Introduction

The tumor protein p53 (encoded by TP53), as described by many studies, suppresses the cell cycle and induces apoptosis on activation after DNA damage. These processes are important for the protection against cancer and TP53 mutations and polymorphisms have been widely associated with cancer.1, 2

A common TP53 polymorphism is at codon 72 (rs1042522), which changes an arginine into a proline. Allele distribution depends on ethnic background. Although the Pro-allele is most common in African regions, the Arg-allele is more common in Caucasians. The Pro-allele of rs1042522 has been associated with several forms of cancer. However, next to increased cancer susceptibility, the Pro-allele has also been associated with longevity,3, 4, 5 raising the hypothesis that this allele increases longevity at the cost of increased cancer susceptibility.

The main causes of death in developed countries are cancer and cardiovascular diseases. Important risk factors for cardio vascular diseases are body composition, type 2 diabetes, dyslipidemia and blood pressure. Evidence is emerging that the p53 protein mediates metabolism. It has been shown that p53 expression in adipose tissue is associated with insulin resistance and subsequently with age-related cardiovascular disorders.6, 7 Therefore, we hypothesized that rs1042522 might have an effect on maintenance of metabolism, body composition and blood pressure during ageing.

In this study, we aimed to investigate the relation of rs1042522 with blood pressure and several anthropometric (BMI and waist circumference) and biochemical parameters as measures of successful ageing. Using a longitudinal design, we are able to monitor changes over time, which reflect the effects of ageing. As rs1042522 is the most frequent non-synonymous SNP in TP53 and the only SNP known in this locus to be involved in longevity, we focused only on this variant.

Materials and methods

Study populations

For this study two longitudinal populations were used. The first sample is from the Dutch Doetinchem Cohort Study (DCS), which has been described in detail previously.8 In short, this population-based study aims to identify the impact of lifestyle factors and biological risk factors on aspects of health, such as the incidence of chronic diseases, physical and cognitive functioning and quality of life. Participants were randomly selected from the municipal population register from the city of Doetinchem, The Netherlands. A random subset of participants was re-invited with intervals of approximately 5 years. Inclusion criteria for genetic studies were stable smoking behavior and absence of pregnancy at time of measurement as described previously.9 Data from four visit rounds are now available. However, waist circumference was not assessed in the first measurement round. Furthermore, no compatible data for blood pressure were available from the fourth visit round. In total, data were available from 2679 participants (49% male). Participants were not fasted before investigation.

The second sample was the Botnia Prospective Study (BPS). This is a family-based study aiming to identify genes increasing susceptibility to type 2 diabetes. Details of the study cohort and sampling strategy have been presented earlier.10 In brief, individuals with type 2 diabetes from the area of five health-care centers in western Finland were invited to participate, together with their family members.11 An oral glucose tolerance test (OGTT) was performed for all participants aged 18–70 years who had fasting plasma glucose concentration <11 mmol/l. Glucose tolerance was defined according to the current World Health Organization criteria.12 Data at baseline and after follow-up were available from 2338 participants (46% male). Mean follow-up time was 8.8 years (±4.2 years).

Characteristics are summarized in Table 1.

Table 1 Characteristics of study populations
Full size table

All participants gave written informed consent. Both studies were approved by the local ethics committees and were in accordance with the principles of the Declaration of Helsinki.

Genotyping and quality control

For genotyping of rs1042522 in DCS, KASPar SNP genotyping Chemistry from KBioscience (KBioscience, Hoddesdon, UK) was used. SNPviewer (version 1.93, KBioscience, Hoddesdon, UK) was used for automated genotype calling. Genotype clusters were visually observed for validation of automated genotype calling. The success rate was 98%. Genotype distribution obeyed HWE (P=0.24).

The BPS was genotyped using an allelic discrimination assay-by-design method on ABI 7900 (Applied Biosystems, Foster City, CA, USA). We obtained a genotyping success rate of 99%. Genotypes were in HWE (P=0.73).

Statistics

To test for differences in changes in the outcome variable over time according to rs1042522 genotype, we used linear mixed models (PROC MIXED), with adjustment for age, gender and follow-up time with a random slope and intercept. Differences in changes over time were assessed by including the interaction between SNP and time in the model. Alpha was set at 0.05.

For analysis of blood pressure, participants who used anti-hypertensive medication were excluded (n=274 for DCS and n=335 for BPS). For the analysis of OGTT measures (BPS only), participants who converted to type 2 diabetes were excluded. As the BPS is a family-based study, we adjusted for family relation by clustering for family number in the mixed model. All analysis were performed using SAS 9.2 (SAS Institute Inc., Cary, NC, USA).

Power calculation

Power calculations were performed using G*Power 3.1.13 The required effect size was calculated using MANOVA for between factors (SNP effect), within factors (time effect) and within-between interactions (SNP*time interaction). The correlation between visit rounds was set at 0.5, which was observed in the DCS and the required power was set at 80%. The estimated minimal effect sizes were 0.05% SD for a SNP effect, 0.02% SD for a time effect and 0.05% SD for a SNP*time interaction.

Results

First, the DCS was genotyped and analyzed for associations with changes in blood pressure. Blood pressure was available from the first three visit rounds. Rs1042522 was significantly associated with changes in both systolic blood pressure (SBP) and diastolic blood pressure (DBP) over time. The C/C genotype (Pro/Pro) showed a blunted increase in SBP and DBP over time (P=0.008 and 0.004, respectively, Figure 1). SBP increased 1.4 mm Hg in C/C carriers, 5.1 mm Hg in C/G carriers and 5.0 Hg in G/G carriers over time. DBP increased 0.8 mm Hg in C/C carriers, 3.4 mm Hg in C/G carriers and 3 mm Hg in G/G carriers between the third and first visit round (Supplementary Table 1). SBP was also significantly lower in C/C carriers compared with C/G and G/G carriers in visit round 3 according to a general linear model (P=0.002).

Figure 1
figure 1

Time-dependent changes in blood pressure in the DCS according to rs1042522 genotype. (a) SBP, P=0.008. (b) DBP, P=0.004. Error bars represent 95% confidence intervals. *P=0.002.

Full size image

As changes in blood pressure are age dependent,14 participants were divided in those below and above the median age at baseline (40 years). Although the trend was the same, the association with changes in SBP over time was stronger in the subgroup above the median age (Pbelow=0.18, Pabove=0.04, data not shown). For DBP, this was the other way around, the association was the strongest in the subgroup below the median age (Pbelow=0.009, Pabove=0.29).

Next, we analyzed changes in BMI and waist circumference over time. BMI data were available from all four visit rounds. Waist circumference data were available from visit round 2 to 4. Although no association was observed with changes in BMI over time (P=0.69, Supplementary Table 1), the rs1042522 variant showed a borderline significant association with changes in waist circumference over time (P=0.03, Figure 2). Carriers of the C/C-genotype (Pro/Pro) showed a blunted increase, compared with G/G and C/G carriers. Waist circumference increased 1.1 cm in C/C carriers, 2.2 cm in C/G carriers and 1.7 cm in G/G carriers over time (Supplementary Table 1). No associations were observed with changes in non-fasting HDL and total cholesterol in relation to time (data not shown).

Figure 2
figure 2

Time-dependent changes in waist circumference in the DCS according to rs1042522 genotype, P=0.03. Error bars represent 95% confidence intervals.

Full size image

In order to replicate our findings, we analyzed changes in blood pressure and waist circumference over time in the BPS. Although our findings for SBP were not confirmed, an association with changes in DBP over time was observed (P=0.64 and 0.02, respectively, Figure 3), where the C/C genotype (Pro/Pro) showed a blunted increase in DBP. DBP decreased 0.4 mm Hg in C/C carriers while it increased 1.7 mm Hg in C/G carriers and 1.1 in G/G carriers between follow-up and baseline measurement (Supplementary Table 1).

Figure 3
figure 3

Time-dependent changes in blood pressure in the BPS according to rs1042522 genotype. (a) SBP, P=0.64. (b) DBP, P=0.02. Error bars represent 95% confidence intervals.

Full size image

Stratification according to the median age at baseline (43.7 years) showed also no evidence for an association with SBP changes over time. However, the association with DBP was the strongest in the subgroup below the median age (Pbelow=0.04, Pabove=0.53), which is consistent with data from DCS. Adjustment for family relation in the BPS resulted in identical outcome (data not shown).

Next, we analyzed changes in waist circumference over time for the different genotypes. Corresponding to the DCS data, we observed a borderline significant association with changes in waist circumference (P=0.05, Figure 4), where C/C carriers (Pro/Pro) showed a blunted increase in waist circumference over time. Waist circumference increased 1.6 cm in C/C carriers, 2.6 cm in C/G carriers and 2.8 cm in G/G carriers (Supplementary Table 1). Correction for family relation did not affect the outcome (data not shown). Adjustment of the association between rs1042522 and changes in DBP for waist circumference showed that the association could not be explained by waist circumference (data not shown).

Figure 4
figure 4

Time-dependent changes in waist circumference in the BPS according to rs1042522 genotype, P=0.05. Error bars represent 95% confidence intervals.

Full size image

Finally, we also analyzed the association of rs1042522 with changes in OGTT measures. Although no association with changes in fasting glucose and 2-h glucose levels over time were observed, there were associations with changes in glucose levels at 30 and 60 min after glucose load between follow-up and baseline measurement (P=0.01 and 0.02, respectively, Supplementary Figure 1, Supplementary Table 2). Again, the C/C genotype showed a blunted increase. Furthermore, testing glucose levels independent of time at baseline and follow-up measurement revealed significant differences between genotypes at 30 and 60 min, only at follow-up measurement (P=0.005 and 0.01, respectively). Adjustment for family relation did not affect the results (data not shown).

Discussion

Here, we show evidence in two independent longitudinal studies that the C/C genotype (Pro/Pro) of the rs1042522 variant might be associated with the most modest increases in DBP over time as compared with the other genotypes. The differences in changes in DBP over time are most pronounced in younger subjects (age below the median age). This corresponds with the observation that DBP increases predominantly below 60 years of age.14 The association with changes in SBP observed in the DCS could not be replicated in the BPS. Furthermore, weak evidence was observed that the same genotype is also associated with the most modest increase in waist circumference in both studies. Our findings implicate that sequence variation in the TP53 gene might be associated with changes in traits of metabolism and cardiovascular function during ageing.

In this study, we have only analyzed rs1042522 located in the TP53 locus. In this locus, several other SNPs are present. However, according to the HapMap database, rs1042522 is the only non-synonymous SNP with a minor allele frequency (MAF) above 5%. Other SNPs with a MAF above 5% are all located in non-coding parts of the gene and not known to be associated with longevity. Therefore, we exclusively focused this research on rs1042522.

Although adjustment of the association between rs1042522 and changes in DBP for waist circumference was not possible in the DCS, the BPS showed that the association with changes in DBP was not explained by waist circumference.

As only one candidate SNP was studied and a limited number of selected quantitative traits a stringent correction for multiple testing is not applicable. In order to decrease the risk of chance findings, results were tested in a second cohort (BPS).

However, there are some differences between the two studies used for this research. The follow-up time between visit rounds and the number of visit rounds differs. Furthermore, the DCS is a population-based study whereas the BPS is a family-based study. This might introduce heterogeneity. Selection of more homogeneous cohorts has proven difficult. Most available population studies are of a cross-sectional design. Other longitudinal studies often include largely elder subjects, which might have complicated analyses for DBP, as the effects were mainly observed in younger subjects. Within our criteria the selected studies were the best choice. The differences between the studies might explain the lack of replication for SBP. Therefore, we cannot rule out an association between rs1042522 and changes in SBP. The possibility of chance findings can, however, also not be excluded.

Genome-wide association studies, which investigated the effect of SNPs on blood pressure did not observe p53 SNPs among their top hits.15, 16, 17, 18, 19, 20, 21, 22, 23 As these studies have a cross-sectional design, it is not possible to examine the effect of a SNP on development of a trait. We suggest that rs1042522 has a role in the development of DBP and waist circumference during ageing and therefore, a SNP effect alone will be less informative compared with a SNP*time interaction in this case. This is highlighted by our findings, showing only a SNP*time interaction but no independent SNP effect.

Data from the BPS, suggest that rs1042522 is also associated with changes in 30- and 60-min glucose levels during OGTT. It can be speculated that rs1042522 affects insulin sensitivity and subsequently cardiovascular functioning and metabolism. This corresponds with previous observations that p53 is involved in insulin resistance and subsequently cardiovascular and metabolic diseases, although there is currently no strong evidence that rs1042522 leads to increased susceptibility for these diseases.7, 24 Our results are based on only a single study, because no OGTT was performed in the DCS. Therefore, this observation merits replication.

Previously, it has been shown that the Pro-allele of rs1042522 is associated with longevity.3, 4 Other studies showed that the Pro-variant is less efficient in inducing apoptosis compared with the Arg-variant.25, 26 This difference seems to become more pronounced with advancing age.27 One hypothesis that has been suggested is that individuals homozygous for the Pro-allele preserve their stem cell stock as a result of decreased apoptosis and therefore maintain better cell renewal and homeostasis compared with Arg-carriers.4 If decreased apoptosis does not lead to mortality caused by cancer, preserved regenerative capacity could increase longevity. Another study suggests that the longevity phenotype is caused by increased survival after diagnosis of cancer in C/C carriers.3, 24 Our data support an additional, not mutually exclusive hypothesis. We suggest that the C/C genotype is associated with blunted increases in blood pressure, waist circumference and OGTT measures in our study, which are hallmarks of human ageing. It is therefore tempting to speculate that the association with longevity may be also caused by these observed blunted increases, possibly initiated by enhanced maintenance of insulin sensitivity. However, since we are the first to observe associations with changes over time and to the best of our knowledge the rs1042522 polymorphism has not been associated with blood pressure, waist circumference or glucose metabolism before, future research is necessary to elaborate on our findings and elucidate the mechanism of p53 in metabolic development and ageing.