Abstract
Objective:
To investigate the associations of different types of carbohydrate intake after breast cancer diagnosis with breast cancer-specific and all-cause mortality.
Methods:
We prospectively assessed post-diagnostic intake of total sugar, added sugar, and natural sugar as well as carbohydrate from different sources, among 8,932 women with stage I-III breast cancer that were identified in the Nurses’ Health Study from 1980 to 2010 and Nurses’ Health Study II from 1991 to 2011. Participants completed a validated food frequency questionnaire every four years after diagnosis and were followed up for death.
Results:
We prospectively documented 1,071 deaths due to breast cancer and 2,532 all-cause deaths, over a mean of 11.5 years of follow-up. After adjustment for confounding variables, greater post-diagnostic total sugar intake was suggestively associated with greater risk of breast cancer-specific mortality [hazard ratio (HR)Q5vsQ1=1.16, 95% confidence interval (CI)=0.95-1.41; Ptrend=0.02] and significantly associated with greater risk of all-cause mortality (HRQ5vsQ1=1.23, 95% CI=1.08-1.41; Ptrend=0.0001). Greater post-diagnostic added sugar intake was significantly associated with greater risk of all-cause mortality (HRQ5vsQ1=1.20, 95% CI=1.06-1.36; Ptrend=0.001). Post-diagnostic natural sugar (occurring in foods and not added as an ingredient) intake was not associated with mortality risk. Greater post-diagnostic fructose intake was significantly associated with greater risk of breast cancer-specific mortality (HRQ5vsQ1=1.34, 95% CI=1.10-1.64; Ptrend=0.005) and all-cause mortality (HRQ5vsQ1=1.16, 95% CI=1.02-1.32; Ptrend=0.01). High post-diagnostic intake of sucrose was associated with higher risk of breast cancer-specific and all-cause mortality. Increased post-diagnostic intake of carbohydrate from fruit juice was significantly associated with higher risk of breast cancer-specific and all-cause mortality and carbohydrate from vegetables was significantly associated with lower risk of all-cause mortality. High post-diagnostic intake of carbohydrate from potatoes was suggestively associated with higher risk of breast cancer-specific mortality and carbohydrate from refined grains was suggestively associated with higher risk of all-cause mortality.
Conclusion:
We found that higher total sugar intake, especially added sugar, sucrose, and fructose, as well as carbohydrate from fruit juice after a breast cancer diagnosis were associated with poorer prognosis. High post-diagnostic intake of carbohydrate from vegetables was associated with reduced risk of mortality.
Introduction
After a breast cancer diagnosis, women tend to modify their dietary intake to prevent or delay recurrence or death following breast cancer treatment [1, 2]. Nutrition therapy may be an effective and practical approach to improve survivorship through mechanisms associated with glucose and insulin. The hypothesis is that the growth of tumor cells depends on glucose, thus, high blood glucose levels may increase tumor progression [3]. In addition, hyperglycemia stimulates insulin secretion. Insulin is a potent growth factor [4] and high levels of insulin may affect breast cancer prognosis [5], in addition to acceleration of glucose uptake into cells. This hypothesis is supported by our recent analyses within the Nurses’ Health Study (NHS) and the Nurses’ Health Study II (NHSII) that showed a diet high in glycemic load (GL) after breast cancer diagnosis was related to higher risk of breast cancer-specific and all-cause mortality [6]. Furthermore, we found that high intake of fruit juice and sugar-sweetened beverages, as well as intake of total carbohydrates, after breast cancer diagnosis were associated with higher risk of breast cancer-specific and all-cause mortality [6-8]. As a matter of fact, dietary carbohydrate can vary greatly based on food sources and types [9, 10] and may disproportionately be associated with risk of mortality in the 3.8 million women living in the US with breast cancer [11].
There is research evidence that supports the role of sugar consumption in development of obesity, cardiovascular disease (CVD), type 2 diabetes, and cancer [12-16]. In healthy populations, the role of high sugar consumption on mortality has been examined in a few studies with inconsistent findings. In the National Institute of Health-AARP Diet and Health Study, intake of fructose was associated with higher risk of all-cause mortality in men and women. In addition, high intake of total sugar was associated with higher risk of all-cause mortality among women [17]. A 2.75-fold higher risk of CVD mortality was observed among adults in the US who consumed more than 25% of total daily energy intake from added sugar, compared with consumption of less than 10% of total daily energy intake from added sugar [18]. In the Takayama Study, total sugar intake was associated with higher total, CVD, and non-cancer non-CVD mortality in men but not in women [19]. In contrast, among Chinese elderly, added sugar intake was associated with lower risk of CVD mortality; however, it was no longer significant after additionally adjusting for change in body fatness [20]. Furthermore, U-shaped associations were reported between added sugar intake and risk of CVD, cancer, and all-cause mortality in the Malmö Diet and Cancer Study [21].
Given the associations of sugar intake with cardiometabolic disease and incidence of cancer and mortality [12-19], and the role of glucose in tumor growth and progression [3], we hypothesized that high sugar intake would be detrimental for breast cancer survivors. Furthermore, we hypothesized that the carbohydrate from different sources may have different effects on breast cancer survival. In this regard, we evaluated the associations of post-diagnostic carbohydrate consumption according to types and sources with breast cancer-specific and all-cause mortality, using combined data from the NHS and NHSII.
Subjects and Methods
Study Population
For this study, we used data from 2 ongoing cohort studies in the US: the NHS with an enrollment of 121,700 female nurses aged 30-55 years in 1976; and the NHSII with an enrollment of 116,429 female nurses aged 25-42 years in 1989. During follow-up from 1980 to 2010 in the NHS, and from 1991 to 2011 in the NHSII, we identified women with invasive breast cancer. Among women with invasive breast cancer who reported dietary intake at least 12 months after diagnosis, we excluded participants who reported implausible total energy intake after diagnosis (<600 or >3500 kcal/day), left blank more than 70 food items on the food frequency questionnaire (FFQ), reported another cancer diagnosis (except non-melanoma skin cancer) before breast cancer, were stage IV disease at initial diagnosis, or had no information regarding disease stage. Hence, a total of 8,932 women with stage I-III breast cancer were included for the analyses.
Completion of the questionnaire was considered to imply informed consent when the study protocol was approved in 1976 (NHS) and 1989 (NHSII) by the institutional review boards of the Brigham and Women’s Hospital (Boston, MA) and Harvard T.H. Chan School of Public Health (Boston, MA), and those of participating registries as required. The studies were conducted in accordance with recognized ethical guidelines (Declaration of Helsinki).
Assessment of dietary intake
Dietary intake was evaluated using a validated semi-quantitative FFQ that was administered in 1980, 1984, 1986 and every four years thereafter in the NHS, and in 1991 and every four years thereafter in the NHSII (questionnaires are available at http://www.nurseshealthstudy.org/participants/questionnaires). Post-diagnostic dietary intake data were obtained from all available FFQs that women completed at least 12 months after diagnosis. The cumulative average of post-diagnostic dietary intake was calculated to reduce within-person variation and evaluate dietary intake over a long-term period after diagnosis. The amounts of total sugars, added sugars, sucrose, fructose, carbohydrate from fruits, fruit juice, vegetables, whole grains, refined grains, legumes, and potatoes as well as energy in foods were obtained from the Harvard University Food Composition Database to reflect food composition during the periods corresponding to the questionnaire dates. Total sugars refer to the sum of sucrose, fructose, lactose, glucose, and maltose in foods and beverages. Added sugars refer to free, monosaccharides and disaccharides added at the table or used as ingredients in both beverage and solid foods, including syrups and honey, and sugars in concentrated fruit or vegetable juices that are higher than 100% fruit or vegetable juices of the same type. Natural sugars are those naturally occurring in foods and not added as an ingredient; they are derived as follows: total sugars minus added sugars. Carbohydrate from fruits was calculated from whole fruits (excluding fruit juice) and carbohydrate from vegetables was calculated from all vegetables (excluding potatoes). Whole grain carbohydrate was calculated from whole grain foods including dark bread, brown rice, oatmeal, whole grain cold breakfast cereal, and popcorn. Carbohydrate from potatoes was calculated from baked, boiled, and mashed potatoes as well as potato and corn chips and French-fried potatoes. Carbohydrate from fruit juice refers to natural sugar in fruit juice. All nutrient intakes were energy-adjusted through the residuals from the regression of nutrient intake on total energy intake [22].
Ascertainment of breast cancer and death
Self-reported breast cancers on the biennial questionnaires were confirmed by the study physician through reviewing medical records and pathology reports. Furthermore, we obtained information on tumor characteristics, disease stage, estrogen receptor (ER), and progesterone receptor (PR) status, and other relevant information from medical records and pathology reports. After reporting deaths by family members or the postal service or identifying through search of the National Death Index, the study physician ascertained the cause of death through reviewing the death certificate and medical records.
For approximately 70% of women with breast cancer, breast tumor samples were collected. Immunohistochemistry was performed on tissue microarrays (TMAs) to determine the status of ER and PR [23-25]. If TMAs were not available, we extracted ER and PR status from medical records. In the 2,501 breast tissues from the NHS, insulin receptor (IR) expression (cytoplasmic and membranous) was determined through Definiens image analysis software (Tissue Studio, Definiens AG, Munich, Germany) [26].
Covariates
We collected data on post-diagnostic body mass index (BMI), smoking status, alcohol consumption, physical activity, and aspirin use from biennial questionnaires that women returned at least 12 months after diagnosis. Because of the possibility of reverse causation, the cumulative averages of post-diagnostic BMI and physical activity were calculated using 4-year lagged values. Changes in BMI from pre-diagnosis (collected from the last questionnaire before diagnosis) to post-diagnosis (4-year lagged cumulative averages of post-diagnostic BMI) were calculated. We also obtained pre-diagnostic information about menopausal status, age at menopause, postmenopausal hormone use, and oral contraceptive use. Furthermore, we obtained information related to breast cancer characteristics (age at diagnosis, disease stage, self-reported treatment including radiation therapy, chemotherapy, and hormonal treatment), using medical records and supplemental questionnaires.
Statistical analysis
We combined data from the NHS and NHSII. Person-time of follow-up was contributed by study participants from the return date of the first FFQ after diagnosis until death, or the end of the study period (June 1, 2014 in the NHS and June 1, 2015 in the NHSII), whichever came first. We evaluated breast cancer-specific and all-cause mortality as endpoints.
Cox proportional hazards regression models were used to estimate hazard ratios (HRs) and 95% confidence intervals (CIs). Participants diagnosed with breast cancer were divided into quintiles according to the cumulative averages of post-diagnostic total sugar, added sugar, natural sugar, sucrose, and fructose intake, as well as carbohydrate intake from fruits, fruit juice, vegetables, whole grains, refined grains, legumes, and potatoes. Models 1 and 2 were stratified by cohort and adjusted for age at diagnosis and calendar year of diagnosis. Model 2 was additionally adjusted for potential predictors of breast cancer survival including pre-diagnostic BMI, changes in BMI from pre- to post-diagnosis, post-diagnostic physical activity, post-diagnostic smoking, post-diagnostic aspirin use, post-diagnostic alcohol consumption, post-diagnostic total energy intake, pre-diagnostic oral contraceptive use, menopausal status, age at menopause, postmenopausal hormone use, race, tumor characteristics, and treatment, as well as time between diagnosis and first FFQ after diagnosis, and calendar year at start of follow-up of each-2-year questionnaire cycle. To handle missing covariate information for post-diagnostic smoking status (<1%); BMI before diagnosis (<1%); BMI after diagnosis (1.4%); menopausal status, age at menopause, and postmenopausal hormone use before diagnosis (7.0%); post-diagnostic aspirin use (5.8%); post-diagnostic physical activity (9.3%); ER/PR status (9.9%); hormonal treatment (10.2%); radiotherapy (11.6%); and chemotherapy (12.6%), we used missing indicator variables. To address the potential confounding role of other dietary factors, we also evaluated the associations after additionally controlling for dietary sugar intake before diagnosis as well as the post-diagnostic modified alternate healthy eating index (AHEI), total protein intake, or dietary GL.
We conducted stratified analyses to evaluate potential effect modifiers of the sugar intake association with survival by IR status (IR positive/negative), ER status (ER positive/negative), post-diagnostic BMI (<25/≥25kg/m2), and history of type 2 diabetes (yes/no). To evaluate whether the observed associations were related to quality of overall dietary intake, we stratified by post-diagnostic modified AHEI (</≥ median). All analyses were conducted using SAS software version 9.4 (SAS Institute, Cary, NC) with a two-sided p value of <0.05.
Results
Among 8,932 eligible women diagnosed with invasive breast cancer, we documented 1,071 deaths due to breast cancer and 2,523 all-cause deaths, over a mean of 11.5 years of follow-up. As seen in Table 1, participants with higher total sugar intake tended to consume more fruits, smoke less, drink less alcohol, consume less animal fat and protein, take less aspirin, and they were less likely to have used oral contraceptives. However, participants were overweight on average across total sugar intake quintiles.
Table 1.
Age-standardized characteristics of 8,932 women with breast cancer in the combined Nurses’ Health Study and Nurses’ Health Study II after breast cancer diagnosis, according to quintiles of post-diagnostic energy-adjusted total sugar intake
| Characteristic | Total Sugar Intake | ||||
|---|---|---|---|---|---|
| Quintile 1 |
Quintile 2 |
Quintile 3 |
Quintile 4 |
Quintile 5 |
|
| Number | 1,799 | 1,762 | 1,796 | 1,783 | 1,792 |
| Mean | |||||
| Total sugar intake, g/day | 61.0 | 83.0 | 97.2 | 112.6 | 143.2 |
| Added sugar intake, g/day | 24.3 | 33.0 | 39.0 | 46.3 | 66.8 |
| Natural sugar intake, g/day | 37.3 | 50.6 | 59.0 | 67.0 | 76.6 |
| Total fiber intake, g/day | 18.6 | 20.4 | 21.1 | 21.0 | 20.9 |
| Animal fat intake, g/day | 30.8 | 27.8 | 26.3 | 24.8 | 21.8 |
| Total carbohydrate intake, g/day | 174 | 200 | 214 | 226 | 250 |
| Total protein intake, g/day | 76.4 | 75.2 | 74.2 | 72.9 | 67.0 |
| Total energy intake, kcal/day | 1,681 | 1,750 | 1,754 | 1,729 | 1,680 |
| Alcohol consumption, g/day | 11.2 | 6.6 | 4.8 | 3.4 | 2.5 |
| Total fruit intake, servings/day | 1.0 | 1.4 | 1.6 | 1.8 | 2.0 |
| Total vegetable intake, servings/day | 3.1 | 3.2 | 3.3 | 3.1 | 2.8 |
| Whole grain intake, servings/day | 1.0 | 1.1 | 1.1 | 1.1 | 0.9 |
| Refined grain intake, servings/day | 1.9 | 2.0 | 1.9 | 1.8 | 1.5 |
| Age at diagnosis, years | 58.3 | 58.1 | 58.9 | 59.2 | 58.8 |
| BMI, kg/m2 | 27.0 | 26.7 | 26.5 | 26.4 | 26.1 |
| Physical activity, MET-hours/week | 16.1 | 17.6 | 18.3 | 18.3 | 18.5 |
| % | |||||
| Current smokers | 14 | 10 | 8 | 7 | 9 |
| Ever used oral contraceptives | 60 | 57 | 58 | 57 | 55 |
| Ever used menopausal hormone | 47 | 49 | 47 | 49 | 47 |
| Current use of aspirin | 46 | 47 | 43 | 43 | 41 |
| Premenopausal at diagnosis | 26 | 27 | 26 | 25 | 26 |
| Stage of breast cancer | |||||
| I | 62 | 61 | 59 | 60 | 59 |
| II | 28 | 30 | 31 | 30 | 31 |
| III | 10 | 9 | 10 | 10 | 10 |
| Estrogen receptor status | |||||
| Positive | 77 | 76 | 77 | 77 | 77 |
| Negative | 16 | 17 | 17 | 18 | 17 |
| Missing | 7 | 7 | 6 | 5 | 6 |
| Treatment | |||||
| Radiotherapy | 57 | 57 | 55 | 57 | 56 |
| Chemotherapy | 44 | 44 | 46 | 47 | 48 |
| Hormonal treatment | 69 | 67 | 70 | 70 | 70 |
After adjustment for potential confounding variables, post-diagnostic total sugar intake was suggestively associated with higher risk of breast cancer-specific mortality: HRQ5vsQ1=1.16, 95% CI=0.95-1.41; Ptrend=0.02 (Table 2). All-cause mortality risk was significantly higher among women with higher total sugar intake after diagnosis: HRQ5vsQ1=1.23, 95% CI=1.08-1.41; Ptrend=0.0001 (Table 2). The association with risk of all-cause mortality remained significant after additionally adjusting for post-diagnostic modified AHEI (excluding alcohol score), post-diagnostic total protein intake, or pre-diagnostic total sugar intake, but the association was attenuated after additionally adjusting for post-diagnostic dietary GL (Table S1).
Table 2.
Cumulative average of post-diagnostic energy-adjusted dietary sugar intake in relation to mortality after breast cancer diagnosis (n=8,932 women, 2,523 deaths including 1,071 deaths from breast cancer), Nurses’ Health Study and Nurses’ Health Study II
| Quintile | Median of intake (grams) |
Breast cancer-specific mortality | All-cause mortality | ||||
|---|---|---|---|---|---|---|---|
| No. of deaths |
Model 1 | Model 2 | No. of deaths |
Model 1 | Model 2 | ||
| Total sugar | |||||||
| 1 | 67.2 | 209 | 1 | 1 | 441 | 1 | 1 |
| 2 | 84.8 | 176 | 0.83 (0.68-1.01) | 0.91 (0.74-1.12) | 433 | 0.95 (0.84-1.09) | 1.06 (0.93-1.21) |
| 3 | 97.4 | 202 | 0.94 (0.77-1.14) | 1.05 (0.86-1.28) | 497 | 1.02 (0.90-1.16) | 1.12 (0.98-1.28) |
| 4 | 111.1 | 240 | 1.11 (0.92-1.34) | 1.23 (1.01-1.50) | 579 | 1.15 (1.02-1.31) | 1.28 (1.13-1.46) |
| 5 | 132.7 | 244 | 1.16 (0.96-1.39) | 1.16 (0.95-1.41) | 573 | 1.13 (0.99-1.28) | 1.23 (1.08-1.41) |
| P trend | 0.009 | 0.02 | 0.004 | 0.0001 | |||
| Added sugar | |||||||
| 1 | 20.1 | 216 | 1 | 1 | 466 | 1 | 1 |
| 2 | 30.1 | 197 | 0.89 (0.73-1.08) | 0.94 (0.77-1.14) | 478 | 1.03 (0.90-1.17) | 1.06 (0.93-1.21) |
| 3 | 38.2 | 204 | 0.91 (0.75-1.10) | 1.01 (0.83-1.22) | 460 | 0.98 (0.86-1.11) | 1.05 (0.92-1.19) |
| 4 | 47.6 | 220 | 0.99 (0.82-1.19) | 1.12 (0.92-1.35) | 545 | 1.13 (1.00-1.28) | 1.22 (1.07-1.38) |
| 5 | 66.8 | 234 | 1.09 (0.90-1.31) | 1.11 (0.92-1.34) | 574 | 1.24 (1.10-1.41) | 1.20 (1.06-1.36) |
| P trend | 0.14 | 0.11 | <0.0001 | 0.001 | |||
| Natural sugar | |||||||
| 1 | 34.5 | 215 | 1 | 1 | 496 | 1 | 1 |
| 2 | 47.1 | 180 | 0.79 (0.65-0.96) | 0.91 (0.74-1.11) | 432 | 0.80 (0.70-0.91) | 0.89 (0.78-1.02) |
| 3 | 56.6 | 207 | 0.90 (0.74-1.09) | 1.00 (0.82-1.22) | 496 | 0.88 (0.78-1.00) | 0.99 (0.87-1.13) |
| 4 | 66.9 | 233 | 0.99 (0.82-1.20) | 1.13 (0.93-1.37) | 536 | 0.89 (0.79-1.01) | 1.03 (0.91-1.17) |
| 5 | 83.3 | 236 | 1.00 (0.83-1.21) | 0.98 (0.81-1.19) | 563 | 0.86 (0.76-0.97) | 0.96 (0.84-1.09) |
| P trend | 0.32 | 0.58 | 0.15 | 0.89 | |||
| Sucrose | |||||||
| 1 | 22.9 | 198 | 1 | 1 | 435 | 1 | 1 |
| 2 | 30.8 | 217 | 1.05 (0.86-1.27) | 1.02 (0.84-1.24) | 451 | 0.99 (0.87-1.13) | 1.00 (0.87-1.14) |
| 3 | 36.9 | 192 | 0.91 (0.75-1.11) | 1.03 (0.84-1.26) | 494 | 1.04 (0.91-1.18) | 1.13 (0.99-1.29) |
| 4 | 43.6 | 216 | 1.02 (0.84-1.24) | 1.17 (0.96-1.42) | 529 | 1.07 (0.94-1.22) | 1.18 (1.04-1.35) |
| 5 | 56.1 | 248 | 1.21 (1.00-1.46) | 1.22 (1.00-1.48) | 614 | 1.25 (1.10-1.41) | 1.25 (1.10-1.42) |
| P trend | 0.04 | 0.02 | <0.0001 | <0.0001 | |||
| Fructose | |||||||
| 1 | 13.0 | 174 | 1 | 1 | 439 | 1 | 1 |
| 2 | 17.5 | 191 | 1.05 (0.85-1.29) | 1.23 (1.00-1.51) | 463 | 1.00 (0.88-1.14) | 1.11 (0.97-1.27) |
| 3 | 20.9 | 210 | 1.12 (0.92-1.37) | 1.35 (1.10-1.66) | 497 | 1.01 (0.89-1.15) | 1.10 (0.96-1.26) |
| 4 | 24.9 | 240 | 1.27 (1.04-1.54) | 1.44 (1.18-1.76) | 563 | 1.09 (0.96-1.24) | 1.26 (1.11-1.43) |
| 5 | 32.0 | 256 | 1.39 (1.14-1.68) | 1.34 (1.10-1.64) | 561 | 1.10 (0.97-1.25) | 1.16 (1.02-1.32) |
| P trend | <0.0001 | 0.005 | 0.05 | 0.01 | |||
Model 1 was stratified by cohort and adjusted for age at diagnosis (year) and calendar year of diagnosis.
Model 2 was stratified by cohort and adjusted for age at diagnosis (year), calendar year of diagnosis, time between diagnosis and first FFQ (year), calendar year at start of follow-up of each-2-year questionnaire cycle, pre-diagnostic BMI (<18.5, 18.5 to <25.0, 25.0 to <30, 30 to <35, ≥35 kg/m2, missing), BMI change after diagnosis [no change (≥−0.5 to ≤0.5 kg/m2), decrease (<−0.5 kg/m2), increase (>0.5-2 kg/m2), increase (>2 kg/m2), missing], post-diagnostic smoking (never, past, current 1-14 cigarettes/day, current 15-24 cigarettes/day, current ≥25 cigarettes/day, missing), post-diagnostic physical activity (<5, 5 to <11.5, 11.5 to <22, ≥22 MET-h/week, missing), oral contraceptive use (ever, never), post-diagnostic alcohol consumption (<0.15, 0.15 to <2.0, 2.0 to 7.5, ≥7.5 g/day), post-diagnostic total energy intake (quintiles, kcal/day), pre-diagnostic menopausal status, age at menopause, and postmenopausal hormone use status (premenopausal; postmenopausal, age at menopause<50 year, and never postmenopausal hormone use; postmenopausal, age at menopause<50 year, and past postmenopausal hormone use; postmenopausal, age at menopause<50 year, and current postmenopausal hormone use; postmenopausal, age at menopause≥50 year, and never postmenopausal hormone use; postmenopausal, age at menopause≥50 year, and past postmenopausal hormone use; postmenopausal, age at menopause≥50 year, and current postmenopausal hormone use; missing), post-diagnostic aspirin use (never, past, current, missing), race (non-Hispanic white, other), stage of disease (I, II, III), ER/PR status (ER/PR positive, ER positive and PR negative, ER/PR negative, missing), radiotherapy (yes, no, missing), chemotherapy (yes, no, missing), and hormonal treatment (yes, no, missing).
Post-diagnostic added sugar intake was associated with greater risk of all-cause mortality (HRQ5vsQ1=1.20, 95% CI=1.06-1.36; Ptrend=0.001) (Table 2). There was no significant association with risk of breast cancer-specific mortality. The association with all-cause mortality risk remained significant after additionally adjusting for pre-diagnostic intake of added sugar, but not after additionally adjusting for post-diagnostic modified AHEI (excluding alcohol score) or total protein intake. The association was attenuated after additional adjustment for post-diagnostic dietary GL (Table S1). Post-diagnostic natural sugar intake was not associated with risk of breast cancer-specific and all-cause mortality (Table 2). When post-diagnostic added sugar and natural sugar intakes were mutually adjusted, a significant greater association was observed for added sugar intake and risk of all-cause mortality (HR=1.20; 95% CI=1.06-1.37; Ptrend=0.001), but not natural sugar intake (HR=0.98; 95% CI=0.86-1.11; Ptrend=0.60). Post-diagnostic sucrose intake was associated with greater risk of breast cancer-specific mortality (HRQ5vsQ1=1.22, 95% CI=1.00-1.48; Ptrend=0.02) and all-cause mortality (HRQ5vsQ1=1.25, 95% CI=1.10-1.42; Ptrend<0.0001). These associations remained significant after additional adjustment for pre-diagnostic sucrose intake. However, there was no significant association with risk of breast cancer-specific mortality after additional adjustment for post-diagnostic dietary GL or total protein intake (Table S1). Post-diagnostic fructose intake was significantly associated with higher risk of breast cancer-specific mortality (HRQ5vsQ1=1.34, 95% CI=1.10-1.64; Ptrend=0.005) and all-cause mortality (HRQ5vsQ1=1.16, 95% CI=1.02-1.32; Ptrend=0.01) (Table 2). However, there was no significant association after additionally adjusting for post-diagnostic dietary GL or total protein intake (Table S1).
Among post-diagnostic carbohydrate intake from different sources (Table 3), carbohydrate from fruit juice was significantly associated with higher risk of breast cancer-specific mortality (HRQ5vsQ1=1.24, 95%CI=1.02-1.50; Ptrend=0.008) and all-cause mortality (HRQ5vsQ1=1.15, 95%CI=1.01-1.30; Ptrend=0.008). However, the associations were not significant after additionally adjusting for dietary GL. The associations were also attenuated after additional adjustment for post-diagnostic protein intake (Table S1). Carbohydrate from vegetables was significantly associated with lower risk of all-cause mortality (HRQ5vsQ1=0.86, 95%CI=0.75-0.97; Ptrend=0.01) but not breast cancer-specific mortality. The association with all-cause mortality remained significant after additionally adjusting for post-diagnostic dietary GL or pre-diagnostic intake of carbohydrate from vegetables, but not after additionally adjusting for post-diagnostic modified AHEI (excluding fruit, vegetable, and alcohol scores) or total protein intake (Table S1). Higher carbohydrate intake from whole grains was not associated with lower risk of breast cancer-specific or all-cause mortality, while higher carbohydrate intake from refined grains was suggestively associated with higher risk of all-cause mortality (HRQ5vsQ1=1.16, 95% CI=1.02-1.32; Ptrend=0.06). However, the association was not significant after additionally adjusting for post-diagnostic modified AHEI (excluding whole grain and alcohol scores), dietary GL, or total protein intake (Table S1). Higher carbohydrate intake from potatoes was suggestively associated with higher risk of breast cancer-specific mortality (HRQ5vsQ1=1.25, 95% CI=1.02-1.52; Ptrend=0.11) (Table 3). However, the association was not significant after additionally adjusting for post-diagnostic modified AHEI (excluding alcohol score), dietary GL, or total protein intake as well as pre-diagnostic carbohydrate from potatoes (Table S1). In the model including carbohydrate intake from different sources (fruits, fruit juice, vegetables, whole grains, refined grains, legumes, and potatoes) simultaneously, we observed similar results: high intake of carbohydrate from fruit juice was associated with higher risk of breast cancer-specific mortality (HRQ5vsQ1=1.22, 95% CI=1.00-1.48; Ptrend=0.008) and all-cause mortality (HRQ5vsQ1=1.15, 95%CI=1.01-1.31; Ptrend=0.006); higher carbohydrate intake from vegetables was suggestively associated with lower risk of breast cancer-specific mortality (HRQ5vsQ1=0.80, 95%CI=0.64-0.99; Ptrend=0.06) and significantly associated with lower risk of all-cause mortality (HRQ5vsQ1=0.87, 95% CI=0.76-0.99; Ptrend=0.04); high intake of carbohydrate from refined grains was suggestively associated with higher risk of all-cause mortality (HRQ5vsQ1=1.14, 95% CI=1.00-1.30; Ptrend=0.10); and higher carbohydrate intake from potatoes was suggestively associated with higher risk of breast cancer-specific mortality (HRQ5vsQ1=1.25, 95% CI=1.02-1.53; Ptrend=0.09).
Table 3.
Cumulative average of post-diagnostic energy-adjusted carbohydrate intake from different sources in relation to mortality after breast cancer diagnosis (n = 8,932 women, 2,523 deaths including 1,071 deaths from breast cancer), Nurses’ Health Study and Nurses’ Health Study II
| Quintile | Median of intake (grams) |
Breast cancer-specific mortality | All-cause mortality | ||||
|---|---|---|---|---|---|---|---|
| No. of deaths |
Model 1 | Model 2 | No. of deaths |
Model 1 | Model 2 | ||
| Carbohydrate from fruits | |||||||
| 1 | 14.2 | 207 | 1 | 1 | 481 | 1 | 1 |
| 2 | 23.8 | 214 | 0.99 (0.82-1.20) | 1.20 (0.99-1.46) | 494 | 0.95 (0.84-1.08) | 1.07 (0.94-1.21) |
| 3 | 31.5 | 213 | 0.95 (0.78-1.15) | 1.14 (0.94-1.39) | 485 | 0.85 (0.75-0.97) | 1.04 (0.91-1.18) |
| 4 | 40.3 | 225 | 1.01 (0.83-1.22) | 1.23 (1.01-1.49) | 521 | 0.88 (0.77-0.99) | 1.03 (0.91-1.17) |
| 5 | 55.5 | 212 | 0.95 (0.78-1.16) | 1.02 (0.83-1.25) | 542 | 0.84 (0.74-0.95) | 0.97 (0.85-1.11) |
| P trend | 0.71 | 0.99 | 0.004 | 0.42 | |||
| Carbohydrate from fruit juice | |||||||
| 1 | 0.9 | 201 | 1 | 1 | 448 | 1 | 1 |
| 2 | 3.7 | 182 | 0.86 (0.70-1.05) | 1.02 (0.83-1.25) | 431 | 0.94 (0.82-1.07) | 0.98 (0.86-1.12) |
| 3 | 7.7 | 194 | 0.89 (0.73-1.08) | 1.07 (0.88-1.31) | 476 | 0.98 (0.86-1.12) | 1.06 (0.93-1.21) |
| 4 | 13.1 | 234 | 1.05 (0.86-1.26) | 1.23 (1.01-1.49) | 543 | 1.03 (0.90-1.16) | 1.08 (0.95-1.23) |
| 5 | 23.3 | 260 | 1.16 (0.96-1.39) | 1.24 (1.02-1.50) | 625 | 1.09 (0.96-1.23) | 1.15 (1.01-1.30) |
| P trend | 0.005 | 0.008 | 0.03 | 0.008 | |||
| Carbohydrate from vegetables | |||||||
| 1 | 8.0 | 225 | 1 | 1 | 596 | 1 | 1 |
| 2 | 11.6 | 204 | 0.88 (0.73-1.06) | 0.88 (0.73-1.07) | 523 | 0.89 (0.79-1.00) | 0.94 (0.83-1.06) |
| 3 | 14.4 | 219 | 0.94 (0.78-1.13) | 0.95 (0.79-1.15) | 489 | 0.85 (0.75-0.95) | 0.92 (0.82-1.04) |
| 4 | 17.7 | 220 | 0.95 (0.79-1.15) | 0.92 (0.76-1.11) | 467 | 0.83 (0.74-0.94) | 0.89 (0.78-1.00) |
| 5 | 23.9 | 203 | 0.91 (0.75-1.10) | 0.84 (0.69-1.02) | 448 | 0.83 (0.73-0.93) | 0.86 (0.75-0.97) |
| P trend | 0.56 | 0.14 | 0.002 | 0.01 | |||
| Carbohydrate from whole grains | |||||||
| 1 | 5.7 | 298 | 1 | 1 | 615 | 1 | 1 |
| 2 | 12.4 | 217 | 0.79 (0.66-0.94) | 1.06 (0.88-1.27) | 512 | 0.90 (0.80-1.01) | 0.97 (0.86-1.09) |
| 3 | 18.6 | 197 | 0.74 (0.61-0.88) | 1.08 (0.89-1.30) | 508 | 0.88 (0.78-0.99) | 0.98 (0.86-1.11) |
| 4 | 25.9 | 165 | 0.65 (0.54-0.79) | 1.04 (0.85-1.29) | 426 | 0.75 (0.66-0.85) | 0.89 (0.78-1.02) |
| 5 | 38.3 | 194 | 0.80 (0.66-0.97) | 1.12 (0.91-1.37) | 462 | 0.80 (0.71-0.91) | 0.92 (0.80-1.05) |
| P trend | 0.01 | 0.36 | <0.0001 | 0.13 | |||
| Carbohydrate from refined grains | |||||||
| 1 | 25.7 | 212 | 1 | 1 | 499 | 1 | 1 |
| 2 | 35.4 | 213 | 0.93 (0.77-1.13) | 1.08 (0.89-1.31) | 506 | 1.03 (0.91-1.16) | 1.10 (0.97-1.24) |
| 3 | 42.7 | 233 | 0.99 (0.82-1.20) | 1.19 (0.99-1.44) | 543 | 1.11 (0.98-1.25) | 1.20 (1.06-1.36) |
| 4 | 50.7 | 207 | 0.89 (0.73-1.08) | 1.03 (0.85-1.25) | 493 | 1.07 (0.94-1.21) | 1.07 (0.94-1.21) |
| 5 | 64.7 | 206 | 0.92 (0.76-1.12) | 0.96 (0.79-1.18) | 482 | 1.13 (0.99-1.28) | 1.16 (1.02-1.32) |
| P trend | 0.35 | 0.50 | 0.06 | 0.06 | |||
| Carbohydrate from legumes | |||||||
| 1 | 2.1 | 226 | 1 | 1 | 540 | 1 | 1 |
| 2 | 3.7 | 223 | 0.98 (0.81-1.18) | 1.15 (0.95-1.39) | 509 | 0.97 (0.86-1.09) | 1.08 (0.95-1.22) |
| 3 | 5.1 | 211 | 0.92 (0.76-1.10) | 1.15 (0.95-1.39) | 533 | 0.99 (0.88-1.12) | 1.12 (0.99-1.27) |
| 4 | 6.7 | 196 | 0.87 (0.72-1.05) | 1.12 (0.92-1.36) | 470 | 0.89 (0.79-1.01) | 0.99 (0.87-1.12) |
| 5 | 10.5 | 215 | 1.01 (0.83-1.21) | 1.12 (0.92-1.36) | 471 | 0.96 (0.85-1.08) | 0.99 (0.88-1.13) |
| P trend | 0.92 | 0.44 | 0.35 | 0.47 | |||
| Carbohydrate from potatoes | |||||||
| 1 | 4.0 | 183 | 1 | 1 | 446 | 1 | 1 |
| 2 | 7.9 | 231 | 1.18 (0.97-1.44) | 1.23 (1.01-1.49) | 503 | 1.13 (0.99-1.28) | 1.11 (0.97-1.26) |
| 3 | 11.6 | 202 | 1.01 (0.83-1.24) | 1.15 (0.94-1.42) | 506 | 1.12 (0.98-1.27) | 1.11 (0.97-1.26) |
| 4 | 15.9 | 216 | 1.07 (0.88-1.30) | 1.16 (0.95-1.41) | 512 | 1.14 (1.00-1.29) | 1.10 (0.97-1.25) |
| 5 | 24.0 | 239 | 1.20 (0.99-1.46) | 1.25 (1.02-1.52) | 556 | 1.22 (1.08-1.38) | 1.13 (0.99-1.28) |
| P trend | 0.17 | 0.11 | 0.005 | 0.14 | |||
Model 1 was stratified by cohort and adjusted for age at diagnosis (year) and calendar year of diagnosis.
Model 2 was stratified by cohort and adjusted for age at diagnosis (year), calendar year of diagnosis, time between diagnosis and first FFQ (year), calendar year at start of follow-up of each-2-year questionnaire cycle, pre-diagnostic BMI (<18.5, 18.5 to <25.0, 25.0 to <30, 30 to <35, ≥35 kg/m2, missing), BMI change after diagnosis [no change (≥−0.5 to ≤0.5 kg/m2), decrease (<−0.5 kg/m2), increase (>0.5-2 kg/m2), increase (>2 kg/m2), missing], post-diagnostic smoking (never, past, current 1-14 cigarettes/day, current 15-24 cigarettes/day, current ≥25 cigarettes/day, missing), post-diagnostic physical activity (<5, 5 to <11.5, 11.5 to <22, ≥22 MET-h/week, missing), oral contraceptive use (ever, never), post-diagnostic alcohol consumption (<0.15, 0.15 to <2.0, 2.0 to 7.5, ≥7.5 g/day), post-diagnostic total energy intake (quintiles, kcal/day), pre-diagnostic menopausal status, age at menopause, and postmenopausal hormone use status (premenopausal; postmenopausal, age at menopause<50 year, and never postmenopausal hormone use; postmenopausal, age at menopause<50 year, and past postmenopausal hormone use; postmenopausal, age at menopause<50 year, and current postmenopausal hormone use; postmenopausal, age at menopause≥50 year, and never postmenopausal hormone use; postmenopausal, age at menopause≥50 year, and past postmenopausal hormone use; postmenopausal, age at menopause≥50 year, and current postmenopausal hormone use; missing), post-diagnostic aspirin use (never, past, current, missing), race (non-Hispanic white, other), stage of disease (I, II, III), ER/PR status (ER/PR positive, ER positive and PR negative, ER/PR negative, missing), radiotherapy (yes, no, missing), chemotherapy (yes, no, missing), and hormonal treatment (yes, no, missing).
We also examined the associations using dietary intake from the first FFQ after diagnosis and last FFQ before diagnosis. For the first FFQ after diagnosis, all associations were weaker except for the greater risk of breast cancer-specific mortality with greater intake of fructose (Table S2). We did not observe any significant associations of pre-diagnostic total sugar, added sugar, natural sugar, sucrose, and fructose intake from last FFQ before diagnosis and risk of breast cancer-specific or all-cause mortality (Table S3). However, pre-diagnostic carbohydrate intake from whole grains was suggestively associated with lower risk of all-cause mortality (HRQ5vsQ1=0.88, 95% CI=0.76-1.01; Ptrend=0.02). In contrast, pre-diagnostic carbohydrate intake from legumes was significantly associated with higher risk of breast cancer-specific mortality (HRQ5vsQ1=1.32, 95% CI=1.08-1.61; Ptrend=0.005) and all-cause mortality (HRQ5vsQ1=1.20, 95% CI=1.05-1.37; Ptrend=0.01).
When we looked at the associations based on IR status (Table 4), post-diagnostic fructose intake was significantly associated with higher risk of breast cancer-specific mortality among women with IR negative tumors (HRQ5vsQ1=1.73, 95% CI=1.09-2.74; Ptrend=0.03, P for interaction=0.03). Furthermore, the association between post-diagnostic total sugar intake and breast cancer-specific mortality differed by IR status (P for interaction=0.03) and the association between post-diagnostic added sugar intake and breast cancer-specific mortality differed by ER status (P for interaction=0.04) (Table 4).
Table 4.
Cumulative average of post-diagnostic energy-adjusted dietary sugar intake in relation to breast cancer-specific mortality after breast cancer diagnosis, stratified by insulin receptor status (n = 2,501 women, n=392 breast cancer deaths) and estrogen receptor status (n=8,384 women, n=982 breast cancer deaths), Nurses’ Health Study and Nurses’ Health Study II
| Quintile | Median of intake (grams) |
IR Status | ER Status | ||||||
|---|---|---|---|---|---|---|---|---|---|
| No. of deaths | IR positive | No. of deaths | IR negative | No. of deaths | ER positive | No. of deaths | ER negative | ||
| Total sugar | |||||||||
| 1 | 67.2 | 41 | 1 | 43 | 1 | 157 | 1 | 37 | 1 |
| 2 | 84.8 | 29 | 0.90 (0.55-1.47) | 37 | 0.81 (0.51-1.29) | 126 | 0.89 (0.70-1.13) | 33 | 0.95 (0.58-1.54) |
| 3 | 97.4 | 30 | 0.86 (0.52-1.42) | 39 | 0.87 (0.55-1.38) | 146 | 1.07 (0.85-1.35) | 39 | 0.96 (0.60-1.55) |
| 4 | 111.1 | 41 | 1.10 (0.69-1.76) | 53 | 1.32 (0.85-2.06) | 169 | 1.15 (0.91-1.44) | 50 | 1.28 (0.81-2.02) |
| 5 | 132.7 | 36 | 0.72 (0.43-1.20) | 43 | 1.22 (0.76-1.94) | 171 | 1.10 (0.87-1.39) | 54 | 1.40 (0.89-2.22) |
| P trend | 0.38 | 0.11 | 0.14 | 0.05 | |||||
| P Interaction | 0.03 | 0.11 | |||||||
| Added sugar | |||||||||
| 1 | 20.1 | 38 | 1 | 42 | 1 | 163 | 1 | 33 | 1 |
| 2 | 30.1 | 35 | 1.05 (0.65-1.69) | 55 | 1.24 (0.81-1.89) | 152 | 0.97 (0.77-1.21) | 32 | 0.99 (0.60-1.64) |
| 3 | 38.2 | 36 | 1.07 (0.66-1.73) | 43 | 0.93 (0.60-1.46) | 147 | 1.03 (0.82-1.29) | 40 | 0.98 (0.60-1.59) |
| 4 | 47.6 | 43 | 1.27 (0.79-2.04) | 36 | 0.89 (0.55-1.43) | 148 | 1.08 (0.86-1.35) | 55 | 1.40 (0.89-2.21) |
| 5 | 66.8 | 25 | 0.60 (0.35-1.03) | 39 | 1.19 (0.73-1.92) | 159 | 1.06 (0.85-1.33) | 53 | 1.31 (0.83-2.08) |
| P trend | 0.10 | 0.90 | 0.43 | 0.09 | |||||
| P Interaction | 0.20 | 0.04 | |||||||
| Natural sugar | |||||||||
| 1 | 34.5 | 27 | 1 | 41 | 1 | 151 | 1 | 50 | 1 |
| 2 | 47.1 | 35 | 1.21 (0.72-2.03) | 34 | 0.77 (0.48-1.25) | 118 | 0.86 (0.68-1.10) | 38 | 0.85 (0.55-1.32) |
| 3 | 56.6 | 32 | 1.25 (0.73-2.14) | 43 | 0.87 (0.55-1.38) | 154 | 1.01 (0.80-1.28) | 35 | 0.89 (0.57-1.39) |
| 4 | 66.9 | 45 | 1.67 (1.00-2.81) | 49 | 0.96 (0.62-1.49) | 172 | 1.17 (0.93-1.47) | 48 | 1.21 (0.79-1.84) |
| 5 | 83.3 | 38 | 0.97 (0.56-1.67) | 48 | 0.98 (0.62-1.56) | 174 | 0.97 (0.76-1.22) | 42 | 1.00 (0.64-1.56) |
| P trend | 0.97 | 0.73 | 0.61 | 0.56 | |||||
| P Interaction | 0.47 | 0.83 | |||||||
| Sucrose | |||||||||
| 1 | 22.9 | 42 | 1 | 39 | 1 | 154 | 1 | 30 | 1 |
| 2 | 30.8 | 32 | 0.89 (0.55-1.43) | 48 | 1.14 (0.73-1.78) | 161 | 1.02 (0.81-1.28) | 36 | 1.39 (0.84-2.29) |
| 3 | 36.9 | 40 | 1.22 (0.77-1.94) | 45 | 1.11 (0.71-1.75) | 141 | 1.06 (0.84-1.34) | 40 | 1.10 (0.67-1.81) |
| 4 | 43.6 | 28 | 0.72 (0.43-1.20) | 42 | 1.04 (0.65-1.66) | 142 | 1.05 (0.83-1.33) | 61 | 1.68 (1.05-2.68) |
| 5 | 56.1 | 35 | 0.80 (0.49-1.30) | 41 | 1.02 (0.63-1.66) | 171 | 1.15 (0.92-1.45) | 46 | 1.31 (0.80-2.14) |
| P trend | 0.25 | 0.90 | 0.20 | 0.24 | |||||
| P Interaction | 0.33 | 0.17 | |||||||
| Fructose | |||||||||
| 1 | 13.0 | 31 | 1 | 35 | 1 | 124 | 1 | 36 | 1 |
| 2 | 17.5 | 33 | 1.10 (0.65-1.84) | 43 | 1.33 (0.83-2.12) | 140 | 1.34 (1.04-1.71) | 39 | 1.22 (0.76-1.95) |
| 3 | 20.9 | 36 | 1.55 (0.93-2.58) | 38 | 1.31 (0.81-2.13) | 139 | 1.31 (1.02-1.68) | 47 | 1.51 (0.96-2.38) |
| 4 | 24.9 | 43 | 1.91 (1.16-3.14) | 45 | 1.28 (0.80-2.03) | 188 | 1.54 (1.22-1.95) | 40 | 1.55 (0.96-2.50) |
| 5 | 32.0 | 34 | 0.80 (0.47-1.38) | 54 | 1.73 (1.09-2.74) | 178 | 1.32 (1.04-1.68) | 51 | 1.38 (0.88-2.16) |
| P trend | 0.62 | 0.03 | 0.03 | 0.17 | |||||
| P Interaction | 0.03 | 0.68 | |||||||
Models were stratified by cohort and adjusted for age at diagnosis (year), calendar year of diagnosis, time between diagnosis and first FFQ (year), calendar year at start of follow-up of each-2-year questionnaire cycle, pre-diagnostic BMI (<18.5, 18.5 to <25.0, 25.0 to <30, 30 to <35, ≥35 kg/m2, missing), BMI change after diagnosis [no change (≥−0.5 to ≤0.5 kg/m2), decrease (<−0.5 kg/m2), increase (>0.5-2 kg/m2), increase (>2 kg/m2), missing], post-diagnostic smoking (never, past, current 1-14 cigarettes/day, current 15-24 cigarettes/day, current ≥25 cigarettes/day, missing), post-diagnostic physical activity (<5, 5 to <11.5, 11.5 to <22, ≥22 MET-h/week, missing), oral contraceptive use (ever, never), post-diagnostic alcohol consumption (<0.15, 0.15 to <2.0, 2.0 to 7.5, ≥7.5 g/day), post-diagnostic total energy intake (quintiles, kcal/day), pre-diagnostic menopausal status, age at menopause, and postmenopausal hormone use status (premenopausal; postmenopausal, age at menopause<50 year, and never postmenopausal hormone use; postmenopausal, age at menopause<50 year, and past postmenopausal hormone use; postmenopausal, age at menopause<50 year, and current postmenopausal hormone use; postmenopausal, age at menopause≥50 year, and never postmenopausal hormone use; postmenopausal, age at menopause≥50 year, and past postmenopausal hormone use; postmenopausal, age at menopause≥50 year, and current postmenopausal hormone use; missing), post-diagnostic aspirin use (never, past, current, missing), race (non-Hispanic white, other), stage of disease (I, II, III), ER/PR status (ER/PR positive, ER positive and PR negative, ER/PR negative, missing), radiotherapy (yes, no, missing), chemotherapy (yes, no, missing), and hormonal treatment (yes, no, missing). In the ER status analysis, we did not adjust for ER/PR status.
When we evaluated the associations based on adherence to AHEI, we did not observe any significant interactions (Table S4). Furthermore, when we looked at the associations based on obesity status, post-diagnostic intake of added sugar was significantly associated with higher risk of breast cancer-specific mortality among women with BMI ≥ 25 kg/m2 (HRQ5vsQ1=1.34, 95%CI=1.04-1.72; Ptrend=0.003, P for interaction=0.01) (Table S5). We did not observe any significant interaction by history of type 2 diabetes (Table S6).
Discussion
In this prospective study of participants with stage I-III breast cancer, we found that high post-diagnostic intakes of sucrose and fructose were associated with increased risk of breast cancer-specific mortality. High intake of total sugar was suggestively associated with higher risk of breast cancer-specific mortality. In addition, greater intakes of total sugar, particularly added sugar, sucrose, and fructose after diagnosis were associated with poorer overall survival. Stratified analysis suggests potential effect modification of fructose on breast cancer-specific mortality by IR status, particularly at the highest level of intake and potential effect modification of added sugar on breast cancer-specific mortality by obesity status. In contrast, higher post-diagnostic intake of carbohydrate from vegetables was associated with a lower risk of all-cause mortality, whereas high intake of carbohydrate from fruit juice was associated with increased risk of mortality. High intake of carbohydrate from potatoes was suggestively associated with higher risk of breast cancer-specific mortality and high intake of carbohydrate from refined grains was suggestively associated with higher risk of all-cause mortality.
Carbohydrate intake may play a role in cancer development and prognosis. Higher dietary carbohydrate not only provides the nutrient for tumor cell growth and proliferation but also stimulates insulin secretion, which promotes breast cancer cell proliferation [27]. Role of carbohydrate intake in cancer prognosis has been evaluated in limited studies. High total carbohydrate intake in patients with stage III colon cancer was associated with poorer overall survival [28], whereas plant-rich, low-carbohydrate diet score was associated with lower risk of mortality among stage I-III colorectal cancer patients [29]. Our previous study showed that post-diagnostic total carbohydrate intake was associated with higher risk of breast cancer-specific and all-cause mortality [6]. This study suggests that carbohydrate from different sources may have different effects on breast cancer outcomes. High post-diagnostic intake of carbohydrate from vegetables is associated with better prognosis, whereas high post-diagnostic intake of carbohydrate from fruit juice is associated with poorer prognosis.
Carbohydrates in vegetables coexist with fiber, in addition to anti-carcinogenic and other protective bioactive substances, which can slow the rate of carbohydrate digestion and absorption, and prevent hyperglycemia. Supported by findings from our previous studies, high intakes of vegetables and fiber were associated with lower overall mortality risk among women with breast cancer [6, 7]. In contrast, the suggestive increased risk of mortality with high intake of carbohydrate from refined grains may result from the high glycemic index and low amount of fiber in these food items.
Potatoes in the FFQ include potato and corn chips, French fried potatoes, and baked, boiled or mashed potatoes, with no information on consumption with skin or without. Cooking methods and type of potato influence the glycemic index [30]. Moreover, potatoes are good sources of some nutrients and phytochemicals [31], with higher fiber when eaten with skin. Therefore, our finding of suggestive poor prognosis with high post-diagnostic potato intake needs to be interpreted with caution.
Fructose exists naturally in fruits, fruit juice, certain vegetables, honey, and table sugar. High-fructose corn syrup is present in sodas, candies, baked goods, and other processed foods. Although fructose raises blood sugar levels slowly compared to glucose, high fructose intake may cause insulin resistance, which can lead to obesity and type 2 diabetes [32], known risk factors of poor breast cancer prognosis [33-35]. However, independently of pre- to post-diagnostic weight change, high intake of fructose was associated with poorer breast cancer prognosis and we did not observe a significant interaction by BMI after diagnosis. This suggests that the adverse effect may be independent of adiposity. Furthermore, high intake of fructose has been found to increase inflammatory markers [36, 37]. However, it is worth mentioning that fructose from different sources may not have similar effects on cancer prognosis. As stated before, coexisting with fiber and other anti-cancer and protective bioactive components in some food items, such as fruits and vegetables, may protect the body from adverse effects of excess fructose. Supported by findings from our previous studies, fruit or vegetable intake was not associated with risk of breast cancer-specific mortality [7]. But high intake of foods high in carbohydrate and low in fiber such as fruit juice and sugar-sweetened beverages, as sources of fructose in the diet, may increase risk of both breast cancer-specific and all-cause mortality [7, 8]. Unfortunately, in this study, we were unable to evaluate the associations of fructose intake based on dietary sources, but this needs to be evaluated in future research. In addition, we found that higher fructose intake was associated with higher risk of breast cancer death among women with IR negative tumors. This may be due to the fact that fructose decreases insulin receptor mRNA levels and insulin receptor numbers [38]. Also, insulin is not required for cells to uptake and metabolize fructose [32]. However, our finding warrants further investigation.
Additionally, while no significant associations were found between post-diagnostic intake of carbohydrate from legumes and breast cancer prognoses, further investigations are needed to understand the significant association found between pre-diagnostic intake of carbohydrate from legumes and risk of breast cancer-specific and all-cause mortality. This, given previous reports have demonstrated significant protective effects of intake of legumes and cancer and overall mortality [39, 40].
The present study has several strengths. Highly detailed prospective diet and lifestyle factor data were collected before and after breast cancer diagnosis which provided a unique opportunity to evaluate the role of dietary sugar and carbohydrate types on breast cancer survival after controlling for other relevant factors. These analyses also highlighted the importance of evaluating the role of diet on breast cancer survival using repeated assessments of diet after diagnosis instead of only one diet assessment. Moreover, we were able to examine associations by hormone receptor status. Long-term follow-up and using standard medical record review are other strengths of the current study.
However, the present study has some limitations. Even though we controlled for several risk factors of breast cancer survival, residual confounding is possible. Since the majority of participants were non-Hispanic white and health professionals, the findings of the current study may not be generalizable to other sociodemographic groups. Type I error is possible due to multiple comparisons. However, our results are biologically plausible and largely consistent with previous findings.
In summary, we observed higher risk of breast cancer-specific mortality with higher post-diagnostic intakes of sucrose and fructose. Poorer overall survival after breast cancer diagnosis was observed among women with higher post-diagnostic intakes of total sugar, added sugar, sucrose, and fructose. Although post-diagnostic intake of carbohydrate from vegetables was associated with better survival, high post-diagnostic intake of carbohydrate from fruit juice was associated with increased risk of mortality. Further large-scale prospective studies are needed to confirm current findings, and more experimental data are needed to clarify the underlying mechanisms.
Supplementary Material
Acknowledgements
We would like to thank the participants and staff of the NHS and NHSII for their valuable contributions as well as the following state cancer registries for their help: AL, AZ, AR, CA, CO, CT, DE, FL, GA, ID, IL, IN, IA, KY, LA, ME, MD, MA, MI, NE, NH, NJ, NY, NC, ND, OH, OK, OR, PA, RI, SC, TN, TX, VA, WA, WY.
Funding
The study was supported by the National Institutes of Health (grants U01 CA176726, UM1 CA186107) and the American Institute for Cancer Research (to MSF). The study sponsors were not involved in the study design and collection, analysis and interpretation of data, or the writing of the article or the decision to submit it for publication. The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health. The authors were independent from study sponsors.
Footnotes
Conflict interest Michelle D. Holmes reported a grant from FHI Solutions, non-financial support from Bayer AG (Bayer supplies aspirin and placebo for the Aspirin after Breast Cancer trial) and personal fees from Arla Foods (participated in a systematic review of dietary intake in Nigerian children for this company) outside the submitted work. The other authors made no disclosures.
References
- 1.Gu Q et al. (2019) Diet Quality among Cancer Survivors and Participants without Cancer: A Population-Based, Cross-Sectional Study in the Atlantic Partnership for Tomorrow's Health Project. Nutrients 11 (12):3027. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2.Lei YY et al. (2018) Dietary changes in the first 3 years after breast cancer diagnosis: a prospective Chinese breast cancer cohort study. Cancer Manag Res 10, 4073–4084. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3.Villarreal-Garza C et al. (2012) Impact of diabetes and hyperglycemia on survival in advanced breast cancer patients. Exp Diabetes Res 2012, 732027. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4.Gupta K et al. (2002) Insulin: a novel factor in carcinogenesis. Am J Med Sci 323 (3), 140–5. [DOI] [PubMed] [Google Scholar]
- 5.Ferroni P et al. (2016) Pretreatment Insulin Levels as a Prognostic Factor for Breast Cancer Progression. Oncologist 21 (9), 1041–9. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6.Farvid MS et al. (2021) Postdiagnostic Dietary Glycemic Index, Glycemic Load, Dietary Insulin Index, and Insulin Load and Breast Cancer Survival. Cancer Epidemiol Biomarkers Prev 30 (2), 335–343. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7.Farvid MS et al. (2020) Postdiagnostic Fruit and Vegetable Consumption and Breast Cancer Survival: Prospective Analyses in the Nurses' Health Studies. Cancer Res 80 (22), 5134–5143. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8.Farvid MS et al. (2021) Consumption of sugar-sweetened and artificially sweetened beverages and breast cancer survival. Cancer 127 (15), 2762–2773. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9.Jenkins DJ et al. (1981) Glycemic index of foods: a physiological basis for carbohydrate exchange. Am J Clin Nutr 34 (3), 362–6. [DOI] [PubMed] [Google Scholar]
- 10.Foster-Powell K et al. (2002) International table of glycemic index and glycemic load values: 2002. Am J Clin Nutr 76 (1), 5–56. [DOI] [PubMed] [Google Scholar]
- 11.American Cancer Society. Breast cancer facts&figures 2019-2020. Atlanta (GA): American Cancer Society; 2019. Available from: https://www.cancer.org/content/dam/cancer-org/research/cancer-facts-and-statistics/breast-cancerfacts-and-figures/breast-cancer-facts-and-figures-2019-2020.pdf. [Google Scholar]
- 12.Stanhope KL (2016) Sugar consumption, metabolic disease and obesity: The state of the controversy. Crit Rev Clin Lab Sci 53 (1), 52–67. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13.Allister Price C and Stanhope KL (2016) Understanding the Impact of Added Sugar Consumption on Risk for Type 2 Diabetes. J Calif Dent Assoc 44 (10), 619–26. [PubMed] [Google Scholar]
- 14.Basu S et al. (2013) The relationship of sugar to population-level diabetes prevalence: an econometric analysis of repeated cross-sectional data. PLoS One 8 (2), e57873. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15.Debras C et al. (2020) Total and added sugar intakes, sugar types, and cancer risk: results from the prospective NutriNet-Sante cohort. Am J Clin Nutr 112 (5), 1267–1279. [DOI] [PubMed] [Google Scholar]
- 16.Maino Vieytes CA et al. (2019) Carbohydrate Nutrition and the Risk of Cancer. Curr Nutr Rep 8 (3), 230–239. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17.Tasevska N et al. (2014) Sugars and risk of mortality in the NIH-AARP Diet and Health Study. Am J Clin Nutr 99 (5), 1077–88. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18.Yang Q et al. (2014) Added sugar intake and cardiovascular diseases mortality among US adults. JAMA Intern Med 174 (4), 516–24. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19.Nagata C et al. (2019) Intake of starch and sugars and total and cause-specific mortality in a Japanese community: the Takayama Study. Br J Nutr 122 (7), 820–828. [DOI] [PubMed] [Google Scholar]
- 20.Liu ZM et al. (2018) Dietary sugar intake was associated with increased body fatness but decreased cardiovascular mortality in Chinese elderly: an 11-year prospective study of Mr and Ms OS of Hong Kong. Int J Obes (Lond) 42 (4), 808–816. [DOI] [PubMed] [Google Scholar]
- 21.Ramne S et al. (2019) Association between added sugar intake and mortality is nonlinear and dependent on sugar source in 2 Swedish population-based prospective cohorts. Am J Clin Nutr 109 (2), 411–423. [DOI] [PubMed] [Google Scholar]
- 22.Willett W Implications of total energy intake for epidemiologic analyses. Willett WC, editor. Nutritional Epidemiology: Oxford University Press; 2013. p. 261–83. [Google Scholar]
- 23.Tamimi RM et al. (2008) Comparison of molecular phenotypes of ductal carcinoma in situ and invasive breast cancer. Breast Cancer Res 10 (4), R67. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24.Collins LC et al. (2008) Comparison of estrogen receptor results from pathology reports with results from central laboratory testing. J Natl Cancer Inst 100 (3), 218–21. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 25.Oh H et al. (2016) Expression of estrogen receptor, progesterone receptor, and Ki67 in normal breast tissue in relation to subsequent risk of breast cancer. NPJ Breast Cancer 2, 16032-. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 26.Wang J et al. (2015) Alcohol Consumption and Risk of Breast Cancer by Tumor Receptor Expression. Horm Cancer 6 (5-6), 237–46. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 27.Klement RJ and Kammerer U (2011) Is there a role for carbohydrate restriction in the treatment and prevention of cancer? Nutr Metab (Lond) 8, 75. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 28.Meyerhardt JA et al. (2012) Dietary glycemic load and cancer recurrence and survival in patients with stage III colon cancer: findings from CALGB 89803. J Natl Cancer Inst 104 (22), 1702–11. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 29.Song M et al. (2018) Low-Carbohydrate Diet Score and Macronutrient Intake in Relation to Survival After Colorectal Cancer Diagnosis. JNCI Cancer Spectr 2 (4), pky077. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 30.Fernandes G et al. (2005) Glycemic index of potatoes commonly consumed in North America. J Am Diet Assoc 105 (4), 557–62. [DOI] [PubMed] [Google Scholar]
- 31.Chu YF et al. (2002) Antioxidant and antiproliferative activities of common vegetables. J Agric Food Chem 50 (23), 6910–6. [DOI] [PubMed] [Google Scholar]
- 32.Taskinen MR et al. (2019) Dietary Fructose and the Metabolic Syndrome. Nutrients 11 (9):1987. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 33.Jiralerspong S and Goodwin PJ (2016) Obesity and Breast Cancer Prognosis: Evidence, Challenges, and Opportunities. J Clin Oncol 34 (35), 4203–4216. [DOI] [PubMed] [Google Scholar]
- 34.Lee KN et al. (2020) Type 2 diabetes, breast cancer specific and overall mortality: Associations by metformin use and modification by race, body mass, and estrogen receptor status. PLoS ONE 15 (5), e0232581. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 35.Picon-Ruiz M et al. (2017) Obesity and adverse breast cancer risk and outcome: Mechanistic insights and strategies for intervention. CA Cancer J Clin 67 (5), 378–397. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 36.Gambaro SE et al. (2018) M1 macrophage subtypes activation and adipocyte dysfunction worsen during prolonged consumption of a fructose-rich diet. J Nutr Biochem 61, 173–182. [DOI] [PubMed] [Google Scholar]
- 37.Pektas MB et al. (2016) Dietary Fructose Activates Insulin Signaling and Inflammation in Adipose Tissue: Modulatory Role of Resveratrol. Biomed Res Int 2016, 8014252. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 38.Catena C et al. (2003) Cellular mechanisms of insulin resistance in rats with fructose-induced hypertension. Am J Hypertens 16 (11 Pt 1), 973–8. [DOI] [PubMed] [Google Scholar]
- 39.Darmadi-Blackberry I et al. (2004) Legumes: the most important dietary predictor of survival in older people of different ethnicities. Asia Pac J Clin Nutr 13 (2), 217–20. [PubMed] [Google Scholar]
- 40.Papandreou C et al. (2019) Legume consumption and risk of all-cause, cardiovascular, and cancer mortality in the PREDIMED study. Clin Nutr 38 (1), 348–356. [DOI] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.