Association between the TP53 Arg72Pro polymorphism and clinical outcomes in acute myeloid leukemia

In acute myeloid leukemia (AML), alterations involving the tumor suppressor gene TP53, including mutations, deletions, or both, are frequently associated with older age and very poor prognosis.1 In addition, they are closely associated with complex aberrant karyotype; of note, TP53 alterations are rarely observed outside this group.2 On the other hand, several polymorphisms in the TP53 gene have been described3 of which the non-synonymous TP53 Arg72Pro (G215C) polymorphism appears to be a promising genetic modifier in human tumors, particularly because of its role as a modulator of the apoptotic activities of the encoded p53 protein.4 In terms of the biological significance of the TP53 Arg72Pro polymorphism, there are discernible functional differences between variants at this site. While the TP53 variant that encodes proline (Pro72) results in 3- to 5-fold decreased apoptotic activity5 and an increased risk of cancer,6 the TP53 arginine (Arg72) variant has an increased ability to induce apoptosis and to repress cellular transformation.5

To date, little is known about the frequency and prognostic impact of the TP53 Arg72Pro polymorphism in hematologic malignancies, particularly in AML. The few available data suggest that the interaction between the TP53 Arg72Pro polymorphism and the MDM2 SNP309 variant could modulate responses to genotoxic therapy and increase the risk of therapy-related AML.7 Here, we assessed the frequency of the TP53 Arg72Pro polymorphism in healthy volunteers and patients with AML, and evaluated its clinical impact on outcomes of non-selected patients with AML (excluding acute promyelocytic leukemia) at two university hospitals who were followed from June 2003 to January 2016.

Overall, 429 subjects were included. Two-hundred and five adult patients with de novo AML were retrospectively analyzed. One hundred and fifty-one (74%) patients were treated in Recife (northeast Brazil), while 54 (26%) patients were treated in Ribeirao Preto (southeast Brazil). Baseline characteristics were similar between centers. The treatment protocol has been described previously.8 Briefly, in patients up to 60 years of age, the treatment protocol was adapted according to performance status and the presence of comorbidities (in particular, cardiac disorders). Conventional chemotherapy consisted of daunorubicin (60–90 mg/m/d for 3 days) and cytarabine (100–200 mg/m/d for 7 days) or TAD-9 as induction, followed by three or four cycles of consolidation therapy with high doses of cytarabine (>1 g/m/d). For patients who did not achieve complete remission (CR) after one course of chemotherapy, a second course was administered between days 28 and 35 after the end of the first course. CR was assessed by bone marrow examination on day 28 after each course of chemotherapy. For those who needed it, a post-remission therapy based on autologous or allogeneic transplantation was performed. Patients older than 60 years were treated with low-dose ARA-C, a combination of etoposide, thioguanine, and idarubicin, or best supportive care.

For the healthy control group, peripheral blood samples from 224 age- and sex-matched healthy volunteers (median age 51 years, range 21–83 years; 55% female) with no history of hematologic disease were obtained from the University Hospital, Federal University of Pernambuco, Recife, Brazil. Patients with therapy-related AML or with a previous history of myelodysplastic syndrome were not included. Informed consent was obtained from all patients and healthy volunteers in accordance with the Declaration of Helsinki and approval was obtained from the local research ethics board. The TP53 Arg72Pro polymorphism was evaluated by polymerase chain reaction-restriction fragment length polymorphism.9 Details of the statistical analysis and clinical end points have been described previously.10

The TP53 Arg72Pro polymorphism was successfully genotyped in 413 (96%) subjects, including 189 of 205 patients with AML (92%) and 224 of 224 healthy volunteers (100%). No deviation from Hardy-Weinberg equilibrium was detected in the patient (P>0.05) or control (P>0.05) groups. In addition, there were no differences in the baseline characteristics or outcomes of the patients included in this study versus patients who were not included because of poor quality genomic DNA or unavailable DNA samples (data not shown). To determine whether the TP53 Arg72Pro polymorphism is associated with risk of AML, we compared the frequency of this polymorphism in patients with AML and healthy subjects. The allelic (P=0.004) and genotypic (P=0.008) frequencies of the TP53 Pro72 variant were higher in patients with AML (Table 1). Next, we restricted our analysis to patients with AML and analyzed the association of the frequency of the TP53 Arg72Pro polymorphism with clinical and laboratory features. We also evaluated its clinical impact on induction and post-induction outcomes. The recessive model (i.e. Pro/Pro vs. Arg/Arg+Arg/Pro, hereafter called non-Pro/Pro) was used because it had the best fit to our data (Table 1). There were no significant differences between patients with the Pro/Pro genotype (43 patients, 23%) versus the non-Pro/Pro genotype (146 patients, 77%) with respect to clinical and laboratory features (Table 2).

Table 1.Distribution of allele and genotype frequencies of the TP53 Arg72Pro polymorphism.

Table 2.Baseline characteristics.

Of the 189 enrolled patients, 2 (1%) patients who started the induction treatment were lost to follow up without being assessed for CR. In addition, 68 (36%) patients did not receive conventional chemotherapy (main reasons for treatment failure have been previously described8) and were considered ineligible for the induction and post-induction therapy analyses. CR was achieved in 68 of 119 (57%) patients, of whom 18 of 27 (67%) and 50 of 92 (54%) patients were assigned to the Pro/Pro and non-Pro/Pro groups, respectively. The TP53 Arg72Pro polymorphism had no impact on CR (P=0.278). Of the 51 patients who failed to reach CR, 16 (33%) experienced early mortality, mainly due to bacterial and fungal infections. Although the early mortality rate was proportionally higher in patients with the non-Pro/Pro genotype (44% vs. 30%) there was no difference between groups (P=0.449).

The median follow up for the entire cohort was 135 days [95% confidence interval (CI): 69–200 days] with an estimated 5-year overall survival (OS) rate of 22% (95%CI: 17%–29%). Patients with the Pro/Pro genotype had significantly longer survival (median 264 days, 95%CI: 201–657 days) than patients with the non-Pro/Pro genotype (median 114 days, 95%CI: 70–158 days). Univariate analysis showed that patients with the Pro/Pro genotype had a higher 5-year OS rate (42%) than patients with non-Pro/Pro genotype (12%) (P=0.031) (Figure 1A), although this difference was no longer significant after adjustment for age, cytogenetic risk stratification, and leukocyte counts at diagnosis [hazard ratio (HR): 0.66, 95%CI: 0.37–1.17; P=0.163]. We also analyzed the clinical impact of the TP53 Arg72Pro polymorphism in each cytogenetic risk group. The TP53 Arg72Pro polymorphism had no impact on the clinical outcome of patients assigned to the favorable (24 patients; P=0.6) and adverse (16 patients; P=0.561) groups. In contrast, the TP53 Pro/Pro genotype was significantly associated with longer survival (P=0.035) for patients assigned to the intermediate cytogenetic risk group (57 patients) (Figure 1B), even though these results were not consistent with the multivariate analysis (HR: 0.44, 95%CI: 0.14–1.37; P=0.164). The TP53 Arg72Pro polymorphism had no impact on disease-free survival (P=0.77).

Figure 1.Patients’ survival. The probability of overall survival (OS) (A) in patients with acute myeloid leukemia (AML) according to the presence of the TP53 Arg72Pro polymorphism (entire cohort). OS (B) in patients with intermediate-risk karyotype according to the presence of the TP53 Arg72Pro polymorphism. Survival curves were estimated using the Kaplan-Meier method and the log-rank test was used for comparison.

Regarding the prognostic relevance of age in AML, we decided to evaluate the prognostic impact of the TP53 Arg72Pro polymorphism separately in younger (<60 years of age; n=125) and older (>60 years; n=64) patients. For patients up to 60 years of age, the TP53 Arg72Pro polymorphism had no impact on CR (P=0.648), and DFS (P=0.856), but there was a trend towards a higher OS rate for patients with Pro/Pro genotype (17% vs. 46%; P=0.059). TP53 Arg72Pro polymorphism was not associated with treatment outcomes in patients over 60 years of age (CR: P=0.154, DFS: P=0.201, OS: P=0.643).

Although most studies have described the TP53 Pro72 allele as a poor prognostic factor in several types of cancer,6 evidence suggests that the Pro72 variant is more efficient in both activating DNA-repair target genes11 and inducing cell-cycle arrest.12 Particularly in myeloid neoplasms, the TP53 Pro72 variant exerts a protective effect against therapy-related AML in individuals with lower levels of the MDM2 protein.7 Finally, Pro72 allele carriers show significantly lower frequency of TP53 mutations in specific types of human non-hematologic tumors.13 Here, we demonstrated that the TP53 Pro/Pro genotype was associated with higher risk of leukemia and favorable outcome in AML patients treated with conventional therapy, particularly those assigned to the intermediate cytogenetic risk group. Therefore, it seems that the TP53 Arg72Pro polymorphism may have different tissue- and context-specific functions, and the prognostic importance of each allele may depend on the type of cancer and on the particular treatment. Importantly, screening for the TP53 mutations was not performed in the present study. Nevertheless, we should point out that, in our cohort, only 6% of patients had complex karyotype, and abnormalities involving chromosome 17 [17p-, monosomy 17, or i(17q)] were not observed. One may argue that TP53 alterations could be responsible for inferior outcomes in our cohort; however, considering the rarity of these mutations in non-complex karyotype AML, this seems unlikely.

Several published data have investigated the relationship between the TP53 Arg72Pro polymorphism and risk of leukemia, but the impact of this polymorphism on predisposition to leukemia remains controversial. Most recently, in a meta-analysis involving seven AML cohorts with 1054 patients and 4337 healthy subjects, Tian et al. reported an absence of association between the TP53 Arg72Pro polymorphism and increased risk of AML.14 Nevertheless, it is important to note that the Authors conducted their analyses based on studies from Asian (eight studies) and Caucasian (five studies) populations. These may not reflect the Brazilian genetic background, which is well-known for its mixed genetic population.

Although we have demonstrated that the TP53 Arg72Pro polymorphism could be prognostically relevant in AML, these results must be treated with caution. First, a sizable number of patients were excluded from induction and post-induction analyses because either they did not receive conventional chemotherapy (36%) or because they experienced early mortality (33%), which could bias our analyses. In addition, because of the relatively small sample size and the current lack of validation in independent cohorts, it is probably premature to use TP53 Arg72Pro genotype information in treatment decisions. It would be reasonable to suppose that our findings will be confirmed by other groups with larger sample sizes and well-designed studies. Despite its limitations, the germline genetic profile should not be overlooked, taking into account the genomic landscape of AML and its role in the clonal evolution of the disease.15