Aggressive tumor proliferation leads to insufficient tumor vascularization, resulting in chronic tumor hypoxia, which forces cancer cells to become highly glycolytic. Here, we show that lowering systemic glucose by the simultaneous reduction in dietary carbohydrates and inhibiting gluconeogenesis delays the development of hypoxic breast cancer in vitro and in vivo.
The results of this study demonstrate that hypoxic tumor tissues are susceptible to even mild glucose limitation. First, we confirmed that breast cancer cells rely on an abnormally high glucose level to survive in a hypoxic environment in tissue culture. Second, using two aggressive breast cancer mouse models, we showed that a glucose-lowering regimen consisting of a combination of two modalities -- a low carbohydrate (ketogenic) diet and metformin -- significantly decreased tumor burden by 2/3 compared to the control or each modality alone. Moreover, tumors in the ketogenic diet-metformin group grew 38% more slowly, resulting in an additional 31 days of overall survival. This life extension equates to more than three human years (52), a significant increase over the current median TNBC survival of 18 months (42). Third, we showed that the median latency of breast tumors in mice using our combination treatment increased by 36% compared to the median latency of other groups. Lastly, since micrometastases are hypoxic due to the lack of newly–formed vascularization (5, 11), we obtained preliminary evidence that metastasis to the lungs may also be delayed (see Supplementary materials).
Limiting glucose with a combination of a ketogenic diet plus metformin regimen to slow cancer growth has been independently proposed (53, 54) and this combination regimen has been safely used in humans for a different purpose (55). Furthermore, timed metformin dosing during transient hypoglycemia caused by intermittent fasting, strongly inhibited melanoma-derived tumor growth (56). Other ways to limit systemic glucose levels are also under investigation. Several studies described the direct cytotoxic action of metformin in low glucose conditions (57, 58). Additionally, glycolytic tumors have been targeted through inhibition of glycolysis (59), the PI3 Kinase/Akt/mTORc growth signaling pathway (60) or by blocking glucose transport (61, 62). However, as with conventional chemotherapies, tumor evolution can circumvent these targeted chemotherapies, leading to cancer recurrence. Additionally, these molecular approaches may be ineffective or toxic, as some molecular targets are redundant or indiscriminate and normal cells may also rely on these activities. In contrast, lowering systemic glucose via the combined regimen proposed here adopts an “organismic” view of cancer (63) by safely modifying organismal physiology rather than targeting a unique cancer activity.
Confirming our findings, diabetic cancer patients taking metformin exhibit a significantly lower incidence of hepatic, colorectal, mammary and pancreatic cancers and increased survival from colorectal, pulmonary and prostate cancers than those on other antidiabetic medications that do not inhibit gluconeogenesis (66, 67). Most probable explanation is that diabetic patients tend to control their carbohydrate intake better than the general population (68), boosting metformin’s anticancer effect. It follows that a low carbohydrate ketogenic diet in combination with metformin may potentiate metformin’s anti-carcinogenic action in cancer patients regardless of their diabetic status, as we observed in our mouse models.
An alternative explanation is that а concurrent decrease in insulin levels caused by low glucose slows tumor growth. This would mean that in the presence of insulin, the normoglycemic and hypoxic environment should allow cancer cells to proliferate. However, our work shows that the normoglycemic (1g/L) insulin–containing growth medium did not support hypoxic PyMT cancer cell viability. Instead, to survive, MET-1 breast cancer cells required a “diabetic” 4.5g/L glucose level in the DMEM culture medium containing insulin. This observation implies a direct effect of glucose levels on cancer cell growth rather than the indirect effect of lower insulin. While insulin is important in the promotional stage of breast tumorigenesis, a large proportion of advanced ER-negative breast adenocarcinomas do not show a mitogenic response upon insulin signaling in culture (69). Moreover, hyperinsulinemia tends to be irrelevant to breast cancer risk for premenopausal women while potentially increasing it for post-menopausal women (70). Evidence in cell culture, mice and humans demonstrates that hyperglycemia is a bona fide risk factor, at least for ER-negative breast cancer such as TNBC.
While we observed a significant decrease in tumor burden, growth rate and an increase in tumor latency with a mild decrease in systemic glucose using a combination of a clinically relevant dose of metformin and a ketogenic diet, the treatment did not inhibit tumor growth altogether. One explanation is that properly oxygenated, and, therefore, nonglycolytic tumor cells would not be susceptible to this regimen. Since well-oxygenated, proliferating cancer cells can be targeted by chemo-, radio- and immunotherapies, our metabolic regimen is a natural candidate for combination with these therapies for synergistic therapeutic effects. Finally, this metabolic regimen may be similarly effective against a broad range of other FDG-PET- positive (glycolytic) tumors in other organs (10, 18, 71).