Abstract
Obesity and alterations in metabolic programming from early diet exposures can affect the propensity to disease in later life. Through dietary manipulation, developing mouse pups were exposed to a hyperinsulinemic, hyperglycemic milieu during three developmental phases: gestation, lactation, and postweaning. Analyses showed that a postweaning high fat/high sugar (HF/HS) diet had the main negative effect on adult body weight, glucose tolerance, and insulin resistance. However, dimethylbenz[a]anthracene (DMBA)-induced carcinogenesis revealed that animals born to a mother fed a HF/HS gestation diet, nursed by a mother on a mildly diet-restricted, low fat/low sugar diet (DR) and weaned onto a HF/HS diet (HF/DR/HF) had the highest mammary tumor incidence, while HF/HF/DR had the lowest tumor incidence. Cox proportional hazards analysis showed that a HF/HS postweaning diet doubled mammary cancer risk, and a HF/HS diet during gestation and postweaning increased risk 5.5 times. Exposure to a HF/HS diet during gestation, when combined with a postweaning DR diet, had a protective effect, reducing mammary tumor risk by 86% (HR = 0.142). Serum adipocytokine analysis revealed significant diet-dependent differences in leptin/adiponectin ratio and IGF-1. Flow cytometry analysis of cells isolated from mammary glands from a high tumor incidence group, DR/HF/HF, showed a significant increase in the size of the mammary stem cell compartment compared with a low tumor group, HF/HF/DR. These results indicate that dietary reprogramming induces an expansion of the mammary stem cell compartment during mammary development, increasing likely carcinogen targets and mammary cancer risk. Cancer Prev Res; 10(10); 553–62. ©2017 AACR.
See related editorial by Freedland, p. 551–2.
Introduction
The increase in obesity, especially in young children, has caused concern due to its broad negative impact on health (1). The most recent National Center for Health Statistics report estimates that 69% of the adult U.S. population is overweight and 35.5% of the overweight are obese. Strikingly, 20.5% of adolescents aged 12 to 19 are obese as well. According to the Centers of Disease Control, early obesity places children and adolescents at higher risk for adult health problems, such as cardiovascular disease, diabetes, and certain cancers, including endometrial, colorectal, and postmenopausal breast cancer (2). There is an established link between metabolic syndrome and breast cancer in humans (3).
Concordant with this increase in obesity, there has been a significant increase in the adjusted incidence of breast cancer in white females in recent years (4). Similarly, in populations at increased risk for hyperinsulinemia and obesity, such as in Hispanic women, weight gain is associated with increased risk of breast cancer in pre- and postmenopausal women (5, 6). A recently acquired high body mass index (BMI), as classified by the World Health Organization, in postmenopausal African American (AA) women is associated with an increased risk of estrogen receptor (ER)-positive breast cancer, and a recently-acquired high waist-to-hip ratio, which indicates a larger amount of intra-abdominal fat, is associated with an increased risk of premenopausal ER-positive tumors as well as postmenopausal breast cancer of all subtypes (7). Excess consumption of high energy and sugary fast foods increases the risk for postmenopausal breast cancer in white women as well as premenopausal breast cancer in AA women (8).
In addition to dietary influences in adulthood, there is strong evidence that the in utero environment, determined by the nutritional status of the mother, permanently affects development and metabolism as well as propensity to later disease in their offspring (9). This is known as the fetal origins, or the developmental origins of health and disease hypothesis (9, 10). The term “metabolic programming” has been coined to refer to the early adaptations to nutritional stimuli or milieu that can induce permanent and persistent changes in metabolism and physiology (11, 12).
Although a correlation between high birth weight and breast cancer risk has been established, it is difficult to determine the precise impact of highly variable factors, such as lifestyle, BMI, and diet in human populations. Studies of birth weight illustrate the long-term impact of the in utero environment on development and health. Both low and high birth weights are associated with a greater risk of obesity in the adult (13), and studies in Pima Indian women strongly suggest that the prenatal environment of the offspring of diabetic women results in the development of obesity in childhood and early adulthood (14). Women with type II diabetes are more likely to have larger offspring (15), who have an increased risk of developing breast cancer as an adult (16).
Studies of populations exposed to famine during the Dutch Hunger Winter demonstrate that prenatal malnutrition is associated with differentially methylated regions (PDMR) of genes related to growth and metabolism, as well as highlighting the importance of timing of famine exposure on DNA methylation. Two PDMRs are found to be associated with birth weight and serum LDL cholesterol (17, 18).
Unlike other organs, the mammary gland develops predominantly postnatally. At birth, the mammary gland has only a rudimentary ductal structure. Branching and the formation of terminal end buds (TEB, alveolar precursors) starts at around age 10 to 11 years (Tanner stage II) in humans and at 4 to 5 weeks of age in mice. Because of this protracted development, there are critical windows of susceptibility during which the tissue is highly proliferative and the hierarchical equilibrium between stem, precursor, and differentiated cells can be affected by hormonal, metabolic, or environmental exposures (19).
Rodent studies have also demonstrated the importance of in utero dietary exposures on metabolic programming and mammary cancer susceptibility. The “pup in a cup” model, in which 1- to 4-day-old mouse pups are fed high carbohydrate diets, shows that they have an immediate onset of hyperinsulinemia that persists throughout the suckling period and into adulthood and they are prone to early obesity (20), even when the diets are normalized at weaning (21). Studies have also shown that gestational exposure to obese mothers can “metabolically program” the pups, increasing propensity to early obesity in their offspring (22). Importantly, the progeny of these obese, hyperinsulinemic, and hyperglycemic mothers, who were not fed a high fat (HF) diet themselves, show the same pattern of early obesity and metabolic dysregulation as their mothers that were fed the lard-based HF diet (22). When rats are fed an isocaloric HF (43% calories from corn oil) or a low fat (LF, 12% calories from corn oil) diet during pregnancy, mammary glands from HF fed offspring show increased epithelial density and an increased number of TEBs compared with LF -fed rats (23).
On the basis of these reports, the goal of the current study was to elucidate the effects of diet-induced metabolic programming during gestational, lactational, and post-weaning exposures on susceptibility to mammary carcinogenesis, and to investigate possible mechanisms by which dietary manipulation increases mammary cancer risk.
Materials and Methods
Mice and diets
All experimental procedures were approved by the Institutional Animal Care and Use Committee. As C57BL/6 mice are highly resistant to mammary carcinogenesis, outbred SENCAR mice were chosen for this study (24). Breeder mice were purchased from the NCI (Frederick, MD) and maintained at our AAALAC-accredited facility under temperature- and light-controlled conditions (24°C and 12-hour light/dark cycle). Defined diets were purchased from Research Diets, Inc., cat#D01060501 was the chow-like, low fat, low sugar control diet, and cat#D04011601 the high fat, high sugar (HF/HS) “Western” diet (Table 1). Upon arrival at 4 weeks of age, SENCAR mice were randomized into two groups and fed either the defined control or HF/HS diets. As these mice were sedentary, animals fed the control diet ad libitum became obese by ≥20 weeks of age. To maintain a normal body weight and provide a nonobese study control, we imposed a mild 12% diet restriction on the control diet group (DR), which would be similar to portion control in sedentary humans to prevent excessive weight gain. DR animals received a daily allotment equal to 88% of the mean daily consumption of the AL group. This mild portion control did not result in an insufficiency of macro- or micronutrients. During lactation, DR mothers were fed the control diet ad libitum to ensure proper nutrition of mother and pups. HF/HS animals were also provided with 10% fructose in their drinking water starting 2 weeks after arrival, simulating the effect of frequent consumption of high-fructose corn syrup sweetened beverages in humans.
Composition of the experimental diets
. | Control, no sucrose . | High fat-high sucrose . | ||
---|---|---|---|---|
. | 10% of calories from fat . | 45% of calories from fat . | ||
Ingredients . | Grams . | % of calories . | Grams . | % of calories . |
Casein | 200 | 20% | 200 | 20% |
L-Cystine | 3 | 3 | ||
Cornstarch | 575 | 57% | 0 | |
Maltodextrin | 125 | 12% | 100 | 10% |
Sucrose | 0 | — | 245.6 | 24% |
Cellulose | 50 | — | 50 | — |
Soybean oil | 25 | 6% | 25 | 6% |
Lard | 20 | 4% | 177.5 | 39% |
Minerals | 45 | 45 | ||
Vitamin mix | 10 | 10 | ||
Choline bitartarate | 2 | 2 | ||
4,057 kcal/1,055 g | 4,057 kcal/858.1 g |
. | Control, no sucrose . | High fat-high sucrose . | ||
---|---|---|---|---|
. | 10% of calories from fat . | 45% of calories from fat . | ||
Ingredients . | Grams . | % of calories . | Grams . | % of calories . |
Casein | 200 | 20% | 200 | 20% |
L-Cystine | 3 | 3 | ||
Cornstarch | 575 | 57% | 0 | |
Maltodextrin | 125 | 12% | 100 | 10% |
Sucrose | 0 | — | 245.6 | 24% |
Cellulose | 50 | — | 50 | — |
Soybean oil | 25 | 6% | 25 | 6% |
Lard | 20 | 4% | 177.5 | 39% |
Minerals | 45 | 45 | ||
Vitamin mix | 10 | 10 | ||
Choline bitartarate | 2 | 2 | ||
4,057 kcal/1,055 g | 4,057 kcal/858.1 g |
HF/HS fed animals became glucose intolerant and insulin resistant after 10 weeks on diet compared with DR controls, as determined by glucose and insulin tolerance tests (Supplementary Fig. S1A and S1B). At 15 weeks of age, normal and hyperinsulinemic/hyperglycemic mice were mated to normal SENCAR males. Male pups of the resulting litters were culled immediately, and female pups were fostered within 24 hours of birth to generate the lactation exposure groups (Fig. 1). At weaning, pups were randomized into the different postwean diet exposure groups for a total of 8 dietary regimens (Table 2) and further randomized into separate groups for GTT, ITT, DMBA treatment, or stem cell analysis. Weekly body weights were recorded for all animals.
Experimental diet groups
Group . | Gestation diet . | Nursing diet . | Postwean diet . |
---|---|---|---|
A | DR | DR | DR |
B | DR | DR | HF/HS |
C | DR | HF/HS | DR |
D | DR | HF/HS | HF/HS |
E | HF/HS | HF/HS | HF/HS |
F | HF/HS | HF/HS | DR |
G | HF/HS | DR | HF/HS |
H | HF/HS | DR | DR |
Group . | Gestation diet . | Nursing diet . | Postwean diet . |
---|---|---|---|
A | DR | DR | DR |
B | DR | DR | HF/HS |
C | DR | HF/HS | DR |
D | DR | HF/HS | HF/HS |
E | HF/HS | HF/HS | HF/HS |
F | HF/HS | HF/HS | DR |
G | HF/HS | DR | HF/HS |
H | HF/HS | DR | DR |
Glucose and insulin tolerance tests
Glucose tolerance tests (GTT) and insulin tolerance tests (ITT) were performed at 10 to 12 weeks of age (n = 7–13/group). Animals were fasted for 6 hours starting at 6:00 am; then, baseline blood glucose levels were taken by drawing blood from the submandibular vein with MEDIPoint Goldenrod animal lancets (MEDIpoint) and measuring blood glucose with a Bayer Breeze2 glucometer (Bayer Corp.). Following intraperitoneal injection of 2 g/kg glucose (GTT) or 0.75 U/kg insulin (ITT), blood glucose levels were measured at 15, 30, 60, and 120 minutes postinjection. GTT and ITT animals were sacrificed at 5 months of age by CO2 asphyxiation. Blood was collected at sacrifice via cardiac puncture, centrifuged, serum collected, and frozen in aliquots at −70°C. Carcasses were scanned with a Lunar PIXImus II Densitometer, and tissues were collected for histologic and molecular analyses.
Mammary carcinogenesis
Starting at 7 to 9 weeks of age, mice in the carcinogenesis study were administered daily 20 μg doses of 7,12-dimethylbenz[a]anthracene (DMBA) by oral gavage, 5 days per week for 6 weeks as described previously (24), n = 17–33 per group. Animals were weighed weekly and monitored daily for health status and tumors. Upon tumor detection, animals were sacrificed by CO2 asphyxiation, blood was collected at sacrifice via cardiac puncture, and carcasses were scanned with a Lunar PIXImus II Densitometer. Tissues and mammary tumors were collected for histology and molecular analyses. In the absence of palpable tumors, animals were allowed to continue on the same diet until they reached 1 year of age, and then, blood and refractory tissues were harvested and assessed for microscopic mammary lesions.
Serum assays
Serum from the 5-month-old animals that were previously evaluated with GTT and ITT tests were analyzed, n = 19 for groups F and H combined, n = 45 for combined groups B, D, and G. MILLIPLEX mouse serum adipokine and adiponectin kits (Millipore Corp.) were purchased and analyzed on a Bio-Plex 200 (Bio-Rad Laboratories, Inc.) system. Calibration was performed before each run with a BioPlex Calibration Kit, and data were acquired using BioPlex Manager software version 4.1. Serum IGF-1 was analyzed using the Quantikine Mouse/Rat IGF-1 ELISA (R&D Systems) at an optical density of 540 nm.
Mammary epithelial cell isolation
For primary cell cultures, abdominal and inguinal mammary glands were harvested from 5-week-old group F or D mice (n = 6 and n = 5, respectively) and were dissociated for 12 hours at 37°C in DMEM/F12 medium, 2% FBS, and a 1:10 dilution of gentle collagenase/hyaluronidase (StemCell Technologies Inc.). After treatment with 0.8% NH4Cl:HBSS:2% FBS (4:1, v/v) to lyse erythrocytes, single cells were obtained by gentle pipetting for 1 minute in prewarmed (37°C) 0.25% trypsin, and then for 2 minutes in 5 mg/mL prewarmed (37°C) dispase plus 10 U DNase I, followed by filtration through a 40-μm mesh cell strainer.
Flow cytometry and cell sorting
Dissociated single mammary epithelial cells were incubated with the following conjugated antibodies at the concentrations shown: anti-CD31-FITC (clone 390; BD Pharmingen) 1:400, anti-CD45-FITC (clone 30-F11; BD Pharmingen) 1:400, anti-TER-119-FITC (clone TER-119; BD Pharmingen) 1:400, anti-CD24-Pacific Blue (clone M1/69; BD Pharmingen) 1:200, anti-CD29-PE-Cy7 (clone HMb1-1; BioLegend) 1:400, anti-Sca1-APC-Cy7 (clone D7; BioLegend) 1:200, and anti-CD49f-BV650 [clone eBioGoH3 (GoH3); eBioscience] 1:200.
Isolated mammary cells (0.5–2.0 × 106) from each group were diluted in 500 μL HBSS and incubated for 30 minutes on ice in the dark. CompBeads (BD Biosciences) were used as the single color and unstained controls. After washing twice, cells were resuspended in 500 μL HBSS and 100,000 labeled cells per sample were analyzed on a FACSAria II flow cytometer (Becton Dickinson). Apoptotic cells were excluded using Pacific Green succinimidyl ester (Life Technologies). Sorted cell populations were routinely reanalyzed and found to be 94% to 98% pure. Data were analyzed using FlowJo software (Tree Star, Inc.). Data from 5 to 6 biological replicates are shown.
Statistical analysis
Body weight at 20 weeks was examined by comparing all diet groups using a two-sample unequal variance t test. Significant differences coincided with a naïve visual inspection of the data. GTT and ITT response between diet regimes was tested using trapezoidal AUC measure as described previously (25) and then compared using a Wilcoxon rank sum test. Survival analysis was carried out using the Kaplan–Meier statistic. All diet groups were compared using the Mantel–Haenszel test, and final tumor incidence proportions were compared using Pearson χ2 test statistic. The influence of diet regime at the different developmental periods on body weights, trapezoidal AUC measures for GTT and ITT, and survival was examined using linear model frameworks. Final models were identified with backwards stepwise selection using Bayesian information criterion. The effect of diet regime on survival outcomes was examined using the Cox proportional hazards model, while the other outcomes were analyzed using multiple linear regression comparing all groups to group A (DR/DR/DR). These analyses were done in the R programming language using the standard coxph, lm, and step functions with default values. Cytokine data were evaluated by combining values of the high tumor incidence and low tumor incidence groups and compared using a two-sample unequal variance t test with GraphPad Prism software.
Results
Effect of diet regime on body weight, composition, and glucose metabolism
Weekly body weights segregated into three clusters (P < 0.0001), as seen in Fig. 2A. Groups D (DR/HF/HF) and E (HF/HF/HF) were the heaviest, while the groups with postweaning DR diet F (HF/HF/DR), C (DR/HF/DR), H (HF/DR/DR), and A (DR/DR/DR) had the lowest body weights (P < 0.0001). Groups G (HF/DR/HF) and B (DR/DR/HF) included DR during gestation and/or lactation and therefore had intermediate weights. When assessing dietary influences with statistical linear model frameworks, it showed that a combination of HF/HS during lactation and postweaning had the strongest effect, and _/HF/HF animals exhibited the heaviest body weights (Table 3).
Body weights, glucose, and insulin tolerance tests. Two bracketed groups are significantly different from each other if they do not share a letter. A, Body weights up to 20 weeks, a ≠ b ≠ c. B, GTT results, a ≠ b ≠ c ≠ d ≠ e. C, ITT results, a ≠ b. D, Fasting blood glucose baseline levels of all animals combined by postwean diet, P ≤ 0.0001.
Body weights, glucose, and insulin tolerance tests. Two bracketed groups are significantly different from each other if they do not share a letter. A, Body weights up to 20 weeks, a ≠ b ≠ c. B, GTT results, a ≠ b ≠ c ≠ d ≠ e. C, ITT results, a ≠ b. D, Fasting blood glucose baseline levels of all animals combined by postwean diet, P ≤ 0.0001.
Multivariate regression analysis
Body weights . | Additive effect . | SE . | P . | . |
---|---|---|---|---|
Intercept | 37.385 | 1.082 | <2e–16a | |
PostHF | +21.855 | 1.101 | <2e–16a | |
LactHF | +7.231 | 1.071 | 7.95E–11a | |
GTT | ||||
Intercept | 1,809.3 | 121.1 | <2e–16a | |
HF:lact | +919.6 | 190.8 | 6.70E–06a | |
HF:post | +1,584.3 | 138.8 | <2e–16a | |
GestHF*lactHF | −922.2 | 198.8 | 1.36E–05a | |
ITT | ||||
Intercept | 570.64 | 80.81 | 7.11E–10a | |
HF:post | +1,279.8 | 113.54 | <2e–16a | |
Survival | HR | Coefficient | SE | P |
HF:gest | 0.142 | −1.954 | 0.756 | 0.0097a |
HF:post | 2.026 | 0.706 | 0.332 | 0.034b |
HF:gest*HF:post | 5.499 | 1.705 | 0.82 | 0.038b |
Body weights . | Additive effect . | SE . | P . | . |
---|---|---|---|---|
Intercept | 37.385 | 1.082 | <2e–16a | |
PostHF | +21.855 | 1.101 | <2e–16a | |
LactHF | +7.231 | 1.071 | 7.95E–11a | |
GTT | ||||
Intercept | 1,809.3 | 121.1 | <2e–16a | |
HF:lact | +919.6 | 190.8 | 6.70E–06a | |
HF:post | +1,584.3 | 138.8 | <2e–16a | |
GestHF*lactHF | −922.2 | 198.8 | 1.36E–05a | |
ITT | ||||
Intercept | 570.64 | 80.81 | 7.11E–10a | |
HF:post | +1,279.8 | 113.54 | <2e–16a | |
Survival | HR | Coefficient | SE | P |
HF:gest | 0.142 | −1.954 | 0.756 | 0.0097a |
HF:post | 2.026 | 0.706 | 0.332 | 0.034b |
HF:gest*HF:post | 5.499 | 1.705 | 0.82 | 0.038b |
NOTE: Linear model frameworks and Cox proportional hazard models assessed prognostic factors of diet regimes on outcomes of body weight, GTT, or ITT results, and mammary tumorigenesis. All groups were compared with group A (DR/DR/DR).
Abbreviations: GTT, glucose tolerance test, ITT insulin tolerance test.
aStatistical significance with P ≤ 0.01. Intercept is the average outcome for the baseline group A (DR/DR/DR), Coef is the relative change compared with control group A.
bStatistical significance with P ≤ 0.05.
DEXA analysis of carcasses showed significant increases in bone mineral content and density and highly significant increases in lean mass, fat mass, and percent body fat in the combined HF/HS postwean fed groups B, D, and G, compared with the combined postwean DR-fed groups F and H (Supplementary Fig. S2A–S2F).
The GTT results in Fig. 2B showed that animals from group D, which were born to DR mothers, nursed by HF/HS–fed mothers and weaned onto HF/HS diets (DR/HF/HF), were significantly more glucose intolerant than all the other groups (P < 0.001). Groups A, F, and H, with the lowest average body weights, had normal glucose tolerance. They rapidly cleared glucose from the bloodstream, and levels returned to baseline within 120 minutes following glucose injection. Groups B and E were significantly less glucose tolerant than groups A, F, and H, indicating that the HF/HS postwean diet had a strong effect on glucose clearance (P < 0.001). Interestingly, animals in groups E and D with similar body weights (Fig. 2A) had different glucose metabolism, with group E being less glucose intolerant than group D (Fig. 2B). Linear model frameworks analysis showed that a HF/HS diet during lactation, _/HF/_, significantly decreased glucose tolerance, and a postwean HF/HS diet had the greatest effect, as indicated by the fact that _/_/HF animals were much more glucose intolerant than _/_/DR animals (Table 3).
The dietary groups segregated into two main clusters in terms of insulin sensitivity (Fig. 2C). Groups F, H, A, and C all received a DR diet postweaning and displayed a normal response to insulin, demonstrated by a drop in blood sugar levels immediately after insulin administration, and recovery within 2 hours. Conversely, blood glucose levels did not change in response to insulin administration in groups B, E, G, and that received HF/HS diet postweaning, indicating that these groups with the heaviest body weights (Fig. 2A) were also profoundly insulin resistant (Fig. 2C, P < 0.000). The _/_/DR groups F, A, H, and C were the lightest adults and also showed a normal response to insulin. Neither gestation nor lactation diets had a measurable effect on insulin response, either alone or in combination; therefore, the postwean HF/HS diet alone was the main contributor to insulin resistance in the adult (Table 3).
Effect of diet regime on susceptibility to mammary carcinogenesis
A separate group of diet-exposed animals were treated with low doses of DMBA over a 6-week period, which has been shown to give the greatest ratio of mammary tumors to lymphomas (24). The Kaplan–Meier survival curve (Fig. 3A) revealed significant differences in susceptibility to mammary carcinogenesis among diet-exposed groups (Table 4). Groups broadly fell into three clusters: groups B, D, and G all received HF/HS postwean diets and exhibited the lowest survival time. Groups F and H had the highest survival times, developing very few mammary tumors with increased latency, many of which were found as latent lesions upon termination of the study at one year of age. Surprisingly, groups A and E, which received consistent diets throughout, had similar intermediate survival times, suggesting that the diet switch during pregnancy and lactation strongly impacted susceptibility to tumorigenesis. While both groups D and E became morbidly obese and had the highest body weights of all the groups (Fig. 2A), group D showed a trend to higher mammary tumor incidence and lower survival time than group E (P = 0.09).
Tumor incidence data paralleled the Kaplan–Meier survival curve. The high tumor incidence groups B, D, and G were significantly different from the low tumor groups F and H (Table 5A). Intermediate tumor incidence groups A and E were different from high tumor groups B and G (Table 5B), but not different from low tumor groups F and H. Group C fell in-between the intermediate and high tumor incidence groups, suggesting a modest effect of lactation diet on tumor outcome. Body weight did not correlate well with tumor incidence in this model (Fig. 3B, r2 = 0.342), and there was also little correlation between tumor incidence and the results from the GTT (Fig. 3C, r2 = 0.017) and ITT tests (Fig. 3D, r2 = 0.124).
Mammary carcinogenesis results. A, Kaplan–Meier survival curve. Two bracketed groups are significantly different from each other if they do not share a letter. a ≠ b ≠ c ≠ d ≠ e ≠ f ≠ g. Statistical differences are shown in Table 4. B, No correlation was seen between the average body weights and tumor incidence (r2 = 0.342, P = 0.128). C, No correlation was seen between GTT results and tumor incidence (r2 = 0.017, P = 0.755). D, No correlation was seen between ITT results and tumor incidence (r2 = 0.124, P = 0.392).
Mammary carcinogenesis results. A, Kaplan–Meier survival curve. Two bracketed groups are significantly different from each other if they do not share a letter. a ≠ b ≠ c ≠ d ≠ e ≠ f ≠ g. Statistical differences are shown in Table 4. B, No correlation was seen between the average body weights and tumor incidence (r2 = 0.342, P = 0.128). C, No correlation was seen between GTT results and tumor incidence (r2 = 0.017, P = 0.755). D, No correlation was seen between ITT results and tumor incidence (r2 = 0.124, P = 0.392).
Kaplan–Meier survival curve significant P values of Fig. 3A
Groups compared . | P . |
---|---|
A vs. B | 0.034 |
A vs. D | 0.024 |
A vs. F | 0.054 |
A vs. G | 0.003 |
B vs. F | >0.001 |
B vs. H | 0.003 |
C vs. F | 0.007 |
C vs. H | 0.021 |
D vs. F | >0.001 |
D vs. H | 0.002 |
E vs. G | 0.018 |
F vs. G | >0.001 |
G vs. H | >0.001 |
Groups compared . | P . |
---|---|
A vs. B | 0.034 |
A vs. D | 0.024 |
A vs. F | 0.054 |
A vs. G | 0.003 |
B vs. F | >0.001 |
B vs. H | 0.003 |
C vs. F | 0.007 |
C vs. H | 0.021 |
D vs. F | >0.001 |
D vs. H | 0.002 |
E vs. G | 0.018 |
F vs. G | >0.001 |
G vs. H | >0.001 |
The first tumors were found in the high tumor incidence groups B, D, and G, with tumors detected as early as 58, 82, and 93 days post-dosing, respectively. In the low tumor incidence groups F and H, tumor latency was increased with the first tumors arising 145 and 156 days post-dosing, respectively. Median tumor latency in groups F and H was 282 and 266 days, whereas median latency of the high incidence groups B, D, and G was 148, 170, and 159 days, respectively. Unlike the high tumor groups, late tumors in groups F and H were found as occult latent lesions after the study was terminated, and the animals had reached 1 year of age (275–285 days post dosing). Groups A, C, and E, whose tumor incidence was intermediate, also had intermediate median latencies compared with the high tumor and low tumor incidence groups.
In order to assess the impact of diet regimen on mammary tumor prognosis, multivariate analysis using the Cox proportional hazards model was performed. When compared with control group A, which was exposed to a low fat, low sugar diet regime throughout, hazard ratios (HR) revealed that a HF/HS postweaning diet (_/_/HF), as seen in the high tumor incidence groups B, D, and G, had a significant negative effect on survival, doubling mammary cancer risk (HR = 2.026, Table 3). Somewhat unexpectedly, exposure to a HF/HS gestation diet (HF/_/_), as in the low tumor incidence groups F and H, had a significant protective effect, reducing mammary tumor risk by 86% when compared with group A (HR = 0.142), which did not experience a diet switch. (Table 3). Conversely, a HF/HS diet during gestation as well as postweaning (HF/_/HF) increased risk 5.5 times (HR = 5.499), as seen in group G, showing that a HF/HS postweaning diet abrogated the protective effect that the gestational HF/HS diet provided.
Mammary tumor incidence
Group . | n . | Tumors . | % Incidence . |
---|---|---|---|
A | 28 | 8 | 29% |
B | 33 | 19 | 58% |
C | 17 | 8 | 47% |
D | 17 | 11 | 65% |
E | 19 | 6 | 32% |
F | 23 | 5 | 22% |
G | 17 | 12 | 71% |
H | 17 | 3 | 18% |
Group . | n . | Tumors . | % Incidence . |
---|---|---|---|
A | 28 | 8 | 29% |
B | 33 | 19 | 58% |
C | 17 | 8 | 47% |
D | 17 | 11 | 65% |
E | 19 | 6 | 32% |
F | 23 | 5 | 22% |
G | 17 | 12 | 71% |
H | 17 | 3 | 18% |
Statistical comparisons of mammary tumor incidence. Significant P values were compared using the Pearson's χ2 test
% Incidence compared . | P . |
---|---|
A vs. B | 0.038 |
A vs. G | 0.020 |
B vs. F | >0.001 |
B vs. H | 0.002 |
C vs. F | 0.013 |
C vs. H | 0.043 |
D vs. F | 0.007 |
D vs. H | 0.026 |
E vs. G | 0.048 |
F vs. G | >0.001 |
G vs. H | 0.001 |
% Incidence compared . | P . |
---|---|
A vs. B | 0.038 |
A vs. G | 0.020 |
B vs. F | >0.001 |
B vs. H | 0.002 |
C vs. F | 0.013 |
C vs. H | 0.043 |
D vs. F | 0.007 |
D vs. H | 0.026 |
E vs. G | 0.048 |
F vs. G | >0.001 |
G vs. H | 0.001 |
Effect of diet regimes on serum cytokines
Some studies point to a role of adipokines as novel risk factors as well as potential diagnostic and prognostic biomarkers in breast cancer (26, 27). Therefore, we assessed the effects of the dietary exposures on leptin, adiponectin, and IGF-1 in the 5-month-old, untreated animals. To better assess this complex dataset, we combined results from groups with the lowest mammary tumor incidence, F and H, and compared them with the combined groups with highest mammary tumor incidence, B, D, and G.
Serum leptin levels were increased (P = 0.059), and adiponectin levels significantly decreased (P < 0.0001) in the high tumor incidence groups (Fig. 4A and B). Therefore, the leptin/adiponectin ratio was also significantly higher in the high tumor incidence groups (Fig. 4C). A high leptin/adiponectin ratio is an indicator for insulin resistance, and we observed the animals of groups B, D, and G to be unresponsive to insulin (Fig. 2C).
Cytokine data of sera of 5-month-old animals. Groups F and H n = 19, groups B, D, and G n = 45. A, Serum leptin levels, P = 0.0591. B, Serum adiponectin, P < 0.0001. C, Leptin/adiponectin ratio, P = 0.011. D, Serum IGF-1, P < 0.0001.
Cytokine data of sera of 5-month-old animals. Groups F and H n = 19, groups B, D, and G n = 45. A, Serum leptin levels, P = 0.0591. B, Serum adiponectin, P < 0.0001. C, Leptin/adiponectin ratio, P = 0.011. D, Serum IGF-1, P < 0.0001.
Serum IGF-1 levels were also significantly higher in the groups with high tumor incidence, B, D, and G compared with the groups with low tumor incidence F and H. (Fig. 4D, P < 0.0001), suggesting a link between exposures to a HF/HS diet and increased free IGF-1 levels.
Effect of diet regimes on the stem cell compartment
Reports indicate that offspring of rat dams fed a HF diet have mammary glands with increased epithelial density, increased numbers of TEBs, and increased tumor incidence compared with low fat (LF)-fed offspring (23), and we have previously reported that mammary glands of the BK5.IGF-1 transgenic mouse model show an increase in the number and size of TEBs, as well as a significantly increased susceptibility to mammary carcinogenesis (28). In addition, recent investigations revealed a link between high levels of tissue IGF-1 and the size of the mammary stem cell (MaSC) compartment (ref. 29, Luo L., and colleagues, unpublished observations). Therefore, we evaluated the size of the MaSC compartment in prepubertal mammary glands of a high tumor and a low tumor incidence group. Results from FACS (Fig. 5A) revealed that prepubertal mice of the high tumor incidence group D had a significantly increased number of MaSC/progenitor cells compared with the low tumor group F (Fig. 5B, P < 0.001), suggesting that gestation and postwean diet exposures have a marked effect on MaSC proliferation. Emerging studies have long identified quiescent stem cells and early progenitors as important carcinogen targets that accumulate mutations, leading to oncogenic transformation (30).
Representative FACS plots of prepubertal mammary gland cells from low tumor group F (A and C) and high tumor group D (B and D) animals. n = 6 (F) and n = 5 (D) per group. The MaSC-enriched population was identified on the basis of the following validated stem cell markers: CD31−/CD45−/TER119−/CD24+/CD29hi/CD49f+/Sca1−. A and B, The CD24+/CD29hi population (arrow) was gated after excluding debris, cell aggregates, and lineage-positive cells (Supplementary Fig. S3) in each group. Lu., luminal cells; Ba, basal cells; Str, stromal cells. C and D, The CD24+/CD29hi population from each group was then further gated into the MaSC-enriched population of CD49f+/Sca1− cells. Arrows, MaSC-enriched population. E, Frequency of MaSCs in the enriched population over total mammary epithelial cells (MEC) between low tumor group F and high tumor group D. MECs include the total number of luminal, basal, and CD24+/CD29hi cells. Unpaired two-sided equal variance t test was used for statistical analysis. Data are presented as mean ± SEM. ***, P < 0.001.
Representative FACS plots of prepubertal mammary gland cells from low tumor group F (A and C) and high tumor group D (B and D) animals. n = 6 (F) and n = 5 (D) per group. The MaSC-enriched population was identified on the basis of the following validated stem cell markers: CD31−/CD45−/TER119−/CD24+/CD29hi/CD49f+/Sca1−. A and B, The CD24+/CD29hi population (arrow) was gated after excluding debris, cell aggregates, and lineage-positive cells (Supplementary Fig. S3) in each group. Lu., luminal cells; Ba, basal cells; Str, stromal cells. C and D, The CD24+/CD29hi population from each group was then further gated into the MaSC-enriched population of CD49f+/Sca1− cells. Arrows, MaSC-enriched population. E, Frequency of MaSCs in the enriched population over total mammary epithelial cells (MEC) between low tumor group F and high tumor group D. MECs include the total number of luminal, basal, and CD24+/CD29hi cells. Unpaired two-sided equal variance t test was used for statistical analysis. Data are presented as mean ± SEM. ***, P < 0.001.
Discussion
Our study explored the effects of dietary manipulation during gestation, lactation, and postweaning periods on body weight, glucose metabolism, susceptibility to mammary carcinogenesis, and the size of the MaSC compartment. One novel finding in our study was that DR does not have to be severe [12% vs. the usual 30%–40% calorie restriction (CR); ref. 31], to have a lasting effect on glucose metabolism and mammary tumor risk. Analyses of metabolic parameters showed that animals on a _/HF/HF diet regime had the heaviest body weights of all the groups, and a HF/HS postweaning diet had the strongest effect on metabolic parameters, causing all animals became glucose intolerant and insulin resistant by 12 weeks of age.
We evaluated serum leptin levels in these animals at 5 months of age. Serum leptin levels were elevated in the high tumor incidence groups, and adiponectin was significantly decreased, resulting in a significantly higher leptin/adiponectin ratio in the high tumor incidence groups. Leptin, which is produced by adipose tissue, exerts its effect on energy balance by activating specific centers in the hypothalamus to decrease food intake, increase energy expenditure, and influence glucose and fat metabolism. As body fat storages increase, circulating leptin concentrations increase, whereas adiponectin levels decrease. It has been reported that higher prediagnostic leptin or leptin/adiponectin ratios are linked to a greater risk of postmenopausal breast cancer (27, 32), and that exposure to excess leptin during pregnancy increases mammary tumorigenesis in parous rats (33).
Together with high leptin/adiponectin ratios, we found that serum IGF-1 levels were significantly elevated in the high tumor groups. IGF-1 stimulates TEB formation and ductal elongation in the mammary gland (34), and IGF-1 transgenic mice have an increased mammary tumor incidence as well as a higher susceptibility to mammary tumorigenesis (28). It has been shown that the addition of leptin increases liver and muscle tissue IGF-1 levels in aged CR mice (35), showing a possible link between leptin and IGF-1. These studies suggest that in our model, a HF/HS diet increased leptin levels, and leptin increased IGF-1 levels, possibly through STAT5 signaling (36). Recently, it has been demonstrated that levels of leptin in human adipose explant-derived conditioned media positively correlated with the size of the normal breast stem cell pool, and a strong linear correlation was found with an increasing leptin/adiponectin ratio and increased breast stem cell pool (37). It has been proposed that an increase in the amount of tissue stem cells due to dysregulation of the self-renewal process of the stem cell compartment may be a key event preceding early tumorigenesis (38).
Mammary tumorigenesis data analysis revealed that while having a HF/HS postwean diet increased mammary tumor risk (HR = 2.026), a HF/HS diet during gestation significantly reduced risk by 86% (HR = 0.142) compared with the control group A. However, when a gestational HF/HS diet was combined with a postweaning HF/HS diet, mammary tumor risk increased 5.5 times, therefore effectively abolishing the protective effect of the gestational HF/HS diet. The strongest protective effect was seen when pregnant mothers consumed a HF/HS diet during gestation and offspring receiving the DR diet postweaning. This supports previous studies showing the benefits of CR on mammary tumorigenesis (39); however, groups A and E had a consistent throughout but opposite metabolic profiles, and their tumor outcomes were still similar. A study by Andrade and colleagues demonstrates that rats fed lard-based HF diet during gestation have significantly reduced tumor incidence and multiplicity (40), and their follow-up study shows that mammary tissues from these protected animals exhibit permanent changes in their fatty acid profiles and transcriptional networks (41). Therefore, this study brings additional insights into the impact of switching diets during early development.
Although extrapolation of these results to humans should be made cautiously, our data indicate that a HF/HS diet during pregnancy followed by mild portion control during childhood may provide protection, while DR during gestation, followed by a HF/HS diet and childhood obesity, may contribute to increased risk for mammary cancer. Similar trends have been found in a Dutch Famine Birth Cohort study that shows that maternal undernutrition increases breast cancer risk of the female progeny (42).
Mammary glands from the high tumor incidence group D (DR/HF/HF) and the low tumor risk group F (HF/HF/DR) were assessed for the size of the MaSC compartment. Group D had a significantly larger MaSC compartment than group F. This points to an increase in the number of symmetric stem cell divisions in developing virgin mammary glands when exposed to DR in utero, then followed by a HF/HS diet postweaning. This early diet switch could create a chain of events, such as increased leptin levels increasing IGF-1 through STAT5 signaling, followed by elevated IGF-1 increasing symmetric stem cell divisions in mammary tissue, therefore producing more stem/progenitor cells. Through this mechanism, our high tumor risk group could have acquired, through diet, a larger pool of MaSCs, increasing potential targets for cancer initiation and transformation into “cancer stem cells.”
The cancer stem cell hypothesis suggests that malignancies associated with cancer originate from a small population of stem-like, tumor-initiating cells. The ability of stem cells to self-renew and their pluripotency gives them long life spans relative to mature, differentiated cells, and it has been hypothesized that because of this long life span, they are more likely to undergo the multiple mutations necessary for tumor formation. The stem cell markers ALDH1 and NOTCH1 gene expression levels in the MMTV-Wnt-1 mouse model are significantly increased in doxorubicin-treated tumors of animals fed a HF diet compared with controls, indicating that metabolic status plays a role in breast cancer stem cell expansion (43). We also know that numerous, but not all, malignant tumor cell lines have highly elevated levels of aerobic glycolysis, up to 200 times higher than their normal tissues of origin. This phenomenon, called the Warburg effect, attributes the glycolytic changes to possible mitochondrial defects (44). Stem cells contain immature mitochondria characterized by perinuclear localization, poorly developed cristae, as well as an electron-lucid matrix compared with differentiated cells (45). Therefore, stem cell glucose metabolism is shifted to glycolysis and lactate production rather than oxidative phosphorylation (46). It is possible that, if a stem cell transforms over time, it retains its immature mitochondria, and with its metabolism shifted toward glycolysis and lactate production, it generates an acidic, protumorigenic microenvironment, therefore accelerating its transformation into a cancer stem cell.
In conclusion, exposure to the DR/HF/HF diet regimen significantly increased the MaSC compartment in prepubertal mammary glands compared with glands from HF/HF/DR animals. Therefore, a diet switch after gestation caused an increase in the number of MaSCs that could later become possible carcinogen targets, transforming some into cancer stem cells.
Disclosure of Potential Conflicts of Interest
No potential conflicts of interest were disclosed.
Authors' Contributions
Conception and design: L. Luo, T.R. Berton, S.D. Hursting, C.J. Conti, R. Fuchs-Young
Development of methodology: I.U. Lambertz, T.R. Berton, R. Fuchs-Young
Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): I.U. Lambertz, L. Luo, T.R. Berton, S.D. Hursting, C.J. Conti
Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): I.U. Lambertz, L. Luo, T.R. Berton, S.L. Schwartz, C.J. Conti, R. Fuchs-Young
Writing, review, and/or revision of the manuscript: I.U. Lambertz, T.R. Berton, S.L. Schwartz, S.D. Hursting, C.J. Conti, R. Fuchs-Young
Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): I.U. Lambertz, R. Fuchs-Young
Study supervision: R. Fuchs-Young
Grant Support
This study was supported by the Department of Defense, grant number W81XWH-08-1-0452.
The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.