Background: Antioxidants may protect normal cells from the oxidative damage that occurs during radiotherapy and certain chemotherapy regimens; however, the same mechanism could protect tumor cells and potentially reduce effectiveness of cancer treatments. We evaluated the association of vitamin supplement use in the first 6 months after breast cancer diagnosis and during cancer treatment with total mortality and recurrence.
Methods: We conducted a population-based prospective cohort study of 4,877 women aged 20 to 75 years diagnosed with invasive breast cancer in Shanghai, China, between March 2002 and April 2006. Women were interviewed approximately 6 months after diagnosis and followed up by in-person interviews and record linkage with the vital statistics registry.
Results: During a mean follow-up of 4.1 years, 444 deaths and 532 recurrences occurred. Vitamin use shortly after breast cancer diagnosis was associated with reduced mortality and recurrence risk, adjusted for multiple lifestyle factors, sociodemographics, and known clinical prognostic factors. Women who used antioxidants (vitamin E, vitamin C, multivitamins) had 18% reduced mortality risk (HR = 0.82, 95% CI: 0.65–1.02) and 22% reduced recurrence risk (HR = 0.78, 95% CI: 0.63–0.95). The inverse association was found regardless of whether vitamin use was concurrent or nonconcurrent with chemotherapy, but was present only among patients who did not receive radiotherapy.
Conclusions: Vitamin supplement use in the first 6 months after breast cancer diagnosis may be associated with reduced risk of mortality and recurrence.
Impact: Our results do not support the current recommendation that breast cancer patients should avoid use of vitamin supplements. Cancer Epidemiol Biomarkers Prev; 20(2); 262–71. ©2010 AACR.
This article is featured in Highlights of This Issue, p. 217
Radiotherapy and certain chemotherapy agents act through various oxidative stress mechanisms to produce free radicals that damage tumor cells (1). Oxidative stress during cancer therapy also harms healthy tissue. Antioxidant supplements may help protect normal cells from oxidative damage and reduce the short- and long-term harmful effects of cancer treatment (1–5). On the other hand, concern has been raised that antioxidant supplements may also protect tumor cells during radiotherapy and chemotherapy, thereby compromising treatment efficacy (1, 5–10). This has resulted in controversy over guidelines for the use of vitamin supplements during cancer treatment (1, 3, 11–13).
Although many investigators and clinicians recommend that vitamin supplements, in particular antioxidants in high doses, should not be used by patients during cancer treatment (1, 11–13), the use of vitamin supplements is common among breast cancer patients (4, 14–17). To our knowledge, no large, prospective cohort study or clinical trial has been conducted to evaluate the influence of vitamin supplement use during breast cancer treatment on long-term outcomes among breast cancer patients. Using data from a prospective cohort study of approximately 5,000 breast cancer survivors, we evaluated the associations of total mortality and breast cancer recurrence with vitamin supplement use following cancer diagnosis and concurrent with cancer treatment.
Materials and Methods
The Shanghai Breast Cancer Survival Study (SBCSS) is a population-based, prospective cohort study of Chinese women diagnosed with breast cancer. Study methods have been previously described (18). Briefly, women newly diagnosed with invasive breast cancer and aged 20 to 75 years were identified from the Shanghai Cancer Registry within approximately 6 months of diagnosis between March 2002 and April 2006. Eligibility criteria included the following: (i) first diagnosis of primary breast cancer; (ii) permanent resident of Shanghai; and (iii) alive at study recruitment. Of the 6,299 eligible cases, 5,042 participated (80.0%) and provided written informed consent. Reasons for nonparticipation included refusal, absent during study enrollment, could not be contacted, and other miscellaneous reasons. For the present study, we excluded women with stage 0 tumors (n = 156) and women who did not have surgery (n = 9), leaving a final sample of 4,877 women for the analyses.
In-person interviews were conducted by trained interviewers, all retired medical professionals, on average 6.5 months (range: 3–11) after diagnosis to obtain information on reproductive history, medical history, selected dietary and other lifestyle factors, complementary and alternative medicine use, sociodemographics, and quality of life. Anthropometric measurements were taken using a standard protocol. Medical records were reviewed for 98.1% of participants to obtain and verify clinical data, including cancer diagnosis, treatment history, and tumor characteristics [e.g., estrogen receptor (ER) and progesterone receptor (PR) status and tumor stage]. The agreement rates between self-reported and medical chart information ranged from 94% to 98%. In-person follow-up surveys have been conducted at 18 months (4,572 completed among survivors), 36 months (4,149 completed among survivors), and 60 months (ongoing follow-up with 2,228 interviews completed to date). The in-person follow-up rate for the 36-month interview was 88.2%. Information on survival status was also obtained by annual linkage to the Shanghai Vital Statistics Registry database. Procedures followed were in accordance with the ethical standards of the involved institutions, and Human Subjects Institutional Review Board approval was obtained from all participating institutions.
Postdiagnosis vitamin use
For women who responded “yes” to ever taking vitamins at least once a week for 1 month or more after diagnosis at the 6-month interview, specific information was collected for the time period from diagnosis to the interview for multivitamins; cod-liver oil; vitamins A, C, D, and E; B vitamins (data on individual types of B vitamins were not collected); and other nonspecified vitamins. Vitamin use was also assessed at the 36-month and ongoing 60-month follow-up surveys. For this report, we were interested in vitamin use during cancer treatments; hence, we used only data from the 6-month interview, which were collected for the time period when most women received their cancer treatments.
Categorical variables were created to examine postdiagnosis use of any type of vitamin supplement, multivitamins, vitamin E, vitamin C, and use of any antioxidants (including multivitamins, vitamins C, and vitamin E). For each type of vitamin, variables were created for any use and for duration of use (≤3 months and > 3 months, 3 months is approximately half the time between diagnosis and the baseline interview). Women were also classified according to the timing of vitamin use in relation to cancer treatment. Information regarding the specific brand, composition, and dosage of vitamin supplements was not available.
Differences in clinical characteristics, lifestyle factors, and sociodemographics by postdiagnosis vitamin supplement use were assessed with the chi-square test. Given the known health benefits of vitamin supplement use in the general population and concern that the use of antioxidant supplements during cancer treatment may reduce the effectiveness of cancer therapies, we chose both total mortality and recurrence as the main outcomes for our analysis. We also conducted analyses with breast cancer–specific mortality as the outcome. Survival time was calculated as the time from breast cancer diagnosis to event, with censoring at non–breast cancer deaths for breast cancer–specific mortality, the last date of in-person contact, or May 31, 2008 (5 months before the most recent linkage to the vital statistics registry, whichever was the latest date).
Adjusted HRs and their corresponding 95% CIs were derived from Cox proportional hazards regression models, using age as the time scale (19). Entry time was defined as age at diagnosis, and exit time was defined as age at event or censoring. The reference group for all analyses was never use of any vitamin supplement after diagnosis.
Potential confounders included known clinical predictors of survival (ER/PR status, TNM stage, chemotherapy, radiotherapy, tamoxifen use) and baseline sociodemographic and lifestyle factors associated with both vitamin use and survival [education, income, body mass index (BMI)], regular tea drinker, regular exercise participation calculated using standard metabolic equivalents (METs; ref. 20) in MET-hours per week, cruciferous vegetable intake, soy protein intake] with statistical definitions shown in Table 1. To address the potential for residual confounding due to inadequate adjustment for categorized confounders, we created a propensity score (21), which combined these potential confounders into a summary variable (and also included smoking and alcohol intake), using logistic regression with vitamin use as the dependent variable and potential confounders as the independent variables (21). We evaluated the associations of vitamin use with mortality and recurrence, adjusting for the propensity score, and results were similar to those generated from analyses with adjustment for individual confounders. Therefore, only the results from the latter are presented.
Radiotherapy, ER/PR status, TNM stage, and tamoxifen were examined as potential modifiers of the associations of vitamin use and breast cancer outcomes. Multiplicative interactions were tested for using −2 log likelihood ratio test statistics, which compared models with and without the interaction terms. All analyses were carried out using SAS version 9.2. Tests of statistical significance were based on 2-sided probability and values of P < 0.05 were considered statistically significant. Results for breast cancer–specific mortality were very similar to those for total mortality; hence, breast cancer–specific mortality results were included only as supplemental information (see Supplementary Tables S1–S3).
After an average of 4.1 years of follow-up (range: 0.5–6.2 years), 4,433 women were alive and 444 died (389 from breast cancer, 55 from other causes). A total of 4,325 women remained disease free during follow-up, and 532 had a breast cancer recurrence. Approximately 36.4% of breast cancer survivors ever used any type of vitamin supplement after diagnosis. Vitamin C was the most common (17.5%), followed by B vitamins (16.3%), vitamin E (7.6%), vitamin A (1.7%), and vitamin D (0.4%); about 11% used multivitamins.
Women who reported vitamin use tended to have higher education, income, daily intake of cruciferous vegetables and soy protein, and were more likely to have a lower BMI as well as to report not smoking, drinking tea, and exercising regularly (Table 1). Vitamin use did not vary significantly by age at diagnosis, joint ER and PR tumor status, TNM stage, chemotherapy, radiotherapy, tamoxifen use, number of pregnancies, family history of breast cancer, alcohol intake, or meat intake (Table 1).
As shown in Table 2, in general, postdiagnosis vitamin use within the first 6 months of cancer diagnosis [including any vitamins, multivitamins, vitamin E alone, vitamin C alone, and any antioxidants (multivitamins, vitamin C, and/or vitamin E)] was associated with reduced risk of total mortality and breast cancer recurrence, although not all HRs reached statistical significance. HRs were adjusted for multiple lifestyle factors (e.g., physical activity in MET-hours per week, soy protein intake), education, income, and clinical characteristics. Age-adjusted results were similar to the fully adjusted results (data not shown). The largest reduction in risk was seen for women who used vitamin C or E for a longer duration after diagnosis, estimated by use longer than approximately half the time period between diagnosis and the baseline interview. Specifically, women who used vitamin C for more than 3 months had a 44% decrease in risk of mortality (adjusted HR: 0.56; 95% CI: 0.37–0.87) and a 38% decrease in risk of recurrence (adjusted HR: 0.62; 95% CI: 0.43–0.90). Similarly, users of vitamin E for more than 3 months had a reduced risk of mortality (adjusted HR: 0.52; 95% CI: 0.27–1.01) and recurrence (adjusted HR: 0.57; 95% CI: 0.32–1.01), although point estimates were of marginal statistical significance (Table 2).
Associations of mortality and recurrence with vitamin use concurrent with chemotherapy and nonconcurrent with chemotherapy (vitamin use before or after chemotherapy) are shown in Table 3. This analysis was limited to women who received chemotherapy (n = 4,497), which comprised about 92% of study participants. We found that vitamin use concurrent or nonconcurrent with chemotherapy was associated with reduced risk of both mortality and recurrence, although the point estimates were not statistically significant, partially due to the reduced sample sizes for these analyses.
We evaluated the associations of mortality and recurrence with vitamin use by radiotherapy status (Table 4). Vitamin use was not associated with breast cancer outcomes among radiotherapy users (n = 1,597). In contrast, among women who did not receive radiotherapy (n = 3,280), vitamin use was associated with decreased risk for both mortality and recurrence, with the strongest association seen for the use of any antioxidant (adjusted HR for mortality: 0.65; 95% CI: 0.47–0.92; adjusted HR for recurrence: 0.63; 95% CI: 0.46–0.86). However, P values for multiplicative interactions were not statistically significant (Table 4). Women who received radiotherapy tended to be younger at diagnosis, have higher education, higher TNM stage at diagnosis, higher total meat intake, lower soy protein intake, were more likely to have received chemotherapy or drink tea, and were less likely to use tamoxifen or exercise regularly compared with women who did not receive radiotherapy (data not shown). Radiotherapy was associated with increased risk of mortality (adjusted HR: 1.40; 95% CI: 1.13–1.73) and breast cancer recurrence (adjusted HR: 1.39; 95% CI: 1.14–1.68). These associations were not significantly modified by other clinical characteristics or lifestyle factors (data not shown).
We also examined the joint associations for radiotherapy and vitamin use with breast cancer outcomes. In comparison to women who did not receive radiotherapy or take vitamin supplements during the first 6 months after diagnosis, women who took antioxidant vitamins and did not receive radiotherapy were at reduced risk of mortality (adjusted HR: 0.67; 95% CI: 0.48–0.94) and recurrence (adjusted HR: 0.66; 95% CI: 0.49–0.89); women who did not take antioxidant vitamins and received radiotherapy were at nonsignificantly increased risk of mortality (adjusted HR: 1.26; 95% CI: 0.92–1.72) and recurrence (adjusted HR: 1.26; 95% CI: 1.00–1.57); and women jointly exposed to radiotherapy and antioxidant vitamins had nonsignificantly increased risk of mortality (adjusted HR: 1.27; 95% CI: 0.99–1.64) and recurrence (adjusted HR: 1.17; 95% CI: 0.88–1.54). Similar results were found for analyses for any type of vitamin use (data not shown).
In stratified analyses, the associations of vitamin use with risk of mortality and recurrence varied little by TNM stage or tamoxifen use (Table 5). Some differences were found by ER/PR status, with stronger inverse associations among women with ER/PR-negative tumors than among women with ER/PR-positive tumors, although the P values for multiplicative interactions were not statistically significant (data not shown).
There is a widespread concern that the use of antioxidant supplements during cancer treatment may protect tumor cells from the oxidative damage induced by cancer therapies, thereby reducing the effectiveness of treatment and increasing risk of mortality (1, 11, 12). The epidemiologic data to support this concern are limited, in particular among breast cancer patients (1, 3). In fact, no large, prospective cohort study to date has reported findings on vitamin supplement use in conjunction with cancer treatment and subsequent mortality and recurrence risk among breast cancer survivors. Given the concern regarding the safety of antioxidant use during cancer treatment, as well as the few previous studies in this area, a randomized controlled trial may not be feasible or appropriate at this time; hence, results from observational studies are particularly warranted. In this first large, prospective cohort study of vitamin use in conjunction with cancer treatment among breast cancer survivors, we found that vitamin supplement use shortly after diagnosis, including antioxidant vitamins C and E, was associated with reduced risk of mortality and recurrence among breast cancer survivors regardless of whether vitamin use was concurrent or not concurrent with chemotherapy. In results stratified by radiotherapy status, the inverse association was found only among women who did not receive radiotherapy.
Few studies have directly evaluated whether there is an association of vitamin use after cancer diagnosis and during cancer treatment with mortality and recurrence, in particular among breast cancer patients (1–3, 22). In a recent comprehensive review of studies of antioxidant supplement use during breast cancer treatment and breast cancer patient outcomes (3), only 5 studies were identified that examined vitamin supplement use in association with recurrence and/or mortality. Four of these studies involved fewer than 55 patients (23–26) and were further limited by lack of a concurrent control group (24) or unclear/unreported statistical analyses (23, 25, 26). The largest study, a retrospective cohort study, identified patients from a medical database maintained by the British Columbia Cancer Agency (BCCA; ref. 27). Exposed women (n = 90) were seen by an orthomolecular physician who prescribed them a regimen of mega-dose vitamin/mineral supplements. Unexposed women (n = 180) were selected from the BCCA database and did not see that orthomolecular physician. After a median follow-up of 68 months, nonsignificantly increased HRs for breast cancer mortality (1.75; 95% CI: 0.83–2.69) and disease-free survival (1.55; 95% CI: 0.94–2.54) were found for women following the regimen of mega-dose vitamin/mineral supplements (27). However, this study was limited because of concerns regarding potential selection bias and lack of data on treatment compliance and over-the-counter vitamin use (27).
Of note, the main study that has been cited to support concern regarding the safety of antioxidant use during cancer treatment was conducted among head and neck cancer patients. This study was a randomized controlled clinical trial of 540 patients who received 400 IU of α-tocopherol and 30 mg of β-carotene or placebo at the start of radiotherapy and for 3 years thereafter (β-carotene was discontinued after enrollment of 156 patients; ref. 9). After a median of 6.5 years of follow-up, all-cause mortality was increased among participants in the supplement arm as compared with the placebo arm (HR: 1.38; 95% CI: 1.03–1.85; ref. 9). However, in a subsequent report, the increased risk for mortality was found to be limited to patients who smoked during radiation therapy (28). Because smoking rates are low among breast cancer survivors (18, 29), and prognosis and treatment of breast cancer differ substantially from that of head and neck cancer, it is questionable whether the results from this trial can be generalized to breast cancer patients.
The biological activity of antioxidants depends on several factors, including oxidative stress level, interactions with other antioxidants, and the concentration of antioxidants available at the cellular level (1). One explanation for a lack of protection from vitamin use among women who received radiotherapy in our study could be that the dosages of vitamin supplements were not high enough to be beneficial among these women. Further studies with a larger sample size and a wide range of vitamin supplements are needed to confirm the association of vitamin use and breast cancer outcomes among radiotherapy users.
The SBCSS is a large, well-designed, prospective cohort study of breast cancer survivors (30). The potential for selection bias is small due to the population-based design and high response and follow-up rates. Standardized in-person interviews collected information on cancer treatment, lifestyle factors, anthropometrics, and disease history, which improved the exposure assessment and allowed for adjustment for many potential confounders.
Several limitations should be considered. First, we did not have complete information on dosages for vitamin supplements. However, among women with available data who reported taking vitamin C or E (in mg) daily, approximately 85% used 400 mg/d or less of vitamin C and 99% used 400 mg/d or less of vitamin E. These are much lower dosages than those found in mega-dose vitamins, which can be well over 1 g (1, 3). Second, we did not have complete dietary information for participants, which prevented an evaluation of dietary vitamin/antioxidant intake. In the Shanghai Women's Health Study (31), a population-based cohort study of women aged 40 to 70 years residing in Shanghai, where the current study was conducted, dietary intakes of vitamins C and E were very weakly correlated with supplement use of these single vitamins (r = 0.04 and r = 0.08, respectively). On the other hand, daily intakes of cruciferous vegetables and dietary vitamin C were highly correlated (r = 0.67). In our analyses, we adjusted for daily intakes of cruciferous vegetable and soy protein, both of which are major sources of dietary antioxidants for Chinese women in Shanghai. Thus, potential confounding by dietary sources of vitamins should not be a major concern in this study.
Although we adjusted for a wide range of clinical prognostic factors, sociodemographics, and lifestyle factors in multivariable analyses, both as independent covariates and by creating propensity scores, and obtained similar results with both approaches, we cannot exclude the possibility of residual confounding from inadequately measured covariates or unmeasured confounders. For example, one potential concern is that vitamin use is more common in women with higher socioeconomic status, a factor that may also be associated with completing recommended cancer therapy. Although we collected detailed information on chemotherapy treatment regimens and duration of the treatment, we did not obtain information on the prescribed length of treatment. Hence, we are not able evaluate whether women completed the full prescribed courses of chemotherapy. However, in our study population, the weeks of total chemotherapy treatment were very similar for users of vitamins [mean = 17.7; range: 13.1 (25th percentile) to 21.9 (75th percentile)] and for nonusers [mean = 17.4; range: 13.1 (25th percentile) to 21.6 (75th percentile)]. Thus, differences in treatment compliance is an unlikely explanation for our findings.
Another concern is that prediagnosis vitamin use, which could be related to both postdiagnosis vitamin use and breast cancer outcomes, was unavailable for all study participants. We did, however, have information on prediagnosis vitamin use for a subset of participants (n = 1,442). The correlation between prediagnosis vitamin use (any type) and use around the time of treatment was 0.18. Results adjusted for prediagnosis vitamin use were similar to overall findings, although not significant because of the smaller sample size. We did not examine vitamin use at 36 months postdiagnosis in relation to breast cancer outcomes, because the focus of this study was to evaluate the association of vitamin use during cancer treatment. In addition, the cohort follow-up time is not yet long enough to evaluate long-term vitamin use in relation to breast cancer outcomes. Continued follow-up of this cohort will allow us to examine this research question in the future. Finally, despite an overall large sample size, the number of women exposed to individual vitamins was small and studies with a larger sample size are warranted.
In conclusion, we found no evidence that vitamin use during the first 6 months following diagnosis had a detrimental effect on breast cancer outcomes. Instead, vitamin use, particularly vitamin C and vitamin E use, may be associated with reduced risk of mortality and recurrence, independent of multiple lifestyle factors, clinical prognostic factors, and sociodemographics. The inverse association was primarily seen among women who did not receive radiotherapy. To our knowledge, this is the first large, prospective cohort study to report on vitamin use during cancer treatment in association with recurrence and mortality among breast cancer survivors and future studies of postdiagnosis vitamin use and breast cancer outcomes are needed.
Disclosure of Potential Conflicts of Interest
No potential conflicts of interest were disclosed.
This study was supported by grants from the Department of Defense Breast Cancer Research Program (DAMD 17-02-1-0607) and the National Cancer Institute (R01 CA118229).
The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
Note: Supplementary data for this article are available at Cancer Epidemiology, Biomarkers and Prevention Online (http://cebp.aacrjournals.org/).
- Received October 12, 2010.
- Revision received December 2, 2010.
- Accepted December 3, 2010.
- ©2010 American Association for Cancer Research.