Background:

Advanced glycation end-products (AGE) are formed through nonenzymatic glycation of free amino groups in proteins or lipid. They are associated with inflammation and oxidative stress, and their accumulation in the body is implicated in chronic disease morbidity and mortality. We examined the association between postdiagnosis dietary Nϵ-carboxymethyl-lysine (CML)–AGE intake and mortality among women diagnosed with breast cancer.

Methods:

Postmenopausal women aged 50 to 79 years were enrolled in the Women's Health Initiative (WHI) between 1993 and 1998 and followed up until death or censoring through March 2018. We included 2,023 women diagnosed with first primary invasive breast cancer during follow-up who completed a food frequency questionnaire (FFQ) after diagnosis. Cox proportional hazards (PH) regression models estimated adjusted hazard ratios (HR) and 95% confidence intervals (CI) of association between tertiles of postdiagnosis CML-AGE intake and mortality risk from all causes, breast cancer, and cardiovascular disease.

Results:

After a median 15.1 years of follow-up, 630 deaths from all causes were reported (193 were breast cancer–related, and 129 were cardiovascular disease–related). Postdiagnosis CML-AGE intake was associated with all-cause (HRT3vsT1, 1.37; 95% CI, 1.09–1.74), breast cancer (HRT3vsT1, 1.49; 95% CI, 0.98–2.24), and cardiovascular disease (HRT3vsT1, 1.91; 95% CI, 1.09–3.32) mortality.

Conclusions:

Higher intake of AGEs was associated with higher risk of major causes of mortality among postmenopausal women diagnosed with breast cancer.

Impact:

Our findings suggest that dietary AGEs may contribute to the risk of mortality after breast cancer diagnosis. Further prospective studies examining dietary AGEs in breast cancer outcomes and intervention studies targeting dietary AGE reduction are needed to confirm our findings.

Advanced glycation end-products (AGE) are compounds that form naturally in the body but can also be consumed in the diet, largely as a result of high-heat cooking of select foods (1). Specifically, AGEs are formed through irreversible nonenzymatic reactions of reducing sugars with proteins or lipids (1). High AGE levels are seen in hyperglycemic conditions and are linked to vascular complications in type II diabetes and the ageing process (2). Further accumulation of circulating AGEs occur through dietary intake of AGE-rich foods. A common mechanistic consequence of AGE accumulation is the activation of the transmembrane receptor for AGE (RAGE). AGE activation of the RAGE signaling cascade leads to persistent increases in oxidative stress and chronic inflammation. This promotes a tissue microenvironment conducive to the onset of chronic diseases such as cardiovascular disease and cancer, as well their associated complications and comorbidities (3–6). In addition, factors such as unhealthy diet, smoking, sedentary lifestyle, and obesity are positive risk factors for breast cancer and cardiovascular disease morbidity and mortality (7–9) and also are positively associated with AGEs (10, 11). Compared with healthy women, higher serum AGE levels have been observed in women with breast cancer and cardiovascular disease, and RAGE is markedly overexpressed in cancerous as compared with noncancerous breast tissue (12–14).

In breast cancer survivors, poor cardiovascular health and further aggravation of preexisting disease may negatively impact survival outcomes (15, 16). AGEs also have been suggested to inhibit the effects of tamoxifen treatment in hormone receptor–positive breast cancer (12). In addition to the naturally produced AGEs in the body, we hypothesize that further AGE accumulation from dietary exposure may represent a mechanism driving both cancer and cardiovascular disease and may be associated with their comorbidity in cancer survivors thus reducing survival outcomes for women with breast cancer. No study to date has examined dietary AGEs and mortality after breast cancer diagnosis. Using data from the Women's Health Initiative (WHI), we examined the association between postdiagnosis intake of a priori–defined dietary AGEs assessed by food frequency questionnaires (FFQ) and all-cause and cause-specific mortality among postmenopausal women diagnosed with invasive breast cancer. The limited number of serum samples available for WHI participants at the time of diagnosis would not allow for sufficient power to use serum concentrations of AGEs in this data set. Whereas serum levels would reflect both endogenous and exogenous sources of AGEs, we focused instead on dietary intake of AGEs as a modifiable risk factor and potential target for future interventions. Specifically, we assessed dietary intake of Nϵ-carboxymethyl-lysine (CML)–AGE, because it is a commonly measured form of AGE in previous epidemiologic studies (17–19).

We hypothesized that higher CML-AGE intake after invasive breast cancer diagnosis would be positively associated with mortality outcomes. We also examined if associations were stronger in hormone receptor–positive breast cancer and among low consumers of fruits and vegetables who may have reduced intake of phytochemicals and nutrients that could counter the proinflammatory and oxidative activity of AGEs. Finally, by incorporating prediagnosis dietary data, we explored whether contrasts in pre- and postdiagnosis CML-AGE intake were differentially associated with mortality risk.

Study population

The WHI enrolled 161,808 postmenopausal women aged between 50 and 79 years from 1993 to 1998 across 40 clinical centers into one or more of three clinical trials (n = 68,132) or an observational study (OS; n = 93,676; refs. 20, 21). The clinical trials included the hormone therapy (HT) trial, dietary modification (DM) trial, and the calcium plus vitamin D supplementation (CaD) trial. The current study used data only from the OS and DM trial of the WHI because FFQs which were used to calculate AGEs were completed multiple times during the course of the study. The methodology of the WHI has been previously published (21). Our study sample included participants who had a first primary diagnosis of invasive breast cancer after enrollment in the WHI, completed a FFQ postdiagnosis, were cancer-free before or at WHI enrollment (except for nonmelanoma skin cancer), had FFQ-assessed energy intakes between ≥600 kcal per day and ≤5000 kcal per day, and had FFQ records in the WHI dietary AGE database (Fig. 1). Our analytical sample contained 2,023 women diagnosed with invasive breast cancer as determined through self report, followed by trained-physician adjudication including review of medical records (22). In the analyses utilizing both pre- and postdiagnosis FFQ data, we further excluded participants with missing baseline (i.e., at time of WHI enrollment) FFQ or with baseline FFQ energy intakes less than 600 or more than 5,000 kcal per day (n = 24).

Figure 1.

Flow chart of participants in the WHI. Only women diagnosed with invasive breast cancer during follow-up and who completed an FFQ after breast cancer diagnosis were included. Additional exclusion criteria are described for the main analyses, the joint analyses of pre- and postdiagnosis dietary intake, and the sensitivity analyses.

Figure 1.

Flow chart of participants in the WHI. Only women diagnosed with invasive breast cancer during follow-up and who completed an FFQ after breast cancer diagnosis were included. Additional exclusion criteria are described for the main analyses, the joint analyses of pre- and postdiagnosis dietary intake, and the sensitivity analyses.

Close modal

Data collection

Participants completed self-administered questionnaires at baseline that obtained information on age at screening for study eligibility, race/ethnicity, education, income, and use of HT. Physical measurements on weight and height from which body mass index (BMI) was calculated were assessed during clinic visits and smoking and physical activity were assessed through personal habit questionnaires. Because BMI after breast cancer diagnosis was missing in 178 participants, BMI at baseline was applied throughout these analyses as it was highly correlated with postdiagnosis BMI (r = 0.81). We also used physical activity and smoking status at baseline because these measures were not updated in the WHI-OS after breast cancer diagnosis.

Usual dietary intake in the past 3 months was assessed through a self-administered FFQ (23). The FFQ consisted of three sections: (i) One hundred and twenty-two foods or food groups soliciting information on usual dietary intake frequency and portion size, (ii) four summary questions on usual intake of fruits and vegetables and added fat for comparison with the information obtained from the line items, and (iii) Nineteen adjustment questions on nutritional content in food, preparation methods, and added fats [Nutrition Data Systems for Research (NDSR®) version 2005]. The nutrition database at the Nutrition Coordinating Center, University of Minnesota was used to derive the nutrients from the FFQ responses (23). FFQs were administered to participants in the WHI-DM at baseline and at one year of follow-up. Subsequently, one third of WHI-DM participants were randomly selected each year and FFQs were administered annually for 9 years. Participants in the WHI-OS were administered FFQs at baseline and after three years of follow-up (23). For prediagnosis dietary assessment we utilized the first FFQ administered at baseline, while the first FFQ that was completed after breast cancer diagnosis (ranged between 1.1–6.3 years; average ± SD was 1.5 ± 1.1 years) was utilized in the postdiagnosis dietary assessment (24).

Exposure assessment

WHI investigators at the Fred Hutchinson Cancer Research Center assigned dietary AGE values to each food item on the FFQ using a published database (1, 10). The database contains CML-AGE content of 549 selected foods commonly consumed in the northeastern metropolitan region of the United States (1). In the database, CML-AGE content was estimated using ELISA based on monoclonal anti-CML antibody (1). CML-AGE values were matched to each food item on the FFQ. CML-AGE has been used as a measure of AGE exposure in previous epidemiologic studies to estimate dietary AGE intake (18, 19, 25). CML-AGE intake was energy adjusted using the nutrient density method (26). Total postdiagnosis CML-AGE intake was categorized into tertiles with the lowest tertile serving as the referent in regression analysis. In the combined pre- and postdiagnosis assessment, CML-AGE intake was categorized into low and high intake categories using median cut points. Categories created include low pre- and low postdiagnosis, low pre- and high postdiagnosis, high pre- and low postdiagnosis, or high pre- and high postdiagnosis. Low pre- and low postdiagnosis CML-AGE intake served as the referent in regression analysis.

Covariate assessment

Potential confounders were identified through a literature search on known or suspected factors implicated in breast cancer survival (9, 27, 28). The factors identified include age at breast cancer diagnosis; time from breast cancer diagnosis to the closest subsequent FFQ (years); race/ethnicity [non-Hispanic white (NHW), black/African American, Hispanic/Latino, or Others); annual household income (missing/don't know, <$20,000, $20,000–<$50,000, or ≥$50,000); education (missing, high school or less, some college, college or some postgraduate, or postgraduate); WHI study arm (WHI-OS, WHI-DM intervention, or WHI-DM control); breast cancer stage (localized, regional, distant, or unknown); estrogen receptor (ER) and progesterone receptor (PR) status [positive, negative, or other (borderline, ordered/results not available, or unknown)]; energy intake (kcal/day); alcohol intake (servings/week); red and processed meat (servings/day); healthy eating index (HEI)-2015 score; baseline BMI (kg/m2; <18.5 or missing, 18.5–25, 25–30, or ≥30); baseline recreational physical activity such as walking, mild, moderate, and strenuous activity (MET-hours/week; missing, 0, ≤3, 3.1–8.9, or ≥9); baseline HT use (never user/missing, past user, or current user); and baseline smoking status (missing, never, past smoker, or current smoker). Because of potential bias if missing covariate data were not missing at random, we included separate categories for missing for physical activity (missing = 147), smoking (missing = 31) and education (missing = 14). For income, missing (n = 76) and “don't know” (n = 30) responses were assumed to be similar and grouped together into one category. For HT use, only two people were missing HT use data and these were grouped with HR never users.

Outcome assessment

Our study outcomes included adjudicated death from all causes, breast cancer, and cardiovascular disease. Cardiovascular disease–related deaths were defined as deaths from definite coronary heart disease, cerebrovascular, pulmonary embolism, possible coronary heart disease, other cardiovascular disease, and unspecified cardiovascular disease (29). Cause of death was ascertained through death certificates, medical records, hospitalization, and autopsy reports (22). Unreported deaths and causes of death were ascertained through data linked to National Death Index of the National Center for Health Statistics (22).

Statistical analyses

Descriptive analyses estimated means and percentages for continuous and categorical characteristics, respectively. The association between CML-AGE intake and risk of mortality from all causes was estimated using hazard ratios (HR) and 95% confidence intervals (CI) calculated from a Cox proportional hazards (PH) model. A competing risk analysis estimated risk of mortality from breast cancer and cardiovascular disease (30). Person time was estimated from the date of diagnosis of invasive breast cancer until death or censoring through March 2018. We adjusted for the time period between breast cancer diagnosis and completion of the FFQ, when no participants were at risk of death (i.e., immortal time; ref. 31) by including a time-dependent covariate which stratified the status of the study outcome before and after completion of postdiagnosis FFQ (29, 32). Adjustment models included a simple model that adjusted for age at breast cancer diagnosis and energy intake. Multivariable Cox PH models included the covariates mentioned previously.

Our analyses were stratified by fruit and vegetable intake which was categorized into tertiles based on the frequency distribution (low intake: 0.3–3.38 servings/day, medium intake: 3.39–5.30 servings/day, or high intake: 5.31–14.29 servings/day). Associations were assessed for differences in the strength of the association by fruit and vegetable intake level. We also stratified our analyses by tumor hormone receptor status (ER, PR, and ER/PR combinations) and assessed for potential interaction with CML-AGE intake by including a multiplicative interaction term in the models. Because dietary changes may occur in the first year after breast cancer diagnosis, we conducted a sensitivity analysis restricting the sample to 1,211 women who completed a FFQ at least 1 year after breast cancer diagnosis. Women enrolled in the WHI-DM intervention arm may have a healthier diet quality thus further analyses were restricted to (i) women enrolled in only the WHI-OS (n = 923), and (ii) WHI-OS and control arm of the WHI-DM (n = 1,599).

PH assumption was assessed using Schoenfeld residual test (33). For covariates violating PH assumption, we fitted stratified Cox PH models for categorical covariates (BMI) and time-dependent Cox PH models for continuous covariates (age at breast cancer diagnosis and intakes of alcohol and red and processed meat). P values for trend (Ptrend) were calculated by modeling CML-AGE as a continuous variable. All statistical analyses were conducted using SAS version 9.4 and statistical significance was set at α = 0.05.

After a median follow-up time of 15.1 years, 630 deaths from all causes were reported, from which 193 were breast cancer–specific and 129 were cardiovascular disease–related deaths. The average daily post-diagnosis CML-AGE consumption was 6,659 ± 2309 kilounits (kU)/1,000 kcal and ranged from 830 kU/1,000 kcal to 19,420 kU/1,000 kcal. Compared with women in the lowest tertile of postdiagnosis CML-AGE intake, women in the highest tertile had higher reported daily energy intake and red and processed meat intake, were more likely to be younger at breast cancer diagnosis, obese, current smokers, physically inactive, and more likely to be diagnosed with PR, regional (locally advanced), or distant breast cancer (Table 1). Women in the highest tertile of postdiagnosis CML-AGE intake had lower intake of fruits and vegetables and were least likely to be current HT users, have a college education or higher, or have a higher income.

Table 1.

Baseline characteristics by tertiles of postdiagnosis CML-AGE (kU/1,000 kcal) intake.

Tertile 1 (n = 674)Tertile 2 (n = 675)Tertile 3 (n = 674)
CML-AGE, kU/1,000 kcal <5,549 5,549–7,311 >7,311 
 Mean (SD) Mean (SD) Mean (SD) 
Years from enrollment to BC diagnosis 2.5 (1.8) 2.8 (2.0) 3.0 (2.0) 
Years from BC diagnosis to FFQ 1.5 (1.0) 1.5 (1.1) 1.6 (1.1) 
Total energy intake, kcal/day 1,477.5 (501.6) 1,568.6 (562.2) 1,578.5 (615.9) 
Alcohol, servings/week 2.5 (6.0) 1.9 (3.4) 2.1 (3.8) 
Red and processed meat, servings/day 0.5 (0.4) 0.7 (0.4) 0.9 (0.6) 
Daily fruit and vegetable intake, portion/day 5.4 (2.4) 4.7 (2.2) 3.8 (1.9) 
Total HEI-2015 score 72.9 (8.8) 68.5 (9.2) 62.3 (8.6) 
Age at BC diagnosis, year 67.4 (7.1) 66.6 (6.8) 66.3 (7.1) 
 N (%) N (%) N (%) 
WHI study arm 
 WHI-OS 348 (51.6) 318 (47.1) 257 (38.1) 
 WHI-DM intervention 194 (28.8) 140 (20.7) 90 (13.4) 
 WHI-DM control 132 (19.6) 217 (32.2) 327 (48.5) 
Race/ethnicity 
 Non-Hispanic White 622 (92.2) 623 (92.3) 573 (85.0) 
 Black or African American 26 (3.9) 28 (4.2) 61 (9.1) 
 Hispanic/Latino 14 (2.1) 14 (2.1) 24 (3.6) 
 Others 12 (1.8) 10 (1.5) 16 (2.4) 
BMI at enrollment (kg/m2
 <18.5 or missing 7 (1.0) 4 (0.6) 12 (1.8) 
 18.5–<25 264 (39.2) 235 (34.8) 159 (23.6) 
 25–<30 232 (34.4) 219 (32.4) 233 (34.6) 
 ≥30 171 (25.4) 217 (32.2) 270 (40.1) 
Physical activity at enrollment, MET-hours/week 
 Missing 35 (5.2) 53 (7.9) 59 (8.8) 
 0 87 (12.9) 85 (12.6) 127 (18.8) 
 ≤3 70 (10.4) 82 (12.2) 90 (13.4) 
 3.1–8.9 154 (22.9) 142 (21.0) 145 (21.5) 
 ≥9 328 (48.7) 313 (46.4) 253 (37.5) 
Income 
 Missing or don't know 32 (4.8) 31 (4.6) 43 (6.4) 
 <$20,000 80 (11.9) 73 (10.8) 86 (12.8) 
 $20,000–<$50,000 296 (43.9) 285 (42.2) 286 (42.4) 
 ≥$50,000 266 (39.5) 286 (42.4) 259 (38.4) 
Education 
 Missing 7 (1.0) 4 (0.6) 3 (0.5) 
 High school or less 107 (15.9) 102 (15.1) 123 (18.3) 
 Some college 233 (34.6) 247 (36.6) 274 (40.7) 
 College or some postgraduate 193 (28.6) 178 (26.4) 159 (23.6) 
 Postgraduate 134 (19.9) 144 (21.3) 115 (17.1) 
Cause of death 
 No death 479 (71.1) 473 (70.1) 441 (65.4) 
 BC-related death 60 (8.9) 52 (7.7) 81 (12.0) 
 CVD-related death 35 (5.2) 46 (6.8) 48 (7.1) 
 Death from other causes 100 (14.8) 104 (15.4) 104 (15.4) 
Cancer stage 
 Localized 514 (76.3) 515 (76.3) 494 (73.3) 
 Regional 151 (22.4) 147 (21.8) 167 (24.8) 
 Distant 3 (0.5) 4 (0.6) 7 (1.0) 
 Unknown 6 (0.9) 9 (1.3) 6 (0.9) 
Smoking status at enrollment 
 Missing 6 (0.9) 11 (1.6) 14 (2.1) 
 Never 339 (50.3) 309 (45.8) 314 (46.6) 
 Past smoker 300 (44.5) 319 (47.3) 310 (46.0) 
 Current smoker 29 (4.3) 36 (5.3) 36 (5.3) 
ER status 
 Positive 508 (75.4) 525 (77.8) 513 (76.1) 
 Negative 94 (14.0) 90 (13.3) 88 (13.1) 
 Other 72 (10.7) 60 (8.9) 73 (10.8) 
PR status 
 Positive 422 (62.6) 438 (64.9) 415 (61.6) 
 Negative 165 (24.5) 165 (24.4) 171 (25.4) 
 Other 87 (12.9) 72 (10.7) 88 (13.1) 
HT use at enrollment 
 Never user or missing 210 (31.2) 231 (34.2) 232 (34.4) 
 Past user 88 (13.1) 79 (11.7) 95 (14.1) 
 Current user 376 (55.8) 365 (54.1) 347 (51.5) 
Tertile 1 (n = 674)Tertile 2 (n = 675)Tertile 3 (n = 674)
CML-AGE, kU/1,000 kcal <5,549 5,549–7,311 >7,311 
 Mean (SD) Mean (SD) Mean (SD) 
Years from enrollment to BC diagnosis 2.5 (1.8) 2.8 (2.0) 3.0 (2.0) 
Years from BC diagnosis to FFQ 1.5 (1.0) 1.5 (1.1) 1.6 (1.1) 
Total energy intake, kcal/day 1,477.5 (501.6) 1,568.6 (562.2) 1,578.5 (615.9) 
Alcohol, servings/week 2.5 (6.0) 1.9 (3.4) 2.1 (3.8) 
Red and processed meat, servings/day 0.5 (0.4) 0.7 (0.4) 0.9 (0.6) 
Daily fruit and vegetable intake, portion/day 5.4 (2.4) 4.7 (2.2) 3.8 (1.9) 
Total HEI-2015 score 72.9 (8.8) 68.5 (9.2) 62.3 (8.6) 
Age at BC diagnosis, year 67.4 (7.1) 66.6 (6.8) 66.3 (7.1) 
 N (%) N (%) N (%) 
WHI study arm 
 WHI-OS 348 (51.6) 318 (47.1) 257 (38.1) 
 WHI-DM intervention 194 (28.8) 140 (20.7) 90 (13.4) 
 WHI-DM control 132 (19.6) 217 (32.2) 327 (48.5) 
Race/ethnicity 
 Non-Hispanic White 622 (92.2) 623 (92.3) 573 (85.0) 
 Black or African American 26 (3.9) 28 (4.2) 61 (9.1) 
 Hispanic/Latino 14 (2.1) 14 (2.1) 24 (3.6) 
 Others 12 (1.8) 10 (1.5) 16 (2.4) 
BMI at enrollment (kg/m2
 <18.5 or missing 7 (1.0) 4 (0.6) 12 (1.8) 
 18.5–<25 264 (39.2) 235 (34.8) 159 (23.6) 
 25–<30 232 (34.4) 219 (32.4) 233 (34.6) 
 ≥30 171 (25.4) 217 (32.2) 270 (40.1) 
Physical activity at enrollment, MET-hours/week 
 Missing 35 (5.2) 53 (7.9) 59 (8.8) 
 0 87 (12.9) 85 (12.6) 127 (18.8) 
 ≤3 70 (10.4) 82 (12.2) 90 (13.4) 
 3.1–8.9 154 (22.9) 142 (21.0) 145 (21.5) 
 ≥9 328 (48.7) 313 (46.4) 253 (37.5) 
Income 
 Missing or don't know 32 (4.8) 31 (4.6) 43 (6.4) 
 <$20,000 80 (11.9) 73 (10.8) 86 (12.8) 
 $20,000–<$50,000 296 (43.9) 285 (42.2) 286 (42.4) 
 ≥$50,000 266 (39.5) 286 (42.4) 259 (38.4) 
Education 
 Missing 7 (1.0) 4 (0.6) 3 (0.5) 
 High school or less 107 (15.9) 102 (15.1) 123 (18.3) 
 Some college 233 (34.6) 247 (36.6) 274 (40.7) 
 College or some postgraduate 193 (28.6) 178 (26.4) 159 (23.6) 
 Postgraduate 134 (19.9) 144 (21.3) 115 (17.1) 
Cause of death 
 No death 479 (71.1) 473 (70.1) 441 (65.4) 
 BC-related death 60 (8.9) 52 (7.7) 81 (12.0) 
 CVD-related death 35 (5.2) 46 (6.8) 48 (7.1) 
 Death from other causes 100 (14.8) 104 (15.4) 104 (15.4) 
Cancer stage 
 Localized 514 (76.3) 515 (76.3) 494 (73.3) 
 Regional 151 (22.4) 147 (21.8) 167 (24.8) 
 Distant 3 (0.5) 4 (0.6) 7 (1.0) 
 Unknown 6 (0.9) 9 (1.3) 6 (0.9) 
Smoking status at enrollment 
 Missing 6 (0.9) 11 (1.6) 14 (2.1) 
 Never 339 (50.3) 309 (45.8) 314 (46.6) 
 Past smoker 300 (44.5) 319 (47.3) 310 (46.0) 
 Current smoker 29 (4.3) 36 (5.3) 36 (5.3) 
ER status 
 Positive 508 (75.4) 525 (77.8) 513 (76.1) 
 Negative 94 (14.0) 90 (13.3) 88 (13.1) 
 Other 72 (10.7) 60 (8.9) 73 (10.8) 
PR status 
 Positive 422 (62.6) 438 (64.9) 415 (61.6) 
 Negative 165 (24.5) 165 (24.4) 171 (25.4) 
 Other 87 (12.9) 72 (10.7) 88 (13.1) 
HT use at enrollment 
 Never user or missing 210 (31.2) 231 (34.2) 232 (34.4) 
 Past user 88 (13.1) 79 (11.7) 95 (14.1) 
 Current user 376 (55.8) 365 (54.1) 347 (51.5) 

Abbreviations: BC, breast cancer; BMI, body mass index; CVD, cardiovascular disease; ER, estrogen receptor; HT, hormone therapy; PR, progesterone receptor.

The results showing HRs and 95% CIs for mortality risk across the tertiles of postdiagnosis CML-AGE intake are presented in Table 2. In multivariable model 2, there was a higher risk for all-cause (HRT3vsT1, 1.33; 95% CI, 1.06–1.66; Ptrend = 0.024), breast cancer–specific (HRT3vsT1, 1.55; 95% CI, 1.05–2.13; Ptrend = 0.098), and cardiovascular disease–related (HRT3vsT1, 1.85; 95% CI, 1.08–3.15; Ptrend = 0.354) mortality. The associations persisted after additional adjustment for red and processed meat intake (model 3) for all-cause mortality (HRT3vsT1, 1.37; 95% CI, 1.09–1.74; Ptrend = 0.009), breast cancer–specific mortality (HRT3vsT1, 1.49; 95% CI, 0.98–2.24; Ptrend = 0.285), and cardiovascular disease–related mortality (HRT3vsT1, 1.91; 95% CI, 1.09–3.32; Ptrend = 0.381). In sensitivity analysis restricting the sample to 1,211 women who completed a FFQ at least one year after breast cancer diagnosis, the results were similar for breast cancer mortality. However, the associations between postdiagnosis CML-AGE intake and all-cause mortality (HRT3vsT1, 1.19; 95% CI, 0.88–1.62; Ptrend = 0.09) and cardiovascular disease mortality (HRT3vsT1, 1.41; 95% CI, 0.69–2.90; Ptrend = 0.576) were attenuated and CIs included the null (Supplementary Table S1). In analyses restricting the sample to include only participants in the WHI-OS and WHI-DM control arm, the results were similar to the main findings for all mortality outcomes (Supplementary Table S2). Among WHI-OS participants only the associations were similar, though the breast cancer mortality association was attenuated (HRT3vsT1, 1.35; 95% CI, 0.79–2.28; Ptrend = 0.264; Supplementary Table S3).

Table 2.

HRs (95% CI) for all-cause mortality, breast cancer–specific mortality, and CVD-related mortality by tertilesa of postdiagnosis CML-AGE intake in women diagnosed with breast cancer in the WHI.

HR (95% CI)
No. of deathsModel 1bModel 2cModel 3d
All-cause mortality 
 Tertile 1 195 Reference Reference Reference 
 Tertile 2 202 1.09 (0.89–1.33) 1.03 (0.84–1.27) 1.04 (0.85–1.29) 
 Tertile 3 233 1.50 (1.24–1.82) 1.33 (1.06–1.66) 1.37 (1.09–1.74) 
Ptrende 630 <0.0001 0.024 0.009 
BC mortality 
 Tertile 1 60 Reference Reference Reference 
 Tertile 2 52 0.85 (0.58–1.23) 0.81 (0.55–1.20) 0.80 (0.54–1.18) 
 Tertile 3 81 1.46 (1.04–2.05) 1.55 (1.05–2.31) 1.49 (0.98–2.24) 
Ptrende 193 0.056 0.098 0.29 
CVD mortalityf 
 Tertile 1 35 Reference Reference Reference 
 Tertile 2 46 1.57 (1.00–2.46) 1.56 (0.97–2.51) 1.58 (0.98–2.55) 
 Tertile 3 48 2.06 (1.31–3.21) 1.85 (1.08–3.15) 1.91 (1.09–3.32) 
Ptrende 129 0.028 0.35 0.38 
HR (95% CI)
No. of deathsModel 1bModel 2cModel 3d
All-cause mortality 
 Tertile 1 195 Reference Reference Reference 
 Tertile 2 202 1.09 (0.89–1.33) 1.03 (0.84–1.27) 1.04 (0.85–1.29) 
 Tertile 3 233 1.50 (1.24–1.82) 1.33 (1.06–1.66) 1.37 (1.09–1.74) 
Ptrende 630 <0.0001 0.024 0.009 
BC mortality 
 Tertile 1 60 Reference Reference Reference 
 Tertile 2 52 0.85 (0.58–1.23) 0.81 (0.55–1.20) 0.80 (0.54–1.18) 
 Tertile 3 81 1.46 (1.04–2.05) 1.55 (1.05–2.31) 1.49 (0.98–2.24) 
Ptrende 193 0.056 0.098 0.29 
CVD mortalityf 
 Tertile 1 35 Reference Reference Reference 
 Tertile 2 46 1.57 (1.00–2.46) 1.56 (0.97–2.51) 1.58 (0.98–2.55) 
 Tertile 3 48 2.06 (1.31–3.21) 1.85 (1.08–3.15) 1.91 (1.09–3.32) 
Ptrende 129 0.028 0.35 0.38 

Abbreviations: BC, breast cancer; CI, confidence interval; CVD, cardiovascular disease; HR, hazard ratio.

aTertile cut-off points in kU/1,000 kcal: Tertile 1: <5,549; Tertile 2: 5,549–7,311; Tertile 3: >7,311.

bAdjusted for age at BC diagnosis and energy intake.

cAdjusted for age at BC diagnosis, energy intake, income, race/ethnicity, study arm, time from BC diagnosis to FFQ, education, physical activity, smoking status, BMI, ER status, PR status, BC stage, HT use, alcohol intake, HEI-2015, and covariate of time-dependent status before and after post-diagnosis FFQ.

dAdjusted for all covariates in c and dietary intake of red and processed meat.

ePtrend estimated by modeling CML-AGE as a continuous variable.

fAdjusted for time-dependent age at BC diagnosis due to PH assumption violation.

In the assessment of combined pre- and postdiagnosis CML-AGE intake, higher CML-AGE intake at both pre- and postdiagnosis was associated with modestly higher risk of all-cause mortality (HR, 1.20; 95% CI, 0.94–1.52), breast cancer mortality (HR, 1.31; 95% CI, 0.84–2.05), and cardiovascular disease mortality (HR, 1.55; 95% CI, 0.89–2.70) compared with women with low pre- and postdiagnosis CML-AGE intake (Fig. 2). Women who reported high postdiagnosis CML-AGE intake, regardless of whether their prediagnosis intake was low or high, had greater risk of death from all causes and breast cancer compared with women with low pre- and postdiagnosis intake, though CIs included the null.

Figure 2.

HRs (95% CI) for all-cause mortality, breast cancer (BC)-specific mortality, and CVD-related mortality by categories of pre/postdiagnosis CML-AGE intake based on the median split. Note: Low/low, Low/high, High/low, and High/high refer to prediagnosis/postdiagnosis intakes of CML-AGE where low prediagnosis refers to <6,752 kU/1,000 kcal, low postdiagnosis refers to <6,362 kU/1,000 kcal, high prediagnosis refers to ≥6,752 kU/1,000 kcal, and high postdiagnosis refers to ≥6,362 kU/1,000 kcal intakes. aAdjusted for age group at breast cancer diagnosis, pre- and postdiagnosis energy intake, income, race/ethnicity, study arm, time from breast cancer diagnosis to FFQ, education, physical activity, smoking status, BMI, ER status, PR status, breast cancer stage, HT use, covariate of time-dependent status before and after postdiagnosis FFQ, and pre- and postdiagnosis alcohol intake, HEI-2015, and dietary intake of red and processed meat.

Figure 2.

HRs (95% CI) for all-cause mortality, breast cancer (BC)-specific mortality, and CVD-related mortality by categories of pre/postdiagnosis CML-AGE intake based on the median split. Note: Low/low, Low/high, High/low, and High/high refer to prediagnosis/postdiagnosis intakes of CML-AGE where low prediagnosis refers to <6,752 kU/1,000 kcal, low postdiagnosis refers to <6,362 kU/1,000 kcal, high prediagnosis refers to ≥6,752 kU/1,000 kcal, and high postdiagnosis refers to ≥6,362 kU/1,000 kcal intakes. aAdjusted for age group at breast cancer diagnosis, pre- and postdiagnosis energy intake, income, race/ethnicity, study arm, time from breast cancer diagnosis to FFQ, education, physical activity, smoking status, BMI, ER status, PR status, breast cancer stage, HT use, covariate of time-dependent status before and after postdiagnosis FFQ, and pre- and postdiagnosis alcohol intake, HEI-2015, and dietary intake of red and processed meat.

Close modal

In Table 3, higher risk for all-cause mortality was observed among all hormone receptor types and significant interaction was detected between CML-AGE and ER status (Pinteraction = 0.042) but not for PR status (Pinteraction = 0.12) or ER/PR combinations (Pinteraction = 0.15). Associations tended to be stronger for hormone receptor–negative subtypes [ER (HRT3vsT1, 1.92; 95% CI, 1.01–3.62); PR (HRT3vsT1, 1.79; 95% CI, 1.10–2.92); ER and PR (HRT3vsT1, 2.02; 95% CI, 1.01–4.05)]. In stratified analyses, positive associations between postdiagnosis CML-AGE intake and risk of breast cancer–specific mortality were strongest among low and medium consumers of fruits and vegetables than among high consumers of fruits and vegetables (Supplementary Table S4). For all-cause and cardiovascular disease–related mortality, higher intake of CML-AGE was associated with higher mortality risk across all strata of fruit and vegetable intake, though CIs included the null.

Table 3.

HRs (95% CI) for all-cause mortality by tertilesa of postdiagnosis CML-AGE intake stratified by hormone receptor status in the WHI.

ER status (n = 1,818)No. of deaths from any causeHR (95% CI)bPinteractionc
ER+ 
 Tertile 1 142 Reference  
 Tertile 2 148 0.99 (0.78–1.27)  
 Tertile 3 171 1.36 (1.04–1.80)  
ER 
 Tertile 1 33 Reference 0.042 
 Tertile 2 32 0.78 (0.43–1.41)  
 Tertile 3 40 1.92 (1.01–3.62)  
PR status (n = 1,776) 
PR+ 
 Tertile 1 115 Reference  
 Tertile 2 122 0.98 (0.75–1.39)  
 Tertile 3 139 1.31 (0.97–1.78)  
PR 
 Tertile 1 56 Reference 0.12 
 Tertile 2 54 1.03 (0.67–1.58)  
 Tertile 3 66 1.79 (1.10–2.92)  
ER/PR combination 
ER+/PR+ (n = 1,578) 
 Tertile 1 145 Reference  
 Tertile 2 152 1.01 (0.79–1.28)  
 Tertile 3 178 1.37 (1.05–1.80)  
ER and PR (n = 234) 
 Tertile 1 28 Reference 0.15 
 Tertile 2 28 0.79 (0.42–1.49)  
 Tertile 3 33 2.02 (1.01–4.05)  
ER status (n = 1,818)No. of deaths from any causeHR (95% CI)bPinteractionc
ER+ 
 Tertile 1 142 Reference  
 Tertile 2 148 0.99 (0.78–1.27)  
 Tertile 3 171 1.36 (1.04–1.80)  
ER 
 Tertile 1 33 Reference 0.042 
 Tertile 2 32 0.78 (0.43–1.41)  
 Tertile 3 40 1.92 (1.01–3.62)  
PR status (n = 1,776) 
PR+ 
 Tertile 1 115 Reference  
 Tertile 2 122 0.98 (0.75–1.39)  
 Tertile 3 139 1.31 (0.97–1.78)  
PR 
 Tertile 1 56 Reference 0.12 
 Tertile 2 54 1.03 (0.67–1.58)  
 Tertile 3 66 1.79 (1.10–2.92)  
ER/PR combination 
ER+/PR+ (n = 1,578) 
 Tertile 1 145 Reference  
 Tertile 2 152 1.01 (0.79–1.28)  
 Tertile 3 178 1.37 (1.05–1.80)  
ER and PR (n = 234) 
 Tertile 1 28 Reference 0.15 
 Tertile 2 28 0.79 (0.42–1.49)  
 Tertile 3 33 2.02 (1.01–4.05)  

aTertile cut-off points in kU/1,000 kcal: Tertile 1: <5,549; Tertile 2: 5,549–7,311; Tertile 3: >7,311.

bAdjusted for age at breast cancer diagnosis, energy intake, income, race/ethnicity, study arm, time from breast cancer diagnosis to FFQ, education, physical activity, smoking status, BMI, breast cancer stage, HT use, alcohol intake, HEI-2015, dietary intake of red and processed meat, and covariate of time-dependent status before and after postdiagnosis FFQ (for ER/PR combination model), ER status (for PR model), and PR status (for ER model).

cInteraction between tertiles of CML-AGE and ER and PR status.

Higher dietary intake of CML-AGE, a modifiable dietary exposure, was associated with higher mortality risk from all causes, breast cancer, and cardiovascular disease in postmenopausal women diagnosed with invasive breast cancer among participants of the large, prospective WHI study. When examining both pre- and postdiagnosis dietary data, higher CML-AGE intake postdiagnosis was associated with higher mortality risk from all causes and breast cancer regardless of prediagnosis intake, suggesting that modification of intake postdiagnosis holds promise to improve outcomes after a breast cancer diagnosis. Positive associations between postdiagnosis CML-AGE intake and mortality from all causes were particularly strong for women with hormone receptor–negative (ER and PR) breast cancer.

There is a scarcity of literature on the association between AGEs and mortality among cancer survivors. Previous epidemiologic studies utilizing the same published AGE database for dietary assessment of CML-AGE are limited to associations with cancer incidence and reported that higher intake levels were associated with the risk of overall (19) and invasive breast cancer (25), and pancreatic cancer in men (18). Whereas our associations persisted even after adjustment for intake of red and processed meats, in previous analyses utilizing the NIH-American Association for Retired Persons Diet and Health Study, the increased risk of invasive breast cancer was attenuated after adjustment for total fat and meat intakes (25). Stratified analyses showed positive associations for breast cancer mortality with high postdiagnosis CML-AGE intake which appeared to be attenuated among high consumers of fruits and vegetables. In the WHI-DM trial, women randomized to a diet characterized by low fat and increased intake of fruits, vegetables, and grains, reduced mortality risk was seen when compared with women in the usual diet comparison group (34), and this effect persisted even after long-term follow-up (median follow-up of 19.6 years; ref. 35). The high amounts of phytochemicals and fiber contained in fruits and vegetables may counter the proinflammatory and oxidative activity of AGEs and be protective of breast cancer (36).

While there has been limited research on postdiagnosis AGEs intake and survival after breast cancer diagnosis, previous studies have examined postdiagnosis overall diet quality, for example by using the HEI-2015 which was developed based on the 2015 to 2020 U.S. Federal Dietary Guidelines for Americans, with higher scores indicating better diet quality (37). In previous studies, better diet quality (higher HEI-2015 score) was associated with a reduced risk of death from all causes but not from breast cancer–specific death in women diagnosed with early-stage breast cancer (38) and invasive breast cancer (24). Similarly, better adherence to the American Cancer Society dietary guidelines post diagnosis was not associated with breast cancer mortality, but was associated with modest decreased risk of death from other causes in the Cancer Prevention Study II Nutrition Cohort (39). In our sample, there was a moderate negative correlation between HEI-2015 score and postdiagnosis CML-AGE intake (r: −0.45 P value: <0.0001). Reduced risk of cardiovascular disease mortality in the WHI among women diagnosed with breast cancer who consumed a more anti-inflammatory diet after diagnosis compared with those consuming a more proinflammatory diet using the dietary inflammatory index (DII) as a measure of inflammatory potential was previously reported (29). While AGEs are not included in the DII scoring algorithm, they have been associated with increased biomarkers of inflammation (40). Chronic diseases and adverse health outcomes have been linked to the Western diet (1, 41–43). The Western diet is characterized by high intake of red and processed meat, fried foods, and products high in sugar and saturated fats, which are also major sources of AGEs (1). Findings from the Nurses' Health Study (NHS) showed a positive relationship between the Western dietary pattern and risk of mortality from all causes in women diagnosed with invasive breast cancer (44). Similarly, higher consumption of grilled/barbecued and smoked meat (sources of AGEs) before and after breast cancer diagnosis compared with low pre- and postdiagnosis intake was associated with increased risk of all-cause mortality among women in the Long Island Breast Cancer Study Project (45).

Accumulation of AGEs in tissues can promote protein structure damage thereby modifying mechanical and physiologic function impacting carcinogenesis and inflammation. AGEs markedly stimulate RAGE activity and therefore increase release of inflammatory cytokines and reactive oxygen species that could induce DNA damage (46–50). RAGE is markedly expressed in breast tumors (51) and increased RAGE expression enhances the proliferation and invasion of breast cancer cells (14, 52). High accumulation of CML in breast cancer tissues (53) and elevated serum CML levels were observed in women with breast cancer (12). Clinical studies in humans suggest a link between high AGE plasma levels and cardiovascular disease outcomes. AGEs can quench nitric oxide and increase production of endothelin favoring vasoconstriction and therefore promote vascular complications (54). These are some of the many potential links between AGEs and cardiovascular disease that may explain the significant association with cardiovascular disease–related mortality demonstrated herein. Plasma CML levels have been linked to the severity of cardiovascular outcomes in patients with coronary heart failure (55). Furthermore, serum AGE levels were associated with increased risk of mortality from all causes and cardiovascular disease in women aged 45 to 64 years who were followed for 18 years in a population registry in Finland (56).

We utilized data from a large prospective study with adequate sample size, long follow-up period, and information on multiple potential confounders. Our study included only women with a postdiagnosis FFQ in the WHI which might introduce bias since women with poorer diet quality may have died or dropped out of the study before completing the FFQ after diagnosis. However, we compared baseline characteristics between participants included in our sample with those excluded who lacked information on postdiagnosis diet (Supplementary Table S5) and noted that average HEI-2015 score from baseline FFQ (prediagnosis) was actually lower among participants in our analytical sample as compared with those who were not included.

Because diet after breast cancer diagnosis was assessed from the first FFQ completed after diagnosis, it is possible that changes to diet during the follow-up period may have occurred. In our sensitivity analyses, we excluded women with a postdiagnosis FFQ completed less than 1 year after diagnosis and the results were unchanged for breast cancer–specific mortality. Of note, the associations for all-cause mortality and cardiovascular disease mortality were attenuated and the CIs included the null, which may be partially due to the smaller sample size and number of cardiovascular disease–related deaths. Comprehensive information on types of treatment received such as chemotherapy could be a clinical predictor for cause-specific mortality. Though information on breast cancer treatment was not included in our analyses, we controlled for breast cancer stage and hormone receptor status, which might serve as an indicator of pharmacologic treatment. In addition, we did not have information on nonsteroidal anti-inflammatory drug use in our dataset. AGEs may increase inflammation and thus, nonsteroidal anti-inflammatory drugs may modify the association between AGEs and mortality.

Based on the distribution of BMI in our study, it is likely that energy intake from diet was underreported and FFQs have known measurement error (57, 58). Since dietary information was obtained both before and after breast cancer diagnosis in all subjects, measurement error that might have occurred from the self-reported FFQ is more likely to be nondifferential (59). The database developed by Uribarri and colleagues that assessed CML-AGE content of over 500 foods and beverages using ELISA was used to estimate total CML-AGE intake (1). Some other analytical methods produced varying CML-AGE contents for certain foods but a standard approach to quantify AGEs is yet to be established (60).

Thus, the true AGE content present in food may be under- or overestimated (17, 60). Other AGE databases measuring AGE contents of foods consumed in other populations have been developed. Some of these databases have included measurement of AGEs other than CML such as carboxyethyllysine (CEL). One database utilized the ultra performance liquid chromatography–MS/MS method to examine CML-AGE contents of 190 food items commonly consumed by the Dutch population (61). The Takeuchi and colleagues database measured CML-AGE contents of 1,650 beverages and foods consumed in Japan using ELISA (62). More recently, an AGE database is being developed by researchers at the Dresden University of Technology and contains AGE contents measured in 537 food items (63). Previous studies reported moderate correlations between estimated dietary AGE intake and serum AGE levels, (64), while one study averaging two assessments of serum CML measured by ELISA and taken 13 weeks apart reported no correlation between intake of foods considered high in dietary AGE and serum CML-AGE (65). Cooking methods utilizing high heat such as grilling or frying are major contributors to total AGE; the FFQ generally did not ascertain food preparation information thus limiting the precision in estimating this exposure. Apart from diet contributing to AGE levels in the body, endogenous production of AGE may contribute to the levels found in serum (1). Also, AGE metabolism may be influenced by the composition of the gut microbiome (66, 67) which may impact serum AGE levels. Thus far, the Uribarri and colleagues AGE database is the most frequently utilized in epidemiologic studies to estimate dietary AGE intake from FFQ responses (18, 19, 25), and thus enhances reproducibility and comparability of our results with other studies. We were unable to explore associations by race/ethnicity due to the small sample size among various racial/ethnic groups. Further studies utilizing large and racially diverse datasets are warranted to explore differential associations by race/ethnicity and breast cancer hormone receptor status.

Higher intake of AGEs was associated with higher risk of major causes of mortality among women diagnosed with breast cancer. Further prospective studies examining dietary AGEs in breast cancer outcomes and intervention studies targeting dietary AGE reduction are needed to evaluate potential benefits on survival outcomes.

L.L. Peterson reports grants from American Cancer Society during the conduct of the study. S.E. Steck reports grants from Susan G. Komen during the conduct of the study. No disclosures were reported by the other authors.

O.O. Omofuma: Conceptualization, formal analysis, methodology, writing–original draft, writing–review and editing. L.L. Peterson: Resources, methodology, writing–review and editing. D.P. Turner: Conceptualization, supervision, methodology, writing–review and editing. A.T. Merchant: Conceptualization, supervision, methodology, writing–review and editing. J. Zhang: Conceptualization, supervision, methodology, writing–review and editing. C.A. Thomson: Methodology, writing–review and editing. M.L. Neuhouser: Methodology, writing–review and editing. L.G. Snetselaar: Methodology, writing–review and editing. B.J. Caan: Methodology, writing–review and editing. A.H. Shadyab: Methodology, writing–review and editing. N. Saquib: Methodology, writing–review and editing. H.R. Banack: Methodology, writing–review and editing. J. Uribarri: Writing–review and editing. S.E. Steck: Conceptualization, supervision, methodology, writing–review and editing.

For a list of all the investigators who have contributed to WHI science, please visit: https://s3-us-west-2.amazonaws.com/www-whi-org/wp-content/uploads/WHI-Investigator-Long-List.pdf. O.O. Omofuma is funded by a graduate training in disparities research grant from Susan G. Komen (GTDR17500160; PI: S.E. Steck; Trainee: O.O. Omofuma); a support to promote advancement of research and creativity (SPARC) grant from the University of South Carolina Office of the Vice President for Research. L.L. Peterson is funded by the American Cancer Society, (MRSG-18-199-01-NEC) and TREC Training Workshop (R25CA203650). D.P. Turner is funded by the NIH/NCI (U54CA210962), NIH/NCI (R21CA218929), and NIH (R01CA245143). S.E. Steck is funded by a graduate training in disparities research grant from Susan G. Komen (GTDR17500160; PI: S.E. Steck). The WHI program is funded by the National Heart, Lung, and Blood Institute, NIH, U.S. Department of Health and Human Services contracts HHSN268201100046C, HHSN268201100001C, HHSN268201100002C, HHSN268201100003C, HHSN268201100004C, and HHSN271201100004C.

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.

1.
Uribarri
J
,
Woodruff
S
,
Goodman
S
,
Cai
W
,
Chen
X
,
Pyzik
R
, et al
Advanced glycation end products in foods and a practical guide to their reduction in the diet
.
J Am Diet Assoc
2010
;
110
:
911
16
.
2.
Luevano-Contreras
C
,
Chapman-Novakofski
K
,
Luevano-Contreras
C
,
Chapman-Novakofski
K
. 
Dietary advanced glycation end products and aging
.
Nutrients
2010
;
2
:
1247
65
.
3.
Ramasamy
R
,
Vannucci
SJ
,
Yan
SSD
,
Herold
K
,
Yan
SF
,
Schmidt
AM
. 
Advanced glycation end products and RAGE: a common thread in aging, diabetes, neurodegeneration, and inflammation
.
Glycobiology
2005
;
15
:
16R
28R
.
4.
Riehl
A
,
Németh
J
,
Angel
P
,
Hess
J
. 
The receptor RAGE: Bridging inflammation and cancer
.
Cell Commun Signal CCS
2009
;
7
:
12
.
5.
Kierdorf
K
,
Fritz
G
. 
RAGE regulation and signaling in inflammation and beyond
.
J Leukoc Biol
2013
;
94
:
55
68
.
6.
Fritz
G
. 
RAGE: a single receptor fits multiple ligands
.
Trends Biochem Sci
2011
;
36
:
625
32
.
7.
Dieterich
M
,
Stubert
J
,
Reimer
T
,
Erickson
N
,
Berling
A
. 
Influence of lifestyle factors on breast cancer risk
.
Breast Care Basel Switz
2014
;
9
:
407
14
.
8.
Buttar
HS
,
Li
T
,
Ravi
N
. 
Prevention of cardiovascular diseases: Role of exercise, dietary interventions, obesity and smoking cessation
.
Exp Clin Cardiol
2005
;
10
:
229
49
.
9.
World Cancer Research Fund/American Institute for Cancer Research
. 
Diet, Nutrition, Physical Activity and Cancer: A Global Perspective
.
Continuous Update Project Expert Report 2018. Available from
: https://www.wcrf.org/dietandcancer.
10.
Uribarri
J
,
del Castillo
MD
,
de la Maza
MP
,
Filip
R
,
Gugliucci
A
,
Luevano-Contreras
C
, et al
Dietary advanced glycation end products and their role in health and Disease12
.
Adv Nutr
2015
;
6
:
461
73
.
11.
Turner
DP
. 
Advanced glycation end-products: a biological consequence of lifestyle contributing to cancer disparity
.
Cancer Res
2015
;
75
:
1925
9
.
12.
Walter
KR
,
Ford
ME
,
Gregoski
MJ
,
Kramer
RM
,
Knight
KD
,
Spruill
L
, et al
Advanced glycation end products are elevated in estrogen receptor-positive breast cancer patients, alter response to therapy, and can be targeted by lifestyle intervention
.
Breast Cancer Res Treat
2019
;
173
:
559
71
.
13.
Tesarová
P
,
Kalousová
M
,
Trnková
B
,
Soukupová
J
,
Argalásová
S
,
Mestek
O
, et al
Carbonyl and oxidative stress in patients with breast cancer–is there a relation to the stage of the disease?
Neoplasma
2007
;
54
:
219
24
.
14.
Nankali
M
,
Karimi
J
,
Goodarzi
MT
,
Saidijam
M
,
Khodadadi
I
,
Razavi
ANE
, et al
Increased expression of the receptor for advanced glycation End-Products (RAGE) Is associated with advanced breast cancer stage
.
Oncol Res Treat
2016
;
39
:
622
8
.
15.
Masoudkabir
F
,
Sarrafzadegan
N
,
Gotay
C
,
Ignaszewski
A
,
Krahn
AD
,
Davis
MK
, et al
Cardiovascular disease and cancer: Evidence for shared disease pathways and pharmacologic prevention
.
Atherosclerosis
2017
;
263
:
343
51
.
16.
Abdel-Qadir
H
,
Austin
PC
,
Lee
DS
,
Amir
E
,
Tu
JV
,
Thavendiranathan
P
, et al
A population-based study of cardiovascular mortality following early-stage breast cancer
.
JAMA Cardiol
2017
;
2
:
88
.
17.
Zhang
Q
,
Ames
JM
,
Smith
RD
,
Baynes
JW
,
Metz
TO
. 
A perspective on the maillard reaction and the analysis of protein glycation by mass spectrometry: Probing the pathogenesis of chronic disease
.
J Proteome Res
2009
;
8
:
754
69
.
18.
Jiao
L
,
Stolzenberg-Solomon
R
,
Zimmerman
TP
,
Duan
Z
,
Chen
L
,
Kahle
L
, et al
Dietary consumption of advanced glycation end products and pancreatic cancer in the prospective NIH-AARP Diet and Health Study
.
Am J Clin Nutr
2015
;
101
:
126
34
.
19.
Omofuma
OO
,
Turner
DP
,
Peterson
LL
,
Merchant
AT
,
Zhang
J
,
Steck
SE
. 
Dietary advanced glycation end-products (AGEs) and risk of breast cancer in the Prostate, Lung, Colorectal and Ovarian Cancer Screening Trial (PLCO)
.
Cancer Prev Res
2020
;
13
:
601
.
20.
Design of the Women's Health Initiative clinical trial and observational study. The Women's Health Initiative Study Group
.
Control Clin Trials
1998
;
19
:
61
109
.
21.
Hays
J
,
Hunt
JR
,
Hubbell
FA
,
Anderson
GL
,
Limacher
M
,
Allen
C
, et al
The women's health initiative recruitment methods and results
.
Ann Epidemiol
2003
;
13
:
S18
77
.
22.
Curb
JD
,
Mctiernan
A
,
Heckbert
SR
,
Kooperberg
C
,
Stanford
J
,
Nevitt
M
, et al
Outcomes ascertainment and adjudication methods in the Women's Health Initiative
.
Ann Epidemiol
2003
;
13
:
S122
128
.
23.
Patterson
RE
,
Kristal
AR
,
Tinker
LF
,
Carter
RA
,
Bolton
MP
,
Agurs-Collins
T
. 
Measurement characteristics of the Women's Health Initiative food frequency questionnaire
.
Ann Epidemiol
1999
;
9
:
178
87
.
24.
George
SM
,
Ballard-Barbash
R
,
Shikany
JM
,
Caan
BJ
,
Freudenheim
JL
,
Kroenke
CH
, et al
Better postdiagnosis diet quality Is associated with reduced risk of death among postmenopausal women with invasive breast cancer in the Women's Health Initiative
.
Cancer Epidemiol Prev Biomark
2014
;
23
:
575
83
.
25.
Peterson
LL
,
Park
S
,
Park
Y
,
Colditz
GA
,
Anbardar
N
,
Turner
DP
. 
Dietary advanced glycation end products and the risk of postmenopausal breast cancer in the National Institutes of Health-AARP Diet and Health Study
.
Cancer
2020
;
126
:
2648
57
.
26.
Willett
WC
,
Howe
GR
,
Kushi
LH
. 
Adjustment for total energy intake in epidemiologic studies
.
Am J Clin Nutr
1997
;
65
:
1220S
8S
.
27.
Caleffi
M
,
Fentiman
IS
,
Birkhead
BG
. 
Factors at presentation influencing the prognosis in breast cancer
.
Eur J Cancer Clin Oncol
1989
;
25
:
51
56
.
28.
Seedhom
AE
,
Kamal
NN
. 
Factors affecting survival of women diagnosed with breast cancer in El-Minia Governorate, Egypt
.
Int J Prev Med
2011
;
2
:
131
8
.
29.
Zheng
J
,
Tabung
FK
,
Zhang
J
,
Liese
AD
,
Shivappa
N
,
Ockene
JK
, et al
Association between post-cancer diagnosis dietary inflammatory potential and mortality among invasive breast cancer survivors in the Women's Health Initiative
.
Cancer Epidemiol Prev Biomark
2018
;
27
:
454
63
.
30.
Fine
JP
. 
Regression modeling of competing crude failure probabilities
.
Biostatistics
2001
;
2
:
85
97
.
31.
Lévesque
LE
,
Hanley
JA
,
Kezouh
A
,
Suissa
S
. 
Problem of immortal time bias in cohort studies: example using statins for preventing progression of diabetes
.
BMJ
2010
;
340
:
b5087
.
32.
Agarwal
P
,
Moshier
E
,
Ru
M
,
Ohri
N
,
Ennis
R
,
Rosenzweig
K
, et al
Immortal time bias in observational studies of time-to-event outcomes
.
Cancer Control J Moffitt Cancer Cent
2018
;
25
.
33.
Schoenfeld
D
. 
Partial residuals for the proportional hazards regression model
.
Biometrika
1982
;
69
:
239
41
.
34.
Chlebowski
RT
,
Aragaki
AK
,
Anderson
GL
,
Thomson
CA
,
Manson
JE
,
Simon
MS
, et al
Low-fat dietary pattern and breast cancer mortality in the Women's Health Initiative Randomized Controlled Trial
.
J Clin Oncol
2017
;
35
:
2919
26
.
35.
Chlebowski
RT
,
Aragaki
AK
,
Anderson
GL
,
Pan
K
,
Neuhouser
ML
,
Manson
JE
, et al
Dietary modification and breast cancer mortality: long-term follow-up of the Women's Health Initiative Randomized Trial
.
J Clin Oncol
2020
;
38
:
1419
28
.
36.
Michels
KB
,
Mohllajee
AP
,
Roset-Bahmanyar
E
,
Beehler
GP
,
Moysich
KB
. 
Diet and breast cancer
.
Cancer
2007
;
109
:
2712
49
.
37.
Krebs-Smith
SM
,
Pannucci
TE
,
Subar
AF
,
Kirkpatrick
SI
,
Lerman
JL
,
Tooze
JA
, et al
Update of the Healthy Eating Index: HEI-2015
.
J Acad Nutr Diet
2018
;
118
:
1591
602
.
38.
George
SM
,
Irwin
ML
,
Smith
AW
,
Neuhouser
ML
,
Reedy
J
,
McTiernan
A
, et al
Postdiagnosis diet quality, the combination of diet quality and recreational physical activity, and prognosis after early-stage breast cancer
.
Cancer Causes Control
2011
;
22
:
589
98
.
39.
McCullough
ML
,
Gapstur
SM
,
Shah
R
,
Campbell
PT
,
Wang
Y
,
Doyle
C
, et al
Pre- and postdiagnostic diet in relation to mortality among breast cancer survivors in the CPS-II Nutrition Cohort
.
Cancer Causes Control
2016
;
27
:
1303
14
.
40.
Almajwal
AM
,
Alam
I
,
Abulmeaty
M
,
Razak
S
,
Pawelec
G
,
Alam
W
. 
Intake of dietary advanced glycation end products influences inflammatory markers, immune phenotypes, and antiradical capacity of healthy elderly in a little-studied population
.
Food Sci Nutr
2020
;
8
:
1046
57
.
41.
Uribarri
J
,
Cai
W
,
Sandu
O
,
Peppa
M
,
Goldberg
T
,
Vlassara
H
. 
Diet-derived advanced glycation end products are major contributors to the body's AGE pool and induce inflammation in healthy subjects
.
Ann N Y Acad Sci
2005
;
1043
:
461
6
.
42.
Uribarri
J
,
Cai
W
,
Peppa
M
,
Goodman
S
,
Ferrucci
L
,
Striker
G
, et al
Circulating glycotoxins and dietary advanced glycation endproducts: two links to inflammatory response, oxidative stress, and aging
.
J Gerontol Ser A
2007
;
62
:
427
33
.
43.
Vlassara
H
. 
Advanced glycation in health and disease: role of the modern environment
.
Ann N Y Acad Sci
2005
;
1043
:
452
60
.
44.
Kroenke
CH
,
Fung
TT
,
Hu
FB
,
Holmes
MD
. 
Dietary patterns and survival after breast cancer diagnosis
.
J Clin Oncol
2005
;
23
:
9295
303
.
45.
Parada
H
,
Steck
SE
,
Bradshaw
PT
,
Engel
LS
,
Conway
K
,
Teitelbaum
SL
, et al
Grilled, barbecued, and smoked meat intake and survival following breast cancer
.
J Natl Cancer Inst
2017
;
109
:
djw299
.
46.
Ward
MS
,
Fotheringham
AK
,
Cooper
ME
,
Forbes
JM
. 
Targeting advanced glycation endproducts and mitochondrial dysfunction in cardiovascular disease
.
Curr Opin Pharmacol
2013
;
13
:
654
61
.
47.
Clarke
RE
,
Dordevic
AL
,
Tan
SM
,
Ryan
L
,
Coughlan
MT
. 
Dietary advanced glycation end products and risk factors for chronic disease: a systematic review of randomised controlled trials
.
Nutrients
2016
;
8
:
125
.
48.
Kellow
NJ
,
Savige
GS
. 
Dietary advanced glycation end-product restriction for the attenuation of insulin resistance, oxidative stress and endothelial dysfunction: a systematic review
.
Eur J Clin Nutr
2013
;
67
:
239
48
.
49.
Bansal
S
,
Siddarth
M
,
Chawla
D
,
Banerjee
BD
,
Madhu
SV
,
Tripathi
AK
. 
Advanced glycation end products enhance reactive oxygen and nitrogen species generation in neutrophils in vitro
.
Mol Cell Biochem
2012
;
361
:
289
96
.
50.
Guimarães
ELM
,
Empsen
C
,
Geerts
A
,
van Grunsven
LA
. 
Advanced glycation end products induce production of reactive oxygen species via the activation of NADPH oxidase in murine hepatic stellate cells
.
J Hepatol
2010
;
52
:
389
97
.
51.
Logsdon
CD
,
Fuentes
MK
,
Huang
EH
,
Arumugam
T
. 
RAGE and RAGE ligands in cancer
.
Curr Mol Med
2007
;
7
:
777
89
.
52.
Sharaf
H
,
Matou-Nasri
S
,
Wang
Q
,
Rabhan
Z
,
Al-Eidi
H
,
Al Abdulrahman
A
, et al
Advanced glycation endproducts increase proliferation, migration and invasion of the breast cancer cell line MDA-MB-231
.
Biochim Biophys Acta
2015
;
1852
:
429
41
.
53.
Nass
N
,
Ignatov
A
,
Andreas
L
,
Weißenborn
C
,
Kalinski
T
,
Sel
S
. 
Accumulation of the advanced glycation end product carboxymethyl lysine in breast cancer is positively associated with estrogen receptor expression and unfavorable prognosis in estrogen receptor-negative cases
.
Histochem Cell Biol
2017
;
147
:
625
34
.
54.
Hartog
JWL
,
Voors
AA
,
Bakker
SJL
,
Smit
AJ
,
van Veldhuisen
DJ
. 
Advanced glycation end-products (AGEs) and heart failure: Pathophysiology and clinical implications
.
Eur J Heart Fail
2007
;
9
:
1146
55
.
55.
Hartog
JWL
,
Voors
AA
,
Schalkwijk
CG
,
Scheijen
J
,
Smilde
TDJ
,
Damman
K
, et al
Clinical and prognostic value of advanced glycation end-products in chronic heart failure
.
Eur Heart J
2007
;
28
:
2879
85
.
56.
Kilhovd
BK
,
Juutilainen
A
,
Lehto
S
,
Rönnemaa
T
,
Torjesen
PA
,
Birkeland
KI
, et al
High serum levels of advanced glycation end products predict increased coronary heart disease mortality in nondiabetic women but not in nondiabetic men
.
Arterioscler Thromb Vasc Biol
2005
;
25
:
815
20
.
57.
Freedman
LS
,
Commins
JM
,
Moler
JE
,
Arab
L
,
Baer
DJ
,
Kipnis
V
, et al
Pooled results from 5 validation studies of dietary self-report instruments using recovery biomarkers for energy and protein intake
.
Am J Epidemiol
2014
;
180
:
172
88
.
58.
Prentice
RL
,
Mossavar-Rahmani
Y
,
Huang
Y
,
Van Horn
L
,
Beresford
SAA
,
Caan
B
, et al
Evaluation and comparison of food records, recalls, and frequencies for energy and protein assessment by using recovery biomarkers
.
Am J Epidemiol
2011
;
174
:
591
603
.
59.
Freedman
LS
,
Schatzkin
A
,
Midthune
D
,
Kipnis
V
. 
Dealing with dietary measurement error in nutritional cohort studies
.
J Natl Cancer Inst
2011
;
103
:
1086
92
.
60.
Nowotny
K
,
Schröter
D
,
Schreiner
M
,
Grune
T
. 
Dietary advanced glycation end products and their relevance for human health
.
Ageing Res Rev
2018
;
47
:
55
66
.
61.
Scheijen
JLJM
,
Clevers
E
,
Engelen
L
,
Dagnelie
PC
,
Brouns
F
,
Stehouwer
CDA
, et al
Analysis of advanced glycation endproducts in selected food items by ultra-performance liquid chromatography tandem mass spectrometry: presentation of a dietary AGE database
.
Food Chem
2016
;
190
:
1145
50
.
62.
Takeuchi
M
,
Takino
J-I
,
Furuno
S
,
Shirai
H
,
Kawakami
M
,
Muramatsu
M
, et al
Assessment of the concentrations of various advanced glycation end-products in beverages and foods that are commonly consumed in Japan
.
PLoS One
2015
;
10
:
e0118652
.
63.
TUD - AGE Database
.
Dresden, Germany
:
Technische Universität Dresden
.
Available from
: https://lemchem.file3.wcms.tu-dresden.de//index.php.
64.
Uribarri
J
,
Peppa
M
,
Cai
W
,
Goldberg
T
,
Lu
M
,
Baliga
S
, et al
Dietary glycotoxins correlate with circulating advanced glycation end product levels in renal failure patients
.
Am J Kidney Dis
2003
;
42
:
532
8
.
65.
Semba
RD
,
Ang
A
,
Talegawkar
S
,
Crasto
C
,
Dalal
M
,
Jardack
P
, et al
Dietary intake associated with serum versus urinary carboxymethyl-lysine, a major advanced glycation end product, in adults: the Energetics Study
.
Eur J Clin Nutr
2012
;
66
:
3
9
.
66.
Seiquer
I
,
Rubio
LA
,
Peinado
MJ
,
Delgado-Andrade
C
,
Navarro
MP
. 
Maillard reaction products modulate gut microbiota composition in adolescents
.
Mol Nutr Food Res
2014
;
58
:
1552
60
.
67.
Snelson
M
,
Coughlan
MT
. 
Dietary advanced glycation end products: digestion, metabolism and modulation of gut microbial ecology
.
Nutrients
2019
;
11
:
215
.