Prior studies found that carbohydrate restriction slows prostate tumor growth in murine xenograft models relative to high-fat/high-carbohydrate diets (5
). Moreover, carbohydrate restriction resulted in marked reductions in serum insulin and/or IGF-I, potentially mediating tumor growth delay. However, all previous studies were done in the context of calorie restriction and/or weight loss, and thus whether results were due to either overall energy restriction or specifically carbohydrate restriction remained unclear. We therefore sought to test carbohydrate restriction without energy restriction to clarify whether carbohydrate restriction alone results in tumor growth delay. Furthermore, our study addressed the concern that carbohydrate-restricted diets contain excessive fat, which may stimulate tumor growth (9
). This concern was particularly relevant to our formulation of a NCKD because it was largely composed of saturated fats from milk and lard for an overall caloric composition of 83% from fat. Our study also provided the opportunity to further elucidate the molecular mechanisms through which variations in dietary carbohydrate may influence prostate tumor growth.
We compared the effects of NCKD, LFD, and MCD on the growth of the commonly used prostate cancer cell line LNCaP. This cell line offers several advantages when testing the role of a NCKD on prostate tumor growth. First, LNCaP is a commonly used model of hormone-responsive prostate cancer, potentially broadening the applicability of our results to previous and future studies. Second, LNCaP exhibits stimulated growth in response to increased dietary fat consumption and thus is a reasonable model to test whether the effects of carbohydrate restriction outweigh the effects of high fat intake on tumor growth (16
). In addition, LNCaP undergoes increased cell proliferation in vitro
in response to insulin (17
) and IGF-I (2
), further supporting the use of this cell line as an appropriate model to examine whether alterations of IGF axis hormones mediated by diet can modulate prostate tumor growth.
Before conducting the present study, we conducted a pilot feeding study (without tumor injection) to determine the calorie intake level necessary to maintain similar body weights among diet groups. Previous observations revealed that NCKD mice consumed more calories than LFD-fed mice in ad libitum
), consistent with our pilot study that revealed that NCKD mice required 12.5% more calories relative to LFD mice to maintain similar body weights. Despite using this information to guide our feeding protocol, we nonetheless observed that the NCKD group remained consistently heavier than the other two groups, even after reducing the daily calorie supplement early in the study from 12.5% to 10%. Prior studies using isocaloric feeding reported weight loss for mice on carbohydrate-restricted diets relative to comparison diets (5
). As such, this weight loss in previous studies may have biased findings in favor of carbohydrate restriction. Conversely, in the current study, weight gain among NCKD mice may have biased the findings against the NCKD. Overall, our experience highlights the complexities of replicating animal feeding behaviors, which may be dependent on genetic and/or environmental factors difficult to control in successive experiments. This is particularly challenging when testing more than two diets because the potential for intergroup variation is greater than in studies involving only two groups (6
Diet group was significantly associated with overall survival, with both the LFD and NCKD groups experiencing significantly longer survival than the MCD group. The survival benefit in the NCKD group was accompanied by favorable changes in both serum levels of key IGF axis hormones and in IGF axis signaling within the tumor as measured by the downstream marker p-AKT. This suggests that reduced IGF axis signaling may underscore the observed survival benefit of a NCKD. Although survival was similarly prolonged in the LFD group, there were fewer significant changes in serum IGF axis hormones and less reduction in p-AKT levels in this diet group, suggesting the LFD may have growth inhibitory effects that are less strongly mediated by IGF signaling.
Our results are consistent with prior studies that examined carbohydrate-restricted diets in the context of energy restriction. In particular, our finding that carbohydrate restriction is associated with the lowest serum levels of IGF-I is consistent with previous investigations (5
). However, in contrast to previous studies (6
), we further clarified this finding by measuring serum levels of IGFBP-1 and IGFBP–3 to assess the fraction of free or bioactive hormone reflected by the values of IGF-I/IGFBP-1 and IGF-I/IGFBP-3 ratios. Ultimately, we found the NCKD led to significant reductions in the IGF-I/IGFBP-3 ratio relative to the MCD. We also observed that IGFBP-1 was significantly higher in the NKCD group relative to both the LFD and MCD groups. In addition, the IGF-I/IGFBP-1 ratio, a marker of fasting IGF axis activity that has not previously been measured in studies of carbohydrate restriction, was lowest in the NCKD group. We also observed that the NCKD led to the highest serum levels of IGFBP-2, which may further reduce circulating free IGF-I. Together, these findings provide the strongest evidence to date that a NCKD is associated with beneficial changes across multiple parameters of the IGF axis to slow prostate tumor growth.
Consistent with the observed IGF axis measurements, AKT signaling in tumor tissue revealed favorable changes for the NCKD, in line with previous results (6
). In particular, both intracellular levels of p-AKT and percent activated AKT, as measured by the p-AKT/t-AKT ratio, were lowest in the NKCD group relative to the MCD and LFD groups. When analyzing these findings in the context of IGF axis hormones, we observed a novel correlation between the serum IGF-I/IGFBP-3 ratio and tumor p-AKT/t-AKT expression. This result has not previously been reported in the context of xenograft studies of carbohydrate restriction and may further underscore the importance of IGF-I to AKT signaling and overall tumor growth in dietary strategies modulating carbohydrate intake.
In the current study, serum insulin levels were lowest for the LFD group and were significantly lower in both the LFD and NCKD groups relative to the MCD group. However, serum IGF-I was highest in the LFD group, which approached statistical significance relative to the MCD group (P
= 0.07) and was significantly higher than the NCKD group (P
= 0.01). Thus, the benefits of a LFD on the IGF axis seems to be less pronounced in the current study compared with clearer benefits observed previously (5
Our prior study comparing these same three diets (5
) suggested that a NCKD prolonged survival relative to a LFD, although this did not reach statistical significance (hazard ratio, 0.71; P
= 0.24). Therefore, we powered the current study to detect a 29% prolongation of survival, which required 50% more mice than our prior study. Ultimately, we observed no significant difference in survival between the LFD and NCKD groups although the NCKD was associated with several metabolic changes suggestive of more favorable outcomes than the LFD. Of note, NCKD mice were significantly heavier than LFD mice during the majority of the experiment. Given that lower body weight may slow tumor growth (8
), this may have biased our results against the NCKD, potentially explaining the lack of difference in tumor growth and/or survival between the LFD and NCKD groups. Alternatively, it is possible that dietary fat may stimulate LNCaP xenograft growth to a greater degree than other cell lines, partially negating the superior benefits of the NCKD on IGF-I signaling. Ultimately, we found no evidence that a LFD, the current recommended diet for cancer patients, was superior to a NCKD. However, it should be noted that the LFD had a higher level of simple carbohydrates in the form of sucrose than what is recommended in humans (19
). However, a similar LFD with high sucrose content was used in prior preclinical studies and led to slower tumor growth and delayed development of androgen resistance relative to a MCD (9
). Ultimately, whether similar results would be obtained by alternative formulations of carbohydrate composition in a LFD remains unknown.
To further explore the complex molecular mechanisms through which dietary carbohydrate variation may affect cancer growth, we performed gene expression analysis of tumor tissue. Our expression analysis showed that pathways related to insulin resistance, obesity, and anti-apoptosis were up-regulated in tumors from MCD mice relative to tumors from NCKD and LFD mice. In particular, NF-κB and MAP kinase pathways were up-regulated in MCD tumors. Although NF-κB is normally expressed at low levels in LNCaP cells (19
), increased NF-κB levels would promote anti-apoptotic pathways (19
). In addition, higher serum insulin levels may stimulate tumor growth by activating MAP kinase pathways (20
). Further increases in MAP kinase signaling by elevated serum IGF-I may synergize with insulin signaling through the MAP kinase pathway to further promote tumor growth (21
) in the MCD group relative to the NCKD and LFD groups.
Assessment of liver histology revealed that the NCKD group had the least fatty infiltration. Similar benefits from carbohydrate restriction were previously observed in patients with nonalcoholic fatty liver disease (22
). In that study, four of five patients showed histologic improvements in fatty infiltration, inflammation, and fibrosis after 6 months of dietary carbohydrate restriction. Further benefits for these patients included marked weight loss and favorable changes in high-density lipoprotein, low-density lipoprotein, and triglycerides, trends indicating reduced cardiovascular risk. Indeed, numerous clinical trials show that carbohydrate restriction leads to sustained weight loss for up to 1 year with beneficial changes in serum lipoproteins such as high-density lipoprotein, total cholesterol, and triglycerides, suggesting that carbohydrate-restricted diets may have cardiac benefits (23
). These benefits are important given that cardiovascular morbidity is a major cause of mortality among prostate cancer patients (29
One primary limitation of the present study was that mice were randomized before tumor injection, whereas previous investigations randomized mice after tumor injection or palpability (6
). Typically, lifestyle modifications, including dietary changes, are used after cancer diagnosis. We used the current model to maximize the time mice consumed their respective diets to show proof of concept before conducting a postinjection study. A second limitation is that a diet devoid of carbohydrates may not be feasible in a clinical setting. Alternatively, further research may determine an upper limit of carbohydrate intake that preserves tumor inhibitory effects. In addition, our formulation of a NCKD may pose practical challenges given its high saturated fat content primarily from milk and lard. Although we purposefully formulated the NCKD in this manner to test whether benefits of carbohydrate restriction were sufficiently robust to be observed with a diet composed mostly of saturated fat, future studies should be aimed at testing NCKDs composed of mixtures of other types of fats (e.g., mono- and poly-unsaturates) that may extend survival beyond that observed in this study. Finally, we cannot be certain that removing dietary carbohydrate via a NCKD was responsible for the tumor growth delay observed in our study. For example, it is possible that fat intake may have a biphasic effect on tumor growth and that extremely high fat intake may slow growth. Alternatively, the high fat content of the NCKD may have facilitated uptake of fat-soluble vitamins, which themselves exert anti-tumor effects. Moreover, it is possible that changes in serum levels of androgen and/or prostate-specific antigen in response to fat intake may account for at least some of the differential effects on growth kinetics of tumors. However, previous studies have argued against such an effect (9
). Similarly, whereas it remains possible that alterations in serum levels of angiogenic factors may have played a role in our observations, the gene expression data did not reveal such changes between diet groups. Lastly, it is possible that ketone bodies themselves are toxic to tumors. Thus, whereas we clearly show significant changes in the IGF axis associated with a NCKD, other mechanisms unrelated to dietary carbohydrate intake per se may have contributed to tumor growth delay.
Overall, our current findings extend observations from previous preclinical studies to show that varying dietary carbohydrate intake is associated with benefits independent from those of energy restriction. As such, it remains plausible that clinical studies may reveal that a carbohydrate-restricted diet coupled with significant weight loss may result in additive benefits to further slow tumor growth than either carbohydrate or energy restriction alone, as revealed in preclinical studies to date.