Antioxidant supplement use after breast cancer diagnosis and mortality in the Life After Cancer Epidemiology (LACE) cohort

Authors


  • The content of this article is solely the responsibility of the authors and does not necessarily represent the official views of the National Cancer Institute.

Abstract

BACKGROUND:

There is concern that antioxidant supplement use during chemotherapy and radiation therapy may decrease treatment effects, yet the effects of such supplements on recurrence and survival are largely unknown.

METHODS:

The authors prospectively examined the associations between antioxidant use after breast cancer (BC) diagnosis and BC outcomes in 2264 women in the Life After Cancer Epidemiology (LACE) cohort. The cohort included women who were diagnosed with early stage, primary BC from 1997 to 2000 who enrolled, on average, 2 years postdiagnosis. Baseline data were collected on antioxidant supplement use since diagnosis and other factors. BC recurrence and mortality were ascertained, and hazard ratios (HRs) and 95% confidence intervals (CIs) were estimated using delayed entry Cox proportional hazards models. All tests of statistical significance were 2-sided.

RESULTS:

Antioxidant supplement use after diagnosis was reported by 81% of women. Among antioxidant users, frequent use of vitamin C and vitamin E was associated with a decreased risk of BC recurrence (vitamin C: HR, 0.73; 95% CI, 0.55-0.97; vitamin E: HR, 0.71; 95% CI, 0.54-0.94); and vitamin E use was associated with a decreased risk of all-cause mortality (HR, 0.76; 95% CI, 0.58-1.00). Conversely, frequent use of combination carotenoids was associated with increased risk of death from BC (HR, 2.07; 95% CI, 1.21-3.56) and all-cause mortality (HR, 1.75; 95% CI, 1.13-2.71).

CONCLUSIONS:

Frequent use of vitamin C and vitamin E in the period after BC diagnosis was associated with a decreased likelihood of recurrence, whereas frequent use of combination carotenoids was associated with increased mortality. The effects of antioxidant supplement use after diagnosis likely differ by type of antioxidant. Cancer 2012. © 2011 American Cancer Society.

INTRODUCTION

Antioxidant supplement use after a breast cancer diagnosis is common,1 but long-term effects are poorly understood.2 Concerns have been raised about antioxidant supplementation during cancer treatment; such supplementation may possibly protect tumor cells from the pro-oxidant effects of chemotherapy and radiation,3 yet there are limited data on recurrence and survival outcomes in women with breast cancer. Previous studies have had important limitations, including a lack of details on antioxidant dose and frequency of use, retrospective data collection, poorly matched control groups, small sample sizes, and limited duration of follow-up.4-8 A recent report from a well designed, prospective breast cancer cohort study suggests that there is no harm from supplementation with multivitamins, vitamin C, and vitamin E in the period after diagnosis and a possible benefit from supplementation in improving recurrence and survival rates.9

We used data from the Life After Cancer Epidemiology (LACE) Study, a prospective cohort study of women with early stage breast cancer, to examine the association between antioxidant use in the 2-year period after diagnosis and breast cancer outcomes, including all-cause mortality, death from breast cancer, and recurrence. Our working hypothesis was that different forms of antioxidant supplements would have differing effects on outcomes.

MATERIALS AND METHODS

Study Participants

The LACE cohort (n = 2264) consists of women primarily recruited from the Kaiser Permanente Northern California (KPNC) (83%) and the Utah (12%) cancer registries who were diagnosed with early stage primary breast cancer between 1997 and 2000. Women enrolled between 2000 and 2002, approximately 1 to 3 years after a breast cancer diagnosis (average, 1.9 years; range, 0.9-3.2 years). Eligibility criteria included a diagnosis of early stage, primary breast cancer (stage I [≥1 cm], stage II, or stage IIIA); ages 18 to 79 years at diagnosis; completion of breast cancer treatment, including surgery, chemotherapy, and radiation therapy (use of hormone therapy was permitted); no evidence of recurrent disease; and no history of other cancers in the 5 years before enrollment. LACE recruitment and data-collection methods have been described elsewhere.10 The study was approved by the institutional review boards of KPNC and the University of Utah.

Data Collection

At enrollment, a mailed, self-administered questionnaire collected detailed data on demographics, medical history, medication use, reproductive history, family history of breast cancer, anthropometric measures (height, weight, waist circumference), weight history, diet, dietary supplement use, physical activity, quality of life, and depressive symptoms. Detailed questionnaire data were collected on the use of antioxidant supplements. Specific questions were asked about the use of multivitamins with and without minerals, supplements containing combinations of multiple carotenoids (beta-carotene, lycopene, lutein, etc), as well as the following individual supplements: beta-carotene, lycopene, vitamin C, vitamin E, selenium, and zinc. For each supplement, questions were asked about any use since diagnosis (yes/no), use in the 5 years before diagnosis (yes/no), and frequency of use in the period between breast cancer diagnosis and study enrollment (<1 day per week, 1-2 days per week, 3-5 days per week, or 6-7 days per week).

Clinical data were obtained using KPNC electronic databases or medical chart review. Data included tumor size, number of positive lymph nodes, tumor hormone receptor status, and treatments received, including surgery, chemotherapy, radiation therapy, and hormone therapy. Tumor stage was determined based on American Joint Committee on Cancer (AJCC) criteria according to the AJCC Cancer Staging Manual (fourth edition).

Outcome Ascertainment

Breast cancer recurrence and death outcomes were ascertained through November 2010 by semiannual mailed questionnaires or annual mailed questionnaires (after April 2005, when the average follow-up was 5 years) that asked participants to report any major health events in the preceding 6 months or 12 months. Nonrespondents were contacted by telephone. Reported outcomes were verified by a medical record review. Participant deaths were ascertained by KPNC electronic databases and by family members responding to mailed questionnaires and telephone calls. Death certificates were obtained to verify primary and underlying causes of death based on International Classification of Diseases, Ninth Revision (ICD-9) codes. Trained abstractors performed the initial reviews, and a physician reviewer was consulted if the cause of death was unclear. For these analyses, breast cancer recurrence was defined as a local/regional cancer recurrence, distant recurrence/metastasis, or the development of a contralateral primary breast cancer. Death from breast cancer was defined as death attributable to breast cancer as the primary or underlying cause based on the death certificate ICD-9 code. All-cause mortality was defined as death from any cause, including breast cancer.

Variable Definitions

A 3-level exposure variable for use of each antioxidant in the period between diagnosis and study enrollment was created with the following categories: no use; occasional use (<1-5 days per week); and frequent use (6-7 days per week). A comorbidity score was calculated using a modified approach to the Charlson comorbidity index.11 Self-reported data that contributed to the score included angina, myocardial infarction, other heart problems, stroke, insulin-dependent diabetes, noninsulin-dependent diabetes, other cancer diagnosis, cirrhosis, peripheral arterial disease, other kidney disease, ulcer, and lupus. Each condition contributed 1 point to the score with a possible range of 0 to 12 points.

Statistical Analyses

The primary goal of the analyses was to relate use of antioxidant supplements in the period after breast cancer diagnosis with subsequent recurrence and death. Delayed-entry Cox proportional hazards models were used to estimate hazard ratios (HRs) and 95% confidence intervals (CIs) for the association between use of individual antioxidant supplements and breast cancer outcomes, with time since diagnosis defined as the timescale. Follow-up began at the date of study entry and ended at the date of first confirmed breast cancer recurrence or the date of death, depending on the specific analysis. Individuals who did not have an event were censored at the date of last contact. All tests of statistical significance were 2-sided and were considered statistically significant at an α level of .05. Initial analyses dichotomized antioxidant use into ever use and never use. To address possible biases related to differences between antioxidant users and nonusers, subsequent analyses were restricted to women who used at least 1 form of antioxidant supplement.12

Covariates that were selected a priori for inclusion in the multivariable adjusted models were age at diagnosis (continuous), race/ethnicity (non-Hispanic white, black, Hispanic, Asian/Pacific Islander, other), education (less than high school, high school graduate/some college, college graduate), AJCC disease stage (I, II, IIIA), number of positive lymph nodes (0, 1-3, ≥4), tumor hormone receptor status (estrogen receptor [ER]-negative and/or progesterone receptor [PR]-positive, ER-negative and PR-negative), chemotherapy received (yes or no), radiation therapy received (yes or no), hormone therapy received (yes or no), body mass index 1 year before diagnosis (<25 kg/m2, 25 to <30 kg/m2, or ≥30 kg/m2), smoking history at enrollment (never, former, current), alcohol consumption at enrollment (≤0.5 g/day, >0.5 to <6 g/day, ≥6 g/day), physical activity at enrollment (nonsedentary metabolic equivalent-hours per week; continuous), daily servings of fruit/vegetable intake at enrollment (continuous), and an enrollment comorbidity score. Additional covariates that were considered for model inclusion were total caloric intake and antioxidant use in the 5 years before diagnosis; however, these were not retained in the final models, because their inclusion did not substantially change the effect estimates. Tests for interaction were performed between treatment received (chemotherapy, radiation therapy, hormone therapy) and use of specific antioxidant supplements. Subgroup analyses, defined a priori, were performed based on types of treatment received. Stratified analyses were performed to examine whether smoking was an effect modifier of the association between antioxidant supplementation and breast cancer outcomes. Analyses were conducted using the SAS statistical software package (version 9.1.3; SAS Institute Inc., Cary, NC).

RESULTS

Participant Characteristics

Among 2264 women, the average age was 58.3 years (range, 25.5-79.9 years) and 80% of women were non-Hispanic white. On average, women were enrolled 1.9 years after breast cancer diagnosis, at which time most had completed chemotherapy and/or radiation therapy and had initiated hormone therapy. The majority of women (80.3%) were diagnosed with stage I or IIA breast cancer; and 57.2% received chemotherapy, 63% received radiation therapy, and 80.4% received hormone therapy. Over 10 years of study follow-up (mean ± standard deviation: 8.3 ± 2.4 years), 375 of the women had recurrences of breast cancer, and we observed 393 deaths, including 214 deaths from breast cancer.

Antioxidant Supplement Use

Of the 2264 women, 1829 (81%) used 1 or more antioxidant-containing supplements. The majority of women (70%) used a multivitamin supplement, all of which contain 1 or more antioxidant vitamins or minerals. Of the multivitamin supplement users, most (69%) also used 1 or more other supplements that contained antioxidants. Aside from multivitamins, other commonly reported antioxidant supplements among the 2264 women were vitamin C (40%), vitamin E (48%), zinc (10%), selenium (7%), combination carotenoids (7%), beta-carotene alone (6%), and lycopene alone (1%). Compared with women who were not using any antioxidant supplements, women who reported using antioxidant supplements were older, more likely to be non-Hispanic white, had higher education, had lower body mass index, were less likely to have smoked, engaged in more physical activity, ate more fruits and vegetables, and had a lower comorbidity score (Table 1).

Table 1. Baseline Demographics, Tumor, and Lifestyle Characteristics of Life After Cancer Epidemiology (LACE) Cohort Participants by Use of Any Antioxidants Since Diagnosis (n = 2264)
 Antioxidant Use Since Diagnosis: No. of Women (%)a 
CharacteristicNo Use, n = 425Occasional Use, n = 335bFrequent Use, n = 1494cP
  • Abbreviations: BMI, body mass index; ER, estrogen receptor; MET, metabolic equivalents; PR, progesterone receptor; SD, standard deviation.

  • a

    Use of antioxidants since diagnosis is defined as the use of any of the following in the period between diagnosis and study enrollment: multivitamins, combination carotenoids, vitamin C, vitamin E, beta-carotene, lycopene, selenium, and zinc.

  • b

    Occasional use is defined as using antioxidant supplements 1 to 5 days per week in the period between diagnosis and study enrollment.

  • c

    Frequent use is defined as using antioxidant supplements 6 or 7 days per week in the period between diagnosis and study enrollment.

Age at diagnosis, y    
 ≤50128 (30.1)115 (34.3)305 (20.4)<.001
 >50297 (69.9)220 (65.7)1189 (79.6) 
Race/ethnicity    
 Non-Hispanic white306 (72.7)251 (74.9)1243 (83.3)<.001
 Black24 (5.7)21 (6.3)66 (4.4) 
 Hispanic40 (9.5)25 (7.5)74 (5) 
 Asian/Pacific Islander41 (9.7)24 (7.2)64 (4.3) 
 Other10 (2.4)14 (4.2)45 (3) 
Education    
 Less than high school32 (7.6)19 (5.7)72 (4.8).001
 High school graduate/some college275 (65.3)197 (59.2)856 (57.4) 
 College graduate114 (27.1)117 (35.1)563 (37.8) 
Stage    
 I210 (49.4)146 (43.6)693 (46.4).75
 IIA136 (32)121 (36.1)504 (33.8) 
 IIB67 (15.8)55 (16.4)252 (16.9) 
 IIIA12 (2.8)13 (3.9)44 (2.9) 
No. of positive lymph nodes    
 0272 (64.6)207 (62.2)938 (63.2).96
 1-3106 (25.2)91 (27.3)396 (26.7) 
 ≥443 (10.2)35 (10.5)151 (10.2) 
ER/PR status    
 ER-positive and/or PR-positive360 (86.1)268 (81)1253 (84.7).14
 ER-negative and PR-negative58 (13.9)63 (19)227 (15.3) 
Treatment    
 Chemotherapy238 (56)208 (62.1)843 (56.6).15
 Radiation261 (61.4)227 (67.8)931 (62.3).13
 Hormone therapy338 (80.1)256 (77.6)1220 (81.8).19
BMI 1 y before diagnosis, kg/m2    
 <25171 (41)136 (41.5)716 (48.3).01
 25 to <29124 (29.7)107 (32.6)433 (29.2) 
 ≥30122 (29.3)85 (25.9)334 (22.5) 
Smoking status    
 Never204 (48.3)201 (60)785 (52.6).001
 Former171 (40.5)109 (32.5)607 (40.7) 
 Current47 (11.1)25 (7.5)100 (6.7) 
Alcohol consumption, g/d    
 ≤0.5167 (51.5)141 (51.3)626 (48.7).78
 0.5-6.080 (24.7)70 (25.5)323 (25.1) 
 ≥6.077 (23.8)64 (23.3)336 (26.1) 
Antioxidant use 5 y before diagnosis    
 Multivitamins31 (8.8)163 (52.2)929 (65.6)<.0001
 Vitamin C14 (4.1)87 (33)521 (44.8)<.0001
 Vitamin E12 (3.5)71 (27.5)572 (45.4)<.0001
Comorbidity score: Mean±SD0.11±0.380.06±0.270.07±0.31.09
Nonsedentary MET-h/wk: Mean±SD45.5±29.551.91±31.0953.42±32.51<.001
Daily fruit/vegetable servings: Mean±SD3.6±2.23.9±2.114.36±2.41<.001

Antioxidant Use and Breast Cancer Outcomes Among All Participants

All-cause mortality

Compared with nonusers of vitamin C, frequent users of vitamin C tended toward a lower risk of all-cause mortality (HR, 0.78; 95% CI, 0.61-1.00) (Table 2). Compared with nonusers of vitamin E, frequent users of vitamin E had significantly lower all-cause mortality (HR, 0.75; 95% CI, 0.59-0.96). In contrast, use of combination carotenoids was associated with significantly increased risk of all-cause mortality versus no use (HR, 1.63; 95% CI, 1.06-2.50). Relatively few women reported using individual carotenoids that were not part of a combination supplement; and, although the CIs associated with individual carotenoids were wide, the point estimates suggested increased risk of mortality (beta-carotene: HR, 1.18; 95% CI, 0.71-1.97; lycopene: HR, 1.38; 95% CI, 0.41-4.61).

Table 2. Associations Between Antioxidant Use Since Diagnosis and Breast Cancer Recurrence, Death From Breast Cancer, and Death From All Causes Among All Participants (n = 2264)
  Death From All CausesDeath From Breast CancerBreast Cancer Recurrence
AntioxidantNo. of WomenNo. of EventsHR (95% CI)aPPTrendNo. of EventsHR (95% CI)aPPTrendNo. of EventsHR (95% CI)aPPTrend
  • Abbreviations: CI, confidence interval; HR, hazard ratio; Ref, referent category.

  • a

    Multivariable HR models were conducted separately for each type of antioxidant supplement and were adjusted for the following potential confounders: age at diagnosis, race/ethnicity, education, breast cancer stage at diagnosis, number positive lymph nodes, tumor hormone receptor status, chemotherapy received, radiation therapy received, hormone therapy received, body mass index 1 year before diagnosis, smoking history at enrollment, alcohol consumption at enrollment, physical activity at enrollment, daily servings of fruits and vegetables at enrollment, and comorbidity score at enrollment.

Multivitamins    .18   .21   .13
 No use520991.00 (Ref)  511.00 (Ref)  941.00 (Ref)  
 Occasional use256370.83 (0.56-1.22).35 220.75 (0.45-1.25).27 480.91 (0.64-1.3).61 
 Frequent use10511800.84 (0.65-1.08).16 950.79 (0.56-1.12).18 1710.82 (0.63-1.06).13 
Vitamin C alone    .05   .25   <.01
 No use10721981.00 (Ref)  1031.00 (Ref)  2031.00 (Ref)  
 Occasional use192250.78 (0.51-1.18).24 160.84 (0.49-1.43).52 300.78 (0.53-1.14).2 
 Frequent use540890.78 (0.61-1.01).06 480.82 (0.58-1.16).26 790.71 (0.54-0.92).01 
Vitamin E alone    .02   .34   <.01
 No use9181741.00 (Ref)  881.00 (Ref)  1751.00 (Ref)  
 Occasional use161270.88 (0.58-1.32).53 181.08 (0.64-1.81).78 351.08 (0.75-1.57).67 
 Frequent use7131130.75 (0.59-0.96).02 620.85(0.61-1.18).33 1020.7 (0.54-0.9)<.01 
Combination carotenoids    .04   .03   .52
 No use16982841.00 (Ref)  1471.00 (Ref)  2861.00 (Ref)  
 Occasional use3360.95 (0.41-2.19).91 30.77 (0.24-2.48).66 60.79 (0.34-1.8).57 
 Frequent use89241.63 (1.06-2.5).03 161.93 (1.14-3.28).02 191.23 (0.76-1.96).40 
Beta-carotene alone    .41   .34   .90
 No use17262951.00 (Ref)  1571.00 (Ref)  2971.00 (Ref)  
 Occasional use2141.65 (0.61-4.46).33 21.56 (0.38-6.4).54 51.7 (0.7-4.15).25 
 Frequent use71161.18 (0.71-1.97).52 101.33 (0.69-2.55).39 120.89 (0.5-1.6).70 
Lycopene alone    .46   .15   .67
 No use18043121.00 (Ref)  1651.00 (Ref)  3091.00 (Ref)  
 Occasional use412.82 (0.39-20.67).31 14.84 (0.64-36.4).13 12.09 (0.29-15.22).47 
 Frequent use1031.38 (0.41-4.61).60 32.09 (0.59-7.43).25 31.17 (0.35-3.89).80 
Selenium alone    .65   .87   .75
 No use16872981.00 (Ref)  1561.00 (Ref)  2911.00 (Ref)  
 Occasional use3471.72 (0.8-3.7).16 41.33 (0.48-3.67).58 71.16 (0.54-2.48).70 
 Frequent use91130.80 (0.45-1.41).43 90.9 (0.45-1.79).76 160.89 (0.53-1.49).66 
Zinc alone    .29   .49   .26
 No use16422891.00 (Ref)  1531.00 (Ref)  2871.00 (Ref)  
 Occasional use51101.11 (0.58-2.12).74 40.83 (0.31-2.27).72 70.72 (0.34-1.54).40 
 Frequent use113180.75 (0.46-1.21).24 110.82 (0.44-1.53).54 180.79 (0.49-1.28).34 

No statistically significant associations were observed between occasional use of antioxidant supplements after diagnosis and any of the breast cancer outcomes (Table 2). Similarly, no statistically significant associations were observed between use of multivitamins, selenium, and zinc and any of the breast cancer outcomes (Table 2).

Death from breast cancer

Analyses were repeated to examine the associations between antioxidant supplement use and the risk of death from breast cancer. Associations were in the same direction as those for all-cause mortality, although the only associations that reached statistical significance were for combination carotenoids (Table 2). Among all 2264 women, compared with nonusers, frequent users of combination carotenoids had increased risk of death from breast cancer (HR, 1.93; 95% CI, 1.14-3.28).

Breast cancer recurrence

Analyses were repeated to examine the associations between antioxidant supplement use and breast cancer recurrence (Table 2). Again, associations were in the same direction as those for all-cause mortality. Among all 2264 participants, frequent use of vitamin C and vitamin E were associated with a lower risk of recurrence versus no vitamin C use (HR, 0.70; 95% CI, 0.54-0.92) and versus no vitamin E use (HR, 0.70; 95% CI, 0.54-0.90), and the use of combination carotenoids was associated with a suggestive increased risk of recurrence versus no use (HR, 1.23; 95% CI, 0.76-1.96).

Antioxidant Use and Breast Cancer Outcomes Among Antioxidant Users

To control for any possible unmeasured confounding between antioxidant users and nonusers, the analyses presented in Table 2 were rerun by restricting the group analyzed to the 1829 women who reported using 1 or more antioxidants after diagnosis (Table 3). The results followed the same directions as those presented in Table 2. Vitamin C and vitamin E tended to be associated with decreased mortality, and frequent use of combination carotenoids was associated with increased mortality compared with no use (HR, 1.75; 95% CI, 1.13-2.71). Frequent use of combination carotenoids was associated with increased risk of dying from breast cancer compared with no use (HR, 2.07; 95% CI, 1.21-3.56). Frequent use of vitamin C and vitamin E was associated with a lower risk of breast cancer recurrence compared with no vitamin C use (HR, 0.73; 95% CI, 0.55-0.97) and no vitamin E use (HR, 0.71; 95% CI, 0.54-0.94).

Table 3. Associations Between Antioxidant Use Since Diagnosis and Death From All Causes, Death From Breast Cancer, and Breast Cancer Recurrence Among All Antioxidant Users (n = 1829)
  Death From All CausesDeath From Breast CancerBreast Cancer Recurrence
AntioxidantNo. of WomenNo. of EventsHR (95% CI)aPPTrendNo. of EventsHR (95% CI)aPPTrendNo. of EventsHR (95% CI)aPPTrend
  • Abbreviations: CI, confidence interval; HR, hazard ratio; Ref, referent category.

  • a

    Multivariable HR models were conducted separately for each type of antioxidant supplement and were adjusted for the following potential confounders: age at diagnosis, race/ethnicity, education, breast cancer stage at diagnosis, number positive lymph nodes, tumor hormone receptor status, chemotherapy received, radiation therapy received, hormone therapy received, body mass index 1 year before diagnosis, smoking history at enrollment, alcohol consumption at enrollment, physical activity at enrollment, daily servings of fruits and vegetables at enrollment, and comorbidity score at enrollment.

Multivitamins    .95   .50   .43
 No use199331.00 (Ref)  191.00 (Ref)  331.00 (Ref)  
 Occasional use256360.95 (0.59-1.55).84 220.78 (0.41-1.47).44 471.04 (0.66-1.64).88 
 Frequent use10511760.98 (0.67-1.42).90 940.81 (0.49-1.33).40 1680.90 (0.62-1.31).58 
Vitamin C alone    .15   .44   .02
 No use7521301.00 (Ref)  701.00 (Ref)  1391.00 (Ref)  
 Occasional use192250.80 (0.52-1.23).31 160.88 (0.51-1.52).63 290.78 (0.52-1.16).22 
 Frequent use540870.82 (0.62-1.08).17 480.87 (0.60-1.26).45 790.73 (0.55-0.97).03 
Vitamin E alone    .05   .60   .02
 No use5981071.00 (Ref)  551.00 (Ref)  1121.00 (Ref)  
 Occasional use161270.86 (0.56-1.33).50 181.13 (0.65-1.95).67 351.12 (0.76-1.65).56 
 Frequent use7131100.76 (0.58-1.00).05 620.91 (0.63-1.32).61  0.71 (0.54-0.94).02 
Combination carotenoids    .01   .02   .32
 No use13762131.00 (Ref)  1141.00 (Ref)  2211.00 (Ref)  
 Occasional use3361.04 (0.45-2.42).93 30.77 (0.24-2.52).67 60.84 (0.36-1.94).68 
 Frequent use89241.75 (1.13-2.71).01 162.07 (1.21-3.56).01 191.33 (0.83-2.15).24 
Beta-carotene alone    .42   .24   .91
 No use14042261.00 (Ref)  1241.00 (Ref)  2321.00 (Ref)  
 Occasional use2141.80 (0.66-4.91).25 21.70 (0.42-6.98).12 51.87 (0.76-4.58).17 
 Frequent use71151.18 (0.69-2.00).55 101.44 (0.74-2.78).24 120.95 (0.53-1.71).86 
Lycopene alone    .41   .14   .59
 No use14832421.00 (Ref)  1321.00 (Ref)  2451.00 (Ref)  
 Occasional use413.24 (0.44-24.02).25 15.03 (0.66-38.31).49 12.13 (0.29-15.68).46 
 Frequent use1031.44 (0.43-4.87).56 32.16 (0.60-7.77).69 31.25(0.37-4.22).72 
Selenium alone    .73   .81   .87
 No use13652281.00 (Ref)  1241.00 (Ref)  2271.00 (Ref)  
 Occasional use3471.78 (0.82-3.85).14 41.43 (0.52-3.97).72 71.23 (0.57-2.65).59 
 Frequent use91130.82 (0.46-1.45).50 90.87 (0.43-1.74).64 160.92 (0.55-1.55).76 
Zinc alone    .50   .59   .34
 No use13222191.00 (Ref)  1211.00 (Ref)  2231.00 (Ref)  
 Occasional use51101.25 (0.65-2.41).50 40.83 (0.30-2.29).72 70.73 (0.34-1.57).42 
 Frequent use113180.80 (0.50-1.31).38 110.86 (0.46-1.61).64 180.82 (0.51-1.34).44 

Antioxidant Use and Breast Cancer Outcomes by Treatment Type

To examine whether associations between antioxidant supplement use and breast cancer outcomes varied by type of breast cancer treatment, we repeated the analyses within groups defined by treatments received (chemotherapy, radiation therapy, and hormone therapy) (Table 4). The associations that were observed within each subgroup had similar directions of effects as in the overall analyses.

Table 4. Associations Between Antioxidant Use Since Diagnosis and Death From All Causes, Death From Breast Cancer, and Breast Cancer Recurrence Among All Antioxidant Users Who Received Treatment
  Death From All CausesDeath From Breast CancerBreast Cancer Recurrence
AntioxidantNo. of WomenNo. of EventsHR (95% CI)aPNo. of EventsHR (95% CI)aPNo. of EventsHR (95% CI)aP
  • Abbreviations: CI, confidence interval; HR, hazard ratio; Ref, referent category.

  • a

    HRs and CIs were calculated for frequent use versus no use, using no use as the referent group. Multivariable HR models were conducted separately for each type of antioxidant supplement and were adjusted for the following potential confounders: age at diagnosis, race/ethnicity, education, breast cancer stage at diagnosis, number positive lymph nodes, tumor hormone receptor status, body mass index 1 year before diagnosis, smoking history at enrollment, alcohol consumption at enrollment, physical activity at enrollment, daily servings of fruits and vegetables at enrollment, and comorbidity score at enrollment. Within each treatment group, the models also were controlled for other treatments received (chemotherapy, radiation therapy, hormone therapy).

Women who received chemotherapy, n = 1051          
 Vitamin C alone2892890.73 (0.49-1.08).12300.74 (0.47-1.18).21510.74 (0.52-1.05).09
 Vitamin E alone3783780.85 (0.58-1.25).42400.86 (0.55-1.34).50650.79 (0.56-1.12).18
 Carotenoids51512.09 (1.21-3.61).01132.54 (1.37-4.70)<.01151.66 (0.96-2.88).07
Women who received radiation therapy, n = 1158          
 Vitamin C alone334450.69 (0.47-1.00).05280.68 (0.42-1.09).11450.60 (0.42-0.86).01
 Vitamin E alone442590.69 (0.48-0.99).04400.86 (0.55-1.36).53630.70 (0.49-0.98).04
 Carotenoids50142.14 (1.20-3.82).01102.54 (1.28-5.05).01111.37 (0.73-2.57).32
Women who received hormonal therapy, n = 1476          
 Vitamin C alone435710.83 (0.61-1.13).23380.94 (0.61-1.43).76600.72 (0.52-0.99).04
 Vitamin E alone597880.68 (0.50-0.92).01480.84 (0.55-1.29).42810.70 (0.51-0.96).03
 Carotenoids70181.66 (1.00-2.73).05122.14 (1.16-3.97).02141.31 (0.75-2.27).35

All-cause mortality

Among patients who received chemotherapy and radiation therapy, the strength of the association between frequent use of combination carotenoids and all-cause mortality was increased relative to the overall analyses versus no use during chemotherapy (HR, 2.09; 95% CI, 1.21-3.61) and versus no use during radiation therapy (HR, 2.14; 95% CI, 1.20-3.82), although the CIs were wide. Among women who received hormone therapy, the strength of the association between the use of combination carotenoids and mortality remained high and had borderline statistical significance (HR, 1.66; 95% CI, 1.00-2.73). Similarly, relative to the overall analyses, a stronger protective effect of vitamin E was observed among women who received radiation therapy and hormone therapy. No interactions were observed between treatment type and the use of specific antioxidant supplements in any analyses (data not shown).

Death from breast cancer

Relative to the overall analyses, strengthened associations were observed between frequent use of combination carotenoids and death from breast cancer among women who received chemotherapy, radiation therapy and hormone therapy versus no use during chemotherapy (HR, 2.54; 95% CI, 1.37-4.70), versus no use during radiation therapy (HR, 2.54; 95% CI, 1.28-5.05), and versus no use during hormone therapy (HR, 2.14; 95% CI, 1.16-3.97).

Breast cancer recurrence

Among women who received radiation and/or hormone therapy, slightly strengthened associations were observed relative to the overall analyses between frequent use of vitamins C and E and decreased risk of recurrence versus no use during radiation therapy (vitamin C: HR, 0.60; 95% CI, 0.42-0.86; vitamin E: HR, 0.70; 95% CI, 0.49-0.98) and versus no use during hormone therapy (vitamin C: HR, 0.72; 95% CI, 0.52-0.99; vitamin E: HR, 0.70; 95% CI, 0.51-0.96). Among women who received chemotherapy, there was a trend toward increased recurrence among frequent users of multiple carotenoids versus no use (HR, 1.66; 95% CI, 0.96-2.88).

Effect of Smoking

We repeated the analyses in subgroups defined by smoking status (ever/never). The associations in ever-smokers and never-smokers were similar to the associations observed in the overall analyses (data not shown).

DISCUSSION

In a prospective cohort study of antioxidant supplement use after breast cancer diagnosis, we report several findings. First, a large proportion of recent breast cancer survivors consumed antioxidant-containing multivitamins and individual supplements in the early (on average, 2 years) postdiagnosis period, which encompasses breast cancer treatment. Second, in the 10 years after breast cancer diagnosis, women who frequently used combination carotenoids after diagnosis had an approximately 2-fold greater risk of all-cause mortality compared with nonusers. This association was observed even after adjustment for clinical and behavioral prognostic risk factors and conventional cancer therapies that may have been potential confounders. Third, frequent users of vitamin C and vitamin E supplements after diagnosis tended to have more favorable outcomes, including decreased risk of breast cancer recurrence and mortality.

In these analyses, we observed that frequent use of combination carotenoids in the period after diagnosis was associated with an increased risk of death from breast cancer and from all causes, but not with an increased risk of breast cancer recurrence. The mechanism through which combination carotenoids may cause death is unclear, and at least 3 scenarios are possible. It is possible that carotenoids may protect cancer cells against oxidative damage during treatment and, thus, decrease the effects of treatment, resulting in increased breast cancer progression.3, 13-16 It is also possible that the increased risk of death is not related to use of carotenoid supplements during treatment but that carotenoid supplementation in general places a woman at increased risk of breast cancer progression. In the randomized trials for lung cancer prevention, beta-carotene appeared to promote existing tumors rather than initiate tumors, as evidenced by the rapid increase in lung cancer among the group that received beta-carotene. After beta-carotene cessation in the Alpha-Tocopherol, Beta-Carotene Prevention (ATBC) Study, there was a compensatory decrease in incidence such that, after 10 years of follow-up, there was no overall increase in incidence among the beta-carotene group.17 Finally, carotenoid supplementation may place women at increased risk of nonbreast cancer death. The Beta-Carotene and Retinol Efficacy Trial (CARET), a trial of beta-carotene plus retinyl palmitate among individuals at high lung cancer risk, reported that, 6 years after the study was closed because of harm, women who received the intervention continued to be at increased risk of death from cardiovascular disease (44%), lung cancer (33%), and all causes (37%).18 Although the mechanism by which carotenoid supplementation may increase mortality is not clear, our results suggest that there may be cause for concern with carotenoid supplementation in the period after a breast cancer diagnosis and during treatment.

We also observed that frequent use of vitamins C and E after diagnosis was associated with reduced risk of all-cause mortality, death from breast cancer, and breast cancer recurrence. Our findings are similar to those recently reported from the Shanghai Breast Cancer Survival Study (SBCSS).9 Nechuta et al examined the associations between use of antioxidants in the 6 months after breast cancer diagnosis and observed that women who used any form of antioxidants (vitamin E, vitamin C, multivitamins) had a 22% reduced risk of recurrence and an 18% reduced risk of death. In previous analyses within the LACE cohort, we reported that continuous use of multivitamins with minerals before and after breast cancer diagnosis was associated with a 20% decreased risk of recurrence, a 32% decreased risk of breast cancer death, and a 22% decreased risk of all-cause death.19 The protective mechanisms by which vitamins C and E could reduce breast cancer recurrence and mortality may be through oxidative stress pathways related to cancer and cardiovascular disease, but the precise mechanisms are unclear.

There are important limitations to consider when interpreting these results. First, data on antioxidant supplement use in the period after diagnosis and during treatment were collected at the point of study enrollment, which, on average, was 1.9 years after diagnosis. Recall bias may have affected our findings. Data on specific dates of treatment and specific dates of antioxidant use were not available, and it is possible that patients may have completed their breast cancer treatment before initiating antioxidant supplement use, which could have induced misclassification. Second, we lacked information on specific doses, formulations, and duration of use. Therefore, we have limited ability to comment on the dose threshold for benefit or harm of supplementation with specific antioxidants. We used frequency of use as our metric of exposure intensity. We observed the strongest effects among those who we defined as frequent users; this category identified women who were daily users and, thus, most likely had the highest cumulative exposure. We also lacked specific information on the various formulations of combination carotenoids and do not know which carotenoids conferred the most risk. Third, because this was an observational study, we had limited ability to establish a causal relationship. The tendency for breast cancer survivors to be more health conscious is well established,20, 21 and the healthy user bias may explain the findings related to vitamins C and E. At the same time, this renders our observation of increased mortality with combination carotenoid use even more striking, because it cannot be explained by the healthy user bias. Finally, we were unable to draw conclusions about agents that were not used frequently (beta-carotene, lycopene, selenium, zinc), because we lacked sufficient power to detect differences between groups.

In summary, we examined the association between use of various antioxidant supplements in the period between breast cancer diagnosis and enrollment (average, 1.9 years) into a prospective study of early stage breast cancer survivors on breast cancer outcomes. We hypothesized that outcomes would differ by type of antioxidant supplement and observed that, in fact, outcomes did differ. We observed protective associations between the use of vitamin C and vitamin E and breast cancer recurrence and death from all causes that may be true associations or could be caused by a healthy user bias. However, we observed that the use of combination carotenoids was associated with increased risks of death from breast cancer and of death from all causes that cannot be explained by healthy user bias. To our knowledge, this is the first report from a prospective study suggesting possible harm from using a specific type of antioxidant supplement after a breast cancer diagnosis. Our findings should be considered hypothesis-generating and need to be replicated in other study settings.

Acknowledgements

We thank the Life After Cancer Epidemiology (LACE) study participants for their contributions.

FUNDING SOURCES

This work was supported by grants from the National Cancer Institute (R01CA129059 to B.J.C. and K23CA141052 to H.G.).

CONFLICT OF INTEREST DISCLOSURES

The authors made no disclosures.

Ancillary