Abstract
Introduction Malignant breast tumors are often hormone-dependent, and to this end, both estrogen and progesterone receptors are good prognostic markers for evaluation of the outcomes after therapy. In addition, HER-2/neu, whose expression is increasingly being associated with poor prognosis of breast cancer, has predictive potential after immunotherapy. Cytochrome P450 3A4 is highly involved in the metabolism of steroids. Thus, we investigated the impact of CYP 3A4 gene variants in association with clinical outcomes in African American (AFAM) versus Caucasian (CAU) patients with breast cancer diagnosis.
Methods Patients who had undergone biopsy procedures for diagnosis or for partial or radical mastectomy were recruited. The CYP 3A4 genotypes (A or G) were detected using polymerase chain reaction amplification and primers designed for a single nucleotide polymorphism. The messenger RNA (mRNA) transcripts were screened by reverse transcription-polymerase chain reaction. Clinical data including tumor staging, pathology grades, and family history were evaluated.
Results Frequency of the CYP 3A4-G (mutated variant) was significantly increased in AFAM patients as compared with controls (P < 0.001). No statistically significant difference was observed between the genotypes comparing the benign versus ductal carcinomas in situ (DCIS) or infiltrating ductal carcinomas (IDCAs). In AFAM patients, GG alleles were increased in IDCA with stage III tumors, and in CAU patients, the AA alleles were increased with stage III tumors. The mRNA expression was reduced in patients with IDCA versus DCIS or benign tumors (benign vs IDCA, P < 0.0009; DCIS vs IDCA, P < 0.005), as well in HER-2/neu-positive tumors versus samples negative for receptors (P < 0.0024).
Conclusions Genotype association was affected by race. Expression levels of total CYP 3A4 mRNA were inversely correlated with clinical diagnosis. This may suggest mRNA testing as an additional tool that accelerates improvement in the diagnosis of the onsets of breast cancer.
Breast cancer is the second leading cause of cancer-related deaths in women,1although a full etiology remains unknown; a synergy of environmental, genetic, and hormonal factors is known risk elements in the development and progression of breast cancer. Prolonged exposure to estrogens has been shown to be an influential factor for the induction of estrogen-associated cancers such as breast and endometrial cancers.2-4In addition, estrogens regulate the expression of human epidermal growth factor (HER-2/neu) protein whose expression is increasingly being associated with poor prognosis of breast cancer. Most breast tumors are hormone-dependent, and they express estrogen- receptors (ERs), progesterone receptors (PRs), or both. Thus, genes that are involved in the metabolizing process of these hormones may affect the development of breast cancer.
CYP 450 is a superfamily of structurally related enzymes, which metabolizes many physiologic compounds including steroids such as estrogen and progesterone.5,6In humans, the CYP 450 3A subfamily mainly represented by CYP 3A4 and CYP 3A5, but the CYP 3A4 is a major enzyme catalyzes the metabolism of both endogenous and exogenous agents, including estrogen that plays a central role in the etiology of breast cancer.7-9In addition, it has been reported that CYP 3A4 is involved in the 4- and 16α-hydroxylation of estrone, the predominant form of estrogens in postmenopausal women,10and is known to be genotoxic and may play a role in the development of cancer.11Many single nucleotide polymorphisms have been identified in the CYP 3A4 gene, of which some are in association with ethnicity and a few are associated with clinical conditions. CYP 3A4 mutated variant, presented with a transition of A→G at position −290 of the 5′-regulatory region of CYP 3A4 gene, is one of the frequently investigated genotypes that is effective in drug disposition.12-14The frequency of the CYP 3A4-G (mutated variant), which is also known as CYP 3A4-V or CYP 3A4*1B, varies substantially in different populations.15In addition, this variant has been association with several disease conditions, including clinical grades of prostate cancer,16-19breast cancer,20,21secondary leukemia,22hypercholesterolemia, diabetes, and allograft rejection.23,24
In our previous study testing breast tissue specimen for association between CYP 450 gene subfamily and breast cancer, we observed a greater association for CYP 3A4 in malignant breast tumors versus other CYP 450 genes including CYP 1A1, CYP 1A2, and CYP 1B1.25Thus, the goal of this study was to address the role of CYP 3A4 allelic variants with respect to prognostic factors such as tumor size, histologic grades, and hormone receptors in breast carcinomas. We also investigated whether there were differences in CYP 3A4 variant expression levels in association with clinical phenotypes and the ethnic background of the study population. Such variation would help early identification of clinical subtypes and therapeutic strategies.
MATERIALS AND METHODS
Patient Recruitment and Clinical Data Collection
Informed consent was obtained as part of a protocol approved by the University of Mississippi Medical Center Institutional Review Board for obtaining blood and tissue specimen from patients. The patients enrolled in this study were those undergoing various biopsy procedures including lumpectomy, needle localization, and simple mastectomy or modified radical mastectomy at the University of Mississippi Medical Center between 2005 and 2006. Patients were eligible for the study if they were 18 years or older and were diagnosed for first time for a possible breast cancer. From a group of 100 patients who signed the document to participate in the study, only 82 patients were included in this study. The remaining patients were excluded because of inadequate clinical documentation including diagnosis, history, and treatment regimen. Of the 82 patients selected to be part of this report, 51 were African American (AFAM) and 31 were Caucasian (CAU). African Americans constitute greater than 50% of the patient population in our clinic at the University of Mississippi Medical Center, which reflects the nonequality of the patient population in this study.
There were 179 female controls of both races. Controls were among the local volunteers from our other study population,26,27who had no history of any kinds of cancer and no autoimmune diseases.
Cancer staging and histologic grading were obtained from the Department of Pathology at University of Mississippi Medical Center. The malignant tumors were staged according to the TNM staging criteria where I is the least advanced and III is the most advanced. In addition, the tumors were graded according to Nottingham Histological criteria by morphologic features (tubule formation, nuclear pleomorphism, and mitotic count), where 1 of 3 is the least aggressive and 3 of 3 is the most aggressive type of tumors. All tumors were tested for ER, PR, and HER-2/neu receptor status.
CYP 3A4 Genotype Determination
Genotypes were determined by amplification of 186-bp CYP 3A4-G or a 187-bp CYP 3A4-A DNA fragment. A 440-bp internal control fragment flanking the mutation at position −290 was also amplified. For detection of variant alleles (A or G), we used modification of primers that was previously reported,18to generate the amplification primers. We excluded the restriction enzyme sites from the primer sequence. The forward primer sequence for 186-bp CYP 3A4-G was 5′-TGA-GGA-CAG-CCA-TAG-AGA-CAA-GGG-TAG-3′ and that for 187-bp CYP 3A4-A was 5′-A-TGA-GGA-CAG-CCA-TAG-AGA-CAA-GGG-TAA-3′. The primer pairs were designed to have a perfect match only with a single allele (A or G). Such genotype testing was based on the principle that was used in polymerase chain reaction (PCR) sequence-specific primer approach to test 1 single-nucleotide allelic variation. This was in part adopted from the technique that was used in Cytokine Genotyping Tray (One Lambda, Inc, Canoga Park, CA). The sequence-specific primers allowed amplification of target sequence when primers were perfectly matched or lack of amplification of target sequence when primers were not perfectly matched. For the detection of a 440-bp internal control fragment, a forward sequence at position −571 upstream from the coding region 5′- AAC-AGG-CGT-GGA-AAC-CCA-AT-3′ was generated.28All 3 forward primers shared the same reverse primer with complementary sequence of 5′- GAA-TCA-CAC-ACA-CAC-CAC-TCA-CTG-ACC-TC-3′ (−131 to −159).28Genotypes were tested by the PCR under a high-stringency temperature cycles to eliminate nonspecific amplifications. DNA was amplified in a total volume of 12.5 mL of reaction mixture by an initial cycle of 94°C for 2 minutes, followed by 25 cycles of 94°C for 30 seconds and 72°C for 60 seconds. The amplified DNA fragments were visualized by 2% agarose gel, and the genotypes were determined based on the presence or absence of the target DNA fragments (Fig. 1A). For verification of the genotypes, the DNA samples from 10 controls that were heterozygous AG were cloned and sequenced. Briefly, the amplified DNA fragment approximately 440 bp (same as the internal control fragment) was ligated with TOPO10 cloning vector (Invitrogen, Carlsbad, CA). Five white colonies each from 2 separate sets of amplification product were used for sequencing using both forward and reverse sequencing primers and a cycle sequencing kit (Applied Biosystems, Foster City, CA). Between a total of 10 colonies from each subject, it was confirmed that some of the colonies contained, for example, a 10-bp-long AGGGCAAGAG and others at the same location AGGGCAGGAG sequence from nucleotide −287 to −296 of the CYP 3A4 gene. The remaining of the sequence for A or G alleles as shown in Figure 1A was identical.
RNA Isolation and Reverse Transcription-PCR
For determination of the level of CYP 3A4 messenger RNA (mRNA) transcripts, RNA was isolated from the patient's peripheral blood mononuclear cells (PBMCs) using RNAeasy Kit according to the manufacturer's guideline (Qiagen, Valencia, CA). RNA concentration and purity were determined by RNA 6000 Nano Assay (Agilent Technologies, Palo Alto, CA). Afterward, RNA was transcribed by ImProm-II Reverse Transcription method (Promega, Madison, WI), followed by reverse transcription-PCR using a forward primer 5′-cct-tac-aca-tac-aca-ccc-ttt-gga-agt-3′ and a reverse primer 5′-agc-tca-atg-cat-gta-cag-aat-ccc-cgg-tta-3′. The resulting reverse transcription-PCR was normalized against β-actin from each samples (Fig. 1B). The intensity of the amplified banding pattern shown in Figure 1B was quantified using GDS-8000 (UVP) Image Analysis System.
Stains of Breast Cancer Tissue
As part of routine clinical diagnosis, tissue sections were stained with standard hematoxylin and eosin (H&E) method for determination of disease (Fig. 2A). Briefly, surgical specimens were fixed in formalin and embedded in paraffin. Five-micrometer-thick sections were prepared for histology and immunohistochemistry (IHC) study. For biopsy samples with invasive carcinomas determined by H&E stain, an IHC analysis was performed to demonstrate the presence of ER, PR, and HER-2/neu receptors. The primary antibody for testing the receptors was rabbit monoclonal antibody purchased from Ventana Medical System, Inc. The IHC stain was performed in BenchMark XT (Ventana Medical System, Inc, Tucson, AZ) robotic instrument. Interpretation of IHC stain was performed by a pathologist. A positive ER or PR stain was defined as more than 1% of nuclei stained; a positive HER-2/neu was defined as more than 30% of the invasive tumor cells strongly stained.
Statistical Analyses
The frequency distribution of CYP 3A4 genotypes for categorical data was computed by χ2 tests. A 2-sided χ2, odds ratio (OR), and P value were obtained from comparison between genotypes of individuals with disease versus control population using SPSS software (provided by University of Mississippi Medical Center computer services). Variables were considered significant for their effect as a risk factor if P ≤ 0.05. If the P > 0.05, the OR value was calculated from 2 × 2 contingency table to evaluate the risk factors for the possession of the CYP 3A4 genotypes. The mRNA transcript values were expressed as the mean (SD). Group differences for mRNA transcript levels between malignant and benign were determined by Student t test.
RESULTS
Demographic and Clinical Characteristics
Samples from a total of 51 (62%) AFAM and 31 (38%) CAU patients were studied. Demographic and clinical characteristics are shown in Table 1. There was no difference between the age of onsets (mean [SD] age, 42.2 [13.2] years for AFAM and 42.8 [12.8] years for CAU). Overall, 31 (61%) AFAM patients and 19 (61.3%) CAU patients were diagnosed with benign tumors with H&E stain. The remaining, which includes patients with ductal carcinoma in situ (DCIS) and infiltrating ductal carcinoma (IDCA), is shown in Table 1. For primary tumors, AFAM patients were presented equally for stage IIA/IIB or III (35.7%), whereas of CAU patients who were diagnosed with IDCA, 10% had stage IIA/IIB and 70% had stage III tumors. The differences were not statistically significant. Although more patients reported a family history of any cancer rather than none, there was statistically a borderline difference between those patients with a family history of breast cancer comparing CAU versus AFAM (AFAM 29.4% vs CAU 42.9%, P < 0.06; OR = 1.7; 95% CI, 0.4-3.52). In addition, a greater percentage of AFAM had a family history of other types of cancer as compared with the CAU (70.6% vs 58.1%, respectively) but was not statistically significant. A greater percentage of CAU patients were identified with ER- or PR-positive (ER+/PR+) tumors than the AFAM (50% vs 30%, respectively). There was no difference between the HER-2/neu-positive receptor in both groups. There were more ER-, PR-, and HER-2/neu-negative tumors in AFAM patients than in CAU (25% vs 8.3%, respectively).
Frequency Distribution of CYP 3A4 Genotypes
In a global analysis, CYP 3A4 homozygous AA alleles predominantly were among the CAU patients and CYP 3A4 homozygous GG alleles were in AFAM patients. However, homozygous GG alleles were only slightly increased in AFAM patients with malignant tumors compared with benign tumors. The homozygous GG alleles were not observed in CAU patients. The CAU patients and controls had a higher frequency of homozygous AA alleles compared with AFAM patients and controls (Table 2). There was an excess frequency distribution of the heterozygous AG alleles among AFAM healthy controls compared with AFAM patients with either malignant or benign tumors (P < 0.01), and homozygous GG alleles were significantly reduced in AFAM controls as compared with patients with either malignant or benign tumors (P < 0.001), suggesting the influence of the G alleles in tumor cell proliferation but not necessarily in the malignancy. The frequency distribution of the genotypes seems not to support the Hardy-Weinberg equilibrium, which may be resulted from population admixture and or a limitation in sample size. Among AFAM patients who had benign tumors with a pathological diagnosis of fibroadenoma, 50% carried homozygous AA alleles compared with 23% with AG and 28.6% GG alleles (Table 2). Such association was not observed in CAU patients, assuming because of the low frequency of G variant or a low rate for the development of fibroadenoma among the CAU, which need further studies.
Correlation Analysis of Genotypes and Clinical Diagnosis
Patients with benign diagnoses such as fibroadenoma or hyperplasia were compared to patients with malignant diagnoses such as DCIS or IDCAs for genotype association. There was no statistically significant difference between carrying homozygous AA or GG alleles versus heterozygous AG alleles for differentiating between the benign versus DCIS or IDCA comparing the patients with the controls in both populations (data not shown).
Correlation Analysis of Genotypes, Tumor Staging, or Nottingham Histologic Grades
The genotypes were tested against the TNM staging and the Nottingham Histological Criteria (Table 3). Such pathological database has significant prognostic values and may help to improve clinical diagnosis and the treatment planning. In AFAM, homozygous G allele was 4-fold increased in patients who were diagnosed with stage III tumors as compared with the other genotypes (GG, 20%; AG, 5%; AA, 0%). In CAU, homozygous AA allele was increased 2.5-fold in patients who were diagnosed with stage III tumors (AA, 41.66%; AG, 16.6%; GG, 0%); however, the data were statistically insignificant. Similarly, most AFAM patients with histologic grade 3/3 carried homozygous GG genotypes and most CAU patients with histologic grade 3/3 carried homozygous AA genotypes. These data, which are shown in Table 3, suggest that ethnic genotype might affect the progression of tumor size and diffusion.
Distribution Profile of mRNA Expression in Association With Disease
All 82 patients' samples were tested for mRNA expression profiling of which 32 were malignant and 50 were benign. The mean (SD) mRNA transcripts in PBMCs of patients with benign or malignant tumors were compared in Figure 2A. The CYP 3A4 expression was significantly reduced in patients diagnosed with IDCA as compared with DCIS or benign tumors (benign vs DCIS, P < 0.7; benign vs IDCA, P < 0.0009; DCIS vs IDCA, P < 0.005).
It has been reported that the G variant is a higher producing genotype over the A, the wild type. Thus, we tested the mRNA levels in the context of genotypes versus clinical status and the controls. Because of an extreme inequality distribution between G and A alleles in the study population, the data from the 2 population were combined where it was needed. As shown in Figure 2B, the levels of mRNA transcript expression were increased in patients as well as in controls that were homozygous for G allele as compared with those with homozygous A. However, the differences between expression levels were statistically insignificant. A significant difference was observed between the levels of mRNA expression and clinical status, which was independent of the genotypes (benign or DCIS vs IDCA, P < 0.004). The mRNA expression in control samples was not statistically different from the samples with benign or DCIS diagnosis.
Correlation Analysis Between Genotypes and Receptors
Genotype association was evaluated in the presence or absence of the receptors. In all samples, PR+ was present with ER+. Because of the limitation in sample size, data from the 2 study population were combined. The percentage of patients with receptor-positive tumors and the genotype they carried is given in Figure 3A. The ER+ samples were positively associated with homozygous AA alleles but inversely associated with homozygous G variant. A true association value for PR+ was unobservable because of the presence of ER+. The HER-2-positive samples were associated with increased frequency of homozygous GG genotypes compared with AA genotypes (GG 71.42% vs AA 14.28%, P < 0.1; OR = 11.6). Because of the sample size, data were statistically insignificant. Because the level of CYP 3A4 expression is sensitive to hormone metabolism, we tested the mRNA levels in the PBMCs of patients whose tumors were either ER+ (ER+ + PR+) or HER-2+ and compared with those samples that were receptor-negative. The mean (SD) mRNA transcript level was reduced in patients with positive receptors, particularly HER-2+, compared with those patients with negative receptor status. Furthermore, as shown in Figure 3B, the level of mRNA was significantly lower in patients whose tumors were HER-2+ compared with receptor-negative and those with ER+ and PR+ (HER-2+ vs receptor-negative, P < 0.0024), regardless of race.
DISCUSSION
We investigated CYP 3A4 gene variants and its role in the differential status of breast carcinoma. Most genetic studies of breast cancer are associated with specific genes such as BRCA-1 and BRCA-2 that have direct impact on the development of breast cancer.29Here, we have studied common genes such as the CYP 450 that have a preferential effect on breast tumors. To this end, most investigations have focused on CYP 1B1, a predominant member of the CYP1 family, which is expressed in normal breast tissue and breast cancer.21,30,31We have focused on the CYP 3A4, a predominant variant of CYP 3A family, which is well known for its variability in the population and the importance in drug interaction and metabolism.9,14The molecular classification of breast carcinoma has impact on early diagnosis and clinical management of malignant breast tumors. The goal of our study was to validate the role of CYP 3A4 gene polymorphism and the association with tumor stage, grade, and hormone receptor status in AFAM and CAU patients with breast cancer.
The peripheral blood cells of women who had undergone breast biopsies for either malignant or benign lesions were tested for variation in genotype and levels of the CYP 3A4 expression in association with clinical status such as tumor characteristics and the progression of disease in the study population. Although the clinical and pathologic approaches offer a significant degree of prognostic information, there are varying levels of diagnosis, originating from technical variations in processing the tissues. Thus, molecular approaches such as genotype association studies are expected to have a significant effect on prognosis and the treatment strategies. The study of CYP 3A4 genotype variation has at least a 2-fold importance. First, the ethnic distribution of the CYP 3A4-G variant, which is 82% in AFAM and 6% in CAU, may confer genetic susceptibility to disease in the population. Second, the CYP 3A4 genotype variant might account for the differences in hormone metabolisms between the individuals. In addition, the genotype frequency differences among the AFAM and CAU patients might have association with clinical manifestation of the disease such as fibroadenoma, tumor staging, and invasiveness (grading) of the tumors, predisposing the individuals to the development of the disease. For example, we observed a higher percentage of homozygous AA genotype, among AFAM patients with fibroadenoma, a CYP 3A4 genotype that has a low frequency in AFAM population, compared with G (mutated) variant. A detailed analysis in a larger study population of both AFAM and CAU patients are required, which is lacking in this study.
The prevalence of CYP 3A4-G genotype in association with pathological conditions such as prostate cancer,16-19breast cancer,20,21and secondary leukemia22has been reported. However, some discrepancies within the association of CYP 3A4 genotypes and the risk of breast cancer have also been reported.32,33In the latter studies, no association between CYP 3A4 polymorphism and breast cancer as well as with estrogen production was found. There could be several reasons, including excess of CYP 3A4 variants such as homozygous AA versus GG genotypes, which influence the outcome of association in the population. In our study, an excess of homozygous GG alleles was observed in AFAM patients with malignant tumors, and an excess of homozygous AA was observed in CAU patients with malignant tumors. In addition, there was an excess of heterozygous AG alleles and an underrepresentation of the GG alleles among the AFAM controls, causing disequilibrium in the Hardy-Weinberg theory and suggesting a possible admixture effects on the outcomes. Frequency distribution of the genotypes among the AFAM controls was different from that reported by other investigators.18However, this discrepancy in the genotype distribution was not effective in the context of clinical outcomes because there was no significant difference between the genotypes of patients with malignant tumors versus patients with benign tumors. The CYP 3A4-G genotype or the mutant allele has been reported the major variant among the population of African origin.18,34An estimated frequency distribution of the G allele ranged from 0.036 to 0.096 in CAUs, 0.089 in Saudis, 0.69 in Ghanaians, 0.48 to 0.80 in AFAM, and none in Chinese, Taiwanese, or Japanese population.33-35
The discrepancy observed in this study might be a result of extreme population admixture, which has increased the likelihood of heterozygosity and a disruption of previously demonstrated genotype frequency. In studies where the genotype frequency differences are associated with ethnic background of the population, such as the case with CYP 3A4 variants, association of a specific allele with clinical characteristics of the disease requires a well-characterized large patient population. We considered our patient population to be a well-characterized group, with a weak power for the study analysis. However, it should be pointed out that our study was not a case-control study, and the control subjects, both AFAMs and CAUs, were from local volunteers who were characterized only based on disease history and on generations of ancestry background.
Genotyping studies of CYP 3A4 have not demonstrated solid evidence for the association with breast cancer.32,33Likewise, in this study, the CYP 3A4 polymorphism did not significantly modify the risk for the disease. However, in AFAM population, we found a greater risk of fibroadenoma formation in carriers of homozygous AA alleles. Fifty percent of the patients who had benign tumors with a pathological diagnosis of fibroadenoma were homozygous AA compared with 28.5% homozygous GG alleles. This further strengthens the association between GG alleles and malignancy in the AFAM population with the notion that the CYP 3A4 variant may have clinical significance in predicting the drug therapy response as well as monitoring the AFAM patients during treatment, which was not the focus of this study. A number of studies demonstrated an increased level of CYP 3A4 mRNA expression in breast cancer tissue in association with poor therapeutic outcomes.25,36-38These studies suggested that the increased expression level of CYP 3A4 in association with drug metabolism might be the result of accelerated degradation of the therapeutic agents in breast cancer therapy. Although mRNA expression data in our study were not equivalent to the functional presentation of the enzymatic activity of the CYP 3A4 gene, the results may indirectly suggest a possible correlation between the interindividual variability in CYP 3A4 mRNA expression and metabolic activities causing an estrogen overload, significant in the development of breast cancer.
Our previous study that was based on the mRNA from breast tissues demonstrated an increased level of CYP 3A4 mRNA transcripts in normal breast tissues compared with malignant breast tissues.25In current study, using the peripheral blood, we demonstrated increased expression levels of CYP 3A4 in PBMCs of patients with benign tumors as well as controls with no history of cancer and autoimmune disease compared with individuals with malignant tumors. These data support the fact that the level of expression might inversely affect clinical outcome of the disease. Whether lower levels of mRNA expression have any association with PR+ was not observable from this study. However, the observation that CYP 3A4-G variant was increased in patients with HER-2/neu receptors (a receptor whose expression is associated with increased disease recurrence and worse prognosis), and the mRNA expression that was reduced in samples that the tumor was positive for HER-2/neu receptors (Fig. 3, A and B), suggests that these 3 data points, namely, G alleles, HER-2/neu, and CYP 3A4 expression level together, may help to improve the prognosis of the disease in AFAM patients with breast cancer.
Overall, this study demonstrated a clinical relevance and significance for subclinical characteristics of breast cancer. Associations between genotypes and ER+ and PR+ were affected by race, whereas the HER-2 receptor had a true association with G allele, suggesting possible additive effects on the risk of invasive disease. Using patient's peripheral blood cells, which are readily accessible, as opposed to current biopsy methods, might provide a possibility to establish a convenient, inexpensive, and noninvasive method for the detection and monitoring of recurrence as well as identification of the onset of breast cancer in AFAM patients.
ACKNOWLEDGMENTS
The authors thank patients and the clinical staff in the breast clinic and surgical units in the Department of Surgery and Cancer Center at the University of Mississippi Medical Center for their support and cooperation. Our special thanks to Ms Smith, RN, for her timely efforts of obtaining the consent documents and for drawing blood samples.