Association Between BRCA Status and P53 Status in Breast Cancer: A Meta-Analysis.

BACKGROUND Research on BRCA mutation has meaningful clinical implications, such as identifying risk of second primary cancers and risk of hereditary cancers. This study seeks to summarize available data to investigate the association between BRCA status and P53 status by meta-analysis. MATERIAL AND METHODS We searched PubMed, Embase, and Cochrane library databases for relevant studies. Meta-analysis was conducted using STATA software. We summarized odds ratios by fixed-effects or random-effects models. RESULTS This study included a total of 4288 cases from 16 articles, which including 681 BRCA1 mutation carriers (BRCA1Mut), 366 carriers of BRCA2 mutation (BRCA2Mut), and 3241 carriers of normal versions of these genes. BRCA1Mut was significantly associated with P53 over-expression compared with BRCA2Mut (OR 1.851, 95% CI=1.393-2.458) or non-carriers (OR=2.503, 95% CI=1.493-4.198). No difference was found between p53 protein expression in BRCA2 Mut carriers and non-carriers (OR=0.881, 95% CI=0.670-1.158). CONCLUSIONS Our meta-analysis suggests that BRCA1Mut breast cancer patients are more likely to have P53 overexpression compared with BRCA2Mut and non-carriers. This information provides valuable information for clinicians who perform related studies in the future.

Background

Approximately 5–10% of all breast cancers are hereditary, and germline mutations in at least 10 genes linked to DNA repair have been found to be associated with an increased risk for breast cancer [1,2]. Six genes confer a high risk for hereditary breast cancer: BRCA1, BRCA2, TP53, PTEN, CDH1, and STK11. Germline mutations in BRCA1, BRCA2, and P53 are the most common known causes of hereditary breast cancer. BRCA1 and BRCA2 are involved in DNA damage repair and recombination, cell cycle checkpoint control, apoptosis, and transcriptional regulation [3]. P53 is a tumor suppressor that plays a pivotal role in the cellular response to DNA damage by inducing pathways involved in apoptosis, cell cycle arrest, and DNA repair [4,5]. The occurrence of TP53 somatic mutation in breast cancer is well known. Mutations and genetic polymorphisms can alter P53 function and activity, thereby impairing the cellular response to DNA damage [5].

Previous studies of the relationship between BRCA status and p53 status have reported inconsistent results. Several investigations have demonstrated that p53 over-expression is more common in breast tumors associated with BRCA1 mutation than with BRCA2 mutation or non-carriers [6-9]. Others have identified p53 over-expression in tumors with BRCA2 mutation in comparison with non-carriers [10,11]. Conversely, several reports have found no association between p53 over-expression and BRCA status [12,13]. Greater insights into the specific DNA repair mechanism in BRCA mutations and P53 status in breast cancer create more opportunities for effective treatment of patients.

We sought to clarify the relationship between BRCA mutation status and p53 expression status by performing a meta-analysis.

Material and Methods

Publication search

The PubMed, Embase, and Cochrane Library databases were searched for suitable studies published before June 2015. Publications with the following search terms in the title, abstract, or key words were included: TP53, p53, p53 expression, BRCA, BRCA1, BRCA1 mutation, BRCA2, BRCA2 mutation, and breast cancer. We also used these search terms for manual examination of referenced studies. When multiple studies were identified, we evaluated their potential eligibility for inclusion by scanning abstracts and full texts.

Selection and exclusion criteria

The inclusion criteria were: 1) original investigation with evaluation of p53 status in 3 groups (BRCA1Mut, BRCA2Mut, and non-carriers); 2) included sufficient data to determine odds ratio (OR) and 95% confidence interval (CI); 3) published as an original article in English; 4) breast cancer patients including both hereditary and sporadic; and 5) being a high-quality case-control study (Newcastle-Ottawa Scale (NOS) ≥7 points. The exclusion criteria were: 1) data obtained from animal models; 2) BRCA1Mut and BRCA2Mut not examined individually; 3) review article or letter; 4) lacking essential data (e.g., number of P53-positive and -negative individuals in each group not separately indicated); and 5) use of duplicate data.

Data extraction

For each of the eligible manuscripts, 2 investigators (PL and XT) independently extracted the following data: first author name; year of publication; country; numbers of BRCA1Mut patients with p53(+), BRCA1Mut patients with p53(−), BRCA2Mut patients with p53(+), BRCA2Mut patients with p53(−), non-carrier patients with p53(+), and non-carrier patients with p53(−); method of BRCA testing; and method of p53 testing. Any disagreements were resolved by a third author.

Statistical analysis

Statistical analyses were conducted using STATA software (version 12.0, Stata Corporation). We evaluated the association between P53 status and BRCA status using OR and 95% CI. The χ2 and I2 test methods were used to evaluate data heterogeneity in the studies. The fixed-effects model was used when I2 <50% and p>0.05 for χ2; otherwise, the random-effects model was used [14].

Results

Study characteristics

A total of 718 articles were identified by initial database searches. After excluding duplicates, laboratory studies, reviews, and other irrelevant studies, 58 reports remained. We then excluded 11 studies that provided only BRCA1Mut and P53 data, 6 studies that provided only BRCA2Mut and P53 data, 11 studies that contained insufficient data, 9 studies that evaluated BRCA1 and 2 together, and 5 studies that used animal models. The 16 remaining eligible articles [12,15-29] were subsequently included in the meta-analysis (Figure 1).

Figure 1

Flow chart used to identify relevant studies. BRCA1Mut – BRCA1 mutation; BRCA2Mut – BRCA2 mutation.

Table 1 provides the main parameters of the meta-analysis. The 16 eligible studies were published between 1999 and 2015 and all were case-control studies. A total of 4288 patients were included in this meta-analysis, with 681 BRCA1Mut carriers, 366 BRCA2Mut carriers, and 3241 non-carriers. All patients came from 1 of 10 countries (Japan, China, Finland, Spain, The Netherlands, Italy, France, Australia, USA, and the UK). Methods used to assess BRCA mutation were fluorescence in situ hybridization (FISH), denaturing high-performance liquid chromatography (DHLPC), single-strand conformation polymorphism (SSCP), protein truncation test (PTT), horseradish peroxidase (HRP), and high-resolution melting (HRM). Immunohistochemistry (IHC), SSCP, and FISH were the methods used for p53 detection.

Table 1

Main Characteristics of eligible studies.

Author	Country	Year	BRCA-1		BRCA-2		Noncarries		NOS	Method of BRCA test	Method of P53 test
			P53(+)	P53(−)	P53(+)	P53(−)	P53(+)	P53(−)
Gretarsdottir S	USA	1998	NA	NA	10	24	62	306	7	FISH	Not mentioned
Lynch BJ	USA	1998	8	14	5	8	2	18	8	Not mentioned	Immunohistochemistry
Noguchi S	Japan	1999	14	5	6	8	29	59	7	SSCP	Immunohistochemistry
Armes JE	Australia	1999	7	3	1	8	9	11	8	PTT	Cycle sequencing, SSCP, sub-cloning
Freneaux P	France	2000	9	7	0	4	13	21	8	Not mentioned	Immunohistochemistry
Lakhani SR	UK	2002	188	137	59	48	126	80	8	Not mentioned	Immunohistochemistry
Kim S	USA	2003	5	1	1	4	19	45	9	Not mentioned	Immunohistochemistry
Palacios	Spain	2003	10	9	2	11	1	26	8	PTT and SSCP	Immunohistochemistry
Berns EM	Netherlands	2003	27	14	7	7	0	0	7	Enzyme-immunoassay	Immunohistochemistry
Sensi E	Italy	2003	6	4	0	9	25	47	7	PTT and SSCP	Immunohistochemistry
Eerola H	Finland	2005	16	27	9	42	32	118	8	Not mentioned	Not mentioned
Musolino A	Italy	2006	7	1	1	4	13	26	7	FISH	FISH
Colombo M	Italy	2008	10	12	1	15	2	31	8	DHLPC	Not mentioned
Lagos-Jaramillo VI	USA	2011	24	19	11	10	33	37	8	Not mentioned	Immunohistochemistry
Xu J	China	2011	18	34	6	22	98	182	7	HRM	Immunohistochemistry
Aleskandarany M	UK	2015	12	33	4	19	492	1278	8	Not mentioned	Immunohistochemistry

SSCP – single strand conformation polymorphism; DHLPC – denaturing high-performance liquid chromatography; PTT – protein truncation test; HRM – high resolution melting; FISH – fluorescence in situ hybridization.

Study results and meta-analysis

We found that BRCA1Mut was significantly associated with p53 overexpression compared with BRCA2Mut (OR=1.929; 95% CI=1.457–2.554; p<0.001 using the random-effects model; Figure 2). There was no obvious evidence of heterogeneity (I2=46.8%, p=0.024). Compared with the non-carrier group, P53 overexpression was more common in BRCA1Mut carriers (OR=1.509; 95% CI=1.221–1.864; p<0.001 using the random-effects model; Figure 3). Moderate heterogeneity was present for risk difference studies (I2=70.2%, p<0.001). We found no difference in P53 protein expression between BRCA2Mut carriers and non-carriers (OR=0.842; 95% CI=0.642–1.13; p=0.211 using the fixed-effects model; Figure 4), with no obvious evidence of heterogeneity.

Figure 2

Forest plot of odds ratio for P53 mutations: BRCA1 mutations versus BRCA2 mutations.

Figure 3

Forest plot of odds ratio for P53 mutations: BRCA1 mutations versus non-carriers.

Figure 4

Forest plot of odds ratio for P53 mutations: BRCA2 mutations versus non-carriers.

Sensitivity analysis and publication bias

Sensitivity analyses was used to verify the influence of each single study on the overall results in each comparison. The results indicated that 1 study, Lakhani et al. [20], had the most significant impact on the total effect (Figure 5) and it had the most cases of all included studies. However, when the study was removed, the ORs were not substantially changed. The OR was 1.882 (95% CI=1.486–2.427) for excluding the study, and the previous OR was 1.929 (95% CI=1.457–2.554). Moreover, the heterogeneity was not significantly changed. I2 was 46.8%, and dropped to 45.6% when the study was removed. The impact of the total combined effect was the largest.

Figure 5

Sensitivity analysis for each individual study.

Begg’s funnel plot was used to test for potential publication bias. Potential publication biases were found in funnel plot analysis (Figure 6).

Figure 6

Begg’s funnel plot on effectinevess to evaluated the publication bias of the studies.

Discussion

Our meta-analysis suggests an intimate link between BRCA1Mut and P53 overexpression. BRCA1Mut was significantly associated with P53 overexpression compared with BRCA2Mut and non-carrier genotypes.

P53 mutation is frequently identified in high-grade estrogen receptor-negative, progesterone receptor-negative, basal-like breast cancer with increased genomic instability and poor prognosis [30,31]. Additionally, BRCA1Mut breast cancer typically behaves in a manner similar to that described above [12,28]. Several explanations for these similarities are possible. First, both P53 and BRCA1 are tumor suppressors involved in many cellular processes ranging from DNA double-strand repair to cell-cycle arrest, cells apoptosis, and transcriptional control [32,33]. Second, both P53 and BRCA1 are located on chromosome 17, and simultaneous loss of heterozygosity (LOH) at P53 and BRCA1 may occur via misregulation of chromosome 17. This could result in protein-truncating P53 mutations occurring in cis with the BRCA1 germline mutation [34]. Finally, LOH at the P53 locus might occur more efficiently in BRCA1-deficient cells because BRCA1 is also involved in the G2-M and spindle assembly checkpoints [35].

Cells possess numerous DNA repair pathways in which several proteins interact with each other in response to damage detection. The spectrum of P53 mutations identified in BRCA1Mut-associated tumors is highly heterogeneous. This may be because of reduced efficiency of DNA repair activity in BRCA1Mut cells. Dong et al. [36] revealed that P53 mediates homologous recombination during DNA repair by inhibiting excessive BRCA1 function via a mechanism of transcriptional regulation. Consistent with this, we found that BRCA1Mut was significantly associated with P53 overexpression compared with BRCA2Mut or non-carriers.

BRCA2 and P53 have been extensively studied through genetic tests by some medical centers. However, the relationship between P53 expression and BRCA2Mut is less clear than that for BRCA1Mut. Biesma et al. [37] identified P53 overexpression in approximately 50% of BRCA2Mut, whereas others reported that less than 20% of such tumors had elevated P53 [38]. Additionally, P53 overexpression was not found in BRCA2Mut tumors in 1 report [39]. Our meta-analysis found no difference in P53 protein expression in BRCA2Mut carriers and non-carriers.

At present, the methods for detection of site-directed mutation of p53 gene include direct method, indirect method, and biosense technique. Direct method is a technique for the detection of pathogenic gene mutations. The premise of direct method is that the normal sequence and structure of the detected genes have been elucidated. Direct sequencing (DS), PCR restriction fragment length polymorphism analysis (PCR-RFLP), single-strand conformation polymorphism analysis (SSCP), and high-resolution melting (HRM) are used in the direct method. Although these methods are used in clinical practice, it is common to have some problems: it is complicated and time-consuming, generates false-positive results, and requires expensive equipment. The indirect method is based on the expression of p53 protein and determines whether there is a point mutation of p53 gene. The indirect method includes immunohistochemistry (IHC), tissue microarray, flow cytometry, and enzyme-linked immunosorbent assay. IHC is widely used in clinical applications, but there are false-positive and false-negative results. Biosense technology has the advantages of good selectivity, high sensitivity, high speed, low cost, high automation, miniaturization, and integration. It provides a fast and simple method for basic medical research and clinical diagnosis.

The relationship between gene mutation and protein expression has long been controversial. Although there is a certain consistency, there are some differences. Gene mutations were detected in tissues of partial protein-negative expression, while the expression of protein-positive tissue showed no gene changes. The following types of mutations are predicted to escape detection: mutations in noncoding regions, which are estimated to account for a minimum of 10% of pathogenetic BRCA1 and BRCA2 mutations; large deletions undetectable by PCR-based assays; and mutations within the first and last 180 nucleotides of the amplicons analyzed by PTT.

Moderate heterogeneity was found in this study, which is reasonable considering the various races, ages, and different detection methods. The method of BRCA testing – asymmetrical funnel plots – indicated publication bias, perhaps due to diverse reasons (e.g., studies with favorable results are more likely to be published, poor methodological quality of small studies, inclusion of numerous studies without registration, true heterogeneity, artifactual results, and other causes).

There are several limitations to the current study. First, the eligible studies examined were conducted in different populations and were case-control studies; therefore, recall bias and selection bias were inevitable. Second, as only English articles were selected, language bias may be a factor. Third, positive results are more likely to be published than negative findings, and this may also be a source of bias. Fourth, various methods for assessing BRCA1Mut and BRCA2Mut were used in different eligible studies, and different standards were used to define positivity. Fourth, while detection of P53 was predominately via IHC, the antibody source, dilution rate, and cut-off value varied in different studies. Finally, in view of the relationship between protein expression and gene mutation, the present results should be regarded as an approximation.

BRCA1Mut and P53Mut are associated with poor prognosis in breast cancer, and P53 loss rescues the proliferative deficiency of BRCA1Mut cells. Assessing the presence of both mutations could serve as a potential auxiliary biomarker for breast cancer prognosis. The biological characteristics of BRCA1Mut or BRCA2Mut in breast cancers are distinctly different. Our findings suggest that the homologous recombination-deficient-targeting therapeutics used in treating p53-deficient tumors might be most effective in those tumors also carrying BRCA1 mutations.

Conclusions

Our meta-analysis suggests that BRCA1Mut breast cancer patients were more likely to have P53 overexpression compared with BRCA2Mut and non-carriers. This information may provide valuable guidance for clinicians performing related studies in the future.