Breast cancer (BC) is the third most common world wide malignancy among women, and is increasing in incidence.1 The high frequency of BC emphasizes the need to understand the mechanisms involved in breast tumorigenesis.2 Reactive oxygen species (ROS), which induce oxidative damage, are also implicated in malignant transformation and may influence cancer risk.3 In fact, ROS may damage DNA, proteins and fatty acids.4 On the other hand, accumulating data suggest that ROS may also have important cancer protective functions.5
Lipid peroxidation is perhaps the most extensively investigated free radical-inducing and ROS-related process. The role of lipid peroxidation in BC initiation is controversial and strongly debated. It is considered harmful and carcinogenic,6, 7 but it also regulates growth inhibition and cell death.8 Several studies have found enhanced lipid peroxidation and depletion of antioxidants in the plasma of BC patients, suggesting that increased oxidative stress may be related to human breast carcinogenesis.9, 10, 11, 12, 13 Moreover, an increased content of lipid peroxidation products was found in BC tissue, supporting the hypothesis of breast tumorigenesis by in situ oxidative stress.14, 15 On the other hand, recent reports suggest that lipid peroxidation may represent a protective mechanism in BC through the induction of apoptosis, thereby maintaining the balance between breast epithelial cell growth and death,8, 16 supporting the hypothesis that lipid peroxidation in tissue of benign breast disease (BBD) is greater than in BC.17
The adult nonpregnant, nonlactating breast secretes fluid into the breast ductal system that can be easily and noninvasively obtained through nipple aspiration. Nipple aspirate fluid (NAF) consists of secreted proteins and cells sloughed from ductal and lobular epithelium,18 and contains several biomarkers19, 20 that are potentially useful as epidemiological and clinical research tools.
Because BC develops from ductal and lobular epithelium, the analysis of NAF has attracted considerable interest as a method to assess metabolic activity within the mammary gland.21 To nonivasively assess the level of lipid peroxidation in the breast as a marker of oxidative stress, and determine the potential utility of lipid peroxidation markers in clinical studies, we examined the levels of the 2 major lipid peroxidation products, malondialdehyde (MDA) and the 8-epimer of prostaglandin F2α (8-iso-PGF2α), in NAF and plasma samples from healthy controls and women with malignant breast disease. The aim of this study was to evaluate the involvement of lipid peroxidation products in BC initiation/evolution, directly analyzing their secretion by breast ductal/lobular cells. To demonstrate if the breast cells are able to produce and secrete/accumulate lipid peroxidation products during physiologic lactation, we also analyzed milk samples from healthy subjects.
Material and methods
Subjects and sample collection
Women were recruited and enrolled as unselected consecutive patient populations at 2 sites, the Center of Senology of Pesaro-Urbino, Italy and Ellis Fischel Cancer Center in Columbia, MO, USA. Demographic data for analyzed subjects are reported in Table I. The present work was carried out in accordance with the ethical standards of the Helsinky Declaration of 1975 (as revised in 1983) and after the approval of the Ethics Committee of the University “Carlo Bo” of Urbino (protocol 18/CE) and the Institutional Review Board at Ellis Fischel Cancer Center. All subjects (170 women) signed informed consent prior to enrollment. We excluded 8 subjects due to pregnancy within 3 years, and 22 patients for the inability in collecting NAF. After the exclusion, NAF samples were analyzed from the remaining 140 nonlactating women (age range 30–67 years): 113 healthy women without evidence of BC (noncancer), 3 patients with BBD (i.e., epithelial hyperplasia, sclerosing adenosis and intraductal papilloma) at the time of NAF collection who subsequently developed biopsy proven BC and 24 patients with biopsy proven infiltrating duct carcinoma (cancer). NAF samples were collected from the affected breast of BC patients prior to biopsy. The fluid was collected noninvasively using a modified breast pump in calibrated capillary tubes as described previously,18, 21 the volume recorded and the ends sealed with clay. The median volume of NAF collected was 25 μL (range 5–300 μL). Without pooling, the supernatants were snap-frozen at −80°C until use. We also collected milk samples from 10 healthy women during the first month of lactation, and after centrifugation the clear supernatants were stored at −80°C. NAF and milk samples were firstly diluted and then analyzed for total protein by the Bradford method (BioRad, Milan, Italy) and lipid peroxidation products. The median age of the lactating women did not significantly differ from that of the other control patients recruited. To avoid possible age differences between cases and controls, we have performed the age-adjustment of our data. Since plasma specimens were not available for all of the women recruited, we have analyzed matched plasma and NAF samples only from noncancer women (n = 20) and patients with BC (n = 15).
Table I. Demographic Data for Women Analyzedin The Present Study (n = 140)
All samples were assayed in duplicate. Intra- and interassay CVs were determined. We also determined the analytical recovery of different concentrations of purified MDA and 8-iso-PGF2α added to the NAF samples.21 All the reagents were purchased from Sigma-Aldrich (Milan, Italy).
MDA was measured by reverse-phase HPLC through a rapid, sensitive extraction method for its determination in biological fluids.22 Sample derivatization was carried out by adding NAF to a reaction mixture containing 0.05% butylated hydroxytoluene, 0.44 M H3PO4 and 42 mM thiobarbituric acid. After vortexing and heating for 1 hr at 100°C, the samples were placed on ice and the MDA-thiobarbituric acid complex was extracted with butanol after centrifugation at 10,000g. An aliquot of the butanol layer was placed into an HPLC (Jasco, Tokyo, Japan) for analyses without evaporation. The assay was performed using an Alltima C18 cartridge column (4.6 × 250 mm, 5 μm) equipped with a guard column Alltima C18 (4.6 × 7.5 mm, 5 μm) (Alltech, Milan, Italy). The eluent phase was a 40:60 (v/v) methanol buffer consisting of 50 mM phosphate pH 6.8. The flow rate was 0.8 mL/min. UV detection was carried out at 532 nm, the fluorescence detector was set at an excitation wavelength of 515 nm and an emission wavelength of 553 nm. The MDA concentration was expressed as μmol/L.
8-Isoprostanes were determined with a competitive immunoenzymatic kit from Assay Designs (Ann Arbor, MI) evaluating 8-iso-PGF2α levels in biological fluids using Sep-Pack procedure. According to the manufacturer's instructions, hydrolysis of lipoprotein or phospholipid coupled to 8-iso-PGF2α was performed to ensure that the measured 8-iso-PGF2α was a reflection of both free and esterified isoprostanes. To hydrolyze the ester bond, samples were treated with 2 N NaOH for 2 hr at 45°C and then neutralized by concentrated HCl. After incubation at 4°C for 18 hr, the color generated by substrate-cleavage of alkaline phosphatase bound to the polyclonal antibody was read on a microplate reader at 405 nm (BGM LabTech, Frankfurt, Germany). The crossreactivities for a number of related eicosanoid compounds were <1.5% for prostaglandin F2α and negligible (<0.02%) for other 8-iso-prostaglandins as well as for prostaglandins A, B, E and thromboxanes. We then determined the analytical recovery of 3 concentrations of purified 8-iso-PGF2α added to the NAF samples to test the stability and recovery of 8-iso-PGF2α through the hydrolysis procedure. The intensity of the bound yellow color was inversely proportional to the concentration of 8-iso-PGF2α in either standards or samples, expressed as pg/mL.
The data are presented as median and range (minimum and maximum). Comparisons between groups (cancer versus noncancer) were conducted by using the Mann–Whitney U-test, since underlying conditions for parametric tests were not met. Spearman test was used for correlation analyses. In all instance significance was set at p < 0.05. Data were analyzed with Prism software for Windows, version 3.1 (Graph-Pad, San Diego, CA).
Among the clinical/demographic variables compared, only age was significantly related to cancer status. Women with BC tended to be significantly older than their noncancer counterparts (Table I). Although the age was different between cases and controls, no statistically significant differences in biomarker levels were found after age-adjustment. NAF levels of MDA and 8-iso-PGF2α showed no statistically significant difference among the women enrolled from the 2 recruitment sites (Table II). Moreover, median values of both MDA and 8-iso-PGF2α in the cancer and noncancer subgroups did not significantly differ between Italian and USA patients (data not shown).
Table II. Lipid Peroxidation Biomarker Median Levels in NAF Samples Based on the Recruitment Sites
Department of Surgery, Ellis Fischel Cancer Center, University of Missouri, Columbia, MO, USA.
8-Epimer prostaglandin F2α (pg/mL)
Consistency of the assays
The correlation coefficients (r2) between lipid peroxidation products and dilution were 0.98 and 0.96 for MDA and 8-iso-PGF2α, respectively. The percentage means of the analytical recovery of purified compounds added to the samples were (94 ± 7)% and (97 ± 5)% for MDA and 8-iso-PGF2α, respectively. Intra- and interassay CVs were 5 and 9% for MDA and 8-iso-PGF2α, respectively. These studies suggest that the NAF “matrix” (i.e., proteins, hormones and lipids present in NAF) did not affect the performance of either assay.
Malondialdehyde is detectable in plasma, NAF and milk
MDA was detected in all plasma samples, in 80% (8 out of 10) of milk and 95% (133 out of 140) of NAF samples. The median level of MDA in both plasma (p = 0.029) and NAF (p = 0.016) were significantly higher than that detected in milk (Table III). Median MDA levels in NAF were 1.7-fold higher than in matched plasma (p = 0.032).
Table III. Median Levels of Lipid Peroxidation Products in All Samples
Plasma MDA levels were not related to the presence of BC, with no significant difference between noncancer and cancer subgroups (Table IV). Only NAF from BBD patients who subsequently developed BC showed significantly higher median levels of MDA compared to noncancer women (p = 0.024).
Table IV. Malondialdehyde and 8-Epimer Prostaglandin F2α Median Levels in NAF and Plasma Samplesbased on the Cancer Presence
Values are expressed as medians and ranges. NAF: 24 cancer patients, 113 noncancer women, 3 patients affected by BBD that developed BC. Plasma: 15 Cancer patients, 20 noncancer women, 3 patients affected by BBD that developed BC.
No significant correlation between MDA levels in plasma and NAF was found.
8-iso-Prostaglandin F2α is detectable in plasma, NAF and milk
8-iso-PGF2α was detected in all samples analyzed in the present study, with the lowest levels in plasma (Table III). The median concentration in milk was 3.2-fold higher (486.5 vs. 154.2 pg/mL, p = 0.0015) and in NAF was 90.6-fold higher (p < 0.0001) than in matched plasma.
Plasma levels of 8-iso-PGF2α were not significantly related to BC (Table IV). 8-iso-PGF2α concentrations were higher in NAF samples from noncancer subjects (18,743 pg/mL, n = 116) than from subjects with BC (1,803 pg/mL, n = 24, p < 0.0001) (Fig. 1).
The median concentrations of 8-iso-PGF2α in NAF samples (n = 3) from patients diagnosed with BBD at the time of NAF collection who within 3 years developed BC were significantly lower than concentrations in both noncancer and cancer patients (p = 0.0064 and p = 0.0187, respectively) (Table IV). No significant correlation between 8-iso-PGF2α levels in plasma and NAF was found.
Finally, even though the correlation test does not reach the significance level, we found an inverse correlation trend between 8-iso-PGF2α levels in NAF from cancer patients and disease stage (r2 = 0.145, p = 0.067) (Fig. 2).
Although several lipid peroxidation products have been previously identified in plasma and tissues of women with breast disease, as well as in human milk,9, 10, 11, 12, 13, 14, 15, 16, 23, 24 we report for the first time the detection of both MDA and 8-iso-PGF2α, 2 major indexes of lipid peroxidation and oxidative stress, in NAF from healthy women and patients with BC.4, 5, 6, 7
We found that breast milk from healthy women contains lower levels of MDA than matched plasma, postulating that this difference may be related to different metabolism of free fatty acids (e.g. lipoprotein lipase activity).25 Our results suggest that lipid peroxidation naturally occurs in the lactating breast and that enhanced lipid metabolism may induce interactions between ROS and polyunsaturated fatty acids, releasing toxic and highly reactive aldehyde metabolites (e.g., MDA, one-end products of lipid peroxidation).3, 4
Although the plasma level of the lipid peroxidation marker isoprostane has been inversely related to a low-fat diet and positively correlated to body mass index,26 it has been recently demonstrated in a population-based case–control study that urinary isoprostane and oxidative marker concentrations were not significantly altered by age, physical activity, fruit and vegetable intake, alcohol intake, cigarette smoking, body mass index, nor menopausal status.23 Moreover, a randomized intervention trial reported that some isoprostanes and prostaglandins are only marginally affected by low fat and high vegetable/fruit diets.27, 28 Our results in plasma and NAF samples suggest that, even though the differences in lifestyle between the 2 recruitment sites (Urbino, Italy and Columbia, MO USA) may exist, they did not significantly affect the lipid peroxidation status. Moreover, after age-adjustment, no significant differences were found in both biomarker expression between cases and controls.
We found higher levels of MDA in NAF than in matched plasma. Our results argue against passive plasma filtration as the sole source of MDA in NAF, and are consistent with previous findings demonstrating that enhanced lipid peroxidation occurs in BC tissue.15 The high levels of MDA detected in NAF samples from women originally diagnosed as BBD who subsequently developed BC provide evidence that MDA (as index of oxidative stress) may be synthesized by highly metabolizing apocrine/ductal cells.29, 30 Previous studies suggest that MDA levels in plasma, tissues and human cell lines may be related to tumor progression,25, 31 and our findings in the NAF of women who progressed to cancer are consistent with these reports.11, 14
8-iso-PGF2α (arachidonic acid products damaged by ROS/free radicals) is esterified, hydrolyzed and released into the extracellular fluid through phospholipase A2 activity.8, 32 Although 8-iso-PGF2α has been widely employed as a quantitative in vivo biomarker of oxidative damage,32 its role and function are controversial.33, 34 In agreement with previous reports,35, 36 we observed that 8-iso-PGF2α levels were significantly higher in NAF than in matched plasma (p < 0.0001). 8-iso-PGF2α levels in NAF and plasma were not correlated, indicating that NAF 8-iso-PGF2α may reflect local production and/or secretion by breast cells into the duct lumen. We also found high levels of 8-iso-PGF2α in human breast milk, suggesting a possible role in normal breast physiology,8 even though high levels are not related to lactation history.37
Increased peroxidation of cellular lipids (e.g., polyunsaturated fatty acids, free and conjugated linoleic acid and arachidonic acid) generates compounds inhibiting the growth of BC cells both in vivo and in vitro.7, 37 Similarly, lipid peroxidation products (such as 8-iso-PGF2α derived from ROS-damaged arachidonic acid) may inhibit cell proliferation and induce differentiation, maturation and apoptotic turnover of breast tissue,8, 38 such as is necessary both during (e.g. to balance proliferation) and after lactation (e.g. to allow involution back to the prepregnant anatomy and physiology of the breast).38
Our results are consistent with the hypothesis that 8-iso-PGF2α is a product produced in normal breast physiology whose production is impaired with the onset of BC.7, 8 In fact, NAF obtained from healthy subjects with no cancer contained significantly higher levels of 8-iso-PGF2α than NAF samples collected from patients with BC. Interestingly, NAF levels of 8-iso-PGF2α in BBD patients who subsequently developed BC were significantly lower than healthy women without cancer (p = 0.007), whereas no significant differences were found between healthy controls and BBD that did not develop BC.
Although ROS may react with polyunsaturated fatty acids, inducing production/release of reactive aldehyde metabolites (e.g. MDA, end product of lipid peroxidation involved in breast tumorigenesis),25 our findings show that MDA does not seem a useful NAF biomarker to discriminate which women do or do not have cancer. On the contrary, high levels of 8-iso-PGF2α in human breast milk and in NAF from healthy women suggest a direct or indirect physiologic role which is lost in BC. Alternatively, 8-iso-PGF2α may exert a protective function against BC initiation, development and/or progression.39, 40 The latter hypothesis is consistent with a report suggesting that decreased lipid peroxidation leads to increased BC risk.8
NAF contains high levels of oxidized products of cholesterol and lipids (e.g. cholesterol epoxides).41, 42, 43 These products have been linked to BC risk.20 Although the role of the lipid peroxidation pathway(s) in breast carcinogenesis is controversial,6, 8 we hypothesize that the high levels of 8-iso-PGF2α may play a role/function in BC protection/prevention, whereas the low synthesis/secretion of 8-iso-PGF2α by breast ductal cells may help to detect, through the noninvasive method of NAF sampling, women with premalignant conditions prone to BC.2, 18
Isoprostanes are prostaglandin-like compounds produced primarily from esterified arachidonic acid by nonenzymatic reactions catalyzed by free radicals, thus not requiring cyclooxygenases for their formation.32 To commence isoprostane formation, ROS can extract hydrogen atoms from polyunsaturated fatty acids (as well as from arachidonic acid released by phospholipase A2 from membrane phospholipids) under aerobic conditions.33 The formed pentadienyl radical combines with an oxygen molecule to generate bicyclic endoperoxy radicals that react with additional oxygen molecules to produce hydroperoxy bicyclic endoperoxy radicals. Finally, in the presence of an appropriate donor molecule (such as glutathione) the radical chain reaction produces G isoprostanes (analogues of G2 prostaglandins), that rapidly isomerize to give a variety of products including analogues of prostaglandins E and D.41 Physiologically, isoprostanes of F series are produced through the reduction of G isoprostanes via natural endogenous reductants (such as glutathione and glutathione peroxidase).
Our results support the hypothesis that under physiologic conditions, breast ductal/lobular cells release arachidonic acid from phospholipids through the action of phospholipase A2. Arachidonate is converted by COX-1 into prostaglandins. In the presence of ROS, the arachidonate molecules or polyunsaturated fatty acids may be catabolized to G isoprostanes G (analogues to G2 prostaglandins) (Fig. 3). In this step either of 2 paths may occur: (i) in women without cancer, in the presence of normal levels of natural endogenous reductants (i.e., glutathione and glutathione peroxidase), G isoprostanes are reduced to F isoprostanes (e.g., 8-iso-PGF2α detected in high levels in NAF from healthy women found in the present study); (ii) in patients with BC, low levels of endogenous reductants44 may strongly limit the reduction of G isoprostanes to F isoprostanes, favoring the isomerization to analogues of prostaglandins E and rearrangement products. Our findings of significantly reduced levels of 8-iso-PGF2α in NAF from cancer patients agrees with this hypothesis. If this is true, then G isoprostanes likely accumulate in breast tissue and may isomerize rapidly to analogues of prostaglandins E and D, which might participate (through autocrine and paracrine fashion) to the increased biosynthesis of estrogens by the intracellular induction of cyclic AMP-dependent of CYP19 aromatase gene, as previously demonstrated for prostaglandins both in vitro and in vivo.45 Actually no data are available about the possible involvement of isoprostanes in estrogen induction. The evidence that prostaglandin E is accumulated in NAFs from BC patients,46 agrees with our hypothesis. Moreover, the lack of effect of Celecoxib (a COX-2 inhibitor) on prostaglandin E reduction46 may be due to the nonenzymatic production of isoprostanes G that, in absence of sufficient endogenous reductants, may rapidly isomerize to analogues of prostaglandins E independently of the COX2-dependent prostaglandin E production47 (Fig. 3). Moreover, prostaglandin E2 (obtained by the coupled activity of COX-2 and prostaglandin E2 syntase), during cancer development may favor cell adhesion, motility, invasion and promote angiogenesis, inhibiting also apoptosis through the activation of serine/threonine protein kinase B or Akt activity.48
On the other hand, it should be kept in mind that abundant experimental evidence (excellently reviewed in Ref. 8) support a direct role of lipid peroxidation products in BC protection. In general, evidences suggest a role of lipid peroxidation products (e.g. MDA, isoprostanes, cyclopentenone compounds, prostaglandin analogues, hydroxy-2E-nonenals, etc.) in the control of cell proliferation and induction of differentiation, maturation and apoptosis,31, 39, 41, 49 suggesting that lipid peroxidation and ROS are triggers and essential mediators of apoptosis. Normal human breast epithelial cells undergo a continuous cycle of cell proliferation, differentiation and apoptosis, processes required for development, maturation and tissue turnover, also crucially linked in mammary tumor development. The proliferative activity of the mammary epithelium, (e.g. during menstrual cycles and pregnancy) may be substantially influenced by lipid peroxidation/ROS which may provide an important component for cell death and turnover, suggesting beneficial effects in eliminating precancerous, cancerous, virus-infected, chemotherapeutic damaged breast cells.38, 39, 40, 50 Although the conventional consideration of lipid peroxidation is an undesirable cytotoxic process, the anticancer protective effect of lipid peroxidation is mainly linked to apoptotic cell death through inhibition of cell proliferation, induction of cell cycle block, cytochrome c release-induced apoptosis and survivin expression. Even though the underlying molecular mechanisms are still not completely understood, evidences strongly rise the possibility that increased lipid peroxidation may be a mechanism responsible “at least in part” for the decreased risk associated with several hormonal and nonhormonal factors for BC.8 For example, prostaglandin E2 (obtained by the coupled activity of COX-2 and prostaglandin E2 syntase) during cancer development may favor cell adhesion, motility, invasion and promote angiogenesis, inhibiting also apoptosis through the activation of Akt/Protein kinase B activity.48 Accordingly, the inhibition of COX by NSAIDs drugs enhanced apoptotic DNA fragmentation, intracellular production of ROS and released high amounts of isoprostanes.51 So, the higher level of ROS and lipid peroxidation products would induce apoptosis of transformed, precancerous or cancerous cells in both highly proliferating tissues and mitotically competent tissue (like the human breast) and therefore, would have the potential to protect against cancer.6, 8, 38, 41
Further biomolecular evaluations are in progress in our laboratories to provide insights on the balance between lipid peroxidation products (e.g., compounds related to phospholipase A2 activity) and antioxidant molecules (e.g., reduced and oxidized glutathione, glutathione peroxidase, glutathione reductase and superoxide dismutase activities), as well as about the possible role/involvement of isoprostanes (i.e., analogues of several prostaglandins), protein peroxidation products (e.g., carbonyls). These studies may clarify the biological and cellular mechanisms of lipid peroxidation pathway(s) occurring in human breast physiology and carcinogenesis.
We would like to thank Dr. M. Sebastiani (Center of Senology of Pesaro, Italy), and Mrs. L. Schlatter (Department of Surgery, University of Missouri, USA) for their assistance in patient recruitment in collecting NAF, milk and plasma samples, and in gathering patients' clinical data.