In contrast to BCS who participated in stretching exercise and consistently lost BMD at the lumbar spine over 1 year, women who participated in a program of combined resistance + impact training (e.g., POWIR) maintained BMD at this site. In addition to BMD, the POWIR program caused more favorable changes in bone turnover, resembling that achieved by anti-resorptive therapy [44
]. POWIR did not improve lean or fat mass; however, moderator analysis detected differences between groups that depended upon adjuvant hormone therapy use. That is, women who were in the POWIR group and on AIs had the greatest increase in lean mass, while women in the FLEX group who were not on AIs appeared to have a slight decrease in lean mass.
The efficacy of our resistance + impact exercise program to prevent declines in BMD at the spine is consistent with the results from our previous trial in premenopausal women without cancer [31
]. We are the first to report that exercise can preserve bone mass at the lumbar spine, a clinically relevant fracture site in BCS, though we failed to achieve similar results at the hip. Programs at our achieved training volume have increased hip BMD in premenopausal women when performed 3–5 times per week [31
], but the postmenopausal hip may require a greater load to adapt [45
]. Combined programs of impact plus resistance training better target the spine than the hip in postmenopausal women without cancer [36
]. Studies with conservative jump training programs have not improved hip BMD in cancer-free older women [45
], whereas those employing atypical loading [47
], more jumps or more intense resistance exercise were effective [48
]. The lower intensity of home exercises, lower adherence to the home program and modest jump number in our study may have contributed to an overall suboptimal stimulus at the hip. The hip may also simply require a longer time period to adapt to moderate-intensity training in older women. Snow et al. [32
] reported that a similar exercise program to ours did not affect hip BMD after 9 months, but was later found to have preserved BMD at a 5-year follow-up in women continuing to exercise compared to losses among inactive women. Future studies should consider longer training programs, inclusion of exercises that load the hip in unusual patterns, greater jump number, increased frequency of supervised sessions and/or strategies to maintain training intensity at home.
The POWIR program did not favorably shift body composition; however, our results suggest that current AI use influences changes in body composition in response to resistance training. The ability of exercise of any type to improve body composition in BCS of any age is unclear since results from RCTs tend to be mixed [22
]. Schmitz et al. [50
] reported significant gains in lean body mass among recently treated BCS who strength trained for 6 months, but did not replicate these findings among BCS with or at risk for lymphedema, who also notably did not use AIs [51
]. In our study, we found greater between group differences in lean body mass over 1 year among AI users that suggest a potentiating influence of AIs on training-induced hypertrophy. Recently, AI use has been shown to increase free testosterone and lean mass in BCS without concomitant changes in % body fat [53
]. Kenney et al. [54
] reported preservation of lean mass with a combination of dihydroepiandrostenedione supplementation plus exercise in frail older women compared to losses in controls. A possible synergistic effect of AIs and resistance training on lean mass in BCS is an intriguing finding that is worth further study.
The strengths of our study are the translation of a targeted bone loading program shown effective in non-clinical populations to improve BMD in cancer survivors, our study of older BCS who are rarely the focus of exercise trials, and exclusion of bisphosphonate use to reduce confounding. Notable limitations of our study include a lack of a usual care control group, inclusion of aerobically active women, and modest attrition. Given the known benefits of physical activity we felt it would be unethical to assign women to a non-exercise control group. We selected an exercise type, e.g., stretching, that was the least likely to produce an osteogenic stimulus to bone and expend energy, but that could produce benefits for range of motion and well being. The 12-month changes in controls were minimal or followed age-expected trends and so we felt the flexibility program was appropriate. We did not exclude women who were aerobically active from participating because impact and resistance exercise load the skeleton in distinctly different ways than aerobic exercise and are superior for osteogenesis [36
]. However, aerobically active participants may have replaced their habitual activity with the intervention program causing no net change in energy expenditure or body fat (). Retention in the study was modest (62%) and this may have affected the ability of the program to shift lean and fat mass and bone mass at the hip, leaving us potentially underpowered for these variables. However, we fell only three participants short of the estimated sample size for the per protocol analysis, and we did not have any borderline P
values where an inadequate sample size might bring into question whether or not we were underpowered to detect significance. Exercise studies that include a BMD outcome should ideally last a year or longer to allow for completion of bone remodeling cycles [27
], but the tradeoff to longer programs is risk of attrition and waning adherence over time. Only three randomized, controlled trials with center-based training programs in BCS have lasted for 1 year and all reported lower attrition [51
]; however, ours was the only study specific to older women and that required women to attend scheduled group exercise classes. Most women cited logistical issues for dropping out (), which we largely attribute to the constrained schedule, unusual class times, inconvenient parking and travel to the facility. Clearly, we need to better understand the conditions that will yield optimal adherence to resistance training in older in BCS to promote this program in clinical practice. Adherence among women who stayed in the program was comparable to the other center-based trials [51
]. Even with drop out, retention in our study is better than that reported for bisphosphonate use with approximately 50% of patients who are prescribed this medication discontinuing use within 6–12 months [58
Our program was able to preserve bone mass at the spine and lower bone turnover rates, both factors that translate to reductions in fracture risk. Fracture risk at the spine and hip increases 26 and 55% after breast cancer diagnosis, respectively [4
], and both fracture types are associated with excess mortality and morbidity [59
]. Fractures in BCS add to the economic and personal costs that already result from treatment aimed to eradicate disease, particularly for older women [61
]. Our POWIR program was able to preserve bone and slow turnover and each could contribute to lower fracture risk at the spine. A 1–2% increase in BMD translates to a 7–14% decrease in fracture risk [63
] and slowing of bone turnover can lower fracture risk at the hip and spine independent of BMD [64
]. It was encouraging that the training effect on spine BMD and turnover were independent of adjuvant hormone therapy use so that women on bone-depleting AIs still benefited from exercise. Whether exercise can offer a lifestyle strategy to reduce hip fracture risk by preserving bone mass at the hip in BCS remains unclear.