Weight Loss and Breast Cancer Biomarkers in a Randomized Dietary Trial Among Overweight/Obese Premenopausal Women
Susan Olivo-Marston, Ph.D, M.P.H., 1,2 Michael Pennell, Ph.D.,1,3 Cecilia R. DeGraffinreid, M.H.S.,1 Mara Z. Vitolins, Dr.P.H.,4 Marisa Bittoni, Ph.D.,1,5 Elizabeth Grainger, R.D.1,6, Ph.D., Steven K. Clinton, M.D., Ph.D.,1,7 and Electra D. Paskett*, Ph.D.1,8
1Comprehensive Cancer Center, The Ohio State University, Columbus, Ohio
2Division of Epidemiology, College of Public Health, The Ohio State University, Columbus, Ohio
3Division of Biostatistics, College of Public Health, The Ohio State University, Columbus, Ohio
4Division of Epidemiology and Prevention, Department of Public Health Sciences, Wake Forest University, Winston-Salem, North Carolina, USA
5 Health and Rehabilitation Sciences, College of Medicine, The Ohio State University, Columbus, Ohio
6Office of Health Sciences, Comprehensive Cancer Center, The Ohio State University, Columbus, Ohio
7Hematology/Oncology, Department of Internal Medicine, College of Medicine, The Ohio State University, Columbus, Ohio
8Division of Cancer Prevention and Control, College of Medicine, The Ohio State University, Columbus, Ohio
Background: This was a pilot study that was conducted to: 1) determine feasibility ofrecruitment, adherence, and follow-up of calorie-restricted low-fat or low-carbohydrate diet plus exercise intervention among overweight and obese premenopausal women; 2) determine if these interventions promoted weight loss; and 3) investigate whether the interventions altered breast cancer biomarkers.
Methods: 79 overweight and obese premenopausal women were enrolled in a 52-weekrandomized intervention trial of calorie-restricted low-fat versus calorie-restricted low-carbohydrate diet along with physical activity. Changes in body mass index (BMI) and breast cancer biomarkers such as insulin-like growth factor 1 (IGF-1) and insulin-like growth factor binding protein-3 (IGFBP-3) were measured at baseline, 12, 34, and 52 weeks.
Results: Among the 550 women screened, 81 were eligible and 79 agreed to participate. Boththe calorie-restricted low-fat and low-carbohydrate diets plus exercise resulted in significant weight loss (average of 3.2 kg across arms, p<0.01). There were no significant changes in IGF-1 or IGFBP-3 in either group; however, IGF-1 had a significant positive relationship with weight loss (p=0.03), with a similar trend observed in each group (p=0.14). Insulin levels decreased an average of 23% between baseline and 52 weeks (p=0.052), with the same trend observed in each group (p=0.88).
Conclusions: The feasibility of adherence to both calorie-restricted diets plus exercise over a52-week intervention was low; however, both resulted in weight loss. Weight loss may lead to an increase in IGF-1 and a decrease in insulin. Although this was a relatively small pilot study, a larger study is needed to follow-up on these results.
Keywords: Breast Cancer; Low-Fat; Low-Carbohydrate; Physical Activity; Insulin-LikeGrowth Factor-1
Breast cancer is the most common type of cancer diagnosed among U.S. women, with approximately 232,670 newly diagnosed cancers estimated in 2014 . It has been suggested that 30-50% of cancers may be attributed to potentially modifiable lifestyle factors, such as diet and exercise . Additionally, the World Cancer Research Fund (WCRF) and American Institute for Cancer Research (AICR) concluded that there is convincing evidence that being overweight or obese increases the risk of postmenopausal breast cancer ; however, obesity has been associated with a decreased risk of premenopausal breast cancer . Furthermore, weight gain during adulthood, especially after menopause, has been consistently associated with an increased risk of breast cancer in several large cohort and case-control studies [5-8]. Therefore, due to its modifiable nature, weight loss prior to menopause is an attractive breast cancer prevention strategy because it would lead to lower rates of postmenopausal obesity.
There is considerable controversy regarding the association of specific dietary patterns with breast cancer risk. Several studies, along with the AICR/WCRF panel, have suggested that a diet high in fruits and vegetables and low in fat and alcohol, as well as regular physical activity, reduces the risk of breast cancer [3, 9, 10]. However, the Nurses’ Health Study demonstrated that a low-carbohydrate diet was associated with decreased risk of estrogen receptor- (ER) negative breast cancer . Unlike specific dietary patterns, the association between physical activity and breast cancer appears to be clearer. Several studies have found a dose-dependent, significant inverse association with breast cancer risk among both pre- and post-menopausal women and the average risk may be between 30-40% [2, 12, 13].
The biological mechanisms linking diet, physical activity, and breast cancer are unclear; however, they may involve alteration of insulin-like growth factor-1 (IGF-1). The bioavailabilityof IGF-1 depends in part on the presence of insulin-like growth factor binding proteins (IGFBPs) and, in particular, IGFBP-3 which binds approximately 75% of IGF-1 . Many epidemiologic studies have shown that elevated levels of IGF-1, or the ratio of IGF-1 to IGFBP-3, are associated with an increased risk for breast cancer among premenopausal women, [15-19] as well as postmenopausal women . Physical activity has been shown to affect IGF-1 and IGFBP-3 levels [21-24]. Finally, IGFBP activity has been noted to increase with physical activity leading to an overall decrease in IGF-1 bioavailability .
To the best of our knowledge, no randomized trials have been conducted among premenopausal women comparing a calorie-restricted low-fat diet plus exercise versus a calorie-restricted low-carbohydrate diet plus exercise on biomarkers of breast cancer risk. Therefore, the current study was conducted to determine the feasibility of recruiting premenopausal women to a calorie-restricted low-fat versus a calorie-restricted low-carbohydrate diet plus exercise intervention to determine adherence. Furthermore, we investigated whether these interventions would result in weight loss, or an alteration in biomarkers of breast cancer risk, including IGF-1 and IGFBP-3 levels.
Materials and Methods
Figure 1. CONSORT Flow Diagram for the LEAF study.
*Significant difference in trend by diet (p=0.050). No difference in net weight loss between diets (p=0.63).
**Significant trend in time across diets (p<0.01).
***Average trend in time across diets marginally significant (p=0.052).
There was a significant difference in the weight loss trend between the two arms (p=0.050) with women on the low-carbohydrate diet experiencing a steeper decline in weight initially, but at around week 22 began to rebound toward the weight loss level of the women on the low-fat diet. At the conclusion of the study (week 52), the net weight loss was similar across arms (3.5 kg for low-fat and 2.9 kg for low-carbohydrate, p=0.63) (Figure 2).
A sensitivity analysis examining weight loss among just those women who were adherent to the interventions demonstrated similar results to when all women were included. There was a significant overall decline in weight (p<0.01), but there was no difference between the two groups (p=0.42) (Table 3). Similar results were observed when looking at women who were adherent to physical activity. Among the 50 women who were adherent to physical activity, there was significant weight loss (p<0.01), but no difference between groups (p=0.59).
Waist circumference and waist-to-hip ratio were also examined in both interventions over the 52-week study
Figure 2. Mean body weight by dietary intervention. Estimates are from a linear mixed model with a cubic trend in time. A significant amount of weight was lost in both the low-fat and the low-carboInsurance hydrate arms (p<0.01). The trend in weight loss differed between groups, though the net weight loss was the same at study conclusion (p=0.63).
(Table 2). There was a significant decrease in waist circumference means across groups (p=0.03), with a similar decline observed in each group (p=0.19). Results for waist circumference means were similar when analyses were restricted to physical activity adherent subjects (significant overall decline in waist circumference (p<0.01), no difference across groups). The overall decline in waist circumference among diet adherent subjects (2.7 cm) was similar to the decline among physical activity adherent subjects (2.6 cm), but was not significant (p=0.12) due to the small sample size (n=20 adherent subjects) (Table 3). Waist-to-hip ratio did not change in either group (p=0.88); similar results were observed among diet adherent and physical activity adherent subjects.
Table 3. Mixed model means of outcome measures among participants adherent to diet. Overall estimates are averages across the two diets. Estimates for weight assume a cubic trend in time. All other estimates assume a linear trend in time.
*Significant difference in trend by diet (p<0.01). No difference in net weight loss between diets (p=0.42).
**Average trend in time across diets significant (p=0.02).
A number of breast cancer biomarkers were measured at baseline, and weeks 12, 34, and 52, including IGF-1, IGFBP-3, and their ratio, and insulin (Table 2). No significant changes were observed in levels of IGF-1, IGFBP-3, or their ratio over the course of the intervention. However, there was a marginally significant decrease in insulin (overall decrease of 1.4 uIU/ml in the geometric mean across arms, p=0.052). There were no differences in the trend of any of the biomarker levels across the two interventions. Furthermore, the results for IGF-1, IGFBP-3, and IGF-1:IGFBP-3 ratio were consistent when examined only among the women who adhered either to the diet or physical activity. While the results for insulin were also consistent when the analysis was restricted to subjects adherent to physical activity, changes in insulin were more pronounced when we focused on diet-adherent subjects: there was a 50% decrease in mean insulin between baseline and 52 weeks for diet adherers compared to only a 23% decrease overall (Table 3).
Lastly, levels of IGF-1 and IGFBP-3 were measured in association with changes in weight. There was no difference across the interventions in changes in IGF-1 associated with changes in weight (p=0.14) (Figure 3A). However, the average trend across groups was significant (p=0.04) and suggests that regardless of diet arm, an increase in weight loss was associated with an increase in IGF-1. IGFBP-3 did not significantly change with weight (p=0.81), with the results being consistent across arms (p=0.40) (data not shown). There was a significant positive association between IGF-1:IGFBP-3 ratio and weight loss (p<0.01), which did not differ by diet (p=0.12) (Figure 3B).
Modifiable risk factors for breast cancer have been identified and are attractive targets for postmenopausal breast cancer prevention. Two of these risk factors, BMI and weight gain, have been associated with an increased risk of postmenopausal breast cancer, thus reduction and prevention of obesity prior to menopause is an appealing breast cancer prevention strategy. While calorie restriction and physical activity are two of the most effective methods to reduce obesity, the most effective diet modification for reducing obesity and impacting breast cancer prevention is currently unknown. The goal of this pilot intervention study was to determine the feasibility
Figure 3. Percent change in IGF-1 and IGF-1:IGF-BP3 ratio with weight loss in the LEAF study. A) There was no difference between the low-fat and low-carbohydrate arms in change in IGF-1 with weight loss (p=0.14). The average trend across groups was significant (p=0.04). B) There was no difference between the low-fat and low-carbohydrate There was no difference between the low- fat and low-carbohydrate arms in change in the ratio with weight loss (p=0.12). The average trend across groups was significant (p<0.01).
of recruiting premenopausal women to a dietary plus exercise intervention and determining adherence and dropout rates over a 52-week period. Secondary goals included whether the interventions were effective on weight loss and altering biomarkers of cancer risk.
In this feasibility study, recruitment was a challenge. Although 550 women were screened, only 81 were eligible. However, of those 81, 79 agreed to participate. Adherence to the dietary intervention over the 52-week period was very low, similar to what others have observed [34, 35]. Only 29% adhered to the low-carbohydrate diet and 22% adhered to the low-fat diet. Adherence to physical activity, however, was higher at 61% and 66% for women in the low-carbohydrate and the low-fat interventions, respectively. In addition, a significant number of women in each arm were lost to follow-up, with 41% women dropping out of the low-fat arm and 55% women dropping out of the low-carbohydrate arm, similar to what has been observed in other studies [34, 35].
Both the calorie-restricted low-fat and low-carbohydrate interventions resulted in significant weight loss; however, no significant difference in weight loss between the two arms was found. Many studies have observed that a low-carbohydrate diet is more successful in creating short-term weight loss compared to a low-fat diet [36-40]. However, other studiesrandomizing men and women to a low-fat versus low-carbohydrate diet for one or two years observed similar results to ours, i.e., that there was weight loss, but no difference between arms [36, 41-44]. These studies also observed that most of the weight loss occurred in the first 6 months and weight loss occurred faster among those assigned to a low-carbohydrate diet [43, 44]. In addition, it is possible that the distribution of macronutrients in our diets may not have been different enough to detect differences in weight, however; our goal was to use nutritionally replete diets for free-living participants. Although the participants in these studies had a mean BMI similar to our study, most consisted of both men and women and many of the women in these studies were postmenopausal.
We did not observe any change in waist-to-hip ratio among women in either the low-fat or low-carbohydrate interventions. Similarly, another randomized study of low-fat or low-carbohydrate diets conducted among obese men and women did not observe a change in waist-to-hip ratio . However, their follow-up was limited to three months, and in weight loss interventions, the biggest changes are usually observed within the first six months [40, 44, 45]. Few studies have examined the effect of these two diets on waist circumference but, of those studies, most have observed a significant decrease in waist circumference in both diets with no significant differences between diets [40, 41, 45, 46]. Furthermore, many studies have not observed a significant association between waist circumference and increased breast cancer risk after adjusting for BMI .
Levels of IGF-1 did not change significantly in either the low-fat or low-carbohydrate diet groups. The same was observed with levels of IGFBP-3. Mouse studies have demonstrated the ability of a low-carbohydrate diet to reduce IGF-1 levels [48-50] and increase IGFBP-3 levels ; however, some of these studies compared low-carbohydrate diets to calorierestriction and observed that the animals on a low-carbohydrate diet had higher levels of IGF-1 . There are few studies in the current literature examining the effects of a low-fat or low-carbohydrate diet on IGF-1 and IGFBP-3 levels in humans. One study done only in men demonstrated an increase in IGF-1 levels among men on a low-fat diet; however, this was in comparison to a high fat diet and was conducted as a crossover feeding study . This same study reported no differences in IGFBP-3 levels. The authors suggested this effect could have been due to a carryover effect . Similar to our study, others have also detected no changes in IGF-1 levels [53, 54] or IGFBP-3 [53-56] in participants following a low-fat diet. There are a few studies, however, that have observed a decrease in IGF-1 [56-59] and an increase in IGFBP-3  when a low-fat diet was consumed. There is a suggestion that simple overall calorie restriction may alter IGF-1 and IGFBP-3 levels in humans [60, 61]. One study conducted in premenopausal women demonstrated that calorie restriction significantly decreased IGF-1 and increased IGFBP-3 levels . However, other studies have shown an increase in IGF-1 with a low calorie diet . There also appears to be a link between changes in IGF-1 with physical activity. Among women, physical activity has been associated with a reduction in IGF-1 [63-66] and increase in IGFBP-3 levels [25, 63, 64]. However, some studies observed increases in IGF-1 with physical activity , while others observed no difference in IGF-1 or IGFBP-3 [67-69]. Therefore, additional investigations are required to elucidate the association between IGF-1, IGFBP-3, diet, and physical activity.
Although both the low-fat and low-carbohydrate diets resulted in significant weight loss, there was no significant change in IGFBP-3 with change in weight. However, IGF-1 was significantly positively associated with weight loss. Other studies have observed a significant increase in IGF-1 following weight loss [70-72]. The peptide hormone ghrelin is reduced inobese individuals and has been demonstrated to increase after weight loss [73-75]. It has been hypothesized that ghrelin may be one factor involved in the increase in IGF-1 after weight loss . Insulin has also been hypothesized to stimulate production of IGF-1 in obese women [71, 76], although our results contradict this theory.
This study has some limitations. First, the number of participants was relatively small with 41 women randomized to a low-fat diet and 38 women randomized to a low-carbohydrate diet. Therefore, because this was a pilot study it is possible that we were underpowered to observe a difference in weight loss and biomarkers between dietary interventions. This study limited participants to premenopausal women with a BMI between 25 and 34, therefore, future studies are necessary to determine if similar results are observed in postmenopausal women and women with a BMI above 34. Second, adherence to the dietary interventions was low. Therefore, it is possible that this low adherence resulted in smaller changes to the biomarkers measured. Third, physical activity was self-monitored by the participants using a daily physical activity log. Although daily logs are useful because activity is recorded each day, they can sometimes be completed inaccurately and can influence a subject’s behavior . Diet records are also a limitation as they are also self-reported, they may not contain complete data, and subjects may alter their eating patterns due to the burden of recording their food consumption or to please the investigators . Finally, there was a high drop-out rate, which is common in diet intervention studies.
Despite the limitations, the study had several strengths. This was a randomized trial comparing low-fat versus low-carbohydrate diets. There are relatively few studies comparing low-fat versus low-carbohydrate diets for greater than six months, and the current study followed women for a year. Furthermore, not only was weight loss measured, but the current study alsomeasured specific biomarkers that have been associated with breast cancer risk. These biomarkers were assessed at several time points during the study to examine changes over the course of the interventions. BMI was also assessed along with waist and hip measurements. Finally, the use of mixed models in our analysis allows the use of all available measurements and provides unbiased estimates of treatment and time effects under the assumption that missing follow-up measurements were missing at random .
Although this is a small randomized study, several important results can inform future studies, which are as follows: both diets may produce equivalent weight loss; adherence to diet changes is difficult to achieve while adherence to exercise regimens may be easier to attain; and biomarkers of breast cancer risk do not appear to be widely impacted by diet and physical activity changes. These findings, suggest the need for larger randomized studies examining diet, physical activity, and biomarkers of breast cancer risk to determine whether these modifiable factors can result in breast cancer prevention.
The project described was supported by grants from the Breast CancerResearch Foundation, the National Center for Advancing Translational Sciences (8UL1TR000090-05), and the National Cancer Institute (K225K22CA140860-02). The content is solely the responsibility of the authors and does not necessarily represent the official views of the Breast Cancer Research Foundation, the National Center for Advancing Translational Sciences or the National Institutes of Health. There are no conflicts of interest to disclose.
Cite this article: Electra P. Weight Loss and Breast Cancer Biomarkers in a Randomized Dietary Trial Among Overweight/Obese Premenopausal Women. J J Food Nutri. 2014, 1(1): 004.