Study Population

Participants were recruited from the Breast Oncology Clinic at the Dana-Farber Cancer Institute between May 2004 and June 2006. Eligibility criteria included histologic evidence of stage I to III invasive breast cancer, completion of any chemotherapy and/or radiation therapy at least 3 months before enrollment, absence of diabetes, no use of corticosteroids, body mass index (BMI) more than 25 and/or body fat percentage more than 30% (Quantum II Bioelectric Body Composition Analyzer; RJL Systems, Clinton Township, MI), and baseline participation in less than 40 minutes of physical activity per week. Hormonal therapy was allowed as long as patients continued the therapy for the duration of the study. Baseline exercise was assessed via interview by study staff. Patients were excluded if they had evidence of persistent or recurrent breast cancer, other malignancy, uncontrolled heart disease, or other contraindications to exercise.

Medical clearance was obtained from potential participants' medical oncologists. The study was approved by the Institutional Review Board at the Dana-Farber Cancer Institute, and informed consent was obtained from participants before enrollment.

Study Design

After enrollment, participants were randomly assigned 1:1 to an exercise intervention group or control group. The intervention group participated in a 16-week exercise intervention. The control group received routine care for 16 weeks and was then offered consultation with an exercise trainer at the end of the control period. Participants were stratified by menopausal status, which was defined by the presence or absence of a menstrual cycle within the past year, at the time of study entry.

Exercise Intervention

The exercise intervention consisted of a 16-week supervised strength training and home-based cardiovascular training protocol. Participants completed two supervised 50-minute strength training sessions per week and were asked to complete 90 minutes of home-based aerobic exercise weekly. All participants were asked to avoid changes in dietary habits for weight loss purposes for the duration of the study. See the Appendix (online only) for a detailed description of the exercise intervention.

Measurements

Demographic data, disease and treatment information, and baseline physical activity information were collected via interview before participant enrollment. All disease and treatment information was confirmed through review of medical records. Participants underwent a series of anthropometric measurements at the time of study enrollment (baseline), and these were repeated at the completion of the 16-week study period by study staff who were not blinded to group assignment. For women in the intervention group, this final sample was obtained 3 to 5 days after the last weight training session, and women were asked to avoid strenuous activity during this time period. For participants in the control group, the second set of measurements was obtained 15 to 17 weeks after study enrollment.

All measurements were collected after a 12-hour, overnight fast. Body weight and height were measured using a calibrated Scale-Tronix scale (Scale-Tronix, White Plains, NY), with participants wearing street clothes and no shoes. These data were used to calculate BMI using the following formula: BMI = weight (kg)/height (m)². Waist circumference was measured at the bending line, and hip measurement was recorded at the point of maximum girth. Waist-to-hip ratio was calculated as follows: waist-to-hip ratio = waist (cm)/hip (cm). Body composition was measured using a bioelectric impedance analyzer (Quantum II Bioelectric Body Composition Analyzer; RJL Systems).

Compliance with the exercise intervention was assessed through attendance of strength training sessions and self-reported aerobic exercise. Participants recorded daily minutes of aerobic activity in exercise logs for the duration of the study. Strength was measured by recording the maximal weight lifted for each exercise during strength training sessions.

Biomarkers

Fasting blood was drawn at baseline and 16 weeks for all study participants. Two 10-mL red-top tubes were collected for serum. After the blood was clotted, the samples were centrifuged at 2,000 rpm for 15 minutes at 4°C. Serum was then separated, transferred into cryovials, labeled with freezer-proof labels with participant identification numbers and date, and stored at −70°C until all samples were collected. Insulin and glucose analyses were performed at Quest Diagnostic Laboratory (Cambridge, MA) by technicians blinded to group assignment. All insulin and glucose samples were run in a single batch to avoid interassay variability. Glucose was measured using a hexokinase ultraviolet assay, and insulin was measured using an immunochemiluminometric assay. Insulin was measured in μU/mL (1 μU/mL = 6.954 pmol/L). Insulin resistance was calculated by the Homeostatic Model Assessment (HOMA), with the following formula: HOMA = [insulin (μU/mL) × glucose (mg/dL)]/405.

Sample Size Justification and Statistical Analysis

Previous studies in non–breast cancer populations suggested that enrolling 40 patients per group would give us 80% power with a two-sided α = .05 to detect a mean difference in the change of insulin levels of 1.3 μU/mL between the exercise and control groups, assuming equal variance and a standard deviation of 2.0 μU/mL.¹⁴ We enrolled 50 patients per group to account for an anticipated dropout rate of approximately 20%.

Analyses for the changes in hormonal and anthropometric measurements included 82 participants for whom both baseline and week 16 blood samples and physical measurements were available. Two-sample t tests with the assumption of unequal variance were used to compare baseline and week 16 anthropometric and hormonal measurements in the exercise and control groups, as well as mean changes in these measurements over the study period. The absolute changes in fasting insulin, glucose, and HOMA values were also calculated, and t tests were used to test the null hypothesis that the average change in these measurements in the control and exercise groups was zero. Finally, a multiple linear regression model was used to adjust for baseline insulin level on changes in insulin over the 16-week study period.

Compliance with the exercise intervention was measured by the percentage of scheduled strength training sessions attended in an intent-to-treat fashion, as well as one-sample t tests comparing the weekly minutes of aerobic exercise performed with baseline minutes of activity and with the target goal of 90 minutes of aerobic exercise per week. Finally, percentage of increased weight lifted from the first session to the final session was calculated, using the last-observation-carried-forward procedure for all participants who did not complete the trial.

Exercise Intervention

Fifty-one participants were randomly assigned to the exercise intervention. Although 11 participants ultimately did not complete the intervention, at least partial exercise data were available for 49 participants. Using intent-to-treat analyses, participants attended a mean of 73% of scheduled strength training sessions and performed 114 minutes of aerobic exercise per week. Weekly aerobic exercise was significantly greater than both baseline aerobic activity (11 min/wk) and the study target goal of 90 minutes of aerobic activity per week (P < .0001 for both analyses).

Participants in the intervention group demonstrated an increase in strength for each of the tested exercises between baseline and the final week of the protocol (Table 2). In an exploratory, intent-to-treat analysis (including women who stopped participating in the protocol after only one to two strength training sessions), strength increased by an average of 42% to 106% for each of the exercises.

Physical Measurements

Baseline and week 16 anthropometric data were available for 82 participants (Table 3). At baseline, participants, on average, weighed approximately 80 kg and had a BMI slightly greater than 30 kg/m² and a body composition including 45% fat. After the intervention, participants in the exercise group demonstrated a significant decrease in hip circumference (P = .02) and a trend towards a decrease in waist circumference (P = .06) compared with the control group. Participants in the control group did not experience any significant changes in physical measures.

Insulin and Glucose Measurements

Fasting insulin and glucose measurements were available for 82 patients at both baseline and 16 weeks (40 patients in the exercise group and 42 patients in the control group). Table 4 lists all hormonal measurements. Baseline levels of insulin were slightly higher in the exercise group, with a broader range of values seen among the participants in this group, but these differences were not statistically significant (P = .39). After the 16-week exercise or control period, insulin levels decreased by an average of 2.86 μU/mL (28%) in the exercise group (P = .03) and 0.27 μU/mL (3%) in the control group (P = .65). There was also a marginally significant improvement in insulin resistance in the exercise group after the 16-week exercise intervention (P = .05), with no significant change in controls (P = .81). A comparison of the change in insulin levels across time in the two groups approached statistical significance (P = .07). There was also a trend toward improvements in insulin sensitivity in the exercise group (P = .09), with no change seen in fasting glucose levels. Given the slightly higher baseline insulin in the exercise group, a multiple linear regression model was performed to evaluate the impact of baseline insulin levels on change in insulin. In the multivariate model, baseline insulin level was strongly associated with change in insulin (r = −0.69, P < .001), and the relationship between exercise and change in insulin remained of borderline statistical significance (P = .076).

Exercise Intervention

Participants randomly assigned to the exercise group participated in a 16-week mixed strength and endurance training intervention. For the strength training portion of the protocol, participants engaged in two supervised 50-minute sessions each week. Groups of one to three women worked with a personal trainer during each of these sessions, which took place in a Fitcorp fitness center (Boston, MA) adjacent to Dana-Farber Cancer Institute. The training sessions began with a 5-minute warm-up session of low-intensity walking on the treadmill followed by 10 minutes of static stretching exercises. The strength training program focused largely on lower body and core muscle strength, given the limited data regarding the impact of upper body exercise on the risk of lymphedema. The following exercises were incorporated into each exercise session: leg press, quadriceps extension, hamstring curl, hip adductor, hip abductor, abdominal crunches, calf press, and leg lifts. For exercises performed on weight machines, baseline strength was assessed with a one-repetition maximum test after participants were familiarized with the equipment. For subsequent strength training sessions, initial weight was set at 80% of the maximum weight from the baseline strength testing. Participants were asked to perform two to four sets of 10 repetitions for each muscle group. Once a participant was able to complete all 10 repetitions for two to four sets during two consecutive training sessions, the weight was increased by 10% for the next session.

Participants were asked to perform 90 minutes of cardiovascular exercise on their own each week. Each participant was given a pedometer and heart rate monitor on enrollment. Participants were asked to use the heart rate monitor when exercising and to record the number of minutes per week of cardiovascular exercise and the number of steps per day in an exercise journal. Participants were allowed to choose their own form of exercise, as long as it produced a heart rate in the target zone (55% to 80% of maximum heart rate). The exercise journals were reviewed by the exercise physiologist with the patient on a weekly basis to confirm weekly minutes of activity and to set goals for the next week of the program.