Prevalence and predictors of antioxidant supplement use during breast cancer treatment

The Long Island Breast Cancer Study Project

Abstract

BACKGROUND.

Although many patients take antioxidant dietary supplements during breast cancer treatment, the benefits of such supplementation are unproven. The authors of this report analyzed the prevalence of and factors associated with antioxidant supplement use during breast cancer (BC) treatment among women who participated in the Long Island Breast Cancer Study Project.

METHODS.

From 2002 through 2004, women with BC who had participated a case-control study from 1996 to 1997 were invited to participate in a follow-up interview. Antioxidant supplement use was defined as any self-reported intake of supplemental vitamin C, vitamin E, β-carotene, or selenium in individual supplements or multivitamins.

RESULTS.

Follow-up interview participants were younger, more predominantly white, and of higher socioeconomic status than women who did not respond. Among 764 participants who completed the follow-up interview, 663 (86.8%) reported receiving adjuvant treatment for their BC. Of those 663 women, 401 (60.5%) reported using antioxidants during adjuvant treatment: One hundred twenty of 310 women (38.7%) used antioxidants during chemotherapy, 196 of 464 women (42.2%) used them during radiation, and 286 of 462 women (61.9%) used them during tamoxifen therapy. Of 401 antioxidant users, 278 women (69.3%) used high doses (doses higher than those contained in a Centrum multivitamin). The factors that were associated with high antioxidant supplement use during treatment were higher fruit and vegetable intake at diagnosis (relative risk [RR], 1.71; 95% confidence interval [CI], 1.13-2.59), tamoxifen use (RR, 3.66; 95% CI, 2.32-5.78), ever using herbal products (RR, 3.49; 95% CI, 2.26-5.38), and ever engaging in mind-body practices (RR, 1.72; 95% CI, 1.13-2.64).

CONCLUSIONS.

Given the common use of antioxidant supplements during BC treatment, often at high doses and in conjunction with other complementary therapies, future research should address the effects of antioxidant supplementation on BC outcomes. Cancer 2009. © 2009 American Cancer Society.

An estimated 45% to 87% of patients with breast cancer use antioxidant supplements after diagnosis.1-9 However, to our knowledge, the characteristics of women who use antioxidants specifically during treatment for breast cancer have not been described previously.

Many patients with breast cancer take antioxidant supplements, because they believe antioxidants will protect them from the side effects of breast cancer treatment, help prevent breast cancer recurrence, and improve their overall health.7 However, the actual consequences of using antioxidant supplements during cancer treatment are poorly understood and controversial.10-18 In the past few decades, the idea that antioxidant supplementation could protect normal tissue from the adverse effects of cancer treatments has become widespread.19

Radiation therapy and some chemotherapeutic agents induce apoptosis by generating reactive oxygen species, which damage tumor cell DNA and disrupt mitochondrial membranes.20-22 Many medical and radiation oncologists believe that antioxidant supplements interfere with this process and that antioxidant use during treatment may reduce therapeutic efficacy.14 However, the effects of taking antioxidant supplements during cancer therapy on cancer recurrence or mortality and the characteristics of antioxidant users have not been well studied.18, 23

We determined the prevalence and predictors of antioxidant supplement use, taken as multivitamins and as single vitamin supplements, during breast cancer treatment among women with breast cancer who participated in the population-based Long Island Breast Cancer Study Project (LIBCSP). To our knowledge, ours is the first detailed study of antioxidant use during treatment in a cohort study setting.

MATERIALS AND METHODS

Participants

The LIBCSP began as a federally mandated, population-based, case-control study to investigate whether breast cancer risk was associated with environmental exposures among women in Nassau and Suffolk counties in New York State.24 A subsequent follow-up study was conducted among cases to determine whether environmental exposures were associated with mortality. The LIBCSP case-control study cases (n = 1508) were women who were diagnosed with a first primary in situ or invasive breast cancer between August 1, 1996 and July 31, 1997. Cases were identified by rapid ascertainment methods through daily or weekly contact with local pathology departments of hospitals known to diagnose and treat residents of Nassau and Suffolk counties with newly diagnosed breast cancer, including some hospitals in New York City. Once a case was identified, her physician was contacted to verify the diagnosis and to obtain permission to contact the patient. Among eligible cases, the response rate for the parent case-control study was 82.1%.

For the LIBCSP follow-up, which was conducted between 2002 and 2004, interviewers attempted to conduct telephone interviews with all cases who previously agreed to be contacted again (n = 1414; 93.8%),24 and 1098 cases (77.7%) completed a full or short-form interview in person or by proxy. The study reported here is based on the 784 cases (55.4%) who personally completed the full follow-up interview.

Written, signed informed consent was obtained from all participants before the beginning of the in-person baseline interview. The study protocol was approved by all institutional review boards of the collaborating institutions.

Data Collection

Baseline case-control questionnaire

The baseline questionnaire was administered to participants in their homes shortly after they were diagnosed (mean, 96 days) with a first primary breast cancer. The questionnaire asked about known and suspected risk factors for breast cancer (available at: http://epi.grants.cancer.gov/LIBCSP/projects/Questionnaire.html; accessed on October 10, 2008). A 101-item, modified Block food frequency questionnaire was self-administered. Baseline questionnaire variables that were used in the analyses presented here included age at diagnosis, race/ethnicity, education, annual household income, marital status, body mass index, family history of breast cancer, menopausal status at diagnosis, mammogram history, oral contraceptive use, hormone-replacement therapy use, lifetime physical activity, smoking history, previous alcohol use, and fruit and vegetable intake.

Follow-up questionnaire among cases

The follow-up questionnaire asked about breast cancer recurrence risk factors, including demographics, health behaviors, medical history, first course of treatment received, and other clinical variables. The questionnaire included detailed questions about complementary and alternative medicine (CAM) therapy use during 3 separate periods: before, during, and after breast cancer diagnosis and treatment. The CAM modalities that were assessed included multivitamins, single vitamins, minerals, herbal products, other over-the-counter health products, mind-body activities (eg, spirituality, support groups, meditation), special treatments (eg, biofeedback, colon cleansing, hydrotherapy), alternative cancer clinics, special diets (eg, vegan, macrobiotic, low-fat diets), and visits with CAM practitioners. The analyses presented here focus on antioxidant supplement use during treatment. Analyses of other CAM use will be presented in a separate article.

Study participants who reported ever using multivitamins were asked whether they used them during specific phases of their treatment, including surgery, chemotherapy, radiation therapy, and tamoxifen therapy. In addition, participants were asked how many tablets were consumed per week since diagnosis. The questionnaire asked about specific types of multivitamins: multivitamins with minerals, multivitamins without minerals, antioxidant combination types with vitamins A, C, and E; stress-tab types, women's formula types; multivitamins with herbs; and other multivitamins. For the current analysis, we assumed that each multivitamin pill contained the doses of specific micronutrients present in the commonly used Centrum multivitamin circa 2002 (60 mg vitamin C, 30 IU vitamin E, 20 μg selenium, 5000 IU β-carotene).25 To estimate the daily doses consumed during treatment of vitamin C, vitamin E, selenium, and β-carotene from multivitamins, we multiplied the Centrum dose of each individual micronutrient by the number of multivitamin tablets a participant reported taking per week and dividing by 7.

Data were collected in a similar fashion on the use of individual antioxidant supplements during breast cancer treatment, including vitamin C, vitamin E, selenium, and β-carotene. For each individual supplement, data were collected on dose per tablet and number of tablets taken per week since diagnosis and were used to estimate daily doses consumed during treatment.

Less than 4% of participants were missing data on individual antioxidant supplement use (β-carotene, 1.3%; vitamin C, 2.5%; vitamin E; 3.5%, selenium, 2.4%). We assumed that participants who reported taking 1 of these supplements but who provided no information on dose had taken the lowest dose. We conducted sensitivity analyses to determine whether assuming the median dose or the highest dose would make any appreciable differences in our results, and it did not.

To estimate individual intake of each antioxidant during treatment (vitamin C, vitamin E, selenium, and β-carotene), we added the daily dose values for the multivitamins and the individual supplements. Then, intake of each supplement was categorized as none, low, or high. High doses were defined as >60 mg vitamin C, >30 IU vitamin E, >20 μg selenium, and >5000 IU β-carotene; these cutoff points were based on the doses in a typical Centrum multivitamin circa 2002. Between 1997 and 2004, the Food and Nutrition Board of the Institute of Medicine revised its nutrient intake recommendations, now termed Dietary Reference Intakes (DRIs).26, 27 The current DRIs are 90 mg vitamin C, 22 IU natural vitamin E (33 IU synthetic vitamin E), and 55 μg selenium.28 There is no DRI for β-carotene.

We created a composite antioxidant index to summarize total intake of supplemental β-carotene, vitamin C, vitamin E, and selenium during treatment. We computed a categorical variable with a range from 0 points to 8 points based on scoring the intake of each antioxidant as never (0), low (1), or high (2). The antioxidant index itself was divided into 3 categories: none (score, 0), low (score,1-4), and high (score, 5-8).

Medical records

Signed medical records release forms were obtained from case women at the baseline and follow-up interviews. Medical records were obtained from inpatient and outpatient facilities involved in the diagnosis, treatment, and follow-up of breast cancer cases. Medical records were abstracted at baseline and at follow-up to ascertain tumor characteristics and first course of treatment for the primary breast cancer diagnosis. For the 598 women who had complete medical records data available on the first course of treatment, the agreement between self-reported breast cancer treatment and treatment recorded on the medical record was high for chemotherapy (κ = 0.96), radiation therapy (κ = 0.97), and hormone therapy (κ = 0.92).29 Thus, for the current analyses, we used treatment information collected through self-report during the follow-up interview.

Statistical Analyses

Chi-square tests and Student t tests were used to evaluate the statistical significance of the associations of the demographic, health behavior, clinical, and CAM use variables with antioxidant supplement use during treatment. Multivariate logistic and polytomous regression analyses were performed to identify independent predictors of antioxidant use, which were categorized as either none/any or none/low dose/high dose. Model-building procedures were performed in the following 3 basic steps. In Step 1, candidate predictors were identified as the predictors that were statistically significant (P < .10) in unadjusted analyses30 as well as any a priori confounders. Step 2 involved fitting a full model that included all variables identified in Step 1, in which variables were retained in the model if the corresponding P value was <.05. Finally, in Step 3, variables that were rejected in the first step were added back to the model, 1 at a time, for reassessment and were kept in the final model if the corresponding P value was <.05. Tests of significance were based on the likelihood ratio test statistic. Analyses were performed using Stata software (version 9.2; StataCorp., College Station, Tex).

RESULTS

Respondents

In total, 784 individuals personally completed the full telephone interview. They were younger, more educated, more likely to be white, and had higher household income than women who did not respond to the follow-up questionnaire (Table 1). Of 784 women, 764 provided data on antioxidant supplement use (Table 2). Their mean age at the time of diagnosis with a first primary breast cancer was 56.3 years (±11.4 years), and 94% were non-Hispanic whites. Greater than 60% of the women had at least some college education. In total, 555 of 764 respondents (72.6%) used multivitamin or single antioxidant supplements before diagnosis, and the number of users increased to 650 of 764 (85.1%) after diagnosis. All 764 women underwent surgical treatment for their breast cancer, and 663 (86.8%) also received adjuvant treatment: 310 (40.6%) received chemotherapy, 464 (60.7%) received radiation therapy, and 462 (60.5%) received tamoxifen. Of 663 women who received adjuvant treatment, 560 (84.5%) reported using antioxidants after their breast cancer diagnosis, and 401 (60.5%) specifically reported using antioxidants during their treatment. Women who received adjuvant treatment did not differ from those who did not receive such treatment with respect to antioxidant use after diagnosis (data not shown).

Table 1. Demographic and Breast Cancer-related Characteristics at Diagnosis of Responders and Nonresponders

Characteristic	Responders (n=764)		Nonresponders (n=724)		P*
	No.	%	No.	%
* P values are based on chi-square tests and do not include individuals who had missing data.
Demographic characteristics
Age at diagnosis, y
<45	136	17.8	80	11.0	<.001
45-54	231	30.2	162	22.4
55-64	204	26.7	160	22.1
≥65	193	25.3	322	44.5
Race/ethnicity
Non-Hispanic white	718	94.0	626	86.5	<.001
Hispanic white	20	2.6	27	3.7
Black/African American	15	2.0	54	7.5
Other	11	1.4	14	1.9
Education
≤High school graduate	301	39.4	409	56.5	<.001
Some college	195	25.5	160	22.1
College graduate	111	14.5	78	10.8
Postgraduate	157	20.5	71	9.8
Annual household income
<$25,000	76	9.9	182	25.1	<.001
$25,000-$49,999	186	24.3	177	24.4
$50,000-$89,999	230	30.1	150	20.7
≥$90,000	182	23.8	95	13.1
Breast cancer-related characteristics
Mammogram
Never	21	2.7	56	7.7	<.001
≥5 y ago	11	1.4	28	3.9
Within past 5 y	721	94.4	623	86.0
Stage at diagnosis
In situ	137	17.9	95	13.1	0.01
Invasive	627	82.1	629	86.9

Table 2. Demographic, Clinical, and Lifestyle Characteristics of Study Participants by Antioxidant Supplement Use During Adjuvant Treatment (n=764)

Characteristic	No Adjuvant Treatment (n=101)		Antioxidant Supplement Use During Chemotherapy, Radiotherapy, or Hormonal Therapy (n=663)						P
			No Use (n=262)		Low-dose Use (n=169)		High-dose Use (n=232)
	No.	%	No.	%	No.	%	No.	%
ER indicates estrogen receptor; −, negative; PR, progesterone receptor; +, positive; NA, not available; BMI, body mass index; CAM, complementary and alternative medicine; NE, not estimable; OTC, over the counter.
Demographic characteristics
Age at diagnosis, y
<45	25	24.8	49	18.7	22	13.0	40	17.2	.026
45-54	32	31.7	75	28.6	43	25.4	81	34.9
55-64	26	25.7	60	22.9	56	33.1	62	26.7
≥65	18	17.8	78	29.8	48	28.4	49	21.1
Race/ethnicity
Non-Hispanic white	98	97.0	246	93.9	161	95.3	213	91.8	.233
Hispanic white	2	2.0	8	3.1	3	1.8	7	3.0
Black/African American	1	1.0	6	2.3	4	2.4	4	1.7
Other	0	0.0	2	0.8	1	0.6	8	3.4
Education
≤High school graduate	27	26.7	117	44.7	81	47.9	76	32.8	.001
Some college	25	24.8	66	25.2	43	25.4	61	26.3
College graduate	18	17.8	40	15.3	19	11.2	34	14.7
Postgraduate	31	30.7	39	14.9	26	15.4	61	26.3
Annual household income at diagnosis
<$25,000	10	9.9	33	12.6	16	9.5	20	8.6	.003
$25,000-$49,999	15	14.9	86	32.8	64	37.9	69	29.7
$50,000-$89,999	45	44.6	79	30.2	61	36.1	81	34.9
≥$90,000	31	30.7	64	24.4	28	16.6	62	26.7
Clinical characteristics
Stage
In situ	72	71.3	24	9.2	18	10.7	23	9.9	<.001
Invasive	29	28.7	238	90.8	151	89.3	209	90.1
Hormone receptor status
ER−/PR−	6	5.9	48	18.3	24	14.2	21	9.1	.156
ER−/PR+	1	1.0	11	4.2	5	3.0	10	4.3
ER+/PR−	5	5.0	16	6.1	17	10.1	19	8.2
ER+/PR+	14	13.9	106	40.5	65	38.5	107	46.1
Treatments received
Chemotherapy	NA		119	45.4	81	47.9	110	47.4	<.001
Radiotherapy	NA		181	69.1	118	69.8	165	71.1	<.001
Hormonal therapy	NA		152	58.0	125	74.0	185	79.7	<.001
BMI at diagnosis, kg/m^2
<25	64	63.4	109	41.6	75	44.4	127	54.7	.001
25-<30	26	25.7	91	34.7	53	31.4	71	30.6
≥30	11	10.9	62	23.7	41	24.3	34	14.7
Family history of breast cancer
No	76	75.2	208	79.4	121	71.6	190	81.9	.082
Yes	20	19.8	45	17.2	45	26.6	41	17.7
Menopausal status at diagnosis
Premenopausal	45	44.6	86	32.8	50	29.6	98	42.2	.011
Menopausal	55	54.5	166	63.4	117	69.2	127	54.7
Health behavior characteristics before diagnosis
Mammogram
Never had mammogram	0	0.0	8	3.1	8	4.7	5	2.2	.325
Had mammogram >5 y ago	2	2.0	2	0.8	3	1.8	4	1.7
Had mammogram ≤5 y ago	99	98.0	246	93.9	157	92.9	219	94.4
Oral contraceptive use
Never	48	47.5	137	52.3	86	50.9	105	45.3	.390
Ever	53	52.5	123	46.9	83	49.1	127	54.7
Hormone replacement therapy use
Never	59	58.4	189	72.1	113	66.9	150	64.7	.070
Ever	42	41.6	73	27.9	56	33.1	82	35.3
Physical activity (from menarche to diagnosis), h/wk
0	25	24.8	74	28.2	48	28.4	38	16.4	.091
0-0.69	21	20.8	59	22.5	39	23.1	53	22.8
0.70-2.6	23	22.8	55	21.0	42	24.9	64	27.6
≥2.7	24	23.8	64	24.4	30	17.8	63	27.2
Smoking history
Never	45	44.6	111	42.4	73	43.2	108	46.6	.196
Current (within 12 mo)	15	14.9	62	23.7	32	18.9	33	14.2
Past/former	41	40.6	89	34.0	64	37.9	91	39.2
Alcohol use
Never	33	32.7	90	34.4	58	34.3	77	33.2	.985
Ever	68	67.3	172	65.6	111	65.7	155	66.8
Fruit and vegetable intake, servings/wk
0-34	65	64.4	181	69.1	103	60.9	126	54.3	.011
≥35	34	33.7	80	30.5	64	37.9	103	44.4
CAM use (ever)
Any CAM	101	100.0	252	96.2	169	100.0	232	100.0	NE
Multivitamins	91	90.1	190	72.5	152	89.9	229	98.7	<.001
Single vitamins	88	87.1	182	69.5	137	81.1	229	98.7	<.001
Herbal supplements	48	47.5	82	31.3	56	33.1	144	62.1	<.001
Other OTC healthcare products	40	39.6	71	27.1	48	28.4	90	38.8	.010
Mind-body activities	54	53.5	112	42.7	76	45.0	147	63.4	<.001
Special treatments	6	5.9	6	2.3	1	0.6	11	4.7	.032
Alternative cancer clinics	0	0.0	0	0.0	0	0.0	0	0.0	NE
Special diets	39	38.6	82	31.3	42	24.9	97	41.8	.002
CAM practitioners	45	44.6	76	29.0	47	27.8	93	40.1	.002

Factors Associated With Antioxidant Use During Specific Treatments

Of 310 women who received chemotherapy, 120 women (38.7%) used antioxidants during chemotherapy. Antioxidant use during chemotherapy was positively associated with ever having used herbal products and high fruit and vegetable intake before diagnosis (Table 3). Of 464 women who received radiation therapy, 196 (42.2%) used antioxidants during radiation therapy. Antioxidant use during radiation therapy was positively associated with ever having used herbal products and having in situ (as opposed to invasive) disease (Table 3). Of 462 women who received tamoxifen treatment for their breast cancer, 286 (61.9%) used antioxidants during that treatment. Antioxidant use during tamoxifen therapy was positively associated with ever having used herbal products, lower body mass index (<25 kg/m²), moderate lifetime alcohol intake, and family history of breast cancer.

Table 3. Adjusted Analyses of Antioxidant Supplement Use During Chemotherapy, Radiation Therapy, and Tamoxifen Therapy

Variable	Antioxidant Use During Chemotherapy, n=309			Antioxidant Use During Radiotherapy, n=464			Antioxidant Use During Tamoxifen Therapy, n=452
	OR	95% CI	P	OR	95% CI	P	OR	95% CI	P
OR indicates odds ratio; CI, confidence interval; BMI, body mass index. *The ORs displayed are for the variables that were included in each model.
Age at diagnosis (per y)	1.02	0.99-1.04	.22	1.01	0.99-1.02	.54	0.99	0.97-1.01	.49
Race (vs non-Hispanic white)			.53			.67			.23
Hispanic white	0.46	0.08-2.58		0.57	0.17-1.96		0.75	0.20-2.80
Black/African American	0.86	0.19-3.91		0.89	0.24-3.30		3.69	0.41-32.88
Other	2.94	0.45-19.42		1.76	0.45-6.92		4.35	0.50-37.45
Education (vs ≤high school graduate)			.66			.43			.10
Some college	0.81	0.43-1.53		1.46	0.91-2.34		1.29	0.77-2.17
College graduate	0.63	0.29-1.39		1.16	0.62-2.16		0.56	0.29-1.09
Postgraduate	0.73	0.37-1.43		1.34	0.78-2.30		1.24	0.68-2.24
BMI (vs <25), kg/m^2									.03
25-<30							0.54	0.34-0.87
≥30							0.60	0.34-1.06
Lifetime alcohol intake (vs nondrinkers), g/d									.009
<15							0.81	0.52-1.27
15-30							2.84	1.23-6.56
≥30							1.51	0.52-4.40
Family history of breast cancer (yes vs no)							1.82	1.07-3.09	.02
Fruit and vegetable intake (vs <35 servings/wk)			.05
≥35 servings/wk	1.64	1.00-2.69
Stage (invasive vs in situ)				0.51	0.28-0.94	.03
Herbal products (ever vs never)	2.71	1.64-4.49	<.001	1.74	1.17-2.57	.006	2.31	1.49-3.58	<.001

Antioxidant Supplement Dose Used During Treatment

Of 401 women who used antioxidants during therapy, 342 (51.6%) used multivitamins, 199 (30%) used individual vitamin C, 215 (32.4%) used individual vitamin E, 54 (8.1%) used individual selenium, and 14 (2.1%) used individual β-carotene (data not shown). Of all 401 women, 278 (69.3%) used high doses of antioxidant supplements; of 376 women who used any form of vitamin C (from multivitamins or as an individual supplement), 211 (56.1%) used >60 mg per day (Table 4); of 392 women who used any form of vitamin E, 224 (57.1%) used >30 IU per day; of 352 women who used any form of selenium, 88 (25%) used >20 μg per day; and, of 345 women who used any form of β-carotene, 54 (15.7%) used >5000 IU per day.

Table 4. Distribution of No Use, Low-dose Use, and High-dose Use for Total Intake of Antioxidant Supplements During Treatment (n=663)

Antioxidant	Antioxidant Supplement Use During Treatment (n=663)
	No Use		Low-dose Use		High-dose Use
	No.	%	No.	%	No.	%
* The high vitamin C dose was >60 mg. † The high vitamin E dose was >30 IU. ‡ The high selenium dose was >20 μg. § The high β-carotene dosewas >5000 IU.
Vitamin C*	287	43.3	165	24.9	211	31.8
Vitamin E†	271	40.9	168	25.3	224	33.8
Selenium‡	311	46.9	264	39.8	88	13.3
β-Carotene§	318	48.0	291	43.9	54	8.1
Antioxidant index	262	39.5	169	25.5	232	35.0

Antioxidant Index

Of 663 women who received any adjuvant therapy treatment, 232 (35%) scored high on the antioxidant index, which is a proxy for simultaneous high-dose use of multiple antioxidant supplements. In the polytomous model for predicting levels of antioxidant use during treatment, the strongest predictor low-dose antioxidant use during treatment was tamoxifen use (Table 5). The strongest predictors of high-dose antioxidant use during treatment were high fruit and vegetable intake before diagnosis, tamoxifen use, ever using herbal products, and ever using mind-body activities.

Table 5. Adjusted Analyses Using Polytomous Logistic Regression for Antioxidant Use During Any Treatment (n=642)

Variable	Low-dose Antioxidant Supplement Use (n=164)*		High-dose Antioxidant Supplement Use (N=228)*		P†
	RR	95% CI	RR	95% CI
RR indicates relative risk; 95% CI, 95% confidence interval; Ref, reference category; OTC products, over the counter products. * Compared with nonusers (n=250) of antioxidant supplements during treatment. † P values were based on a likelihood ratio test of significance for each variable in the final, adjusted, polytomous regression model.
Age at diagnosis (per y)	1.00	0.98-1.02	1.00	0.98-1.02	.85
Race
Non-Hispanic white	Ref		Ref		.11
Hispanic white	0.71	0.17-2.90	1.04	0.33-3.20
Black	1.85	0.46-7.48	1.05	0.25-4.49
Other	1.88	0.11-31.24	12.40	1.36-112.82
Education
≤High school	Ref		Ref		.62
Some college	0.96	0.58-1.59	1.06	0.63-1.77
College graduate	0.56	0.28-1.10	0.79	0.42-1.46
Postgraduate	0.86	0.47-1.59	1.23	0.70-2.18
Fruit and vegtables, servings/wk
<35	Ref		Ref		.03
≥35	1.42	0.92-2.19	1.71	1.13-2.59
Family history of breast cancer
No	Ref		Ref
Yes	1.66	1.02-2.70	0.86	0.51-1.44	.03
Tamoxifen
Never	Ref		Ref
Ever	2.07	1.32-3.23	3.66	2.32-5.78	<.0001
Herbal products
Never	Ref		Ref
Ever	1.19	0.75-1.90	3.49	2.26-5.38	<.0001
Other OTC products
Never	Ref		Ref
Ever	1.27	0.80-2.01	1.50	0.97-2.32	.18
Mind-body activities
Never	Ref		Ref
Ever	1.17	0.76-1.80	1.72	1.13-2.64	.04
Special diets
Never	Ref		Ref
Ever	0.66	0.41-1.05	1.08	0.71-1.65	.09

DISCUSSION

Among women with breast cancer who participated in the LIBCSP follow-up study, 38.7% reported antioxidant supplement use during chemotherapy, 42.2% reported antioxidant supplement use during radiation therapy, and 61.9% reported antioxidant supplement use during tamoxifen therapy. Greater than 50% of the study participants who used antioxidants during treatment consumed doses that exceeded the DRI.

The majority of participants in our study were well educated, non-Hispanic white, postmenopausal women with high household incomes who engaged in other health-oriented behaviors before diagnosis (eg, regular physical activity, high fruit and vegetable intake, not smoking), and we believe our results are generalizable to similar populations. Almost all participants reported previous use of 1 or more forms of CAM. Most women who reported antioxidant supplement use during treatment, especially those who used high doses, also engaged in other health-oriented behaviors. Women who used high-dose antioxidant supplements during treatment were much more likely ever to have used herbal products and ever to have received tamoxifen therapy than women who did not use antioxidant supplements.

Antioxidant supplement use appears to have been more prevalent among the Long Island breast cancer patients in our sample than among respondents to the 2000 National Health Interview Survey. In that sample 51% of women, including 62% of non-Hispanic white women, reported using vitamins or minerals anytime in the past year.31 However, in a literature review of women who had a history of breast cancer, from 67% to 87% reported using vitamins and minerals9; those rates are similar to our observations.

We observed a strong association between antioxidant supplement use and tamoxifen use. This association may reflect tamoxifen users' strategies to manage vasomotor symptoms associated with this therapy (eg, taking vitamin E supplements to help with hot flashes32). We did not ask women about their motives for dietary supplement use during treatment; however, previous studies have demonstrated that women may take supplements to counteract toxicities due to breast cancer treatment.7

To our knowledge, this study is the first to report on antioxidant use during specific phases of breast cancer treatment. A recent review identified 8 articles that reported on vitamin/mineral supplement use among breast cancer survivors,9 but only 1 of those articles focused on the 12 months postresection.33 A report from the Nurses Health Study indicated that 20% of breast cancer survivors used high-dose vitamins during the 2 years before the assessment (mean, 3.2 years postdiagnosis).6 Among Canadian breast cancer patients (mean, 2.4 years postdiagnosis), 51% reported ever using vitamins and minerals other than those in a multivitamin; to treat symptoms associated with breast cancer, 13.2% used vitamin E, 12.3% used vitamin C, and 6% used β-carotene.8 Neither study reported on use during specific breast cancer treatments.

Another strength of the current study is our cumulative antioxidant index and categorization of study participants as high-dose, low-dose, or nonusers of multiple forms of antioxidants. We developed this approach to exposure measurement because antioxidants rarely are taken alone and may have cumulative physiologic effects. Although our categorization of antioxidant use as low-dose or high-dose was based on detailed questionnaire data, at best, it is an approximation of actual intake. We were conservative in calculating the specific doses of antioxidants consumed in multivitamins; therefore, we may have underestimated some actual doses.

The main limitation of our study is its relatively low response rate, especially compared with the high response rate of the original LIBCSP case-control study. However, we have complete first-hand follow-up questionnaire data on 764 of 1414 women (55.4%) who agreed to be contacted again for the follow-up study, including the CAM questionnaire. Our response bias analyses (data not shown) revealed that women who completed the full questionnaire were of higher socioeconomic status than those who did not. Because such women also are more likely than others to use CAM,34 our study participants may have been heavier users of antioxidants than breast cancer patients in the general population.

The follow-up survey was conducted at least 5 years after participants were diagnosed with a first primary breast cancer. The passage of time may have led to poor recall of supplement use before diagnosis and during treatment. Despite possible misclassification, our analyses identified strong predictors of supplement use that are consistent with those reported by other investigators.3, 35, 36

Data on the actual effects of antioxidant supplements during breast cancer treatment are limited.37 Some studies of antioxidant supplement use during treatment of other cancer sites have suggested harm.18 In a randomized clinical trial of patients who were undergoing radiation therapy for head and neck cancers, α-tocopherol supplementation (400 IU/day for 3 years beginning at the initiation of radiation therapy) was associated with a decrease in both adverse side effects during treatment38 and overall survival at 8 years.39 However, patients with head and neck cancer may differ from breast cancer patients in behaviors, such as tobacco use, that may be relevant to the effects of antioxidants. The effects of antioxidant supplementation during treatment also may depend on the doses and types of supplements, intake of fruits and vegetables, tumor site and stage, type of treatment, and genetic polymorphisms in endogenous antioxidant enzymes.

Areas in which further research is needed include the physiologic interactions between antioxidants and radiation therapy, chemotherapy, and hormone therapy; whether there is a dose threshold above which an antioxidant exerts benefit or harm; and the long-term effect of antioxidants on breast cancer recurrence and survival. Prospective observational studies and clinical trials need to elucidate whether antioxidants affect treatment toxicities, treatment efficacy, and recurrence and survival. Our data suggest that many women diagnosed with breast cancer who use antioxidant supplements also are engaging in other behaviors that may affect their risk of recurrence, and these behaviors should be accounted for in studies of antioxidant supplements during treatment. The results of these studies will shed light on whether or not patients can use antioxidant supplements to enhance the effects of conventional treatment or alleviate its side effects without adversely affecting their prospects for survival.

In summary, we observed that the majority of breast cancer patients in Long Island, many of whom were of high socioeconomic status, used antioxidants during treatment. We believe that oncologists should discuss supplement use and dosing with their patients. More specifically, oncologists can inform patients that antioxidant supplements may dampen the effects of chemotherapy and radiation therapy but that clear evidence of benefit or harm is not yet available.9, 18, 37

Acknowledgements

We thank the women of Nassau and Suffolk Counties, New York who participated in the Long Island Breast Cancer Study Project for their valuable contributions.

Conflict of Interest Disclosure

Supported by a grant from the Lance Armstrong Foundation; by National Cancer Institute and National Institutes of Environmental Health and Sciences grants U01CA/ES66572, U01CA66572, CA52283, and P30ES10126; and by National Cancer Institute grant R25 CA09406.