Complementary Medicine Use before and 1 Year Following Genetic Testing for BRCA1/2 Mutations

Complementary and alternative medicine (CAM) has become a much more accepted option for individuals facing cancer and other chronic diseases in recent years (1, 2). Use of at least 1 of 16 CAM interventions in the preceding year increased from 33% of the U.S. population in 1990 to 42% in 1997, and the percentage of individuals visiting a CAM practitioner increased from 36% to 46% during this time period (3). CAM is broadly defined as medical practices that are neither generally taught in medical schools nor are widely available in U.S. hospitals (3). CAM includes a wide variety of modalities typically not considered a part of conventional medicine, such as acupuncture, herbs, exercise/yoga, and relaxation therapies (i.e., reiki, meditation, or massage) (4). CAM is more often used by women, younger individuals, and those with higher educational and income levels (3, 5, 6). CAM use is typically higher among cancer patients than individuals in the general population, with reports ranging from 63% (7) to 83% (8) of cancer patients indicating use of at least one type of CAM. Breast cancer survivors have been found to often use dietary CAM (i.e., soy products) for menopause symptom management (9) and to use certain herbs (i.e., St. John's Wort) for easing of emotional distress (10).

High-risk women undergoing BRCA1/2 testing face the possibility of confronting difficult health maintenance decisions (11). Women who test mutation positive must consider, for example, when or whether to undergo prophylactic mastectomy and/or oophorectomy (12). In an earlier study, we found that 50% of our cohort of high-risk women used CAM as part of their on-going health maintenance program (13). In this study, we sought to explore the types of CAM high-risk women use for cancer prevention following receipt of BRCA1/2 genetic test results, as well as whether women alter their CAM-related practices in the year following genetic testing. In addition, we assessed the association between number of CAMs used at 1 year follow-up with use of other health behaviors, as well as any relationship between CAM use and distress, breast cancer risk perception, and BRCA1/2 mutation status.

Six cancer risk assessment clinics or clinical oncology practices throughout New England participated in a National Human Genome Research Institute–funded randomized study of two genetic counseling interventions between December 1, 1998, and July 1, 2001. A total of 442 women who presented for cancer risk evaluation and met the protocol eligibility criteria were offered enrollment at the study site of their choosing. Of that number, 310 women (70%) enrolled and 132 women (30%) declined participation in the program. Sixteen women with breast cancer were excluded from this analysis because they reported either ongoing cancer treatment or recurrent or metastatic disease. This was done to minimize the possibility that reported CAM use might be intended as part of active cancer treatment rather than for cancer prevention. Additionally, 51 women did not complete the 1-year questionnaire. Thus, this paper is based on a sample of 243 (83%) women from a total of 216 families who were not receiving treatment for active breast cancer and who completed both the baseline and 1 year post disclosure follow-up questionnaires.

Participants were required to meet the following eligibility criteria: (a) female; (b) age 18 years or older; (c) personal or family history of breast, ovarian, or other cancer consistent with BRCA1/2 heredity with posterior probability of carrying an altered gene of ≥10% based on published probabilities and Bayesian calculations; and (d) documentation of participant or family member cancer diagnosis. Individuals who met the eligibility criteria were mailed an enrollment packet, which contained the baseline questionnaire, informed consent, medical insurance information form, and project brochure. On receipt of the completed enrollment packet materials at the study center at the Dana-Farber Cancer Institute, project staff enrolled and randomized each participant to receive either formal genetic counseling conducted by a genetic counselor or an enhanced consent intervention administered by a medical oncology nurse. (A comprehensive comparison of the two interventions will be reported separately.) Participants were required to attend two visits at their study site, and received their BRCA1/2 test result during their second appointment. Structured standardized interview guides were designed for both the genetic counseling and enhanced consent arms of our study. The genetic counseling intervention was designed as a client-centered model, a psychological approach in which the use of scenario-based decision counseling taught as part of genetic counseling masters degree programs was encouraged. The enhanced consent intervention was adapted from the medical model of informed consent, in which oncology nurses reviewed the informed consent document section by section, soliciting questions after each. No specific discussion or recommendations regarding any aspect of CAM use were included in the structured components of the genetic counselor, nurse, or physician sessions. Women were followed for 1 year after receiving their genetic test results. A federal Certificate of Confidentiality was obtained to prevent undesired disclosure of personal information. The Institutional Review Board at the Dana-Farber Cancer Institute and participating institutions approved this study.

Demographic information, including cancer history, was collected from all interested study participants before the mailing of the enrollment packet. Completion of the enrollment questionnaire was required for entry into the program; therefore, the response rate for this survey is 100% of program enrollees. Questionnaires were also completed 1 year following results disclosure by 83% of the sample. The enrollment, 1 year follow-up, and demographic instruments provide the data sources for this paper.

Complementary Medicine Use. To assess CAM use at baseline enrollment into the program and 1 year following genetic testing, participants were asked, “Which of the following complementary therapies have you used to prevent cancer?” We derived a modified list from early CAM instruments designed by Eisenberg et al. (1) and Richardson et al. (8), consisting of vitamins, special diet/nutrition, herbal remedies, exercise, meditation/yoga, massage therapy, energy work (i.e., polarity, reiki), and acupuncture/acupressure. To compare overall change in CAM use from baseline enrollment to 1 year follow-up, participants were classified as using “any CAM” if they reported using at least one therapy of the eight modalities specifically for cancer prevention. Two additive scales of the number of CAMs used at enrollment and 1 year were also created (range 0-8) to assess association between participant characteristics and number of CAMs used at 1 year. Relationship of the covariates to individual CAMs was also explored.

Patient Characteristics. Standard demographic and health history information was obtained and dichotomized for this analysis: cancer history (cancer versus unaffected), age (46 years or less versus older than 46 years), education (less than college versus college or more), ethnic derivation (Ashkenazi versus non-Ashkenazi Jewish), annual household income (<$65,000 versus $65,000 or more), and randomization intervention (genetic counseling versus oncology nurse consent).

BRCA1/2 Mutation Status. BRCA1/2 test results were classified as positive, true negative (negative test for a documented familial mutation), uninformative negative (negative test in the absence of a documented familial mutation), and genetic variant of uncertain significance. Because most variants have been shown to have no functional significance, for the purposes of this analysis, genetic test results were dichotomized into “positive” versus “other.”

Cancer-Related Health Practices. Health surveillance behaviors may have an influence on cancer outcome and may be related to CAM use (13, 14). Participants were asked several questions regarding their cancer-related health behaviors 1 year following disclosure. The following items were selected for this analysis due to the variability of these behaviors among this sample—“How many clinical breast exams (CBE) have you had in the last 12 months?” Responses to this item ranged from 0 to 2 or more—with two or more indicating medical adherence for women with high breast cancer risk. “Do you see a dermatologist annually for skin exams or to have moles screened for signs of cancer?” Responses were coded as “yes at least once per year;” “yes less often than once per year;” “only when I have a problem;” or “no.” “Do you typically wear sunscreen with an SPF of 15 or more when you are in the summer sun for >15 minutes (15)?” Responses were coded as “always,” “often,” “sometimes,” “rarely,” and “never.” Alcohol and cigarette use were explored in preliminary analysis, however, not included in final analysis due to low prevalence of these behaviors among our sample. Participants were not asked about their frequency of obtaining pap tests or colorectal cancer screening.

Anxiety/Depression. We and others have reported a preliminary relationship between increased depression and CAM use (13, 16). To further explore this association, we used the standardized depression and anxiety subscales of the 53-item Brief Symptom Inventory (17). Participants were asked about their emotional well being in the past 7 days, with response categories on a 5-point Likert scale. Several standardized subscales were calculated, including indices of anxiety and depression, which were dichotomized at the sample medians into “lower anxiety score” and “higher anxiety score” and “lower depression score” and “higher depression score.”

Risk Perception. Our cohort was asked to provide a numerical response to the following item separately for breast and ovarian cancer: “On a scale from 0% to 100%, where 0% is no chance at all and 100% is absolutely certain, what do you think are the chances that you will develop (another) breast/ovarian cancer at some point in your lifetime?” For the purposes of this analysis, responses were dichotomized at each of the sample medians for breast and ovarian cancer.

Statistical analyses were conducted using SAS software (18). Participant characteristics comparing CAM users and nonusers were explored using cross-tabular analysis and Fisher's exact tests (19). Cross-tabular comparison of change in CAM use from baseline to 1 year follow-up was made using McNemar's test, which analyzes for change in proportion between two samples (19). All P values reported were two-tailed, with a level of P < 0.05 to determine statistical significance. Covariates with a P < 0.20 in preliminary univariable regression analysis were entered into multivariable linear regression models to explore their relationship with number of CAMs used at 1 year for cancer prevention. The stepwise method was used for variable selection. Randomization assignment was controlled for in all analyses. Cancer survivors and unaffected participants were analyzed separately for all analyses. Exploratory analysis was conducted to investigate the correlation between the individual CAMs and selected covariates. The fact that the large majority of families (n = 271) had only one participant precluded the use of generalized linear models to control for any potential effects of having some families with more than one participant.

No significant differences were found in standard demographic indicators between CAM users and CAM nonusers with regard to age, Jewish ancestry, education, or annual household income (Table 1). All participants were Caucasian. At enrollment into the program, 53% of CAM users and 46% of nonusers had had breast cancer; 6% of CAM users and 6% of non-CAM users had had ovarian cancer. Twenty-eight percent of CAM users and 20% of nonusers were mutation positive.

A total of 54% of our overall sample reported using any CAM use for cancer prevention at program enrollment and 62% reported any CAM use for cancer prevention 1 year following BRCA1/2 results disclosure. The number of CAMs used for cancer prevention by our cohort increased from baseline to 1 year follow-up, with 32% of our sample reporting use of two or more CAMs at baseline, compared with 41% using two or more CAMs at follow-up. The most common type of CAMs reported at 1 year follow-up were as follows: vitamins (40% unaffected women versus 48% cancer survivors), exercise (38% unaffected versus 41% survivors), and special diet (24% unaffected versus 21% survivors; Table 2). No relationship was found between BRCA1/2 mutation status and these individual CAMs.

Among women who changed their CAM behavior from baseline enrollment to 1 year following BRCA1/2 results disclosure, we found that there was a significantly higher proportion who changed from no CAM use to CAM use among the overall cohort (17%, P = 0.01, McNemar's test), among women without cancer at enrollment (22%, P = 0.003), among BRCA1/2 mutation carriers (21%, P = 0.03), and among women randomized to the genetic counseling intervention (22%, P = 0.009; Table 3).

Multivariable linear stepwise regression models were conducted on the entire cohort (n = 243), and separately for each cancer status, to explore the relationship of the following predictors with number of CAMs used for cancer prevention 1 year following genetic testing: number of CAMs used at baseline, randomization intervention, age, education, ethnic derivation, BRCA1/2 mutation status, cancer-related health behaviors [clinical breast examinations (CBE), dermatologist visits, and sunscreen use], anxiety, depression, and breast cancer risk perception (Table 4). We also explored the relationship to CAM use of the following items in preliminary analysis—frequency of mammography use, breast self-examinations, and ovarian cancer risk perception—no relationships were found with these items and, therefore, not pursued in final analysis.

In all cases, we found that the number of CAMs used at baseline was significantly associated with the number of CAMs used at 1 year follow-up (P < 0. 001). No relationship was found between any of the demographic indicators (age, education, income, or ethnic derivation) or randomization intervention and 1 year CAM use. For the entire sample, there was a significant association between having a positive BRCA1/2 mutation and the number of 1 year CAMs used (P < 0.04). In exploring the relationship of 1 year CAM use to cancer-related health behaviors, we found a significant association between greater sunscreen use and CAM use for the entire sample (P < 0.007) and cancer survivors (P < 0.001). For unaffected women, an inverse relationship was found between number of annual CBEs reported and number of CAMs used at 1 year (P < 0.004). Whereas preliminary univariate analysis did find an association between a greater number of dermatologist visits and the number of 1 year CAMs used, this relationship was not substantiated in the multivariable models. Additionally, for unaffected participants, we found a significant association between greater levels of anxiety and 1 year CAM use (P < 0.05). An inverse relationship was discovered between breast cancer risk perception and 1 year CAM use (overall sample P < 0.001, survivors P < 0.009, and unaffected participants P < 0.09). We further explored the negative relationships of annual CBEs and breast cancer risk perception to CAM use and found a positive although nonsignificant relationship of CBE to two of the individual CAMs—vitamin use (P < 0.24) and special diet (P < 0.13; data not shown). We found significant negative correlations between breast cancer risk perception and two of the individual CAMs—massage therapy (P < 0.02) and energy work (P < 0.003).

This study explored whether high-risk women modify their CAM use from entry into a genetic testing program for breast/ovarian cancer susceptibility to 1 year following results disclosure. In addition, we assessed predictors of the number of CAMs use for cancer prevention at 1 year follow-up, including CAM use at baseline, health behaviors, anxiety, BRCA1/2 mutation status, and breast cancer risk perception. As is typical for women undergoing genetic testing for BRCA mutations, all women in our cohort were Caucasian and the majority had at least a college education and an annual income of $65,000 or more (12, 20, 21). Fifty-four percent of the overall cohort used any CAM for cancer prevention at program enrollment, increasing to 62% at 1 year follow-up. Among women who changed their CAM behavior from baseline enrollment to 1 year following BRCA1/2 results disclosure, there was a significantly higher proportion that began to use CAM among the overall cohort, among unaffected women, among BRCA1/2 mutation carriers, and among women randomized to receive genetic counseling. For all participants, CAM use at baseline was significantly associated with 1 year CAM use. Overall, a relationship was found between certain cancer-related health surveillance behaviors and 1 year CAM use and BRCA1/2 mutation status and CAM use. For unaffected women, there was a significant association between greater levels of anxiety and CAM use, and a negative relationship between annual CBEs and CAM use. For all participants, a negative association was found between breast cancer risk perception and CAM use. We did not find a relationship in our data between CAM use and ovarian cancer risk perception, frequency of mammography use, or breast self-examinations.

Women at high genetic risk tend to practice better health behaviors than average-risk women (11). This pattern was reflected in our cohort—only 3% reported that they currently smoked, whereas 54% regularly engaged in strenuous exercise, such as running, swimming, or aerobics, and 71% consistently protected themselves while in the sun. U.S. general population data on women report higher smoking rates (21%; 22), less regular physical activity (40%; 23), and less regular sunscreen use (33%; 24) than our high-risk women. Lifestyle incorporation of cancer prevention health practices after genetic testing is further reflected in our cohort with unaffected women, mutation carriers, and recipients of the genetic counseling intervention significantly reporting new CAM use 1 year following genetic testing. Although standardized interview guides were used for women randomized to both study arms, the genetic counseling intervention was structured as the more client-centered, psychological approach. Although the interview guides were not specifically designed to elicit discussion of CAM, it is plausible that this topic may have arisen as an area of interest during discussion of cancer-related lifestyle choices. Our findings indicate that high-risk women might undergo health behavior changes after receiving genetic test results. This may be an important moment in which to influence modifiable health practices, particularly in pursuit of healthier lifestyles.

Some studies have shown that modifiable behaviors, such as physical activity, diet, and body mass index, may be associated with breast cancer risk for high-risk women. King et al. (25) documented a significant relationship between increased physical activity, lack of obesity, and decreased breast cancer risk for women with BRCA1/2 mutations. Carpenter et al. (26) explored the effect of family history, exercise, and obesity on postmenopausal breast cancer risk, and found that breast cancer risk significantly increased with higher body mass index levels for women with a strong family history. Although it was previously thought that certain diets, such as those higher in soy and other phytoestrogens, might decrease breast cancer risk, those relationships remain unclear (27). Further, several studies have concluded that soy-related products might not be beneficial in reducing hot flashes for menopausal or postmenopausal breast cancer survivors (28-30). However, it is noteworthy that Fung et al. (31) did report a significant relationship between higher consumption of fruits and vegetables and a decreased risk for estrogen receptor–negative breast cancer.

BRCA1/2 mutation carriers are typically faced with making unique health care decisions for cancer risk management (12). Some high-risk women may elect chemoprevention or prophylactic surgical options and might also include regular practices, such as mammography, breast MRI, or breast ultrasound, as part of their cancer screening regimens (32). Meijers-Heijober et al. (33) reported that 55% of mutation carriers from their program in the Netherlands underwent a prophylactic mastectomy and 60% had a prophylactic oophorectomy within 2 years after genetic testing. In contrast, Lerman et al. (12, 34) found that many BRCA1/2 carriers opted not to undergo prophylactic surgery in spite of their increased cancer risk. Botkin et al. (34) documented few prophylactic mastectomies in their cohort of mutation carriers; however, 46% of carriers had undergone a prophylactic oophorectomy by 2 years postdisclosure. Whereas it is possible that our study was not significantly powered to detect an effect, we did not find any relationship between choosing prophylactic surgery and CAM use at 1 year follow-up for unaffected mutation carriers. The association in our data between other cancer-related health choices and CAM indicates that healthy high-risk women may elect to optimize cancer risk management options other than prophylactic surgery to minimize mutation-associated cancer risk.

Cancer risk behavioral decisions made by high-risk women are clearly complex and multifaceted. However, healthy nutrition and exercise guidelines for women at high cancer risk, as well as for the general population, have become standard (35). Lovegrove et al. (36) explored factors that may be important in cancer-related behavioral decisions made by high-risk women. The authors found that half of their cohort who had overestimated their cancer risk and who declined participation in a chemoprevention trial were significantly more aware of traditional lifestyle factors thought to play a role in cancer risk reduction, including diet and exercise, than those who enrolled in the trial. In our study, we were puzzled by the negative correlation between CAM use and risk perception and CAM use and annual CBEs for unaffected women. To further investigate these findings, we separated the eight-item CAM scale into individual CAMs, which allowed us to explore the relationship of the covariates with the more traditional health behaviors—exercise, nutrition, and diet—separately from the more alternative approaches, including massage and energy healing. Although we did not have sufficient power to detect strong differences, the negative association between risk perception and the more traditional health behaviors did not persist, but it was maintained for more alternative activities. Perhaps women with higher-risk perception may be so apprehensive about their cancer risk that they are not willing to experiment with newer approaches and may be more comfortable with strategies such as vitamins, nutrition, and exercise, which have been associated with cancer risk reduction. Similarly, it is possible that women who maintain more frequent cancer surveillance practices, such as obtaining two or more annual CBEs, are more interested in more established CAM modalities. These observations merit further investigation.

Depression and anxiety have been found to affect health behavior choices made by women with high cancer risk (37). CAM use may help reduce stress in individuals facing chronic diseases (3, 38). In our earlier study, we reported an association between higher levels of depression and CAM use for cancer survivors at baseline entry into our BRCA1/2 testing program (13). Our current follow-up data do not reflect a relationship between depression and CAM use for either cancer survivors or unaffected women. However, we did find an association between anxiety and CAM for unaffected women at 1 year follow-up, indicating that healthy women may seek CAM as one strategy to help manage stressful issues arising from receiving genetic testing information.

Some limitations to our study should be noted. First, previous studies have shown that socioeconomic status is significantly correlated with CAM use (3, 5, 6). The majority of our cohort had high educational and annual income levels, as has been typical for many studies of women receiving BRCA1/2 testing, limiting the generalizability of our findings (39). Second, our study instruments were designed in 1998 when CAM research was emergent. Thus, our CAM categories—although based on the best available data at the time—provided us with limited detail on the types of CAM used. For example, it would have been helpful to know the types of exercise, special diets, and vitamins used by our cohort for cancer prevention. Additionally, our cohort was not asked items that have become increasingly relevant in the exploration of CAM use, including when CAM use began or motivations for using CAM. This information would have been particularly helpful in interpreting our results, especially with regard to whether cancer survivors began using CAM before or after their cancer diagnosis. It should be noted that with regard to cancer survivors, we did explore the relationship between CAM use and length of time since cancer diagnosis (<5 years ago versus 5 years or more), and CAM use and age at cancer diagnosis (younger than age 46 years versus age 46 years or older); we did not find a relationship between these items. Further, we would have explored predictors of CAM use ideally by creating a high-risk breast management guideline adherence scale. However, we felt that this was not feasible based on our available data. At the time our study was designed, breast-imaging techniques were not yet routinely recommended for high-risk women as reflected in our data (conducted regular breast MRI and breast ultrasound 4%, respectively). Mammography adherence also became difficult to explore due to the fact that 40 women in our cohort had no breasts remaining by 1 year postdisclosure. To delete this group from the analysis would have resulted in significant loss of sample size. We opted instead to focus on CBE as an indicator of breast management adherence due to the fact that there was sufficient variability in this item to explore its relationship to CAM use. Finally, due to the lack of association between CAM use and ovarian cancer risk perception in our data, we were unable to fully explore the effect of CAM use on ovarian cancer risk management specifically. This remains an area needing further exploration.

Although there is minimal scientific information on the true role of CAM-related approaches in cancer prevention (40), our study indicates that high-risk women may adopt new CAM behaviors following genetic testing. As high-risk women consider their nonsurgical cancer risk reduction options, such as increased physical activity, healthy body weight, and other behavioral strategies, providers may consider encouraging women to participate in related chemoprevention trials. Providers should be aware that high-risk women may already be using CAM-related behaviors when reviewing their approaches for coping with their cancer risk and should be prepared to address available data on the advisability or inadvisability of these practices. Further, randomized nutrition and physical intervention studies for high-risk women are warranted to assist in determining the actual effectiveness of these strategies for cancer risk reduction.

We thank the following study investigators: Caroline Block, M.D., Kevin Hughes, M.D., Joel Schwartz, M.D., Michael Seiden, M.D., Tracey Weisberg, M.D.; genetic counselors: Anu Chittenden, M.S., Katherine Schneider, M.P.H., Kristen Shannon, M.S.; project management: Kylie Smith, M.S.; psychological components: Nancy Borstelmann, LICSW, M.P.H., Michel Dorval, Ph.D., Andrea Patenaude, Ph.D.; study participants; and David S. Rosenthal, M.D., Director, Zakim Center for Complementary Medicine at the Dana-Farber Cancer Institute, for his careful review of the manuscript.