p53 tumor suppressor gene plays a central role in the induction of cell cycle arrest, senescence and apoptosis.1 Mutations of the p53 gene are associated with poor clinical outcome in many types of cancer, including breast cancer.2–6 A common polymorphism at codon 72 of p53 gene results in an arginine (R) to a proline (P) change. This polymorphism is located in the proline-rich domain, which is important in the apoptosis function of p53. Studies suggest that the two variants have different biochemical and apoptotic capacity, with the 72 R form of wild-type p53 harboring a greater apoptosis-inducing potential than the 72 P variant of wild-type p53.7–9 These functional differences between the 2 variants may alter the tumor response to systemic chemotherapy and affect the patient's clinical outcome. Our previous study suggested that breast cancer patients with the P/P variant are less likely to respond to primary anthracyclin-based chemotherapy.10
The prognostic role of p53 codon 72 genotype in cancer patient's outcome has not been fully established. A recent study showed that the codon 72 P/P variant is associated with poor clinical outcome in breast cancer, and the association is independent of p53 protein expression.11 This study examined the p53 status by using immunohistochemistry rather than DNA sequencing assay11; however, numerous studies indicate that at least 30% of p53 mutations cannot be detected by immunohistochemistry.6, 12, 13 Since in vitro studies stressed that the p53 codon 72 genotype modulates the p53 apoptotic capacity depending on whether the tumor cells retain a wild-type p53 gene,7, 8 it would be of great interest to determine whether the effect of codon 72 genotype on breast cancer survival was dependent on the p53 gene status in a large series of breast cancer patients.
In this study, we genotyped the p53 codon 72 genotype using restriction fragment length polymorphism assay (RFLP) and determined p53 mutations by sequencing p53 gene through exons 4–9 in paraffin-embedded specimens from 414 breast cancer patients with a long-term follow-up (median, 8.2 years). We aimed to investigate the correlation between the codon 72 genotype and survival in this cohort of 414 patients and to further analyze whether such correlation was associated with the p53 gene status.
Materials and methods
A total of 608 consecutive patients with operable primary breast cancer (stage: I–III) were treated at Breast Center, Peking University School of Oncology from December 1994 to June 1998, as previously described.14 Paraffin blocks for both negative axillary lymph-node and tumor tissue are available for 439 patients, in 25 of 439 available specimens, a PCR product could not be obtained because of poor-quality DNA. Thus, 414 patients with operable primary breast cancer were analyzed in this study. Pathological diagnosis was performed for all patients, the patient ages ranged from 25 to 86 years, with a median of 49 years. A total of 214 patients were premenopausal, and 200 patients were postmenopausal. The stage of the tumors was classified according to the tumor-node-metastasis classification of the Union Internationale Contre Le Cancer. Patients received radical or modified radical mastectomy; the axillary lymph nodes were routinely dissected at least level I and II, and whether the lymph node metastasis or not was determined based on the histological examination. The majority of patients received adjuvant treatment, including chemotherapy, endocrine therapy, or combined treatment; the adjuvant treatment is summarized in Table I. The follow-up data were available for all patients, with a median follow-up of 8.2 years (range, 0.3–11.1 years). During the follow-up period, 125 patients had developed distant metastases or local recurrences, and among of them, 90 patients had died of breast cancer. This study was approved by the Research and Ethical Committee of Peking University School of Oncology.
Table I. Association Between the p53 Codon 72 Genotype, Clinicopathologic Characteristics, and Adjuvant Treatment
C, chemotherapy; TAM, tamoxifen; ER, estrogen receptor; PR, progesterone receptor.
Pearson's χ2 test.
Lymph node status
DNA was extracted from paraffin-embedded specimens by using phenol–chloroform assay. For p53 codon 72 genotype analysis, the genomic DNA was extracted from negative axillary lymph nodes (histological examination), whereas for p53 mutations analysis, tumor DNA was extracted from the breast tumor specimens. Briefly, 3 to 4 sections, 10 μm thick, were cut from each sample. After xylene deparaffination and absolute ethanol washing, the sections were digested by proteinase K (0.1 mg/ml) in 200 μl of DNA extraction buffer at 56°C for overnight, following the phenol-chloroform extraction.
p53 codon 72 genotyping
p53 codon 72 genotype was detected using polymerase chain reaction-restriction fragment length polymorphism (PCR-RFLP) assay as described previously,10 with a minor modification. In brief, PCR reaction was composed of 100 ng of genomic DNA, 0.8 mM dNTP, 1.5 mM MgCl2, 1.0× PCR buffer, 0.5 μM each primer (Forward: 5′-TTTCACCCATCTACAGTCCC-3′, Reverse: 5′-CGGTGTAGGAGCTGCTG-3′), 1.0× Q-solution and 0.4 unit HotStar Taq DNA polymerase (Qiagen, Chatsworth, CA) in a total volume of 20 μl. The PCR product was 166 bp; 14 μl of PCR product was digested with BstU1 restriction enzyme (New England Biolabs) for 2 hr at 60°C. The digestion products were then resolved on a 2.5% agarose gel containing ethidium bromide. The P/P variant was identified by a single band (166 bp), the R/R variant produced 2 bands (135 and 31 bp), and heterozygous P/R variant displayed 3 bands (166, 135 and 31 bp) (Fig. 1). Twenty percent of the samples were genotyped in duplicates, and the results were fully concordant. To further verify the RFLP results, we randomly selected 9 PCR products from the 3 genotypes analyzed by PCR-RFLP for DNA sequencing analysis. The genotypes were confirmed by DNA sequencing analysis.
p53 mutation analysis
Polymerase chain reaction and direct DNA sequencing analysis were applied to examine exons 4 through 9 of the p53 gene for somatic mutations. Nested-PCR was performed to amplify exons 4–9 of the p53 gene. A total of 100 ng of tumor DNA from each specimen was used for PCR reaction. The primers used to amplify exons 4–9 of the p53 gene are available when requested. The PCR product was isolated and directly sequenced by using BigDye Termininator Cycle Sequencing Kit and ABI 3730 automated sequencer (Applied Biosystems, Foster City, CA). p53 mutations were confirmed by 2 independent experiments.
The correlation between the genotype variants, clinicopathologic characteristics and adjuvant treatment was determined using Pearson's χ2 test. Disease-free survival or disease-specific survival curves were estimated with the Kaplan–Meier method. The endpoint was the date of relapse or death, or the last point of follow-up, and comparisons between survival curves were performed using the log rank test. A Cox regression model was applied to determine whether a factor was independent predictor of disease-free survival or disease-specific survival in multivariate analysis. All statistical tests were two-sided, and p values less than 0.05 were considered as statistically significant. The statistical analyses were performed using SPSS 13.0 software.
Characteristics of the patients
p53 codon 72 genotype was determined in 414 patients; 22% (91 of 414) of the patients were homozygous for P/P variant, 55% (229 of 414) were heterozygous for P/R and 23% (94 of 414) were homozygous for R/R variant. The frequency of the codon 72 genotypes was similar to previous reports in Chinese population,10, 15 but it was not consistent with Hardy–Weinberg equilibrium (p = 0.03), which means that the cohort may exist a selected bias. However, those 414 patients are derived from the 608 consecutive operable primary breast cancers, and it accounts for 68.1% of the total patients. Furthermore, the clinicopathological characteristics of the 414 patients were similar to that of the 608 patients.
The codon 72 genotype was not statistically significantly associated with the age, tumor size, clinical stage, erbB2 status, lymph-node involvement, estrogen receptor (ER) or progesterone receptor (PR) status and p53 gene status (Table I). Previous studies suggested that p53 mutations were more frequently occurred in R allele16; however, we did not find such association in this cohort. The genotype was not statistically significantly associated with the mean age at diagnosis either in the entire study group or in the subgroup of patients whose tumor had a wild-type p53 (data not shown). Furthermore, the adjuvant treatment, i.e., chemotherapy, endocrine therapy, alone or combined, was evenly distributed among the 3 genotypes, and no statistically significant differences were observed between the genotype groups (p = 0.52; Table I).
p53 mutations and survival
Exons 4–9 were examined specifically for p53 mutations because these exons encompass most p53 mutations detected in breast cancer.6, 17, 18 Of the 414 assessable tumors, 68 (16%) had a mutation in exons 4 through 9 of p53. These mutations resulted in amino acid substitutions or deletions, truncated proteins or coding-reading frame shifts (supplementary Table I). Five-year disease-free survival (DFS) and 5-year disease-specific survival (DSS) were significantly lower in patients with a mutant p53 in their tumors than in those with a wild-type p53 in their tumors (DFS: p = 0.01; DSS: p = 0.04; respectively, Table II).
Table II. Association Between the p53 Codon 72 Genotype, Clinicopathologic Characteristics, and 5-Year Disease-Free Survival and 5-Year Disease-Specific Survival
The 5-year DFS and 5-year DSS in the entire study population (n = 414) were 78 and 82%, respectively. Among the total of 414 patients, p53 codon 72 genotype was statistically significantly associated with DFS (p = 0.02; Table II and Fig. 2a), but not with DSS (p = 0.24; Table II and Fig. 2b); patients with the P/P variant had a worse DFS than did patients with the P/R or R/R variant in the entire study population. Furthermore, the association between the codon 72 genotype and survival was investigated according to p53 gene status, and the association between the codon 72 genotype and survival was more pronounced among 346 patients whose tumor retained a wild-type p53. In patients whose tumors retained a wild-type p53, patients with the P/P variant had a 5-year DFS rate of 65%, as compared with 80 or 83% in patients with the P/R or R/R variant, respectively (p = 0.001; Table II and Fig. 3a); the association of the P/P with worse DSS was also persisted in this subgroup of patients (p = 0.04; Table II and Fig. 3b). We further stratified the p53 codon 72 genotype to clinical stages. The codon 72 genotype was associated with survival in patients with stage I disease whose tumors retain a wild-type p53, but no association was found in patients with stage II or stage III disease whether the tumors retain a wild-type or mutant p53 (data not shown).
In contrast, in patients whose tumor harbored a mutant p53 gene, we found a trend correlation that patients with the P/P variant had better DFS and DSS than did patients with the P/R or R/R variant; however, such association did not reach significant difference (DFS: p = 0.33; DSS: p = 0.22; respectively, Table II, Figs. 3c and 3d).
Genotype and survival in multivariate analysis
In a multivariate analysis, the p53 codon 72 genotype was not an independent marker of DFS in the entire study population (data not shown). In contrast, among 346 patients whose tumors retained a wild-type p53, the codon 72 genotype remained an independent prognostic marker of DFS (P/P vs P/R+R/R, hazard ratio, 2.5; 95% confidence interval, 1.4 to 4.4, p = 0.003) after adjusting for clinical stage, age, tumor size, lymph node status, ER or PR status, erbB2 status and adjuvant treatment (Table III). The risk of patients with the P/P variant relapsed from the disease was almost 2.5 times higher than the risk for patients with the P/R or R/R variant. However, the codon 72 genotype was not an independent marker for DSS in this subgroup of 346 patients (Table III).
Table III. Multivariate Analyses of Disease-Free Survival and Disease-Specific Survival in 346 Breast Cancer Patients with a Wild-Type p53 Tumor
Adjusted for age, tumor size, lymph node status, estrogen receptor or progesterone receptor status; erbB2 expression, and adjuvant treatment.
Cox regression analysis.
To our knowledge, our study is first report to examine both p53 codon 72 genotype and p53 mutations in a relatively large series of breast cancer. We found the codon 72 genotype was associated with DFS but not with DSS in the entire study population. In contrast, the codon 72 genotype was strongly associated with both DFS and DSS only when the 346 patients whose tumors retained a wild-type p53 gene were included in analysis, and more importantly, as compared with the R/R or P/R variant, the P/P variant remained an independent prognostic marker of DFS in this subgroup of patients. Our data suggested that the association between codon 72 genotype and breast cancer survival was more pronounced among patients with a wild-type p53 tumor.
The correlation of the p53 codon 72 genotype with cancer risk has been investigated extensively in a variety of tumors,11, 19–23 but the results are controversial. In our study, the codon 72 genotype was not associated with breast cancer risk or disease onset either in the entire study population, or in the group of patients whose tumors retained a wild-type p53 (data not shown). On the other hand, the association of p53 codon 72 genotype with survival has also been investigated in several types of cancer,24–27 but few studies have determined the codon 72 genotype and p53 mutations in clinical samples.8, 28, 29 Sullivan et al. found that patients with a wild-type 72 R tumor are more likely to respond to chemotherapy and display a longer survival than those with a wild-type 72P tumor in 70 advanced head and neck tumors,8 implying that the association between the codon 72 genotype and survival is modified by p53 gene status.
We detected the p53 mutations through exons 4–9 in 414 breast tumors. The frequency of p53 mutations was about 16% in this study, which was consistent with a meta-analysis that the p53 mutations occur in approximately 18% of breast tumors.4 In agreement with previous studies, we found patients with p53 mutations had worse DFS and DSS than did those with a wild-type p53 in our study. The p53 mutations were screened in the total of 414 tumors, and we were able to analyze the association between the codon 72 genotype and survival according to p53 gene status. Among patients with a wild-type p53 tumor, the P/P variant was significantly associated with poor DFS and DSS and remained an independent marker of DFS after adjustment of adjuvant treatment and other clinicopathologic variables. Interestingly, our finding that the P/P variant was an independent unfavorable marker of DFS in wild-type p53 tumors supported the in vitro notion that 72 P variant of wild-type p53 is less effective at inducing apoptosis than the 72 R variant.7, 8 In contrast, among patients with a mutant p53 tumor, patients with the P/P variant tended to have a better survival than those with the R/R or P/R variant, but this association did not reach statistical significance, possibly because of a small number of patients in this subgroup. The DNA was isolated from the paraffin-embedded block, we failed to clone DNA (1–2 kb) from the P/R heterozygous tumors to determine the mutant p53 allele. Previous studies have demonstrated that certain p53 mutations selectively gain the ability to bind and inactivate p73-mediated apoptosis30–32 and that this effect is influenced by the p53 codon 72 genotype, with the mutant 72R allele disrupting p73 function more efficiently than the mutant 72P allele.33 Consistent with the in vitro studies, we found that prognosis was less favorable when the p53 mutations occurring in the 72R rather than 72P allele. The number of patients who carried the p53 mutations in their tumors is relatively small, and it should carefully interpret the data, and further studies with large sample size are warranted to confirm this finding. Our data suggest that the p53 codon 72 genotype might possess opposing effects on wild-type or mutant p53 with respect to breast cancer survival.
In this study, we have demonstrated that the effect of the p53 codon 72 genotype on the breast cancer survival is dependent on p53 gene status, and the genotype remains an independent prognostic factor of DFS in patients with a wild-type p53 tumor. We previously showed that patients with p53 codon 72 P/P variant are less likely than those with P/R or R/R variant to respond to primary anthracyclin-based chemotherapy.10 Toyama et al. also found that the codon 72 P/P variant is associated with poorer DFS in breast cancer patients who received adjuvant chemotherapy, in contrast, there was no association between the P/P variant and survival in patients who received tamoxifen treatment or who did not receive adjuvant therapy, suggesting that patients with P/P variant having worse survival are largely due to resistance to adjuvant chemotherapy.34 Taken together, our findings may have clinical implications, among breast cancer patients who retain a wild-type p53 in their tumors, the P/P carriers may not respond to anthracyclin-based adjuvant chemotherapy and have a worse DFS and, therefore, an alternative chemotherapy regimen (like paclitaxel-based regimen) may be applied for this subgroup of patients. Nevertheless, further studies are needed to confirm our present findings.