Introduction

Prostate cancer (PrCa) is the most frequent tumor and the third leading cause of cancer related death in men in the western industrial world [1]. In the etiology of PrCa a considerable degree of heritability is assumed to be involved [2]. To date, more than 30 moderately risk-associated single nucleotide polymorphisms have been identified in genome wide association studies, which explain a minor fraction of heritability [3]. In contrast, high risk genes which could be responsible for the majority of the observed familial clustering are still unknown.

There is growing evidence that genes involved in DNA damage response play a predisposing role for PrCa [4]. This includes DNA repair processes as well related mechanisms like apoptotic pathways. TP53AIP1 (tumor protein p53 regulated apoptosis inducing protein 1) plays a key role in the tumor suppressor p53 dependent apoptotic signaling. TP53AIP1 is localized in the mitochondrial membrane and mediates apoptosis through cytochrome c release [5]. The expression of TP53AIP1 is triggered upon severe damage by phosphorylation of p53 at Ser46 [6]. TP53AIP1 was found to be mutated in PrCa tissue in a recent study [7]. The two identified truncating mutations p.Q22fs and p.S32X turned out to be recurrent germ-line variants and were strongly associated with PrCa risk (OR of 5.1), thus suggesting TP53AIP1 as a promising susceptibility gene. In the present study we have evaluated the risk effect of these two TP53AIP1 mutations in two independent German PrCa cohorts.

Results

Both truncation variants were observed in our cohorts. While the frameshift allele p.Q22fs was present in 1.8% cases and 1.6% controls, the nonsense mutation p.S32X appeared to be comparably rare (0.1% carriers in both cases and controls). Overall, p.Q22fs and p.S32X were not associated with prostate cancer risk, neither in the combined series of Ulm and Hannover, nor in each individual sample (Table 1). No evidence against homogeneity was observed among the Ulm and Hannover series (p = 0.39). Accumulation in familial prostate cancer cases could indicate a high risk effect mediated by the variants. This hypothesis was tested in the Ulm series, which contained a collection of PrCa cases with positive family history. However no enrichment of the mutations was observed in familial cases (6 out of 377 (1.6%)) when compared to sporadic cases (7 out of 325 (2.1%)). Finally, clinical subgroups of cases were defined in order to elucidate trends towards aggressive tumor forms. The mutation carriers showed no significant difference from non-carriers in severity of prostate cancer, as determined by age of onset, Gleason score, or PSA level prior treatment (Table 2).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 1. Genotype specific odds ratio (OR), exact 95% confidence interval (CI 95%) and p-values of truncating TP53AIP1 variants in case control comparisons.

https://doi.org/10.1371/journal.pone.0034128.t001

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 2. Comparison of age at diagnosis, Gleason Score and PSA at diagnosis between carriers of the truncating TP53AIP1 variants and non-carriers.

https://doi.org/10.1371/journal.pone.0034128.t002

Discussion

TP53AIP1 has been suggested as a PrCa susceptibility gene, since two recurrent germline mutations (p.Q22fs and p.S32X) were found considerably overrepresented among PrCa patients within a previous case-control study [7]. For the purpose of confirmation, we have genotyped p.Q22fs and p.S32X in approximately 1,200 prostate cancer cases and 1,500 controls, but we failed to replicate an association of these variants with PrCa risk.

The discrepancies between the present and the initial study could be explained by several factors, including variable ethnic backgrounds, different sample sizes and diverging inclusion criteria. Population-specific differences have become evident in the case of p.S32X, as this variant has been found rarely in German probands, and thus could not be evaluated for disease association. The most obvious difference between the study samples is the frequency of variant p.Q22fs, which appeared much lower in the control group of Wang and colleagues as compared to our controls (0.3% versus 1.6%, respectively). Noteworthy, Wang et al. have recruited healthy, age matched controls, which could be more powerful to detect associations with prostate cancer risk, despite of their 4.5-fold smaller sample size. In contrast, the unselected controls for our comparisons may contain misclassified prostate cancer cases at population prevalence levels, so that true disease effects would appear slightly understated. However, power estimations suggest that the loss of power by choosing population controls instead of “super controls” would be small for a disease with an about 10% incidence rate, and that this loss can be efficiently compensated for by modest increases in sample size [8]. The fact that we observed virtually equal carrier frequencies in both cases and controls strongly argues against an association of p.Q22fs with PrCa. Based on the sample sizes enrolled here, a resulting interval of confidence rules out any risk effect greater than 2.2 for p.Q22fs and p.S32X.

In subgroup analyses we have considered predisposing roles of TP53AIP1 mutations particularly for familial and for aggressive PrCa. No accumulation of p.Q22fs and p.S32X was observed in familial cases, rejecting the hypothesis of a high penetrance of these mutations. Finally, the lack of genotype/phenotype correlation with respect to clinical parameters argues against susceptibility to more severe forms of PrCa for the mutation carriers.

Aside from truncating mutations examined in the present study, a missense substitution (p.A7V) in TP53AIP1 has previously been studied in the Hannover series. Similarly, the p.A7V variant also had not shown an increased frequency in prostate cancer cases [9]. We conclude that there is little support at present to regard TP53AIP1 as a prostate cancer susceptibility gene.

Materials and Methods

Acknowledgments

Manuel Luedeke is participant of the International Graduate School in Molecular Medicine Ulm.