PCR. Then, libraries for the NGS were prepared and quantified by the Qubit 4.0 Fluorometer (Invitrogen by Thermo Fisher Scientific Inc. Wilmington, DE, USA). The panel allowed us to obtain 263 amplicons, with lengths from 92 to 184 bases, covering the hot-spots and surrounding regions. Data analysis, including the alignment to the reference human genome hg19, and the variant calling, was performed using a Myriapod NGS Workstation (Diatech Pharmacogenetics) equipped with Myriapod NGS Data Analysis Software (Diatech Pharmacogenetics, Jesi, Italy). Filtered variants were annotated in the NCBI RefSeq database, and alignments were verified using the Integrative Genomics Viewer. A comparison with a reference population was then carried out with data from the 1000GenomeProject [5] on healthy subjects from central Italy (Tuscany). The statistical analysis was carried out with the online Odds Ratio Calculator [6]. This study has been reported according to STrengthening the REporting of Genetic Association studies (STREGA), which is an extension of the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) Statement [7].

3. Results

3.1. Mountain Gene Results

Patients’ characteristics are listed in

Table 1. PaC cohort: patients’ characteristics.

Clinical Characteristics
Sex (M/F)	20/15
Age (yr)	68 (range 45–81)
M0/1	29/6
TP53 rs1042522	85.7%
TP53 homozygous C/C	14.3%
Variant frequency range	49.93–100%

. Over a 3-year span, 35 patients affected by non-metastatic PDAC were recruited. Eventually, six patients were found to be metastatic at diagnosis. The study was therefore stopped due to under-recruitment. After cfDNA extraction and PCR amplification, QuBit quantification yielded a mean DNA concentration of 3.71 ± 1.75 ng/mL. NGS results on 56 analyzed genes were filtered for PASS quality for a total of 621 mutations; synonymous and intron variants (including 8 intron-variant SMAD4 mutations) were then excluded, leaving 251 mutations. Concerning the mountain genes involved in PDAC, NGS sequencing revealed no significant CDKN2A and SMAD4 mutations in this PaC population, KRAS missense mutation rs121913530 in 1 patient—this variant in exon 2, also known as G12C, has a demonstrated pathogenicity—and 25 TP53 mutations in 23 (65.71%) patients: 23 rs1042522 mutations, discussed in the next paragraph, 1 rs121912651 missense mutation, and 1 rs863224500 stop-gained mutation. The complete data obtained in this study are freely available in the online public repository OSF at https://osf.io/b28ne/ (accessed on 13 May 2022).

3.2. TP53 Polymorphism

As an incidental finding, NGS revealed the rs1042522 variant of gene TP53 in 30 out of 35 patients. In four patients, it yielded a low coverage (<500 reads) with a variant frequency of >99%, therefore making them included in the analysis. In all 30 patients with rs1042522, a cytosine (C) was replaced by a guanine (G), leading to an amino acid change from Proline to Arginine in position 72. The five patients without any call for rs1042522 were classified as homozygous C/C, with a Pro72 phenotype. Nineteen patients with rs1042522 and a VF > 99% were deemed polymorphic G/G homozygotes, with an Arg72 phenotype. Eleven patients in between were presumed to be heterozygous C/G-G/C, with a mixed phenotype.

3.3. TP53 Arg72Pro in Local Population

A comparison with the public repository Ensemble [5] from the 1000GenomeProject was conducted. The healthy population from Tuscany (central Italy) covered in this library accounts for 6 homozygous G/G, 54 homozygous C/C, and 47 heterozygous C/G-G/C [8]. The presence of the rs1042522 variant confers an OR of 6.11 (95%CI 2.20–16.94, p = 0.0005) toward pancreatic cancer when comparing our cohort of PaC patients to a healthy population coming from the same geographical area of Tuscany, central Italy (

Table 2. TP53 polymorphism, PaC cohort vs. healthy subjects (TSI: Tuscany population), in G allele carriers (G/G-G/C-C/G, rs1042522+) vs. C/C homozygous (rs1042522 absent).

ODDS RATIO = 6.11
	PaC patients	TSI healthy individuals
G/G-C/G-G/C	30	53
HOMOZYGOUS C/C	5	54

). Moreover, if we consider only subjects deemed homozygous for G alleles in exon 4, locus 72, they present an OR of 20 (95%CI 6.93–57.61, p < 0.0001) toward pancreatic cancer (

Table 3. TP53 polymorphism, PaC cohort vs. healthy subjects (TSI: Tuscany population), in homozygous G/G vs. C allele carriers (C/C-G/C-C/G).

ODDS RATIO = 20
	PaC patients	TSI healthy individuals
HOMOZYGOUS G/G	19	6
C/C-C/G-G/C	16	101

).

4. Discussion

The original research study has been closed due to under-recruitment. Nevertheless, we incidentally describe an interesting incidence of TP53 polymorphism in this PDAC population.

The economic burden of a hypothetical pancreatic screening prevents second-level imaging from being offered to populations with an increased odds ratio, from 1.7 to 2.2, for pancreatic cancer. This includes those affected by obesity, pancreatitis, and diabetes [9]. Pancreatic cancer screening in asymptomatic patients is impractical. To detect a case, one should screen over 17,000 people, and to find a resectable, hence curable, patient, over 35,000 people should be screened [10]. Finding new genomic susceptibilities to PaC in these high-risk populations might narrow the field for early diagnosis.

First discovered in 1986, p53 has two main species: one has an Arginine in position 72 that has a long positively charged side chain, and the other, which has a Proline in the same position, has a small nonpolar side chain [11]. Rs1042522 is a single-nucleotide variant (SNV) caused by the substitution of cytosine with the wild-type guanine (G) as in our findings, or adenosine (A) or thymine (T) [8]. According to the reference SNP (rs) report from the National Library of Medicine [12], G frequency ranges from 0.11 to 0.495. 1000Genomes Project Phase 3 reports a frequency of the G allele of 46% worldwide, being 0.276 in a Tuscan population. This European subpopulation accounts for 107 patients from central Italy, including the 6 patients who presented homozygous GG alleles in position 72 of the TP53 gene [8].

Assuming that all patients from our cohort with a variant frequency > 99% are likely homozygous G/G and that patients without any call for rs1042522 are C/C homozygous, our population had a C allele frequency of 0.3 and a G allele frequency of 0.7.

Regarding the Hardy–Weinberg Equilibrium effect on case–control studies [13], according to which disequilibrium in the genetic variation in a population can be attributed to different genetic backgrounds, inbreeding, or genotyping problems, we believe that the multicentric enrollment along with the comparison with a public database overcomes its effect on our results.

The Arg72 variant induces apoptosis five times better than the Pro72 variant, while Pro72 represses the G1 phase of the cell cycle [14,15]. This higher apoptotic potential is the mechanism thought to weaken the reserves of pancreatic beta cells, leading to insulin depletion, causing diabetes. As a matter of fact, rs1042522 has been associated with increased odds of diabetes mellitus [16], one of the conditions tightly linked to pancreatic cancer, both as a cause and as an early effect. New-onset diabetic patients, within 3 years from diagnosis, have a prevalence of PaC of 1%, while long-standing type 2 DM patients have an RR of 1.8–2.1 for developing PaC [17]. If confirmed, the association of GG features with an increased risk of PaC could identify a very-high-risk population deserving of being screened.

Interestingly, rs1042522 gives a susceptibility to gastric and esophageal cancer to Northwestern Chinese populations [18], as well to other cancer types such as gliomas, non-Hodgkin lymphoma, and lung, bladder, prostate, colorectal, and breast cancer [19,20,21]. On the other hand, a recent meta-analysis showed that no association has been found between TP53 Arg72Pro and risk of ovarian cancer [22].

To the best of our knowledge, TP53 Pro72Arg polymorphism has been described in three pancreatic cancer populations from Japan [23,24] and the Czech Republic [25]. One study genotyped from FFPE tissues and showed an abnormally high C allele frequency of 0.8 [23]. The second and the latter genotyped from peripheral lymphocytes and had comparable C allele frequency of 0.29–0.39 [24,25]. In both cases, it was hypothesized that there is a higher risk of PaC in C allele carriers.

Despite our results in allele frequency being in line with those of Sonoyama [24] and Naccarati [26], we draw opposite conclusions for our population. In a similar way, p53 exon 4 (Pro) seems to also be protective for head and neck cancer in the Italian population [26], whereas a recent meta-analysis on TP53 rs1042522 polymorphism and cervical cancer showed a significant association in Asian and European ethnicities, not in other populations [27].

Limitations of this study are the lack of a matched control population, as it compares a cohort of patients to a public database; moreover, the small sample size along with the great variability in rs1042522 across the Italian population makes the results not transposable on a larger scale, hence the need for a prospective large-scale validation.

5. Conclusions

PaC is a major cause of death worldwide due to late diagnosis and the lack of screening programs. This results in a need to pinpoint high-risk populations. TP53 is one of the four mountain genes responsible for PaC cancerogenesis, and in our PaC population, its polymorphism rs1042522 seemed to confer a 6- to 20-fold-increased OR for PaC compared to a reference population. Further studies are needed to validate this hypothesis in order to improve patient selection for second-level investigations and early PaC diagnosis.