Quality of Life in a Randomized Breast Cancer Prevention Trial of Low-Dose Tamoxifen and Fenretinide in Premenopausal Women

Chemoprevention has a tremendous potential, especially for breast and colon cancer (1, 2). Assuming for any treatment the oft-quoted Hippocratic dictum “first do no harm,” this is a must in the field of preventive medicine, and evaluation of the resulting quality of life (QoL) is of paramount importance. Strategies to improve QoL and prompt management of side effects are essential for preventive program acceptability and adherence (3).

In 1999, the FDA approved tamoxifen for primary prevention of breast cancer. Later, raloxifene and then aromatase inhibitors (AI) proved their efficacy in postmenopausal women at high risk for breast cancers (4–6).

Despite the strong risk reduction, selective estrogen receptor modulators and AIs are not well accepted by high-risk women for the possible impact on QoL and adverse events (7). In parallel, even physicians may fail to recognize and/or advise high-risk women of such preventive opportunities (8).

Menopausal symptoms (hot flashes and night sweats) were a main reason for full-dose tamoxifen treatment withdrawal in the IBIS I trial (9). Maintaining a high QoL is therefore crucial to implement therapeutic prevention on a large scale. We have investigated for several years a strategy to reduce side effects and retain efficacy by lowering the dose or using an intermittent schedule of tamoxifen and or combining treatments to neutralize side effects (10–12). At the biological level, we demonstrated an equivalent efficacy of 5 or 1 mg day relative to 20 mg day in reducing Ki-67 (13) or modulating circulating biomarkers and mammographic density (10, 11). Moreover, a lower dose compared with the standard dose, used as cancer therapy, can be more appealing to healthy women at increased risk. At the clinical level, tamoxifen at 5 mg was studied in a phase III prevention trial in women taking hormone replacement therapy with a 20% nonsignificant reduction of breast cancer incidence overall and a significant 68% reduction in luminal A breast cancers (12). Furthermore, an indirect comparison of tamoxifen side effects between the standard dose and the low-dose tamoxifen can be made across the two phase III prevention studies conducted by our group (12, 14). A trend toward a lower symptom rate has been shown by low-dose tamoxifen, with an annual rate of grade 2 to 3 hot flashes of 6.5%, 9.7%, and 11.9% and vaginal discharge of 1.2%, 2.1%, and 6% for placebo, tamoxifen 5 mg and 20 mg, respectively.

The Menopause Quality of Life (MenQoL) questionnaire is validated self-administered instrument to evaluate the QoL for climacteric symptoms (15, 16). The purpose of the current study was to evaluate the QoL of women participating in a chemoprevention trial through the self-administered MenQoL. The study, a 2 × 2 factorial breast cancer prevention trial, included 235 premenopausal women with an increased risk for breast cancer who were randomized to either tamoxifen at a lower dose, fenretinide, the combination, or placebo for 2 years, with 1 year of follow up. The primary endpoint biomarkers were the changes in IGF-I and mammographic percent density values from baseline to 24 months (10).

A total of 235 high-risk premenopausal women were randomly allocated to either tamoxifen 5 mg/daily, fenretinide 200 mg/daily, the combination, or placebo in a double-blind and double-dummy fashion. The study was conducted under the approval of the European Institute of Oncology (IEO) Review Board, and all the participants signed an inform consent. The study duration was 2 years of treatment and 1 year of follow-up. Risk categories were surgically removed intraepithelial neoplasia (IEN) or pT1mic breast cancer (160 and 21 women, respectively) or increased risk based on the Gail model (54 women). A detailed description of the study has been previously reported (10).

The aim of the present study was to evaluate the modification in QoL in the different intervention groups using a validated questionnaire specific for climacteric symptoms, namely the MenQoL questionnaire. The MenQoL intervention questionnaire is based on 29 items divided into four domains (vasomotor, physical, psychosocial, and sexual). Each item is scored from 1 to 8: 1 means no symptom, 2 indicates presence of the symptoms but not bothersome, 3 to 8 mean an increasing grade of discomfort. The vasomotor domain includes three items to investigate hot flashes and sweating. In the physical domain, there are 16 items covering general symptoms, skin, gastrointestinal, sleeping problems, and urinary symptoms. The psychosocial domain includes seven items, evaluating states of anxiousness, memory, and loneliness. The sexual domain has three items: sexual desire, vaginal dryness, and avoiding intimacy (15).

Women were clinically assessed at baseline and every 6 months up to the 36th month. The trial received Institutional Review approvals and was conducted in accordance with Good Clinical Practice procedures. All subjects gave their written-informed consent. Adverse events were assessed using the National Cancer Institute Common Terminology Criteria of Adverse Events (version 4.0). Compliance was measured by pill count and circulating drug levels.

Genomic DNA was extracted from whole-blood specimens with a QIAamp DNA blood kit (Qiagen). The INFINITI analyzer (AutoGenomics) was employed for CYP2D6 genotyping, according to the manufacturer's protocol as described elsewhere (17).

We classified the resulting genotype as poor metabolizers (PM) when there were nonfunctional variants on both alleles; intermediate metabolizers (IM) if they carried two reduced function alleles or one reduced and one nonfunctional allele; extensive metabolizers (EM) when carrying at least one fully functional allele; and ultrarapid metabolizers (UM) when duplication of normal alleles was present. For the statistical analysis, usually the four genotypes are grouped in two groups (EM + UM genotype as rapid metabolizers and IM + PM genotype as slow metabolizers, as previously described; ref. 17). Because there were no UM subjects in this study, comparison is between extensive and slow metabolizers.

Percentages of symptoms of each domain at baseline and three time points are represented graphically by trial arms and CYP2D6 genotypes.

In the remaining analyses, data of the MenQoL questionnaire were preliminarily summarized, within each patient at each assessment time, by computing the mean score over distinct items of each domain. For descriptive purposes, such means were used to compute average scores of each domain according to time and treatment arm. Furthermore, to assess the clinical significance of 12, 24, and 36 months versus baseline differences, we determined the effect size as M_{12 month} – M_baseline)/SD _baseline, where M is the average (at baseline or 12 months) and SD is the standard deviation (at baseline).

Differences in frequencies were assessed by χ² tests and Mantel–Haenszel χ² tests. Differences in median values of continuous variables were assessed by the Wilcoxon rank tests.

Multivariate mixed models were used to assess differences in effect sizes between treatment arms, considering the patient effect as a random factor and including as fixed factors the treatment arms (tamoxifen, fenretinide, tamoxifen+ fenretinide, or placebo) and time (baseline, 12, and 24 months). Exploratory ANCOVA analysis was carried out adjusting for confounding factors such as age, body mass index (BMI), change in menopausal status during time, as well as baseline values.

A total of 235 subjects were randomized to either tamoxifen 5 mg day (N = 58), or fenretinide (N = 59), or fenretinide and tamoxifen (N = 60), or placebo (N = 58) in a double-blind, double-dummy fashion. Baseline characteristics are shown in Table 1, and more details are published elsewhere (10). Even though all women were premenopausal, several women already had some climacteric symptoms. Within the vasomotor domain, 38% of women in the placebo group, 45% in the tamoxifen and the combination group, and 53% in the fenretinide arm developed at least one symptom, with no statistical significant differences among arms. By the end of the 2-year treatment, 34% of women became postmenopausal, justifying a physiologic symptoms increment.

MenQoL was administered at each visit and analyzed at baseline, 6, 12, and 24 months, while subjects were on treatment, and then after 12 months from treatment completion.

Table 2 shows the MenQoL effect size score per arm for each domain (vasomotor, physical, psychosocial, and sexual). Effect sizes, that estimate changes from baseline, show no different effects in the tamoxifen arms compared with no tamoxifen arms. In particular, for the vasomotor domain, we can see a nonsignificant increment in the mean score from 1.03 at one year up to 1.21 at the second year and a decline to 0.70 in the follow-up year for the tamoxifen arm. Similar results were found for the combination of tamoxifen and fenretinide. Fenretinide and placebo had a parallel trend with a steady increment in time irrespective of treatment.

Figure 1 shows the mean score for each domain by treatment arms. For the physical, psychosocial, and sexual domains, curves were superimposable and stable overtime. The curve shapes for vasomotor symptoms were also similar among arms but with an increasing trend overall.

The analysis was also conducted combining the two arms with tamoxifen versus no tamoxifen, and again no differences were observed for all domains. For instance, the percentage of women with hot flashes at baseline, 12, 24, and 36 months in the tamoxifen groups versus no tamoxifen was, respectively, 20% versus 19.3%, 44.9% versus 33.3%, 47.6% versus 39.2%, and 38% versus 50.5%. Likewise, the vasomotor symptom score in the two groups did not show significant difference at any time points.

We further looked at bothersome symptoms, namely, the percentage of women experiencing bothersome symptoms with score ≥3 for each domain by arm (Fig. 2). Again, no statistically significant differences were detected with a superimposable trend on the physical, psychosocial, and sexual curves. The curves for bothersome vasomotor symptoms showed a slight nonsignificant tendency to a higher rate in the combination arm at 24 months.

Exploratory analyses were conducted to correlate the MenQoL score with adherence to the treatment. Within the study, a total of 37 subjects dropped out, 9 in the tamoxifen arm, 8 in the fenretinide arm, 10 in the combination, and 10 in the placebo group. The drop-out rate was not associated with a worse MenQoL score (data not shown). Furthermore, we investigated whether the CYP2D6 genotype was associated with symptom occurrence in the tamoxifen arms (Fig. 3). Out of 109 subjects taking tamoxifen, 92 were EM and 17 slow metabolizers (10 intermediate and 7 PMs). No ultrarapid metabolizer was found. EM CYP2D6 genotype was associated with higher percentage of bothersome symptoms (score ≥3) in the vasomotor, psychosocial, and sexual domain (P value = 0.01, 0.007, and 0.007, respectively).

Despite the positive results, the uptake of tamoxifen in high-risk women is below 1% (7). Several factors may contribute to the low uptake of tamoxifen, among which the increase of menopausal side effects is of prominent importance (8, 18).

To address QoL from the women's prospective, within a randomized phase II prevention trial, we used a self-administered questionnaire “MenQoL,” which focuses on climacteric symptoms. In this study, tamoxifen was administered at 5 mg per day, and the menopausal symptoms related to tamoxifen did not significantly exceed the fenretinide or placebo. The percentage of subjects reporting vasomotor symptoms and bothersome vasomotor symptoms seems to increase more rapidly in the two tamoxifen-containing arms, but the difference was not significant. By the end of study, we observed an increased symptom score in all four arms, probably due to the physiologic aging of the participants, as 34% of the subjects became postmenopausal while on study. For all other domains, the percentage of subjects reporting any symptoms was overlapping among arms.

It has been reported that not only acceptability but also adherence to the intervention can hamper the implementation of a widespread preventive intervention program. Some studies showed that nearly 50% of subjects who began a prevention regimen as well as an adjuvant therapy with tamoxifen did not finish the 5 years of treatment (19, 20). In our trial, the compliance was much higher, and the claimed discomfort reported on the questionnaire was not associated with dropout. A possible explanation is that within a relatively small clinical trial such as ours, it is easier to maintain a high retention rate by developing personalized counseling with each individual in the study.

Recently in the NSABP P-1 trial, it was reported that the mental component of QoL has more impact on treatment adherence compared with the physical component (3). A prompt symptom management and safety reassurance by the physician helped women to be motivated to adhere to the treatment. The recent analysis from the IBIS I study showed a higher drop-out rates in the tamoxifen arm, but the adherence did not differ between tamoxifen and placebo considering the overall reported symptoms (9). This leads to a further issue, i.e., the nocebo effect, that might be crucial in a preventive treatment. It is a difficult topic to study for the duty to inform properly the participants on the possible side effects and the widespread use of the Internet to get more information, which makes crucial how the physician presents the possible side effects to the participant (21).

CYP2D6 genotype modulates the level of endoxifen, the main tamoxifen metabolite (22), but its clinical relevance is still controversial (23). Also tamoxifen side effects and CYP2D6 genotype showed contradictory results; in the IBIS I and in the TADE studies, CYP2D6 genotype showed no correlation with vasomotor symptoms (24, 25), whereas the BIG 1-98 study showed an association of IM and PM with an increased risk of hot flashes (26), and an opposite trend was reported by Goetz and colleagues in their study (27). These results are not truly comparable because there is no homogeneous categorization and methodology across studies. Our findings support the hypothesis that EM may have a higher incidence of menopausal symptoms on tamoxifen, but these symptoms did not affect adherence to tamoxifen. Because we did not test CYP2D6 genotype in the no-tamoxifen arms, we are unable to assess if the CYP2D6 genotype “per se” has an effect on menopausal symptoms.

The intersubject variability of tamoxifen and its metabolite levels suggests a flexible therapeutic window. Prior studies also showed the tamoxifen antagonist/agonist effect may have a better ratio at low dose (13, 28).

Fenretinide, a synthetic retinoic acid derivative, has not been related to menopausal symptoms, and its profile overlapped the placebo arm for the symptoms related to the MenQoL questionnaire.

Our experience with low-dose tamoxifen is quite encouraging based on its biological, clinical, and safety profile (10, 12, 29). Concerning menopausal symptoms, low-dose tamoxifen exhibits a good safety profile and tolerability in pre/perimenopausal women, which is thus reassuring regarding the use of this agent for breast cancer risk reduction. An ongoing phase III clinical trial of low-dose tamoxifen in women with a resected breast IEN (Ductal Carcinoma In Situ Lobular Carcinoma In Situ [DCIS/LCIS]) will definitely address its clinical efficacy and safety (30). Our findings in a large observational study of DCIS (31) show that low-dose tamoxifen is more acceptable, and adherence is greater in patients with a histologic report of a premalignant disease, where the nocebo effect is lower.

No potential conflicts of interest were disclosed.

Conception and design: B. Bonanni, A. DeCensi

Development of methodology: D. Macis, B. Bonanni, A. DeCensi

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): D. Serrano, A. Guerrieri-Gonzaga, H. Johansson, D. Macis, V. Aristarco

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): D. Serrano, S. Gandini, A. Guerrieri-Gonzaga, A. DeCensi

Writing, review, and/or revision of the manuscript: D. Serrano, S. Gandini, A. Guerrieri-Gonzaga, H. Johansson, D. Macis, B. Bonanni, A. DeCensi

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): A. Guerrieri-Gonzaga, I. Feroce, H. Johansson, D. Macis

Study supervision: B. Bonanni, A. DeCensi

Other (study funding): A. DeCensi

This study was supported by NCI grant no. CA-77188, a regional grant no. 1068/2005 on second tumors from the Associazione Italiana per la Ricerca sul Cancro (AIRC), and Italian Ministry of Health.

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.